source: src/molecules.cpp@ edcda5

Action_Thermostats Add_AtomRandomPerturbation Add_FitFragmentPartialChargesAction Add_RotateAroundBondAction Add_SelectAtomByNameAction Added_ParseSaveFragmentResults AddingActions_SaveParseParticleParameters Adding_Graph_to_ChangeBondActions Adding_MD_integration_tests Adding_ParticleName_to_Atom Adding_StructOpt_integration_tests AtomFragments Automaking_mpqc_open AutomationFragmentation_failures Candidate_v1.5.4 Candidate_v1.6.0 Candidate_v1.6.1 Candidate_v1.7.0 ChangeBugEmailaddress ChangingTestPorts ChemicalSpaceEvaluator CombiningParticlePotentialParsing Combining_Subpackages Debian_Package_split Debian_package_split_molecuildergui_only Disabling_MemDebug Docu_Python_wait EmpiricalPotential_contain_HomologyGraph EmpiricalPotential_contain_HomologyGraph_documentation Enable_parallel_make_install Enhance_userguide Enhanced_StructuralOptimization Enhanced_StructuralOptimization_continued Example_ManyWaysToTranslateAtom Exclude_Hydrogens_annealWithBondGraph FitPartialCharges_GlobalError Fix_BoundInBox_CenterInBox_MoleculeActions Fix_ChargeSampling_PBC Fix_ChronosMutex Fix_FitPartialCharges Fix_FitPotential_needs_atomicnumbers Fix_ForceAnnealing Fix_IndependentFragmentGrids Fix_ParseParticles Fix_ParseParticles_split_forward_backward_Actions Fix_PopActions Fix_QtFragmentList_sorted_selection Fix_Restrictedkeyset_FragmentMolecule Fix_StatusMsg Fix_StepWorldTime_single_argument Fix_Verbose_Codepatterns Fix_fitting_potentials Fixes ForceAnnealing_goodresults ForceAnnealing_oldresults ForceAnnealing_tocheck ForceAnnealing_with_BondGraph ForceAnnealing_with_BondGraph_continued ForceAnnealing_with_BondGraph_continued_betteresults ForceAnnealing_with_BondGraph_contraction-expansion FragmentAction_writes_AtomFragments FragmentMolecule_checks_bonddegrees GeometryObjects Gui_Fixes Gui_displays_atomic_force_velocity ImplicitCharges IndependentFragmentGrids IndependentFragmentGrids_IndividualZeroInstances IndependentFragmentGrids_IntegrationTest IndependentFragmentGrids_Sole_NN_Calculation JobMarket_RobustOnKillsSegFaults JobMarket_StableWorkerPool JobMarket_unresolvable_hostname_fix MoreRobust_FragmentAutomation ODR_violation_mpqc_open PartialCharges_OrthogonalSummation PdbParser_setsAtomName PythonUI_with_named_parameters QtGui_reactivate_TimeChanged_changes Recreated_GuiChecks Rewrite_FitPartialCharges RotateToPrincipalAxisSystem_UndoRedo SaturateAtoms_findBestMatching SaturateAtoms_singleDegree StoppableMakroAction Subpackage_CodePatterns Subpackage_JobMarket Subpackage_LinearAlgebra Subpackage_levmar Subpackage_mpqc_open Subpackage_vmg Switchable_LogView ThirdParty_MPQC_rebuilt_buildsystem TrajectoryDependenant_MaxOrder TremoloParser_IncreasedPrecision TremoloParser_MultipleTimesteps TremoloParser_setsAtomName Ubuntu_1604_changes stable
Last change on this file since edcda5 was 4158ba, checked in by Frederik Heber <heber@…>, 17 years ago

BUGFIX: molecule::CreateAdjacencyList() used CandidateBondNo in output even if no candidate had been found

+ If no Candidate has been found, output message would declare to be unable to correct bond degree for a unspecified bond (CandidateBondNo set to no sensible value)

  • Property mode set to 100644
File size: 237.3 KB
Line 
1/** \file molecules.cpp
2 *
3 * Functions for the class molecule.
4 *
5 */
6
7#include "molecules.hpp"
8
9/************************************* Other Functions *************************************/
10
11/** Determines sum of squared distances of \a X to all \a **vectors.
12 * \param *x reference vector
13 * \param *params
14 * \return sum of square distances
15 */
16double LSQ (const gsl_vector * x, void * params)
17{
18 double sum = 0.;
19 struct LSQ_params *par = (struct LSQ_params *)params;
20 Vector **vectors = par->vectors;
21 int num = par->num;
22
23 for (int i=num;i--;) {
24 for(int j=NDIM;j--;)
25 sum += (gsl_vector_get(x,j) - (vectors[i])->x[j])*(gsl_vector_get(x,j) - (vectors[i])->x[j]);
26 }
27
28 return sum;
29};
30
31/************************************* Functions for class molecule *********************************/
32
33/** Constructor of class molecule.
34 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
35 */
36molecule::molecule(periodentafel *teil)
37{
38 // init atom chain list
39 start = new atom;
40 end = new atom;
41 start->father = NULL;
42 end->father = NULL;
43 link(start,end);
44 // init bond chain list
45 first = new bond(start, end, 1, -1);
46 last = new bond(start, end, 1, -1);
47 link(first,last);
48 // other stuff
49 MDSteps = 0;
50 last_atom = 0;
51 elemente = teil;
52 AtomCount = 0;
53 BondCount = 0;
54 NoNonBonds = 0;
55 NoNonHydrogen = 0;
56 NoCyclicBonds = 0;
57 ListOfBondsPerAtom = NULL;
58 NumberOfBondsPerAtom = NULL;
59 ElementCount = 0;
60 for(int i=MAX_ELEMENTS;i--;)
61 ElementsInMolecule[i] = 0;
62 cell_size[0] = cell_size[2] = cell_size[5]= 20.;
63 cell_size[1] = cell_size[3] = cell_size[4]= 0.;
64};
65
66/** Destructor of class molecule.
67 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
68 */
69molecule::~molecule()
70{
71 if (ListOfBondsPerAtom != NULL)
72 for(int i=AtomCount;i--;)
73 Free((void **)&ListOfBondsPerAtom[i], "molecule::~molecule: ListOfBondsPerAtom[i]");
74 Free((void **)&ListOfBondsPerAtom, "molecule::~molecule: ListOfBondsPerAtom");
75 Free((void **)&NumberOfBondsPerAtom, "molecule::~molecule: NumberOfBondsPerAtom");
76 CleanupMolecule();
77 delete(first);
78 delete(last);
79 delete(end);
80 delete(start);
81};
82
83/** Adds given atom \a *pointer from molecule list.
84 * Increases molecule::last_atom and gives last number to added atom and names it according to its element::abbrev and molecule::AtomCount
85 * \param *pointer allocated and set atom
86 * \return true - succeeded, false - atom not found in list
87 */
88bool molecule::AddAtom(atom *pointer)
89{
90 if (pointer != NULL) {
91 pointer->sort = &pointer->nr;
92 pointer->nr = last_atom++; // increase number within molecule
93 AtomCount++;
94 if (pointer->type != NULL) {
95 if (ElementsInMolecule[pointer->type->Z] == 0)
96 ElementCount++;
97 ElementsInMolecule[pointer->type->Z]++; // increase number of elements
98 if (pointer->type->Z != 1)
99 NoNonHydrogen++;
100 if (pointer->Name == NULL) {
101 Free((void **)&pointer->Name, "molecule::AddAtom: *pointer->Name");
102 pointer->Name = (char *) Malloc(sizeof(char)*6, "molecule::AddAtom: *pointer->Name");
103 sprintf(pointer->Name, "%2s%02d", pointer->type->symbol, pointer->nr+1);
104 }
105 }
106 return add(pointer, end);
107 } else
108 return false;
109};
110
111/** Adds a copy of the given atom \a *pointer from molecule list.
112 * Increases molecule::last_atom and gives last number to added atom.
113 * \param *pointer allocated and set atom
114 * \return true - succeeded, false - atom not found in list
115 */
116atom * molecule::AddCopyAtom(atom *pointer)
117{
118 if (pointer != NULL) {
119 atom *walker = new atom();
120 walker->type = pointer->type; // copy element of atom
121 walker->x.CopyVector(&pointer->x); // copy coordination
122 walker->v.CopyVector(&pointer->v); // copy velocity
123 walker->FixedIon = pointer->FixedIon;
124 walker->sort = &walker->nr;
125 walker->nr = last_atom++; // increase number within molecule
126 walker->father = pointer; //->GetTrueFather();
127 walker->Name = (char *) Malloc(sizeof(char)*strlen(pointer->Name)+1, "molecule::AddCopyAtom: *Name");
128 strcpy (walker->Name, pointer->Name);
129 add(walker, end);
130 if ((pointer->type != NULL) && (pointer->type->Z != 1))
131 NoNonHydrogen++;
132 AtomCount++;
133 return walker;
134 } else
135 return NULL;
136};
137
138/** Adds a Hydrogen atom in replacement for the given atom \a *partner in bond with a *origin.
139 * Here, we have to distinguish between single, double or triple bonds as stated by \a BondDegree, that each demand
140 * a different scheme when adding \a *replacement atom for the given one.
141 * -# Single Bond: Simply add new atom with bond distance rescaled to typical hydrogen one
142 * -# Double Bond: Here, we need the **BondList of the \a *origin atom, by scanning for the other bonds instead of
143 * *Bond, we use the through these connected atoms to determine the plane they lie in, vector::MakeNormalvector().
144 * The orthonormal vector to this plane along with the vector in *Bond direction determines the plane the two
145 * replacing hydrogens shall lie in. Now, all remains to do is take the usual hydrogen double bond angle for the
146 * element of *origin and form the sin/cos admixture of both plane vectors for the new coordinates of the two
147 * hydrogens forming this angle with *origin.
148 * -# Triple Bond: The idea is to set up a tetraoid (C1-H1-H2-H3) (however the lengths \f$b\f$ of the sides of the base
149 * triangle formed by the to be added hydrogens are not equal to the typical bond distance \f$l\f$ but have to be
150 * determined from the typical angle \f$\alpha\f$ for a hydrogen triple connected to the element of *origin):
151 * We have the height \f$d\f$ as the vector in *Bond direction (from triangle C1-H1-H2).
152 * \f[ h = l \cdot \cos{\left (\frac{\alpha}{2} \right )} \qquad b = 2l \cdot \sin{\left (\frac{\alpha}{2} \right)} \quad \rightarrow \quad d = l \cdot \sqrt{\cos^2{\left (\frac{\alpha}{2} \right)}-\frac{1}{3}\cdot\sin^2{\left (\frac{\alpha}{2}\right )}}
153 * \f]
154 * vector::GetNormalvector() creates one orthonormal vector from this *Bond vector and vector::MakeNormalvector creates
155 * the third one from the former two vectors. The latter ones form the plane of the base triangle mentioned above.
156 * The lengths for these are \f$f\f$ and \f$g\f$ (from triangle H1-H2-(center of H1-H2-H3)) with knowledge that
157 * the median lines in an isosceles triangle meet in the center point with a ratio 2:1.
158 * \f[ f = \frac{b}{\sqrt{3}} \qquad g = \frac{b}{2}
159 * \f]
160 * as the coordination of all three atoms in the coordinate system of these three vectors:
161 * \f$\pmatrix{d & f & 0}\f$, \f$\pmatrix{d & -0.5 \cdot f & g}\f$ and \f$\pmatrix{d & -0.5 \cdot f & -g}\f$.
162 *
163 * \param *out output stream for debugging
164 * \param *Bond pointer to bond between \a *origin and \a *replacement
165 * \param *TopOrigin son of \a *origin of upper level molecule (the atom added to this molecule as a copy of \a *origin)
166 * \param *origin pointer to atom which acts as the origin for scaling the added hydrogen to correct bond length
167 * \param *replacement pointer to the atom which shall be copied as a hydrogen atom in this molecule
168 * \param **BondList list of bonds \a *replacement has (necessary to determine plane for double and triple bonds)
169 * \param NumBond number of bonds in \a **BondList
170 * \param isAngstroem whether the coordination of the given atoms is in AtomicLength (false) or Angstrom(true)
171 * \return number of atoms added, if < bond::BondDegree then something went wrong
172 * \todo double and triple bonds splitting (always use the tetraeder angle!)
173 */
174bool molecule::AddHydrogenReplacementAtom(ofstream *out, bond *TopBond, atom *BottomOrigin, atom *TopOrigin, atom *TopReplacement, bond **BondList, int NumBond, bool IsAngstroem)
175{
176 double bondlength; // bond length of the bond to be replaced/cut
177 double bondangle; // bond angle of the bond to be replaced/cut
178 double BondRescale; // rescale value for the hydrogen bond length
179 bool AllWentWell = true; // flag gathering the boolean return value of molecule::AddAtom and other functions, as return value on exit
180 bond *FirstBond = NULL, *SecondBond = NULL; // Other bonds in double bond case to determine "other" plane
181 atom *FirstOtherAtom = NULL, *SecondOtherAtom = NULL, *ThirdOtherAtom = NULL; // pointer to hydrogen atoms to be added
182 double b,l,d,f,g, alpha, factors[NDIM]; // hold temporary values in triple bond case for coordination determination
183 Vector Orthovector1, Orthovector2; // temporary vectors in coordination construction
184 Vector InBondvector; // vector in direction of *Bond
185 bond *Binder = NULL;
186 double *matrix;
187
188// *out << Verbose(3) << "Begin of AddHydrogenReplacementAtom." << endl;
189 // create vector in direction of bond
190 InBondvector.CopyVector(&TopReplacement->x);
191 InBondvector.SubtractVector(&TopOrigin->x);
192 bondlength = InBondvector.Norm();
193
194 // is greater than typical bond distance? Then we have to correct periodically
195 // the problem is not the H being out of the box, but InBondvector have the wrong direction
196 // due to TopReplacement or Origin being on the wrong side!
197 if (bondlength > BondDistance) {
198// *out << Verbose(4) << "InBondvector is: ";
199// InBondvector.Output(out);
200// *out << endl;
201 Orthovector1.Zero();
202 for (int i=NDIM;i--;) {
203 l = TopReplacement->x.x[i] - TopOrigin->x.x[i];
204 if (fabs(l) > BondDistance) { // is component greater than bond distance
205 Orthovector1.x[i] = (l < 0) ? -1. : +1.;
206 } // (signs are correct, was tested!)
207 }
208 matrix = ReturnFullMatrixforSymmetric(cell_size);
209 Orthovector1.MatrixMultiplication(matrix);
210 InBondvector.SubtractVector(&Orthovector1); // subtract just the additional translation
211 Free((void **)&matrix, "molecule::AddHydrogenReplacementAtom: *matrix");
212 bondlength = InBondvector.Norm();
213// *out << Verbose(4) << "Corrected InBondvector is now: ";
214// InBondvector.Output(out);
215// *out << endl;
216 } // periodic correction finished
217
218 InBondvector.Normalize();
219 // get typical bond length and store as scale factor for later
220 BondRescale = TopOrigin->type->HBondDistance[TopBond->BondDegree-1];
221 if (BondRescale == -1) {
222 cerr << Verbose(3) << "WARNING: There is no typical bond distance for bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") of degree " << TopBond->BondDegree << "!" << endl;
223 BondRescale = bondlength;
224 } else {
225 if (!IsAngstroem)
226 BondRescale /= (1.*AtomicLengthToAngstroem);
227 }
228
229 // discern single, double and triple bonds
230 switch(TopBond->BondDegree) {
231 case 1:
232 FirstOtherAtom = new atom(); // new atom
233 FirstOtherAtom->type = elemente->FindElement(1); // element is Hydrogen
234 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
235 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
236 if (TopReplacement->type->Z == 1) { // neither rescale nor replace if it's already hydrogen
237 FirstOtherAtom->father = TopReplacement;
238 BondRescale = bondlength;
239 } else {
240 FirstOtherAtom->father = NULL; // if we replace hydrogen, we mark it as our father, otherwise we are just an added hydrogen with no father
241 }
242 InBondvector.Scale(&BondRescale); // rescale the distance vector to Hydrogen bond length
243 FirstOtherAtom->x.CopyVector(&TopOrigin->x); // set coordination to origin ...
244 FirstOtherAtom->x.AddVector(&InBondvector); // ... and add distance vector to replacement atom
245 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
246// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
247// FirstOtherAtom->x.Output(out);
248// *out << endl;
249 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
250 Binder->Cyclic = false;
251 Binder->Type = TreeEdge;
252 break;
253 case 2:
254 // determine two other bonds (warning if there are more than two other) plus valence sanity check
255 for (int i=0;i<NumBond;i++) {
256 if (BondList[i] != TopBond) {
257 if (FirstBond == NULL) {
258 FirstBond = BondList[i];
259 FirstOtherAtom = BondList[i]->GetOtherAtom(TopOrigin);
260 } else if (SecondBond == NULL) {
261 SecondBond = BondList[i];
262 SecondOtherAtom = BondList[i]->GetOtherAtom(TopOrigin);
263 } else {
264 *out << Verbose(3) << "WARNING: Detected more than four bonds for atom " << TopOrigin->Name;
265 }
266 }
267 }
268 if (SecondOtherAtom == NULL) { // then we have an atom with valence four, but only 3 bonds: one to replace and one which is TopBond (third is FirstBond)
269 SecondBond = TopBond;
270 SecondOtherAtom = TopReplacement;
271 }
272 if (FirstOtherAtom != NULL) { // then we just have this double bond and the plane does not matter at all
273// *out << Verbose(3) << "Regarding the double bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") to be constructed: Taking " << FirstOtherAtom->Name << " and " << SecondOtherAtom->Name << " along with " << TopOrigin->Name << " to determine orthogonal plane." << endl;
274
275 // determine the plane of these two with the *origin
276 AllWentWell = AllWentWell && Orthovector1.MakeNormalVector(&TopOrigin->x, &FirstOtherAtom->x, &SecondOtherAtom->x);
277 } else {
278 Orthovector1.GetOneNormalVector(&InBondvector);
279 }
280 //*out << Verbose(3)<< "Orthovector1: ";
281 //Orthovector1.Output(out);
282 //*out << endl;
283 // orthogonal vector and bond vector between origin and replacement form the new plane
284 Orthovector1.MakeNormalVector(&InBondvector);
285 Orthovector1.Normalize();
286 //*out << Verbose(3) << "ReScaleCheck: " << Orthovector1.Norm() << " and " << InBondvector.Norm() << "." << endl;
287
288 // create the two Hydrogens ...
289 FirstOtherAtom = new atom();
290 SecondOtherAtom = new atom();
291 FirstOtherAtom->type = elemente->FindElement(1);
292 SecondOtherAtom->type = elemente->FindElement(1);
293 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
294 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
295 SecondOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
296 SecondOtherAtom->FixedIon = TopReplacement->FixedIon;
297 FirstOtherAtom->father = NULL; // we are just an added hydrogen with no father
298 SecondOtherAtom->father = NULL; // we are just an added hydrogen with no father
299 bondangle = TopOrigin->type->HBondAngle[1];
300 if (bondangle == -1) {
301 *out << Verbose(3) << "WARNING: There is no typical bond angle for bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") of degree " << TopBond->BondDegree << "!" << endl;
302 bondangle = 0;
303 }
304 bondangle *= M_PI/180./2.;
305// *out << Verbose(3) << "ReScaleCheck: InBondvector ";
306// InBondvector.Output(out);
307// *out << endl;
308// *out << Verbose(3) << "ReScaleCheck: Orthovector ";
309// Orthovector1.Output(out);
310// *out << endl;
311// *out << Verbose(3) << "Half the bond angle is " << bondangle << ", sin and cos of it: " << sin(bondangle) << ", " << cos(bondangle) << endl;
312 FirstOtherAtom->x.Zero();
313 SecondOtherAtom->x.Zero();
314 for(int i=NDIM;i--;) { // rotate by half the bond angle in both directions (InBondvector is bondangle = 0 direction)
315 FirstOtherAtom->x.x[i] = InBondvector.x[i] * cos(bondangle) + Orthovector1.x[i] * (sin(bondangle));
316 SecondOtherAtom->x.x[i] = InBondvector.x[i] * cos(bondangle) + Orthovector1.x[i] * (-sin(bondangle));
317 }
318 FirstOtherAtom->x.Scale(&BondRescale); // rescale by correct BondDistance
319 SecondOtherAtom->x.Scale(&BondRescale);
320 //*out << Verbose(3) << "ReScaleCheck: " << FirstOtherAtom->x.Norm() << " and " << SecondOtherAtom->x.Norm() << "." << endl;
321 for(int i=NDIM;i--;) { // and make relative to origin atom
322 FirstOtherAtom->x.x[i] += TopOrigin->x.x[i];
323 SecondOtherAtom->x.x[i] += TopOrigin->x.x[i];
324 }
325 // ... and add to molecule
326 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
327 AllWentWell = AllWentWell && AddAtom(SecondOtherAtom);
328// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
329// FirstOtherAtom->x.Output(out);
330// *out << endl;
331// *out << Verbose(4) << "Added " << *SecondOtherAtom << " at: ";
332// SecondOtherAtom->x.Output(out);
333// *out << endl;
334 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
335 Binder->Cyclic = false;
336 Binder->Type = TreeEdge;
337 Binder = AddBond(BottomOrigin, SecondOtherAtom, 1);
338 Binder->Cyclic = false;
339 Binder->Type = TreeEdge;
340 break;
341 case 3:
342 // take the "usual" tetraoidal angle and add the three Hydrogen in direction of the bond (height of the tetraoid)
343 FirstOtherAtom = new atom();
344 SecondOtherAtom = new atom();
345 ThirdOtherAtom = new atom();
346 FirstOtherAtom->type = elemente->FindElement(1);
347 SecondOtherAtom->type = elemente->FindElement(1);
348 ThirdOtherAtom->type = elemente->FindElement(1);
349 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
350 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
351 SecondOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
352 SecondOtherAtom->FixedIon = TopReplacement->FixedIon;
353 ThirdOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
354 ThirdOtherAtom->FixedIon = TopReplacement->FixedIon;
355 FirstOtherAtom->father = NULL; // we are just an added hydrogen with no father
356 SecondOtherAtom->father = NULL; // we are just an added hydrogen with no father
357 ThirdOtherAtom->father = NULL; // we are just an added hydrogen with no father
358
359 // we need to vectors orthonormal the InBondvector
360 AllWentWell = AllWentWell && Orthovector1.GetOneNormalVector(&InBondvector);
361// *out << Verbose(3) << "Orthovector1: ";
362// Orthovector1.Output(out);
363// *out << endl;
364 AllWentWell = AllWentWell && Orthovector2.MakeNormalVector(&InBondvector, &Orthovector1);
365// *out << Verbose(3) << "Orthovector2: ";
366// Orthovector2.Output(out);
367// *out << endl;
368
369 // create correct coordination for the three atoms
370 alpha = (TopOrigin->type->HBondAngle[2])/180.*M_PI/2.; // retrieve triple bond angle from database
371 l = BondRescale; // desired bond length
372 b = 2.*l*sin(alpha); // base length of isosceles triangle
373 d = l*sqrt(cos(alpha)*cos(alpha) - sin(alpha)*sin(alpha)/3.); // length for InBondvector
374 f = b/sqrt(3.); // length for Orthvector1
375 g = b/2.; // length for Orthvector2
376// *out << Verbose(3) << "Bond length and half-angle: " << l << ", " << alpha << "\t (b,d,f,g) = " << b << ", " << d << ", " << f << ", " << g << ", " << endl;
377// *out << Verbose(3) << "The three Bond lengths: " << sqrt(d*d+f*f) << ", " << sqrt(d*d+(-0.5*f)*(-0.5*f)+g*g) << ", " << sqrt(d*d+(-0.5*f)*(-0.5*f)+g*g) << endl;
378 factors[0] = d;
379 factors[1] = f;
380 factors[2] = 0.;
381 FirstOtherAtom->x.LinearCombinationOfVectors(&InBondvector, &Orthovector1, &Orthovector2, factors);
382 factors[1] = -0.5*f;
383 factors[2] = g;
384 SecondOtherAtom->x.LinearCombinationOfVectors(&InBondvector, &Orthovector1, &Orthovector2, factors);
385 factors[2] = -g;
386 ThirdOtherAtom->x.LinearCombinationOfVectors(&InBondvector, &Orthovector1, &Orthovector2, factors);
387
388 // rescale each to correct BondDistance
389// FirstOtherAtom->x.Scale(&BondRescale);
390// SecondOtherAtom->x.Scale(&BondRescale);
391// ThirdOtherAtom->x.Scale(&BondRescale);
392
393 // and relative to *origin atom
394 FirstOtherAtom->x.AddVector(&TopOrigin->x);
395 SecondOtherAtom->x.AddVector(&TopOrigin->x);
396 ThirdOtherAtom->x.AddVector(&TopOrigin->x);
397
398 // ... and add to molecule
399 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
400 AllWentWell = AllWentWell && AddAtom(SecondOtherAtom);
401 AllWentWell = AllWentWell && AddAtom(ThirdOtherAtom);
402// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
403// FirstOtherAtom->x.Output(out);
404// *out << endl;
405// *out << Verbose(4) << "Added " << *SecondOtherAtom << " at: ";
406// SecondOtherAtom->x.Output(out);
407// *out << endl;
408// *out << Verbose(4) << "Added " << *ThirdOtherAtom << " at: ";
409// ThirdOtherAtom->x.Output(out);
410// *out << endl;
411 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
412 Binder->Cyclic = false;
413 Binder->Type = TreeEdge;
414 Binder = AddBond(BottomOrigin, SecondOtherAtom, 1);
415 Binder->Cyclic = false;
416 Binder->Type = TreeEdge;
417 Binder = AddBond(BottomOrigin, ThirdOtherAtom, 1);
418 Binder->Cyclic = false;
419 Binder->Type = TreeEdge;
420 break;
421 default:
422 cerr << "ERROR: BondDegree does not state single, double or triple bond!" << endl;
423 AllWentWell = false;
424 break;
425 }
426
427// *out << Verbose(3) << "End of AddHydrogenReplacementAtom." << endl;
428 return AllWentWell;
429};
430
431/** Adds given atom \a *pointer from molecule list.
432 * Increases molecule::last_atom and gives last number to added atom.
433 * \param filename name and path of xyz file
434 * \return true - succeeded, false - file not found
435 */
436bool molecule::AddXYZFile(string filename)
437{
438 istringstream *input = NULL;
439 int NumberOfAtoms = 0; // atom number in xyz read
440 int i, j; // loop variables
441 atom *Walker = NULL; // pointer to added atom
442 char shorthand[3]; // shorthand for atom name
443 ifstream xyzfile; // xyz file
444 string line; // currently parsed line
445 double x[3]; // atom coordinates
446
447 xyzfile.open(filename.c_str());
448 if (!xyzfile)
449 return false;
450
451 getline(xyzfile,line,'\n'); // Read numer of atoms in file
452 input = new istringstream(line);
453 *input >> NumberOfAtoms;
454 cout << Verbose(0) << "Parsing " << NumberOfAtoms << " atoms in file." << endl;
455 getline(xyzfile,line,'\n'); // Read comment
456 cout << Verbose(1) << "Comment: " << line << endl;
457
458 if (MDSteps == 0) // no atoms yet present
459 MDSteps++;
460 for(i=0;i<NumberOfAtoms;i++){
461 Walker = new atom;
462 getline(xyzfile,line,'\n');
463 istringstream *item = new istringstream(line);
464 //istringstream input(line);
465 //cout << Verbose(1) << "Reading: " << line << endl;
466 *item >> shorthand;
467 *item >> x[0];
468 *item >> x[1];
469 *item >> x[2];
470 Walker->type = elemente->FindElement(shorthand);
471 if (Walker->type == NULL) {
472 cerr << "Could not parse the element at line: '" << line << "', setting to H.";
473 Walker->type = elemente->FindElement(1);
474 }
475 if (Trajectories[Walker].R.size() <= (unsigned int)MDSteps) {
476 Trajectories[Walker].R.resize(MDSteps+10);
477 Trajectories[Walker].U.resize(MDSteps+10);
478 Trajectories[Walker].F.resize(MDSteps+10);
479 }
480 for(j=NDIM;j--;) {
481 Walker->x.x[j] = x[j];
482 Trajectories[Walker].R.at(MDSteps-1).x[j] = x[j];
483 Trajectories[Walker].U.at(MDSteps-1).x[j] = 0;
484 Trajectories[Walker].F.at(MDSteps-1).x[j] = 0;
485 }
486 AddAtom(Walker); // add to molecule
487 delete(item);
488 }
489 xyzfile.close();
490 delete(input);
491 return true;
492};
493
494/** Creates a copy of this molecule.
495 * \return copy of molecule
496 */
497molecule *molecule::CopyMolecule()
498{
499 molecule *copy = new molecule(elemente);
500 atom *CurrentAtom = NULL;
501 atom *LeftAtom = NULL, *RightAtom = NULL;
502 atom *Walker = NULL;
503
504 // copy all atoms
505 Walker = start;
506 while(Walker->next != end) {
507 Walker = Walker->next;
508 CurrentAtom = copy->AddCopyAtom(Walker);
509 }
510
511 // copy all bonds
512 bond *Binder = first;
513 bond *NewBond = NULL;
514 while(Binder->next != last) {
515 Binder = Binder->next;
516 // get the pendant atoms of current bond in the copy molecule
517 LeftAtom = copy->start;
518 while (LeftAtom->next != copy->end) {
519 LeftAtom = LeftAtom->next;
520 if (LeftAtom->father == Binder->leftatom)
521 break;
522 }
523 RightAtom = copy->start;
524 while (RightAtom->next != copy->end) {
525 RightAtom = RightAtom->next;
526 if (RightAtom->father == Binder->rightatom)
527 break;
528 }
529 NewBond = copy->AddBond(LeftAtom, RightAtom, Binder->BondDegree);
530 NewBond->Cyclic = Binder->Cyclic;
531 if (Binder->Cyclic)
532 copy->NoCyclicBonds++;
533 NewBond->Type = Binder->Type;
534 }
535 // correct fathers
536 Walker = copy->start;
537 while(Walker->next != copy->end) {
538 Walker = Walker->next;
539 if (Walker->father->father == Walker->father) // same atom in copy's father points to itself
540 Walker->father = Walker; // set father to itself (copy of a whole molecule)
541 else
542 Walker->father = Walker->father->father; // set father to original's father
543 }
544 // copy values
545 copy->CountAtoms((ofstream *)&cout);
546 copy->CountElements();
547 if (first->next != last) { // if adjaceny list is present
548 copy->BondDistance = BondDistance;
549 copy->CreateListOfBondsPerAtom((ofstream *)&cout);
550 }
551
552 return copy;
553};
554
555/** Adds a bond to a the molecule specified by two atoms, \a *first and \a *second.
556 * Also updates molecule::BondCount and molecule::NoNonBonds.
557 * \param *first first atom in bond
558 * \param *second atom in bond
559 * \return pointer to bond or NULL on failure
560 */
561bond * molecule::AddBond(atom *atom1, atom *atom2, int degree=1)
562{
563 bond *Binder = NULL;
564 if ((atom1 != NULL) && (FindAtom(atom1->nr) != NULL) && (atom2 != NULL) && (FindAtom(atom2->nr) != NULL)) {
565 Binder = new bond(atom1, atom2, degree, BondCount++);
566 if ((atom1->type != NULL) && (atom1->type->Z != 1) && (atom2->type != NULL) && (atom2->type->Z != 1))
567 NoNonBonds++;
568 add(Binder, last);
569 } else {
570 cerr << Verbose(1) << "ERROR: Could not add bond between " << atom1->Name << " and " << atom2->Name << " as one or both are not present in the molecule." << endl;
571 }
572 return Binder;
573};
574
575/** Remove bond from bond chain list.
576 * \todo Function not implemented yet
577 * \param *pointer bond pointer
578 * \return true - bound found and removed, false - bond not found/removed
579 */
580bool molecule::RemoveBond(bond *pointer)
581{
582 //cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
583 removewithoutcheck(pointer);
584 return true;
585};
586
587/** Remove every bond from bond chain list that atom \a *BondPartner is a constituent of.
588 * \todo Function not implemented yet
589 * \param *BondPartner atom to be removed
590 * \return true - bounds found and removed, false - bonds not found/removed
591 */
592bool molecule::RemoveBonds(atom *BondPartner)
593{
594 cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
595 return false;
596};
597
598/** Sets the molecule::cell_size to the components of \a *dim (rectangular box)
599 * \param *dim vector class
600 */
601void molecule::SetBoxDimension(Vector *dim)
602{
603 cell_size[0] = dim->x[0];
604 cell_size[1] = 0.;
605 cell_size[2] = dim->x[1];
606 cell_size[3] = 0.;
607 cell_size[4] = 0.;
608 cell_size[5] = dim->x[2];
609};
610
611/** Centers the molecule in the box whose lengths are defined by vector \a *BoxLengths.
612 * \param *out output stream for debugging
613 * \param *BoxLengths box lengths
614 */
615bool molecule::CenterInBox(ofstream *out, Vector *BoxLengths)
616{
617 bool status = true;
618 atom *ptr = NULL;
619 Vector *min = new Vector;
620 Vector *max = new Vector;
621
622 // gather min and max for each axis
623 ptr = start->next; // start at first in list
624 if (ptr != end) { //list not empty?
625 for (int i=NDIM;i--;) {
626 max->x[i] = ptr->x.x[i];
627 min->x[i] = ptr->x.x[i];
628 }
629 while (ptr->next != end) { // continue with second if present
630 ptr = ptr->next;
631 //ptr->Output(1,1,out);
632 for (int i=NDIM;i--;) {
633 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
634 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
635 }
636 }
637 }
638 // sanity check
639 for(int i=NDIM;i--;) {
640 if (max->x[i] - min->x[i] > BoxLengths->x[i])
641 status = false;
642 }
643 // warn if check failed
644 if (!status)
645 *out << "WARNING: molecule is bigger than defined box!" << endl;
646 else { // else center in box
647 max->AddVector(min);
648 max->Scale(-1.);
649 max->AddVector(BoxLengths);
650 max->Scale(0.5);
651 Translate(max);
652 }
653
654 // free and exit
655 delete(min);
656 delete(max);
657 return status;
658};
659
660/** Centers the edge of the atoms at (0,0,0).
661 * \param *out output stream for debugging
662 * \param *max coordinates of other edge, specifying box dimensions.
663 */
664void molecule::CenterEdge(ofstream *out, Vector *max)
665{
666 Vector *min = new Vector;
667
668// *out << Verbose(3) << "Begin of CenterEdge." << endl;
669 atom *ptr = start->next; // start at first in list
670 if (ptr != end) { //list not empty?
671 for (int i=NDIM;i--;) {
672 max->x[i] = ptr->x.x[i];
673 min->x[i] = ptr->x.x[i];
674 }
675 while (ptr->next != end) { // continue with second if present
676 ptr = ptr->next;
677 //ptr->Output(1,1,out);
678 for (int i=NDIM;i--;) {
679 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
680 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
681 }
682 }
683// *out << Verbose(4) << "Maximum is ";
684// max->Output(out);
685// *out << ", Minimum is ";
686// min->Output(out);
687// *out << endl;
688 min->Scale(-1.);
689 max->AddVector(min);
690 Translate(min);
691 }
692 delete(min);
693// *out << Verbose(3) << "End of CenterEdge." << endl;
694};
695
696/** Centers the center of the atoms at (0,0,0).
697 * \param *out output stream for debugging
698 * \param *center return vector for translation vector
699 */
700void molecule::CenterOrigin(ofstream *out, Vector *center)
701{
702 int Num = 0;
703 atom *ptr = start->next; // start at first in list
704
705 for(int i=NDIM;i--;) // zero center vector
706 center->x[i] = 0.;
707
708 if (ptr != end) { //list not empty?
709 while (ptr->next != end) { // continue with second if present
710 ptr = ptr->next;
711 Num++;
712 center->AddVector(&ptr->x);
713 }
714 center->Scale(-1./Num); // divide through total number (and sign for direction)
715 Translate(center);
716 }
717};
718
719/** Returns vector pointing to center of gravity.
720 * \param *out output stream for debugging
721 * \return pointer to center of gravity vector
722 */
723Vector * molecule::DetermineCenterOfAll(ofstream *out)
724{
725 atom *ptr = start->next; // start at first in list
726 Vector *a = new Vector();
727 Vector tmp;
728 double Num = 0;
729
730 a->Zero();
731
732 if (ptr != end) { //list not empty?
733 while (ptr->next != end) { // continue with second if present
734 ptr = ptr->next;
735 Num += 1.;
736 tmp.CopyVector(&ptr->x);
737 a->AddVector(&tmp);
738 }
739 a->Scale(-1./Num); // divide through total mass (and sign for direction)
740 }
741 //cout << Verbose(1) << "Resulting center of gravity: ";
742 //a->Output(out);
743 //cout << endl;
744 return a;
745};
746
747/** Returns vector pointing to center of gravity.
748 * \param *out output stream for debugging
749 * \return pointer to center of gravity vector
750 */
751Vector * molecule::DetermineCenterOfGravity(ofstream *out)
752{
753 atom *ptr = start->next; // start at first in list
754 Vector *a = new Vector();
755 Vector tmp;
756 double Num = 0;
757
758 a->Zero();
759
760 if (ptr != end) { //list not empty?
761 while (ptr->next != end) { // continue with second if present
762 ptr = ptr->next;
763 Num += ptr->type->mass;
764 tmp.CopyVector(&ptr->x);
765 tmp.Scale(ptr->type->mass); // scale by mass
766 a->AddVector(&tmp);
767 }
768 a->Scale(-1./Num); // divide through total mass (and sign for direction)
769 }
770// *out << Verbose(1) << "Resulting center of gravity: ";
771// a->Output(out);
772// *out << endl;
773 return a;
774};
775
776/** Centers the center of gravity of the atoms at (0,0,0).
777 * \param *out output stream for debugging
778 * \param *center return vector for translation vector
779 */
780void molecule::CenterGravity(ofstream *out, Vector *center)
781{
782 if (center == NULL) {
783 DetermineCenter(*center);
784 Translate(center);
785 delete(center);
786 } else {
787 Translate(center);
788 }
789};
790
791/** Scales all atoms by \a *factor.
792 * \param *factor pointer to scaling factor
793 */
794void molecule::Scale(double **factor)
795{
796 atom *ptr = start;
797
798 while (ptr->next != end) {
799 ptr = ptr->next;
800 for (int j=0;j<MDSteps;j++)
801 Trajectories[ptr].R.at(j).Scale(factor);
802 ptr->x.Scale(factor);
803 }
804};
805
806/** Translate all atoms by given vector.
807 * \param trans[] translation vector.
808 */
809void molecule::Translate(const Vector *trans)
810{
811 atom *ptr = start;
812
813 while (ptr->next != end) {
814 ptr = ptr->next;
815 for (int j=0;j<MDSteps;j++)
816 Trajectories[ptr].R.at(j).Translate(trans);
817 ptr->x.Translate(trans);
818 }
819};
820
821/** Mirrors all atoms against a given plane.
822 * \param n[] normal vector of mirror plane.
823 */
824void molecule::Mirror(const Vector *n)
825{
826 atom *ptr = start;
827
828 while (ptr->next != end) {
829 ptr = ptr->next;
830 for (int j=0;j<MDSteps;j++)
831 Trajectories[ptr].R.at(j).Mirror(n);
832 ptr->x.Mirror(n);
833 }
834};
835
836/** Determines center of molecule (yet not considering atom masses).
837 * \param Center reference to return vector
838 */
839void molecule::DetermineCenter(Vector &Center)
840{
841 atom *Walker = start;
842 bond *Binder = NULL;
843 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
844 double tmp;
845 bool flag;
846 Vector Testvector, Translationvector;
847
848 do {
849 Center.Zero();
850 flag = true;
851 while (Walker->next != end) {
852 Walker = Walker->next;
853#ifdef ADDHYDROGEN
854 if (Walker->type->Z != 1) {
855#endif
856 Testvector.CopyVector(&Walker->x);
857 Testvector.InverseMatrixMultiplication(matrix);
858 Translationvector.Zero();
859 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
860 Binder = ListOfBondsPerAtom[Walker->nr][i];
861 if (Walker->nr < Binder->GetOtherAtom(Walker)->nr) // otherwise we shift one to, the other fro and gain nothing
862 for (int j=0;j<NDIM;j++) {
863 tmp = Walker->x.x[j] - Binder->GetOtherAtom(Walker)->x.x[j];
864 if ((fabs(tmp)) > BondDistance) {
865 flag = false;
866 cout << Verbose(0) << "Hit: atom " << Walker->Name << " in bond " << *Binder << " has to be shifted due to " << tmp << "." << endl;
867 if (tmp > 0)
868 Translationvector.x[j] -= 1.;
869 else
870 Translationvector.x[j] += 1.;
871 }
872 }
873 }
874 Testvector.AddVector(&Translationvector);
875 Testvector.MatrixMultiplication(matrix);
876 Center.AddVector(&Testvector);
877 cout << Verbose(1) << "vector is: ";
878 Testvector.Output((ofstream *)&cout);
879 cout << endl;
880#ifdef ADDHYDROGEN
881 // now also change all hydrogens
882 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
883 Binder = ListOfBondsPerAtom[Walker->nr][i];
884 if (Binder->GetOtherAtom(Walker)->type->Z == 1) {
885 Testvector.CopyVector(&Binder->GetOtherAtom(Walker)->x);
886 Testvector.InverseMatrixMultiplication(matrix);
887 Testvector.AddVector(&Translationvector);
888 Testvector.MatrixMultiplication(matrix);
889 Center.AddVector(&Testvector);
890 cout << Verbose(1) << "Hydrogen vector is: ";
891 Testvector.Output((ofstream *)&cout);
892 cout << endl;
893 }
894 }
895 }
896#endif
897 }
898 } while (!flag);
899 Free((void **)&matrix, "molecule::DetermineCenter: *matrix");
900 Center.Scale(1./(double)AtomCount);
901};
902
903/** Transforms/Rotates the given molecule into its principal axis system.
904 * \param *out output stream for debugging
905 * \param DoRotate whether to rotate (true) or only to determine the PAS.
906 */
907void molecule::PrincipalAxisSystem(ofstream *out, bool DoRotate)
908{
909 atom *ptr = start; // start at first in list
910 double InertiaTensor[NDIM*NDIM];
911 Vector *CenterOfGravity = DetermineCenterOfGravity(out);
912
913 CenterGravity(out, CenterOfGravity);
914
915 // reset inertia tensor
916 for(int i=0;i<NDIM*NDIM;i++)
917 InertiaTensor[i] = 0.;
918
919 // sum up inertia tensor
920 while (ptr->next != end) {
921 ptr = ptr->next;
922 Vector x;
923 x.CopyVector(&ptr->x);
924 //x.SubtractVector(CenterOfGravity);
925 InertiaTensor[0] += ptr->type->mass*(x.x[1]*x.x[1] + x.x[2]*x.x[2]);
926 InertiaTensor[1] += ptr->type->mass*(-x.x[0]*x.x[1]);
927 InertiaTensor[2] += ptr->type->mass*(-x.x[0]*x.x[2]);
928 InertiaTensor[3] += ptr->type->mass*(-x.x[1]*x.x[0]);
929 InertiaTensor[4] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[2]*x.x[2]);
930 InertiaTensor[5] += ptr->type->mass*(-x.x[1]*x.x[2]);
931 InertiaTensor[6] += ptr->type->mass*(-x.x[2]*x.x[0]);
932 InertiaTensor[7] += ptr->type->mass*(-x.x[2]*x.x[1]);
933 InertiaTensor[8] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[1]*x.x[1]);
934 }
935 // print InertiaTensor for debugging
936 *out << "The inertia tensor is:" << endl;
937 for(int i=0;i<NDIM;i++) {
938 for(int j=0;j<NDIM;j++)
939 *out << InertiaTensor[i*NDIM+j] << " ";
940 *out << endl;
941 }
942 *out << endl;
943
944 // diagonalize to determine principal axis system
945 gsl_eigen_symmv_workspace *T = gsl_eigen_symmv_alloc(NDIM);
946 gsl_matrix_view m = gsl_matrix_view_array(InertiaTensor, NDIM, NDIM);
947 gsl_vector *eval = gsl_vector_alloc(NDIM);
948 gsl_matrix *evec = gsl_matrix_alloc(NDIM, NDIM);
949 gsl_eigen_symmv(&m.matrix, eval, evec, T);
950 gsl_eigen_symmv_free(T);
951 gsl_eigen_symmv_sort(eval, evec, GSL_EIGEN_SORT_ABS_DESC);
952
953 for(int i=0;i<NDIM;i++) {
954 *out << Verbose(1) << "eigenvalue = " << gsl_vector_get(eval, i);
955 *out << ", eigenvector = (" << evec->data[i * evec->tda + 0] << "," << evec->data[i * evec->tda + 1] << "," << evec->data[i * evec->tda + 2] << ")" << endl;
956 }
957
958 // check whether we rotate or not
959 if (DoRotate) {
960 *out << Verbose(1) << "Transforming molecule into PAS ... ";
961 // the eigenvectors specify the transformation matrix
962 ptr = start;
963 while (ptr->next != end) {
964 ptr = ptr->next;
965 for (int j=0;j<MDSteps;j++)
966 Trajectories[ptr].R.at(j).MatrixMultiplication(evec->data);
967 ptr->x.MatrixMultiplication(evec->data);
968 }
969 *out << "done." << endl;
970
971 // summing anew for debugging (resulting matrix has to be diagonal!)
972 // reset inertia tensor
973 for(int i=0;i<NDIM*NDIM;i++)
974 InertiaTensor[i] = 0.;
975
976 // sum up inertia tensor
977 ptr = start;
978 while (ptr->next != end) {
979 ptr = ptr->next;
980 Vector x;
981 x.CopyVector(&ptr->x);
982 //x.SubtractVector(CenterOfGravity);
983 InertiaTensor[0] += ptr->type->mass*(x.x[1]*x.x[1] + x.x[2]*x.x[2]);
984 InertiaTensor[1] += ptr->type->mass*(-x.x[0]*x.x[1]);
985 InertiaTensor[2] += ptr->type->mass*(-x.x[0]*x.x[2]);
986 InertiaTensor[3] += ptr->type->mass*(-x.x[1]*x.x[0]);
987 InertiaTensor[4] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[2]*x.x[2]);
988 InertiaTensor[5] += ptr->type->mass*(-x.x[1]*x.x[2]);
989 InertiaTensor[6] += ptr->type->mass*(-x.x[2]*x.x[0]);
990 InertiaTensor[7] += ptr->type->mass*(-x.x[2]*x.x[1]);
991 InertiaTensor[8] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[1]*x.x[1]);
992 }
993 // print InertiaTensor for debugging
994 *out << "The inertia tensor is:" << endl;
995 for(int i=0;i<NDIM;i++) {
996 for(int j=0;j<NDIM;j++)
997 *out << InertiaTensor[i*NDIM+j] << " ";
998 *out << endl;
999 }
1000 *out << endl;
1001 }
1002
1003 // free everything
1004 delete(CenterOfGravity);
1005 gsl_vector_free(eval);
1006 gsl_matrix_free(evec);
1007};
1008
1009/** Evaluates the potential energy used for constrained molecular dynamics.
1010 * \f$V_i^{con} = c^{bond} \cdot | r_{P(i)} - R_i | + sum_{i \neq j} C^{min} \cdot \frac{1}{C_{ij}} + C^{inj} \Bigl (1 - \theta \bigl (\prod_{i \neq j} (P(i) - P(j)) \bigr ) \Bigr )\f$
1011 * where the first term points to the target in minimum distance, the second is a penalty for trajectories lying too close to each other (\f$C_{ij}$ is minimum distance between
1012 * trajectories i and j) and the third term is a penalty for two atoms trying to each the same target point.
1013 * Note that for the second term we have to solve the following linear system:
1014 * \f$-c_1 \cdot n_1 + c_2 \cdot n_2 + C \cdot n_3 = - p_2 + p_1\f$, where \f$c_1\f$, \f$c_2\f$ and \f$C\f$ are constants,
1015 * offset vector \f$p_1\f$ in direction \f$n_1\f$, offset vector \f$p_2\f$ in direction \f$n_2\f$,
1016 * \f$n_3\f$ is the normal vector to both directions. \f$C\f$ would be the minimum distance between the two lines.
1017 * \sa molecule::MinimiseConstrainedPotential(), molecule::VerletForceIntegration()
1018 * \param *out output stream for debugging
1019 * \param *PermutationMap gives target ptr for each atom, array of size molecule::AtomCount (this is "x" in \f$V^{con}(x)\f$)
1020 * \param startstep start configuration (MDStep in molecule::trajectories)
1021 * \param endstep end configuration (MDStep in molecule::trajectories)
1022 * \param *constants constant in front of each term
1023 * \param IsAngstroem whether coordinates are in angstroem (true) or bohrradius (false)
1024 * \return potential energy
1025 * \note This routine is scaling quadratically which is not optimal.
1026 * \todo There's a bit double counting going on for the first time, bu nothing to worry really about.
1027 */
1028double molecule::ConstrainedPotential(ofstream *out, atom **PermutationMap, int startstep, int endstep, double *constants, bool IsAngstroem)
1029{
1030 double result = 0., tmp, Norm1, Norm2;
1031 atom *Walker = NULL, *Runner = NULL, *Sprinter = NULL;
1032 Vector trajectory1, trajectory2, normal, TestVector;
1033 gsl_matrix *A = gsl_matrix_alloc(NDIM,NDIM);
1034 gsl_vector *x = gsl_vector_alloc(NDIM);
1035
1036 // go through every atom
1037 Walker = start;
1038 while (Walker->next != end) {
1039 Walker = Walker->next;
1040 // first term: distance to target
1041 Runner = PermutationMap[Walker->nr]; // find target point
1042 tmp = (Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(endstep)));
1043 tmp *= IsAngstroem ? 1. : 1./AtomicLengthToAngstroem;
1044 result += constants[0] * tmp;
1045 //*out << Verbose(4) << "Adding " << tmp*constants[0] << "." << endl;
1046
1047 // second term: sum of distances to other trajectories
1048 Runner = start;
1049 while (Runner->next != end) {
1050 Runner = Runner->next;
1051 if (Runner == Walker) // hence, we only go up to the Walker, not beyond (similar to i=0; i<j; i++)
1052 break;
1053 // determine normalized trajectories direction vector (n1, n2)
1054 Sprinter = PermutationMap[Walker->nr]; // find first target point
1055 trajectory1.CopyVector(&Trajectories[Sprinter].R.at(endstep));
1056 trajectory1.SubtractVector(&Trajectories[Walker].R.at(startstep));
1057 trajectory1.Normalize();
1058 Norm1 = trajectory1.Norm();
1059 Sprinter = PermutationMap[Runner->nr]; // find second target point
1060 trajectory2.CopyVector(&Trajectories[Sprinter].R.at(endstep));
1061 trajectory2.SubtractVector(&Trajectories[Runner].R.at(startstep));
1062 trajectory2.Normalize();
1063 Norm2 = trajectory1.Norm();
1064 // check whether either is zero()
1065 if ((Norm1 < MYEPSILON) && (Norm2 < MYEPSILON)) {
1066 tmp = Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(startstep));
1067 } else if (Norm1 < MYEPSILON) {
1068 Sprinter = PermutationMap[Walker->nr]; // find first target point
1069 trajectory1.CopyVector(&Trajectories[Sprinter].R.at(endstep)); // copy first offset
1070 trajectory1.SubtractVector(&Trajectories[Runner].R.at(startstep)); // subtract second offset
1071 trajectory2.Scale( trajectory1.ScalarProduct(&trajectory2) ); // trajectory2 is scaled to unity, hence we don't need to divide by anything
1072 trajectory1.SubtractVector(&trajectory2); // project the part in norm direction away
1073 tmp = trajectory1.Norm(); // remaining norm is distance
1074 } else if (Norm2 < MYEPSILON) {
1075 Sprinter = PermutationMap[Runner->nr]; // find second target point
1076 trajectory2.CopyVector(&Trajectories[Sprinter].R.at(endstep)); // copy second offset
1077 trajectory2.SubtractVector(&Trajectories[Walker].R.at(startstep)); // subtract first offset
1078 trajectory1.Scale( trajectory2.ScalarProduct(&trajectory1) ); // trajectory1 is scaled to unity, hence we don't need to divide by anything
1079 trajectory2.SubtractVector(&trajectory1); // project the part in norm direction away
1080 tmp = trajectory2.Norm(); // remaining norm is distance
1081 } else if ((fabs(trajectory1.ScalarProduct(&trajectory2)/Norm1/Norm2) - 1.) < MYEPSILON) { // check whether they're linear dependent
1082// *out << Verbose(3) << "Both trajectories of " << *Walker << " and " << *Runner << " are linear dependent: ";
1083// *out << trajectory1;
1084// *out << " and ";
1085// *out << trajectory2;
1086 tmp = Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(startstep));
1087// *out << " with distance " << tmp << "." << endl;
1088 } else { // determine distance by finding minimum distance
1089// *out << Verbose(3) << "Both trajectories of " << *Walker << " and " << *Runner << " are linear independent ";
1090// *out << endl;
1091// *out << "First Trajectory: ";
1092// *out << trajectory1 << endl;
1093// *out << "Second Trajectory: ";
1094// *out << trajectory2 << endl;
1095 // determine normal vector for both
1096 normal.MakeNormalVector(&trajectory1, &trajectory2);
1097 // print all vectors for debugging
1098// *out << "Normal vector in between: ";
1099// *out << normal << endl;
1100 // setup matrix
1101 for (int i=NDIM;i--;) {
1102 gsl_matrix_set(A, 0, i, trajectory1.x[i]);
1103 gsl_matrix_set(A, 1, i, trajectory2.x[i]);
1104 gsl_matrix_set(A, 2, i, normal.x[i]);
1105 gsl_vector_set(x,i, (Trajectories[Walker].R.at(startstep).x[i] - Trajectories[Runner].R.at(startstep).x[i]));
1106 }
1107 // solve the linear system by Householder transformations
1108 gsl_linalg_HH_svx(A, x);
1109 // distance from last component
1110 tmp = gsl_vector_get(x,2);
1111// *out << " with distance " << tmp << "." << endl;
1112 // test whether we really have the intersection (by checking on c_1 and c_2)
1113 TestVector.CopyVector(&Trajectories[Runner].R.at(startstep));
1114 trajectory2.Scale(gsl_vector_get(x,1));
1115 TestVector.AddVector(&trajectory2);
1116 normal.Scale(gsl_vector_get(x,2));
1117 TestVector.AddVector(&normal);
1118 TestVector.SubtractVector(&Trajectories[Walker].R.at(startstep));
1119 trajectory1.Scale(gsl_vector_get(x,0));
1120 TestVector.SubtractVector(&trajectory1);
1121 if (TestVector.Norm() < MYEPSILON) {
1122// *out << Verbose(2) << "Test: ok.\tDistance of " << tmp << " is correct." << endl;
1123 } else {
1124// *out << Verbose(2) << "Test: failed.\tIntersection is off by ";
1125// *out << TestVector;
1126// *out << "." << endl;
1127 }
1128 }
1129 // add up
1130 tmp *= IsAngstroem ? 1. : 1./AtomicLengthToAngstroem;
1131 if (fabs(tmp) > MYEPSILON) {
1132 result += constants[1] * 1./tmp;
1133 //*out << Verbose(4) << "Adding " << 1./tmp*constants[1] << "." << endl;
1134 }
1135 }
1136
1137 // third term: penalty for equal targets
1138 Runner = start;
1139 while (Runner->next != end) {
1140 Runner = Runner->next;
1141 if ((PermutationMap[Walker->nr] == PermutationMap[Runner->nr]) && (Walker->nr < Runner->nr)) {
1142 Sprinter = PermutationMap[Walker->nr];
1143// *out << *Walker << " and " << *Runner << " are heading to the same target at ";
1144// *out << Trajectories[Sprinter].R.at(endstep);
1145// *out << ", penalting." << endl;
1146 result += constants[2];
1147 //*out << Verbose(4) << "Adding " << constants[2] << "." << endl;
1148 }
1149 }
1150 }
1151
1152 return result;
1153};
1154
1155void PrintPermutationMap(ofstream *out, atom **PermutationMap, int Nr)
1156{
1157 stringstream zeile1, zeile2;
1158 int *DoubleList = (int *) Malloc(Nr*sizeof(int), "PrintPermutationMap: *DoubleList");
1159 int doubles = 0;
1160 for (int i=0;i<Nr;i++)
1161 DoubleList[i] = 0;
1162 zeile1 << "PermutationMap: ";
1163 zeile2 << " ";
1164 for (int i=0;i<Nr;i++) {
1165 DoubleList[PermutationMap[i]->nr]++;
1166 zeile1 << i << " ";
1167 zeile2 << PermutationMap[i]->nr << " ";
1168 }
1169 for (int i=0;i<Nr;i++)
1170 if (DoubleList[i] > 1)
1171 doubles++;
1172 // *out << "Found " << doubles << " Doubles." << endl;
1173 Free((void **)&DoubleList, "PrintPermutationMap: *DoubleList");
1174// *out << zeile1.str() << endl << zeile2.str() << endl;
1175};
1176
1177/** Minimises the extra potential for constrained molecular dynamics and gives forces and the constrained potential energy.
1178 * We do the following:
1179 * -# Generate a distance list from all source to all target points
1180 * -# Sort this per source point
1181 * -# Take for each source point the target point with minimum distance, use this as initial permutation
1182 * -# check whether molecule::ConstrainedPotential() is greater than injective penalty
1183 * -# If so, we go through each source point, stepping down in the sorted target point distance list and re-checking potential.
1184 * -# Next, we only apply transformations that keep the injectivity of the permutations list.
1185 * -# Hence, for one source point we step down the ladder and seek the corresponding owner of this new target
1186 * point and try to change it for one with lesser distance, or for the next one with greater distance, but only
1187 * if this decreases the conditional potential.
1188 * -# finished.
1189 * -# Then, we calculate the forces by taking the spatial derivative, where we scale the potential to such a degree,
1190 * that the total force is always pointing in direction of the constraint force (ensuring that we move in the
1191 * right direction).
1192 * -# Finally, we calculate the potential energy and return.
1193 * \param *out output stream for debugging
1194 * \param **PermutationMap on return: mapping between the atom label of the initial and the final configuration
1195 * \param startstep current MD step giving initial position between which and \a endstep we perform the constrained MD (as further steps are always concatenated)
1196 * \param endstep step giving final position in constrained MD
1197 * \param IsAngstroem whether coordinates are in angstroem (true) or bohrradius (false)
1198 * \sa molecule::VerletForceIntegration()
1199 * \return potential energy (and allocated **PermutationMap (array of molecule::AtomCount ^2)
1200 * \todo The constrained potential's constants are set to fixed values right now, but they should scale based on checks of the system in order
1201 * to ensure they're properties (e.g. constants[2] always greater than the energy of the system).
1202 * \bug this all is not O(N log N) but O(N^2)
1203 */
1204double molecule::MinimiseConstrainedPotential(ofstream *out, atom **&PermutationMap, int startstep, int endstep, bool IsAngstroem)
1205{
1206 double Potential, OldPotential, OlderPotential;
1207 PermutationMap = (atom **) Malloc(AtomCount*sizeof(atom *), "molecule::MinimiseConstrainedPotential: **PermutationMap");
1208 DistanceMap **DistanceList = (DistanceMap **) Malloc(AtomCount*sizeof(DistanceMap *), "molecule::MinimiseConstrainedPotential: **DistanceList");
1209 DistanceMap::iterator *DistanceIterators = (DistanceMap::iterator *) Malloc(AtomCount*sizeof(DistanceMap::iterator), "molecule::MinimiseConstrainedPotential: *DistanceIterators");
1210 int *DoubleList = (int *) Malloc(AtomCount*sizeof(int), "molecule::MinimiseConstrainedPotential: *DoubleList");
1211 DistanceMap::iterator *StepList = (DistanceMap::iterator *) Malloc(AtomCount*sizeof(DistanceMap::iterator), "molecule::MinimiseConstrainedPotential: *StepList");
1212 double constants[3];
1213 int round;
1214 atom *Walker = NULL, *Runner = NULL, *Sprinter = NULL;
1215 DistanceMap::iterator Rider, Strider;
1216
1217 /// Minimise the potential
1218 // set Lagrange multiplier constants
1219 constants[0] = 10.;
1220 constants[1] = 1.;
1221 constants[2] = 1e+7; // just a huge penalty
1222 // generate the distance list
1223 *out << Verbose(1) << "Creating the distance list ... " << endl;
1224 for (int i=AtomCount; i--;) {
1225 DoubleList[i] = 0; // stores for how many atoms in startstep this atom is a target in endstep
1226 DistanceList[i] = new DistanceMap; // is the distance sorted target list per atom
1227 DistanceList[i]->clear();
1228 }
1229 *out << Verbose(1) << "Filling the distance list ... " << endl;
1230 Walker = start;
1231 while (Walker->next != end) {
1232 Walker = Walker->next;
1233 Runner = start;
1234 while(Runner->next != end) {
1235 Runner = Runner->next;
1236 DistanceList[Walker->nr]->insert( DistancePair(Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(endstep)), Runner) );
1237 }
1238 }
1239 // create the initial PermutationMap (source -> target)
1240 Walker = start;
1241 while (Walker->next != end) {
1242 Walker = Walker->next;
1243 StepList[Walker->nr] = DistanceList[Walker->nr]->begin(); // stores the step to the next iterator that could be a possible next target
1244 PermutationMap[Walker->nr] = DistanceList[Walker->nr]->begin()->second; // always pick target with the smallest distance
1245 DoubleList[DistanceList[Walker->nr]->begin()->second->nr]++; // increase this target's source count (>1? not injective)
1246 DistanceIterators[Walker->nr] = DistanceList[Walker->nr]->begin(); // and remember which one we picked
1247 *out << *Walker << " starts with distance " << DistanceList[Walker->nr]->begin()->first << "." << endl;
1248 }
1249 *out << Verbose(1) << "done." << endl;
1250 // make the PermutationMap injective by checking whether we have a non-zero constants[2] term in it
1251 *out << Verbose(1) << "Making the PermutationMap injective ... " << endl;
1252 Walker = start;
1253 DistanceMap::iterator NewBase;
1254 OldPotential = fabs(ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem));
1255 while ((OldPotential) > constants[2]) {
1256 PrintPermutationMap(out, PermutationMap, AtomCount);
1257 Walker = Walker->next;
1258 if (Walker == end) // round-robin at the end
1259 Walker = start->next;
1260 if (DoubleList[DistanceIterators[Walker->nr]->second->nr] <= 1) // no need to make those injective that aren't
1261 continue;
1262 // now, try finding a new one
1263 NewBase = DistanceIterators[Walker->nr]; // store old base
1264 do {
1265 NewBase++; // take next further distance in distance to targets list that's a target of no one
1266 } while ((DoubleList[NewBase->second->nr] != 0) && (NewBase != DistanceList[Walker->nr]->end()));
1267 if (NewBase != DistanceList[Walker->nr]->end()) {
1268 PermutationMap[Walker->nr] = NewBase->second;
1269 Potential = fabs(ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem));
1270 if (Potential > OldPotential) { // undo
1271 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second;
1272 } else { // do
1273 DoubleList[DistanceIterators[Walker->nr]->second->nr]--; // decrease the old entry in the doubles list
1274 DoubleList[NewBase->second->nr]++; // increase the old entry in the doubles list
1275 DistanceIterators[Walker->nr] = NewBase;
1276 OldPotential = Potential;
1277 *out << Verbose(3) << "Found a new permutation, new potential is " << OldPotential << "." << endl;
1278 }
1279 }
1280 }
1281 for (int i=AtomCount; i--;) // now each single entry in the DoubleList should be <=1
1282 if (DoubleList[i] > 1) {
1283 cerr << "Failed to create an injective PermutationMap!" << endl;
1284 exit(1);
1285 }
1286 *out << Verbose(1) << "done." << endl;
1287 Free((void **)&DoubleList, "molecule::MinimiseConstrainedPotential: *DoubleList");
1288 // argument minimise the constrained potential in this injective PermutationMap
1289 *out << Verbose(1) << "Argument minimising the PermutationMap, at current potential " << OldPotential << " ... " << endl;
1290 OldPotential = 1e+10;
1291 round = 0;
1292 do {
1293 *out << "Starting round " << ++round << " ... " << endl;
1294 OlderPotential = OldPotential;
1295 do {
1296 Walker = start;
1297 while (Walker->next != end) { // pick one
1298 Walker = Walker->next;
1299 PrintPermutationMap(out, PermutationMap, AtomCount);
1300 Sprinter = DistanceIterators[Walker->nr]->second; // store initial partner
1301 Strider = DistanceIterators[Walker->nr]; //remember old iterator
1302 DistanceIterators[Walker->nr] = StepList[Walker->nr];
1303 if (DistanceIterators[Walker->nr] == DistanceList[Walker->nr]->end()) {// stop, before we run through the list and still on
1304 DistanceIterators[Walker->nr] == DistanceList[Walker->nr]->begin();
1305 break;
1306 }
1307 //*out << Verbose(2) << "Current Walker: " << *Walker << " with old/next candidate " << *Sprinter << "/" << *DistanceIterators[Walker->nr]->second << "." << endl;
1308 // find source of the new target
1309 Runner = start->next;
1310 while(Runner != end) { // find the source whose toes we might be stepping on (Walker's new target should be in use by another already)
1311 if (PermutationMap[Runner->nr] == DistanceIterators[Walker->nr]->second) {
1312 //*out << Verbose(2) << "Found the corresponding owner " << *Runner << " to " << *PermutationMap[Runner->nr] << "." << endl;
1313 break;
1314 }
1315 Runner = Runner->next;
1316 }
1317 if (Runner != end) { // we found the other source
1318 // then look in its distance list for Sprinter
1319 Rider = DistanceList[Runner->nr]->begin();
1320 for (; Rider != DistanceList[Runner->nr]->end(); Rider++)
1321 if (Rider->second == Sprinter)
1322 break;
1323 if (Rider != DistanceList[Runner->nr]->end()) { // if we have found one
1324 //*out << Verbose(2) << "Current Other: " << *Runner << " with old/next candidate " << *PermutationMap[Runner->nr] << "/" << *Rider->second << "." << endl;
1325 // exchange both
1326 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second; // put next farther distance into PermutationMap
1327 PermutationMap[Runner->nr] = Sprinter; // and hand the old target to its respective owner
1328 PrintPermutationMap(out, PermutationMap, AtomCount);
1329 // calculate the new potential
1330 //*out << Verbose(2) << "Checking new potential ..." << endl;
1331 Potential = ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem);
1332 if (Potential > OldPotential) { // we made everything worse! Undo ...
1333 //*out << Verbose(3) << "Nay, made the potential worse: " << Potential << " vs. " << OldPotential << "!" << endl;
1334 //*out << Verbose(3) << "Setting " << *Runner << "'s source to " << *DistanceIterators[Runner->nr]->second << "." << endl;
1335 // Undo for Runner (note, we haven't moved the iteration yet, we may use this)
1336 PermutationMap[Runner->nr] = DistanceIterators[Runner->nr]->second;
1337 // Undo for Walker
1338 DistanceIterators[Walker->nr] = Strider; // take next farther distance target
1339 //*out << Verbose(3) << "Setting " << *Walker << "'s source to " << *DistanceIterators[Walker->nr]->second << "." << endl;
1340 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second;
1341 } else {
1342 DistanceIterators[Runner->nr] = Rider; // if successful also move the pointer in the iterator list
1343 *out << Verbose(3) << "Found a better permutation, new potential is " << Potential << " vs." << OldPotential << "." << endl;
1344 OldPotential = Potential;
1345 }
1346 if (Potential > constants[2]) {
1347 cerr << "ERROR: The two-step permutation procedure did not maintain injectivity!" << endl;
1348 exit(255);
1349 }
1350 //*out << endl;
1351 } else {
1352 cerr << "ERROR: " << *Runner << " was not the owner of " << *Sprinter << "!" << endl;
1353 exit(255);
1354 }
1355 } else {
1356 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second; // new target has no source!
1357 }
1358 StepList[Walker->nr]++; // take next farther distance target
1359 }
1360 } while (Walker->next != end);
1361 } while ((OlderPotential - OldPotential) > 1e-3);
1362 *out << Verbose(1) << "done." << endl;
1363
1364
1365 /// free memory and return with evaluated potential
1366 for (int i=AtomCount; i--;)
1367 DistanceList[i]->clear();
1368 Free((void **)&DistanceList, "molecule::MinimiseConstrainedPotential: **DistanceList");
1369 Free((void **)&DistanceIterators, "molecule::MinimiseConstrainedPotential: *DistanceIterators");
1370 return ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem);
1371};
1372
1373/** Evaluates the (distance-related part) of the constrained potential for the constrained forces.
1374 * \param *out output stream for debugging
1375 * \param startstep current MD step giving initial position between which and \a endstep we perform the constrained MD (as further steps are always concatenated)
1376 * \param endstep step giving final position in constrained MD
1377 * \param **PermutationMap mapping between the atom label of the initial and the final configuration
1378 * \param *Force ForceMatrix containing force vectors from the external energy functional minimisation.
1379 * \todo the constant for the constrained potential distance part is hard-coded independently of the hard-coded value in MinimiseConstrainedPotential()
1380 */
1381void molecule::EvaluateConstrainedForces(ofstream *out, int startstep, int endstep, atom **PermutationMap, ForceMatrix *Force)
1382{
1383 double constant = 10.;
1384 atom *Walker = NULL, *Sprinter = NULL;
1385
1386 /// evaluate forces (only the distance to target dependent part) with the final PermutationMap
1387 *out << Verbose(1) << "Calculating forces and adding onto ForceMatrix ... " << endl;
1388 Walker = start;
1389 while (Walker->next != NULL) {
1390 Walker = Walker->next;
1391 Sprinter = PermutationMap[Walker->nr];
1392 // set forces
1393 for (int i=NDIM;i++;)
1394 Force->Matrix[0][Walker->nr][5+i] += 2.*constant*sqrt(Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Sprinter].R.at(endstep)));
1395 }
1396 *out << Verbose(1) << "done." << endl;
1397};
1398
1399/** Performs a linear interpolation between two desired atomic configurations with a given number of steps.
1400 * Note, step number is config::MaxOuterStep
1401 * \param *out output stream for debugging
1402 * \param startstep stating initial configuration in molecule::Trajectories
1403 * \param endstep stating final configuration in molecule::Trajectories
1404 * \param &config configuration structure
1405 * \return true - success in writing step files, false - error writing files or only one step in molecule::Trajectories
1406 */
1407bool molecule::LinearInterpolationBetweenConfiguration(ofstream *out, int startstep, int endstep, const char *prefix, config &configuration)
1408{
1409 bool status = true;
1410 int MaxSteps = configuration.MaxOuterStep;
1411 MoleculeListClass *MoleculePerStep = new MoleculeListClass(MaxSteps+1, AtomCount);
1412 // Get the Permutation Map by MinimiseConstrainedPotential
1413 atom **PermutationMap = NULL;
1414 atom *Walker = NULL, *Sprinter = NULL;
1415 MinimiseConstrainedPotential(out, PermutationMap, startstep, endstep, configuration.GetIsAngstroem());
1416
1417 // check whether we have sufficient space in Trajectories for each atom
1418 Walker = start;
1419 while (Walker->next != end) {
1420 Walker = Walker->next;
1421 if (Trajectories[Walker].R.size() <= (unsigned int)(MaxSteps)) {
1422 //cout << "Increasing size for trajectory array of " << keyword << " to " << (MaxSteps+1) << "." << endl;
1423 Trajectories[Walker].R.resize(MaxSteps+1);
1424 Trajectories[Walker].U.resize(MaxSteps+1);
1425 Trajectories[Walker].F.resize(MaxSteps+1);
1426 }
1427 }
1428 // push endstep to last one
1429 Walker = start;
1430 while (Walker->next != end) { // remove the endstep (is now the very last one)
1431 Walker = Walker->next;
1432 for (int n=NDIM;n--;) {
1433 Trajectories[Walker].R.at(MaxSteps).x[n] = Trajectories[Walker].R.at(endstep).x[n];
1434 Trajectories[Walker].U.at(MaxSteps).x[n] = Trajectories[Walker].U.at(endstep).x[n];
1435 Trajectories[Walker].F.at(MaxSteps).x[n] = Trajectories[Walker].F.at(endstep).x[n];
1436 }
1437 }
1438 endstep = MaxSteps;
1439
1440 // go through all steps and add the molecular configuration to the list and to the Trajectories of \a this molecule
1441 *out << Verbose(1) << "Filling intermediate " << MaxSteps << " steps with MDSteps of " << MDSteps << "." << endl;
1442 for (int step = 0; step <= MaxSteps; step++) {
1443 MoleculePerStep->ListOfMolecules[step] = new molecule(elemente);
1444 Walker = start;
1445 while (Walker->next != end) {
1446 Walker = Walker->next;
1447 // add to molecule list
1448 Sprinter = MoleculePerStep->ListOfMolecules[step]->AddCopyAtom(Walker);
1449 for (int n=NDIM;n--;) {
1450 Sprinter->x.x[n] = Trajectories[Walker].R.at(startstep).x[n] + (Trajectories[PermutationMap[Walker->nr]].R.at(endstep).x[n] - Trajectories[Walker].R.at(startstep).x[n])*((double)step/(double)MaxSteps);
1451 // add to Trajectories
1452 //*out << Verbose(3) << step << ">=" << MDSteps-1 << endl;
1453 if (step < MaxSteps) {
1454 Trajectories[Walker].R.at(step).x[n] = Trajectories[Walker].R.at(startstep).x[n] + (Trajectories[PermutationMap[Walker->nr]].R.at(endstep).x[n] - Trajectories[Walker].R.at(startstep).x[n])*((double)step/(double)MaxSteps);
1455 Trajectories[Walker].U.at(step).x[n] = 0.;
1456 Trajectories[Walker].F.at(step).x[n] = 0.;
1457 }
1458 }
1459 }
1460 }
1461 MDSteps = MaxSteps+1; // otherwise new Trajectories' points aren't stored on save&exit
1462
1463 // store the list to single step files
1464 int *SortIndex = (int *) Malloc(AtomCount*sizeof(int), "molecule::LinearInterpolationBetweenConfiguration: *SortIndex");
1465 for (int i=AtomCount; i--; )
1466 SortIndex[i] = i;
1467 status = MoleculePerStep->OutputConfigForListOfFragments(out, "ConstrainedStep", &configuration, SortIndex, false, false);
1468
1469 // free and return
1470 Free((void **)&PermutationMap, "molecule::MinimiseConstrainedPotential: *PermutationMap");
1471 delete(MoleculePerStep);
1472 return status;
1473};
1474
1475/** Parses nuclear forces from file and performs Verlet integration.
1476 * Note that we assume the parsed forces to be in atomic units (hence, if coordinates are in angstroem, we
1477 * have to transform them).
1478 * This adds a new MD step to the config file.
1479 * \param *out output stream for debugging
1480 * \param *file filename
1481 * \param delta_t time step width in atomic units
1482 * \param IsAngstroem whether coordinates are in angstroem (true) or bohrradius (false)
1483 * \param DoConstrained whether we perform a constrained (>0, target step in molecule::trajectories) or unconstrained (0) molecular dynamics, \sa molecule::MinimiseConstrainedPotential()
1484 * \return true - file found and parsed, false - file not found or imparsable
1485 * \todo This is not yet checked if it is correctly working with DoConstrained set to true.
1486 */
1487bool molecule::VerletForceIntegration(ofstream *out, char *file, double delta_t, bool IsAngstroem, int DoConstrained)
1488{
1489 element *runner = elemente->start;
1490 atom *walker = NULL;
1491 int AtomNo;
1492 ifstream input(file);
1493 string token;
1494 stringstream item;
1495 double a, IonMass, Vector[NDIM], ConstrainedPotentialEnergy;
1496 ForceMatrix Force;
1497
1498 CountElements(); // make sure ElementsInMolecule is up to date
1499
1500 // check file
1501 if (input == NULL) {
1502 return false;
1503 } else {
1504 // parse file into ForceMatrix
1505 if (!Force.ParseMatrix(file, 0,0,0)) {
1506 cerr << "Could not parse Force Matrix file " << file << "." << endl;
1507 return false;
1508 }
1509 if (Force.RowCounter[0] != AtomCount) {
1510 cerr << "Mismatch between number of atoms in file " << Force.RowCounter[0] << " and in molecule " << AtomCount << "." << endl;
1511 return false;
1512 }
1513 // correct Forces
1514 for(int d=0;d<NDIM;d++)
1515 Vector[d] = 0.;
1516 for(int i=0;i<AtomCount;i++)
1517 for(int d=0;d<NDIM;d++) {
1518 Vector[d] += Force.Matrix[0][i][d+5];
1519 }
1520 for(int i=0;i<AtomCount;i++)
1521 for(int d=0;d<NDIM;d++) {
1522 Force.Matrix[0][i][d+5] -= Vector[d]/(double)AtomCount;
1523 }
1524 // solve a constrained potential if we are meant to
1525 if (DoConstrained) {
1526 // calculate forces and potential
1527 atom **PermutationMap = NULL;
1528 ConstrainedPotentialEnergy = MinimiseConstrainedPotential(out, PermutationMap, DoConstrained, 0, IsAngstroem);
1529 EvaluateConstrainedForces(out, DoConstrained, 0, PermutationMap, &Force);
1530 Free((void **)&PermutationMap, "molecule::MinimiseConstrainedPotential: *PermutationMap");
1531 }
1532
1533 // and perform Verlet integration for each atom with position, velocity and force vector
1534 runner = elemente->start;
1535 while (runner->next != elemente->end) { // go through every element
1536 runner = runner->next;
1537 IonMass = runner->mass;
1538 a = delta_t*0.5/IonMass; // (F+F_old)/2m = a and thus: v = (F+F_old)/2m * t = (F + F_old) * a
1539 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
1540 AtomNo = 0;
1541 walker = start;
1542 while (walker->next != end) { // go through every atom of this element
1543 walker = walker->next;
1544 if (walker->type == runner) { // if this atom fits to element
1545 // check size of vectors
1546 if (Trajectories[walker].R.size() <= (unsigned int)(MDSteps)) {
1547 //out << "Increasing size for trajectory array of " << *walker << " to " << (size+10) << "." << endl;
1548 Trajectories[walker].R.resize(MDSteps+10);
1549 Trajectories[walker].U.resize(MDSteps+10);
1550 Trajectories[walker].F.resize(MDSteps+10);
1551 }
1552
1553 // Update R (and F)
1554 for (int d=0; d<NDIM; d++) {
1555 Trajectories[walker].F.at(MDSteps).x[d] = -Force.Matrix[0][AtomNo][d+5]*(IsAngstroem ? AtomicLengthToAngstroem : 1.);
1556 Trajectories[walker].R.at(MDSteps).x[d] = Trajectories[walker].R.at(MDSteps-1).x[d];
1557 Trajectories[walker].R.at(MDSteps).x[d] += delta_t*(Trajectories[walker].U.at(MDSteps-1).x[d]);
1558 Trajectories[walker].R.at(MDSteps).x[d] += delta_t*a*(Trajectories[walker].F.at(MDSteps).x[d]); // F = m * a and s = 0.5 * F/m * t^2 = F * a * t
1559 }
1560 // Update U
1561 for (int d=0; d<NDIM; d++) {
1562 Trajectories[walker].U.at(MDSteps).x[d] = Trajectories[walker].U.at(MDSteps-1).x[d];
1563 Trajectories[walker].U.at(MDSteps).x[d] += a * (Trajectories[walker].F.at(MDSteps).x[d]+Trajectories[walker].F.at(MDSteps-1).x[d]);
1564 }
1565// out << "Integrated position&velocity of step " << (MDSteps) << ": (";
1566// for (int d=0;d<NDIM;d++)
1567// out << Trajectories[walker].R.at(MDSteps).x[d] << " "; // next step
1568// out << ")\t(";
1569// for (int d=0;d<NDIM;d++)
1570// cout << Trajectories[walker].U.at(MDSteps).x[d] << " "; // next step
1571// out << ")" << endl;
1572 // next atom
1573 AtomNo++;
1574 }
1575 }
1576 }
1577 }
1578 }
1579// // correct velocities (rather momenta) so that center of mass remains motionless
1580// for(int d=0;d<NDIM;d++)
1581// Vector[d] = 0.;
1582// IonMass = 0.;
1583// walker = start;
1584// while (walker->next != end) { // go through every atom
1585// walker = walker->next;
1586// IonMass += walker->type->mass; // sum up total mass
1587// for(int d=0;d<NDIM;d++) {
1588// Vector[d] += Trajectories[walker].U.at(MDSteps).x[d]*walker->type->mass;
1589// }
1590// }
1591// walker = start;
1592// while (walker->next != end) { // go through every atom of this element
1593// walker = walker->next;
1594// for(int d=0;d<NDIM;d++) {
1595// Trajectories[walker].U.at(MDSteps).x[d] -= Vector[d]*walker->type->mass/IonMass;
1596// }
1597// }
1598 MDSteps++;
1599
1600
1601 // exit
1602 return true;
1603};
1604
1605/** Align all atoms in such a manner that given vector \a *n is along z axis.
1606 * \param n[] alignment vector.
1607 */
1608void molecule::Align(Vector *n)
1609{
1610 atom *ptr = start;
1611 double alpha, tmp;
1612 Vector z_axis;
1613 z_axis.x[0] = 0.;
1614 z_axis.x[1] = 0.;
1615 z_axis.x[2] = 1.;
1616
1617 // rotate on z-x plane
1618 cout << Verbose(0) << "Begin of Aligning all atoms." << endl;
1619 alpha = atan(-n->x[0]/n->x[2]);
1620 cout << Verbose(1) << "Z-X-angle: " << alpha << " ... ";
1621 while (ptr->next != end) {
1622 ptr = ptr->next;
1623 tmp = ptr->x.x[0];
1624 ptr->x.x[0] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
1625 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
1626 for (int j=0;j<MDSteps;j++) {
1627 tmp = Trajectories[ptr].R.at(j).x[0];
1628 Trajectories[ptr].R.at(j).x[0] = cos(alpha) * tmp + sin(alpha) * Trajectories[ptr].R.at(j).x[2];
1629 Trajectories[ptr].R.at(j).x[2] = -sin(alpha) * tmp + cos(alpha) * Trajectories[ptr].R.at(j).x[2];
1630 }
1631 }
1632 // rotate n vector
1633 tmp = n->x[0];
1634 n->x[0] = cos(alpha) * tmp + sin(alpha) * n->x[2];
1635 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
1636 cout << Verbose(1) << "alignment vector after first rotation: ";
1637 n->Output((ofstream *)&cout);
1638 cout << endl;
1639
1640 // rotate on z-y plane
1641 ptr = start;
1642 alpha = atan(-n->x[1]/n->x[2]);
1643 cout << Verbose(1) << "Z-Y-angle: " << alpha << " ... ";
1644 while (ptr->next != end) {
1645 ptr = ptr->next;
1646 tmp = ptr->x.x[1];
1647 ptr->x.x[1] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
1648 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
1649 for (int j=0;j<MDSteps;j++) {
1650 tmp = Trajectories[ptr].R.at(j).x[1];
1651 Trajectories[ptr].R.at(j).x[1] = cos(alpha) * tmp + sin(alpha) * Trajectories[ptr].R.at(j).x[2];
1652 Trajectories[ptr].R.at(j).x[2] = -sin(alpha) * tmp + cos(alpha) * Trajectories[ptr].R.at(j).x[2];
1653 }
1654 }
1655 // rotate n vector (for consistency check)
1656 tmp = n->x[1];
1657 n->x[1] = cos(alpha) * tmp + sin(alpha) * n->x[2];
1658 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
1659
1660 cout << Verbose(1) << "alignment vector after second rotation: ";
1661 n->Output((ofstream *)&cout);
1662 cout << Verbose(1) << endl;
1663 cout << Verbose(0) << "End of Aligning all atoms." << endl;
1664};
1665
1666/** Removes atom from molecule list.
1667 * \param *pointer atom to be removed
1668 * \return true - succeeded, false - atom not found in list
1669 */
1670bool molecule::RemoveAtom(atom *pointer)
1671{
1672 if (ElementsInMolecule[pointer->type->Z] != 0) // this would indicate an error
1673 ElementsInMolecule[pointer->type->Z]--; // decrease number of atom of this element
1674 else
1675 cerr << "ERROR: Atom " << pointer->Name << " is of element " << pointer->type->Z << " but the entry in the table of the molecule is 0!" << endl;
1676 if (ElementsInMolecule[pointer->type->Z] == 0) // was last atom of this element?
1677 ElementCount--;
1678 Trajectories.erase(pointer);
1679 return remove(pointer, start, end);
1680};
1681
1682/** Removes every atom from molecule list.
1683 * \return true - succeeded, false - atom not found in list
1684 */
1685bool molecule::CleanupMolecule()
1686{
1687 return (cleanup(start,end) && cleanup(first,last));
1688};
1689
1690/** Finds an atom specified by its continuous number.
1691 * \param Nr number of atom withim molecule
1692 * \return pointer to atom or NULL
1693 */
1694atom * molecule::FindAtom(int Nr) const{
1695 atom * walker = find(&Nr, start,end);
1696 if (walker != NULL) {
1697 //cout << Verbose(0) << "Found Atom Nr. " << walker->nr << endl;
1698 return walker;
1699 } else {
1700 cout << Verbose(0) << "Atom not found in list." << endl;
1701 return NULL;
1702 }
1703};
1704
1705/** Asks for atom number, and checks whether in list.
1706 * \param *text question before entering
1707 */
1708atom * molecule::AskAtom(string text)
1709{
1710 int No;
1711 atom *ion = NULL;
1712 do {
1713 //cout << Verbose(0) << "============Atom list==========================" << endl;
1714 //mol->Output((ofstream *)&cout);
1715 //cout << Verbose(0) << "===============================================" << endl;
1716 cout << Verbose(0) << text;
1717 cin >> No;
1718 ion = this->FindAtom(No);
1719 } while (ion == NULL);
1720 return ion;
1721};
1722
1723/** Checks if given coordinates are within cell volume.
1724 * \param *x array of coordinates
1725 * \return true - is within, false - out of cell
1726 */
1727bool molecule::CheckBounds(const Vector *x) const
1728{
1729 bool result = true;
1730 int j =-1;
1731 for (int i=0;i<NDIM;i++) {
1732 j += i+1;
1733 result = result && ((x->x[i] >= 0) && (x->x[i] < cell_size[j]));
1734 }
1735 //return result;
1736 return true; /// probably not gonna use the check no more
1737};
1738
1739/** Calculates sum over least square distance to line hidden in \a *x.
1740 * \param *x offset and direction vector
1741 * \param *params pointer to lsq_params structure
1742 * \return \f$ sum_i^N | y_i - (a + t_i b)|^2\f$
1743 */
1744double LeastSquareDistance (const gsl_vector * x, void * params)
1745{
1746 double res = 0, t;
1747 Vector a,b,c,d;
1748 struct lsq_params *par = (struct lsq_params *)params;
1749 atom *ptr = par->mol->start;
1750
1751 // initialize vectors
1752 a.x[0] = gsl_vector_get(x,0);
1753 a.x[1] = gsl_vector_get(x,1);
1754 a.x[2] = gsl_vector_get(x,2);
1755 b.x[0] = gsl_vector_get(x,3);
1756 b.x[1] = gsl_vector_get(x,4);
1757 b.x[2] = gsl_vector_get(x,5);
1758 // go through all atoms
1759 while (ptr != par->mol->end) {
1760 ptr = ptr->next;
1761 if (ptr->type == ((struct lsq_params *)params)->type) { // for specific type
1762 c.CopyVector(&ptr->x); // copy vector to temporary one
1763 c.SubtractVector(&a); // subtract offset vector
1764 t = c.ScalarProduct(&b); // get direction parameter
1765 d.CopyVector(&b); // and create vector
1766 d.Scale(&t);
1767 c.SubtractVector(&d); // ... yielding distance vector
1768 res += d.ScalarProduct((const Vector *)&d); // add squared distance
1769 }
1770 }
1771 return res;
1772};
1773
1774/** By minimizing the least square distance gains alignment vector.
1775 * \bug this is not yet working properly it seems
1776 */
1777void molecule::GetAlignvector(struct lsq_params * par) const
1778{
1779 int np = 6;
1780
1781 const gsl_multimin_fminimizer_type *T =
1782 gsl_multimin_fminimizer_nmsimplex;
1783 gsl_multimin_fminimizer *s = NULL;
1784 gsl_vector *ss;
1785 gsl_multimin_function minex_func;
1786
1787 size_t iter = 0, i;
1788 int status;
1789 double size;
1790
1791 /* Initial vertex size vector */
1792 ss = gsl_vector_alloc (np);
1793
1794 /* Set all step sizes to 1 */
1795 gsl_vector_set_all (ss, 1.0);
1796
1797 /* Starting point */
1798 par->x = gsl_vector_alloc (np);
1799 par->mol = this;
1800
1801 gsl_vector_set (par->x, 0, 0.0); // offset
1802 gsl_vector_set (par->x, 1, 0.0);
1803 gsl_vector_set (par->x, 2, 0.0);
1804 gsl_vector_set (par->x, 3, 0.0); // direction
1805 gsl_vector_set (par->x, 4, 0.0);
1806 gsl_vector_set (par->x, 5, 1.0);
1807
1808 /* Initialize method and iterate */
1809 minex_func.f = &LeastSquareDistance;
1810 minex_func.n = np;
1811 minex_func.params = (void *)par;
1812
1813 s = gsl_multimin_fminimizer_alloc (T, np);
1814 gsl_multimin_fminimizer_set (s, &minex_func, par->x, ss);
1815
1816 do
1817 {
1818 iter++;
1819 status = gsl_multimin_fminimizer_iterate(s);
1820
1821 if (status)
1822 break;
1823
1824 size = gsl_multimin_fminimizer_size (s);
1825 status = gsl_multimin_test_size (size, 1e-2);
1826
1827 if (status == GSL_SUCCESS)
1828 {
1829 printf ("converged to minimum at\n");
1830 }
1831
1832 printf ("%5d ", (int)iter);
1833 for (i = 0; i < (size_t)np; i++)
1834 {
1835 printf ("%10.3e ", gsl_vector_get (s->x, i));
1836 }
1837 printf ("f() = %7.3f size = %.3f\n", s->fval, size);
1838 }
1839 while (status == GSL_CONTINUE && iter < 100);
1840
1841 for (i=0;i<(size_t)np;i++)
1842 gsl_vector_set(par->x, i, gsl_vector_get(s->x, i));
1843 //gsl_vector_free(par->x);
1844 gsl_vector_free(ss);
1845 gsl_multimin_fminimizer_free (s);
1846};
1847
1848/** Prints molecule to *out.
1849 * \param *out output stream
1850 */
1851bool molecule::Output(ofstream *out)
1852{
1853 element *runner;
1854 atom *walker = NULL;
1855 int ElementNo, AtomNo;
1856 CountElements();
1857
1858 if (out == NULL) {
1859 return false;
1860 } else {
1861 *out << "#Ion_TypeNr._Nr.R[0] R[1] R[2] MoveType (0 MoveIon, 1 FixedIon)" << endl;
1862 ElementNo = 0;
1863 runner = elemente->start;
1864 while (runner->next != elemente->end) { // go through every element
1865 runner = runner->next;
1866 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
1867 ElementNo++;
1868 AtomNo = 0;
1869 walker = start;
1870 while (walker->next != end) { // go through every atom of this element
1871 walker = walker->next;
1872 if (walker->type == runner) { // if this atom fits to element
1873 AtomNo++;
1874 walker->Output(ElementNo, AtomNo, out); // removed due to trajectories
1875 }
1876 }
1877 }
1878 }
1879 return true;
1880 }
1881};
1882
1883/** Prints molecule with all atomic trajectory positions to *out.
1884 * \param *out output stream
1885 */
1886bool molecule::OutputTrajectories(ofstream *out)
1887{
1888 element *runner = NULL;
1889 atom *walker = NULL;
1890 int ElementNo, AtomNo;
1891 CountElements();
1892
1893 if (out == NULL) {
1894 return false;
1895 } else {
1896 for (int step = 0; step < MDSteps; step++) {
1897 if (step == 0) {
1898 *out << "#Ion_TypeNr._Nr.R[0] R[1] R[2] MoveType (0 MoveIon, 1 FixedIon)" << endl;
1899 } else {
1900 *out << "# ====== MD step " << step << " =========" << endl;
1901 }
1902 ElementNo = 0;
1903 runner = elemente->start;
1904 while (runner->next != elemente->end) { // go through every element
1905 runner = runner->next;
1906 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
1907 ElementNo++;
1908 AtomNo = 0;
1909 walker = start;
1910 while (walker->next != end) { // go through every atom of this element
1911 walker = walker->next;
1912 if (walker->type == runner) { // if this atom fits to element
1913 AtomNo++;
1914 *out << "Ion_Type" << ElementNo << "_" << AtomNo << "\t" << fixed << setprecision(9) << showpoint;
1915 *out << Trajectories[walker].R.at(step).x[0] << "\t" << Trajectories[walker].R.at(step).x[1] << "\t" << Trajectories[walker].R.at(step).x[2];
1916 *out << "\t" << walker->FixedIon;
1917 if (Trajectories[walker].U.at(step).Norm() > MYEPSILON)
1918 *out << "\t" << scientific << setprecision(6) << Trajectories[walker].U.at(step).x[0] << "\t" << Trajectories[walker].U.at(step).x[1] << "\t" << Trajectories[walker].U.at(step).x[2] << "\t";
1919 if (Trajectories[walker].F.at(step).Norm() > MYEPSILON)
1920 *out << "\t" << scientific << setprecision(6) << Trajectories[walker].F.at(step).x[0] << "\t" << Trajectories[walker].F.at(step).x[1] << "\t" << Trajectories[walker].F.at(step).x[2] << "\t";
1921 *out << "\t# Number in molecule " << walker->nr << endl;
1922 }
1923 }
1924 }
1925 }
1926 }
1927 return true;
1928 }
1929};
1930
1931/** Outputs contents of molecule::ListOfBondsPerAtom.
1932 * \param *out output stream
1933 */
1934void molecule::OutputListOfBonds(ofstream *out) const
1935{
1936 *out << Verbose(2) << endl << "From Contents of ListOfBondsPerAtom, all non-hydrogen atoms:" << endl;
1937 atom *Walker = start;
1938 while (Walker->next != end) {
1939 Walker = Walker->next;
1940#ifdef ADDHYDROGEN
1941 if (Walker->type->Z != 1) { // regard only non-hydrogen
1942#endif
1943 *out << Verbose(2) << "Atom " << Walker->Name << " has Bonds: "<<endl;
1944 for(int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
1945 *out << Verbose(3) << *(ListOfBondsPerAtom)[Walker->nr][j] << endl;
1946 }
1947#ifdef ADDHYDROGEN
1948 }
1949#endif
1950 }
1951 *out << endl;
1952};
1953
1954/** Output of element before the actual coordination list.
1955 * \param *out stream pointer
1956 */
1957bool molecule::Checkout(ofstream *out) const
1958{
1959 return elemente->Checkout(out, ElementsInMolecule);
1960};
1961
1962/** Prints molecule with all its trajectories to *out as xyz file.
1963 * \param *out output stream
1964 */
1965bool molecule::OutputTrajectoriesXYZ(ofstream *out)
1966{
1967 atom *walker = NULL;
1968 int No = 0;
1969 time_t now;
1970
1971 now = time((time_t *)NULL); // Get the system time and put it into 'now' as 'calender time'
1972 walker = start;
1973 while (walker->next != end) { // go through every atom and count
1974 walker = walker->next;
1975 No++;
1976 }
1977 if (out != NULL) {
1978 for (int step=0;step<MDSteps;step++) {
1979 *out << No << "\n\tCreated by molecuilder, step " << step << ", on " << ctime(&now);
1980 walker = start;
1981 while (walker->next != end) { // go through every atom of this element
1982 walker = walker->next;
1983 *out << walker->type->symbol << "\t" << Trajectories[walker].R.at(step).x[0] << "\t" << Trajectories[walker].R.at(step).x[1] << "\t" << Trajectories[walker].R.at(step).x[2] << endl;
1984 }
1985 }
1986 return true;
1987 } else
1988 return false;
1989};
1990
1991/** Prints molecule to *out as xyz file.
1992* \param *out output stream
1993 */
1994bool molecule::OutputXYZ(ofstream *out) const
1995{
1996 atom *walker = NULL;
1997 int No = 0;
1998 time_t now;
1999
2000 now = time((time_t *)NULL); // Get the system time and put it into 'now' as 'calender time'
2001 walker = start;
2002 while (walker->next != end) { // go through every atom and count
2003 walker = walker->next;
2004 No++;
2005 }
2006 if (out != NULL) {
2007 *out << No << "\n\tCreated by molecuilder on " << ctime(&now);
2008 walker = start;
2009 while (walker->next != end) { // go through every atom of this element
2010 walker = walker->next;
2011 walker->OutputXYZLine(out);
2012 }
2013 return true;
2014 } else
2015 return false;
2016};
2017
2018/** Brings molecule::AtomCount and atom::*Name up-to-date.
2019 * \param *out output stream for debugging
2020 */
2021void molecule::CountAtoms(ofstream *out)
2022{
2023 int i = 0;
2024 atom *Walker = start;
2025 while (Walker->next != end) {
2026 Walker = Walker->next;
2027 i++;
2028 }
2029 if ((AtomCount == 0) || (i != AtomCount)) {
2030 *out << Verbose(3) << "Mismatch in AtomCount " << AtomCount << " and recounted number " << i << ", renaming all." << endl;
2031 AtomCount = i;
2032
2033 // count NonHydrogen atoms and give each atom a unique name
2034 if (AtomCount != 0) {
2035 i=0;
2036 NoNonHydrogen = 0;
2037 Walker = start;
2038 while (Walker->next != end) {
2039 Walker = Walker->next;
2040 Walker->nr = i; // update number in molecule (for easier referencing in FragmentMolecule lateron)
2041 if (Walker->type->Z != 1) // count non-hydrogen atoms whilst at it
2042 NoNonHydrogen++;
2043 Free((void **)&Walker->Name, "molecule::CountAtoms: *walker->Name");
2044 Walker->Name = (char *) Malloc(sizeof(char)*6, "molecule::CountAtoms: *walker->Name");
2045 sprintf(Walker->Name, "%2s%02d", Walker->type->symbol, Walker->nr+1);
2046 *out << "Naming atom nr. " << Walker->nr << " " << Walker->Name << "." << endl;
2047 i++;
2048 }
2049 } else
2050 *out << Verbose(3) << "AtomCount is still " << AtomCount << ", thus counting nothing." << endl;
2051 }
2052};
2053
2054/** Brings molecule::ElementCount and molecule::ElementsInMolecule up-to-date.
2055 */
2056void molecule::CountElements()
2057{
2058 int i = 0;
2059 for(i=MAX_ELEMENTS;i--;)
2060 ElementsInMolecule[i] = 0;
2061 ElementCount = 0;
2062
2063 atom *walker = start;
2064 while (walker->next != end) {
2065 walker = walker->next;
2066 ElementsInMolecule[walker->type->Z]++;
2067 i++;
2068 }
2069 for(i=MAX_ELEMENTS;i--;)
2070 ElementCount += (ElementsInMolecule[i] != 0 ? 1 : 0);
2071};
2072
2073/** Counts all cyclic bonds and returns their number.
2074 * \note Hydrogen bonds can never by cyclic, thus no check for that
2075 * \param *out output stream for debugging
2076 * \return number opf cyclic bonds
2077 */
2078int molecule::CountCyclicBonds(ofstream *out)
2079{
2080 int No = 0;
2081 int *MinimumRingSize = NULL;
2082 MoleculeLeafClass *Subgraphs = NULL;
2083 class StackClass<bond *> *BackEdgeStack = NULL;
2084 bond *Binder = first;
2085 if ((Binder->next != last) && (Binder->next->Type == Undetermined)) {
2086 *out << Verbose(0) << "No Depth-First-Search analysis performed so far, calling ..." << endl;
2087 Subgraphs = DepthFirstSearchAnalysis(out, BackEdgeStack);
2088 while (Subgraphs->next != NULL) {
2089 Subgraphs = Subgraphs->next;
2090 delete(Subgraphs->previous);
2091 }
2092 delete(Subgraphs);
2093 delete[](MinimumRingSize);
2094 }
2095 while(Binder->next != last) {
2096 Binder = Binder->next;
2097 if (Binder->Cyclic)
2098 No++;
2099 }
2100 delete(BackEdgeStack);
2101 return No;
2102};
2103/** Returns Shading as a char string.
2104 * \param color the Shading
2105 * \return string of the flag
2106 */
2107string molecule::GetColor(enum Shading color)
2108{
2109 switch(color) {
2110 case white:
2111 return "white";
2112 break;
2113 case lightgray:
2114 return "lightgray";
2115 break;
2116 case darkgray:
2117 return "darkgray";
2118 break;
2119 case black:
2120 return "black";
2121 break;
2122 default:
2123 return "uncolored";
2124 break;
2125 };
2126};
2127
2128
2129/** Counts necessary number of valence electrons and returns number and SpinType.
2130 * \param configuration containing everything
2131 */
2132void molecule::CalculateOrbitals(class config &configuration)
2133{
2134 configuration.MaxPsiDouble = configuration.PsiMaxNoDown = configuration.PsiMaxNoUp = configuration.PsiType = 0;
2135 for(int i=MAX_ELEMENTS;i--;) {
2136 if (ElementsInMolecule[i] != 0) {
2137 //cout << "CalculateOrbitals: " << elemente->FindElement(i)->name << " has a valence of " << (int)elemente->FindElement(i)->Valence << " and there are " << ElementsInMolecule[i] << " of it." << endl;
2138 configuration.MaxPsiDouble += ElementsInMolecule[i]*((int)elemente->FindElement(i)->Valence);
2139 }
2140 }
2141 configuration.PsiMaxNoDown = configuration.MaxPsiDouble/2 + (configuration.MaxPsiDouble % 2);
2142 configuration.PsiMaxNoUp = configuration.MaxPsiDouble/2;
2143 configuration.MaxPsiDouble /= 2;
2144 configuration.PsiType = (configuration.PsiMaxNoDown == configuration.PsiMaxNoUp) ? 0 : 1;
2145 if ((configuration.PsiType == 1) && (configuration.ProcPEPsi < 2)) {
2146 configuration.ProcPEGamma /= 2;
2147 configuration.ProcPEPsi *= 2;
2148 } else {
2149 configuration.ProcPEGamma *= configuration.ProcPEPsi;
2150 configuration.ProcPEPsi = 1;
2151 }
2152 configuration.InitMaxMinStopStep = configuration.MaxMinStopStep = configuration.MaxPsiDouble;
2153};
2154
2155/** Creates an adjacency list of the molecule.
2156 * Generally, we use the CSD approach to bond recognition, that is the the distance
2157 * between two atoms A and B must be within [Rcov(A)+Rcov(B)-t,Rcov(A)+Rcov(B)+t] with
2158 * a threshold t = 0.4 Angstroem.
2159 * To make it O(N log N) the function uses the linked-cell technique as follows:
2160 * The procedure is step-wise:
2161 * -# Remove every bond in list
2162 * -# Count the atoms in the molecule with CountAtoms()
2163 * -# partition cell into smaller linked cells of size \a bonddistance
2164 * -# put each atom into its corresponding cell
2165 * -# go through every cell, check the atoms therein against all possible bond partners in the 27 adjacent cells, add bond if true
2166 * -# create the list of bonds via CreateListOfBondsPerAtom()
2167 * -# correct the bond degree iteratively (single->double->triple bond)
2168 * -# finally print the bond list to \a *out if desired
2169 * \param *out out stream for printing the matrix, NULL if no output
2170 * \param bonddistance length of linked cells (i.e. maximum minimal length checked)
2171 * \param IsAngstroem whether coordinate system is gauged to Angstroem or Bohr radii
2172 */
2173void molecule::CreateAdjacencyList(ofstream *out, double bonddistance, bool IsAngstroem)
2174{
2175 atom *Walker = NULL, *OtherWalker = NULL, *Candidate = NULL;
2176 int No, NoBonds, CandidateBondNo;
2177 int NumberCells, divisor[NDIM], n[NDIM], N[NDIM], index, Index, j;
2178 molecule **CellList;
2179 double distance, MinDistance, MaxDistance;
2180 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
2181 Vector x;
2182 int FalseBondDegree = 0;
2183
2184 BondDistance = bonddistance; // * ((IsAngstroem) ? 1. : 1./AtomicLengthToAngstroem);
2185 *out << Verbose(0) << "Begin of CreateAdjacencyList." << endl;
2186 // remove every bond from the list
2187 if ((first->next != last) && (last->previous != first)) { // there are bonds present
2188 cleanup(first,last);
2189 }
2190
2191 // count atoms in molecule = dimension of matrix (also give each unique name and continuous numbering)
2192 CountAtoms(out);
2193 *out << Verbose(1) << "AtomCount " << AtomCount << "." << endl;
2194
2195 if (AtomCount != 0) {
2196 // 1. find divisor for each axis, such that a sphere with radius of at least bonddistance can be placed into each cell
2197 j=-1;
2198 for (int i=0;i<NDIM;i++) {
2199 j += i+1;
2200 divisor[i] = (int)floor(cell_size[j]/bonddistance); // take smaller value such that size of linked cell is at least bonddistance
2201 //*out << Verbose(1) << "divisor[" << i << "] = " << divisor[i] << "." << endl;
2202 }
2203 // 2a. allocate memory for the cell list
2204 NumberCells = divisor[0]*divisor[1]*divisor[2];
2205 *out << Verbose(1) << "Allocating " << NumberCells << " cells." << endl;
2206 CellList = (molecule **) Malloc(sizeof(molecule *)*NumberCells, "molecule::CreateAdjacencyList - ** CellList");
2207 for (int i=NumberCells;i--;)
2208 CellList[i] = NULL;
2209
2210 // 2b. put all atoms into its corresponding list
2211 Walker = start;
2212 while(Walker->next != end) {
2213 Walker = Walker->next;
2214 //*out << Verbose(1) << "Current atom is " << *Walker << " with coordinates ";
2215 //Walker->x.Output(out);
2216 //*out << "." << endl;
2217 // compute the cell by the atom's coordinates
2218 j=-1;
2219 for (int i=0;i<NDIM;i++) {
2220 j += i+1;
2221 x.CopyVector(&(Walker->x));
2222 x.KeepPeriodic(out, matrix);
2223 n[i] = (int)floor(x.x[i]/cell_size[j]*(double)divisor[i]);
2224 }
2225 index = n[2] + (n[1] + n[0] * divisor[1]) * divisor[2];
2226 //*out << Verbose(1) << "Atom " << *Walker << " goes into cell number [" << n[0] << "," << n[1] << "," << n[2] << "] = " << index << "." << endl;
2227 // add copy atom to this cell
2228 if (CellList[index] == NULL) // allocate molecule if not done
2229 CellList[index] = new molecule(elemente);
2230 OtherWalker = CellList[index]->AddCopyAtom(Walker); // add a copy of walker to this atom, father will be walker for later reference
2231 //*out << Verbose(1) << "Copy Atom is " << *OtherWalker << "." << endl;
2232 }
2233 //for (int i=0;i<NumberCells;i++)
2234 //*out << Verbose(1) << "Cell number " << i << ": " << CellList[i] << "." << endl;
2235
2236 // 3a. go through every cell
2237 for (N[0]=divisor[0];N[0]--;)
2238 for (N[1]=divisor[1];N[1]--;)
2239 for (N[2]=divisor[2];N[2]--;) {
2240 Index = N[2] + (N[1] + N[0] * divisor[1]) * divisor[2];
2241 if (CellList[Index] != NULL) { // if there atoms in this cell
2242 //*out << Verbose(1) << "Current cell is " << Index << "." << endl;
2243 // 3b. for every atom therein
2244 Walker = CellList[Index]->start;
2245 while (Walker->next != CellList[Index]->end) { // go through every atom
2246 Walker = Walker->next;
2247 //*out << Verbose(0) << "Current Atom is " << *Walker << "." << endl;
2248 // 3c. check for possible bond between each atom in this and every one in the 27 cells
2249 for (n[0]=-1;n[0]<=1;n[0]++)
2250 for (n[1]=-1;n[1]<=1;n[1]++)
2251 for (n[2]=-1;n[2]<=1;n[2]++) {
2252 // compute the index of this comparison cell and make it periodic
2253 index = ((N[2]+n[2]+divisor[2])%divisor[2]) + (((N[1]+n[1]+divisor[1])%divisor[1]) + ((N[0]+n[0]+divisor[0])%divisor[0]) * divisor[1]) * divisor[2];
2254 //*out << Verbose(1) << "Number of comparison cell is " << index << "." << endl;
2255 if (CellList[index] != NULL) { // if there are any atoms in this cell
2256 OtherWalker = CellList[index]->start;
2257 while(OtherWalker->next != CellList[index]->end) { // go through every atom in this cell
2258 OtherWalker = OtherWalker->next;
2259 //*out << Verbose(0) << "Current comparison atom is " << *OtherWalker << "." << endl;
2260 /// \todo periodic check is missing here!
2261 //*out << Verbose(1) << "Checking distance " << OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size) << " against typical bond length of " << bonddistance*bonddistance << "." << endl;
2262 MinDistance = OtherWalker->type->CovalentRadius + Walker->type->CovalentRadius;
2263 MinDistance *= (IsAngstroem) ? 1. : 1./AtomicLengthToAngstroem;
2264 MaxDistance = MinDistance + BONDTHRESHOLD;
2265 MinDistance -= BONDTHRESHOLD;
2266 distance = OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size);
2267 if ((OtherWalker->father->nr > Walker->father->nr) && (distance <= MaxDistance*MaxDistance) && (distance >= MinDistance*MinDistance)) { // create bond if distance is smaller
2268 //*out << Verbose(0) << "Adding Bond between " << *Walker << " and " << *OtherWalker << "." << endl;
2269 AddBond(Walker->father, OtherWalker->father, 1); // also increases molecule::BondCount
2270 BondCount++;
2271 } else {
2272 //*out << Verbose(1) << "Not Adding: Wrong label order or distance too great." << endl;
2273 }
2274 }
2275 }
2276 }
2277 }
2278 }
2279 }
2280 // 4. free the cell again
2281 for (int i=NumberCells;i--;)
2282 if (CellList[i] != NULL) {
2283 delete(CellList[i]);
2284 }
2285 Free((void **)&CellList, "molecule::CreateAdjacencyList - ** CellList");
2286
2287 // create the adjacency list per atom
2288 CreateListOfBondsPerAtom(out);
2289
2290 // correct Bond degree of each bond by checking both bond partners for a mismatch between valence and current sum of bond degrees,
2291 // iteratively increase the one first where the other bond partner has the fewest number of bonds (i.e. in general bonds oxygene
2292 // preferred over carbon bonds). Beforehand, we had picked the first mismatching partner, which lead to oxygenes with single instead of
2293 // double bonds as was expected.
2294 if (BondCount != 0) {
2295 NoCyclicBonds = 0;
2296 *out << Verbose(1) << "Correcting Bond degree of each bond ... ";
2297 do {
2298 No = 0; // No acts as breakup flag (if 1 we still continue)
2299 Walker = start;
2300 while (Walker->next != end) { // go through every atom
2301 Walker = Walker->next;
2302 // count valence of first partner
2303 NoBonds = 0;
2304 for(j=0;j<NumberOfBondsPerAtom[Walker->nr];j++)
2305 NoBonds += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
2306 *out << Verbose(3) << "Walker " << *Walker << ": " << (int)Walker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
2307 if ((int)(Walker->type->NoValenceOrbitals) > NoBonds) { // we have a mismatch, check all bonding partners for mismatch
2308 Candidate = NULL;
2309 CandidateBondNo = -1;
2310 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through each of its bond partners
2311 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
2312 // count valence of second partner
2313 NoBonds = 0;
2314 for(j=0;j<NumberOfBondsPerAtom[OtherWalker->nr];j++)
2315 NoBonds += ListOfBondsPerAtom[OtherWalker->nr][j]->BondDegree;
2316 *out << Verbose(3) << "OtherWalker " << *OtherWalker << ": " << (int)OtherWalker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
2317 if ((int)(OtherWalker->type->NoValenceOrbitals) > NoBonds) { // check if possible candidate
2318 if ((Candidate == NULL) || (NumberOfBondsPerAtom[Candidate->nr] > NumberOfBondsPerAtom[OtherWalker->nr])) { // pick the one with fewer number of bonds first
2319 Candidate = OtherWalker;
2320 CandidateBondNo = i;
2321 *out << Verbose(3) << "New candidate is " << *Candidate << "." << endl;
2322 }
2323 }
2324 }
2325 if ((Candidate != NULL) && (CandidateBondNo != -1)) {
2326 ListOfBondsPerAtom[Walker->nr][CandidateBondNo]->BondDegree++;
2327 *out << Verbose(2) << "Increased bond degree for bond " << *ListOfBondsPerAtom[Walker->nr][CandidateBondNo] << "." << endl;
2328 } else
2329 *out << Verbose(2) << "Could not find correct degree for atom " << *Walker << "." << endl;
2330 FalseBondDegree++;
2331 }
2332 }
2333 } while (No);
2334 *out << " done." << endl;
2335 } else
2336 *out << Verbose(1) << "BondCount is " << BondCount << ", no bonds between any of the " << AtomCount << " atoms." << endl;
2337 *out << Verbose(1) << "I detected " << BondCount << " bonds in the molecule with distance " << bonddistance << ", " << FalseBondDegree << " bonds could not be corrected." << endl;
2338
2339 // output bonds for debugging (if bond chain list was correctly installed)
2340 *out << Verbose(1) << endl << "From contents of bond chain list:";
2341 bond *Binder = first;
2342 while(Binder->next != last) {
2343 Binder = Binder->next;
2344 *out << *Binder << "\t" << endl;
2345 }
2346 *out << endl;
2347 } else
2348 *out << Verbose(1) << "AtomCount is " << AtomCount << ", thus no bonds, no connections!." << endl;
2349 *out << Verbose(0) << "End of CreateAdjacencyList." << endl;
2350 Free((void **)&matrix, "molecule::CreateAdjacencyList: *matrix");
2351};
2352
2353/** Performs a Depth-First search on this molecule.
2354 * Marks bonds in molecule as cyclic, bridge, ... and atoms as
2355 * articulations points, ...
2356 * We use the algorithm from [Even, Graph Algorithms, p.62].
2357 * \param *out output stream for debugging
2358 * \param *&BackEdgeStack NULL pointer to StackClass with all the found back edges, allocated and filled on return
2359 * \return list of each disconnected subgraph as an individual molecule class structure
2360 */
2361MoleculeLeafClass * molecule::DepthFirstSearchAnalysis(ofstream *out, class StackClass<bond *> *&BackEdgeStack)
2362{
2363 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
2364 BackEdgeStack = new StackClass<bond *> (BondCount);
2365 MoleculeLeafClass *SubGraphs = new MoleculeLeafClass(NULL);
2366 MoleculeLeafClass *LeafWalker = SubGraphs;
2367 int CurrentGraphNr = 0, OldGraphNr;
2368 int ComponentNumber = 0;
2369 atom *Walker = NULL, *OtherAtom = NULL, *Root = start->next;
2370 bond *Binder = NULL;
2371 bool BackStepping = false;
2372
2373 *out << Verbose(0) << "Begin of DepthFirstSearchAnalysis" << endl;
2374
2375 ResetAllBondsToUnused();
2376 ResetAllAtomNumbers();
2377 InitComponentNumbers();
2378 BackEdgeStack->ClearStack();
2379 while (Root != end) { // if there any atoms at all
2380 // (1) mark all edges unused, empty stack, set atom->GraphNr = 0 for all
2381 AtomStack->ClearStack();
2382
2383 // put into new subgraph molecule and add this to list of subgraphs
2384 LeafWalker = new MoleculeLeafClass(LeafWalker);
2385 LeafWalker->Leaf = new molecule(elemente);
2386 LeafWalker->Leaf->AddCopyAtom(Root);
2387
2388 OldGraphNr = CurrentGraphNr;
2389 Walker = Root;
2390 do { // (10)
2391 do { // (2) set number and Lowpoint of Atom to i, increase i, push current atom
2392 if (!BackStepping) { // if we don't just return from (8)
2393 Walker->GraphNr = CurrentGraphNr;
2394 Walker->LowpointNr = CurrentGraphNr;
2395 *out << Verbose(1) << "Setting Walker[" << Walker->Name << "]'s number to " << Walker->GraphNr << " with Lowpoint " << Walker->LowpointNr << "." << endl;
2396 AtomStack->Push(Walker);
2397 CurrentGraphNr++;
2398 }
2399 do { // (3) if Walker has no unused egdes, go to (5)
2400 BackStepping = false; // reset backstepping flag for (8)
2401 if (Binder == NULL) // if we don't just return from (11), Binder is already set to next unused
2402 Binder = FindNextUnused(Walker);
2403 if (Binder == NULL)
2404 break;
2405 *out << Verbose(2) << "Current Unused Bond is " << *Binder << "." << endl;
2406 // (4) Mark Binder used, ...
2407 Binder->MarkUsed(black);
2408 OtherAtom = Binder->GetOtherAtom(Walker);
2409 *out << Verbose(2) << "(4) OtherAtom is " << OtherAtom->Name << "." << endl;
2410 if (OtherAtom->GraphNr != -1) {
2411 // (4a) ... if "other" atom has been visited (GraphNr != 0), set lowpoint to minimum of both, go to (3)
2412 Binder->Type = BackEdge;
2413 BackEdgeStack->Push(Binder);
2414 Walker->LowpointNr = ( Walker->LowpointNr < OtherAtom->GraphNr ) ? Walker->LowpointNr : OtherAtom->GraphNr;
2415 *out << Verbose(3) << "(4a) Visited: Setting Lowpoint of Walker[" << Walker->Name << "] to " << Walker->LowpointNr << "." << endl;
2416 } else {
2417 // (4b) ... otherwise set OtherAtom as Ancestor of Walker and Walker as OtherAtom, go to (2)
2418 Binder->Type = TreeEdge;
2419 OtherAtom->Ancestor = Walker;
2420 Walker = OtherAtom;
2421 *out << Verbose(3) << "(4b) Not Visited: OtherAtom[" << OtherAtom->Name << "]'s Ancestor is now " << OtherAtom->Ancestor->Name << ", Walker is OtherAtom " << OtherAtom->Name << "." << endl;
2422 break;
2423 }
2424 Binder = NULL;
2425 } while (1); // (3)
2426 if (Binder == NULL) {
2427 *out << Verbose(2) << "No more Unused Bonds." << endl;
2428 break;
2429 } else
2430 Binder = NULL;
2431 } while (1); // (2)
2432
2433 // if we came from backstepping, yet there were no more unused bonds, we end up here with no Ancestor, because Walker is Root! Then we are finished!
2434 if ((Walker == Root) && (Binder == NULL))
2435 break;
2436
2437 // (5) if Ancestor of Walker is ...
2438 *out << Verbose(1) << "(5) Number of Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "] is " << Walker->Ancestor->GraphNr << "." << endl;
2439 if (Walker->Ancestor->GraphNr != Root->GraphNr) {
2440 // (6) (Ancestor of Walker is not Root)
2441 if (Walker->LowpointNr < Walker->Ancestor->GraphNr) {
2442 // (6a) set Ancestor's Lowpoint number to minimum of of its Ancestor and itself, go to Step(8)
2443 Walker->Ancestor->LowpointNr = (Walker->Ancestor->LowpointNr < Walker->LowpointNr) ? Walker->Ancestor->LowpointNr : Walker->LowpointNr;
2444 *out << Verbose(2) << "(6) Setting Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s Lowpoint to " << Walker->Ancestor->LowpointNr << "." << endl;
2445 } else {
2446 // (7) (Ancestor of Walker is a separating vertex, remove all from stack till Walker (including), these and Ancestor form a component
2447 Walker->Ancestor->SeparationVertex = true;
2448 *out << Verbose(2) << "(7) Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s is a separating vertex, creating component." << endl;
2449 SetNextComponentNumber(Walker->Ancestor, ComponentNumber);
2450 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Ancestor's Compont is " << ComponentNumber << "." << endl;
2451 SetNextComponentNumber(Walker, ComponentNumber);
2452 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Compont is " << ComponentNumber << "." << endl;
2453 do {
2454 OtherAtom = AtomStack->PopLast();
2455 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
2456 SetNextComponentNumber(OtherAtom, ComponentNumber);
2457 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
2458 } while (OtherAtom != Walker);
2459 ComponentNumber++;
2460 }
2461 // (8) Walker becomes its Ancestor, go to (3)
2462 *out << Verbose(2) << "(8) Walker[" << Walker->Name << "] is now its Ancestor " << Walker->Ancestor->Name << ", backstepping. " << endl;
2463 Walker = Walker->Ancestor;
2464 BackStepping = true;
2465 }
2466 if (!BackStepping) { // coming from (8) want to go to (3)
2467 // (9) remove all from stack till Walker (including), these and Root form a component
2468 AtomStack->Output(out);
2469 SetNextComponentNumber(Root, ComponentNumber);
2470 *out << Verbose(3) << "(9) Root[" << Root->Name << "]'s Component is " << ComponentNumber << "." << endl;
2471 SetNextComponentNumber(Walker, ComponentNumber);
2472 *out << Verbose(3) << "(9) Walker[" << Walker->Name << "]'s Component is " << ComponentNumber << "." << endl;
2473 do {
2474 OtherAtom = AtomStack->PopLast();
2475 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
2476 SetNextComponentNumber(OtherAtom, ComponentNumber);
2477 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
2478 } while (OtherAtom != Walker);
2479 ComponentNumber++;
2480
2481 // (11) Root is separation vertex, set Walker to Root and go to (4)
2482 Walker = Root;
2483 Binder = FindNextUnused(Walker);
2484 *out << Verbose(1) << "(10) Walker is Root[" << Root->Name << "], next Unused Bond is " << Binder << "." << endl;
2485 if (Binder != NULL) { // Root is separation vertex
2486 *out << Verbose(1) << "(11) Root is a separation vertex." << endl;
2487 Walker->SeparationVertex = true;
2488 }
2489 }
2490 } while ((BackStepping) || (Binder != NULL)); // (10) halt only if Root has no unused edges
2491
2492 // From OldGraphNr to CurrentGraphNr ranges an disconnected subgraph
2493 *out << Verbose(0) << "Disconnected subgraph ranges from " << OldGraphNr << " to " << CurrentGraphNr << "." << endl;
2494 LeafWalker->Leaf->Output(out);
2495 *out << endl;
2496
2497 // step on to next root
2498 while ((Root != end) && (Root->GraphNr != -1)) {
2499 //*out << Verbose(1) << "Current next subgraph root candidate is " << Root->Name << "." << endl;
2500 if (Root->GraphNr != -1) // if already discovered, step on
2501 Root = Root->next;
2502 }
2503 }
2504 // set cyclic bond criterium on "same LP" basis
2505 Binder = first;
2506 while(Binder->next != last) {
2507 Binder = Binder->next;
2508 if (Binder->rightatom->LowpointNr == Binder->leftatom->LowpointNr) { // cyclic ??
2509 Binder->Cyclic = true;
2510 NoCyclicBonds++;
2511 }
2512 }
2513
2514
2515 *out << Verbose(1) << "Final graph info for each atom is:" << endl;
2516 Walker = start;
2517 while (Walker->next != end) {
2518 Walker = Walker->next;
2519 *out << Verbose(2) << "Atom " << Walker->Name << " is " << ((Walker->SeparationVertex) ? "a" : "not a") << " separation vertex, components are ";
2520 OutputComponentNumber(out, Walker);
2521 *out << " with Lowpoint " << Walker->LowpointNr << " and Graph Nr. " << Walker->GraphNr << "." << endl;
2522 }
2523
2524 *out << Verbose(1) << "Final graph info for each bond is:" << endl;
2525 Binder = first;
2526 while(Binder->next != last) {
2527 Binder = Binder->next;
2528 *out << Verbose(2) << ((Binder->Type == TreeEdge) ? "TreeEdge " : "BackEdge ") << *Binder << ": <";
2529 *out << ((Binder->leftatom->SeparationVertex) ? "SP," : "") << "L" << Binder->leftatom->LowpointNr << " G" << Binder->leftatom->GraphNr << " Comp.";
2530 OutputComponentNumber(out, Binder->leftatom);
2531 *out << " === ";
2532 *out << ((Binder->rightatom->SeparationVertex) ? "SP," : "") << "L" << Binder->rightatom->LowpointNr << " G" << Binder->rightatom->GraphNr << " Comp.";
2533 OutputComponentNumber(out, Binder->rightatom);
2534 *out << ">." << endl;
2535 if (Binder->Cyclic) // cyclic ??
2536 *out << Verbose(3) << "Lowpoint at each side are equal: CYCLIC!" << endl;
2537 }
2538
2539 // free all and exit
2540 delete(AtomStack);
2541 *out << Verbose(0) << "End of DepthFirstSearchAnalysis" << endl;
2542 return SubGraphs;
2543};
2544
2545/** Analyses the cycles found and returns minimum of all cycle lengths.
2546 * We begin with a list of Back edges found during DepthFirstSearchAnalysis(). We go through this list - one end is the Root,
2547 * the other our initial Walker - and do a Breadth First Search for the Root. We mark down each Predecessor and as soon as
2548 * we have found the Root via BFS, we may climb back the closed cycle via the Predecessors. Thereby we mark atoms and bonds
2549 * as cyclic and print out the cycles.
2550 * \param *out output stream for debugging
2551 * \param *BackEdgeStack stack with all back edges found during DFS scan. Beware: This stack contains the bonds from the total molecule, not from the subgraph!
2552 * \param *&MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found, if set is maximum search distance
2553 * \todo BFS from the not-same-LP to find back to starting point of tributary cycle over more than one bond
2554 */
2555void molecule::CyclicStructureAnalysis(ofstream *out, class StackClass<bond *> * BackEdgeStack, int *&MinimumRingSize)
2556{
2557 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CyclicStructureAnalysis: **PredecessorList");
2558 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CyclicStructureAnalysis: *ShortestPathList");
2559 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CyclicStructureAnalysis: *ColorList");
2560 class StackClass<atom *> *BFSStack = new StackClass<atom *> (AtomCount); // will hold the current ring
2561 class StackClass<atom *> *TouchedStack = new StackClass<atom *> (AtomCount); // contains all "touched" atoms (that need to be reset after BFS loop)
2562 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL;
2563 bond *Binder = NULL, *BackEdge = NULL;
2564 int RingSize, NumCycles, MinRingSize = -1;
2565
2566 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
2567 for (int i=AtomCount;i--;) {
2568 PredecessorList[i] = NULL;
2569 ShortestPathList[i] = -1;
2570 ColorList[i] = white;
2571 }
2572
2573 *out << Verbose(1) << "Back edge list - ";
2574 BackEdgeStack->Output(out);
2575
2576 *out << Verbose(1) << "Analysing cycles ... " << endl;
2577 NumCycles = 0;
2578 while (!BackEdgeStack->IsEmpty()) {
2579 BackEdge = BackEdgeStack->PopFirst();
2580 // this is the target
2581 Root = BackEdge->leftatom;
2582 // this is the source point
2583 Walker = BackEdge->rightatom;
2584 ShortestPathList[Walker->nr] = 0;
2585 BFSStack->ClearStack(); // start with empty BFS stack
2586 BFSStack->Push(Walker);
2587 TouchedStack->Push(Walker);
2588 *out << Verbose(1) << "---------------------------------------------------------------------------------------------------------" << endl;
2589 OtherAtom = NULL;
2590 do { // look for Root
2591 Walker = BFSStack->PopFirst();
2592 *out << Verbose(2) << "Current Walker is " << *Walker << ", we look for SP to Root " << *Root << "." << endl;
2593 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2594 Binder = ListOfBondsPerAtom[Walker->nr][i];
2595 if (Binder != BackEdge) { // only walk along DFS spanning tree (otherwise we always find SP of one being backedge Binder)
2596 OtherAtom = Binder->GetOtherAtom(Walker);
2597#ifdef ADDHYDROGEN
2598 if (OtherAtom->type->Z != 1) {
2599#endif
2600 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2601 if (ColorList[OtherAtom->nr] == white) {
2602 TouchedStack->Push(OtherAtom);
2603 ColorList[OtherAtom->nr] = lightgray;
2604 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2605 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2606 *out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " lightgray, its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
2607 //if (ShortestPathList[OtherAtom->nr] < MinimumRingSize[Walker->GetTrueFather()->nr]) { // Check for maximum distance
2608 *out << Verbose(3) << "Putting OtherAtom into queue." << endl;
2609 BFSStack->Push(OtherAtom);
2610 //}
2611 } else {
2612 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
2613 }
2614 if (OtherAtom == Root)
2615 break;
2616#ifdef ADDHYDROGEN
2617 } else {
2618 *out << Verbose(2) << "Skipping hydrogen atom " << *OtherAtom << "." << endl;
2619 ColorList[OtherAtom->nr] = black;
2620 }
2621#endif
2622 } else {
2623 *out << Verbose(2) << "Bond " << *Binder << " not Visiting, is the back edge." << endl;
2624 }
2625 }
2626 ColorList[Walker->nr] = black;
2627 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2628 if (OtherAtom == Root) { // if we have found the root, check whether this cycle wasn't already found beforehand
2629 // step through predecessor list
2630 while (OtherAtom != BackEdge->rightatom) {
2631 if (!OtherAtom->GetTrueFather()->IsCyclic) // if one bond in the loop is not marked as cyclic, we haven't found this cycle yet
2632 break;
2633 else
2634 OtherAtom = PredecessorList[OtherAtom->nr];
2635 }
2636 if (OtherAtom == BackEdge->rightatom) { // if each atom in found cycle is cyclic, loop's been found before already
2637 *out << Verbose(3) << "This cycle was already found before, skipping and removing seeker from search." << endl;\
2638 do {
2639 OtherAtom = TouchedStack->PopLast();
2640 if (PredecessorList[OtherAtom->nr] == Walker) {
2641 *out << Verbose(4) << "Removing " << *OtherAtom << " from lists and stacks." << endl;
2642 PredecessorList[OtherAtom->nr] = NULL;
2643 ShortestPathList[OtherAtom->nr] = -1;
2644 ColorList[OtherAtom->nr] = white;
2645 BFSStack->RemoveItem(OtherAtom);
2646 }
2647 } while ((!TouchedStack->IsEmpty()) && (PredecessorList[OtherAtom->nr] == NULL));
2648 TouchedStack->Push(OtherAtom); // last was wrongly popped
2649 OtherAtom = BackEdge->rightatom; // set to not Root
2650 } else
2651 OtherAtom = Root;
2652 }
2653 } while ((!BFSStack->IsEmpty()) && (OtherAtom != Root) && (OtherAtom != NULL)); // || (ShortestPathList[OtherAtom->nr] < MinimumRingSize[Walker->GetTrueFather()->nr])));
2654
2655 if (OtherAtom == Root) {
2656 // now climb back the predecessor list and thus find the cycle members
2657 NumCycles++;
2658 RingSize = 1;
2659 Root->GetTrueFather()->IsCyclic = true;
2660 *out << Verbose(1) << "Found ring contains: ";
2661 Walker = Root;
2662 while (Walker != BackEdge->rightatom) {
2663 *out << Walker->Name << " <-> ";
2664 Walker = PredecessorList[Walker->nr];
2665 Walker->GetTrueFather()->IsCyclic = true;
2666 RingSize++;
2667 }
2668 *out << Walker->Name << " with a length of " << RingSize << "." << endl << endl;
2669 // walk through all and set MinimumRingSize
2670 Walker = Root;
2671 MinimumRingSize[Walker->GetTrueFather()->nr] = RingSize;
2672 while (Walker != BackEdge->rightatom) {
2673 Walker = PredecessorList[Walker->nr];
2674 if (RingSize < MinimumRingSize[Walker->GetTrueFather()->nr])
2675 MinimumRingSize[Walker->GetTrueFather()->nr] = RingSize;
2676 }
2677 if ((RingSize < MinRingSize) || (MinRingSize == -1))
2678 MinRingSize = RingSize;
2679 } else {
2680 *out << Verbose(1) << "No ring containing " << *Root << " with length equal to or smaller than " << MinimumRingSize[Walker->GetTrueFather()->nr] << " found." << endl;
2681 }
2682
2683 // now clean the lists
2684 while (!TouchedStack->IsEmpty()){
2685 Walker = TouchedStack->PopFirst();
2686 PredecessorList[Walker->nr] = NULL;
2687 ShortestPathList[Walker->nr] = -1;
2688 ColorList[Walker->nr] = white;
2689 }
2690 }
2691 if (MinRingSize != -1) {
2692 // go over all atoms
2693 Root = start;
2694 while(Root->next != end) {
2695 Root = Root->next;
2696
2697 if (MinimumRingSize[Root->GetTrueFather()->nr] == AtomCount) { // check whether MinimumRingSize is set, if not BFS to next where it is
2698 Walker = Root;
2699 ShortestPathList[Walker->nr] = 0;
2700 BFSStack->ClearStack(); // start with empty BFS stack
2701 BFSStack->Push(Walker);
2702 TouchedStack->Push(Walker);
2703 //*out << Verbose(1) << "---------------------------------------------------------------------------------------------------------" << endl;
2704 OtherAtom = Walker;
2705 while (OtherAtom != NULL) { // look for Root
2706 Walker = BFSStack->PopFirst();
2707 //*out << Verbose(2) << "Current Walker is " << *Walker << ", we look for SP to Root " << *Root << "." << endl;
2708 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2709 Binder = ListOfBondsPerAtom[Walker->nr][i];
2710 if ((Binder != BackEdge) || (NumberOfBondsPerAtom[Walker->nr] == 1)) { // only walk along DFS spanning tree (otherwise we always find SP of 1 being backedge Binder), but terminal hydrogens may be connected via backedge, hence extra check
2711 OtherAtom = Binder->GetOtherAtom(Walker);
2712 //*out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2713 if (ColorList[OtherAtom->nr] == white) {
2714 TouchedStack->Push(OtherAtom);
2715 ColorList[OtherAtom->nr] = lightgray;
2716 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2717 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2718 //*out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " lightgray, its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
2719 if (OtherAtom->GetTrueFather()->IsCyclic) { // if the other atom is connected to a ring
2720 MinimumRingSize[Root->GetTrueFather()->nr] = ShortestPathList[OtherAtom->nr]+MinimumRingSize[OtherAtom->GetTrueFather()->nr];
2721 OtherAtom = NULL; //break;
2722 break;
2723 } else
2724 BFSStack->Push(OtherAtom);
2725 } else {
2726 //*out << Verbose(3) << "Not Adding, has already been visited." << endl;
2727 }
2728 } else {
2729 //*out << Verbose(3) << "Not Visiting, is a back edge." << endl;
2730 }
2731 }
2732 ColorList[Walker->nr] = black;
2733 //*out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2734 }
2735
2736 // now clean the lists
2737 while (!TouchedStack->IsEmpty()){
2738 Walker = TouchedStack->PopFirst();
2739 PredecessorList[Walker->nr] = NULL;
2740 ShortestPathList[Walker->nr] = -1;
2741 ColorList[Walker->nr] = white;
2742 }
2743 }
2744 *out << Verbose(1) << "Minimum ring size of " << *Root << " is " << MinimumRingSize[Root->GetTrueFather()->nr] << "." << endl;
2745 }
2746 *out << Verbose(1) << "Minimum ring size is " << MinRingSize << ", over " << NumCycles << " cycles total." << endl;
2747 } else
2748 *out << Verbose(1) << "No rings were detected in the molecular structure." << endl;
2749
2750 Free((void **)&PredecessorList, "molecule::CyclicStructureAnalysis: **PredecessorList");
2751 Free((void **)&ShortestPathList, "molecule::CyclicStructureAnalysis: **ShortestPathList");
2752 Free((void **)&ColorList, "molecule::CyclicStructureAnalysis: **ColorList");
2753 delete(BFSStack);
2754};
2755
2756/** Sets the next component number.
2757 * This is O(N) as the number of bonds per atom is bound.
2758 * \param *vertex atom whose next atom::*ComponentNr is to be set
2759 * \param nr number to use
2760 */
2761void molecule::SetNextComponentNumber(atom *vertex, int nr)
2762{
2763 int i=0;
2764 if (vertex != NULL) {
2765 for(;i<NumberOfBondsPerAtom[vertex->nr];i++) {
2766 if (vertex->ComponentNr[i] == -1) { // check if not yet used
2767 vertex->ComponentNr[i] = nr;
2768 break;
2769 }
2770 else if (vertex->ComponentNr[i] == nr) // if number is already present, don't add another time
2771 break; // breaking here will not cause error!
2772 }
2773 if (i == NumberOfBondsPerAtom[vertex->nr])
2774 cerr << "Error: All Component entries are already occupied!" << endl;
2775 } else
2776 cerr << "Error: Given vertex is NULL!" << endl;
2777};
2778
2779/** Output a list of flags, stating whether the bond was visited or not.
2780 * \param *out output stream for debugging
2781 */
2782void molecule::OutputComponentNumber(ofstream *out, atom *vertex)
2783{
2784 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
2785 *out << vertex->ComponentNr[i] << " ";
2786};
2787
2788/** Allocates memory for all atom::*ComponentNr in this molecule and sets each entry to -1.
2789 */
2790void molecule::InitComponentNumbers()
2791{
2792 atom *Walker = start;
2793 while(Walker->next != end) {
2794 Walker = Walker->next;
2795 if (Walker->ComponentNr != NULL)
2796 Free((void **)&Walker->ComponentNr, "molecule::InitComponentNumbers: **Walker->ComponentNr");
2797 Walker->ComponentNr = (int *) Malloc(sizeof(int)*NumberOfBondsPerAtom[Walker->nr], "molecule::InitComponentNumbers: *Walker->ComponentNr");
2798 for (int i=NumberOfBondsPerAtom[Walker->nr];i--;)
2799 Walker->ComponentNr[i] = -1;
2800 }
2801};
2802
2803/** Returns next unused bond for this atom \a *vertex or NULL of none exists.
2804 * \param *vertex atom to regard
2805 * \return bond class or NULL
2806 */
2807bond * molecule::FindNextUnused(atom *vertex)
2808{
2809 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
2810 if (ListOfBondsPerAtom[vertex->nr][i]->IsUsed() == white)
2811 return(ListOfBondsPerAtom[vertex->nr][i]);
2812 return NULL;
2813};
2814
2815/** Resets bond::Used flag of all bonds in this molecule.
2816 * \return true - success, false - -failure
2817 */
2818void molecule::ResetAllBondsToUnused()
2819{
2820 bond *Binder = first;
2821 while (Binder->next != last) {
2822 Binder = Binder->next;
2823 Binder->ResetUsed();
2824 }
2825};
2826
2827/** Resets atom::nr to -1 of all atoms in this molecule.
2828 */
2829void molecule::ResetAllAtomNumbers()
2830{
2831 atom *Walker = start;
2832 while (Walker->next != end) {
2833 Walker = Walker->next;
2834 Walker->GraphNr = -1;
2835 }
2836};
2837
2838/** Output a list of flags, stating whether the bond was visited or not.
2839 * \param *out output stream for debugging
2840 * \param *list
2841 */
2842void OutputAlreadyVisited(ofstream *out, int *list)
2843{
2844 *out << Verbose(4) << "Already Visited Bonds:\t";
2845 for(int i=1;i<=list[0];i++) *out << Verbose(0) << list[i] << " ";
2846 *out << endl;
2847};
2848
2849/** Estimates by educated guessing (using upper limit) the expected number of fragments.
2850 * The upper limit is
2851 * \f[
2852 * n = N \cdot C^k
2853 * \f]
2854 * where \f$C=2^c\f$ and c is the maximum bond degree over N number of atoms.
2855 * \param *out output stream for debugging
2856 * \param order bond order k
2857 * \return number n of fragments
2858 */
2859int molecule::GuesstimateFragmentCount(ofstream *out, int order)
2860{
2861 int c = 0;
2862 int FragmentCount;
2863 // get maximum bond degree
2864 atom *Walker = start;
2865 while (Walker->next != end) {
2866 Walker = Walker->next;
2867 c = (NumberOfBondsPerAtom[Walker->nr] > c) ? NumberOfBondsPerAtom[Walker->nr] : c;
2868 }
2869 FragmentCount = NoNonHydrogen*(1 << (c*order));
2870 *out << Verbose(1) << "Upper limit for this subgraph is " << FragmentCount << " for " << NoNonHydrogen << " non-H atoms with maximum bond degree of " << c << "." << endl;
2871 return FragmentCount;
2872};
2873
2874/** Scans a single line for number and puts them into \a KeySet.
2875 * \param *out output stream for debugging
2876 * \param *buffer buffer to scan
2877 * \param &CurrentSet filled KeySet on return
2878 * \return true - at least one valid atom id parsed, false - CurrentSet is empty
2879 */
2880bool molecule::ScanBufferIntoKeySet(ofstream *out, char *buffer, KeySet &CurrentSet)
2881{
2882 stringstream line;
2883 int AtomNr;
2884 int status = 0;
2885
2886 line.str(buffer);
2887 while (!line.eof()) {
2888 line >> AtomNr;
2889 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
2890 CurrentSet.insert(AtomNr); // insert at end, hence in same order as in file!
2891 status++;
2892 } // else it's "-1" or else and thus must not be added
2893 }
2894 *out << Verbose(1) << "The scanned KeySet is ";
2895 for(KeySet::iterator runner = CurrentSet.begin(); runner != CurrentSet.end(); runner++) {
2896 *out << (*runner) << "\t";
2897 }
2898 *out << endl;
2899 return (status != 0);
2900};
2901
2902/** Parses the KeySet file and fills \a *FragmentList from the known molecule structure.
2903 * Does two-pass scanning:
2904 * -# Scans the keyset file and initialises a temporary graph
2905 * -# Scans TEFactors file and sets the TEFactor of each key set in the temporary graph accordingly
2906 * Finally, the temporary graph is inserted into the given \a FragmentList for return.
2907 * \param *out output stream for debugging
2908 * \param *path path to file
2909 * \param *FragmentList empty, filled on return
2910 * \return true - parsing successfully, false - failure on parsing (FragmentList will be NULL)
2911 */
2912bool molecule::ParseKeySetFile(ofstream *out, char *path, Graph *&FragmentList)
2913{
2914 bool status = true;
2915 ifstream InputFile;
2916 stringstream line;
2917 GraphTestPair testGraphInsert;
2918 int NumberOfFragments = 0;
2919 double TEFactor;
2920 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - filename");
2921
2922 if (FragmentList == NULL) { // check list pointer
2923 FragmentList = new Graph;
2924 }
2925
2926 // 1st pass: open file and read
2927 *out << Verbose(1) << "Parsing the KeySet file ... " << endl;
2928 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, KEYSETFILE);
2929 InputFile.open(filename);
2930 if (InputFile != NULL) {
2931 // each line represents a new fragment
2932 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - *buffer");
2933 // 1. parse keysets and insert into temp. graph
2934 while (!InputFile.eof()) {
2935 InputFile.getline(buffer, MAXSTRINGSIZE);
2936 KeySet CurrentSet;
2937 if ((strlen(buffer) > 0) && (ScanBufferIntoKeySet(out, buffer, CurrentSet))) { // if at least one valid atom was added, write config
2938 testGraphInsert = FragmentList->insert(GraphPair (CurrentSet,pair<int,double>(NumberOfFragments++,1))); // store fragment number and current factor
2939 if (!testGraphInsert.second) {
2940 cerr << "KeySet file must be corrupt as there are two equal key sets therein!" << endl;
2941 }
2942 //FragmentList->ListOfMolecules[NumberOfFragments++] = StoreFragmentFromKeySet(out, CurrentSet, IsAngstroem);
2943 }
2944 }
2945 // 2. Free and done
2946 InputFile.close();
2947 InputFile.clear();
2948 Free((void **)&buffer, "molecule::ParseKeySetFile - *buffer");
2949 *out << Verbose(1) << "done." << endl;
2950 } else {
2951 *out << Verbose(1) << "File " << filename << " not found." << endl;
2952 status = false;
2953 }
2954
2955 // 2nd pass: open TEFactors file and read
2956 *out << Verbose(1) << "Parsing the TEFactors file ... " << endl;
2957 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, TEFACTORSFILE);
2958 InputFile.open(filename);
2959 if (InputFile != NULL) {
2960 // 3. add found TEFactors to each keyset
2961 NumberOfFragments = 0;
2962 for(Graph::iterator runner = FragmentList->begin();runner != FragmentList->end(); runner++) {
2963 if (!InputFile.eof()) {
2964 InputFile >> TEFactor;
2965 (*runner).second.second = TEFactor;
2966 *out << Verbose(2) << "Setting " << ++NumberOfFragments << " fragment's TEFactor to " << (*runner).second.second << "." << endl;
2967 } else {
2968 status = false;
2969 break;
2970 }
2971 }
2972 // 4. Free and done
2973 InputFile.close();
2974 *out << Verbose(1) << "done." << endl;
2975 } else {
2976 *out << Verbose(1) << "File " << filename << " not found." << endl;
2977 status = false;
2978 }
2979
2980 // free memory
2981 Free((void **)&filename, "molecule::ParseKeySetFile - filename");
2982
2983 return status;
2984};
2985
2986/** Stores keysets and TEFactors to file.
2987 * \param *out output stream for debugging
2988 * \param KeySetList Graph with Keysets and factors
2989 * \param *path path to file
2990 * \return true - file written successfully, false - writing failed
2991 */
2992bool molecule::StoreKeySetFile(ofstream *out, Graph &KeySetList, char *path)
2993{
2994 ofstream output;
2995 bool status = true;
2996 string line;
2997
2998 // open KeySet file
2999 line = path;
3000 line.append("/");
3001 line += FRAGMENTPREFIX;
3002 line += KEYSETFILE;
3003 output.open(line.c_str(), ios::out);
3004 *out << Verbose(1) << "Saving key sets of the total graph ... ";
3005 if(output != NULL) {
3006 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++) {
3007 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
3008 if (sprinter != (*runner).first.begin())
3009 output << "\t";
3010 output << *sprinter;
3011 }
3012 output << endl;
3013 }
3014 *out << "done." << endl;
3015 } else {
3016 cerr << "Unable to open " << line << " for writing keysets!" << endl;
3017 status = false;
3018 }
3019 output.close();
3020 output.clear();
3021
3022 // open TEFactors file
3023 line = path;
3024 line.append("/");
3025 line += FRAGMENTPREFIX;
3026 line += TEFACTORSFILE;
3027 output.open(line.c_str(), ios::out);
3028 *out << Verbose(1) << "Saving TEFactors of the total graph ... ";
3029 if(output != NULL) {
3030 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++)
3031 output << (*runner).second.second << endl;
3032 *out << Verbose(1) << "done." << endl;
3033 } else {
3034 *out << Verbose(1) << "failed to open " << line << "." << endl;
3035 status = false;
3036 }
3037 output.close();
3038
3039 return status;
3040};
3041
3042/** Storing the bond structure of a molecule to file.
3043 * Simply stores Atom::nr and then the Atom::nr of all bond partners per line.
3044 * \param *out output stream for debugging
3045 * \param *path path to file
3046 * \return true - file written successfully, false - writing failed
3047 */
3048bool molecule::StoreAdjacencyToFile(ofstream *out, char *path)
3049{
3050 ofstream AdjacencyFile;
3051 atom *Walker = NULL;
3052 stringstream line;
3053 bool status = true;
3054
3055 line << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
3056 AdjacencyFile.open(line.str().c_str(), ios::out);
3057 *out << Verbose(1) << "Saving adjacency list ... ";
3058 if (AdjacencyFile != NULL) {
3059 Walker = start;
3060 while(Walker->next != end) {
3061 Walker = Walker->next;
3062 AdjacencyFile << Walker->nr << "\t";
3063 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
3064 AdjacencyFile << ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker)->nr << "\t";
3065 AdjacencyFile << endl;
3066 }
3067 AdjacencyFile.close();
3068 *out << Verbose(1) << "done." << endl;
3069 } else {
3070 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
3071 status = false;
3072 }
3073
3074 return status;
3075};
3076
3077/** Checks contents of adjacency file against bond structure in structure molecule.
3078 * \param *out output stream for debugging
3079 * \param *path path to file
3080 * \param **ListOfAtoms allocated (molecule::AtomCount) and filled lookup table for ids (Atom::nr) to *Atom
3081 * \return true - structure is equal, false - not equivalence
3082 */
3083bool molecule::CheckAdjacencyFileAgainstMolecule(ofstream *out, char *path, atom **ListOfAtoms)
3084{
3085 ifstream File;
3086 stringstream filename;
3087 bool status = true;
3088 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
3089
3090 filename << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
3091 File.open(filename.str().c_str(), ios::out);
3092 *out << Verbose(1) << "Looking at bond structure stored in adjacency file and comparing to present one ... ";
3093 if (File != NULL) {
3094 // allocate storage structure
3095 int NonMatchNumber = 0; // will number of atoms with differing bond structure
3096 int *CurrentBonds = (int *) Malloc(sizeof(int)*8, "molecule::CheckAdjacencyFileAgainstMolecule - CurrentBonds"); // contains parsed bonds of current atom
3097 int CurrentBondsOfAtom;
3098
3099 // Parse the file line by line and count the bonds
3100 while (!File.eof()) {
3101 File.getline(buffer, MAXSTRINGSIZE);
3102 stringstream line;
3103 line.str(buffer);
3104 int AtomNr = -1;
3105 line >> AtomNr;
3106 CurrentBondsOfAtom = -1; // we count one too far due to line end
3107 // parse into structure
3108 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
3109 while (!line.eof())
3110 line >> CurrentBonds[ ++CurrentBondsOfAtom ];
3111 // compare against present bonds
3112 //cout << Verbose(2) << "Walker is " << *Walker << ", bond partners: ";
3113 if (CurrentBondsOfAtom == NumberOfBondsPerAtom[AtomNr]) {
3114 for(int i=0;i<NumberOfBondsPerAtom[AtomNr];i++) {
3115 int id = ListOfBondsPerAtom[AtomNr][i]->GetOtherAtom(ListOfAtoms[AtomNr])->nr;
3116 int j = 0;
3117 for (;(j<CurrentBondsOfAtom) && (CurrentBonds[j++] != id);); // check against all parsed bonds
3118 if (CurrentBonds[j-1] != id) { // no match ? Then mark in ListOfAtoms
3119 ListOfAtoms[AtomNr] = NULL;
3120 NonMatchNumber++;
3121 status = false;
3122 //out << "[" << id << "]\t";
3123 } else {
3124 //out << id << "\t";
3125 }
3126 }
3127 //out << endl;
3128 } else {
3129 *out << "Number of bonds for Atom " << *ListOfAtoms[AtomNr] << " does not match, parsed " << CurrentBondsOfAtom << " against " << NumberOfBondsPerAtom[AtomNr] << "." << endl;
3130 status = false;
3131 }
3132 }
3133 }
3134 File.close();
3135 File.clear();
3136 if (status) { // if equal we parse the KeySetFile
3137 *out << Verbose(1) << "done: Equal." << endl;
3138 status = true;
3139 } else
3140 *out << Verbose(1) << "done: Not equal by " << NonMatchNumber << " atoms." << endl;
3141 Free((void **)&CurrentBonds, "molecule::CheckAdjacencyFileAgainstMolecule - **CurrentBonds");
3142 } else {
3143 *out << Verbose(1) << "Adjacency file not found." << endl;
3144 status = false;
3145 }
3146 *out << endl;
3147 Free((void **)&buffer, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
3148
3149 return status;
3150};
3151
3152/** Checks whether the OrderAtSite is still below \a Order at some site.
3153 * \param *out output stream for debugging
3154 * \param *AtomMask defines true/false per global Atom::nr to mask in/out each nuclear site, used to activate given number of site to increment order adaptively
3155 * \param *GlobalKeySetList list of keysets with global ids (valid in "this" molecule) needed for adaptive increase
3156 * \param Order desired Order if positive, desired exponent in threshold criteria if negative (0 is single-step)
3157 * \param *MinimumRingSize array of max. possible order to avoid loops
3158 * \param *path path to ENERGYPERFRAGMENT file (may be NULL if Order is non-negative)
3159 * \return true - needs further fragmentation, false - does not need fragmentation
3160 */
3161bool molecule::CheckOrderAtSite(ofstream *out, bool *AtomMask, Graph *GlobalKeySetList, int Order, int *MinimumRingSize, char *path)
3162{
3163 atom *Walker = start;
3164 bool status = false;
3165 ifstream InputFile;
3166
3167 // initialize mask list
3168 for(int i=AtomCount;i--;)
3169 AtomMask[i] = false;
3170
3171 if (Order < 0) { // adaptive increase of BondOrder per site
3172 if (AtomMask[AtomCount] == true) // break after one step
3173 return false;
3174 // parse the EnergyPerFragment file
3175 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckOrderAtSite: *buffer");
3176 sprintf(buffer, "%s/%s%s.dat", path, FRAGMENTPREFIX, ENERGYPERFRAGMENT);
3177 InputFile.open(buffer, ios::in);
3178 if ((InputFile != NULL) && (GlobalKeySetList != NULL)) {
3179 // transmorph graph keyset list into indexed KeySetList
3180 map<int,KeySet> IndexKeySetList;
3181 for(Graph::iterator runner = GlobalKeySetList->begin(); runner != GlobalKeySetList->end(); runner++) {
3182 IndexKeySetList.insert( pair<int,KeySet>(runner->second.first,runner->first) );
3183 }
3184 int lines = 0;
3185 // count the number of lines, i.e. the number of fragments
3186 InputFile.getline(buffer, MAXSTRINGSIZE); // skip comment lines
3187 InputFile.getline(buffer, MAXSTRINGSIZE);
3188 while(!InputFile.eof()) {
3189 InputFile.getline(buffer, MAXSTRINGSIZE);
3190 lines++;
3191 }
3192 //*out << Verbose(2) << "Scanned " << lines-1 << " lines." << endl; // one endline too much
3193 InputFile.clear();
3194 InputFile.seekg(ios::beg);
3195 map<int, pair<double,int> > AdaptiveCriteriaList; // (Root No., (Value, Order)) !
3196 int No, FragOrder;
3197 double Value;
3198 // each line represents a fragment root (Atom::nr) id and its energy contribution
3199 InputFile.getline(buffer, MAXSTRINGSIZE); // skip comment lines
3200 InputFile.getline(buffer, MAXSTRINGSIZE);
3201 while(!InputFile.eof()) {
3202 InputFile.getline(buffer, MAXSTRINGSIZE);
3203 if (strlen(buffer) > 2) {
3204 //*out << Verbose(2) << "Scanning: " << buffer << endl;
3205 stringstream line(buffer);
3206 line >> FragOrder;
3207 line >> ws >> No;
3208 line >> ws >> Value; // skip time entry
3209 line >> ws >> Value;
3210 No -= 1; // indices start at 1 in file, not 0
3211 //*out << Verbose(2) << " - yields (" << No << "," << Value << ", " << FragOrder << ")" << endl;
3212
3213 // clean the list of those entries that have been superceded by higher order terms already
3214 map<int,KeySet>::iterator marker = IndexKeySetList.find(No); // find keyset to Frag No.
3215 if (marker != IndexKeySetList.end()) { // if found
3216 Value *= 1 + MYEPSILON*(*((*marker).second.begin())); // in case of equal energies this makes em not equal without changing anything actually
3217 // as the smallest number in each set has always been the root (we use global id to keep the doubles away), seek smallest and insert into AtomMask
3218 pair <map<int, pair<double,int> >::iterator, bool> InsertedElement = AdaptiveCriteriaList.insert( make_pair(*((*marker).second.begin()), pair<double,int>( fabs(Value), FragOrder) ));
3219 map<int, pair<double,int> >::iterator PresentItem = InsertedElement.first;
3220 if (!InsertedElement.second) { // this root is already present
3221 if ((*PresentItem).second.second < FragOrder) // if order there is lower, update entry with higher-order term
3222 //if ((*PresentItem).second.first < (*runner).first) // as higher-order terms are not always better, we skip this part (which would always include this site into adaptive increase)
3223 { // if value is smaller, update value and order
3224 (*PresentItem).second.first = fabs(Value);
3225 (*PresentItem).second.second = FragOrder;
3226 *out << Verbose(2) << "Updated element (" << (*PresentItem).first << ",[" << (*PresentItem).second.first << "," << (*PresentItem).second.second << "])." << endl;
3227 } else {
3228 *out << Verbose(2) << "Did not update element " << (*PresentItem).first << " as " << FragOrder << " is less than or equal to " << (*PresentItem).second.second << "." << endl;
3229 }
3230 } else {
3231 *out << Verbose(2) << "Inserted element (" << (*PresentItem).first << ",[" << (*PresentItem).second.first << "," << (*PresentItem).second.second << "])." << endl;
3232 }
3233 } else {
3234 *out << Verbose(1) << "No Fragment under No. " << No << "found." << endl;
3235 }
3236 }
3237 }
3238 // then map back onto (Value, (Root Nr., Order)) (i.e. sorted by value to pick the highest ones)
3239 map<double, pair<int,int> > FinalRootCandidates;
3240 *out << Verbose(1) << "Root candidate list is: " << endl;
3241 for(map<int, pair<double,int> >::iterator runner = AdaptiveCriteriaList.begin(); runner != AdaptiveCriteriaList.end(); runner++) {
3242 Walker = FindAtom((*runner).first);
3243 if (Walker != NULL) {
3244 //if ((*runner).second.second >= Walker->AdaptiveOrder) { // only insert if this is an "active" root site for the current order
3245 if (!Walker->MaxOrder) {
3246 *out << Verbose(2) << "(" << (*runner).first << ",[" << (*runner).second.first << "," << (*runner).second.second << "])" << endl;
3247 FinalRootCandidates.insert( make_pair( (*runner).second.first, pair<int,int>((*runner).first, (*runner).second.second) ) );
3248 } else {
3249 *out << Verbose(2) << "Excluding (" << *Walker << ", " << (*runner).first << ",[" << (*runner).second.first << "," << (*runner).second.second << "]), as it has reached its maximum order." << endl;
3250 }
3251 } else {
3252 cerr << "Atom No. " << (*runner).second.first << " was not found in this molecule." << endl;
3253 }
3254 }
3255 // pick the ones still below threshold and mark as to be adaptively updated
3256 for(map<double, pair<int,int> >::iterator runner = FinalRootCandidates.upper_bound(pow(10.,Order)); runner != FinalRootCandidates.end(); runner++) {
3257 No = (*runner).second.first;
3258 Walker = FindAtom(No);
3259 //if (Walker->AdaptiveOrder < MinimumRingSize[Walker->nr]) {
3260 *out << Verbose(2) << "Root " << No << " is still above threshold (10^{" << Order <<"}: " << runner->first << ", setting entry " << No << " of Atom mask to true." << endl;
3261 AtomMask[No] = true;
3262 status = true;
3263 //} else
3264 //*out << Verbose(2) << "Root " << No << " is still above threshold (10^{" << Order <<"}: " << runner->first << ", however MinimumRingSize of " << MinimumRingSize[Walker->nr] << " does not allow further adaptive increase." << endl;
3265 }
3266 // close and done
3267 InputFile.close();
3268 InputFile.clear();
3269 } else {
3270 cerr << "Unable to parse " << buffer << " file, incrementing all." << endl;
3271 while (Walker->next != end) {
3272 Walker = Walker->next;
3273 #ifdef ADDHYDROGEN
3274 if (Walker->type->Z != 1) // skip hydrogen
3275 #endif
3276 {
3277 AtomMask[Walker->nr] = true; // include all (non-hydrogen) atoms
3278 status = true;
3279 }
3280 }
3281 }
3282 Free((void **)&buffer, "molecule::CheckOrderAtSite: *buffer");
3283 // pick a given number of highest values and set AtomMask
3284 } else { // global increase of Bond Order
3285 while (Walker->next != end) {
3286 Walker = Walker->next;
3287 #ifdef ADDHYDROGEN
3288 if (Walker->type->Z != 1) // skip hydrogen
3289 #endif
3290 {
3291 AtomMask[Walker->nr] = true; // include all (non-hydrogen) atoms
3292 if ((Order != 0) && (Walker->AdaptiveOrder < Order)) // && (Walker->AdaptiveOrder < MinimumRingSize[Walker->nr]))
3293 status = true;
3294 }
3295 }
3296 if ((Order == 0) && (AtomMask[AtomCount] == false)) // single stepping, just check
3297 status = true;
3298
3299 if (!status) {
3300 if (Order == 0)
3301 *out << Verbose(1) << "Single stepping done." << endl;
3302 else
3303 *out << Verbose(1) << "Order at every site is already equal or above desired order " << Order << "." << endl;
3304 }
3305 }
3306
3307 // print atom mask for debugging
3308 *out << " ";
3309 for(int i=0;i<AtomCount;i++)
3310 *out << (i % 10);
3311 *out << endl << "Atom mask is: ";
3312 for(int i=0;i<AtomCount;i++)
3313 *out << (AtomMask[i] ? "t" : "f");
3314 *out << endl;
3315
3316 return status;
3317};
3318
3319/** Create a SortIndex to map from atomic labels to the sequence in which the atoms are given in the config file.
3320 * \param *out output stream for debugging
3321 * \param *&SortIndex Mapping array of size molecule::AtomCount
3322 * \return true - success, false - failure of SortIndex alloc
3323 */
3324bool molecule::CreateMappingLabelsToConfigSequence(ofstream *out, int *&SortIndex)
3325{
3326 element *runner = elemente->start;
3327 int AtomNo = 0;
3328 atom *Walker = NULL;
3329
3330 if (SortIndex != NULL) {
3331 *out << Verbose(1) << "SortIndex is " << SortIndex << " and not NULL as expected." << endl;
3332 return false;
3333 }
3334 SortIndex = (int *) Malloc(sizeof(int)*AtomCount, "molecule::FragmentMolecule: *SortIndex");
3335 for(int i=AtomCount;i--;)
3336 SortIndex[i] = -1;
3337 while (runner->next != elemente->end) { // go through every element
3338 runner = runner->next;
3339 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
3340 Walker = start;
3341 while (Walker->next != end) { // go through every atom of this element
3342 Walker = Walker->next;
3343 if (Walker->type->Z == runner->Z) // if this atom fits to element
3344 SortIndex[Walker->nr] = AtomNo++;
3345 }
3346 }
3347 }
3348 return true;
3349};
3350
3351/** Performs a many-body bond order analysis for a given bond order.
3352 * -# parses adjacency, keysets and orderatsite files
3353 * -# performs DFS to find connected subgraphs (to leave this in was a design decision: might be useful later)
3354 * -# RootStack is created for every subgraph (here, later we implement the "update 10 sites with highest energ
3355y contribution", and that's why this consciously not done in the following loop)
3356 * -# in a loop over all subgraphs
3357 * -# calls FragmentBOSSANOVA with this RootStack and within the subgraph molecule structure
3358 * -# creates molecule (fragment)s from the returned keysets (StoreFragmentFromKeySet)
3359 * -# combines the generated molecule lists from all subgraphs
3360 * -# saves to disk: fragment configs, adjacency, orderatsite, keyset files
3361 * Note that as we split "this" molecule up into a list of subgraphs, i.e. a MoleculeListClass, we have two sets
3362 * of vertex indices: Global always means the index in "this" molecule, whereas local refers to the molecule or
3363 * subgraph in the MoleculeListClass.
3364 * \param *out output stream for debugging
3365 * \param Order up to how many neighbouring bonds a fragment contains in BondOrderScheme::BottumUp scheme
3366 * \param *configuration configuration for writing config files for each fragment
3367 * \return 1 - continue, 2 - stop (no fragmentation occured)
3368 */
3369int molecule::FragmentMolecule(ofstream *out, int Order, config *configuration)
3370{
3371 MoleculeListClass *BondFragments = NULL;
3372 int *SortIndex = NULL;
3373 int *MinimumRingSize = new int[AtomCount];
3374 int FragmentCounter;
3375 MoleculeLeafClass *MolecularWalker = NULL;
3376 MoleculeLeafClass *Subgraphs = NULL; // list of subgraphs from DFS analysis
3377 fstream File;
3378 bool FragmentationToDo = true;
3379 class StackClass<bond *> *BackEdgeStack = NULL, *LocalBackEdgeStack = NULL;
3380 bool CheckOrder = false;
3381 Graph **FragmentList = NULL;
3382 Graph *ParsedFragmentList = NULL;
3383 Graph TotalGraph; // graph with all keysets however local numbers
3384 int TotalNumberOfKeySets = 0;
3385 atom **ListOfAtoms = NULL;
3386 atom ***ListOfLocalAtoms = NULL;
3387 bool *AtomMask = NULL;
3388
3389 *out << endl;
3390#ifdef ADDHYDROGEN
3391 *out << Verbose(0) << "I will treat hydrogen special and saturate dangling bonds with it." << endl;
3392#else
3393 *out << Verbose(0) << "Hydrogen is treated just like the rest of the lot." << endl;
3394#endif
3395
3396 // ++++++++++++++++++++++++++++ INITIAL STUFF: Bond structure analysis, file parsing, ... ++++++++++++++++++++++++++++++++++++++++++
3397
3398 // ===== 1. Check whether bond structure is same as stored in files ====
3399
3400 // fill the adjacency list
3401 CreateListOfBondsPerAtom(out);
3402
3403 // create lookup table for Atom::nr
3404 FragmentationToDo = FragmentationToDo && CreateFatherLookupTable(out, start, end, ListOfAtoms, AtomCount);
3405
3406 // === compare it with adjacency file ===
3407 FragmentationToDo = FragmentationToDo && CheckAdjacencyFileAgainstMolecule(out, configuration->configpath, ListOfAtoms);
3408 Free((void **)&ListOfAtoms, "molecule::FragmentMolecule - **ListOfAtoms");
3409
3410 // ===== 2. perform a DFS analysis to gather info on cyclic structure and a list of disconnected subgraphs =====
3411 Subgraphs = DepthFirstSearchAnalysis(out, BackEdgeStack);
3412 // fill the bond structure of the individually stored subgraphs
3413 Subgraphs->next->FillBondStructureFromReference(out, this, (FragmentCounter = 0), ListOfLocalAtoms, false); // we want to keep the created ListOfLocalAtoms
3414 // analysis of the cycles (print rings, get minimum cycle length) for each subgraph
3415 for(int i=AtomCount;i--;)
3416 MinimumRingSize[i] = AtomCount;
3417 MolecularWalker = Subgraphs;
3418 FragmentCounter = 0;
3419 while (MolecularWalker->next != NULL) {
3420 MolecularWalker = MolecularWalker->next;
3421 *out << Verbose(0) << "Analysing the cycles of subgraph " << MolecularWalker->Leaf << " with nr. " << FragmentCounter << "." << endl;
3422 LocalBackEdgeStack = new StackClass<bond *> (MolecularWalker->Leaf->BondCount);
3423// // check the list of local atoms for debugging
3424// *out << Verbose(0) << "ListOfLocalAtoms for this subgraph is:" << endl;
3425// for (int i=0;i<AtomCount;i++)
3426// if (ListOfLocalAtoms[FragmentCounter][i] == NULL)
3427// *out << "\tNULL";
3428// else
3429// *out << "\t" << ListOfLocalAtoms[FragmentCounter][i]->Name;
3430 MolecularWalker->Leaf->PickLocalBackEdges(out, ListOfLocalAtoms[FragmentCounter++], BackEdgeStack, LocalBackEdgeStack);
3431 MolecularWalker->Leaf->CyclicStructureAnalysis(out, LocalBackEdgeStack, MinimumRingSize);
3432 delete(LocalBackEdgeStack);
3433 }
3434
3435 // ===== 3. if structure still valid, parse key set file and others =====
3436 FragmentationToDo = FragmentationToDo && ParseKeySetFile(out, configuration->configpath, ParsedFragmentList);
3437
3438 // ===== 4. check globally whether there's something to do actually (first adaptivity check)
3439 FragmentationToDo = FragmentationToDo && ParseOrderAtSiteFromFile(out, configuration->configpath);
3440
3441 // =================================== Begin of FRAGMENTATION ===============================
3442 // ===== 6a. assign each keyset to its respective subgraph =====
3443 Subgraphs->next->AssignKeySetsToFragment(out, this, ParsedFragmentList, ListOfLocalAtoms, FragmentList, (FragmentCounter = 0), true);
3444
3445 // ===== 6b. prepare and go into the adaptive (Order<0), single-step (Order==0) or incremental (Order>0) cycle
3446 KeyStack *RootStack = new KeyStack[Subgraphs->next->Count()];
3447 AtomMask = new bool[AtomCount+1];
3448 AtomMask[AtomCount] = false;
3449 FragmentationToDo = false; // if CheckOrderAtSite just ones recommends fragmentation, we will save fragments afterwards
3450 while ((CheckOrder = CheckOrderAtSite(out, AtomMask, ParsedFragmentList, Order, MinimumRingSize, configuration->configpath))) {
3451 FragmentationToDo = FragmentationToDo || CheckOrder;
3452 AtomMask[AtomCount] = true; // last plus one entry is used as marker that we have been through this loop once already in CheckOrderAtSite()
3453 // ===== 6b. fill RootStack for each subgraph (second adaptivity check) =====
3454 Subgraphs->next->FillRootStackForSubgraphs(out, RootStack, AtomMask, (FragmentCounter = 0));
3455
3456 // ===== 7. fill the bond fragment list =====
3457 FragmentCounter = 0;
3458 MolecularWalker = Subgraphs;
3459 while (MolecularWalker->next != NULL) {
3460 MolecularWalker = MolecularWalker->next;
3461 *out << Verbose(1) << "Fragmenting subgraph " << MolecularWalker << "." << endl;
3462 // output ListOfBondsPerAtom for debugging
3463 MolecularWalker->Leaf->OutputListOfBonds(out);
3464 if (MolecularWalker->Leaf->first->next != MolecularWalker->Leaf->last) {
3465
3466 // call BOSSANOVA method
3467 *out << Verbose(0) << endl << " ========== BOND ENERGY of subgraph " << FragmentCounter << " ========================= " << endl;
3468 MolecularWalker->Leaf->FragmentBOSSANOVA(out, FragmentList[FragmentCounter], RootStack[FragmentCounter], MinimumRingSize);
3469 } else {
3470 cerr << "Subgraph " << MolecularWalker << " has no atoms!" << endl;
3471 }
3472 FragmentCounter++; // next fragment list
3473 }
3474 }
3475 delete[](RootStack);
3476 delete[](AtomMask);
3477 delete(ParsedFragmentList);
3478 delete[](MinimumRingSize);
3479
3480
3481 // ==================================== End of FRAGMENTATION ============================================
3482
3483 // ===== 8a. translate list into global numbers (i.e. ones that are valid in "this" molecule, not in MolecularWalker->Leaf)
3484 Subgraphs->next->TranslateIndicesToGlobalIDs(out, FragmentList, (FragmentCounter = 0), TotalNumberOfKeySets, TotalGraph);
3485
3486 // free subgraph memory again
3487 FragmentCounter = 0;
3488 if (Subgraphs != NULL) {
3489 while (Subgraphs->next != NULL) {
3490 Subgraphs = Subgraphs->next;
3491 delete(FragmentList[FragmentCounter++]);
3492 delete(Subgraphs->previous);
3493 }
3494 delete(Subgraphs);
3495 }
3496 Free((void **)&FragmentList, "molecule::FragmentMolecule - **FragmentList");
3497
3498 // ===== 8b. gather keyset lists (graphs) from all subgraphs and transform into MoleculeListClass =====
3499 //if (FragmentationToDo) { // we should always store the fragments again as coordination might have changed slightly without changing bond structure
3500 // allocate memory for the pointer array and transmorph graphs into full molecular fragments
3501 BondFragments = new MoleculeListClass(TotalGraph.size(), AtomCount);
3502 int k=0;
3503 for(Graph::iterator runner = TotalGraph.begin(); runner != TotalGraph.end(); runner++) {
3504 KeySet test = (*runner).first;
3505 *out << "Fragment No." << (*runner).second.first << " with TEFactor " << (*runner).second.second << "." << endl;
3506 BondFragments->ListOfMolecules[k] = StoreFragmentFromKeySet(out, test, configuration);
3507 k++;
3508 }
3509 *out << k << "/" << BondFragments->NumberOfMolecules << " fragments generated from the keysets." << endl;
3510
3511 // ===== 9. Save fragments' configuration and keyset files et al to disk ===
3512 if (BondFragments->NumberOfMolecules != 0) {
3513 // create the SortIndex from BFS labels to order in the config file
3514 CreateMappingLabelsToConfigSequence(out, SortIndex);
3515
3516 *out << Verbose(1) << "Writing " << BondFragments->NumberOfMolecules << " possible bond fragmentation configs" << endl;
3517 if (BondFragments->OutputConfigForListOfFragments(out, FRAGMENTPREFIX, configuration, SortIndex, true, true))
3518 *out << Verbose(1) << "All configs written." << endl;
3519 else
3520 *out << Verbose(1) << "Some config writing failed." << endl;
3521
3522 // store force index reference file
3523 BondFragments->StoreForcesFile(out, configuration->configpath, SortIndex);
3524
3525 // store keysets file
3526 StoreKeySetFile(out, TotalGraph, configuration->configpath);
3527
3528 // store Adjacency file
3529 StoreAdjacencyToFile(out, configuration->configpath);
3530
3531 // store Hydrogen saturation correction file
3532 BondFragments->AddHydrogenCorrection(out, configuration->configpath);
3533
3534 // store adaptive orders into file
3535 StoreOrderAtSiteFile(out, configuration->configpath);
3536
3537 // restore orbital and Stop values
3538 CalculateOrbitals(*configuration);
3539
3540 // free memory for bond part
3541 *out << Verbose(1) << "Freeing bond memory" << endl;
3542 delete(FragmentList); // remove bond molecule from memory
3543 Free((void **)&SortIndex, "molecule::FragmentMolecule: *SortIndex");
3544 } else
3545 *out << Verbose(1) << "FragmentList is zero on return, splitting failed." << endl;
3546 //} else
3547 // *out << Verbose(1) << "No fragments to store." << endl;
3548 *out << Verbose(0) << "End of bond fragmentation." << endl;
3549
3550 return ((int)(!FragmentationToDo)+1); // 1 - continue, 2 - stop (no fragmentation occured)
3551};
3552
3553
3554/** Picks from a global stack with all back edges the ones in the fragment.
3555 * \param *out output stream for debugging
3556 * \param **ListOfLocalAtoms array of father atom::nr to local atom::nr (reverse of atom::father)
3557 * \param *ReferenceStack stack with all the back egdes
3558 * \param *LocalStack stack to be filled
3559 * \return true - everything ok, false - ReferenceStack was empty
3560 */
3561bool molecule::PickLocalBackEdges(ofstream *out, atom **ListOfLocalAtoms, class StackClass<bond *> *&ReferenceStack, class StackClass<bond *> *&LocalStack)
3562{
3563 bool status = true;
3564 if (ReferenceStack->IsEmpty()) {
3565 cerr << "ReferenceStack is empty!" << endl;
3566 return false;
3567 }
3568 bond *Binder = ReferenceStack->PopFirst();
3569 bond *FirstBond = Binder; // mark the first bond, so that we don't loop through the stack indefinitely
3570 atom *Walker = NULL, *OtherAtom = NULL;
3571 ReferenceStack->Push(Binder);
3572
3573 do { // go through all bonds and push local ones
3574 Walker = ListOfLocalAtoms[Binder->leftatom->nr]; // get one atom in the reference molecule
3575 if (Walker == NULL) // if this Walker exists in the subgraph ...
3576 continue;
3577 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through the local list of bonds
3578 OtherAtom = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
3579 if (OtherAtom == ListOfLocalAtoms[Binder->rightatom->nr]) { // found the bond
3580 LocalStack->Push(ListOfBondsPerAtom[Walker->nr][i]);
3581 break;
3582 }
3583 }
3584 Binder = ReferenceStack->PopFirst(); // loop the stack for next item
3585 ReferenceStack->Push(Binder);
3586 } while (FirstBond != Binder);
3587
3588 return status;
3589};
3590
3591/** Stores pairs (Atom::nr, Atom::AdaptiveOrder) into file.
3592 * Atoms not present in the file get "-1".
3593 * \param *out output stream for debugging
3594 * \param *path path to file ORDERATSITEFILE
3595 * \return true - file writable, false - not writable
3596 */
3597bool molecule::StoreOrderAtSiteFile(ofstream *out, char *path)
3598{
3599 stringstream line;
3600 ofstream file;
3601
3602 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
3603 file.open(line.str().c_str());
3604 *out << Verbose(1) << "Writing OrderAtSite " << ORDERATSITEFILE << " ... " << endl;
3605 if (file != NULL) {
3606 atom *Walker = start;
3607 while (Walker->next != end) {
3608 Walker = Walker->next;
3609 file << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "\t" << (int)Walker->MaxOrder << endl;
3610 *out << Verbose(2) << "Storing: " << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "\t" << (int)Walker->MaxOrder << "." << endl;
3611 }
3612 file.close();
3613 *out << Verbose(1) << "done." << endl;
3614 return true;
3615 } else {
3616 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
3617 return false;
3618 }
3619};
3620
3621/** Parses pairs(Atom::nr, Atom::AdaptiveOrder) from file and stores in molecule's Atom's.
3622 * Atoms not present in the file get "0".
3623 * \param *out output stream for debugging
3624 * \param *path path to file ORDERATSITEFILEe
3625 * \return true - file found and scanned, false - file not found
3626 * \sa ParseKeySetFile() and CheckAdjacencyFileAgainstMolecule() as this is meant to be used in conjunction with the two
3627 */
3628bool molecule::ParseOrderAtSiteFromFile(ofstream *out, char *path)
3629{
3630 unsigned char *OrderArray = (unsigned char *) Malloc(sizeof(unsigned char)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
3631 bool *MaxArray = (bool *) Malloc(sizeof(bool)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *MaxArray");
3632 bool status;
3633 int AtomNr, value;
3634 stringstream line;
3635 ifstream file;
3636
3637 *out << Verbose(1) << "Begin of ParseOrderAtSiteFromFile" << endl;
3638 for(int i=AtomCount;i--;)
3639 OrderArray[i] = 0;
3640 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
3641 file.open(line.str().c_str());
3642 if (file != NULL) {
3643 for (int i=AtomCount;i--;) { // initialise with 0
3644 OrderArray[i] = 0;
3645 MaxArray[i] = 0;
3646 }
3647 while (!file.eof()) { // parse from file
3648 AtomNr = -1;
3649 file >> AtomNr;
3650 if (AtomNr != -1) { // test whether we really parsed something (this is necessary, otherwise last atom is set twice and to 0 on second time)
3651 file >> value;
3652 OrderArray[AtomNr] = value;
3653 file >> value;
3654 MaxArray[AtomNr] = value;
3655 //*out << Verbose(2) << "AtomNr " << AtomNr << " with order " << (int)OrderArray[AtomNr] << " and max order set to " << (int)MaxArray[AtomNr] << "." << endl;
3656 }
3657 }
3658 atom *Walker = start;
3659 while (Walker->next != end) { // fill into atom classes
3660 Walker = Walker->next;
3661 Walker->AdaptiveOrder = OrderArray[Walker->nr];
3662 Walker->MaxOrder = MaxArray[Walker->nr];
3663 *out << Verbose(2) << *Walker << " gets order " << (int)Walker->AdaptiveOrder << " and is " << (!Walker->MaxOrder ? "not " : " ") << "maxed." << endl;
3664 }
3665 file.close();
3666 *out << Verbose(1) << "done." << endl;
3667 status = true;
3668 } else {
3669 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
3670 status = false;
3671 }
3672 Free((void **)&OrderArray, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
3673 Free((void **)&MaxArray, "molecule::ParseOrderAtSiteFromFile - *MaxArray");
3674
3675 *out << Verbose(1) << "End of ParseOrderAtSiteFromFile" << endl;
3676 return status;
3677};
3678
3679/** Creates an 2d array of pointer with an entry for each atom and each bond it has.
3680 * Updates molecule::ListOfBondsPerAtom, molecule::NumberOfBondsPerAtom by parsing through
3681 * bond chain list, using molecule::AtomCount and molecule::BondCount.
3682 * Allocates memory, fills the array and exits
3683 * \param *out output stream for debugging
3684 */
3685void molecule::CreateListOfBondsPerAtom(ofstream *out)
3686{
3687 bond *Binder = NULL;
3688 atom *Walker = NULL;
3689 int TotalDegree;
3690 *out << Verbose(1) << "Begin of Creating ListOfBondsPerAtom: AtomCount = " << AtomCount << "\tBondCount = " << BondCount << "\tNoNonBonds = " << NoNonBonds << "." << endl;
3691
3692 // re-allocate memory
3693 *out << Verbose(2) << "(Re-)Allocating memory." << endl;
3694 if (ListOfBondsPerAtom != NULL) {
3695 for(int i=AtomCount;i--;)
3696 Free((void **)&ListOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom[i]");
3697 Free((void **)&ListOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom");
3698 }
3699 if (NumberOfBondsPerAtom != NULL)
3700 Free((void **)&NumberOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: NumberOfBondsPerAtom");
3701 ListOfBondsPerAtom = (bond ***) Malloc(sizeof(bond **)*AtomCount, "molecule::CreateListOfBondsPerAtom: ***ListOfBondsPerAtom");
3702 NumberOfBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfBondsPerAtom: *NumberOfBondsPerAtom");
3703
3704 // reset bond counts per atom
3705 for(int i=AtomCount;i--;)
3706 NumberOfBondsPerAtom[i] = 0;
3707 // count bonds per atom
3708 Binder = first;
3709 while (Binder->next != last) {
3710 Binder = Binder->next;
3711 NumberOfBondsPerAtom[Binder->leftatom->nr]++;
3712 NumberOfBondsPerAtom[Binder->rightatom->nr]++;
3713 }
3714 for(int i=AtomCount;i--;) {
3715 // allocate list of bonds per atom
3716 ListOfBondsPerAtom[i] = (bond **) Malloc(sizeof(bond *)*NumberOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: **ListOfBondsPerAtom[]");
3717 // clear the list again, now each NumberOfBondsPerAtom marks current free field
3718 NumberOfBondsPerAtom[i] = 0;
3719 }
3720 // fill the list
3721 Binder = first;
3722 while (Binder->next != last) {
3723 Binder = Binder->next;
3724 ListOfBondsPerAtom[Binder->leftatom->nr][NumberOfBondsPerAtom[Binder->leftatom->nr]++] = Binder;
3725 ListOfBondsPerAtom[Binder->rightatom->nr][NumberOfBondsPerAtom[Binder->rightatom->nr]++] = Binder;
3726 }
3727
3728 // output list for debugging
3729 *out << Verbose(3) << "ListOfBondsPerAtom for each atom:" << endl;
3730 Walker = start;
3731 while (Walker->next != end) {
3732 Walker = Walker->next;
3733 *out << Verbose(4) << "Atom " << Walker->Name << "/" << Walker->nr << " with " << NumberOfBondsPerAtom[Walker->nr] << " bonds: ";
3734 TotalDegree = 0;
3735 for (int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
3736 *out << *ListOfBondsPerAtom[Walker->nr][j] << "\t";
3737 TotalDegree += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
3738 }
3739 *out << " -- TotalDegree: " << TotalDegree << endl;
3740 }
3741 *out << Verbose(1) << "End of Creating ListOfBondsPerAtom." << endl << endl;
3742};
3743
3744/** Adds atoms up to \a BondCount distance from \a *Root and notes them down in \a **AddedAtomList.
3745 * Gray vertices are always enqueued in an StackClass<atom *> FIFO queue, the rest is usual BFS with adding vertices found was
3746 * white and putting into queue.
3747 * \param *out output stream for debugging
3748 * \param *Mol Molecule class to add atoms to
3749 * \param **AddedAtomList list with added atom pointers, index is atom father's number
3750 * \param **AddedBondList list with added bond pointers, index is bond father's number
3751 * \param *Root root vertex for BFS
3752 * \param *Bond bond not to look beyond
3753 * \param BondOrder maximum distance for vertices to add
3754 * \param IsAngstroem lengths are in angstroem or bohrradii
3755 */
3756void molecule::BreadthFirstSearchAdd(ofstream *out, molecule *Mol, atom **&AddedAtomList, bond **&AddedBondList, atom *Root, bond *Bond, int BondOrder, bool IsAngstroem)
3757{
3758 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::BreadthFirstSearchAdd: **PredecessorList");
3759 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::BreadthFirstSearchAdd: *ShortestPathList");
3760 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::BreadthFirstSearchAdd: *ColorList");
3761 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
3762 atom *Walker = NULL, *OtherAtom = NULL;
3763 bond *Binder = NULL;
3764
3765 // add Root if not done yet
3766 AtomStack->ClearStack();
3767 if (AddedAtomList[Root->nr] == NULL) // add Root if not yet present
3768 AddedAtomList[Root->nr] = Mol->AddCopyAtom(Root);
3769 AtomStack->Push(Root);
3770
3771 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
3772 for (int i=AtomCount;i--;) {
3773 PredecessorList[i] = NULL;
3774 ShortestPathList[i] = -1;
3775 if (AddedAtomList[i] != NULL) // mark already present atoms (i.e. Root and maybe others) as visited
3776 ColorList[i] = lightgray;
3777 else
3778 ColorList[i] = white;
3779 }
3780 ShortestPathList[Root->nr] = 0;
3781
3782 // and go on ... Queue always contains all lightgray vertices
3783 while (!AtomStack->IsEmpty()) {
3784 // we have to pop the oldest atom from stack. This keeps the atoms on the stack always of the same ShortestPath distance.
3785 // e.g. if current atom is 2, push to end of stack are of length 3, but first all of length 2 would be popped. They again
3786 // append length of 3 (their neighbours). Thus on stack we have always atoms of a certain length n at bottom of stack and
3787 // followed by n+1 till top of stack.
3788 Walker = AtomStack->PopFirst(); // pop oldest added
3789 *out << Verbose(1) << "Current Walker is: " << Walker->Name << ", and has " << NumberOfBondsPerAtom[Walker->nr] << " bonds." << endl;
3790 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
3791 Binder = ListOfBondsPerAtom[Walker->nr][i];
3792 if (Binder != NULL) { // don't look at bond equal NULL
3793 OtherAtom = Binder->GetOtherAtom(Walker);
3794 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
3795 if (ColorList[OtherAtom->nr] == white) {
3796 if (Binder != Bond) // let other atom white if it's via Root bond. In case it's cyclic it has to be reached again (yet Root is from OtherAtom already black, thus no problem)
3797 ColorList[OtherAtom->nr] = lightgray;
3798 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
3799 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
3800 *out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " " << ((ColorList[OtherAtom->nr] == white) ? "white" : "lightgray") << ", its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
3801 if ((((ShortestPathList[OtherAtom->nr] < BondOrder) && (Binder != Bond))) ) { // Check for maximum distance
3802 *out << Verbose(3);
3803 if (AddedAtomList[OtherAtom->nr] == NULL) { // add if it's not been so far
3804 AddedAtomList[OtherAtom->nr] = Mol->AddCopyAtom(OtherAtom);
3805 *out << "Added OtherAtom " << OtherAtom->Name;
3806 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3807 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3808 AddedBondList[Binder->nr]->Type = Binder->Type;
3809 *out << " and bond " << *(AddedBondList[Binder->nr]) << ", ";
3810 } else { // this code should actually never come into play (all white atoms are not yet present in BondMolecule, that's why they are white in the first place)
3811 *out << "Not adding OtherAtom " << OtherAtom->Name;
3812 if (AddedBondList[Binder->nr] == NULL) {
3813 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3814 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3815 AddedBondList[Binder->nr]->Type = Binder->Type;
3816 *out << ", added Bond " << *(AddedBondList[Binder->nr]);
3817 } else
3818 *out << ", not added Bond ";
3819 }
3820 *out << ", putting OtherAtom into queue." << endl;
3821 AtomStack->Push(OtherAtom);
3822 } else { // out of bond order, then replace
3823 if ((AddedAtomList[OtherAtom->nr] == NULL) && (Binder->Cyclic))
3824 ColorList[OtherAtom->nr] = white; // unmark if it has not been queued/added, to make it available via its other bonds (cyclic)
3825 if (Binder == Bond)
3826 *out << Verbose(3) << "Not Queueing, is the Root bond";
3827 else if (ShortestPathList[OtherAtom->nr] >= BondOrder)
3828 *out << Verbose(3) << "Not Queueing, is out of Bond Count of " << BondOrder;
3829 if (!Binder->Cyclic)
3830 *out << ", is not part of a cyclic bond, saturating bond with Hydrogen." << endl;
3831 if (AddedBondList[Binder->nr] == NULL) {
3832 if ((AddedAtomList[OtherAtom->nr] != NULL)) { // .. whether we add or saturate
3833 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3834 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3835 AddedBondList[Binder->nr]->Type = Binder->Type;
3836 } else {
3837#ifdef ADDHYDROGEN
3838 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
3839#endif
3840 }
3841 }
3842 }
3843 } else {
3844 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
3845 // This has to be a cyclic bond, check whether it's present ...
3846 if (AddedBondList[Binder->nr] == NULL) {
3847 if ((Binder != Bond) && (Binder->Cyclic) && (((ShortestPathList[Walker->nr]+1) < BondOrder))) {
3848 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3849 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3850 AddedBondList[Binder->nr]->Type = Binder->Type;
3851 } else { // if it's root bond it has to broken (otherwise we would not create the fragments)
3852#ifdef ADDHYDROGEN
3853 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
3854#endif
3855 }
3856 }
3857 }
3858 }
3859 }
3860 ColorList[Walker->nr] = black;
3861 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
3862 }
3863 Free((void **)&PredecessorList, "molecule::BreadthFirstSearchAdd: **PredecessorList");
3864 Free((void **)&ShortestPathList, "molecule::BreadthFirstSearchAdd: **ShortestPathList");
3865 Free((void **)&ColorList, "molecule::BreadthFirstSearchAdd: **ColorList");
3866 delete(AtomStack);
3867};
3868
3869/** Adds bond structure to this molecule from \a Father molecule.
3870 * This basically causes this molecule to become an induced subgraph of the \a Father, i.e. for every bond in Father
3871 * with end points present in this molecule, bond is created in this molecule.
3872 * Special care was taken to ensure that this is of complexity O(N), where N is the \a Father's molecule::AtomCount.
3873 * \param *out output stream for debugging
3874 * \param *Father father molecule
3875 * \return true - is induced subgraph, false - there are atoms with fathers not in \a Father
3876 * \todo not checked, not fully working probably
3877 */
3878bool molecule::BuildInducedSubgraph(ofstream *out, const molecule *Father)
3879{
3880 atom *Walker = NULL, *OtherAtom = NULL;
3881 bool status = true;
3882 atom **ParentList = (atom **) Malloc(sizeof(atom *)*Father->AtomCount, "molecule::BuildInducedSubgraph: **ParentList");
3883
3884 *out << Verbose(2) << "Begin of BuildInducedSubgraph." << endl;
3885
3886 // reset parent list
3887 *out << Verbose(3) << "Resetting ParentList." << endl;
3888 for (int i=Father->AtomCount;i--;)
3889 ParentList[i] = NULL;
3890
3891 // fill parent list with sons
3892 *out << Verbose(3) << "Filling Parent List." << endl;
3893 Walker = start;
3894 while (Walker->next != end) {
3895 Walker = Walker->next;
3896 ParentList[Walker->father->nr] = Walker;
3897 // Outputting List for debugging
3898 *out << Verbose(4) << "Son["<< Walker->father->nr <<"] of " << Walker->father << " is " << ParentList[Walker->father->nr] << "." << endl;
3899 }
3900
3901 // check each entry of parent list and if ok (one-to-and-onto matching) create bonds
3902 *out << Verbose(3) << "Creating bonds." << endl;
3903 Walker = Father->start;
3904 while (Walker->next != Father->end) {
3905 Walker = Walker->next;
3906 if (ParentList[Walker->nr] != NULL) {
3907 if (ParentList[Walker->nr]->father != Walker) {
3908 status = false;
3909 } else {
3910 for (int i=0;i<Father->NumberOfBondsPerAtom[Walker->nr];i++) {
3911 OtherAtom = Father->ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
3912 if (ParentList[OtherAtom->nr] != NULL) { // if otheratom is also a father of an atom on this molecule, create the bond
3913 *out << Verbose(4) << "Endpoints of Bond " << Father->ListOfBondsPerAtom[Walker->nr][i] << " are both present: " << ParentList[Walker->nr]->Name << " and " << ParentList[OtherAtom->nr]->Name << "." << endl;
3914 AddBond(ParentList[Walker->nr], ParentList[OtherAtom->nr], Father->ListOfBondsPerAtom[Walker->nr][i]->BondDegree);
3915 }
3916 }
3917 }
3918 }
3919 }
3920
3921 Free((void **)&ParentList, "molecule::BuildInducedSubgraph: **ParentList");
3922 *out << Verbose(2) << "End of BuildInducedSubgraph." << endl;
3923 return status;
3924};
3925
3926
3927/** Looks through a StackClass<atom *> and returns the likeliest removal candiate.
3928 * \param *out output stream for debugging messages
3929 * \param *&Leaf KeySet to look through
3930 * \param *&ShortestPathList list of the shortest path to decide which atom to suggest as removal candidate in the end
3931 * \param index of the atom suggested for removal
3932 */
3933int molecule::LookForRemovalCandidate(ofstream *&out, KeySet *&Leaf, int *&ShortestPathList)
3934{
3935 atom *Runner = NULL;
3936 int SP, Removal;
3937
3938 *out << Verbose(2) << "Looking for removal candidate." << endl;
3939 SP = -1; //0; // not -1, so that Root is never removed
3940 Removal = -1;
3941 for (KeySet::iterator runner = Leaf->begin(); runner != Leaf->end(); runner++) {
3942 Runner = FindAtom((*runner));
3943 if (Runner->type->Z != 1) { // skip all those added hydrogens when re-filling snake stack
3944 if (ShortestPathList[(*runner)] > SP) { // remove the oldest one with longest shortest path
3945 SP = ShortestPathList[(*runner)];
3946 Removal = (*runner);
3947 }
3948 }
3949 }
3950 return Removal;
3951};
3952
3953/** Stores a fragment from \a KeySet into \a molecule.
3954 * First creates the minimal set of atoms from the KeySet, then creates the bond structure from the complete
3955 * molecule and adds missing hydrogen where bonds were cut.
3956 * \param *out output stream for debugging messages
3957 * \param &Leaflet pointer to KeySet structure
3958 * \param IsAngstroem whether we have Ansgtroem or bohrradius
3959 * \return pointer to constructed molecule
3960 */
3961molecule * molecule::StoreFragmentFromKeySet(ofstream *out, KeySet &Leaflet, bool IsAngstroem)
3962{
3963 atom *Runner = NULL, *FatherOfRunner = NULL, *OtherFather = NULL;
3964 atom **SonList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::StoreFragmentFromStack: **SonList");
3965 molecule *Leaf = new molecule(elemente);
3966 bool LonelyFlag = false;
3967 int size;
3968
3969// *out << Verbose(1) << "Begin of StoreFragmentFromKeyset." << endl;
3970
3971 Leaf->BondDistance = BondDistance;
3972 for(int i=NDIM*2;i--;)
3973 Leaf->cell_size[i] = cell_size[i];
3974
3975 // initialise SonList (indicates when we need to replace a bond with hydrogen instead)
3976 for(int i=AtomCount;i--;)
3977 SonList[i] = NULL;
3978
3979 // first create the minimal set of atoms from the KeySet
3980 size = 0;
3981 for(KeySet::iterator runner = Leaflet.begin(); runner != Leaflet.end(); runner++) {
3982 FatherOfRunner = FindAtom((*runner)); // find the id
3983 SonList[FatherOfRunner->nr] = Leaf->AddCopyAtom(FatherOfRunner);
3984 size++;
3985 }
3986
3987 // create the bonds between all: Make it an induced subgraph and add hydrogen
3988// *out << Verbose(2) << "Creating bonds from father graph (i.e. induced subgraph creation)." << endl;
3989 Runner = Leaf->start;
3990 while (Runner->next != Leaf->end) {
3991 Runner = Runner->next;
3992 LonelyFlag = true;
3993 FatherOfRunner = Runner->father;
3994 if (SonList[FatherOfRunner->nr] != NULL) { // check if this, our father, is present in list
3995 // create all bonds
3996 for (int i=0;i<NumberOfBondsPerAtom[FatherOfRunner->nr];i++) { // go through every bond of father
3997 OtherFather = ListOfBondsPerAtom[FatherOfRunner->nr][i]->GetOtherAtom(FatherOfRunner);
3998// *out << Verbose(2) << "Father " << *FatherOfRunner << " of son " << *SonList[FatherOfRunner->nr] << " is bound to " << *OtherFather;
3999 if (SonList[OtherFather->nr] != NULL) {
4000// *out << ", whose son is " << *SonList[OtherFather->nr] << "." << endl;
4001 if (OtherFather->nr > FatherOfRunner->nr) { // add bond (nr check is for adding only one of both variants: ab, ba)
4002// *out << Verbose(3) << "Adding Bond: ";
4003// *out <<
4004 Leaf->AddBond(Runner, SonList[OtherFather->nr], ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree);
4005// *out << "." << endl;
4006 //NumBonds[Runner->nr]++;
4007 } else {
4008// *out << Verbose(3) << "Not adding bond, labels in wrong order." << endl;
4009 }
4010 LonelyFlag = false;
4011 } else {
4012// *out << ", who has no son in this fragment molecule." << endl;
4013#ifdef ADDHYDROGEN
4014 //*out << Verbose(3) << "Adding Hydrogen to " << Runner->Name << " and a bond in between." << endl;
4015 Leaf->AddHydrogenReplacementAtom(out, ListOfBondsPerAtom[FatherOfRunner->nr][i], Runner, FatherOfRunner, OtherFather, ListOfBondsPerAtom[FatherOfRunner->nr],NumberOfBondsPerAtom[FatherOfRunner->nr], IsAngstroem);
4016#endif
4017 //NumBonds[Runner->nr] += ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree;
4018 }
4019 }
4020 } else {
4021 *out << Verbose(0) << "ERROR: Son " << Runner->Name << " has father " << FatherOfRunner->Name << " but its entry in SonList is " << SonList[FatherOfRunner->nr] << "!" << endl;
4022 }
4023 if ((LonelyFlag) && (size > 1)) {
4024 *out << Verbose(0) << *Runner << "has got bonds only to hydrogens!" << endl;
4025 }
4026#ifdef ADDHYDROGEN
4027 while ((Runner->next != Leaf->end) && (Runner->next->type->Z == 1)) // skip added hydrogen
4028 Runner = Runner->next;
4029#endif
4030 }
4031 Leaf->CreateListOfBondsPerAtom(out);
4032 //Leaflet->Leaf->ScanForPeriodicCorrection(out);
4033 Free((void **)&SonList, "molecule::StoreFragmentFromStack: **SonList");
4034// *out << Verbose(1) << "End of StoreFragmentFromKeyset." << endl;
4035 return Leaf;
4036};
4037
4038/** Creates \a MoleculeListClass of all unique fragments of the \a molecule containing \a Order atoms or vertices.
4039 * The picture to have in mind is that of a DFS "snake" of a certain length \a Order, i.e. as in the infamous
4040 * computer game, that winds through the connected graph representing the molecule. Color (white,
4041 * lightgray, darkgray, black) indicates whether a vertex has been discovered so far or not. Labels will help in
4042 * creating only unique fragments and not additional ones with vertices simply in different sequence.
4043 * The Predecessor is always the one that came before in discovering, needed on backstepping. And
4044 * finally, the ShortestPath is needed for removing vertices from the snake stack during the back-
4045 * stepping.
4046 * \param *out output stream for debugging
4047 * \param Order number of atoms in each fragment
4048 * \param *configuration configuration for writing config files for each fragment
4049 * \return List of all unique fragments with \a Order atoms
4050 */
4051/*
4052MoleculeListClass * molecule::CreateListOfUniqueFragmentsOfOrder(ofstream *out, int Order, config *configuration)
4053{
4054 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
4055 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
4056 int *Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
4057 enum Shading *ColorVertexList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
4058 enum Shading *ColorEdgeList = (enum Shading *) Malloc(sizeof(enum Shading)*BondCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorBondList");
4059 StackClass<atom *> *RootStack = new StackClass<atom *>(AtomCount);
4060 StackClass<atom *> *TouchedStack = new StackClass<atom *>((int)pow(4,Order)+2); // number of atoms reached from one with maximal 4 bonds plus Root itself
4061 StackClass<atom *> *SnakeStack = new StackClass<atom *>(Order+1); // equal to Order is not possible, as then the StackClass<atom *> cannot discern between full and empty stack!
4062 MoleculeLeafClass *Leaflet = NULL, *TempLeaf = NULL;
4063 MoleculeListClass *FragmentList = NULL;
4064 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Removal = NULL;
4065 bond *Binder = NULL;
4066 int RunningIndex = 0, FragmentCounter = 0;
4067
4068 *out << Verbose(1) << "Begin of CreateListOfUniqueFragmentsOfOrder." << endl;
4069
4070 // reset parent list
4071 *out << Verbose(3) << "Resetting labels, parent, predecessor, color and shortest path lists." << endl;
4072 for (int i=0;i<AtomCount;i++) { // reset all atom labels
4073 // initialise each vertex as white with no predecessor, empty queue, color lightgray, not labelled, no sons
4074 Labels[i] = -1;
4075 SonList[i] = NULL;
4076 PredecessorList[i] = NULL;
4077 ColorVertexList[i] = white;
4078 ShortestPathList[i] = -1;
4079 }
4080 for (int i=0;i<BondCount;i++)
4081 ColorEdgeList[i] = white;
4082 RootStack->ClearStack(); // clearstack and push first atom if exists
4083 TouchedStack->ClearStack();
4084 Walker = start->next;
4085 while ((Walker != end)
4086#ifdef ADDHYDROGEN
4087 && (Walker->type->Z == 1)
4088#endif
4089 ) { // search for first non-hydrogen atom
4090 *out << Verbose(4) << "Current Root candidate is " << Walker->Name << "." << endl;
4091 Walker = Walker->next;
4092 }
4093 if (Walker != end)
4094 RootStack->Push(Walker);
4095 else
4096 *out << Verbose(0) << "ERROR: Could not find an appropriate Root atom!" << endl;
4097 *out << Verbose(3) << "Root " << Walker->Name << " is on AtomStack, beginning loop through all vertices ..." << endl;
4098
4099 ///// OUTER LOOP ////////////
4100 while (!RootStack->IsEmpty()) {
4101 // get new root vertex from atom stack
4102 Root = RootStack->PopFirst();
4103 ShortestPathList[Root->nr] = 0;
4104 if (Labels[Root->nr] == -1)
4105 Labels[Root->nr] = RunningIndex++; // prevent it from getting again on AtomStack
4106 PredecessorList[Root->nr] = Root;
4107 TouchedStack->Push(Root);
4108 *out << Verbose(0) << "Root for this loop is: " << Root->Name << ".\n";
4109
4110 // clear snake stack
4111 SnakeStack->ClearStack();
4112 //SnakeStack->TestImplementation(out, start->next);
4113
4114 ///// INNER LOOP ////////////
4115 // Problems:
4116 // - what about cyclic bonds?
4117 Walker = Root;
4118 do {
4119 *out << Verbose(1) << "Current Walker is: " << Walker->Name;
4120 // initial setting of the new Walker: label, color, shortest path and put on stacks
4121 if (Labels[Walker->nr] == -1) { // give atom a unique, monotonely increasing number
4122 Labels[Walker->nr] = RunningIndex++;
4123 RootStack->Push(Walker);
4124 }
4125 *out << ", has label " << Labels[Walker->nr];
4126 if ((ColorVertexList[Walker->nr] == white) || ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white))) { // color it if newly discovered and push on stacks (and if within reach!)
4127 if ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white)) {
4128 // Binder ought to be set still from last neighbour search
4129 *out << ", coloring bond " << *Binder << " black";
4130 ColorEdgeList[Binder->nr] = black; // mark this bond as used
4131 }
4132 if (ShortestPathList[Walker->nr] == -1) {
4133 ShortestPathList[Walker->nr] = ShortestPathList[PredecessorList[Walker->nr]->nr]+1;
4134 TouchedStack->Push(Walker); // mark every atom for lists cleanup later, whose shortest path has been changed
4135 }
4136 if ((ShortestPathList[Walker->nr] < Order) && (ColorVertexList[Walker->nr] != darkgray)) { // if not already on snake stack
4137 SnakeStack->Push(Walker);
4138 ColorVertexList[Walker->nr] = darkgray; // mark as dark gray of on snake stack
4139 }
4140 }
4141 *out << ", SP of " << ShortestPathList[Walker->nr] << " and its color is " << GetColor(ColorVertexList[Walker->nr]) << "." << endl;
4142
4143 // then check the stack for a newly stumbled upon fragment
4144 if (SnakeStack->ItemCount() == Order) { // is stack full?
4145 // store the fragment if it is one and get a removal candidate
4146 Removal = StoreFragmentFromStack(out, Root, Walker, Leaflet, SnakeStack, ShortestPathList, SonList, Labels, &FragmentCounter, configuration);
4147 // remove the candidate if one was found
4148 if (Removal != NULL) {
4149 *out << Verbose(2) << "Removing item " << Removal->Name << " with SP of " << ShortestPathList[Removal->nr] << " from snake stack." << endl;
4150 SnakeStack->RemoveItem(Removal);
4151 ColorVertexList[Removal->nr] = lightgray; // return back to not on snake stack but explored marking
4152 if (Walker == Removal) { // if the current atom is to be removed, we also have to take a step back
4153 Walker = PredecessorList[Removal->nr];
4154 *out << Verbose(2) << "Stepping back to " << Walker->Name << "." << endl;
4155 }
4156 }
4157 } else
4158 Removal = NULL;
4159
4160 // finally, look for a white neighbour as the next Walker
4161 Binder = NULL;
4162 if ((Removal == NULL) || (Walker != PredecessorList[Removal->nr])) { // don't look, if a new walker has been set above
4163 *out << Verbose(2) << "Snake has currently " << SnakeStack->ItemCount() << " item(s)." << endl;
4164 OtherAtom = NULL; // this is actually not needed, every atom has at least one neighbour
4165 if (ShortestPathList[Walker->nr] < Order) {
4166 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
4167 Binder = ListOfBondsPerAtom[Walker->nr][i];
4168 *out << Verbose(2) << "Current bond is " << *Binder << ": ";
4169 OtherAtom = Binder->GetOtherAtom(Walker);
4170 if ((Labels[OtherAtom->nr] != -1) && (Labels[OtherAtom->nr] < Labels[Root->nr])) { // we don't step up to labels bigger than us
4171 *out << "Label " << Labels[OtherAtom->nr] << " is smaller than Root's " << Labels[Root->nr] << "." << endl;
4172 //ColorVertexList[OtherAtom->nr] = lightgray; // mark as explored
4173 } else { // otherwise check its colour and element
4174 if (
4175#ifdef ADDHYDROGEN
4176 (OtherAtom->type->Z != 1) &&
4177#endif
4178 (ColorEdgeList[Binder->nr] == white)) { // skip hydrogen, look for unexplored vertices
4179 *out << "Moving along " << GetColor(ColorEdgeList[Binder->nr]) << " bond " << Binder << " to " << ((ColorVertexList[OtherAtom->nr] == white) ? "unexplored" : "explored") << " item: " << OtherAtom->Name << "." << endl;
4180 // i find it currently rather sensible to always set the predecessor in order to find one's way back
4181 //if (PredecessorList[OtherAtom->nr] == NULL) {
4182 PredecessorList[OtherAtom->nr] = Walker;
4183 *out << Verbose(3) << "Setting Predecessor of " << OtherAtom->Name << " to " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
4184 //} else {
4185 // *out << Verbose(3) << "Predecessor of " << OtherAtom->Name << " is " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
4186 //}
4187 Walker = OtherAtom;
4188 break;
4189 } else {
4190 if (OtherAtom->type->Z == 1)
4191 *out << "Links to a hydrogen atom." << endl;
4192 else
4193 *out << "Bond has not white but " << GetColor(ColorEdgeList[Binder->nr]) << " color." << endl;
4194 }
4195 }
4196 }
4197 } else { // means we have stepped beyond the horizon: Return!
4198 Walker = PredecessorList[Walker->nr];
4199 OtherAtom = Walker;
4200 *out << Verbose(3) << "We have gone too far, stepping back to " << Walker->Name << "." << endl;
4201 }
4202 if (Walker != OtherAtom) { // if no white neighbours anymore, color it black
4203 *out << Verbose(2) << "Coloring " << Walker->Name << " black." << endl;
4204 ColorVertexList[Walker->nr] = black;
4205 Walker = PredecessorList[Walker->nr];
4206 }
4207 }
4208 } while ((Walker != Root) || (ColorVertexList[Root->nr] != black));
4209 *out << Verbose(2) << "Inner Looping is finished." << endl;
4210
4211 // if we reset all AtomCount atoms, we have again technically O(N^2) ...
4212 *out << Verbose(2) << "Resetting lists." << endl;
4213 Walker = NULL;
4214 Binder = NULL;
4215 while (!TouchedStack->IsEmpty()) {
4216 Walker = TouchedStack->PopLast();
4217 *out << Verbose(3) << "Re-initialising entries of " << *Walker << "." << endl;
4218 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
4219 ColorEdgeList[ListOfBondsPerAtom[Walker->nr][i]->nr] = white;
4220 PredecessorList[Walker->nr] = NULL;
4221 ColorVertexList[Walker->nr] = white;
4222 ShortestPathList[Walker->nr] = -1;
4223 }
4224 }
4225 *out << Verbose(1) << "Outer Looping over all vertices is done." << endl;
4226
4227 // copy together
4228 *out << Verbose(1) << "Copying all fragments into MoleculeList structure." << endl;
4229 FragmentList = new MoleculeListClass(FragmentCounter, AtomCount);
4230 RunningIndex = 0;
4231 while ((Leaflet != NULL) && (RunningIndex < FragmentCounter)) {
4232 FragmentList->ListOfMolecules[RunningIndex++] = Leaflet->Leaf;
4233 Leaflet->Leaf = NULL; // prevent molecule from being removed
4234 TempLeaf = Leaflet;
4235 Leaflet = Leaflet->previous;
4236 delete(TempLeaf);
4237 };
4238
4239 // free memory and exit
4240 Free((void **)&PredecessorList, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
4241 Free((void **)&ShortestPathList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
4242 Free((void **)&Labels, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
4243 Free((void **)&ColorVertexList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
4244 delete(RootStack);
4245 delete(TouchedStack);
4246 delete(SnakeStack);
4247
4248 *out << Verbose(1) << "End of CreateListOfUniqueFragmentsOfOrder." << endl;
4249 return FragmentList;
4250};
4251*/
4252
4253/** Structure containing all values in power set combination generation.
4254 */
4255struct UniqueFragments {
4256 config *configuration;
4257 atom *Root;
4258 Graph *Leaflet;
4259 KeySet *FragmentSet;
4260 int ANOVAOrder;
4261 int FragmentCounter;
4262 int CurrentIndex;
4263 double TEFactor;
4264 int *ShortestPathList;
4265 bool **UsedList;
4266 bond **BondsPerSPList;
4267 int *BondsPerSPCount;
4268};
4269
4270/** From a given set of Bond sorted by Shortest Path distance, create all possible fragments of size \a SetDimension.
4271 * -# loops over every possible combination (2^dimension of edge set)
4272 * -# inserts current set, if there's still space left
4273 * -# yes: calls SPFragmentGenerator with structure, created new edge list and size respective to root dist
4274ance+1
4275 * -# no: stores fragment into keyset list by calling InsertFragmentIntoGraph
4276 * -# removes all items added into the snake stack (in UniqueFragments structure) added during level (root
4277distance) and current set
4278 * \param *out output stream for debugging
4279 * \param FragmentSearch UniqueFragments structure with all values needed
4280 * \param RootDistance current shortest path level, whose set of edges is represented by **BondsSet
4281 * \param SetDimension Number of possible bonds on this level (i.e. size of the array BondsSet[])
4282 * \param SubOrder remaining number of allowed vertices to add
4283 */
4284void molecule::SPFragmentGenerator(ofstream *out, struct UniqueFragments *FragmentSearch, int RootDistance, bond **BondsSet, int SetDimension, int SubOrder)
4285{
4286 atom *OtherWalker = NULL;
4287 int verbosity = 0; //FragmentSearch->ANOVAOrder-SubOrder;
4288 int NumCombinations;
4289 bool bit;
4290 int bits, TouchedIndex, SubSetDimension, SP, Added;
4291 int Removal;
4292 int SpaceLeft;
4293 int *TouchedList = (int *) Malloc(sizeof(int)*(SubOrder+1), "molecule::SPFragmentGenerator: *TouchedList");
4294 bond *Binder = NULL;
4295 bond **BondsList = NULL;
4296 KeySetTestPair TestKeySetInsert;
4297
4298 NumCombinations = 1 << SetDimension;
4299
4300 // Hier muessen von 1 bis NumberOfBondsPerAtom[Walker->nr] alle Kombinationen
4301 // von Endstuecken (aus den Bonds) hinzugefᅵᅵgt werden und fᅵᅵr verbleibende ANOVAOrder
4302 // rekursiv GraphCrawler in der nᅵᅵchsten Ebene aufgerufen werden
4303
4304 *out << Verbose(1+verbosity) << "Begin of SPFragmentGenerator." << endl;
4305 *out << Verbose(1+verbosity) << "We are " << RootDistance << " away from Root, which is " << *FragmentSearch->Root << ", SubOrder is " << SubOrder << ", SetDimension is " << SetDimension << " and this means " << NumCombinations-1 << " combination(s)." << endl;
4306
4307 // initialised touched list (stores added atoms on this level)
4308 *out << Verbose(1+verbosity) << "Clearing touched list." << endl;
4309 for (TouchedIndex=SubOrder+1;TouchedIndex--;) // empty touched list
4310 TouchedList[TouchedIndex] = -1;
4311 TouchedIndex = 0;
4312
4313 // create every possible combination of the endpieces
4314 *out << Verbose(1+verbosity) << "Going through all combinations of the power set." << endl;
4315 for (int i=1;i<NumCombinations;i++) { // sweep through all power set combinations (skip empty set!)
4316 // count the set bit of i
4317 bits = 0;
4318 for (int j=SetDimension;j--;)
4319 bits += (i & (1 << j)) >> j;
4320
4321 *out << Verbose(1+verbosity) << "Current set is " << Binary(i | (1 << SetDimension)) << ", number of bits is " << bits << "." << endl;
4322 if (bits <= SubOrder) { // if not greater than additional atoms allowed on stack, continue
4323 // --1-- add this set of the power set of bond partners to the snake stack
4324 Added = 0;
4325 for (int j=0;j<SetDimension;j++) { // pull out every bit by shifting
4326 bit = ((i & (1 << j)) != 0); // mask the bit for the j-th bond
4327 if (bit) { // if bit is set, we add this bond partner
4328 OtherWalker = BondsSet[j]->rightatom; // rightatom is always the one more distant, i.e. the one to add
4329 //*out << Verbose(1+verbosity) << "Current Bond is " << ListOfBondsPerAtom[Walker->nr][i] << ", checking on " << *OtherWalker << "." << endl;
4330 *out << Verbose(2+verbosity) << "Adding " << *OtherWalker << " with nr " << OtherWalker->nr << "." << endl;
4331 TestKeySetInsert = FragmentSearch->FragmentSet->insert(OtherWalker->nr);
4332 if (TestKeySetInsert.second) {
4333 TouchedList[TouchedIndex++] = OtherWalker->nr; // note as added
4334 Added++;
4335 } else {
4336 *out << Verbose(2+verbosity) << "This was item was already present in the keyset." << endl;
4337 }
4338 //FragmentSearch->UsedList[OtherWalker->nr][i] = true;
4339 //}
4340 } else {
4341 *out << Verbose(2+verbosity) << "Not adding." << endl;
4342 }
4343 }
4344
4345 SpaceLeft = SubOrder - Added ;// SubOrder - bits; // due to item's maybe being already present, this does not work anymore
4346 if (SpaceLeft > 0) {
4347 *out << Verbose(1+verbosity) << "There's still some space left on stack: " << SpaceLeft << "." << endl;
4348 if (SubOrder > 1) { // Due to Added above we have to check extra whether we're not already reaching beyond the desired Order
4349 // --2-- look at all added end pieces of this combination, construct bond subsets and sweep through a power set of these by recursion
4350 SP = RootDistance+1; // this is the next level
4351 // first count the members in the subset
4352 SubSetDimension = 0;
4353 Binder = FragmentSearch->BondsPerSPList[2*SP]; // start node for this level
4354 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) { // compare to end node of this level
4355 Binder = Binder->next;
4356 for (int k=TouchedIndex;k--;) {
4357 if (Binder->Contains(TouchedList[k])) // if we added this very endpiece
4358 SubSetDimension++;
4359 }
4360 }
4361 // then allocate and fill the list
4362 BondsList = (bond **) Malloc(sizeof(bond *)*SubSetDimension, "molecule::SPFragmentGenerator: **BondsList");
4363 SubSetDimension = 0;
4364 Binder = FragmentSearch->BondsPerSPList[2*SP];
4365 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) {
4366 Binder = Binder->next;
4367 for (int k=0;k<TouchedIndex;k++) {
4368 if (Binder->leftatom->nr == TouchedList[k]) // leftatom is always the close one
4369 BondsList[SubSetDimension++] = Binder;
4370 }
4371 }
4372 *out << Verbose(2+verbosity) << "Calling subset generator " << SP << " away from root " << *FragmentSearch->Root << " with sub set dimension " << SubSetDimension << "." << endl;
4373 SPFragmentGenerator(out, FragmentSearch, SP, BondsList, SubSetDimension, SubOrder-bits);
4374 Free((void **)&BondsList, "molecule::SPFragmentGenerator: **BondsList");
4375 }
4376 } else {
4377 // --2-- otherwise store the complete fragment
4378 *out << Verbose(1+verbosity) << "Enough items on stack for a fragment!" << endl;
4379 // store fragment as a KeySet
4380 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch->FragmentCounter << "], local nr.s are: ";
4381 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
4382 *out << (*runner) << " ";
4383 *out << endl;
4384 //if (!CheckForConnectedSubgraph(out, FragmentSearch->FragmentSet))
4385 //*out << Verbose(0) << "ERROR: The found fragment is not a connected subgraph!" << endl;
4386 InsertFragmentIntoGraph(out, FragmentSearch);
4387 //Removal = LookForRemovalCandidate(out, FragmentSearch->FragmentSet, FragmentSearch->ShortestPathList);
4388 //Removal = StoreFragmentFromStack(out, FragmentSearch->Root, FragmentSearch->Leaflet, FragmentSearch->FragmentStack, FragmentSearch->ShortestPathList, &FragmentSearch->FragmentCounter, FragmentSearch->configuration);
4389 }
4390
4391 // --3-- remove all added items in this level from snake stack
4392 *out << Verbose(1+verbosity) << "Removing all items that were added on this SP level " << RootDistance << "." << endl;
4393 for(int j=0;j<TouchedIndex;j++) {
4394 Removal = TouchedList[j];
4395 *out << Verbose(2+verbosity) << "Removing item nr. " << Removal << " from snake stack." << endl;
4396 FragmentSearch->FragmentSet->erase(Removal);
4397 TouchedList[j] = -1;
4398 }
4399 *out << Verbose(2) << "Remaining local nr.s on snake stack are: ";
4400 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
4401 *out << (*runner) << " ";
4402 *out << endl;
4403 TouchedIndex = 0; // set Index to 0 for list of atoms added on this level
4404 } else {
4405 *out << Verbose(2+verbosity) << "More atoms to add for this set (" << bits << ") than space left on stack " << SubOrder << ", skipping this set." << endl;
4406 }
4407 }
4408 Free((void **)&TouchedList, "molecule::SPFragmentGenerator: *TouchedList");
4409 *out << Verbose(1+verbosity) << "End of SPFragmentGenerator, " << RootDistance << " away from Root " << *FragmentSearch->Root << " and SubOrder is " << SubOrder << "." << endl;
4410};
4411
4412/** For a given keyset \a *Fragment, checks whether it is connected in the current molecule.
4413 * \param *out output stream for debugging
4414 * \param *Fragment Keyset of fragment's vertices
4415 * \return true - connected, false - disconnected
4416 * \note this is O(n^2) for it's just a bug checker not meant for permanent use!
4417 */
4418bool molecule::CheckForConnectedSubgraph(ofstream *out, KeySet *Fragment)
4419{
4420 atom *Walker = NULL, *Walker2 = NULL;
4421 bool BondStatus = false;
4422 int size;
4423
4424 *out << Verbose(1) << "Begin of CheckForConnectedSubgraph" << endl;
4425 *out << Verbose(2) << "Disconnected atom: ";
4426
4427 // count number of atoms in graph
4428 size = 0;
4429 for(KeySet::iterator runner = Fragment->begin(); runner != Fragment->end(); runner++)
4430 size++;
4431 if (size > 1)
4432 for(KeySet::iterator runner = Fragment->begin(); runner != Fragment->end(); runner++) {
4433 Walker = FindAtom(*runner);
4434 BondStatus = false;
4435 for(KeySet::iterator runners = Fragment->begin(); runners != Fragment->end(); runners++) {
4436 Walker2 = FindAtom(*runners);
4437 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
4438 if (ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker) == Walker2) {
4439 BondStatus = true;
4440 break;
4441 }
4442 if (BondStatus)
4443 break;
4444 }
4445 }
4446 if (!BondStatus) {
4447 *out << (*Walker) << endl;
4448 return false;
4449 }
4450 }
4451 else {
4452 *out << "none." << endl;
4453 return true;
4454 }
4455 *out << "none." << endl;
4456
4457 *out << Verbose(1) << "End of CheckForConnectedSubgraph" << endl;
4458
4459 return true;
4460}
4461
4462/** Creates a list of all unique fragments of certain vertex size from a given graph \a Fragment for a given root vertex in the context of \a this molecule.
4463 * -# initialises UniqueFragments structure
4464 * -# fills edge list via BFS
4465 * -# creates the fragment by calling recursive function SPFragmentGenerator with UniqueFragments structure, 0 as
4466 root distance, the edge set, its dimension and the current suborder
4467 * -# Free'ing structure
4468 * Note that we may use the fact that the atoms are SP-ordered on the atomstack. I.e. when popping always the last, we first get all
4469 * with SP of 2, then those with SP of 3, then those with SP of 4 and so on.
4470 * \param *out output stream for debugging
4471 * \param Order bond order (limits BFS exploration and "number of digits" in power set generation
4472 * \param FragmentSearch UniqueFragments structure containing TEFactor, root atom and so on
4473 * \param RestrictedKeySet Restricted vertex set to use in context of molecule
4474 * \return number of inserted fragments
4475 * \note ShortestPathList in FragmentSearch structure is probably due to NumberOfAtomsSPLevel and SP not needed anymore
4476 */
4477int molecule::PowerSetGenerator(ofstream *out, int Order, struct UniqueFragments &FragmentSearch, KeySet RestrictedKeySet)
4478{
4479 int SP, AtomKeyNr;
4480 atom *Walker = NULL, *OtherWalker = NULL, *Predecessor = NULL;
4481 bond *Binder = NULL;
4482 bond *CurrentEdge = NULL;
4483 bond **BondsList = NULL;
4484 int RootKeyNr = FragmentSearch.Root->GetTrueFather()->nr;
4485 int Counter = FragmentSearch.FragmentCounter;
4486 int RemainingWalkers;
4487
4488 *out << endl;
4489 *out << Verbose(0) << "Begin of PowerSetGenerator with order " << Order << " at Root " << *FragmentSearch.Root << "." << endl;
4490
4491 // prepare Label and SP arrays of the BFS search
4492 FragmentSearch.ShortestPathList[FragmentSearch.Root->nr] = 0;
4493
4494 // prepare root level (SP = 0) and a loop bond denoting Root
4495 for (int i=1;i<Order;i++)
4496 FragmentSearch.BondsPerSPCount[i] = 0;
4497 FragmentSearch.BondsPerSPCount[0] = 1;
4498 Binder = new bond(FragmentSearch.Root, FragmentSearch.Root);
4499 add(Binder, FragmentSearch.BondsPerSPList[1]);
4500
4501 // do a BFS search to fill the SP lists and label the found vertices
4502 // Actually, we should construct a spanning tree vom the root atom and select all edges therefrom and put them into
4503 // according shortest path lists. However, we don't. Rather we fill these lists right away, as they do form a spanning
4504 // tree already sorted into various SP levels. That's why we just do loops over the depth (CurrentSP) and breadth
4505 // (EdgeinSPLevel) of this tree ...
4506 // In another picture, the bonds always contain a direction by rightatom being the one more distant from root and hence
4507 // naturally leftatom forming its predecessor, preventing the BFS"seeker" from continuing in the wrong direction.
4508 *out << endl;
4509 *out << Verbose(0) << "Starting BFS analysis ..." << endl;
4510 for (SP = 0; SP < (Order-1); SP++) {
4511 *out << Verbose(1) << "New SP level reached: " << SP << ", creating new SP list with " << FragmentSearch.BondsPerSPCount[SP] << " item(s)";
4512 if (SP > 0) {
4513 *out << ", old level closed with " << FragmentSearch.BondsPerSPCount[SP-1] << " item(s)." << endl;
4514 FragmentSearch.BondsPerSPCount[SP] = 0;
4515 } else
4516 *out << "." << endl;
4517
4518 RemainingWalkers = FragmentSearch.BondsPerSPCount[SP];
4519 CurrentEdge = FragmentSearch.BondsPerSPList[2*SP]; /// start of this SP level's list
4520 while (CurrentEdge->next != FragmentSearch.BondsPerSPList[2*SP+1]) { /// end of this SP level's list
4521 CurrentEdge = CurrentEdge->next;
4522 RemainingWalkers--;
4523 Walker = CurrentEdge->rightatom; // rightatom is always the one more distant
4524 Predecessor = CurrentEdge->leftatom; // ... and leftatom is predecessor
4525 AtomKeyNr = Walker->nr;
4526 *out << Verbose(0) << "Current Walker is: " << *Walker << " with nr " << Walker->nr << " and SP of " << SP << ", with " << RemainingWalkers << " remaining walkers on this level." << endl;
4527 // check for new sp level
4528 // go through all its bonds
4529 *out << Verbose(1) << "Going through all bonds of Walker." << endl;
4530 for (int i=0;i<NumberOfBondsPerAtom[AtomKeyNr];i++) {
4531 Binder = ListOfBondsPerAtom[AtomKeyNr][i];
4532 OtherWalker = Binder->GetOtherAtom(Walker);
4533 if ((RestrictedKeySet.find(OtherWalker->nr) != RestrictedKeySet.end())
4534 #ifdef ADDHYDROGEN
4535 && (OtherWalker->type->Z != 1)
4536 #endif
4537 ) { // skip hydrogens and restrict to fragment
4538 *out << Verbose(2) << "Current partner is " << *OtherWalker << " with nr " << OtherWalker->nr << " in bond " << *Binder << "." << endl;
4539 // set the label if not set (and push on root stack as well)
4540 if ((OtherWalker != Predecessor) && (OtherWalker->GetTrueFather()->nr > RootKeyNr)) { // only pass through those with label bigger than Root's
4541 FragmentSearch.ShortestPathList[OtherWalker->nr] = SP+1;
4542 *out << Verbose(3) << "Set Shortest Path to " << FragmentSearch.ShortestPathList[OtherWalker->nr] << "." << endl;
4543 // add the bond in between to the SP list
4544 Binder = new bond(Walker, OtherWalker); // create a new bond in such a manner, that bond::rightatom is always the one more distant
4545 add(Binder, FragmentSearch.BondsPerSPList[2*(SP+1)+1]);
4546 FragmentSearch.BondsPerSPCount[SP+1]++;
4547 *out << Verbose(3) << "Added its bond to SP list, having now " << FragmentSearch.BondsPerSPCount[SP+1] << " item(s)." << endl;
4548 } else {
4549 if (OtherWalker != Predecessor)
4550 *out << Verbose(3) << "Not passing on, as index of " << *OtherWalker << " " << OtherWalker->GetTrueFather()->nr << " is smaller than that of Root " << RootKeyNr << "." << endl;
4551 else
4552 *out << Verbose(3) << "This is my predecessor " << *Predecessor << "." << endl;
4553 }
4554 } else *out << Verbose(2) << "Is not in the restricted keyset or skipping hydrogen " << *OtherWalker << "." << endl;
4555 }
4556 }
4557 }
4558
4559 // outputting all list for debugging
4560 *out << Verbose(0) << "Printing all found lists." << endl;
4561 for(int i=1;i<Order;i++) { // skip the root edge in the printing
4562 Binder = FragmentSearch.BondsPerSPList[2*i];
4563 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
4564 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
4565 Binder = Binder->next;
4566 *out << Verbose(2) << *Binder << endl;
4567 }
4568 }
4569
4570 // creating fragments with the found edge sets (may be done in reverse order, faster)
4571 SP = -1; // the Root <-> Root edge must be subtracted!
4572 for(int i=Order;i--;) { // sum up all found edges
4573 Binder = FragmentSearch.BondsPerSPList[2*i];
4574 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
4575 Binder = Binder->next;
4576 SP ++;
4577 }
4578 }
4579 *out << Verbose(0) << "Total number of edges is " << SP << "." << endl;
4580 if (SP >= (Order-1)) {
4581 // start with root (push on fragment stack)
4582 *out << Verbose(0) << "Starting fragment generation with " << *FragmentSearch.Root << ", local nr is " << FragmentSearch.Root->nr << "." << endl;
4583 FragmentSearch.FragmentSet->clear();
4584 *out << Verbose(0) << "Preparing subset for this root and calling generator." << endl;
4585 // prepare the subset and call the generator
4586 BondsList = (bond **) Malloc(sizeof(bond *)*FragmentSearch.BondsPerSPCount[0], "molecule::PowerSetGenerator: **BondsList");
4587 BondsList[0] = FragmentSearch.BondsPerSPList[0]->next; // on SP level 0 there's only the root bond
4588
4589 SPFragmentGenerator(out, &FragmentSearch, 0, BondsList, FragmentSearch.BondsPerSPCount[0], Order);
4590
4591 Free((void **)&BondsList, "molecule::PowerSetGenerator: **BondsList");
4592 } else {
4593 *out << Verbose(0) << "Not enough total number of edges to build " << Order << "-body fragments." << endl;
4594 }
4595
4596 // as FragmentSearch structure is used only once, we don't have to clean it anymore
4597 // remove root from stack
4598 *out << Verbose(0) << "Removing root again from stack." << endl;
4599 FragmentSearch.FragmentSet->erase(FragmentSearch.Root->nr);
4600
4601 // free'ing the bonds lists
4602 *out << Verbose(0) << "Free'ing all found lists. and resetting index lists" << endl;
4603 for(int i=Order;i--;) {
4604 *out << Verbose(1) << "Current SP level is " << i << ": ";
4605 Binder = FragmentSearch.BondsPerSPList[2*i];
4606 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
4607 Binder = Binder->next;
4608 // *out << "Removing atom " << Binder->leftatom->nr << " and " << Binder->rightatom->nr << "." << endl; // make sure numbers are local
4609 FragmentSearch.ShortestPathList[Binder->leftatom->nr] = -1;
4610 FragmentSearch.ShortestPathList[Binder->rightatom->nr] = -1;
4611 }
4612 // delete added bonds
4613 cleanup(FragmentSearch.BondsPerSPList[2*i], FragmentSearch.BondsPerSPList[2*i+1]);
4614 // also start and end node
4615 *out << "cleaned." << endl;
4616 }
4617
4618 // return list
4619 *out << Verbose(0) << "End of PowerSetGenerator." << endl;
4620 return (FragmentSearch.FragmentCounter - Counter);
4621};
4622
4623/** Corrects the nuclei position if the fragment was created over the cell borders.
4624 * Scans all bonds, checks the distance, if greater than typical, we have a candidate for the correction.
4625 * We remove the bond whereafter the graph probably separates. Then, we translate the one component periodically
4626 * and re-add the bond. Looping on the distance check.
4627 * \param *out ofstream for debugging messages
4628 */
4629void molecule::ScanForPeriodicCorrection(ofstream *out)
4630{
4631 bond *Binder = NULL;
4632 bond *OtherBinder = NULL;
4633 atom *Walker = NULL;
4634 atom *OtherWalker = NULL;
4635 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
4636 enum Shading *ColorList = NULL;
4637 double tmp;
4638 Vector Translationvector;
4639 //class StackClass<atom *> *CompStack = NULL;
4640 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
4641 bool flag = true;
4642
4643 *out << Verbose(2) << "Begin of ScanForPeriodicCorrection." << endl;
4644
4645 ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::ScanForPeriodicCorrection: *ColorList");
4646 while (flag) {
4647 // remove bonds that are beyond bonddistance
4648 for(int i=NDIM;i--;)
4649 Translationvector.x[i] = 0.;
4650 // scan all bonds
4651 Binder = first;
4652 flag = false;
4653 while ((!flag) && (Binder->next != last)) {
4654 Binder = Binder->next;
4655 for (int i=NDIM;i--;) {
4656 tmp = fabs(Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i]);
4657 //*out << Verbose(3) << "Checking " << i << "th distance of " << *Binder->leftatom << " to " << *Binder->rightatom << ": " << tmp << "." << endl;
4658 if (tmp > BondDistance) {
4659 OtherBinder = Binder->next; // note down binding partner for later re-insertion
4660 unlink(Binder); // unlink bond
4661 *out << Verbose(2) << "Correcting at bond " << *Binder << "." << endl;
4662 flag = true;
4663 break;
4664 }
4665 }
4666 }
4667 if (flag) {
4668 // create translation vector from their periodically modified distance
4669 for (int i=NDIM;i--;) {
4670 tmp = Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i];
4671 if (fabs(tmp) > BondDistance)
4672 Translationvector.x[i] = (tmp < 0) ? +1. : -1.;
4673 }
4674 Translationvector.MatrixMultiplication(matrix);
4675 //*out << Verbose(3) << "Translation vector is ";
4676 Translationvector.Output(out);
4677 *out << endl;
4678 // apply to all atoms of first component via BFS
4679 for (int i=AtomCount;i--;)
4680 ColorList[i] = white;
4681 AtomStack->Push(Binder->leftatom);
4682 while (!AtomStack->IsEmpty()) {
4683 Walker = AtomStack->PopFirst();
4684 //*out << Verbose (3) << "Current Walker is: " << *Walker << "." << endl;
4685 ColorList[Walker->nr] = black; // mark as explored
4686 Walker->x.AddVector(&Translationvector); // translate
4687 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through all binding partners
4688 if (ListOfBondsPerAtom[Walker->nr][i] != Binder) {
4689 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
4690 if (ColorList[OtherWalker->nr] == white) {
4691 AtomStack->Push(OtherWalker); // push if yet unexplored
4692 }
4693 }
4694 }
4695 }
4696 // re-add bond
4697 link(Binder, OtherBinder);
4698 } else {
4699 *out << Verbose(3) << "No corrections for this fragment." << endl;
4700 }
4701 //delete(CompStack);
4702 }
4703
4704 // free allocated space from ReturnFullMatrixforSymmetric()
4705 delete(AtomStack);
4706 Free((void **)&ColorList, "molecule::ScanForPeriodicCorrection: *ColorList");
4707 Free((void **)&matrix, "molecule::ScanForPeriodicCorrection: *matrix");
4708 *out << Verbose(2) << "End of ScanForPeriodicCorrection." << endl;
4709};
4710
4711/** Blows the 6-dimensional \a cell_size array up to a full NDIM by NDIM matrix.
4712 * \param *symm 6-dim array of unique symmetric matrix components
4713 * \return allocated NDIM*NDIM array with the symmetric matrix
4714 */
4715double * molecule::ReturnFullMatrixforSymmetric(double *symm)
4716{
4717 double *matrix = (double *) Malloc(sizeof(double)*NDIM*NDIM, "molecule::ReturnFullMatrixforSymmetric: *matrix");
4718 matrix[0] = symm[0];
4719 matrix[1] = symm[1];
4720 matrix[2] = symm[3];
4721 matrix[3] = symm[1];
4722 matrix[4] = symm[2];
4723 matrix[5] = symm[4];
4724 matrix[6] = symm[3];
4725 matrix[7] = symm[4];
4726 matrix[8] = symm[5];
4727 return matrix;
4728};
4729
4730bool KeyCompare::operator() (const KeySet SubgraphA, const KeySet SubgraphB) const
4731{
4732 //cout << "my check is used." << endl;
4733 if (SubgraphA.size() < SubgraphB.size()) {
4734 return true;
4735 } else {
4736 if (SubgraphA.size() > SubgraphB.size()) {
4737 return false;
4738 } else {
4739 KeySet::iterator IteratorA = SubgraphA.begin();
4740 KeySet::iterator IteratorB = SubgraphB.begin();
4741 while ((IteratorA != SubgraphA.end()) && (IteratorB != SubgraphB.end())) {
4742 if ((*IteratorA) < (*IteratorB))
4743 return true;
4744 else if ((*IteratorA) > (*IteratorB)) {
4745 return false;
4746 } // else, go on to next index
4747 IteratorA++;
4748 IteratorB++;
4749 } // end of while loop
4750 }// end of check in case of equal sizes
4751 }
4752 return false; // if we reach this point, they are equal
4753};
4754
4755//bool operator < (KeySet SubgraphA, KeySet SubgraphB)
4756//{
4757// return KeyCompare(SubgraphA, SubgraphB);
4758//};
4759
4760/** Checking whether KeySet is not already present in Graph, if so just adds factor.
4761 * \param *out output stream for debugging
4762 * \param &set KeySet to insert
4763 * \param &graph Graph to insert into
4764 * \param *counter pointer to unique fragment count
4765 * \param factor energy factor for the fragment
4766 */
4767inline void InsertFragmentIntoGraph(ofstream *out, struct UniqueFragments *Fragment)
4768{
4769 GraphTestPair testGraphInsert;
4770
4771 testGraphInsert = Fragment->Leaflet->insert(GraphPair (*Fragment->FragmentSet,pair<int,double>(Fragment->FragmentCounter,Fragment->TEFactor))); // store fragment number and current factor
4772 if (testGraphInsert.second) {
4773 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " successfully inserted." << endl;
4774 Fragment->FragmentCounter++;
4775 } else {
4776 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
4777 ((*(testGraphInsert.first)).second).second += Fragment->TEFactor; // increase the "created" counter
4778 *out << Verbose(2) << "New factor is " << ((*(testGraphInsert.first)).second).second << "." << endl;
4779 }
4780};
4781//void inline InsertIntoGraph(ofstream *out, KeyStack &stack, Graph &graph, int *counter, double factor)
4782//{
4783// // copy stack contents to set and call overloaded function again
4784// KeySet set;
4785// for(KeyStack::iterator runner = stack.begin(); runner != stack.begin(); runner++)
4786// set.insert((*runner));
4787// InsertIntoGraph(out, set, graph, counter, factor);
4788//};
4789
4790/** Inserts each KeySet in \a graph2 into \a graph1.
4791 * \param *out output stream for debugging
4792 * \param graph1 first (dest) graph
4793 * \param graph2 second (source) graph
4794 * \param *counter keyset counter that gets increased
4795 */
4796inline void InsertGraphIntoGraph(ofstream *out, Graph &graph1, Graph &graph2, int *counter)
4797{
4798 GraphTestPair testGraphInsert;
4799
4800 for(Graph::iterator runner = graph2.begin(); runner != graph2.end(); runner++) {
4801 testGraphInsert = graph1.insert(GraphPair ((*runner).first,pair<int,double>((*counter)++,((*runner).second).second))); // store fragment number and current factor
4802 if (testGraphInsert.second) {
4803 *out << Verbose(2) << "KeySet " << (*counter)-1 << " successfully inserted." << endl;
4804 } else {
4805 *out << Verbose(2) << "KeySet " << (*counter)-1 << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
4806 ((*(testGraphInsert.first)).second).second += (*runner).second.second;
4807 *out << Verbose(2) << "New factor is " << (*(testGraphInsert.first)).second.second << "." << endl;
4808 }
4809 }
4810};
4811
4812
4813/** Performs BOSSANOVA decomposition at selected sites, increasing the cutoff by one at these sites.
4814 * -# constructs a complete keyset of the molecule
4815 * -# In a loop over all possible roots from the given rootstack
4816 * -# increases order of root site
4817 * -# calls PowerSetGenerator with this order, the complete keyset and the rootkeynr
4818 * -# for all consecutive lower levels PowerSetGenerator is called with the suborder, the higher order keyset
4819as the restricted one and each site in the set as the root)
4820 * -# these are merged into a fragment list of keysets
4821 * -# All fragment lists (for all orders, i.e. from all destination fields) are merged into one list for return
4822 * Important only is that we create all fragments, it is not important if we create them more than once
4823 * as these copies are filtered out via use of the hash table (KeySet).
4824 * \param *out output stream for debugging
4825 * \param Fragment&*List list of already present keystacks (adaptive scheme) or empty list
4826 * \param &RootStack stack with all root candidates (unequal to each atom in complete molecule if adaptive scheme is applied)
4827 * \param *MinimumRingSize minimum ring size for each atom (molecule::Atomcount)
4828 * \return pointer to Graph list
4829 */
4830void molecule::FragmentBOSSANOVA(ofstream *out, Graph *&FragmentList, KeyStack &RootStack, int *MinimumRingSize)
4831{
4832 Graph ***FragmentLowerOrdersList = NULL;
4833 int NumLevels, NumMolecules, TotalNumMolecules = 0, *NumMoleculesOfOrder = NULL;
4834 int counter = 0, Order;
4835 int UpgradeCount = RootStack.size();
4836 KeyStack FragmentRootStack;
4837 int RootKeyNr, RootNr;
4838 struct UniqueFragments FragmentSearch;
4839
4840 *out << Verbose(0) << "Begin of FragmentBOSSANOVA." << endl;
4841
4842 // FragmentLowerOrdersList is a 2D-array of pointer to MoleculeListClass objects, one dimension represents the ANOVA expansion of a single order (i.e. 5)
4843 // with all needed lower orders that are subtracted, the other dimension is the BondOrder (i.e. from 1 to 5)
4844 NumMoleculesOfOrder = (int *) Malloc(sizeof(int)*UpgradeCount, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
4845 FragmentLowerOrdersList = (Graph ***) Malloc(sizeof(Graph **)*UpgradeCount, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
4846
4847 // initialise the fragments structure
4848 FragmentSearch.ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *ShortestPathList");
4849 FragmentSearch.FragmentCounter = 0;
4850 FragmentSearch.FragmentSet = new KeySet;
4851 FragmentSearch.Root = FindAtom(RootKeyNr);
4852 for (int i=AtomCount;i--;) {
4853 FragmentSearch.ShortestPathList[i] = -1;
4854 }
4855
4856 // Construct the complete KeySet which we need for topmost level only (but for all Roots)
4857 atom *Walker = start;
4858 KeySet CompleteMolecule;
4859 while (Walker->next != end) {
4860 Walker = Walker->next;
4861 CompleteMolecule.insert(Walker->GetTrueFather()->nr);
4862 }
4863
4864 // this can easily be seen: if Order is 5, then the number of levels for each lower order is the total sum of the number of levels above, as
4865 // each has to be split up. E.g. for the second level we have one from 5th, one from 4th, two from 3th (which in turn is one from 5th, one from 4th),
4866 // hence we have overall four 2th order levels for splitting. This also allows for putting all into a single array (FragmentLowerOrdersList[])
4867 // with the order along the cells as this: 5433222211111111 for BondOrder 5 needing 16=pow(2,5-1) cells (only we use bit-shifting which is faster)
4868 RootNr = 0; // counts through the roots in RootStack
4869 while ((RootNr < UpgradeCount) && (!RootStack.empty())) {
4870 RootKeyNr = RootStack.front();
4871 RootStack.pop_front();
4872 Walker = FindAtom(RootKeyNr);
4873 // check cyclic lengths
4874 //if ((MinimumRingSize[Walker->GetTrueFather()->nr] != -1) && (Walker->GetTrueFather()->AdaptiveOrder+1 > MinimumRingSize[Walker->GetTrueFather()->nr])) {
4875 // *out << Verbose(0) << "Bond order " << Walker->GetTrueFather()->AdaptiveOrder << " of Root " << *Walker << " greater than or equal to Minimum Ring size of " << MinimumRingSize << " found is not allowed." << endl;
4876 //} else
4877 {
4878 // increase adaptive order by one
4879 Walker->GetTrueFather()->AdaptiveOrder++;
4880 Order = Walker->AdaptiveOrder = Walker->GetTrueFather()->AdaptiveOrder;
4881
4882 // initialise Order-dependent entries of UniqueFragments structure
4883 FragmentSearch.BondsPerSPList = (bond **) Malloc(sizeof(bond *)*Order*2, "molecule::PowerSetGenerator: ***BondsPerSPList");
4884 FragmentSearch.BondsPerSPCount = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *BondsPerSPCount");
4885 for (int i=Order;i--;) {
4886 FragmentSearch.BondsPerSPList[2*i] = new bond(); // start node
4887 FragmentSearch.BondsPerSPList[2*i+1] = new bond(); // end node
4888 FragmentSearch.BondsPerSPList[2*i]->next = FragmentSearch.BondsPerSPList[2*i+1]; // intertwine these two
4889 FragmentSearch.BondsPerSPList[2*i+1]->previous = FragmentSearch.BondsPerSPList[2*i];
4890 FragmentSearch.BondsPerSPCount[i] = 0;
4891 }
4892
4893 // allocate memory for all lower level orders in this 1D-array of ptrs
4894 NumLevels = 1 << (Order-1); // (int)pow(2,Order);
4895 FragmentLowerOrdersList[RootNr] = (Graph **) Malloc(sizeof(Graph *)*NumLevels, "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
4896 for (int i=0;i<NumLevels;i++)
4897 FragmentLowerOrdersList[RootNr][i] = NULL;
4898
4899 // create top order where nothing is reduced
4900 *out << Verbose(0) << "==============================================================================================================" << endl;
4901 *out << Verbose(0) << "Creating KeySets of Bond Order " << Order << " for " << *Walker << ", " << (RootStack.size()-RootNr) << " Roots remaining." << endl; // , NumLevels is " << NumLevels << "
4902
4903 // Create list of Graphs of current Bond Order (i.e. F_{ij})
4904 FragmentLowerOrdersList[RootNr][0] = new Graph;
4905 FragmentSearch.TEFactor = 1.;
4906 FragmentSearch.Leaflet = FragmentLowerOrdersList[RootNr][0]; // set to insertion graph
4907 FragmentSearch.Root = Walker;
4908 NumMoleculesOfOrder[RootNr] = PowerSetGenerator(out, Walker->AdaptiveOrder, FragmentSearch, CompleteMolecule);
4909 *out << Verbose(1) << "Number of resulting KeySets is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
4910 if (NumMoleculesOfOrder[RootNr] != 0) {
4911 NumMolecules = 0;
4912
4913 // we don't have to dive into suborders! These keysets are all already created on lower orders!
4914 // this was all ancient stuff, when we still depended on the TEFactors (and for those the suborders were needed)
4915
4916// if ((NumLevels >> 1) > 0) {
4917// // create lower order fragments
4918// *out << Verbose(0) << "Creating list of unique fragments of lower Bond Order terms to be subtracted." << endl;
4919// Order = Walker->AdaptiveOrder;
4920// for (int source=0;source<(NumLevels >> 1);source++) { // 1-terms don't need any more splitting, that's why only half is gone through (shift again)
4921// // step down to next order at (virtual) boundary of powers of 2 in array
4922// while (source >= (1 << (Walker->AdaptiveOrder-Order))) // (int)pow(2,Walker->AdaptiveOrder-Order))
4923// Order--;
4924// *out << Verbose(0) << "Current Order is: " << Order << "." << endl;
4925// for (int SubOrder=Order-1;SubOrder>0;SubOrder--) {
4926// int dest = source + (1 << (Walker->AdaptiveOrder-(SubOrder+1)));
4927// *out << Verbose(0) << "--------------------------------------------------------------------------------------------------------------" << endl;
4928// *out << Verbose(0) << "Current SubOrder is: " << SubOrder << " with source " << source << " to destination " << dest << "." << endl;
4929//
4930// // every molecule is split into a list of again (Order - 1) molecules, while counting all molecules
4931// //*out << Verbose(1) << "Splitting the " << (*FragmentLowerOrdersList[RootNr][source]).size() << " molecules of the " << source << "th cell in the array." << endl;
4932// //NumMolecules = 0;
4933// FragmentLowerOrdersList[RootNr][dest] = new Graph;
4934// for(Graph::iterator runner = (*FragmentLowerOrdersList[RootNr][source]).begin();runner != (*FragmentLowerOrdersList[RootNr][source]).end(); runner++) {
4935// for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
4936// Graph TempFragmentList;
4937// FragmentSearch.TEFactor = -(*runner).second.second;
4938// FragmentSearch.Leaflet = &TempFragmentList; // set to insertion graph
4939// FragmentSearch.Root = FindAtom(*sprinter);
4940// NumMoleculesOfOrder[RootNr] += PowerSetGenerator(out, SubOrder, FragmentSearch, (*runner).first);
4941// // insert new keysets FragmentList into FragmentLowerOrdersList[Walker->AdaptiveOrder-1][dest]
4942// *out << Verbose(1) << "Merging resulting key sets with those present in destination " << dest << "." << endl;
4943// InsertGraphIntoGraph(out, *FragmentLowerOrdersList[RootNr][dest], TempFragmentList, &NumMolecules);
4944// }
4945// }
4946// *out << Verbose(1) << "Number of resulting molecules for SubOrder " << SubOrder << " is: " << NumMolecules << "." << endl;
4947// }
4948// }
4949// }
4950 } else {
4951 Walker->GetTrueFather()->MaxOrder = true;
4952// *out << Verbose(1) << "Hence, we don't dive into SubOrders ... " << endl;
4953 }
4954 // now, we have completely filled each cell of FragmentLowerOrdersList[] for the current Walker->AdaptiveOrder
4955 //NumMoleculesOfOrder[Walker->AdaptiveOrder-1] = NumMolecules;
4956 TotalNumMolecules += NumMoleculesOfOrder[RootNr];
4957// *out << Verbose(1) << "Number of resulting molecules for Order " << (int)Walker->GetTrueFather()->AdaptiveOrder << " is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
4958 RootStack.push_back(RootKeyNr); // put back on stack
4959 RootNr++;
4960
4961 // free Order-dependent entries of UniqueFragments structure for next loop cycle
4962 Free((void **)&FragmentSearch.BondsPerSPCount, "molecule::PowerSetGenerator: *BondsPerSPCount");
4963 for (int i=Order;i--;) {
4964 delete(FragmentSearch.BondsPerSPList[2*i]);
4965 delete(FragmentSearch.BondsPerSPList[2*i+1]);
4966 }
4967 Free((void **)&FragmentSearch.BondsPerSPList, "molecule::PowerSetGenerator: ***BondsPerSPList");
4968 }
4969 }
4970 *out << Verbose(0) << "==============================================================================================================" << endl;
4971 *out << Verbose(1) << "Total number of resulting molecules is: " << TotalNumMolecules << "." << endl;
4972 *out << Verbose(0) << "==============================================================================================================" << endl;
4973
4974 // cleanup FragmentSearch structure
4975 Free((void **)&FragmentSearch.ShortestPathList, "molecule::PowerSetGenerator: *ShortestPathList");
4976 delete(FragmentSearch.FragmentSet);
4977
4978 // now, FragmentLowerOrdersList is complete, it looks - for BondOrder 5 - as this (number is the ANOVA Order of the terms therein)
4979 // 5433222211111111
4980 // 43221111
4981 // 3211
4982 // 21
4983 // 1
4984
4985 // Subsequently, we combine all into a single list (FragmentList)
4986
4987 *out << Verbose(0) << "Combining the lists of all orders per order and finally into a single one." << endl;
4988 if (FragmentList == NULL) {
4989 FragmentList = new Graph;
4990 counter = 0;
4991 } else {
4992 counter = FragmentList->size();
4993 }
4994 RootNr = 0;
4995 while (!RootStack.empty()) {
4996 RootKeyNr = RootStack.front();
4997 RootStack.pop_front();
4998 Walker = FindAtom(RootKeyNr);
4999 NumLevels = 1 << (Walker->AdaptiveOrder - 1);
5000 for(int i=0;i<NumLevels;i++) {
5001 if (FragmentLowerOrdersList[RootNr][i] != NULL) {
5002 InsertGraphIntoGraph(out, *FragmentList, (*FragmentLowerOrdersList[RootNr][i]), &counter);
5003 delete(FragmentLowerOrdersList[RootNr][i]);
5004 }
5005 }
5006 Free((void **)&FragmentLowerOrdersList[RootNr], "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
5007 RootNr++;
5008 }
5009 Free((void **)&FragmentLowerOrdersList, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
5010 Free((void **)&NumMoleculesOfOrder, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
5011
5012 *out << Verbose(0) << "End of FragmentBOSSANOVA." << endl;
5013};
5014
5015/** Comparison function for GSL heapsort on distances in two molecules.
5016 * \param *a
5017 * \param *b
5018 * \return <0, \a *a less than \a *b, ==0 if equal, >0 \a *a greater than \a *b
5019 */
5020inline int CompareDoubles (const void * a, const void * b)
5021{
5022 if (*(double *)a > *(double *)b)
5023 return -1;
5024 else if (*(double *)a < *(double *)b)
5025 return 1;
5026 else
5027 return 0;
5028};
5029
5030/** Determines whether two molecules actually contain the same atoms and coordination.
5031 * \param *out output stream for debugging
5032 * \param *OtherMolecule the molecule to compare this one to
5033 * \param threshold upper limit of difference when comparing the coordination.
5034 * \return NULL - not equal, otherwise an allocated (molecule::AtomCount) permutation map of the atom numbers (which corresponds to which)
5035 */
5036int * molecule::IsEqualToWithinThreshold(ofstream *out, molecule *OtherMolecule, double threshold)
5037{
5038 int flag;
5039 double *Distances = NULL, *OtherDistances = NULL;
5040 Vector CenterOfGravity, OtherCenterOfGravity;
5041 size_t *PermMap = NULL, *OtherPermMap = NULL;
5042 int *PermutationMap = NULL;
5043 atom *Walker = NULL;
5044 bool result = true; // status of comparison
5045
5046 *out << Verbose(3) << "Begin of IsEqualToWithinThreshold." << endl;
5047 /// first count both their atoms and elements and update lists thereby ...
5048 //*out << Verbose(0) << "Counting atoms, updating list" << endl;
5049 CountAtoms(out);
5050 OtherMolecule->CountAtoms(out);
5051 CountElements();
5052 OtherMolecule->CountElements();
5053
5054 /// ... and compare:
5055 /// -# AtomCount
5056 if (result) {
5057 if (AtomCount != OtherMolecule->AtomCount) {
5058 *out << Verbose(4) << "AtomCounts don't match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
5059 result = false;
5060 } else *out << Verbose(4) << "AtomCounts match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
5061 }
5062 /// -# ElementCount
5063 if (result) {
5064 if (ElementCount != OtherMolecule->ElementCount) {
5065 *out << Verbose(4) << "ElementCount don't match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
5066 result = false;
5067 } else *out << Verbose(4) << "ElementCount match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
5068 }
5069 /// -# ElementsInMolecule
5070 if (result) {
5071 for (flag=MAX_ELEMENTS;flag--;) {
5072 //*out << Verbose(5) << "Element " << flag << ": " << ElementsInMolecule[flag] << " <-> " << OtherMolecule->ElementsInMolecule[flag] << "." << endl;
5073 if (ElementsInMolecule[flag] != OtherMolecule->ElementsInMolecule[flag])
5074 break;
5075 }
5076 if (flag < MAX_ELEMENTS) {
5077 *out << Verbose(4) << "ElementsInMolecule don't match." << endl;
5078 result = false;
5079 } else *out << Verbose(4) << "ElementsInMolecule match." << endl;
5080 }
5081 /// then determine and compare center of gravity for each molecule ...
5082 if (result) {
5083 *out << Verbose(5) << "Calculating Centers of Gravity" << endl;
5084 DetermineCenter(CenterOfGravity);
5085 OtherMolecule->DetermineCenter(OtherCenterOfGravity);
5086 *out << Verbose(5) << "Center of Gravity: ";
5087 CenterOfGravity.Output(out);
5088 *out << endl << Verbose(5) << "Other Center of Gravity: ";
5089 OtherCenterOfGravity.Output(out);
5090 *out << endl;
5091 if (CenterOfGravity.Distance(&OtherCenterOfGravity) > threshold) {
5092 *out << Verbose(4) << "Centers of gravity don't match." << endl;
5093 result = false;
5094 }
5095 }
5096
5097 /// ... then make a list with the euclidian distance to this center for each atom of both molecules
5098 if (result) {
5099 *out << Verbose(5) << "Calculating distances" << endl;
5100 Distances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: Distances");
5101 OtherDistances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: OtherDistances");
5102 Walker = start;
5103 while (Walker->next != end) {
5104 Walker = Walker->next;
5105 Distances[Walker->nr] = CenterOfGravity.Distance(&Walker->x);
5106 }
5107 Walker = OtherMolecule->start;
5108 while (Walker->next != OtherMolecule->end) {
5109 Walker = Walker->next;
5110 OtherDistances[Walker->nr] = OtherCenterOfGravity.Distance(&Walker->x);
5111 }
5112
5113 /// ... sort each list (using heapsort (o(N log N)) from GSL)
5114 *out << Verbose(5) << "Sorting distances" << endl;
5115 PermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermMap");
5116 OtherPermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
5117 gsl_heapsort_index (PermMap, Distances, AtomCount, sizeof(double), CompareDoubles);
5118 gsl_heapsort_index (OtherPermMap, OtherDistances, AtomCount, sizeof(double), CompareDoubles);
5119 PermutationMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermutationMap");
5120 *out << Verbose(5) << "Combining Permutation Maps" << endl;
5121 for(int i=AtomCount;i--;)
5122 PermutationMap[PermMap[i]] = (int) OtherPermMap[i];
5123
5124 /// ... and compare them step by step, whether the difference is individiually(!) below \a threshold for all
5125 *out << Verbose(4) << "Comparing distances" << endl;
5126 flag = 0;
5127 for (int i=0;i<AtomCount;i++) {
5128 *out << Verbose(5) << "Distances: |" << Distances[PermMap[i]] << " - " << OtherDistances[OtherPermMap[i]] << "| = " << fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) << " ?<? " << threshold << endl;
5129 if (fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) > threshold)
5130 flag = 1;
5131 }
5132 Free((void **)&PermMap, "molecule::IsEqualToWithinThreshold: *PermMap");
5133 Free((void **)&OtherPermMap, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
5134
5135 /// free memory
5136 Free((void **)&Distances, "molecule::IsEqualToWithinThreshold: Distances");
5137 Free((void **)&OtherDistances, "molecule::IsEqualToWithinThreshold: OtherDistances");
5138 if (flag) { // if not equal
5139 Free((void **)&PermutationMap, "molecule::IsEqualToWithinThreshold: *PermutationMap");
5140 result = false;
5141 }
5142 }
5143 /// return pointer to map if all distances were below \a threshold
5144 *out << Verbose(3) << "End of IsEqualToWithinThreshold." << endl;
5145 if (result) {
5146 *out << Verbose(3) << "Result: Equal." << endl;
5147 return PermutationMap;
5148 } else {
5149 *out << Verbose(3) << "Result: Not equal." << endl;
5150 return NULL;
5151 }
5152};
5153
5154/** Returns an index map for two father-son-molecules.
5155 * The map tells which atom in this molecule corresponds to which one in the other molecul with their fathers.
5156 * \param *out output stream for debugging
5157 * \param *OtherMolecule corresponding molecule with fathers
5158 * \return allocated map of size molecule::AtomCount with map
5159 * \todo make this with a good sort O(n), not O(n^2)
5160 */
5161int * molecule::GetFatherSonAtomicMap(ofstream *out, molecule *OtherMolecule)
5162{
5163 atom *Walker = NULL, *OtherWalker = NULL;
5164 *out << Verbose(3) << "Begin of GetFatherAtomicMap." << endl;
5165 int *AtomicMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::GetAtomicMap: *AtomicMap"); //Calloc
5166 for (int i=AtomCount;i--;)
5167 AtomicMap[i] = -1;
5168 if (OtherMolecule == this) { // same molecule
5169 for (int i=AtomCount;i--;) // no need as -1 means already that there is trivial correspondence
5170 AtomicMap[i] = i;
5171 *out << Verbose(4) << "Map is trivial." << endl;
5172 } else {
5173 *out << Verbose(4) << "Map is ";
5174 Walker = start;
5175 while (Walker->next != end) {
5176 Walker = Walker->next;
5177 if (Walker->father == NULL) {
5178 AtomicMap[Walker->nr] = -2;
5179 } else {
5180 OtherWalker = OtherMolecule->start;
5181 while (OtherWalker->next != OtherMolecule->end) {
5182 OtherWalker = OtherWalker->next;
5183 //for (int i=0;i<AtomCount;i++) { // search atom
5184 //for (int j=0;j<OtherMolecule->AtomCount;j++) {
5185 //*out << Verbose(4) << "Comparing father " << Walker->father << " with the other one " << OtherWalker->father << "." << endl;
5186 if (Walker->father == OtherWalker)
5187 AtomicMap[Walker->nr] = OtherWalker->nr;
5188 }
5189 }
5190 *out << AtomicMap[Walker->nr] << "\t";
5191 }
5192 *out << endl;
5193 }
5194 *out << Verbose(3) << "End of GetFatherAtomicMap." << endl;
5195 return AtomicMap;
5196};
5197
5198/** Stores the temperature evaluated from velocities in molecule::Trajectories.
5199 * We simply use the formula equivaleting temperature and kinetic energy:
5200 * \f$k_B T = \sum_i m_i v_i^2\f$
5201 * \param *out output stream for debugging
5202 * \param startstep first MD step in molecule::Trajectories
5203 * \param endstep last plus one MD step in molecule::Trajectories
5204 * \param *output output stream of temperature file
5205 * \return file written (true), failure on writing file (false)
5206 */
5207bool molecule::OutputTemperatureFromTrajectories(ofstream *out, int startstep, int endstep, ofstream *output)
5208{
5209 double temperature;
5210 atom *Walker = NULL;
5211 // test stream
5212 if (output == NULL)
5213 return false;
5214 else
5215 *output << "# Step Temperature [K] Temperature [a.u.]" << endl;
5216 for (int step=startstep;step < endstep; step++) { // loop over all time steps
5217 temperature = 0.;
5218 Walker = start;
5219 while (Walker->next != end) {
5220 Walker = Walker->next;
5221 for (int i=NDIM;i--;)
5222 temperature += Walker->type->mass * Trajectories[Walker].U.at(step).x[i]* Trajectories[Walker].U.at(step).x[i];
5223 }
5224 *output << step << "\t" << temperature*AtomicEnergyToKelvin << "\t" << temperature << endl;
5225 }
5226 return true;
5227};
Note: See TracBrowser for help on using the repository browser.