source: molecuilder/src/molecules.cpp@ c75363

Last change on this file since c75363 was c75363, checked in by Frederik Heber <heber@…>, 17 years ago

HUGE REWRITE to allow for adaptive increase of the bond order, first working commit

Lots of code was thrown out:
-BottomUp, TopDown and GetAtomicFragments
-TEFactors are now used as "CreationCounters", i.e. the count how often a fragment as been created (ideal would be only once)
-ReduceToUniqueOnes and stuff all thrown out, since they are out-dated since use of hash table
Other changes:
-CreateListofUniqueFragments renamed to PowerSetGenerator
-PowerSetGenerator goes not over all reachable roots, but one given by calling function FragmentBOSSANOVA
-FragmentBOSSANOVA loops over all possible root sites and hands this over to PowerSetGenerator
-by the virtue of the hash table it is not important anymore whether created keysets are unique or not, as this is recognized in log(n). Hence, the label < label is not important anymore (and not applicable in an adaptive scheme with old, parsed keysets and unknown labels) (THIS IS HOWEVER NOT DONE YET!)
The setup is then as follows:

  1. FragmentMolecule
    1. parses adjacency, keysets and orderatsite files
    2. performs DFS to find connected subgraphs (to leave this in was a design decision: might be useful later)
    3. a RootStack is created for every subgraph (here, later we implement the "update 10 sites with highest energy contribution", and that's why this consciously not done in the following loop)
    4. in a loop over all subgraphs

d1. calls FragmentBOSSANOVA with this RootStack and within the subgraph molecule structure
d2. creates molecule (fragment)s from the returned keysets (StoreFragmentFromKeySet)

  1. combines the generated molecule lists from all subgraphs
  2. saves to disk: fragment configs, adjacency, orderatsite, keyset files
  1. FragmentBOSSANOVA
    1. constructs a complete keyset of the molecule
    2. In a loop over all possible roots from the given rootstack

b1. increases order of root site
b2. calls PowerSetGenerator with this order, the complete keyset and the rootkeynr
b3. for all consecutive lower levels PowerSetGenerator is called with the suborder, the higher order keyset as the restricted one and each site in the set as the root)
b4. these are merged into a fragment list of keysets

  1. All fragment lists (for all orders, i.e. from all destination fields) are merged into one list for return
  1. PowerSetGenerator
    1. initialises UniqueFragments structure
    2. fills edge list via BFS
    3. creates the fragment by calling recursive function SPFragmentGenerator with UniqueFragments structure, 0 as root distance, the edge set, its dimension and the current suborder
    4. Free'ing structure
  2. SPFragmentGenerator (nothing much has changed here)
    1. loops over every possible combination (2dimension of edge set)

a1. inserts current set, if there's still space left

a11. yes: calls SPFragmentGenerator with structure, created new edge list and size respective to root distance+1
a12. no: stores fragment into keyset list by calling InsertFragmentIntoGraph

a2. removes all items added into the snake stack (in UniqueFragments structure) added during level (root distance) and current set

  • Property mode set to 100644
File size: 169.3 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=0;i<num;i++) {
24 for(int j=0;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 last_atom = 0;
50 elemente = teil;
51 AtomCount = 0;
52 BondCount = 0;
53 NoNonBonds = 0;
54 NoNonHydrogen = 0;
55 NoCyclicBonds = 0;
56 ListOfBondsPerAtom = NULL;
57 NumberOfBondsPerAtom = NULL;
58 ElementCount = 0;
59 for(int i=0;i<MAX_ELEMENTS;i++)
60 ElementsInMolecule[i] = 0;
61 cell_size[0] = cell_size[2] = cell_size[5]= 20.;
62 cell_size[1] = cell_size[3] = cell_size[4]= 0.;
63};
64
65/** Destructor of class molecule.
66 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
67 */
68molecule::~molecule()
69{
70 if (ListOfBondsPerAtom != NULL)
71 for(int i=0;i<AtomCount;i++)
72 Free((void **)&ListOfBondsPerAtom[i], "molecule::~molecule: ListOfBondsPerAtom[i]");
73 Free((void **)&ListOfBondsPerAtom, "molecule::~molecule: ListOfBondsPerAtom");
74 Free((void **)&NumberOfBondsPerAtom, "molecule::~molecule: NumberOfBondsPerAtom");
75 CleanupMolecule();
76 delete(first);
77 delete(last);
78 delete(end);
79 delete(start);
80};
81
82/** Adds given atom \a *pointer from molecule list.
83 * Increases molecule::last_atom and gives last number to added atom and names it according to its element::abbrev and molecule::AtomCount
84 * \param *pointer allocated and set atom
85 * \return true - succeeded, false - atom not found in list
86 */
87bool molecule::AddAtom(atom *pointer)
88{
89 if (pointer != NULL) {
90 pointer->sort = &pointer->nr;
91 pointer->nr = last_atom++; // increase number within molecule
92 AtomCount++;
93 if (pointer->type != NULL) {
94 if (ElementsInMolecule[pointer->type->Z] == 0)
95 ElementCount++;
96 ElementsInMolecule[pointer->type->Z]++; // increase number of elements
97 if (pointer->type->Z != 1)
98 NoNonHydrogen++;
99 if (pointer->Name == NULL) {
100 Free((void **)&pointer->Name, "molecule::AddAtom: *pointer->Name");
101 pointer->Name = (char *) Malloc(sizeof(char)*6, "molecule::AddAtom: *pointer->Name");
102 sprintf(pointer->Name, "%2s%02d", pointer->type->symbol, pointer->nr+1);
103 }
104 }
105 return add(pointer, end);
106 } else
107 return false;
108};
109
110/** Adds a copy of the given atom \a *pointer from molecule list.
111 * Increases molecule::last_atom and gives last number to added atom.
112 * \param *pointer allocated and set atom
113 * \return true - succeeded, false - atom not found in list
114 */
115atom * molecule::AddCopyAtom(atom *pointer)
116{
117 if (pointer != NULL) {
118 atom *walker = new atom();
119 walker->type = pointer->type; // copy element of atom
120 walker->x.CopyVector(&pointer->x); // copy coordination
121 walker->v.CopyVector(&pointer->v); // copy velocity
122 walker->FixedIon = pointer->FixedIon;
123 walker->sort = &walker->nr;
124 walker->nr = last_atom++; // increase number within molecule
125 walker->father = pointer; //->GetTrueFather();
126 walker->Name = (char *) Malloc(sizeof(char)*strlen(pointer->Name)+1, "molecule::AddCopyAtom: *Name");
127 strcpy (walker->Name, pointer->Name);
128 add(walker, end);
129 if ((pointer->type != NULL) && (pointer->type->Z != 1))
130 NoNonHydrogen++;
131 AtomCount++;
132 return walker;
133 } else
134 return NULL;
135};
136
137/** Adds a Hydrogen atom in replacement for the given atom \a *partner in bond with a *origin.
138 * Here, we have to distinguish between single, double or triple bonds as stated by \a BondDegree, that each demand
139 * a different scheme when adding \a *replacement atom for the given one.
140 * -# Single Bond: Simply add new atom with bond distance rescaled to typical hydrogen one
141 * -# Double Bond: Here, we need the **BondList of the \a *origin atom, by scanning for the other bonds instead of
142 * *Bond, we use the through these connected atoms to determine the plane they lie in, vector::MakeNormalVector().
143 * The orthonormal vector to this plane along with the vector in *Bond direction determines the plane the two
144 * replacing hydrogens shall lie in. Now, all remains to do is take the usual hydrogen double bond angle for the
145 * element of *origin and form the sin/cos admixture of both plane vectors for the new coordinates of the two
146 * hydrogens forming this angle with *origin.
147 * -# 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
148 * triangle formed by the to be added hydrogens are not equal to the typical bond distance \f$l\f$ but have to be
149 * determined from the typical angle \f$\alpha\f$ for a hydrogen triple connected to the element of *origin):
150 * We have the height \f$d\f$ as the vector in *Bond direction (from triangle C1-H1-H2).
151 * \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 )}}
152 * \f]
153 * vector::GetNormalVector() creates one orthonormal vector from this *Bond vector and vector::MakeNormalVector creates
154 * the third one from the former two vectors. The latter ones form the plane of the base triangle mentioned above.
155 * The lengths for these are \f$f\f$ and \f$g\f$ (from triangle H1-H2-(center of H1-H2-H3)) with knowledge that
156 * the median lines in an isosceles triangle meet in the center point with a ratio 2:1.
157 * \f[ f = \frac{b}{\sqrt{3}} \qquad g = \frac{b}{2}
158 * \f]
159 * as the coordination of all three atoms in the coordinate system of these three vectors:
160 * \f$\pmatrix{d & f & 0}\f$, \f$\pmatrix{d & -0.5 \cdot f & g}\f$ and \f$\pmatrix{d & -0.5 \cdot f & -g}\f$.
161 *
162 * \param *out output stream for debugging
163 * \param *Bond pointer to bond between \a *origin and \a *replacement
164 * \param *TopOrigin son of \a *origin of upper level molecule (the atom added to this molecule as a copy of \a *origin)
165 * \param *origin pointer to atom which acts as the origin for scaling the added hydrogen to correct bond length
166 * \param *replacement pointer to the atom which shall be copied as a hydrogen atom in this molecule
167 * \param **BondList list of bonds \a *replacement has (necessary to determine plane for double and triple bonds)
168 * \param NumBond number of bonds in \a **BondList
169 * \param isAngstroem whether the coordination of the given atoms is in AtomicLength (false) or Angstrom(true)
170 * \return number of atoms added, if < bond::BondDegree then something went wrong
171 * \todo double and triple bonds splitting (always use the tetraeder angle!)
172 */
173bool molecule::AddHydrogenReplacementAtom(ofstream *out, bond *TopBond, atom *BottomOrigin, atom *TopOrigin, atom *TopReplacement, bond **BondList, int NumBond, bool IsAngstroem)
174{
175 double bondlength; // bond length of the bond to be replaced/cut
176 double bondangle; // bond angle of the bond to be replaced/cut
177 double BondRescale; // rescale value for the hydrogen bond length
178 bool AllWentWell = true; // flag gathering the boolean return value of molecule::AddAtom and other functions, as return value on exit
179 bond *FirstBond = NULL, *SecondBond = NULL; // Other bonds in double bond case to determine "other" plane
180 atom *FirstOtherAtom = NULL, *SecondOtherAtom = NULL, *ThirdOtherAtom = NULL; // pointer to hydrogen atoms to be added
181 int i; // loop variable
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=0;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 (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(i=0;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(i=0;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 *first = 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 for(i=0;i<NumberOfAtoms;i++){
459 first = new atom;
460 getline(xyzfile,line,'\n');
461 istringstream *item = new istringstream(line);
462 //istringstream input(line);
463 cout << Verbose(1) << "Reading: " << line << endl;
464 *item >> shorthand;
465 *item >> x[0];
466 *item >> x[1];
467 *item >> x[2];
468 first->type = elemente->FindElement(shorthand);
469 if (first->type == NULL) {
470 cerr << "Could not parse the element at line: '" << line << "', setting to H.";
471 first->type = elemente->FindElement(1);
472 }
473 for(j=0;j<NDIM;j++)
474 first->x.x[j] = x[j];
475 AddAtom(first); // add to molecule
476 delete(item);
477 }
478 xyzfile.close();
479 delete(input);
480 return true;
481};
482
483/** Creates a copy of this molecule.
484 * \return copy of molecule
485 */
486molecule *molecule::CopyMolecule()
487{
488 molecule *copy = new molecule(elemente);
489 atom *CurrentAtom = NULL;
490 atom *LeftAtom = NULL, *RightAtom = NULL;
491 atom *Walker = NULL;
492
493 // copy all atoms
494 Walker = start;
495 while(Walker->next != end) {
496 Walker = Walker->next;
497 CurrentAtom = copy->AddCopyAtom(Walker);
498 }
499
500 // copy all bonds
501 bond *Binder = first;
502 bond *NewBond = NULL;
503 while(Binder->next != last) {
504 Binder = Binder->next;
505 // get the pendant atoms of current bond in the copy molecule
506 LeftAtom = copy->start;
507 while (LeftAtom->next != copy->end) {
508 LeftAtom = LeftAtom->next;
509 if (LeftAtom->father == Binder->leftatom)
510 break;
511 }
512 RightAtom = copy->start;
513 while (RightAtom->next != copy->end) {
514 RightAtom = RightAtom->next;
515 if (RightAtom->father == Binder->rightatom)
516 break;
517 }
518 NewBond = copy->AddBond(LeftAtom, RightAtom, Binder->BondDegree);
519 NewBond->Cyclic = Binder->Cyclic;
520 if (Binder->Cyclic)
521 copy->NoCyclicBonds++;
522 NewBond->Type = Binder->Type;
523 }
524 // correct fathers
525 Walker = copy->start;
526 while(Walker->next != copy->end) {
527 Walker = Walker->next;
528 if (Walker->father->father == Walker->father) // same atom in copy's father points to itself
529 Walker->father = Walker; // set father to itself (copy of a whole molecule)
530 else
531 Walker->father = Walker->father->father; // set father to original's father
532 }
533 // copy values
534 copy->CountAtoms((ofstream *)&cout);
535 copy->CountElements();
536 if (first->next != last) { // if adjaceny list is present
537 copy->BondDistance = BondDistance;
538 copy->CreateListOfBondsPerAtom((ofstream *)&cout);
539 }
540
541 return copy;
542};
543
544/** Adds a bond to a the molecule specified by two atoms, \a *first and \a *second.
545 * Also updates molecule::BondCount and molecule::NoNonBonds.
546 * \param *first first atom in bond
547 * \param *second atom in bond
548 * \return pointer to bond or NULL on failure
549 */
550bond * molecule::AddBond(atom *atom1, atom *atom2, int degree=1)
551{
552 bond *Binder = NULL;
553 if ((atom1 != NULL) && (FindAtom(atom1->nr) != NULL) && (atom2 != NULL) && (FindAtom(atom2->nr) != NULL)) {
554 Binder = new bond(atom1, atom2, degree, BondCount++);
555 if ((atom1->type != NULL) && (atom1->type->Z != 1) && (atom2->type != NULL) && (atom2->type->Z != 1))
556 NoNonBonds++;
557 add(Binder, last);
558 } else {
559 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;
560 }
561 return Binder;
562};
563
564/** Remove bond from bond chain list.
565 * \todo Function not implemented yet
566 * \param *pointer bond pointer
567 * \return true - bound found and removed, false - bond not found/removed
568 */
569bool molecule::RemoveBond(bond *pointer)
570{
571 //cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
572 removewithoutcheck(pointer);
573 return true;
574};
575
576/** Remove every bond from bond chain list that atom \a *BondPartner is a constituent of.
577 * \todo Function not implemented yet
578 * \param *BondPartner atom to be removed
579 * \return true - bounds found and removed, false - bonds not found/removed
580 */
581bool molecule::RemoveBonds(atom *BondPartner)
582{
583 cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
584 return false;
585};
586
587/** Sets the molecule::cell_size to the components of \a *dim (rectangular box)
588 * \param *dim vector class
589 */
590void molecule::SetBoxDimension(vector *dim)
591{
592 cell_size[0] = dim->x[0];
593 cell_size[1] = 0.;
594 cell_size[2] = dim->x[1];
595 cell_size[3] = 0.;
596 cell_size[4] = 0.;
597 cell_size[5] = dim->x[2];
598};
599
600/** Centers the edge of the atoms at (0,0,0).
601 * \param *out output stream for debugging
602 * \param *max coordinates of other edge, specifying box dimensions.
603 */
604void molecule::CenterEdge(ofstream *out, vector *max)
605{
606 vector *min = new vector;
607
608// *out << Verbose(3) << "Begin of CenterEdge." << endl;
609 atom *ptr = start->next; // start at first in list
610 if (ptr != end) { //list not empty?
611 for (int i=0;i<NDIM;i++) {
612 max->x[i] = ptr->x.x[i];
613 min->x[i] = ptr->x.x[i];
614 }
615 while (ptr->next != end) { // continue with second if present
616 ptr = ptr->next;
617 //ptr->Output(1,1,out);
618 for (int i=0;i<NDIM;i++) {
619 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
620 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
621 }
622 }
623// *out << Verbose(4) << "Maximum is ";
624// max->Output(out);
625// *out << ", Minimum is ";
626// min->Output(out);
627// *out << endl;
628
629 for (int i=0;i<NDIM;i++) {
630 min->x[i] *= -1.;
631 max->x[i] += min->x[i];
632 }
633 Translate(min);
634 }
635 delete(min);
636// *out << Verbose(3) << "End of CenterEdge." << endl;
637};
638
639/** Centers the center of the atoms at (0,0,0).
640 * \param *out output stream for debugging
641 * \param *center return vector for translation vector
642 */
643void molecule::CenterOrigin(ofstream *out, vector *center)
644{
645 int Num = 0;
646 atom *ptr = start->next; // start at first in list
647
648 for(int i=0;i<NDIM;i++) // zero center vector
649 center->x[i] = 0.;
650
651 if (ptr != end) { //list not empty?
652 while (ptr->next != end) { // continue with second if present
653 ptr = ptr->next;
654 Num++;
655 center->AddVector(&ptr->x);
656 }
657 center->Scale(-1./Num); // divide through total number (and sign for direction)
658 Translate(center);
659 }
660};
661
662/** Centers the center of gravity of the atoms at (0,0,0).
663 * \param *out output stream for debugging
664 * \param *center return vector for translation vector
665 */
666void molecule::CenterGravity(ofstream *out, vector *center)
667{
668 double Num = 0;
669 atom *ptr = start->next; // start at first in list
670 vector tmp;
671
672 for(int i=0;i<NDIM;i++) // zero center vector
673 center->x[i] = 0.;
674
675 if (ptr != end) { //list not empty?
676 while (ptr->next != end) { // continue with second if present
677 ptr = ptr->next;
678 Num += ptr->type->mass;
679 tmp.CopyVector(&ptr->x);
680 tmp.Scale(ptr->type->mass); // scale by mass
681 center->AddVector(&tmp);
682 }
683 center->Scale(-1./Num); // divide through total mass (and sign for direction)
684 Translate(center);
685 }
686};
687
688/** Scales all atoms by \a *factor.
689 * \param *factor pointer to scaling factor
690 */
691void molecule::Scale(double **factor)
692{
693 atom *ptr = start;
694
695 while (ptr->next != end) {
696 ptr = ptr->next;
697 ptr->x.Scale(factor);
698 }
699};
700
701/** Translate all atoms by given vector.
702 * \param trans[] translation vector.
703 */
704void molecule::Translate(const vector *trans)
705{
706 atom *ptr = start;
707
708 while (ptr->next != end) {
709 ptr = ptr->next;
710 ptr->x.Translate(trans);
711 }
712};
713
714/** Mirrors all atoms against a given plane.
715 * \param n[] normal vector of mirror plane.
716 */
717void molecule::Mirror(const vector *n)
718{
719 atom *ptr = start;
720
721 while (ptr->next != end) {
722 ptr = ptr->next;
723 ptr->x.Mirror(n);
724 }
725};
726
727/** Determines center of gravity (yet not considering atom masses).
728 * \param CenterOfGravity reference to return vector
729 */
730void molecule::DetermineCenterOfGravity(vector &CenterOfGravity)
731{
732 atom *Walker = start;
733 bond *Binder = NULL;
734 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
735 double tmp;
736 bool flag;
737 vector TestVector, TranslationVector;
738
739 do {
740 CenterOfGravity.Zero();
741 flag = true;
742 while (Walker->next != end) {
743 Walker = Walker->next;
744#ifdef ADDHYDROGEN
745 if (Walker->type->Z != 1) {
746#endif
747 TestVector.CopyVector(&Walker->x);
748 TestVector.InverseMatrixMultiplication(matrix);
749 TranslationVector.Zero();
750 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
751 Binder = ListOfBondsPerAtom[Walker->nr][i];
752 if (Walker->nr < Binder->GetOtherAtom(Walker)->nr) // otherwise we shift one to, the other fro and gain nothing
753 for (int j=0;j<NDIM;j++) {
754 tmp = Walker->x.x[j] - Binder->GetOtherAtom(Walker)->x.x[j];
755 if ((fabs(tmp)) > BondDistance) {
756 flag = false;
757 cout << Verbose(0) << "Hit: atom " << Walker->Name << " in bond " << *Binder << " has to be shifted due to " << tmp << "." << endl;
758 if (tmp > 0)
759 TranslationVector.x[j] -= 1.;
760 else
761 TranslationVector.x[j] += 1.;
762 }
763 }
764 }
765 TestVector.AddVector(&TranslationVector);
766 TestVector.MatrixMultiplication(matrix);
767 CenterOfGravity.AddVector(&TestVector);
768 cout << Verbose(1) << "Vector is: ";
769 TestVector.Output((ofstream *)&cout);
770 cout << endl;
771#ifdef ADDHYDROGEN
772 // now also change all hydrogens
773 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
774 Binder = ListOfBondsPerAtom[Walker->nr][i];
775 if (Binder->GetOtherAtom(Walker)->type->Z == 1) {
776 TestVector.CopyVector(&Binder->GetOtherAtom(Walker)->x);
777 TestVector.InverseMatrixMultiplication(matrix);
778 TestVector.AddVector(&TranslationVector);
779 TestVector.MatrixMultiplication(matrix);
780 CenterOfGravity.AddVector(&TestVector);
781 cout << Verbose(1) << "Hydrogen Vector is: ";
782 TestVector.Output((ofstream *)&cout);
783 cout << endl;
784 }
785 }
786 }
787#endif
788 }
789 } while (!flag);
790 Free((void **)&matrix, "molecule::DetermineCenterOfGravity: *matrix");
791 CenterOfGravity.Scale(1./(double)AtomCount);
792};
793
794/** Align all atoms in such a manner that given vector \a *n is along z axis.
795 * \param n[] alignment vector.
796 */
797void molecule::Align(vector *n)
798{
799 atom *ptr = start;
800 double alpha, tmp;
801 vector z_axis;
802 z_axis.x[0] = 0.;
803 z_axis.x[1] = 0.;
804 z_axis.x[2] = 1.;
805
806 // rotate on z-x plane
807 cout << Verbose(0) << "Begin of Aligning all atoms." << endl;
808 alpha = atan(-n->x[0]/n->x[2]);
809 cout << Verbose(1) << "Z-X-angle: " << alpha << " ... ";
810 while (ptr->next != end) {
811 ptr = ptr->next;
812 tmp = ptr->x.x[0];
813 ptr->x.x[0] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
814 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
815 }
816 // rotate n vector
817 tmp = n->x[0];
818 n->x[0] = cos(alpha) * tmp + sin(alpha) * n->x[2];
819 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
820 cout << Verbose(1) << "alignment vector after first rotation: ";
821 n->Output((ofstream *)&cout);
822 cout << endl;
823
824 // rotate on z-y plane
825 ptr = start;
826 alpha = atan(-n->x[1]/n->x[2]);
827 cout << Verbose(1) << "Z-Y-angle: " << alpha << " ... ";
828 while (ptr->next != end) {
829 ptr = ptr->next;
830 tmp = ptr->x.x[1];
831 ptr->x.x[1] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
832 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
833 }
834 // rotate n vector (for consistency check)
835 tmp = n->x[1];
836 n->x[1] = cos(alpha) * tmp + sin(alpha) * n->x[2];
837 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
838
839 cout << Verbose(1) << "alignment vector after second rotation: ";
840 n->Output((ofstream *)&cout);
841 cout << Verbose(1) << endl;
842 cout << Verbose(0) << "End of Aligning all atoms." << endl;
843};
844
845/** Removes atom from molecule list.
846 * \param *pointer atom to be removed
847 * \return true - succeeded, false - atom not found in list
848 */
849bool molecule::RemoveAtom(atom *pointer)
850{
851 if (ElementsInMolecule[pointer->type->Z] != 0) // this would indicate an error
852 ElementsInMolecule[pointer->type->Z]--; // decrease number of atom of this element
853 else
854 cerr << "ERROR: Atom " << pointer->Name << " is of element " << pointer->type->Z << " but the entry in the table of the molecule is 0!" << endl;
855 if (ElementsInMolecule[pointer->type->Z] == 0) // was last atom of this element?
856 ElementCount--;
857 return remove(pointer, start, end);
858};
859
860/** Removes every atom from molecule list.
861 * \return true - succeeded, false - atom not found in list
862 */
863bool molecule::CleanupMolecule()
864{
865 return (cleanup(start,end) && cleanup(first,last));
866};
867
868/** Finds an atom specified by its continuous number.
869 * \param Nr number of atom withim molecule
870 * \return pointer to atom or NULL
871 */
872atom * molecule::FindAtom(int Nr) const{
873 atom * walker = find(&Nr, start,end);
874 if (walker != NULL) {
875 //cout << Verbose(0) << "Found Atom Nr. " << walker->nr << endl;
876 return walker;
877 } else {
878 cout << Verbose(0) << "Atom not found in list." << endl;
879 return NULL;
880 }
881};
882
883/** Asks for atom number, and checks whether in list.
884 * \param *text question before entering
885 */
886atom * molecule::AskAtom(char *text)
887{
888 int No;
889 atom *ion = NULL;
890 do {
891 //cout << Verbose(0) << "============Atom list==========================" << endl;
892 //mol->Output((ofstream *)&cout);
893 //cout << Verbose(0) << "===============================================" << endl;
894 cout << Verbose(0) << text;
895 cin >> No;
896 ion = this->FindAtom(No);
897 } while (ion == NULL);
898 return ion;
899};
900
901/** Checks if given coordinates are within cell volume.
902 * \param *x array of coordinates
903 * \return true - is within, false - out of cell
904 */
905bool molecule::CheckBounds(const vector *x) const
906{
907 bool result = true;
908 int j =-1;
909 for (int i=0;i<3;i++) {
910 j += i+1;
911 result = result && ((x->x[i] >= 0) && (x->x[i] < cell_size[j]));
912 }
913 //return result;
914 return true; /// probably not gonna use the check no more
915};
916
917/** Calculates sum over least square distance to line hidden in \a *x.
918 * \param *x offset and direction vector
919 * \param *params pointer to lsq_params structure
920 * \return \f$ sum_i^N | y_i - (a + t_i b)|^2\f$
921 */
922double LeastSquareDistance (const gsl_vector * x, void * params)
923{
924 double res = 0, t;
925 vector a,b,c,d;
926 struct lsq_params *par = (struct lsq_params *)params;
927 atom *ptr = par->mol->start;
928
929 // initialize vectors
930 a.x[0] = gsl_vector_get(x,0);
931 a.x[1] = gsl_vector_get(x,1);
932 a.x[2] = gsl_vector_get(x,2);
933 b.x[0] = gsl_vector_get(x,3);
934 b.x[1] = gsl_vector_get(x,4);
935 b.x[2] = gsl_vector_get(x,5);
936 // go through all atoms
937 while (ptr != par->mol->end) {
938 ptr = ptr->next;
939 if (ptr->type == ((struct lsq_params *)params)->type) { // for specific type
940 c.CopyVector(&ptr->x); // copy vector to temporary one
941 c.SubtractVector(&a); // subtract offset vector
942 t = c.ScalarProduct(&b); // get direction parameter
943 d.CopyVector(&b); // and create vector
944 d.Scale(&t);
945 c.SubtractVector(&d); // ... yielding distance vector
946 res += d.ScalarProduct((const vector *)&d); // add squared distance
947 }
948 }
949 return res;
950};
951
952/** By minimizing the least square distance gains alignment vector.
953 * \bug this is not yet working properly it seems
954 */
955void molecule::GetAlignVector(struct lsq_params * par) const
956{
957 int np = 6;
958
959 const gsl_multimin_fminimizer_type *T =
960 gsl_multimin_fminimizer_nmsimplex;
961 gsl_multimin_fminimizer *s = NULL;
962 gsl_vector *ss;
963 gsl_multimin_function minex_func;
964
965 size_t iter = 0, i;
966 int status;
967 double size;
968
969 /* Initial vertex size vector */
970 ss = gsl_vector_alloc (np);
971
972 /* Set all step sizes to 1 */
973 gsl_vector_set_all (ss, 1.0);
974
975 /* Starting point */
976 par->x = gsl_vector_alloc (np);
977 par->mol = this;
978
979 gsl_vector_set (par->x, 0, 0.0); // offset
980 gsl_vector_set (par->x, 1, 0.0);
981 gsl_vector_set (par->x, 2, 0.0);
982 gsl_vector_set (par->x, 3, 0.0); // direction
983 gsl_vector_set (par->x, 4, 0.0);
984 gsl_vector_set (par->x, 5, 1.0);
985
986 /* Initialize method and iterate */
987 minex_func.f = &LeastSquareDistance;
988 minex_func.n = np;
989 minex_func.params = (void *)par;
990
991 s = gsl_multimin_fminimizer_alloc (T, np);
992 gsl_multimin_fminimizer_set (s, &minex_func, par->x, ss);
993
994 do
995 {
996 iter++;
997 status = gsl_multimin_fminimizer_iterate(s);
998
999 if (status)
1000 break;
1001
1002 size = gsl_multimin_fminimizer_size (s);
1003 status = gsl_multimin_test_size (size, 1e-2);
1004
1005 if (status == GSL_SUCCESS)
1006 {
1007 printf ("converged to minimum at\n");
1008 }
1009
1010 printf ("%5d ", (int)iter);
1011 for (i = 0; i < (size_t)np; i++)
1012 {
1013 printf ("%10.3e ", gsl_vector_get (s->x, i));
1014 }
1015 printf ("f() = %7.3f size = %.3f\n", s->fval, size);
1016 }
1017 while (status == GSL_CONTINUE && iter < 100);
1018
1019 for (i=0;i<(size_t)np;i++)
1020 gsl_vector_set(par->x, i, gsl_vector_get(s->x, i));
1021 //gsl_vector_free(par->x);
1022 gsl_vector_free(ss);
1023 gsl_multimin_fminimizer_free (s);
1024};
1025
1026/** Prints molecule to *out.
1027 * \param *out output stream
1028 */
1029bool molecule::Output(ofstream *out)
1030{
1031 element *runner = elemente->start;
1032 atom *walker = NULL;
1033 int ElementNo, AtomNo;
1034 CountElements();
1035
1036 if (out == NULL) {
1037 return false;
1038 } else {
1039 *out << "#Ion_TypeNr._Nr.R[0] R[1] R[2] MoveType (0 MoveIon, 1 FixedIon)" << endl;
1040 ElementNo = 0;
1041 while (runner->next != elemente->end) { // go through every element
1042 runner = runner->next;
1043 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
1044 ElementNo++;
1045 AtomNo = 0;
1046 walker = start;
1047 while (walker->next != end) { // go through every atom of this element
1048 walker = walker->next;
1049 if (walker->type == runner) { // if this atom fits to element
1050 AtomNo++;
1051 walker->Output(ElementNo, AtomNo, out);
1052 }
1053 }
1054 }
1055 }
1056 return true;
1057 }
1058};
1059
1060/** Output of element before the actual coordination list.
1061 * \param *out stream pointer
1062 */
1063bool molecule::Checkout(ofstream *out) const
1064{
1065 return elemente->Checkout(out, ElementsInMolecule);
1066};
1067
1068/** Prints molecule to *out as xyz file.
1069 * \param *out output stream
1070 */
1071bool molecule::OutputXYZ(ofstream *out) const
1072{
1073 atom *walker = NULL;
1074 int No = 0;
1075 time_t now;
1076
1077 now = time((time_t *)NULL); // Get the system time and put it into 'now' as 'calender time'
1078 walker = start;
1079 while (walker->next != end) { // go through every atom and count
1080 walker = walker->next;
1081 No++;
1082 }
1083 if (out != NULL) {
1084 *out << No << "\n\tCreated by molecuilder on " << ctime(&now);
1085 walker = start;
1086 while (walker->next != end) { // go through every atom of this element
1087 walker = walker->next;
1088 walker->OutputXYZLine(out);
1089 }
1090 return true;
1091 } else
1092 return false;
1093};
1094
1095/** Brings molecule::AtomCount and atom::*Name up-to-date.
1096 * \param *out output stream for debugging
1097 */
1098void molecule::CountAtoms(ofstream *out)
1099{
1100 int i = 0;
1101 atom *Walker = start;
1102 while (Walker->next != end) {
1103 Walker = Walker->next;
1104 i++;
1105 }
1106 if ((AtomCount == 0) || (i != AtomCount)) {
1107 *out << Verbose(3) << "Mismatch in AtomCount " << AtomCount << " and recounted number " << i << ", renaming all." << endl;
1108 AtomCount = i;
1109
1110 // count NonHydrogen atoms and give each atom a unique name
1111 if (AtomCount != 0) {
1112 i=0;
1113 NoNonHydrogen = 0;
1114 Walker = start;
1115 while (Walker->next != end) {
1116 Walker = Walker->next;
1117 Walker->nr = i; // update number in molecule (for easier referencing in FragmentMolecule lateron)
1118 if (Walker->type->Z != 1) // count non-hydrogen atoms whilst at it
1119 NoNonHydrogen++;
1120 Free((void **)&Walker->Name, "molecule::CountAtoms: *walker->Name");
1121 Walker->Name = (char *) Malloc(sizeof(char)*6, "molecule::CountAtoms: *walker->Name");
1122 sprintf(Walker->Name, "%2s%02d", Walker->type->symbol, Walker->nr+1);
1123 *out << "Naming atom nr. " << Walker->nr << " " << Walker->Name << "." << endl;
1124 i++;
1125 }
1126 } else
1127 *out << Verbose(3) << "AtomCount is still " << AtomCount << ", thus counting nothing." << endl;
1128 }
1129};
1130
1131/** Brings molecule::ElementCount and molecule::ElementsInMolecule up-to-date.
1132 */
1133void molecule::CountElements()
1134{
1135 int i = 0;
1136 for(i=0;i<MAX_ELEMENTS;i++)
1137 ElementsInMolecule[i] = 0;
1138 ElementCount = 0;
1139
1140 atom *walker = start;
1141 while (walker->next != end) {
1142 walker = walker->next;
1143 ElementsInMolecule[walker->type->Z]++;
1144 i++;
1145 }
1146 for(i=0;i<MAX_ELEMENTS;i++)
1147 ElementCount += (ElementsInMolecule[i] != 0 ? 1 : 0);
1148};
1149
1150/** Counts all cyclic bonds and returns their number.
1151 * \note Hydrogen bonds can never by cyclic, thus no check for that
1152 * \param *out output stream for debugging
1153 * \return number opf cyclic bonds
1154 */
1155int molecule::CountCyclicBonds(ofstream *out)
1156{
1157 int No = 0;
1158 int MinimumRingSize;
1159 MoleculeLeafClass *Subgraphs = NULL;
1160 bond *Binder = first;
1161 if ((Binder->next != last) && (Binder->next->Type == Undetermined)) {
1162 *out << Verbose(0) << "No Depth-First-Search analysis performed so far, calling ..." << endl;
1163 Subgraphs = DepthFirstSearchAnalysis(out, false, MinimumRingSize);
1164 while (Subgraphs->next != NULL) {
1165 Subgraphs = Subgraphs->next;
1166 delete(Subgraphs->previous);
1167 }
1168 delete(Subgraphs);
1169 }
1170 while(Binder->next != last) {
1171 Binder = Binder->next;
1172 if (Binder->Cyclic)
1173 No++;
1174 }
1175 return No;
1176};
1177
1178/** Returns Shading as a char string.
1179 * \param color the Shading
1180 * \return string of the flag
1181 */
1182char * molecule::GetColor(enum Shading color)
1183{
1184 switch(color) {
1185 case white:
1186 return "white";
1187 break;
1188 case lightgray:
1189 return "lightgray";
1190 break;
1191 case darkgray:
1192 return "darkgray";
1193 break;
1194 case black:
1195 return "black";
1196 break;
1197 default:
1198 return "uncolored";
1199 break;
1200 };
1201};
1202
1203
1204/** Counts necessary number of valence electrons and returns number and SpinType.
1205 * \param configuration containing everything
1206 */
1207void molecule::CalculateOrbitals(class config &configuration)
1208{
1209 configuration.MaxPsiDouble = configuration.PsiMaxNoDown = configuration.PsiMaxNoUp = configuration.PsiType = 0;
1210 for(int i=0;i<MAX_ELEMENTS;i++) {
1211 if (ElementsInMolecule[i] != 0) {
1212 //cout << "CalculateOrbitals: " << elemente->FindElement(i)->name << " has a valence of " << (int)elemente->FindElement(i)->Valence << " and there are " << ElementsInMolecule[i] << " of it." << endl;
1213 configuration.MaxPsiDouble += ElementsInMolecule[i]*((int)elemente->FindElement(i)->Valence);
1214 }
1215 }
1216 configuration.PsiMaxNoDown = configuration.MaxPsiDouble/2 + (configuration.MaxPsiDouble % 2);
1217 configuration.PsiMaxNoUp = configuration.MaxPsiDouble/2;
1218 configuration.MaxPsiDouble /= 2;
1219 configuration.PsiType = (configuration.PsiMaxNoDown == configuration.PsiMaxNoUp) ? 0 : 1;
1220 if ((configuration.PsiType == 1) && (configuration.ProcPEPsi < 2)) {
1221 configuration.ProcPEGamma /= 2;
1222 configuration.ProcPEPsi *= 2;
1223 } else {
1224 configuration.ProcPEGamma *= configuration.ProcPEPsi;
1225 configuration.ProcPEPsi = 1;
1226 }
1227 configuration.InitMaxMinStopStep = configuration.MaxMinStopStep = configuration.MaxPsiDouble;
1228};
1229
1230/** Creates an adjacency list of the molecule.
1231 * Generally, we use the CSD approach to bond recognition, that is the the distance
1232 * between two atoms A and B must be within [Rcov(A)+Rcov(B)-t,Rcov(A)+Rcov(B)+t] with
1233 * a threshold t = 0.4 Angstroem.
1234 * To make it O(N log N) the function uses the linked-cell technique as follows:
1235 * The procedure is step-wise:
1236 * -# Remove every bond in list
1237 * -# Count the atoms in the molecule with CountAtoms()
1238 * -# partition cell into smaller linked cells of size \a bonddistance
1239 * -# put each atom into its corresponding cell
1240 * -# go through every cell, check the atoms therein against all possible bond partners in the 27 adjacent cells, add bond if true
1241 * -# create the list of bonds via CreateListOfBondsPerAtom()
1242 * -# correct the bond degree iteratively (single->double->triple bond)
1243 * -# finally print the bond list to \a *out if desired
1244 * \param *out out stream for printing the matrix, NULL if no output
1245 * \param bonddistance length of linked cells (i.e. maximum minimal length checked)
1246 */
1247void molecule::CreateAdjacencyList(ofstream *out, double bonddistance)
1248{
1249 atom *Walker = NULL, *OtherWalker = NULL;
1250 int No, NoBonds;
1251 bond *Binder = NULL;
1252 int NumberCells, divisor[NDIM], n[NDIM], N[NDIM], index, Index, j;
1253 molecule **CellList;
1254 double distance, MinDistance, MaxDistance;
1255 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
1256 vector x;
1257
1258 BondDistance = bonddistance;
1259 *out << Verbose(0) << "Begin of CreateAdjacencyList." << endl;
1260 // remove every bond from the list
1261 if ((first->next != last) && (last->previous != first)) { // there are bonds present
1262 cleanup(first,last);
1263 }
1264
1265 // count atoms in molecule = dimension of matrix (also give each unique name and continuous numbering)
1266 CountAtoms(out);
1267 *out << Verbose(1) << "AtomCount " << AtomCount << "." << endl;
1268
1269 if (AtomCount != 0) {
1270 // 1. find divisor for each axis, such that a sphere with radius of at least bonddistance can be placed into each cell
1271 j=-1;
1272 for (int i=0;i<NDIM;i++) {
1273 j += i+1;
1274 divisor[i] = (int)floor(cell_size[j]/bonddistance); // take smaller value such that size of linked cell is at least bonddistance
1275 *out << Verbose(1) << "divisor[" << i << "] = " << divisor[i] << "." << endl;
1276 }
1277 // 2a. allocate memory for the cell list
1278 NumberCells = divisor[0]*divisor[1]*divisor[2];
1279 *out << Verbose(1) << "Allocating " << NumberCells << " cells." << endl;
1280 CellList = (molecule **) Malloc(sizeof(molecule *)*NumberCells, "molecule::CreateAdjacencyList - ** CellList");
1281 for (int i=0;i<NumberCells;i++)
1282 CellList[i] = NULL;
1283
1284 // 2b. put all atoms into its corresponding list
1285 Walker = start;
1286 while(Walker->next != end) {
1287 Walker = Walker->next;
1288 //*out << Verbose(1) << "Current atom is " << *Walker << " with coordinates ";
1289 //Walker->x.Output(out);
1290 //*out << "." << endl;
1291 // compute the cell by the atom's coordinates
1292 j=-1;
1293 for (int i=0;i<NDIM;i++) {
1294 j += i+1;
1295 x.CopyVector(&(Walker->x));
1296 x.KeepPeriodic(out, matrix);
1297 n[i] = (int)floor(x.x[i]/cell_size[j]*(double)divisor[i]);
1298 }
1299 index = n[2] + (n[1] + n[0] * divisor[1]) * divisor[2];
1300 *out << Verbose(1) << "Atom " << *Walker << " goes into cell number [" << n[0] << "," << n[1] << "," << n[2] << "] = " << index << "." << endl;
1301 // add copy atom to this cell
1302 if (CellList[index] == NULL) // allocate molecule if not done
1303 CellList[index] = new molecule(elemente);
1304 OtherWalker = CellList[index]->AddCopyAtom(Walker); // add a copy of walker to this atom, father will be walker for later reference
1305 //*out << Verbose(1) << "Copy Atom is " << *OtherWalker << "." << endl;
1306 }
1307 //for (int i=0;i<NumberCells;i++)
1308 //*out << Verbose(1) << "Cell number " << i << ": " << CellList[i] << "." << endl;
1309
1310 // 3a. go through every cell
1311 for (N[0]=0;N[0]<divisor[0];N[0]++)
1312 for (N[1]=0;N[1]<divisor[1];N[1]++)
1313 for (N[2]=0;N[2]<divisor[2];N[2]++) {
1314 Index = N[2] + (N[1] + N[0] * divisor[1]) * divisor[2];
1315 if (CellList[Index] != NULL) { // if there atoms in this cell
1316 //*out << Verbose(1) << "Current cell is " << Index << "." << endl;
1317 // 3b. for every atom therein
1318 Walker = CellList[Index]->start;
1319 while (Walker->next != CellList[Index]->end) { // go through every atom
1320 Walker = Walker->next;
1321 //*out << Verbose(0) << "Current Atom is " << *Walker << "." << endl;
1322 // 3c. check for possible bond between each atom in this and every one in the 27 cells
1323 for (n[0]=-1;n[0]<=1;n[0]++)
1324 for (n[1]=-1;n[1]<=1;n[1]++)
1325 for (n[2]=-1;n[2]<=1;n[2]++) {
1326 // compute the index of this comparison cell and make it periodic
1327 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];
1328 //*out << Verbose(1) << "Number of comparison cell is " << index << "." << endl;
1329 if (CellList[index] != NULL) { // if there are any atoms in this cell
1330 OtherWalker = CellList[index]->start;
1331 while(OtherWalker->next != CellList[index]->end) { // go through every atom in this cell
1332 OtherWalker = OtherWalker->next;
1333 //*out << Verbose(0) << "Current comparison atom is " << *OtherWalker << "." << endl;
1334 /// \todo periodic check is missing here!
1335 //*out << Verbose(1) << "Checking distance " << OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size) << " against typical bond length of " << bonddistance*bonddistance << "." << endl;
1336 MinDistance = OtherWalker->type->CovalentRadius + Walker->type->CovalentRadius;
1337 MaxDistance = MinDistance + BONDTHRESHOLD;
1338 MinDistance -= BONDTHRESHOLD;
1339 distance = OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size);
1340 if ((OtherWalker->father->nr > Walker->father->nr) && (distance <= MaxDistance*MaxDistance) && (distance >= MinDistance*MinDistance)) { // create bond if distance is smaller
1341 *out << Verbose(0) << "Adding Bond between " << *Walker << " and " << *OtherWalker << "." << endl;
1342 AddBond(Walker->father, OtherWalker->father, 1); // also increases molecule::BondCount
1343 } else {
1344 //*out << Verbose(1) << "Not Adding: Wrong label order or distance too great." << endl;
1345 }
1346 }
1347 }
1348 }
1349 }
1350 }
1351 }
1352 // 4. free the cell again
1353 for (int i=0;i<NumberCells;i++)
1354 if (CellList[i] != NULL) {
1355 delete(CellList[i]);
1356 }
1357 Free((void **)&CellList, "molecule::CreateAdjacencyList - ** CellList");
1358
1359 // create the adjacency list per atom
1360 CreateListOfBondsPerAtom(out);
1361
1362 // correct Bond degree of each bond by checking of updated(!) sum of bond degree for an atom match its valence count
1363 // bond degrres are correctled iteratively by one, so that 2-2 instead of 1-3 or 3-1 corrections are favoured: We want
1364 // a rather symmetric distribution of higher bond degrees
1365 if (BondCount != 0) {
1366 NoCyclicBonds = 0;
1367 *out << Verbose(1) << "correct Bond degree of each bond" << endl;
1368 do {
1369 No = 0; // No acts as breakup flag (if 1 we still continue)
1370 Walker = start;
1371 while (Walker->next != end) { // go through every atom
1372 Walker = Walker->next;
1373 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through each of its bond partners
1374 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
1375 // count valence of first partner (updated!), might have changed during last bond partner
1376 NoBonds = 0;
1377 for(j=0;j<NumberOfBondsPerAtom[Walker->nr];j++)
1378 NoBonds += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
1379 *out << Verbose(3) << "Walker: " << (int)Walker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
1380 if ((int)(Walker->type->NoValenceOrbitals) > NoBonds) { // we have a mismatch, check NoBonds of other atom
1381 // count valence of second partner
1382 NoBonds = 0;
1383 for(j=0;j<NumberOfBondsPerAtom[OtherWalker->nr];j++)
1384 NoBonds += ListOfBondsPerAtom[OtherWalker->nr][j]->BondDegree;
1385 *out << Verbose(3) << "OtherWalker: " << (int)OtherWalker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
1386 if ((int)(OtherWalker->type->NoValenceOrbitals) > NoBonds) // increase bond degree by one
1387 ListOfBondsPerAtom[Walker->nr][i]->BondDegree++;
1388 }
1389 }
1390 }
1391 } while (No);
1392
1393 } else
1394 *out << Verbose(1) << "BondCount is " << BondCount << ", no bonds between any of the " << AtomCount << " atoms." << endl;
1395 *out << Verbose(1) << "I detected " << BondCount << " bonds in the molecule with distance " << bonddistance << "." << endl;
1396
1397 // output bonds for debugging (if bond chain list was correctly installed)
1398 *out << Verbose(1) << endl << "From contents of bond chain list:";
1399 Binder = first;
1400 while(Binder->next != last) {
1401 Binder = Binder->next;
1402 *out << *Binder << "\t" << endl;
1403 }
1404 *out << endl;
1405 } else
1406 *out << Verbose(1) << "AtomCount is " << AtomCount << ", thus no bonds, no connections!." << endl;
1407 *out << Verbose(0) << "End of CreateAdjacencyList." << endl;
1408 Free((void **)&matrix, "molecule::CreateAdjacencyList: *matrix");
1409};
1410
1411/** Performs a Depth-First search on this molecule.
1412 * Marks bonds in molecule as cyclic, bridge, ... and atoms as
1413 * articulations points, ...
1414 * We use the algorithm from [Even, Graph Algorithms, p.62].
1415 * \param *out output stream for debugging
1416 * \param ReturnStack true - return pointer to atom stack of separable components, false - return NULL
1417 * \param MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found
1418 * \return list of each disconnected subgraph as an individual molecule class structure
1419 */
1420MoleculeLeafClass * molecule::DepthFirstSearchAnalysis(ofstream *out, bool ReturnStack, int &MinimumRingSize)
1421{
1422 AtomStackClass *AtomStack = new AtomStackClass(AtomCount);
1423 MoleculeLeafClass *SubGraphs = new MoleculeLeafClass(NULL);
1424 MoleculeLeafClass *LeafWalker = SubGraphs;
1425 int CurrentGraphNr = 0, OldGraphNr;
1426 int CyclicBonds;
1427 int ComponentNumber = 0;
1428 atom *Walker = NULL, *OtherAtom = NULL, *Root = start->next;
1429 bond *Binder = NULL;
1430 bool BackStepping = false;
1431
1432 *out << Verbose(0) << "Begin of DepthFirstSearchAnalysis" << endl;
1433
1434 ResetAllBondsToUnused();
1435 ResetAllAtomNumbers();
1436 InitComponentNumbers();
1437 while (Root != end) { // if there any atoms at all
1438 // (1) mark all edges unused, empty stack, set atom->GraphNr = 0 for all
1439 AtomStack->ClearStack();
1440
1441 // put into new subgraph molecule and add this to list of subgraphs
1442 LeafWalker = new MoleculeLeafClass(LeafWalker);
1443 LeafWalker->Leaf = new molecule(elemente);
1444 LeafWalker->Leaf->AddCopyAtom(Root);
1445
1446 OldGraphNr = CurrentGraphNr;
1447 Walker = Root;
1448 do { // (10)
1449 do { // (2) set number and Lowpoint of Atom to i, increase i, push current atom
1450 if (!BackStepping) { // if we don't just return from (8)
1451 Walker->GraphNr = CurrentGraphNr;
1452 Walker->LowpointNr = CurrentGraphNr;
1453 *out << Verbose(1) << "Setting Walker[" << Walker->Name << "]'s number to " << Walker->GraphNr << " with Lowpoint " << Walker->LowpointNr << "." << endl;
1454 AtomStack->Push(Walker);
1455 CurrentGraphNr++;
1456 }
1457 do { // (3) if Walker has no unused egdes, go to (5)
1458 BackStepping = false; // reset backstepping flag for (8)
1459 if (Binder == NULL) // if we don't just return from (11), Binder is already set to next unused
1460 Binder = FindNextUnused(Walker);
1461 if (Binder == NULL)
1462 break;
1463 *out << Verbose(2) << "Current Unused Bond is " << *Binder << "." << endl;
1464 // (4) Mark Binder used, ...
1465 Binder->MarkUsed(black);
1466 OtherAtom = Binder->GetOtherAtom(Walker);
1467 *out << Verbose(2) << "(4) OtherAtom is " << OtherAtom->Name << "." << endl;
1468 if (OtherAtom->GraphNr != -1) {
1469 // (4a) ... if "other" atom has been visited (GraphNr != 0), set lowpoint to minimum of both, go to (3)
1470 Binder->Type = BackEdge;
1471 Walker->LowpointNr = ( Walker->LowpointNr < OtherAtom->GraphNr ) ? Walker->LowpointNr : OtherAtom->GraphNr;
1472 *out << Verbose(3) << "(4a) Visited: Setting Lowpoint of Walker[" << Walker->Name << "] to " << Walker->LowpointNr << "." << endl;
1473 } else {
1474 // (4b) ... otherwise set OtherAtom as Ancestor of Walker and Walker as OtherAtom, go to (2)
1475 Binder->Type = TreeEdge;
1476 OtherAtom->Ancestor = Walker;
1477 Walker = OtherAtom;
1478 *out << Verbose(3) << "(4b) Not Visited: OtherAtom[" << OtherAtom->Name << "]'s Ancestor is now " << OtherAtom->Ancestor->Name << ", Walker is OtherAtom " << OtherAtom->Name << "." << endl;
1479 break;
1480 }
1481 Binder = NULL;
1482 } while (1); // (3)
1483 if (Binder == NULL) {
1484 *out << Verbose(2) << "No more Unused Bonds." << endl;
1485 break;
1486 } else
1487 Binder = NULL;
1488 } while (1); // (2)
1489
1490 // 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!
1491 if ((Walker == Root) && (Binder == NULL))
1492 break;
1493
1494 // (5) if Ancestor of Walker is ...
1495 *out << Verbose(1) << "(5) Number of Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "] is " << Walker->Ancestor->GraphNr << "." << endl;
1496 if (Walker->Ancestor->GraphNr != Root->GraphNr) {
1497 // (6) (Ancestor of Walker is not Root)
1498 if (Walker->LowpointNr < Walker->Ancestor->GraphNr) {
1499 // (6a) set Ancestor's Lowpoint number to minimum of of its Ancestor and itself, go to Step(8)
1500 Walker->Ancestor->LowpointNr = (Walker->Ancestor->LowpointNr < Walker->LowpointNr) ? Walker->Ancestor->LowpointNr : Walker->LowpointNr;
1501 *out << Verbose(2) << "(6) Setting Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s Lowpoint to " << Walker->Ancestor->LowpointNr << "." << endl;
1502 } else {
1503 // (7) (Ancestor of Walker is a separating vertex, remove all from stack till Walker (including), these and Ancestor form a component
1504 Walker->Ancestor->SeparationVertex = true;
1505 *out << Verbose(2) << "(7) Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s is a separating vertex, creating component." << endl;
1506 SetNextComponentNumber(Walker->Ancestor, ComponentNumber);
1507 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Ancestor's Compont is " << ComponentNumber << "." << endl;
1508 SetNextComponentNumber(Walker, ComponentNumber);
1509 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1510 do {
1511 OtherAtom = AtomStack->PopLast();
1512 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
1513 SetNextComponentNumber(OtherAtom, ComponentNumber);
1514 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1515 } while (OtherAtom != Walker);
1516 ComponentNumber++;
1517 }
1518 // (8) Walker becomes its Ancestor, go to (3)
1519 *out << Verbose(2) << "(8) Walker[" << Walker->Name << "] is now its Ancestor " << Walker->Ancestor->Name << ", backstepping. " << endl;
1520 Walker = Walker->Ancestor;
1521 BackStepping = true;
1522 }
1523 if (!BackStepping) { // coming from (8) want to go to (3)
1524 // (9) remove all from stack till Walker (including), these and Root form a component
1525 AtomStack->Output(out);
1526 SetNextComponentNumber(Root, ComponentNumber);
1527 *out << Verbose(3) << "(9) Root[" << Root->Name << "]'s Component is " << ComponentNumber << "." << endl;
1528 SetNextComponentNumber(Walker, ComponentNumber);
1529 *out << Verbose(3) << "(9) Walker[" << Walker->Name << "]'s Component is " << ComponentNumber << "." << endl;
1530 do {
1531 OtherAtom = AtomStack->PopLast();
1532 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
1533 SetNextComponentNumber(OtherAtom, ComponentNumber);
1534 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1535 } while (OtherAtom != Walker);
1536 ComponentNumber++;
1537
1538 // (11) Root is separation vertex, set Walker to Root and go to (4)
1539 Walker = Root;
1540 Binder = FindNextUnused(Walker);
1541 *out << Verbose(1) << "(10) Walker is Root[" << Root->Name << "], next Unused Bond is " << Binder << "." << endl;
1542 if (Binder != NULL) { // Root is separation vertex
1543 *out << Verbose(1) << "(11) Root is a separation vertex." << endl;
1544 Walker->SeparationVertex = true;
1545 }
1546 }
1547 } while ((BackStepping) || (Binder != NULL)); // (10) halt only if Root has no unused edges
1548
1549 // From OldGraphNr to CurrentGraphNr ranges an disconnected subgraph
1550 *out << Verbose(0) << "Disconnected subgraph ranges from " << OldGraphNr << " to " << CurrentGraphNr << "." << endl;
1551 LeafWalker->Leaf->Output(out);
1552 *out << endl;
1553
1554 // step on to next root
1555 while ((Root != end) && (Root->GraphNr != -1)) {
1556 //*out << Verbose(1) << "Current next subgraph root candidate is " << Root->Name << "." << endl;
1557 if (Root->GraphNr != -1) // if already discovered, step on
1558 Root = Root->next;
1559 }
1560 }
1561 // set cyclic bond criterium on "same LP" basis
1562 Binder = first;
1563 while(Binder->next != last) {
1564 Binder = Binder->next;
1565 if (Binder->rightatom->LowpointNr == Binder->leftatom->LowpointNr) { // cyclic ??
1566 Binder->Cyclic = true;
1567 NoCyclicBonds++;
1568 }
1569 }
1570
1571 // correct cyclic bonds that are not included in "same LP" argument
1572 Binder = first;
1573 while (Binder->next != last) {
1574 Binder = Binder->next;
1575 Walker = Binder->leftatom;
1576 OtherAtom = Binder->rightatom;
1577 // now check whether both have a cyclic bond in their list
1578 CyclicBonds = 0; // counts cyclic bonds;
1579 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
1580 if ((CyclicBonds == 0) && (ListOfBondsPerAtom[Walker->nr][i]->Cyclic))
1581 CyclicBonds = 1;
1582 for(int i=0;i<NumberOfBondsPerAtom[OtherAtom->nr];i++)
1583 if ((CyclicBonds == 1) && (ListOfBondsPerAtom[OtherAtom->nr][i]->Cyclic))
1584 CyclicBonds = 2;
1585 Binder->Cyclic = (Binder->Cyclic) || (CyclicBonds == 2); // set the Cyclic criterium either or ...
1586 }
1587
1588 // further analysis of the found cycles (print rings, get minimum cycle length)
1589 CyclicStructureAnalysis(out, MinimumRingSize);
1590 *out << Verbose(1) << "Final graph info for each atom is:" << endl;
1591 Walker = start;
1592 while (Walker->next != end) {
1593 Walker = Walker->next;
1594 *out << Verbose(2) << "Atom " << Walker->Name << " is " << ((Walker->SeparationVertex) ? "a" : "not a") << " separation vertex, components are ";
1595 OutputComponentNumber(out, Walker);
1596 *out << " with Lowpoint " << Walker->LowpointNr << " and Graph Nr. " << Walker->GraphNr << "." << endl;
1597 }
1598
1599 *out << Verbose(1) << "Final graph info for each bond is:" << endl;
1600 Binder = first;
1601 while(Binder->next != last) {
1602 Binder = Binder->next;
1603 *out << Verbose(2) << ((Binder->Type == TreeEdge) ? "TreeEdge " : "BackEdge ") << *Binder << ": <";
1604 *out << ((Binder->leftatom->SeparationVertex) ? "SP," : "") << "L" << Binder->leftatom->LowpointNr << " G" << Binder->leftatom->GraphNr << " Comp.";
1605 OutputComponentNumber(out, Binder->leftatom);
1606 *out << " === ";
1607 *out << ((Binder->rightatom->SeparationVertex) ? "SP," : "") << "L" << Binder->rightatom->LowpointNr << " G" << Binder->rightatom->GraphNr << " Comp.";
1608 OutputComponentNumber(out, Binder->rightatom);
1609 *out << ">." << endl;
1610 if (Binder->Cyclic) // cyclic ??
1611 *out << Verbose(3) << "Lowpoint at each side are equal: CYCLIC!" << endl;
1612 }
1613
1614 // further analysis of the found cycles (print rings, get minimum cycle length)
1615 CyclicStructureAnalysis(out, MinimumRingSize);
1616
1617 // free all and exit
1618 delete(AtomStack);
1619 *out << Verbose(0) << "End of DepthFirstSearchAnalysis" << endl;
1620 return SubGraphs;
1621};
1622
1623/** Analyses the cycles found and returns minimum of all cycle lengths.
1624 * \param *out output stream for debugging
1625 * \param MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found
1626 * \todo BFS from the not-same-LP to find back to starting point of tributary cycle over more than one bond
1627 */
1628void molecule::CyclicStructureAnalysis(ofstream *out, int &MinimumRingSize)
1629{
1630 AtomStackClass *AtomStack = new AtomStackClass(AtomCount);
1631 int LP;
1632 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Runner = NULL;
1633 bond *Binder = NULL;
1634 int RingSize, NumCycles;
1635
1636 // go through every atom
1637 AtomStack->ClearStack();
1638 int *NoCyclicBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CyclicStructureAnalysis: *NoCyclicBondsPerAtom");
1639 Walker = start;
1640 while (Walker->next != end) {
1641 Walker = Walker->next;
1642 NoCyclicBondsPerAtom[Walker->nr] = 0;
1643 // check whether it's connected to cyclic bonds and count per atom
1644 // 0 means not part of a cycle, 2 means in a cycle, 3 or more means interconnection site of cycles
1645 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
1646 Binder = ListOfBondsPerAtom[Walker->nr][i];
1647 NoCyclicBondsPerAtom[Walker->nr] += (int) Binder->Cyclic;
1648 if (NoCyclicBondsPerAtom[Walker->nr] == 3) //push all intersections
1649 AtomStack->Push(Walker);
1650 }
1651 }
1652 *out << Verbose(1) << "NoCyclicBondsPerAtom: ";
1653 for(int i=0;i<AtomCount;i++) {
1654 *out << NoCyclicBondsPerAtom[i] << " ";
1655 }
1656 *out << endl;
1657 *out << Verbose(1) << "Analysing cycles ... " << endl;
1658 MinimumRingSize = -1;
1659 NumCycles = 0;
1660 while (!AtomStack->IsEmpty()) {
1661 Walker = AtomStack->PopFirst();
1662 if (NoCyclicBondsPerAtom[Walker->nr] > 1) {
1663 NoCyclicBondsPerAtom[Walker->nr]--; // remove one for being intersection
1664 RingSize = 0;
1665 *out << Verbose(2) << "Current intersection is " << *Walker << ", expect to find " << NoCyclicBondsPerAtom[Walker->nr] << " cycles." << endl;
1666 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
1667 Binder = ListOfBondsPerAtom[Walker->nr][i];
1668 OtherAtom = Binder->GetOtherAtom(Walker);
1669 // note down the LowPoint number of this cycle
1670 if (NoCyclicBondsPerAtom[OtherAtom->nr] > 1) {
1671 LP = OtherAtom->LowpointNr;
1672 NoCyclicBondsPerAtom[Walker->nr]--; // walker is start of cycle
1673 if (LP != Walker->LowpointNr)
1674 *out << Verbose(2) << "Tributary cycle: ... <-> " << Walker->Name;
1675 else
1676 *out << Verbose(2) << "Main cycle: ... <-> " << Walker->Name;
1677 Root = Walker; // root acts as predecessor marker so that we don't step back accidentally
1678 RingSize = 1;
1679 do {
1680 for(int j=0;j<NumberOfBondsPerAtom[OtherAtom->nr];j++) { // search among its bonds for next in cycle (same lowpoint nr)
1681 Runner = ListOfBondsPerAtom[OtherAtom->nr][j]->GetOtherAtom(OtherAtom);
1682 if (((Runner->LowpointNr == LP) || (Runner->LowpointNr == Walker->LowpointNr)) && (Runner != Root)) {
1683 // first check is to stay in the cycle
1684 // middle check is allow for getting back into main cycle briefly from tributary cycle (just one step, then while further down stops)
1685 // last check is not step back
1686 *out << " <-> " << OtherAtom->Name;
1687 NoCyclicBondsPerAtom[OtherAtom->nr]--;
1688 Root = OtherAtom;
1689 OtherAtom = Runner;
1690 NoCyclicBondsPerAtom[Root->nr]--;
1691 RingSize++;
1692 break;
1693 }
1694 }
1695 } while ((OtherAtom->LowpointNr == LP) && (Walker != OtherAtom) && (Root->LowpointNr == OtherAtom->LowpointNr));
1696 // now check if the LP is different from Walker's, as then there is one more bond to follow
1697 if (LP != Walker->LowpointNr) {
1698 // find last bond to home base
1699 for(int j=0;j<NumberOfBondsPerAtom[OtherAtom->nr];j++)
1700 if (ListOfBondsPerAtom[OtherAtom->nr][j]->GetOtherAtom(OtherAtom) == Root) {
1701 *out << " <-> " << OtherAtom->Name;
1702 RingSize++;
1703 NoCyclicBondsPerAtom[OtherAtom->nr]--;
1704 }
1705 } else {
1706 // we have made the complete cycle
1707 }
1708 *out << " <-> ... with cycle length of " << RingSize << "." << endl;
1709 NumCycles++;
1710 if ((RingSize < MinimumRingSize) || (MinimumRingSize == -1))
1711 MinimumRingSize = RingSize;
1712 }
1713 }
1714 }
1715 }
1716
1717 // print NoCyclicBondsPerAtom to visually check of all are zero
1718 *out << Verbose(1) << "NoCyclicBondsPerAtom: ";
1719 for(int i=0;i<AtomCount;i++) {
1720 if (NoCyclicBondsPerAtom[i] > 0)
1721 cerr << "There was an error in molecule::CyclicStructureAnalysis!" << endl;
1722 *out << NoCyclicBondsPerAtom[i] << " ";
1723 }
1724 *out << endl;
1725
1726 if (MinimumRingSize != -1)
1727 *out << Verbose(1) << "Minimum ring size is " << MinimumRingSize << ", over " << NumCycles << " cycles total." << endl;
1728 else
1729 *out << Verbose(1) << "No rings were detected in the molecular structure." << endl;
1730
1731 Free((void **)&NoCyclicBondsPerAtom, "molecule::CyclicStructureAnalysis: *NoCyclicBondsPerAtom");
1732 delete(AtomStack);
1733};
1734
1735/** Sets the next component number.
1736 * This is O(N) as the number of bonds per atom is bound.
1737 * \param *vertex atom whose next atom::*ComponentNr is to be set
1738 * \param nr number to use
1739 */
1740void molecule::SetNextComponentNumber(atom *vertex, int nr)
1741{
1742 int i=0;
1743 if (vertex != NULL) {
1744 for(;i<NumberOfBondsPerAtom[vertex->nr];i++) {
1745 if (vertex->ComponentNr[i] == -1) { // check if not yet used
1746 vertex->ComponentNr[i] = nr;
1747 break;
1748 }
1749 else if (vertex->ComponentNr[i] == nr) // if number is already present, don't add another time
1750 break; // breaking here will not cause error!
1751 }
1752 if (i == NumberOfBondsPerAtom[vertex->nr])
1753 cerr << "Error: All Component entries are already occupied!" << endl;
1754 } else
1755 cerr << "Error: Given vertex is NULL!" << endl;
1756};
1757
1758/** Output a list of flags, stating whether the bond was visited or not.
1759 * \param *out output stream for debugging
1760 */
1761void molecule::OutputComponentNumber(ofstream *out, atom *vertex)
1762{
1763 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
1764 *out << vertex->ComponentNr[i] << " ";
1765};
1766
1767/** Allocates memory for all atom::*ComponentNr in this molecule and sets each entry to -1.
1768 */
1769void molecule::InitComponentNumbers()
1770{
1771 atom *Walker = start;
1772 while(Walker->next != end) {
1773 Walker = Walker->next;
1774 if (Walker->ComponentNr != NULL)
1775 Free((void **)&Walker->ComponentNr, "molecule::InitComponentNumbers: **Walker->ComponentNr");
1776 Walker->ComponentNr = (int *) Malloc(sizeof(int)*NumberOfBondsPerAtom[Walker->nr], "molecule::InitComponentNumbers: *Walker->ComponentNr");
1777 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
1778 Walker->ComponentNr[i] = -1;
1779 }
1780};
1781
1782/** Returns next unused bond for this atom \a *vertex or NULL of none exists.
1783 * \param *vertex atom to regard
1784 * \return bond class or NULL
1785 */
1786bond * molecule::FindNextUnused(atom *vertex)
1787{
1788 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
1789 if (ListOfBondsPerAtom[vertex->nr][i]->IsUsed() == white)
1790 return(ListOfBondsPerAtom[vertex->nr][i]);
1791 return NULL;
1792};
1793
1794/** Resets bond::Used flag of all bonds in this molecule.
1795 * \return true - success, false - -failure
1796 */
1797void molecule::ResetAllBondsToUnused()
1798{
1799 bond *Binder = first;
1800 while (Binder->next != last) {
1801 Binder = Binder->next;
1802 Binder->ResetUsed();
1803 }
1804};
1805
1806/** Resets atom::nr to -1 of all atoms in this molecule.
1807 */
1808void molecule::ResetAllAtomNumbers()
1809{
1810 atom *Walker = start;
1811 while (Walker->next != end) {
1812 Walker = Walker->next;
1813 Walker->GraphNr = -1;
1814 }
1815};
1816
1817/** Output a list of flags, stating whether the bond was visited or not.
1818 * \param *out output stream for debugging
1819 * \param *list
1820 */
1821void OutputAlreadyVisited(ofstream *out, int *list)
1822{
1823 *out << Verbose(4) << "Already Visited Bonds:\t";
1824 for(int i=1;i<=list[0];i++) *out << Verbose(0) << list[i] << " ";
1825 *out << endl;
1826};
1827
1828/** Estimates by educated guessing (using upper limit) the expected number of fragments.
1829 * The upper limit is
1830 * \f[
1831 * n = N \cdot C^k
1832 * \f]
1833 * where \f$C=2^c\f$ and c is the maximum bond degree over N number of atoms.
1834 * \param *out output stream for debugging
1835 * \param order bond order k
1836 * \return number n of fragments
1837 */
1838int molecule::GuesstimateFragmentCount(ofstream *out, int order)
1839{
1840 int c = 0;
1841 int FragmentCount;
1842 // get maximum bond degree
1843 atom *Walker = start;
1844 while (Walker->next != end) {
1845 Walker = Walker->next;
1846 c = (NumberOfBondsPerAtom[Walker->nr] > c) ? NumberOfBondsPerAtom[Walker->nr] : c;
1847 }
1848 FragmentCount = NoNonHydrogen*(1 << (c*order));
1849 *out << Verbose(1) << "Upper limit for this subgraph is " << FragmentCount << " for " << NoNonHydrogen << " non-H atoms with maximum bond degree of " << c << "." << endl;
1850 return FragmentCount;
1851};
1852
1853/** Scans a single line for number and puts them into \a KeySet.
1854 * \param *out output stream for debugging
1855 * \param *buffer buffer to scan
1856 * \param &CurrentSet filled KeySet on return
1857 * \return true - at least one valid atom id parsed, false - CurrentSet is empty
1858 */
1859bool molecule::ScanBufferIntoKeySet(ofstream *out, char *buffer, KeySet &CurrentSet)
1860{
1861 stringstream line;
1862 int AtomNr;
1863 int status = 0;
1864
1865 line.str(buffer);
1866 while (!line.eof()) {
1867 line >> AtomNr;
1868 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
1869 CurrentSet.insert(AtomNr); // insert at end, hence in same order as in file!
1870 status++;
1871 } // else it's "-1" or else and thus must not be added
1872 }
1873 *out << Verbose(1) << "The scanned KeySet is ";
1874 for(KeySet::iterator runner = CurrentSet.begin(); runner != CurrentSet.end(); runner++) {
1875 *out << (*runner) << "\t";
1876 }
1877 *out << endl;
1878 return (status != 0);
1879};
1880
1881/** Parses the KeySet file and fills \a *FragmentList from the known molecule structure.
1882 * \param *out output stream for debugging
1883 * \param *path path to file
1884 * \param **ListOfAtoms allocated (molecule::AtomCount) and filled lookup table for ids (Atom::nr) to *Atom
1885 * \param *FragmentList NULL, filled on return
1886 * \param IsAngstroem whether we have Ansgtroem or bohrradius
1887 * \return true - parsing successfully, false - failure on parsing (FragmentList will be NULL)
1888 */
1889bool molecule::ParseKeySetFile(ofstream *out, char *path, atom **ListOfAtoms, MoleculeListClass *&FragmentList, bool IsAngstroem)
1890{
1891 bool status = true;
1892 ifstream KeySetFile;
1893 stringstream line;
1894 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - filename");
1895
1896 if (FragmentList != NULL) { // check list pointer
1897 cerr << "Error: FragmentList was not NULL as supposed to be, already atoms present therein?" << endl;
1898 return false;
1899 }
1900 cout << Verbose(1) << "Parsing the KeySet file ... " << endl;
1901 // open file and read
1902 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, KEYSETFILE);
1903 KeySetFile.open(filename);
1904 if (KeySetFile != NULL) {
1905 // each line represents a new fragment
1906 int NumberOfFragments = 0;
1907 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - *buffer");
1908 // 1. scan through file to know number of fragments
1909 while (!KeySetFile.eof()) {
1910 KeySetFile.getline(buffer, MAXSTRINGSIZE);
1911 if (strlen(buffer) > 0) // there is at least on possible number on the parsed line
1912 NumberOfFragments++;
1913 }
1914 // 2. allocate the MoleculeListClass accordingly
1915 FragmentList = new MoleculeListClass(NumberOfFragments, AtomCount);
1916 // 3. Clear File, go to beginning and parse again, now adding found ids to each keyset and converting into molecules
1917 KeySetFile.clear();
1918 KeySetFile.seekg(ios::beg);
1919 NumberOfFragments = 0;
1920 while ((!KeySetFile.eof()) && (FragmentList->NumberOfMolecules > NumberOfFragments)) {
1921 KeySetFile.getline(buffer, MAXSTRINGSIZE);
1922 KeySet CurrentSet;
1923 if ((strlen(buffer) > 0) && (ScanBufferIntoKeySet(out, buffer, CurrentSet))) // if at least one valid atom was added, write config
1924 FragmentList->ListOfMolecules[NumberOfFragments++] = StoreFragmentFromKeySet(out, CurrentSet, IsAngstroem);
1925 }
1926 // 4. Free and done
1927 Free((void **)&buffer, "molecule::ParseKeySetFile - *buffer");
1928 cout << "done." << endl;
1929 } else {
1930 cout << "File not found." << endl;
1931 status = false;
1932 }
1933 Free((void **)&filename, "molecule::ParseKeySetFile - filename");
1934
1935 return status;
1936};
1937
1938/** Storing the bond structure of a molecule to file.
1939 * Simply stores Atom::nr and then the Atom::nr of all bond partners per line.
1940 * \param *out output stream for debugging
1941 * \param *path path to file
1942 * \return true - file written successfully, false - writing failed
1943 */
1944bool molecule::StoreAdjacencyToFile(ofstream *out, char *path)
1945{
1946 ofstream AdjacencyFile;
1947 atom *Walker = NULL;
1948 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::StoreAdjacencyToFile - filename");
1949 bool status = true;
1950
1951 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, ADJACENCYFILE);
1952 AdjacencyFile.open(filename);
1953 cout << Verbose(1) << "Saving adjacency list ... ";
1954 if (AdjacencyFile != NULL) {
1955 Walker = start;
1956 while(Walker->next != end) {
1957 Walker = Walker->next;
1958 AdjacencyFile << Walker->nr << "\t";
1959 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
1960 AdjacencyFile << ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker)->nr << "\t";
1961 AdjacencyFile << endl;
1962 }
1963 AdjacencyFile.close();
1964 cout << "done." << endl;
1965 } else {
1966 cout << "failed." << endl;
1967 status = false;
1968 }
1969 Free((void **)&filename, "molecule::StoreAdjacencyToFile - filename");
1970
1971 return status;
1972};
1973
1974/** Checks contents of adjacency file against bond structure in structure molecule.
1975 * \param *out output stream for debugging
1976 * \param *path path to file
1977 * \param **ListOfAtoms allocated (molecule::AtomCount) and filled lookup table for ids (Atom::nr) to *Atom
1978 * \return true - structure is equal, false - not equivalence
1979 */
1980bool molecule::CheckAdjacencyFileAgainstMolecule(ofstream *out, char *path, atom **ListOfAtoms)
1981{
1982 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckAdjacencyFileAgainstMolecule - filename");
1983 ifstream File;
1984 bool status = true;
1985
1986 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, ADJACENCYFILE);
1987 File.open(filename);
1988 *out << Verbose(1) << "Looking at bond structure stored in adjacency file and comparing to present one ...";
1989 if (File != NULL) {
1990 // allocate storage structure
1991 int NonMatchNumber = 0; // will number of atoms with differing bond structure
1992 int *CurrentBonds = (int *) Malloc(sizeof(int)*8, "molecule::CheckAdjacencyFileAgainstMolecule - CurrentBonds"); // contains parsed bonds of current atom
1993 int CurrentBondsOfAtom;
1994
1995 // Parse the file line by line and count the bonds
1996 while (!File.eof()) {
1997 File.getline(filename, MAXSTRINGSIZE);
1998 stringstream line;
1999 line.str(filename);
2000 int AtomNr = -1;
2001 line >> AtomNr;
2002 CurrentBondsOfAtom = -1; // we count one too far due to line end
2003 // parse into structure
2004 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
2005 while (!line.eof())
2006 line >> CurrentBonds[ ++CurrentBondsOfAtom ];
2007 // compare against present bonds
2008 //cout << Verbose(2) << "Walker is " << *Walker << ", bond partners: ";
2009 if (CurrentBondsOfAtom == NumberOfBondsPerAtom[AtomNr]) {
2010 for(int i=0;i<NumberOfBondsPerAtom[AtomNr];i++) {
2011 int id = ListOfBondsPerAtom[AtomNr][i]->GetOtherAtom(ListOfAtoms[AtomNr])->nr;
2012 int j = 0;
2013 for (;(j<CurrentBondsOfAtom) && (CurrentBonds[j++] != id);); // check against all parsed bonds
2014 if (CurrentBonds[j-1] != id) { // no match ? Then mark in ListOfAtoms
2015 ListOfAtoms[AtomNr] = NULL;
2016 NonMatchNumber++;
2017 status = false;
2018 //out << "[" << id << "]\t";
2019 } else {
2020 //out << id << "\t";
2021 }
2022 }
2023 //out << endl;
2024 } else {
2025 *out << "Number of bonds for Atom " << *ListOfAtoms[AtomNr] << " does not match, parsed " << CurrentBondsOfAtom << " against " << NumberOfBondsPerAtom[AtomNr] << "." << endl;
2026 status = false;
2027 }
2028 }
2029 }
2030 File.close();
2031 File.clear();
2032 if (status) { // if equal we parse the KeySetFile
2033 *out << " done: Equal." << endl;
2034 status = true;
2035 } else
2036 *out << " done: Not equal by " << NonMatchNumber << " atoms." << endl;
2037 Free((void **)&CurrentBonds, "molecule::CheckAdjacencyFileAgainstMolecule - **CurrentBonds");
2038 } else {
2039 *out << " Adjacency file not found." << endl;
2040 status = false;
2041 }
2042 Free((void **)&filename, "molecule::CheckAdjacencyFileAgainstMolecule - filename");
2043
2044 return status;
2045};
2046
2047/** Performs a many-body bond order analysis for a given bond order.
2048 * Writes for each fragment a config file.
2049 * \param *out output stream for debugging
2050 * \param Order up to how many neighbouring bonds a fragment contains in BondOrderScheme::BottumUp scheme
2051 * \param *configuration configuration for writing config files for each fragment
2052 */
2053void molecule::FragmentMolecule(ofstream *out, int Order, config *configuration)
2054{
2055 MoleculeListClass **BondFragments = NULL;
2056 MoleculeListClass *FragmentList = NULL;
2057 atom *Walker = NULL;
2058 int *SortIndex = NULL;
2059 element *runner = NULL;
2060 int AtomNo;
2061 int MinimumRingSize;
2062 int TotalFragmentCounter;
2063 int FragmentCounter;
2064 MoleculeLeafClass *MolecularWalker = NULL;
2065 MoleculeLeafClass *Subgraphs = NULL; // list of subgraphs from DFS analysis
2066 fstream File;
2067 bool FragmentationToDo = true;
2068
2069 *out << endl;
2070#ifdef ADDHYDROGEN
2071 *out << Verbose(0) << "I will treat hydrogen special and saturate dangling bonds with it." << endl;
2072#else
2073 *out << Verbose(0) << "Hydrogen is treated just like the rest of the lot." << endl;
2074#endif
2075
2076 // fill the adjacency list
2077 CreateListOfBondsPerAtom(out);
2078
2079 // === compare it with adjacency file ===
2080 atom **ListOfAtoms = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::FragmentMolecule - **ListOfAtoms");
2081 Walker = start;
2082 while (Walker->next != end) { // create a lookup table (Atom::nr -> atom) used as a marker table lateron
2083 Walker = Walker->next;
2084 if ((Walker->nr >= 0) && (Walker->nr < AtomCount)) {
2085 ListOfAtoms[Walker->nr] = Walker;
2086 } else
2087 break;
2088 }
2089 if (Walker->next != end) { // everything went alright
2090 *out << " range of nuclear ids exceeded [0, AtomCount)." << endl;
2091 FragmentationToDo = false;
2092 }
2093 FragmentationToDo = FragmentationToDo && CheckAdjacencyFileAgainstMolecule(out, configuration->configpath, ListOfAtoms);
2094 if (FragmentationToDo) // NULL entries in ListOfAtoms contain NonMatches
2095 FragmentationToDo = FragmentationToDo && ParseKeySetFile(out, configuration->configpath, ListOfAtoms, FragmentList, configuration->GetIsAngstroem());
2096 if (FragmentationToDo) // parse the adaptive order per atom/site/vertex
2097 FragmentationToDo = FragmentationToDo && ParseOrderAtSiteFromFile(out, configuration->configpath);
2098 Free((void **)&ListOfAtoms, "molecule::FragmentMolecule - **ListOfAtoms");
2099
2100 // =================================== Begin of FRAGMENTATION ===============================
2101 if (FragmentationToDo) { // if we parsed Adjacancy, check whether OrderAtSite is above Order everywhere
2102 Walker = start;
2103 while (Walker->next != end) { // create a lookup table (Atom::nr -> atom) used as a marker table lateron
2104 Walker = Walker->next;
2105#ifdef ADDHYDROGEN
2106 if (Walker->type->Z != 1) // skip hydrogen
2107#endif
2108 if (Walker->AdaptiveOrder < Order)
2109 FragmentationToDo = false;
2110 }
2111 }
2112
2113 if (!FragmentationToDo) { // if we have still do something, FragmentationToDo will be false
2114 // === first perform a DFS analysis to gather info on cyclic structure and a list of disconnected subgraphs ===
2115 Subgraphs = DepthFirstSearchAnalysis((ofstream *)&*out, false, MinimumRingSize);
2116 MolecularWalker = Subgraphs;
2117 // fill the bond structure of the individually stored subgraphs
2118 while (MolecularWalker->next != NULL) {
2119 MolecularWalker = MolecularWalker->next;
2120 *out << Verbose(1) << "Creating adjacency list for subgraph " << MolecularWalker << "." << endl;
2121 MolecularWalker->Leaf->CreateAdjacencyList(out, BondDistance);
2122 MolecularWalker->Leaf->CreateListOfBondsPerAtom(out);
2123 }
2124
2125 // === fragment all subgraphs ===
2126 if ((MinimumRingSize != -1) && (Order >= MinimumRingSize)) {
2127 *out << Verbose(0) << "Bond order greater than or equal to Minimum Ring size of " << MinimumRingSize << " found is not allowed." << endl;
2128 } else {
2129 FragmentCounter = 0;
2130 MolecularWalker = Subgraphs;
2131 // count subgraphs and allocate fragments
2132 while (MolecularWalker->next != NULL) {
2133 MolecularWalker = MolecularWalker->next;
2134 FragmentCounter++;
2135 }
2136 BondFragments = (MoleculeListClass **) Malloc(sizeof(MoleculeListClass *)*FragmentCounter, "molecule::FragmentMolecule - **BondFragments");
2137
2138 // create RootStack for each Subgraph
2139 // NOTE: (keep this extern of following while loop, as lateron we may here look for which site to add to which subgraph)
2140 KeyStack RootStack[FragmentCounter];
2141
2142 FragmentCounter = 0;
2143 MolecularWalker = Subgraphs;
2144 // count subgraphs and allocate fragments
2145 while (MolecularWalker->next != NULL) {
2146 MolecularWalker = MolecularWalker->next;
2147 RootStack[FragmentCounter].clear();
2148 // find first root candidates
2149 Walker = MolecularWalker->Leaf->start;
2150 while (Walker->next != MolecularWalker->Leaf->end) { // go through all (non-hydrogen) atoms
2151 Walker = Walker->next;
2152 #ifdef ADDHYDROGEN
2153 if (Walker->type->Z != 1) // skip hydrogen
2154 #endif
2155 if (Walker->GetTrueFather()->AdaptiveOrder < Order) // only if Order is still greater
2156 RootStack[FragmentCounter].push_front(Walker->nr);
2157 }
2158 FragmentCounter++;
2159 }
2160
2161 // fill the bond fragment list
2162 FragmentCounter = 0;
2163 TotalFragmentCounter = 0;
2164 MolecularWalker = Subgraphs;
2165 while (MolecularWalker->next != NULL) {
2166 MolecularWalker = MolecularWalker->next;
2167 *out << Verbose(1) << "Fragmenting subgraph " << MolecularWalker << "." << endl;
2168 if (MolecularWalker->Leaf->first->next != MolecularWalker->Leaf->last) {
2169 // output ListOfBondsPerAtom for debugging
2170 *out << Verbose(0) << endl << "From Contents of ListOfBondsPerAtom, all non-hydrogen atoms:" << endl;
2171 Walker = MolecularWalker->Leaf->start;
2172 while (Walker->next != MolecularWalker->Leaf->end) {
2173 Walker = Walker->next;
2174 #ifdef ADDHYDROGEN
2175 if (Walker->type->Z != 1) { // regard only non-hydrogen
2176 #endif
2177 *out << Verbose(0) << "Atom " << Walker->Name << " has Bonds: "<<endl;
2178 for(int j=0;j<MolecularWalker->Leaf->NumberOfBondsPerAtom[Walker->nr];j++) {
2179 *out << Verbose(1) << *(MolecularWalker->Leaf->ListOfBondsPerAtom)[Walker->nr][j] << endl;
2180 }
2181 #ifdef ADDHYDROGEN
2182 }
2183 #endif
2184 }
2185 *out << endl;
2186
2187 *out << Verbose(0) << endl << " ========== BOND ENERGY ========================= " << endl;
2188 *out << Verbose(0) << "Begin of bond fragmentation." << endl;
2189 BondFragments[FragmentCounter] = NULL;
2190 // call BOSSANOVA method
2191 Graph *FragmentList = MolecularWalker->Leaf->FragmentBOSSANOVA(out, RootStack[FragmentCounter]);
2192
2193 // allocate memory for the pointer array and transmorph graphs into full molecular fragments
2194 int TotalNumberOfMolecules = 0;
2195 for(Graph::iterator runner = FragmentList->begin(); runner != FragmentList->end(); runner++)
2196 TotalNumberOfMolecules++;
2197 BondFragments[FragmentCounter] = new MoleculeListClass(TotalNumberOfMolecules, AtomCount);
2198 int k=0;
2199 for(Graph::iterator runner = FragmentList->begin(); runner != FragmentList->end(); runner++) {
2200 KeySet test = (*runner).first;
2201 *out << "Fragment No." << (*runner).second.first << " was created " << (int)(*runner).second.second << " time(s)." << endl;
2202 BondFragments[FragmentCounter]->ListOfMolecules[k] = MolecularWalker->Leaf->StoreFragmentFromKeySet(out, test, configuration);
2203 k++;
2204 }
2205 *out << k << "/" << BondFragments[FragmentCounter]->NumberOfMolecules << " fragments generated from the keysets." << endl;
2206 } else {
2207 *out << Verbose(0) << "Connection matrix has not yet been generated!" << endl;
2208 }
2209 TotalFragmentCounter += BondFragments[FragmentCounter]->NumberOfMolecules;
2210 FragmentCounter++; // next fragment list
2211 }
2212 }
2213
2214 // === combine bond fragments list into a single one ===
2215 FragmentList = new MoleculeListClass(TotalFragmentCounter, AtomCount);
2216 TotalFragmentCounter = 0;
2217 for (int i=0;i<FragmentCounter;i++) {
2218 for(int j=0;j<BondFragments[i]->NumberOfMolecules;j++) {
2219 FragmentList->ListOfMolecules[TotalFragmentCounter] = BondFragments[i]->ListOfMolecules[j];
2220 BondFragments[i]->ListOfMolecules[j] = NULL;
2221 TotalFragmentCounter++;
2222 }
2223 delete(BondFragments[i]);
2224 }
2225 Free((void **)&BondFragments, "molecule::FragmentMolecule - **BondFragments");
2226 } else
2227 *out << Verbose(0) << "Nothing to do, using only fragments reconstructed from the KeySetFile." << endl;
2228 // ==================================== End of FRAGMENTATION ================================
2229
2230 // === Save fragments' configuration to disk ===
2231 if (FragmentList != NULL) {
2232 // create a SortIndex to map from BFS labels to the sequence in which the atoms are given in the config file
2233 SortIndex = (int *) Malloc(sizeof(int)*AtomCount, "molecule::FragmentMolecule: *SortIndex");
2234 for(int i=0;i<AtomCount;i++)
2235 SortIndex[i] = -1;
2236 runner = elemente->start;
2237 AtomNo = 0;
2238 while (runner->next != elemente->end) { // go through every element
2239 runner = runner->next;
2240 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
2241 Walker = start;
2242 while (Walker->next != end) { // go through every atom of this element
2243 Walker = Walker->next;
2244 if (Walker->type->Z == runner->Z) // if this atom fits to element
2245 SortIndex[Walker->nr] = AtomNo++;
2246 }
2247 }
2248 }
2249 *out << Verbose(1) << "Writing " << FragmentList->NumberOfMolecules << " possible bond fragmentation configs" << endl;
2250 if (FragmentList->OutputConfigForListOfFragments(out, configuration, SortIndex))
2251 *out << Verbose(1) << "All configs written." << endl;
2252 else
2253 *out << Verbose(1) << "Some configs' writing failed." << endl;
2254 Free((void **)&SortIndex, "molecule::FragmentMolecule: *SortIndex");
2255
2256 // === store Adjacency file ===
2257 StoreAdjacencyToFile(out, configuration->configpath);
2258
2259 // Store adaptive orders into file
2260 StoreOrderAtSiteFile(out, configuration->configpath);
2261
2262 // restore orbital and Stop values
2263 CalculateOrbitals(*configuration);
2264
2265 // free memory for bond part
2266 *out << Verbose(1) << "Freeing bond memory" << endl;
2267 delete(FragmentList); // remove bond molecule from memory
2268 FragmentList = NULL;
2269 } else
2270 *out << Verbose(1) << "FragmentList is zero on return, splitting failed." << endl;
2271
2272 // free subgraph memory again
2273 if (Subgraphs != NULL) {
2274 while (Subgraphs->next != NULL) {
2275 Subgraphs = Subgraphs->next;
2276 delete(Subgraphs->previous);
2277 }
2278 delete(Subgraphs);
2279 }
2280
2281 *out << Verbose(0) << "End of bond fragmentation." << endl;
2282};
2283
2284/** Stores pairs (Atom::nr, Atom::AdaptiveOrder) into file.
2285 * Atoms not present in the file get "-1".
2286 * \param *out output stream for debugging
2287 * \param *path path to file ORDERATSITEFILE
2288 * \return true - file writable, false - not writable
2289 */
2290bool molecule::StoreOrderAtSiteFile(ofstream *out, char *path)
2291{
2292 stringstream line;
2293 ofstream file;
2294
2295 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
2296 file.open(line.str().c_str());
2297 *out << Verbose(0) << "Writing OrderAtSite " << ORDERATSITEFILE << " ... " << endl;
2298 if (file != NULL) {
2299 atom *Walker = start;
2300 while (Walker->next != end) {
2301 Walker = Walker->next;
2302 file << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << endl;
2303 *out << Verbose(2) << "Storing: " << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "." << endl;
2304 }
2305 file.close();
2306 return true;
2307 } else {
2308 return false;
2309 }
2310};
2311
2312/** Parses pairs(Atom::nr, Atom::AdaptiveOrder) from file and stores in molecule's Atom's.
2313 * Atoms not present in the file get "0".
2314 * \param *out output stream for debugging
2315 * \param *path path to file ORDERATSITEFILEe
2316 * \return true - file found and scanned, false - file not found
2317 * \sa ParseKeySetFile() and CheckAdjacencyFileAgainstMolecule() as this is meant to be used in conjunction with the two
2318 */
2319bool molecule::ParseOrderAtSiteFromFile(ofstream *out, char *path)
2320{
2321 unsigned char *OrderArray = (unsigned char *) Malloc(sizeof(unsigned char)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
2322 bool status;
2323 int AtomNr;
2324 stringstream line;
2325 ifstream file;
2326 int Order;
2327
2328 *out << Verbose(1) << "Begin of ParseOrderAtSiteFromFile" << endl;
2329 for(int i=0;i<AtomCount;i++)
2330 OrderArray[i] = 0;
2331 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
2332 file.open(line.str().c_str());
2333 if (file != NULL) {
2334 for (int i=0;i<AtomCount;i++) // initialise with 0
2335 OrderArray[i] = 0;
2336 while (!file.eof()) { // parse from file
2337 file >> AtomNr;
2338 file >> Order;
2339 OrderArray[AtomNr] = (unsigned char) Order;
2340 //*out << Verbose(2) << "AtomNr " << AtomNr << " with order " << (int)OrderArray[AtomNr] << "." << endl;
2341 }
2342 atom *Walker = start;
2343 while (Walker->next != end) { // fill into atom classes
2344 Walker = Walker->next;
2345 Walker->AdaptiveOrder = OrderArray[Walker->nr];
2346 *out << Verbose(2) << *Walker << " gets order " << (int)Walker->AdaptiveOrder << "." << endl;
2347 }
2348 file.close();
2349 status = true;
2350 } else {
2351 status = false;
2352 }
2353 Free((void **)&OrderArray, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
2354
2355 *out << Verbose(1) << "End of ParseOrderAtSiteFromFile" << endl;
2356 return status;
2357};
2358
2359/** Creates an 2d array of pointer with an entry for each atom and each bond it has.
2360 * Updates molecule::ListOfBondsPerAtom, molecule::NumberOfBondsPerAtom by parsing through
2361 * bond chain list, using molecule::AtomCount and molecule::BondCount.
2362 * Allocates memory, fills the array and exits
2363 * \param *out output stream for debugging
2364 */
2365void molecule::CreateListOfBondsPerAtom(ofstream *out)
2366{
2367 bond *Binder = NULL;
2368 atom *Walker = NULL;
2369 int TotalDegree;
2370 *out << Verbose(1) << "Begin of Creating ListOfBondsPerAtom: AtomCount = " << AtomCount << "\tBondCount = " << BondCount << "\tNoNonBonds = " << NoNonBonds << "." << endl;
2371
2372 // re-allocate memory
2373 *out << Verbose(2) << "(Re-)Allocating memory." << endl;
2374 if (ListOfBondsPerAtom != NULL) {
2375 for(int i=0;i<AtomCount;i++)
2376 Free((void **)&ListOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom[i]");
2377 Free((void **)&ListOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom");
2378 }
2379 if (NumberOfBondsPerAtom != NULL)
2380 Free((void **)&NumberOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: NumberOfBondsPerAtom");
2381 ListOfBondsPerAtom = (bond ***) Malloc(sizeof(bond **)*AtomCount, "molecule::CreateListOfBondsPerAtom: ***ListOfBondsPerAtom");
2382 NumberOfBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfBondsPerAtom: *NumberOfBondsPerAtom");
2383
2384 // reset bond counts per atom
2385 for(int i=0;i<AtomCount;i++)
2386 NumberOfBondsPerAtom[i] = 0;
2387 // count bonds per atom
2388 Binder = first;
2389 while (Binder->next != last) {
2390 Binder = Binder->next;
2391 NumberOfBondsPerAtom[Binder->leftatom->nr]++;
2392 NumberOfBondsPerAtom[Binder->rightatom->nr]++;
2393 }
2394 // allocate list of bonds per atom
2395 for(int i=0;i<AtomCount;i++)
2396 ListOfBondsPerAtom[i] = (bond **) Malloc(sizeof(bond *)*NumberOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: **ListOfBondsPerAtom[]");
2397 // clear the list again, now each NumberOfBondsPerAtom marks current free field
2398 for(int i=0;i<AtomCount;i++)
2399 NumberOfBondsPerAtom[i] = 0;
2400 // fill the list
2401 Binder = first;
2402 while (Binder->next != last) {
2403 Binder = Binder->next;
2404 ListOfBondsPerAtom[Binder->leftatom->nr][NumberOfBondsPerAtom[Binder->leftatom->nr]++] = Binder;
2405 ListOfBondsPerAtom[Binder->rightatom->nr][NumberOfBondsPerAtom[Binder->rightatom->nr]++] = Binder;
2406 }
2407
2408 // output list for debugging
2409 *out << Verbose(3) << "ListOfBondsPerAtom for each atom:" << endl;
2410 Walker = start;
2411 while (Walker->next != end) {
2412 Walker = Walker->next;
2413 *out << Verbose(4) << "Atom " << Walker->Name << " with " << NumberOfBondsPerAtom[Walker->nr] << " bonds: ";
2414 TotalDegree = 0;
2415 for (int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
2416 *out << *ListOfBondsPerAtom[Walker->nr][j] << "\t";
2417 TotalDegree += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
2418 }
2419 *out << " -- TotalDegree: " << TotalDegree << endl;
2420 }
2421 *out << Verbose(1) << "End of Creating ListOfBondsPerAtom." << endl << endl;
2422};
2423
2424/** Adds atoms up to \a BondCount distance from \a *Root and notes them down in \a **AddedAtomList.
2425 * Gray vertices are always enqueued in an AtomStackClass FIFO queue, the rest is usual BFS with adding vertices found was
2426 * white and putting into queue.
2427 * \param *out output stream for debugging
2428 * \param *Mol Molecule class to add atoms to
2429 * \param **AddedAtomList list with added atom pointers, index is atom father's number
2430 * \param **AddedBondList list with added bond pointers, index is bond father's number
2431 * \param *Root root vertex for BFS
2432 * \param *Bond bond not to look beyond
2433 * \param BondOrder maximum distance for vertices to add
2434 * \param IsAngstroem lengths are in angstroem or bohrradii
2435 */
2436void molecule::BreadthFirstSearchAdd(ofstream *out, molecule *Mol, atom **&AddedAtomList, bond **&AddedBondList, atom *Root, bond *Bond, int BondOrder, bool IsAngstroem)
2437{
2438 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::BreadthFirstSearchAdd: **PredecessorList");
2439 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::BreadthFirstSearchAdd: *ShortestPathList");
2440 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::BreadthFirstSearchAdd: *ColorList");
2441 AtomStackClass *AtomStack = new AtomStackClass(AtomCount);
2442 atom *Walker = NULL, *OtherAtom = NULL;
2443 bond *Binder = NULL;
2444
2445 // add Root if not done yet
2446 AtomStack->ClearStack();
2447 if (AddedAtomList[Root->nr] == NULL) // add Root if not yet present
2448 AddedAtomList[Root->nr] = Mol->AddCopyAtom(Root);
2449 AtomStack->Push(Root);
2450
2451 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
2452 for (int i=0;i<AtomCount;i++) {
2453 PredecessorList[i] = NULL;
2454 ShortestPathList[i] = -1;
2455 if (AddedAtomList[i] != NULL) // mark already present atoms (i.e. Root and maybe others) as visited
2456 ColorList[i] = lightgray;
2457 else
2458 ColorList[i] = white;
2459 }
2460 ShortestPathList[Root->nr] = 0;
2461
2462 // and go on ... Queue always contains all lightgray vertices
2463 while (!AtomStack->IsEmpty()) {
2464 // we have to pop the oldest atom from stack. This keeps the atoms on the stack always of the same ShortestPath distance.
2465 // 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
2466 // append length of 3 (their neighbours). Thus on stack we have always atoms of a certain length n at bottom of stack and
2467 // followed by n+1 till top of stack.
2468 Walker = AtomStack->PopFirst(); // pop oldest added
2469 *out << Verbose(1) << "Current Walker is: " << Walker->Name << ", and has " << NumberOfBondsPerAtom[Walker->nr] << " bonds." << endl;
2470 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2471 Binder = ListOfBondsPerAtom[Walker->nr][i];
2472 if (Binder != NULL) { // don't look at bond equal NULL
2473 OtherAtom = Binder->GetOtherAtom(Walker);
2474 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2475 if (ColorList[OtherAtom->nr] == white) {
2476 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)
2477 ColorList[OtherAtom->nr] = lightgray;
2478 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2479 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2480 *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;
2481 if ((((ShortestPathList[OtherAtom->nr] < BondOrder) && (Binder != Bond))) ) { // Check for maximum distance
2482 *out << Verbose(3);
2483 if (AddedAtomList[OtherAtom->nr] == NULL) { // add if it's not been so far
2484 AddedAtomList[OtherAtom->nr] = Mol->AddCopyAtom(OtherAtom);
2485 *out << "Added OtherAtom " << OtherAtom->Name;
2486 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2487 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2488 AddedBondList[Binder->nr]->Type = Binder->Type;
2489 *out << " and bond " << *(AddedBondList[Binder->nr]) << ", ";
2490 } 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)
2491 *out << "Not adding OtherAtom " << OtherAtom->Name;
2492 if (AddedBondList[Binder->nr] == NULL) {
2493 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2494 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2495 AddedBondList[Binder->nr]->Type = Binder->Type;
2496 *out << ", added Bond " << *(AddedBondList[Binder->nr]);
2497 } else
2498 *out << ", not added Bond ";
2499 }
2500 *out << ", putting OtherAtom into queue." << endl;
2501 AtomStack->Push(OtherAtom);
2502 } else { // out of bond order, then replace
2503 if ((AddedAtomList[OtherAtom->nr] == NULL) && (Binder->Cyclic))
2504 ColorList[OtherAtom->nr] = white; // unmark if it has not been queued/added, to make it available via its other bonds (cyclic)
2505 if (Binder == Bond)
2506 *out << Verbose(3) << "Not Queueing, is the Root bond";
2507 else if (ShortestPathList[OtherAtom->nr] >= BondOrder)
2508 *out << Verbose(3) << "Not Queueing, is out of Bond Count of " << BondOrder;
2509 if (!Binder->Cyclic)
2510 *out << ", is not part of a cyclic bond, saturating bond with Hydrogen." << endl;
2511 if (AddedBondList[Binder->nr] == NULL) {
2512 if ((AddedAtomList[OtherAtom->nr] != NULL)) { // .. whether we add or saturate
2513 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2514 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2515 AddedBondList[Binder->nr]->Type = Binder->Type;
2516 } else {
2517#ifdef ADDHYDROGEN
2518 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
2519#endif
2520 }
2521 }
2522 }
2523 } else {
2524 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
2525 // This has to be a cyclic bond, check whether it's present ...
2526 if (AddedBondList[Binder->nr] == NULL) {
2527 if ((Binder != Bond) && (Binder->Cyclic) && (((ShortestPathList[Walker->nr]+1) < BondOrder))) {
2528 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2529 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2530 AddedBondList[Binder->nr]->Type = Binder->Type;
2531 } else { // if it's root bond it has to broken (otherwise we would not create the fragments)
2532#ifdef ADDHYDROGEN
2533 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
2534#endif
2535 }
2536 }
2537 }
2538 }
2539 }
2540 ColorList[Walker->nr] = black;
2541 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2542 }
2543 Free((void **)&PredecessorList, "molecule::BreadthFirstSearchAdd: **PredecessorList");
2544 Free((void **)&ShortestPathList, "molecule::BreadthFirstSearchAdd: **ShortestPathList");
2545 Free((void **)&ColorList, "molecule::BreadthFirstSearchAdd: **ColorList");
2546 delete(AtomStack);
2547};
2548
2549/** Adds bond structure to this molecule from \a Father molecule.
2550 * This basically causes this molecule to become an induced subgraph of the \a Father, i.e. for every bond in Father
2551 * with end points present in this molecule, bond is created in this molecule.
2552 * Special care was taken to ensure that this is of complexity O(N), where N is the \a Father's molecule::AtomCount.
2553 * \param *out output stream for debugging
2554 * \param *Father father molecule
2555 * \return true - is induced subgraph, false - there are atoms with fathers not in \a Father
2556 * \todo not checked, not fully working probably
2557 */
2558bool molecule::BuildInducedSubgraph(ofstream *out, const molecule *Father)
2559{
2560 atom *Walker = NULL, *OtherAtom = NULL;
2561 bool status = true;
2562 atom **ParentList = (atom **) Malloc(sizeof(atom *)*Father->AtomCount, "molecule::BuildInducedSubgraph: **ParentList");
2563
2564 *out << Verbose(2) << "Begin of BuildInducedSubgraph." << endl;
2565
2566 // reset parent list
2567 *out << Verbose(3) << "Resetting ParentList." << endl;
2568 for (int i=0;i<Father->AtomCount;i++)
2569 ParentList[i] = NULL;
2570
2571 // fill parent list with sons
2572 *out << Verbose(3) << "Filling Parent List." << endl;
2573 Walker = start;
2574 while (Walker->next != end) {
2575 Walker = Walker->next;
2576 ParentList[Walker->father->nr] = Walker;
2577 // Outputting List for debugging
2578 *out << Verbose(4) << "Son["<< Walker->father->nr <<"] of " << Walker->father << " is " << ParentList[Walker->father->nr] << "." << endl;
2579 }
2580
2581 // check each entry of parent list and if ok (one-to-and-onto matching) create bonds
2582 *out << Verbose(3) << "Creating bonds." << endl;
2583 Walker = Father->start;
2584 while (Walker->next != Father->end) {
2585 Walker = Walker->next;
2586 if (ParentList[Walker->nr] != NULL) {
2587 if (ParentList[Walker->nr]->father != Walker) {
2588 status = false;
2589 } else {
2590 for (int i=0;i<Father->NumberOfBondsPerAtom[Walker->nr];i++) {
2591 OtherAtom = Father->ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
2592 if (ParentList[OtherAtom->nr] != NULL) { // if otheratom is also a father of an atom on this molecule, create the bond
2593 *out << Verbose(4) << "Endpoints of Bond " << Father->ListOfBondsPerAtom[Walker->nr][i] << " are both present: " << ParentList[Walker->nr]->Name << " and " << ParentList[OtherAtom->nr]->Name << "." << endl;
2594 AddBond(ParentList[Walker->nr], ParentList[OtherAtom->nr], Father->ListOfBondsPerAtom[Walker->nr][i]->BondDegree);
2595 }
2596 }
2597 }
2598 }
2599 }
2600
2601 Free((void **)&ParentList, "molecule::BuildInducedSubgraph: **ParentList");
2602 *out << Verbose(2) << "End of BuildInducedSubgraph." << endl;
2603 return status;
2604};
2605
2606
2607/** Looks through a AtomStackClass and returns the likeliest removal candiate.
2608 * \param *out output stream for debugging messages
2609 * \param *&Leaf KeySet to look through
2610 * \param *&ShortestPathList list of the shortest path to decide which atom to suggest as removal candidate in the end
2611 * \param index of the atom suggested for removal
2612 */
2613int molecule::LookForRemovalCandidate(ofstream *&out, KeySet *&Leaf, int *&ShortestPathList)
2614{
2615 atom *Runner = NULL;
2616 int SP, Removal;
2617
2618 *out << Verbose(2) << "Looking for removal candidate." << endl;
2619 SP = -1; //0; // not -1, so that Root is never removed
2620 Removal = -1;
2621 for (KeySet::iterator runner = Leaf->begin(); runner != Leaf->end(); runner++) {
2622 Runner = FindAtom((*runner));
2623 if (Runner->type->Z != 1) { // skip all those added hydrogens when re-filling snake stack
2624 if (ShortestPathList[(*runner)] > SP) { // remove the oldest one with longest shortest path
2625 SP = ShortestPathList[(*runner)];
2626 Removal = (*runner);
2627 }
2628 }
2629 }
2630 return Removal;
2631};
2632
2633/** Stores a fragment from \a KeySet into \a molecule.
2634 * First creates the minimal set of atoms from the KeySet, then creates the bond structure from the complete
2635 * molecule and adds missing hydrogen where bonds were cut.
2636 * \param *out output stream for debugging messages
2637 * \param &Leaflet pointer to KeySet structure
2638 * \param IsAngstroem whether we have Ansgtroem or bohrradius
2639 * \return pointer to constructed molecule
2640 */
2641molecule * molecule::StoreFragmentFromKeySet(ofstream *out, KeySet &Leaflet, bool IsAngstroem)
2642{
2643 atom *Runner = NULL, *FatherOfRunner = NULL, *OtherFather = NULL;
2644 atom **SonList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::StoreFragmentFromStack: **SonList");
2645 molecule *Leaf = new molecule(elemente);
2646
2647// *out << Verbose(1) << "Begin of StoreFragmentFromKeyset." << endl;
2648
2649 Leaf->BondDistance = BondDistance;
2650 for(int i=0;i<NDIM*2;i++)
2651 Leaf->cell_size[i] = cell_size[i];
2652
2653 // initialise SonList (indicates when we need to replace a bond with hydrogen instead)
2654 for(int i=0;i<AtomCount;i++)
2655 SonList[i] = NULL;
2656
2657 // first create the minimal set of atoms from the KeySet
2658 for(KeySet::iterator runner = Leaflet.begin(); runner != Leaflet.end(); runner++) {
2659 FatherOfRunner = FindAtom((*runner)); // find the id
2660 SonList[FatherOfRunner->nr] = Leaf->AddCopyAtom(FatherOfRunner);
2661 }
2662
2663 // create the bonds between all: Make it an induced subgraph and add hydrogen
2664// *out << Verbose(2) << "Creating bonds from father graph (i.e. induced subgraph creation)." << endl;
2665 Runner = Leaf->start;
2666 while (Runner->next != Leaf->end) {
2667 Runner = Runner->next;
2668 FatherOfRunner = Runner->father;
2669 if (SonList[FatherOfRunner->nr] != NULL) { // check if this, our father, is present in list
2670 // create all bonds
2671 for (int i=0;i<NumberOfBondsPerAtom[FatherOfRunner->nr];i++) { // go through every bond of father
2672 OtherFather = ListOfBondsPerAtom[FatherOfRunner->nr][i]->GetOtherAtom(FatherOfRunner);
2673// *out << Verbose(2) << "Father " << *FatherOfRunner << " of son " << *SonList[FatherOfRunner->nr] << " is bound to " << *OtherFather;
2674 if (SonList[OtherFather->nr] != NULL) {
2675// *out << ", whose son is " << *SonList[OtherFather->nr] << "." << endl;
2676 if (OtherFather->nr > FatherOfRunner->nr) { // add bond (nr check is for adding only one of both variants: ab, ba)
2677// *out << Verbose(3) << "Adding Bond: " << Leaf->AddBond(Runner, SonList[OtherFather->nr], ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree) << "." << endl;
2678 //NumBonds[Runner->nr]++;
2679 } else {
2680// *out << Verbose(3) << "Not adding bond, labels in wrong order." << endl;
2681 }
2682 } else {
2683// *out << ", who has no son in this fragment molecule." << endl;
2684#ifdef ADDHYDROGEN
2685// *out << Verbose(3) << "Adding Hydrogen to " << Runner->Name << " and a bond in between." << endl;
2686 Leaf->AddHydrogenReplacementAtom(out, ListOfBondsPerAtom[FatherOfRunner->nr][i], Runner, FatherOfRunner, OtherFather, ListOfBondsPerAtom[FatherOfRunner->nr],NumberOfBondsPerAtom[FatherOfRunner->nr], IsAngstroem);
2687#endif
2688 //NumBonds[Runner->nr] += ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree;
2689 }
2690 }
2691 } else {
2692 *out << Verbose(0) << "ERROR: Son " << Runner->Name << " has father " << FatherOfRunner->Name << " but its entry in SonList is " << SonList[FatherOfRunner->nr] << "!" << endl;
2693 }
2694#ifdef ADDHYDROGEN
2695 while ((Runner->next != Leaf->end) && (Runner->next->type->Z == 1)) // skip added hydrogen
2696 Runner = Runner->next;
2697#endif
2698 }
2699 Leaf->CreateListOfBondsPerAtom(out);
2700 //Leaflet->Leaf->ScanForPeriodicCorrection(out);
2701 Free((void **)&SonList, "molecule::StoreFragmentFromStack: **SonList");
2702// *out << Verbose(1) << "End of StoreFragmentFromKeyset." << endl;
2703 return Leaf;
2704};
2705
2706/** Creates \a MoleculeListClass of all unique fragments of the \a molecule containing \a Order atoms or vertices.
2707 * The picture to have in mind is that of a DFS "snake" of a certain length \a Order, i.e. as in the infamous
2708 * computer game, that winds through the connected graph representing the molecule. Color (white,
2709 * lightgray, darkgray, black) indicates whether a vertex has been discovered so far or not. Labels will help in
2710 * creating only unique fragments and not additional ones with vertices simply in different sequence.
2711 * The Predecessor is always the one that came before in discovering, needed on backstepping. And
2712 * finally, the ShortestPath is needed for removing vertices from the snake stack during the back-
2713 * stepping.
2714 * \param *out output stream for debugging
2715 * \param Order number of atoms in each fragment
2716 * \param *configuration configuration for writing config files for each fragment
2717 * \return List of all unique fragments with \a Order atoms
2718 */
2719/*
2720MoleculeListClass * molecule::CreateListOfUniqueFragmentsOfOrder(ofstream *out, int Order, config *configuration)
2721{
2722 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
2723 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
2724 int *Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
2725 enum Shading *ColorVertexList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
2726 enum Shading *ColorEdgeList = (enum Shading *) Malloc(sizeof(enum Shading)*BondCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorBondList");
2727 AtomStackClass *RootStack = new AtomStackClass(AtomCount);
2728 AtomStackClass *TouchedStack = new AtomStackClass((int)pow(4,Order)+2); // number of atoms reached from one with maximal 4 bonds plus Root itself
2729 AtomStackClass *SnakeStack = new AtomStackClass(Order+1); // equal to Order is not possible, as then the AtomStackClass cannot discern between full and empty stack!
2730 MoleculeLeafClass *Leaflet = NULL, *TempLeaf = NULL;
2731 MoleculeListClass *FragmentList = NULL;
2732 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Removal = NULL;
2733 bond *Binder = NULL;
2734 int RunningIndex = 0, FragmentCounter = 0;
2735
2736 *out << Verbose(1) << "Begin of CreateListOfUniqueFragmentsOfOrder." << endl;
2737
2738 // reset parent list
2739 *out << Verbose(3) << "Resetting labels, parent, predecessor, color and shortest path lists." << endl;
2740 for (int i=0;i<AtomCount;i++) { // reset all atom labels
2741 // initialise each vertex as white with no predecessor, empty queue, color lightgray, not labelled, no sons
2742 Labels[i] = -1;
2743 SonList[i] = NULL;
2744 PredecessorList[i] = NULL;
2745 ColorVertexList[i] = white;
2746 ShortestPathList[i] = -1;
2747 }
2748 for (int i=0;i<BondCount;i++)
2749 ColorEdgeList[i] = white;
2750 RootStack->ClearStack(); // clearstack and push first atom if exists
2751 TouchedStack->ClearStack();
2752 Walker = start->next;
2753 while ((Walker != end)
2754#ifdef ADDHYDROGEN
2755 && (Walker->type->Z == 1)
2756#endif
2757 ) { // search for first non-hydrogen atom
2758 *out << Verbose(4) << "Current Root candidate is " << Walker->Name << "." << endl;
2759 Walker = Walker->next;
2760 }
2761 if (Walker != end)
2762 RootStack->Push(Walker);
2763 else
2764 *out << Verbose(0) << "ERROR: Could not find an appropriate Root atom!" << endl;
2765 *out << Verbose(3) << "Root " << Walker->Name << " is on AtomStack, beginning loop through all vertices ..." << endl;
2766
2767 ///// OUTER LOOP ////////////
2768 while (!RootStack->IsEmpty()) {
2769 // get new root vertex from atom stack
2770 Root = RootStack->PopFirst();
2771 ShortestPathList[Root->nr] = 0;
2772 if (Labels[Root->nr] == -1)
2773 Labels[Root->nr] = RunningIndex++; // prevent it from getting again on AtomStack
2774 PredecessorList[Root->nr] = Root;
2775 TouchedStack->Push(Root);
2776 *out << Verbose(0) << "Root for this loop is: " << Root->Name << ".\n";
2777
2778 // clear snake stack
2779 SnakeStack->ClearStack();
2780 //SnakeStack->TestImplementation(out, start->next);
2781
2782 ///// INNER LOOP ////////////
2783 // Problems:
2784 // - what about cyclic bonds?
2785 Walker = Root;
2786 do {
2787 *out << Verbose(1) << "Current Walker is: " << Walker->Name;
2788 // initial setting of the new Walker: label, color, shortest path and put on stacks
2789 if (Labels[Walker->nr] == -1) { // give atom a unique, monotonely increasing number
2790 Labels[Walker->nr] = RunningIndex++;
2791 RootStack->Push(Walker);
2792 }
2793 *out << ", has label " << Labels[Walker->nr];
2794 if ((ColorVertexList[Walker->nr] == white) || ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white))) { // color it if newly discovered and push on stacks (and if within reach!)
2795 if ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white)) {
2796 // Binder ought to be set still from last neighbour search
2797 *out << ", coloring bond " << *Binder << " black";
2798 ColorEdgeList[Binder->nr] = black; // mark this bond as used
2799 }
2800 if (ShortestPathList[Walker->nr] == -1) {
2801 ShortestPathList[Walker->nr] = ShortestPathList[PredecessorList[Walker->nr]->nr]+1;
2802 TouchedStack->Push(Walker); // mark every atom for lists cleanup later, whose shortest path has been changed
2803 }
2804 if ((ShortestPathList[Walker->nr] < Order) && (ColorVertexList[Walker->nr] != darkgray)) { // if not already on snake stack
2805 SnakeStack->Push(Walker);
2806 ColorVertexList[Walker->nr] = darkgray; // mark as dark gray of on snake stack
2807 }
2808 }
2809 *out << ", SP of " << ShortestPathList[Walker->nr] << " and its color is " << GetColor(ColorVertexList[Walker->nr]) << "." << endl;
2810
2811 // then check the stack for a newly stumbled upon fragment
2812 if (SnakeStack->ItemCount() == Order) { // is stack full?
2813 // store the fragment if it is one and get a removal candidate
2814 Removal = StoreFragmentFromStack(out, Root, Walker, Leaflet, SnakeStack, ShortestPathList, SonList, Labels, &FragmentCounter, configuration);
2815 // remove the candidate if one was found
2816 if (Removal != NULL) {
2817 *out << Verbose(2) << "Removing item " << Removal->Name << " with SP of " << ShortestPathList[Removal->nr] << " from snake stack." << endl;
2818 SnakeStack->RemoveItem(Removal);
2819 ColorVertexList[Removal->nr] = lightgray; // return back to not on snake stack but explored marking
2820 if (Walker == Removal) { // if the current atom is to be removed, we also have to take a step back
2821 Walker = PredecessorList[Removal->nr];
2822 *out << Verbose(2) << "Stepping back to " << Walker->Name << "." << endl;
2823 }
2824 }
2825 } else
2826 Removal = NULL;
2827
2828 // finally, look for a white neighbour as the next Walker
2829 Binder = NULL;
2830 if ((Removal == NULL) || (Walker != PredecessorList[Removal->nr])) { // don't look, if a new walker has been set above
2831 *out << Verbose(2) << "Snake has currently " << SnakeStack->ItemCount() << " item(s)." << endl;
2832 OtherAtom = NULL; // this is actually not needed, every atom has at least one neighbour
2833 if (ShortestPathList[Walker->nr] < Order) {
2834 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2835 Binder = ListOfBondsPerAtom[Walker->nr][i];
2836 *out << Verbose(2) << "Current bond is " << *Binder << ": ";
2837 OtherAtom = Binder->GetOtherAtom(Walker);
2838 if ((Labels[OtherAtom->nr] != -1) && (Labels[OtherAtom->nr] < Labels[Root->nr])) { // we don't step up to labels bigger than us
2839 *out << "Label " << Labels[OtherAtom->nr] << " is smaller than Root's " << Labels[Root->nr] << "." << endl;
2840 //ColorVertexList[OtherAtom->nr] = lightgray; // mark as explored
2841 } else { // otherwise check its colour and element
2842 if (
2843#ifdef ADDHYDROGEN
2844 (OtherAtom->type->Z != 1) &&
2845#endif
2846 (ColorEdgeList[Binder->nr] == white)) { // skip hydrogen, look for unexplored vertices
2847 *out << "Moving along " << GetColor(ColorEdgeList[Binder->nr]) << " bond " << Binder << " to " << ((ColorVertexList[OtherAtom->nr] == white) ? "unexplored" : "explored") << " item: " << OtherAtom->Name << "." << endl;
2848 // i find it currently rather sensible to always set the predecessor in order to find one's way back
2849 //if (PredecessorList[OtherAtom->nr] == NULL) {
2850 PredecessorList[OtherAtom->nr] = Walker;
2851 *out << Verbose(3) << "Setting Predecessor of " << OtherAtom->Name << " to " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
2852 //} else {
2853 // *out << Verbose(3) << "Predecessor of " << OtherAtom->Name << " is " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
2854 //}
2855 Walker = OtherAtom;
2856 break;
2857 } else {
2858 if (OtherAtom->type->Z == 1)
2859 *out << "Links to a hydrogen atom." << endl;
2860 else
2861 *out << "Bond has not white but " << GetColor(ColorEdgeList[Binder->nr]) << " color." << endl;
2862 }
2863 }
2864 }
2865 } else { // means we have stepped beyond the horizon: Return!
2866 Walker = PredecessorList[Walker->nr];
2867 OtherAtom = Walker;
2868 *out << Verbose(3) << "We have gone too far, stepping back to " << Walker->Name << "." << endl;
2869 }
2870 if (Walker != OtherAtom) { // if no white neighbours anymore, color it black
2871 *out << Verbose(2) << "Coloring " << Walker->Name << " black." << endl;
2872 ColorVertexList[Walker->nr] = black;
2873 Walker = PredecessorList[Walker->nr];
2874 }
2875 }
2876 } while ((Walker != Root) || (ColorVertexList[Root->nr] != black));
2877 *out << Verbose(2) << "Inner Looping is finished." << endl;
2878
2879 // if we reset all AtomCount atoms, we have again technically O(N^2) ...
2880 *out << Verbose(2) << "Resetting lists." << endl;
2881 Walker = NULL;
2882 Binder = NULL;
2883 while (!TouchedStack->IsEmpty()) {
2884 Walker = TouchedStack->PopLast();
2885 *out << Verbose(3) << "Re-initialising entries of " << *Walker << "." << endl;
2886 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
2887 ColorEdgeList[ListOfBondsPerAtom[Walker->nr][i]->nr] = white;
2888 PredecessorList[Walker->nr] = NULL;
2889 ColorVertexList[Walker->nr] = white;
2890 ShortestPathList[Walker->nr] = -1;
2891 }
2892 }
2893 *out << Verbose(1) << "Outer Looping over all vertices is done." << endl;
2894
2895 // copy together
2896 *out << Verbose(1) << "Copying all fragments into MoleculeList structure." << endl;
2897 FragmentList = new MoleculeListClass(FragmentCounter, AtomCount);
2898 RunningIndex = 0;
2899 while ((Leaflet != NULL) && (RunningIndex < FragmentCounter)) {
2900 FragmentList->ListOfMolecules[RunningIndex++] = Leaflet->Leaf;
2901 Leaflet->Leaf = NULL; // prevent molecule from being removed
2902 TempLeaf = Leaflet;
2903 Leaflet = Leaflet->previous;
2904 delete(TempLeaf);
2905 };
2906
2907 // free memory and exit
2908 Free((void **)&PredecessorList, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
2909 Free((void **)&ShortestPathList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
2910 Free((void **)&Labels, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
2911 Free((void **)&ColorVertexList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
2912 delete(RootStack);
2913 delete(TouchedStack);
2914 delete(SnakeStack);
2915
2916 *out << Verbose(1) << "End of CreateListOfUniqueFragmentsOfOrder." << endl;
2917 return FragmentList;
2918};
2919*/
2920
2921/** Structure containing all values in power set combination generation.
2922 */
2923struct UniqueFragments {
2924 config *configuration;
2925 atom *Root;
2926 Graph *Leaflet;
2927 KeySet *FragmentSet;
2928 int ANOVAOrder;
2929 int FragmentCounter;
2930 int CurrentIndex;
2931 int *Labels;
2932 int *ShortestPathList;
2933 bool **UsedList;
2934 bond **BondsPerSPList;
2935 int *BondsPerSPCount;
2936};
2937
2938/** From a given set of Bond sorted by Shortest Path distance, create all possible fragments of size \a SetDimension.
2939 * This basically involves recursion to create all power set combinations.
2940 * \param *out output stream for debugging
2941 * \param FragmentSearch UniqueFragments structure with all values needed
2942 * \param RootDistance current shortest path level, whose set of edges is represented by **BondsSet
2943 * \param SetDimension Number of possible bonds on this level (i.e. size of the array BondsSet[])
2944 * \param SubOrder remaining number of allowed vertices to add
2945 */
2946void molecule::SPFragmentGenerator(ofstream *out, struct UniqueFragments *FragmentSearch, int RootDistance, bond **BondsSet, int SetDimension, int SubOrder)
2947{
2948 atom *OtherWalker = NULL;
2949 int verbosity = 0; //FragmentSearch->ANOVAOrder-SubOrder;
2950 int NumCombinations;
2951 bool bit;
2952 int bits, TouchedIndex, SubSetDimension, SP;
2953 int Removal;
2954 int *TouchedList = (int *) Malloc(sizeof(int)*(SubOrder+1), "molecule::SPFragmentGenerator: *TouchedList");
2955 bond *Binder = NULL;
2956 bond **BondsList = NULL;
2957
2958 NumCombinations = 1 << SetDimension;
2959
2960 // Hier muessen von 1 bis NumberOfBondsPerAtom[Walker->nr] alle Kombinationen
2961 // von Endstuecken (aus den Bonds) hinzugefÃŒgt werden und fÃŒr verbleibende ANOVAOrder
2962 // rekursiv GraphCrawler in der nÀchsten Ebene aufgerufen werden
2963
2964 *out << Verbose(1+verbosity) << "Begin of SPFragmentGenerator." << endl;
2965 *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;
2966
2967 // initialised touched list (stores added atoms on this level)
2968 *out << Verbose(1+verbosity) << "Clearing touched list." << endl;
2969 for (TouchedIndex=0;TouchedIndex<=SubOrder;TouchedIndex++) // empty touched list
2970 TouchedList[TouchedIndex] = -1;
2971 TouchedIndex = 0;
2972
2973 // create every possible combination of the endpieces
2974 *out << Verbose(1+verbosity) << "Going through all combinations of the power set." << endl;
2975 for (int i=1;i<NumCombinations;i++) { // sweep through all power set combinations (skip empty set!)
2976 // count the set bit of i
2977 bits = 0;
2978 for (int j=0;j<SetDimension;j++)
2979 bits += (i & (1 << j)) >> j;
2980
2981 *out << Verbose(1+verbosity) << "Current set is " << Binary(i | (1 << SetDimension)) << ", number of bits is " << bits << "." << endl;
2982 if (bits <= SubOrder) { // if not greater than additional atoms allowed on stack, continue
2983 // --1-- add this set of the power set of bond partners to the snake stack
2984 for (int j=0;j<SetDimension;j++) { // pull out every bit by shifting
2985 bit = ((i & (1 << j)) != 0); // mask the bit for the j-th bond
2986 if (bit) { // if bit is set, we add this bond partner
2987 OtherWalker = BondsSet[j]->rightatom; // rightatom is always the one more distant, i.e. the one to add
2988 //*out << Verbose(1+verbosity) << "Current Bond is " << ListOfBondsPerAtom[Walker->nr][i] << ", checking on " << *OtherWalker << "." << endl;
2989 //if ((!FragmentSearch->UsedList[OtherWalker->nr][i]) && (FragmentSearch->Labels[OtherWalker->nr] > FragmentSearch->Labels[FragmentSearch->Root->nr])) {
2990 //*out << Verbose(2+verbosity) << "Not used so far, label " << FragmentSearch->Labels[OtherWalker->nr] << " is bigger than Root's " << FragmentSearch->Labels[FragmentSearch->Root->nr] << "." << endl;
2991 *out << Verbose(2+verbosity) << "Adding " << *OtherWalker << " with nr " << OtherWalker->nr << "." << endl;
2992 TouchedList[TouchedIndex++] = OtherWalker->nr; // note as added
2993 FragmentSearch->FragmentSet->insert(OtherWalker->nr);
2994 //FragmentSearch->UsedList[OtherWalker->nr][i] = true;
2995 //}
2996 } else {
2997 *out << Verbose(2+verbosity) << "Not adding." << endl;
2998 }
2999 }
3000
3001 if (bits < SubOrder) {
3002 *out << Verbose(1+verbosity) << "There's still some space left on stack: " << (SubOrder - bits) << "." << endl;
3003 // --2-- look at all added end pieces of this combination, construct bond subsets and sweep through a power set of these by recursion
3004 SP = RootDistance+1; // this is the next level
3005 // first count the members in the subset
3006 SubSetDimension = 0;
3007 Binder = FragmentSearch->BondsPerSPList[2*SP]; // start node for this level
3008 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) { // compare to end node of this level
3009 Binder = Binder->next;
3010 for (int k=0;k<TouchedIndex;k++) {
3011 if (Binder->Contains(TouchedList[k])) // if we added this very endpiece
3012 SubSetDimension++;
3013 }
3014 }
3015 // then allocate and fill the list
3016 BondsList = (bond **) Malloc(sizeof(bond *)*SubSetDimension, "molecule::SPFragmentGenerator: **BondsList");
3017 SubSetDimension = 0;
3018 Binder = FragmentSearch->BondsPerSPList[2*SP];
3019 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) {
3020 Binder = Binder->next;
3021 for (int k=0;k<TouchedIndex;k++) {
3022 if (Binder->leftatom->nr == TouchedList[k]) // leftatom is always the close one
3023 BondsList[SubSetDimension++] = Binder;
3024 }
3025 }
3026 *out << Verbose(2+verbosity) << "Calling subset generator " << SP << " away from root " << *FragmentSearch->Root << " with sub set dimension " << SubSetDimension << "." << endl;
3027 SPFragmentGenerator(out, FragmentSearch, SP, BondsList, SubSetDimension, SubOrder-bits);
3028 Free((void **)&BondsList, "molecule::SPFragmentGenerator: **BondsList");
3029 } else {
3030 // --2-- otherwise store the complete fragment
3031 *out << Verbose(1+verbosity) << "Enough items on stack for a fragment!" << endl;
3032 // store fragment as a KeySet
3033 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch->FragmentCounter << "], local nr.s are: ";
3034 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
3035 *out << (*runner) << " ";
3036 InsertFragmentIntoGraph(out, FragmentSearch);
3037 //Removal = LookForRemovalCandidate(out, FragmentSearch->FragmentSet, FragmentSearch->ShortestPathList);
3038 //Removal = StoreFragmentFromStack(out, FragmentSearch->Root, FragmentSearch->Leaflet, FragmentSearch->FragmentStack, FragmentSearch->ShortestPathList,FragmentSearch->Labels, &FragmentSearch->FragmentCounter, FragmentSearch->configuration);
3039 }
3040
3041 // --3-- remove all added items in this level from snake stack
3042 *out << Verbose(1+verbosity) << "Removing all items that were added on this SP level " << RootDistance << "." << endl;
3043 for(int j=0;j<TouchedIndex;j++) {
3044 Removal = TouchedList[j];
3045 *out << Verbose(2+verbosity) << "Removing item nr. " << Removal << " from snake stack." << endl;
3046 FragmentSearch->FragmentSet->erase(Removal);
3047 TouchedList[j] = -1;
3048 }
3049 *out << Verbose(2) << "Remaining local nr.s on snake stack are: ";
3050 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
3051 *out << (*runner) << " ";
3052 *out << endl;
3053 TouchedIndex = 0; // set Index to 0 for list of atoms added on this level
3054 } else {
3055 *out << Verbose(2+verbosity) << "More atoms to add for this set (" << bits << ") than space left on stack " << SubOrder << ", skipping this set." << endl;
3056 }
3057 }
3058 Free((void **)&TouchedList, "molecule::SPFragmentGenerator: *TouchedList");
3059 *out << Verbose(1+verbosity) << "End of SPFragmentGenerator, " << RootDistance << " away from Root " << *FragmentSearch->Root << " and SubOrder is " << SubOrder << "." << endl;
3060};
3061
3062/** 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.
3063 * 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
3064 * with SP of 2, then those with SP of 3, then those with SP of 4 and so on.
3065 * \param *out output stream for debugging
3066 * \param Order bond order (limits BFS exploration and "number of digits" in power set generation
3067 * \param *ReturnKeySets Graph structure to insert found keysets/fragments into
3068 * \param RestrictedKeySet Restricted vertex set to use in context of molecule
3069 * \param RootKeyNr Atom::nr of the atom acting as current fragment root
3070 * \return number of inserted fragments
3071 * \note ShortestPathList in FragmentSearch structure is probably due to NumberOfAtomsSPLevel and SP not needed anymore
3072 */
3073int molecule::PowerSetGenerator(ofstream *out, int Order, Graph *ReturnKeySets, KeySet RestrictedKeySet, int RootKeyNr)
3074{
3075 int SP, UniqueIndex, AtomKeyNr;
3076 int *NumberOfAtomsSPLevel = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *SPLevelCount");
3077 atom *Walker = NULL, *OtherWalker = NULL;
3078 bond *Binder = NULL;
3079 bond **BondsList = NULL;
3080 KeyStack AtomStack;
3081 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::PowerSetGenerator: **PredecessorList");
3082 KeySet::iterator runner;
3083 //int Count = RestrictedKeySet.size();
3084
3085 // initialise the fragments structure
3086 struct UniqueFragments FragmentSearch;
3087 FragmentSearch.BondsPerSPList = (bond **) Malloc(sizeof(bond *)*Order*2, "molecule::PowerSetGenerator: ***BondsPerSPList");
3088 FragmentSearch.BondsPerSPCount = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *BondsPerSPCount");
3089 FragmentSearch.ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *ShortestPathList");
3090 FragmentSearch.Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *Labels");
3091 FragmentSearch.FragmentCounter = 0;
3092 FragmentSearch.FragmentSet = new KeySet;
3093 FragmentSearch.Leaflet = ReturnKeySets; // set to insertion graph
3094 FragmentSearch.Root = FindAtom(RootKeyNr);
3095 for (int i=0;i<AtomCount;i++) {
3096 FragmentSearch.Labels[i] = -1;
3097 FragmentSearch.ShortestPathList[i] = -1;
3098 PredecessorList[i] = NULL;
3099 }
3100 for (int i=0;i<Order;i++) {
3101 FragmentSearch.BondsPerSPList[2*i] = new bond(); // start node
3102 FragmentSearch.BondsPerSPList[2*i+1] = new bond(); // end node
3103 FragmentSearch.BondsPerSPList[2*i]->next = FragmentSearch.BondsPerSPList[2*i+1]; // intertwine these two
3104 FragmentSearch.BondsPerSPList[2*i+1]->previous = FragmentSearch.BondsPerSPList[2*i];
3105 FragmentSearch.BondsPerSPCount[i] = 0;
3106 }
3107
3108 *out << endl;
3109 *out << Verbose(0) << "Begin of PowerSetGenerator with order " << Order << " at Root " << *FragmentSearch.Root << "." << endl;
3110
3111 UniqueIndex = 0;
3112 if (FragmentSearch.Labels[RootKeyNr] == -1)
3113 FragmentSearch.Labels[RootKeyNr] = UniqueIndex++;
3114 FragmentSearch.ShortestPathList[RootKeyNr] = 0;
3115 // prepare the atom stack counters (number of atoms with certain SP on stack)
3116 for (int i=0;i<Order;i++)
3117 NumberOfAtomsSPLevel[i] = 0;
3118 NumberOfAtomsSPLevel[0] = 1; // for root
3119 SP = -1;
3120 *out << endl;
3121 *out << Verbose(0) << "Starting BFS analysis ..." << endl;
3122 // push as first on atom stack and goooo ...
3123 AtomStack.clear();
3124 AtomStack.push_back(RootKeyNr);
3125 *out << Verbose(0) << "Cleared AtomStack and pushed root as first item onto it." << endl;
3126 // do a BFS search to fill the SP lists and label the found vertices
3127 while (!AtomStack.empty()) {
3128 // pop next atom
3129 AtomKeyNr = AtomStack.front();
3130 AtomStack.pop_front();
3131 if (SP != -1)
3132 NumberOfAtomsSPLevel[SP]--;
3133 if ((SP == -1) || (NumberOfAtomsSPLevel[SP] == -1)) {
3134 SP++;
3135 NumberOfAtomsSPLevel[SP]--; // carry over "-1" to next level
3136 *out << Verbose(1) << "New SP level reached: " << SP << ", creating new SP list with 0 item(s)";
3137 if (SP > 0)
3138 *out << ", old level closed with " << FragmentSearch.BondsPerSPCount[SP-1] << " item(s)." << endl;
3139 else
3140 *out << "." << endl;
3141 FragmentSearch.BondsPerSPCount[SP] = 0;
3142 } else {
3143 *out << Verbose(1) << "Still " << NumberOfAtomsSPLevel[SP]+1 << " on this SP (" << SP << ") level, currently having " << FragmentSearch.BondsPerSPCount[SP] << " item(s)." << endl;
3144 }
3145 Walker = FindAtom(AtomKeyNr);
3146 *out << Verbose(0) << "Current Walker is: " << *Walker << " with nr " << Walker->nr << " and label " << FragmentSearch.Labels[AtomKeyNr] << " and SP of " << SP << "." << endl;
3147 // check for new sp level
3148 // go through all its bonds
3149 *out << Verbose(1) << "Going through all bonds of Walker." << endl;
3150 for (int i=0;i<NumberOfBondsPerAtom[AtomKeyNr];i++) {
3151 Binder = ListOfBondsPerAtom[AtomKeyNr][i];
3152 OtherWalker = Binder->GetOtherAtom(Walker);
3153 if ((RestrictedKeySet.find(OtherWalker->nr) != RestrictedKeySet.end())
3154#ifdef ADDHYDROGEN
3155 && (OtherWalker->type->Z != 1)
3156#endif
3157 ) { // skip hydrogens and restrict to fragment
3158 *out << Verbose(2) << "Current partner is " << *OtherWalker << " with nr " << OtherWalker->nr << " in bond " << *Binder << "." << endl;
3159 // set the label if not set (and push on root stack as well)
3160 if (FragmentSearch.Labels[OtherWalker->nr] == -1) {
3161 FragmentSearch.Labels[OtherWalker->nr] = UniqueIndex++;
3162 *out << Verbose(3) << "Set label to " << FragmentSearch.Labels[OtherWalker->nr] << "." << endl;
3163 } else {
3164 *out << Verbose(3) << "Label is already " << FragmentSearch.Labels[OtherWalker->nr] << "." << endl;
3165 }
3166 if ((OtherWalker != PredecessorList[AtomKeyNr]) && (FragmentSearch.Labels[OtherWalker->nr] > FragmentSearch.Labels[RootKeyNr])) { // only pass through those with label bigger than Root's
3167 FragmentSearch.ShortestPathList[OtherWalker->nr] = SP+1;
3168 *out << Verbose(3) << "Set Shortest Path to " << FragmentSearch.ShortestPathList[OtherWalker->nr] << "." << endl;
3169 if (FragmentSearch.ShortestPathList[OtherWalker->nr] < Order) { // only pass through those within reach of Order
3170 // push for exploration on stack (only if the SP of OtherWalker is longer than of Walker! (otherwise it has been added already!)
3171 if (FragmentSearch.ShortestPathList[OtherWalker->nr] > SP) {
3172 if (FragmentSearch.ShortestPathList[OtherWalker->nr] < (Order-1)) {
3173 *out << Verbose(3) << "Putting on atom stack for further exploration." << endl;
3174 PredecessorList[OtherWalker->nr] = Walker; // note down the one further up the exploration tree
3175 AtomStack.push_back(OtherWalker->nr);
3176 NumberOfAtomsSPLevel[FragmentSearch.ShortestPathList[OtherWalker->nr]]++;
3177 } else {
3178 *out << Verbose(3) << "Not putting on atom stack, is at end of reach." << endl;
3179 }
3180 // add the bond in between to the SP list
3181 Binder = new bond(Walker, OtherWalker); // create a new bond in such a manner, that bond::rightatom is always the one more distant
3182 add(Binder, FragmentSearch.BondsPerSPList[2*SP+1]);
3183 FragmentSearch.BondsPerSPCount[SP]++;
3184 *out << Verbose(3) << "Added its bond to SP list, having now " << FragmentSearch.BondsPerSPCount[SP] << " item(s)." << endl;
3185 } else *out << Verbose(3) << "Not putting on atom stack, nor adding bond, as " << *OtherWalker << "'s SP is shorter than Walker's." << endl;
3186 } else *out << Verbose(3) << "Not continuing, as " << *OtherWalker << " is out of reach of order " << Order << "." << endl;
3187 } else *out << Verbose(3) << "Not passing on, as label of " << *OtherWalker << " " << FragmentSearch.Labels[OtherWalker->nr] << " is smaller than that of Root " << FragmentSearch.Labels[RootKeyNr] << " or this is my predecessor." << endl;
3188 } else *out << Verbose(2) << "Is not in the retstricted keyset or skipping hydrogen " << *OtherWalker << "." << endl;
3189 }
3190 }
3191 // reset predecessor list
3192 for(int i=0;i<Order;i++) {
3193 Binder = FragmentSearch.BondsPerSPList[2*i];
3194 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
3195 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3196 Binder = Binder->next;
3197 PredecessorList[Binder->rightatom->nr] = NULL; // By construction "OtherAtom" is always Bond::rightatom!
3198 }
3199 }
3200 *out << endl;
3201
3202 // outputting all list for debugging
3203 *out << Verbose(0) << "Printing all found lists." << endl;
3204 for(int i=0;i<Order;i++) {
3205 Binder = FragmentSearch.BondsPerSPList[2*i];
3206 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
3207 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3208 Binder = Binder->next;
3209 *out << Verbose(2) << *Binder << endl;
3210 }
3211 }
3212
3213 // creating fragments with the found edge sets
3214 SP = 0;
3215 for(int i=0;i<Order;i++) { // sum up all found edges
3216 Binder = FragmentSearch.BondsPerSPList[2*i];
3217 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3218 Binder = Binder->next;
3219 SP ++;
3220 }
3221 }
3222 *out << Verbose(0) << "Total number of edges is " << SP << "." << endl;
3223 if (SP >= (Order-1)) {
3224 // start with root (push on fragment stack)
3225 *out << Verbose(0) << "Starting fragment generation with " << *FragmentSearch.Root << ", local nr is " << FragmentSearch.Root->nr << "." << endl;
3226 FragmentSearch.FragmentSet->clear();
3227 FragmentSearch.FragmentSet->insert(FragmentSearch.Root->nr);
3228
3229 if (FragmentSearch.FragmentSet->size() == (unsigned int) Order) {
3230 *out << Verbose(0) << "Enough items on stack already for a fragment!" << endl;
3231 // store fragment as a KeySet
3232 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch.FragmentCounter << "], local nr.s are: ";
3233 for(KeySet::iterator runner = FragmentSearch.FragmentSet->begin(); runner != FragmentSearch.FragmentSet->end(); runner++) {
3234 *out << (*runner) << " ";
3235 }
3236 *out << endl;
3237 InsertFragmentIntoGraph(out, &FragmentSearch);
3238 } else {
3239 *out << Verbose(0) << "Preparing subset for this root and calling generator." << endl;
3240 // prepare the subset and call the generator
3241 BondsList = (bond **) Malloc(sizeof(bond *)*FragmentSearch.BondsPerSPCount[0], "molecule::PowerSetGenerator: **BondsList");
3242 Binder = FragmentSearch.BondsPerSPList[0];
3243 for(int i=0;i<FragmentSearch.BondsPerSPCount[0];i++) {
3244 Binder = Binder->next;
3245 BondsList[i] = Binder;
3246 }
3247 SPFragmentGenerator(out, &FragmentSearch, 0, BondsList, FragmentSearch.BondsPerSPCount[0], Order-1);
3248 Free((void **)&BondsList, "molecule::PowerSetGenerator: **BondsList");
3249 }
3250 } else {
3251 *out << Verbose(0) << "Not enough total number of edges to build " << Order << "-body fragments." << endl;
3252 }
3253
3254/* // as FragmentSearch structure is used only once, we don't have to clean it anymore
3255 // remove root from stack
3256 *out << Verbose(0) << "Removing root again from stack." << endl;
3257 FragmentSearch.FragmentSet->erase(FragmentSearch.Root->nr);
3258
3259 // free'ing the bonds lists
3260 *out << Verbose(0) << "Free'ing all found lists. and resetting index lists" << endl;
3261 for(int i=0;i<Order;i++) {
3262 *out << Verbose(1) << "Current SP level is " << i << ": ";
3263 Binder = FragmentSearch.BondsPerSPList[2*i];
3264 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3265 Binder = Binder->next;
3266 // *out << "Removing atom " << Binder->leftatom->nr << " and " << Binder->rightatom->nr << "." << endl; // make sure numbers are local
3267 FragmentSearch.ShortestPathList[Binder->leftatom->nr] = -1;
3268 FragmentSearch.ShortestPathList[Binder->rightatom->nr] = -1;
3269 }
3270 // delete added bonds
3271 cleanup(FragmentSearch.BondsPerSPList[2*i], FragmentSearch.BondsPerSPList[2*i+1]);
3272 // also start and end node
3273 *out << "cleaned." << endl;
3274 }
3275*/
3276 // free allocated memory
3277 Free((void **)&NumberOfAtomsSPLevel, "molecule::PowerSetGenerator: *SPLevelCount");
3278 Free((void **)&PredecessorList, "molecule::PowerSetGenerator: **PredecessorList");
3279 for(int i=0;i<Order;i++) { // delete start and end of each list
3280 delete(FragmentSearch.BondsPerSPList[2*i]);
3281 delete(FragmentSearch.BondsPerSPList[2*i+1]);
3282 }
3283 Free((void **)&FragmentSearch.BondsPerSPList, "molecule::PowerSetGenerator: ***BondsPerSPList");
3284 Free((void **)&FragmentSearch.BondsPerSPCount, "molecule::PowerSetGenerator: *BondsPerSPCount");
3285 Free((void **)&FragmentSearch.ShortestPathList, "molecule::PowerSetGenerator: *ShortestPathList");
3286 Free((void **)&FragmentSearch.Labels, "molecule::PowerSetGenerator: *Labels");
3287 delete(FragmentSearch.FragmentSet);
3288
3289 // return list
3290 *out << Verbose(0) << "End of PowerSetGenerator." << endl;
3291 return FragmentSearch.FragmentCounter;
3292};
3293
3294/** Corrects the nuclei position if the fragment was created over the cell borders.
3295 * Scans all bonds, checks the distance, if greater than typical, we have a candidate for the correction.
3296 * We remove the bond whereafter the graph probably separates. Then, we translate the one component periodically
3297 * and re-add the bond. Looping on the distance check.
3298 * \param *out ofstream for debugging messages
3299 */
3300void molecule::ScanForPeriodicCorrection(ofstream *out)
3301{
3302 bond *Binder = NULL;
3303 bond *OtherBinder = NULL;
3304 atom *Walker = NULL;
3305 atom *OtherWalker = NULL;
3306 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
3307 enum Shading *ColorList = NULL;
3308 double tmp;
3309 vector TranslationVector;
3310 //AtomStackClass *CompStack = NULL;
3311 AtomStackClass *AtomStack = new AtomStackClass(AtomCount);
3312 bool flag = true;
3313
3314// *out << Verbose(1) << "Begin of ScanForPeriodicCorrection." << endl;
3315
3316 ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::ScanForPeriodicCorrection: *ColorList");
3317 while (flag) {
3318 // remove bonds that are beyond bonddistance
3319 for(int i=0;i<NDIM;i++)
3320 TranslationVector.x[i] = 0.;
3321 // scan all bonds
3322 Binder = first;
3323 flag = false;
3324 while ((!flag) && (Binder->next != last)) {
3325 Binder = Binder->next;
3326 for (int i=0;i<NDIM;i++) {
3327 tmp = fabs(Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i]);
3328 //*out << Verbose(3) << "Checking " << i << "th distance of " << *Binder->leftatom << " to " << *Binder->rightatom << ": " << tmp << "." << endl;
3329 if (tmp > BondDistance) {
3330 OtherBinder = Binder->next; // note down binding partner for later re-insertion
3331 unlink(Binder); // unlink bond
3332// *out << Verbose(2) << "Correcting at bond " << *Binder << "." << endl;
3333 flag = true;
3334 break;
3335 }
3336 }
3337 }
3338 if (flag) {
3339 // create translation vector from their periodically modified distance
3340 for (int i=0;i<NDIM;i++) {
3341 tmp = Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i];
3342 if (fabs(tmp) > BondDistance)
3343 TranslationVector.x[i] = (tmp < 0) ? +1. : -1.;
3344 }
3345 TranslationVector.MatrixMultiplication(matrix);
3346 //*out << "Translation vector is ";
3347 //TranslationVector.Output(out);
3348 //*out << endl;
3349 // apply to all atoms of first component via BFS
3350 for (int i=0;i<AtomCount;i++)
3351 ColorList[i] = white;
3352 AtomStack->Push(Binder->leftatom);
3353 while (!AtomStack->IsEmpty()) {
3354 Walker = AtomStack->PopFirst();
3355 //*out << Verbose (3) << "Current Walker is: " << *Walker << "." << endl;
3356 ColorList[Walker->nr] = black; // mark as explored
3357 Walker->x.AddVector(&TranslationVector); // translate
3358 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through all binding partners
3359 if (ListOfBondsPerAtom[Walker->nr][i] != Binder) {
3360 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
3361 if (ColorList[OtherWalker->nr] == white) {
3362 AtomStack->Push(OtherWalker); // push if yet unexplored
3363 }
3364 }
3365 }
3366 }
3367 // re-add bond
3368 link(Binder, OtherBinder);
3369 } else {
3370// *out << Verbose(2) << "No corrections for this fragment." << endl;
3371 }
3372 //delete(CompStack);
3373 }
3374
3375 // free allocated space from ReturnFullMatrixforSymmetric()
3376 delete(AtomStack);
3377 Free((void **)&ColorList, "molecule::ScanForPeriodicCorrection: *ColorList");
3378 Free((void **)&matrix, "molecule::ScanForPeriodicCorrection: *matrix");
3379// *out << Verbose(1) << "End of ScanForPeriodicCorrection." << endl;
3380};
3381
3382/** Blows the 6-dimensional \a cell_size array up to a full NDIM by NDIM matrix.
3383 * \param *symm 6-dim array of unique symmetric matrix components
3384 * \return allocated NDIM*NDIM array with the symmetric matrix
3385 */
3386double * molecule::ReturnFullMatrixforSymmetric(double *symm)
3387{
3388 double *matrix = (double *) Malloc(sizeof(double)*NDIM*NDIM, "molecule::ReturnFullMatrixforSymmetric: *matrix");
3389 matrix[0] = symm[0];
3390 matrix[1] = symm[1];
3391 matrix[2] = symm[3];
3392 matrix[3] = symm[1];
3393 matrix[4] = symm[2];
3394 matrix[5] = symm[4];
3395 matrix[6] = symm[3];
3396 matrix[7] = symm[4];
3397 matrix[8] = symm[5];
3398 return matrix;
3399};
3400
3401bool KeyCompare::operator() (const KeySet SubgraphA, const KeySet SubgraphB) const
3402{
3403 //cout << "my check is used." << endl;
3404 if (SubgraphA.size() < SubgraphB.size()) {
3405 return true;
3406 } else {
3407 if (SubgraphA.size() > SubgraphB.size()) {
3408 return false;
3409 } else {
3410 KeySet::iterator IteratorA = SubgraphA.begin();
3411 KeySet::iterator IteratorB = SubgraphB.begin();
3412 while ((IteratorA != SubgraphA.end()) && (IteratorB != SubgraphB.end())) {
3413 if ((*IteratorA) < (*IteratorB))
3414 return true;
3415 else if ((*IteratorA) > (*IteratorB)) {
3416 return false;
3417 } // else, go on to next index
3418 IteratorA++;
3419 IteratorB++;
3420 } // end of while loop
3421 }// end of check in case of equal sizes
3422 }
3423 return false; // if we reach this point, they are equal
3424};
3425
3426//bool operator < (KeySet SubgraphA, KeySet SubgraphB)
3427//{
3428// return KeyCompare(SubgraphA, SubgraphB);
3429//};
3430
3431/** Checking whether KeySet is not already present in Graph, if so just adds factor.
3432 * \param *out output stream for debugging
3433 * \param &set KeySet to insert
3434 * \param &graph Graph to insert into
3435 * \param *counter pointer to unique fragment count
3436 * \param factor energy factor for the fragment
3437 */
3438inline void InsertFragmentIntoGraph(ofstream *out, struct UniqueFragments *Fragment)
3439{
3440 GraphTestPair testGraphInsert;
3441
3442 testGraphInsert = Fragment->Leaflet->insert(GraphPair (*Fragment->FragmentSet,pair<int,double>(Fragment->FragmentCounter,1))); // store fragment number and current factor
3443 if (testGraphInsert.second) {
3444 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " successfully inserted." << endl;
3445 Fragment->FragmentCounter++;
3446 } else {
3447 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
3448 ((*(testGraphInsert.first)).second).second ++; // increase the "created" counter
3449 *out << Verbose(2) << "New factor is " << ((*(testGraphInsert.first)).second).second << "." << endl;
3450 }
3451};
3452//void inline InsertIntoGraph(ofstream *out, KeyStack &stack, Graph &graph, int *counter, double factor)
3453//{
3454// // copy stack contents to set and call overloaded function again
3455// KeySet set;
3456// for(KeyStack::iterator runner = stack.begin(); runner != stack.begin(); runner++)
3457// set.insert((*runner));
3458// InsertIntoGraph(out, set, graph, counter, factor);
3459//};
3460
3461/** Inserts each KeySet in \a graph2 into \a graph1.
3462 * \param *out output stream for debugging
3463 * \param graph1 first (dest) graph
3464 * \param graph2 second (source) graph
3465 * \param *counter keyset counter that gets increased
3466 */
3467inline void InsertGraphIntoGraph(ofstream *out, Graph &graph1, Graph &graph2, int *counter)
3468{
3469 GraphTestPair testGraphInsert;
3470
3471 for(Graph::iterator runner = graph2.begin(); runner != graph2.end(); runner++) {
3472 testGraphInsert = graph1.insert(GraphPair ((*runner).first,pair<int,double>((*counter)++,((*runner).second).second))); // store fragment number and current factor
3473 if (testGraphInsert.second) {
3474 *out << Verbose(2) << "KeySet " << (*counter)-1 << " successfully inserted." << endl;
3475 } else {
3476 *out << Verbose(2) << "KeySet " << (*counter)-1 << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
3477 ((*(testGraphInsert.first)).second).second += (*runner).second.second;
3478 *out << Verbose(2) << "New factor is " << (*(testGraphInsert.first)).second.second << "." << endl;
3479 }
3480 }
3481};
3482
3483
3484/** Performs BOSSANOVA decomposition at selected sites, increasing the cutoff by one at these sites.
3485 * Important only is that we create all fragments, it is not important if we create them more than once
3486 * as these copies are filtered out via use of the hash table (KeySet).
3487 * \param *out output stream for debugging
3488 * \return pointer to Graph list
3489 */
3490Graph * molecule::FragmentBOSSANOVA(ofstream *out, KeyStack &RootStack)
3491{
3492 Graph *FragmentList = NULL, ***FragmentLowerOrdersList = NULL;
3493 int Order, NumLevels, NumMolecules, TotalNumMolecules = 0, *NumMoleculesOfOrder = NULL;
3494 int counter = 0;
3495 int UpgradeCount = RootStack.size();
3496 KeyStack FragmentRootStack;
3497 int RootKeyNr, RootNr;
3498
3499 *out << Verbose(0) << "Begin of FragmentBOSSANOVA." << endl;
3500
3501 // FragmentLowerOrdersList is a 2D-array of pointer to MoleculeListClass objects, one dimension represents the ANOVA expansion of a single order (i.e. 5)
3502 // with all needed lower orders that are subtracted, the other dimension is the BondOrder (i.e. from 1 to 5)
3503 NumMoleculesOfOrder = (int *) Malloc(sizeof(int)*UpgradeCount, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
3504 FragmentLowerOrdersList = (Graph ***) Malloc(sizeof(Graph **)*UpgradeCount, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
3505
3506 // Construct the complete KeySet which we need for topmost level only (but for all Roots)
3507 atom *Walker = start;
3508 KeySet CompleteMolecule;
3509 while (Walker->next != end) {
3510 Walker = Walker->next;
3511 CompleteMolecule.insert(Walker->GetTrueFather()->nr);
3512 }
3513
3514 // 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
3515 // 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),
3516 // hence we have overall four 2th order levels for splitting. This also allows for putting all into a single array (FragmentLowerOrdersList[])
3517 // 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)
3518 RootNr = 0; // counts through the roots in RootStack
3519 while (RootNr < UpgradeCount) {
3520 RootKeyNr = RootStack.front();
3521 RootStack.pop_front();
3522 // increase adaptive order by one
3523 Walker = FindAtom(RootKeyNr);
3524 Walker->GetTrueFather()->AdaptiveOrder++;
3525 Order = Walker->AdaptiveOrder = Walker->GetTrueFather()->AdaptiveOrder;
3526
3527 // allocate memory for all lower level orders in this 1D-array of ptrs
3528 NumLevels = 1 << (Order); // (int)pow(2,Order);
3529 FragmentLowerOrdersList[RootNr] = (Graph **) Malloc(sizeof(Graph *)*NumLevels, "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
3530
3531 // create top order where nothing is reduced
3532 *out << Verbose(0) << "==============================================================================================================" << endl;
3533 *out << Verbose(0) << "Creating KeySets of Bond Order " << Order << " for " << *Walker << ", NumLevels is " << NumLevels << ", " << RootStack.size() << " Roots remaining." << endl;
3534
3535 // Create list of Graphs of current Bond Order (i.e. F_{ij})
3536 FragmentLowerOrdersList[RootNr][0] = new Graph;
3537 NumMoleculesOfOrder[RootNr] = PowerSetGenerator(out, Walker->AdaptiveOrder, FragmentLowerOrdersList[RootNr][0], CompleteMolecule, RootKeyNr);
3538 *out << Verbose(1) << "Number of resulting KeySets is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
3539 NumMolecules = 0;
3540
3541 if ((NumLevels >> 1) > 0) {
3542 // create lower order fragments
3543 *out << Verbose(0) << "Creating list of unique fragments of lower Bond Order terms to be subtracted." << endl;
3544 Order = Walker->AdaptiveOrder;
3545 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)
3546 // step down to next order at (virtual) boundary of powers of 2 in array
3547 while (source >= (1 << (Walker->AdaptiveOrder-Order))) // (int)pow(2,Walker->AdaptiveOrder-Order))
3548 Order--;
3549 *out << Verbose(0) << "Current Order is: " << Order << "." << endl;
3550 for (int SubOrder=Order;SubOrder>1;SubOrder--) {
3551 int dest = source + (1 << (Walker->AdaptiveOrder-SubOrder));
3552 *out << Verbose(0) << "--------------------------------------------------------------------------------------------------------------" << endl;
3553 *out << Verbose(0) << "Current SubOrder is: " << SubOrder-1 << " with source " << source << " to destination " << dest << "." << endl;
3554
3555 // every molecule is split into a list of again (Order - 1) molecules, while counting all molecules
3556 //*out << Verbose(1) << "Splitting the " << (*FragmentLowerOrdersList[RootNr][source]).size() << " molecules of the " << source << "th cell in the array." << endl;
3557 //NumMolecules = 0;
3558 FragmentLowerOrdersList[RootNr][dest] = new Graph;
3559 for(Graph::iterator runner = (*FragmentLowerOrdersList[RootNr][source]).begin();runner != (*FragmentLowerOrdersList[RootNr][source]).end(); runner++) {
3560 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
3561 FragmentList = new Graph();
3562 PowerSetGenerator(out, SubOrder-1, FragmentList, (*runner).first, *sprinter);
3563 // insert new keysets FragmentList into FragmentLowerOrdersList[Walker->AdaptiveOrder-1][dest]
3564 *out << Verbose(1) << "Merging resulting key sets with those present in destination " << dest << "." << endl;
3565 InsertGraphIntoGraph(out, *FragmentLowerOrdersList[RootNr][dest], *FragmentList, &NumMolecules);
3566 delete(FragmentList);
3567 }
3568 }
3569 *out << Verbose(1) << "Number of resulting molecules for SubOrder " << SubOrder << " is: " << NumMolecules << "." << endl;
3570 }
3571 }
3572 }
3573 // now, we have completely filled each cell of FragmentLowerOrdersList[] for the current Walker->AdaptiveOrder
3574 //NumMoleculesOfOrder[Walker->AdaptiveOrder-1] = NumMolecules;
3575 TotalNumMolecules += NumMoleculesOfOrder[RootNr];
3576 *out << Verbose(1) << "Number of resulting molecules for Order " << Walker->AdaptiveOrder << " is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
3577 RootStack.push_back(RootKeyNr); // put back on stack
3578 RootNr++;
3579 }
3580 *out << Verbose(0) << "==============================================================================================================" << endl;
3581 *out << Verbose(1) << "Total number of resulting molecules is: " << TotalNumMolecules << "." << endl;
3582 *out << Verbose(0) << "==============================================================================================================" << endl;
3583 // now, FragmentLowerOrdersList is complete, it looks - for BondOrder 5 - as this (number is the ANOVA Order of the terms therein)
3584 // 5433222211111111
3585 // 43221111
3586 // 3211
3587 // 21
3588 // 1
3589 // Subsequently, we combine all into a single list (FragmentList)
3590
3591 *out << Verbose(0) << "Combining the lists of all orders per order and finally into a single one." << endl;
3592 FragmentList = new Graph;
3593 RootNr = 0;
3594 while (!RootStack.empty()) {
3595 RootKeyNr = RootStack.front();
3596 RootStack.pop_front();
3597 Walker = FindAtom(RootKeyNr);
3598 NumLevels = 1 << (Walker->AdaptiveOrder - 1);
3599 for(int i=0;i<NumLevels;i++) {
3600 InsertGraphIntoGraph(out, *FragmentList, (*FragmentLowerOrdersList[RootNr][i]), &counter);
3601 delete(FragmentLowerOrdersList[RootNr][i]);
3602 }
3603 Free((void **)&FragmentLowerOrdersList[RootNr], "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
3604 RootNr++;
3605 }
3606 Free((void **)&FragmentLowerOrdersList, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
3607 Free((void **)&NumMoleculesOfOrder, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
3608
3609 *out << Verbose(0) << "End of FragmentBOSSANOVA." << endl;
3610 return FragmentList;
3611};
3612
3613/** Comparision function for GSL heapsort on distances in two molecules.
3614 * \param *a
3615 * \param *b
3616 * \return <0, \a *a less than \a *b, ==0 if equal, >0 \a *a greater than \a *b
3617 */
3618int CompareDoubles (const void * a, const void * b)
3619{
3620 if (*(double *)a > *(double *)b)
3621 return -1;
3622 else if (*(double *)a < *(double *)b)
3623 return 1;
3624 else
3625 return 0;
3626};
3627
3628/** Determines whether two molecules actually contain the same atoms and coordination.
3629 * \param *out output stream for debugging
3630 * \param *OtherMolecule the molecule to compare this one to
3631 * \param threshold upper limit of difference when comparing the coordination.
3632 * \return NULL - not equal, otherwise an allocated (molecule::AtomCount) permutation map of the atom numbers (which corresponds to which)
3633 */
3634int * molecule::IsEqualToWithinThreshold(ofstream *out, molecule *OtherMolecule, double threshold)
3635{
3636 int flag;
3637 double *Distances = NULL, *OtherDistances = NULL;
3638 vector CenterOfGravity, OtherCenterOfGravity;
3639 size_t *PermMap = NULL, *OtherPermMap = NULL;
3640 int *PermutationMap = NULL;
3641 atom *Walker = NULL;
3642 bool result = true; // status of comparison
3643
3644 *out << Verbose(3) << "Begin of IsEqualToWithinThreshold." << endl;
3645 /// first count both their atoms and elements and update lists thereby ...
3646 //*out << Verbose(0) << "Counting atoms, updating list" << endl;
3647 CountAtoms(out);
3648 OtherMolecule->CountAtoms(out);
3649 CountElements();
3650 OtherMolecule->CountElements();
3651
3652 /// ... and compare:
3653 /// -# AtomCount
3654 if (result) {
3655 if (AtomCount != OtherMolecule->AtomCount) {
3656 *out << Verbose(4) << "AtomCounts don't match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
3657 result = false;
3658 } else *out << Verbose(4) << "AtomCounts match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
3659 }
3660 /// -# ElementCount
3661 if (result) {
3662 if (ElementCount != OtherMolecule->ElementCount) {
3663 *out << Verbose(4) << "ElementCount don't match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
3664 result = false;
3665 } else *out << Verbose(4) << "ElementCount match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
3666 }
3667 /// -# ElementsInMolecule
3668 if (result) {
3669 for (flag=0;flag<MAX_ELEMENTS;flag++) {
3670 //*out << Verbose(5) << "Element " << flag << ": " << ElementsInMolecule[flag] << " <-> " << OtherMolecule->ElementsInMolecule[flag] << "." << endl;
3671 if (ElementsInMolecule[flag] != OtherMolecule->ElementsInMolecule[flag])
3672 break;
3673 }
3674 if (flag < MAX_ELEMENTS) {
3675 *out << Verbose(4) << "ElementsInMolecule don't match." << endl;
3676 result = false;
3677 } else *out << Verbose(4) << "ElementsInMolecule match." << endl;
3678 }
3679 /// then determine and compare center of gravity for each molecule ...
3680 if (result) {
3681 *out << Verbose(5) << "Calculating Centers of Gravity" << endl;
3682 DetermineCenterOfGravity(CenterOfGravity);
3683 OtherMolecule->DetermineCenterOfGravity(OtherCenterOfGravity);
3684 *out << Verbose(5) << "Center of Gravity: ";
3685 CenterOfGravity.Output(out);
3686 *out << endl << Verbose(5) << "Other Center of Gravity: ";
3687 OtherCenterOfGravity.Output(out);
3688 *out << endl;
3689 if (CenterOfGravity.Distance(&OtherCenterOfGravity) > threshold) {
3690 *out << Verbose(4) << "Centers of gravity don't match." << endl;
3691 result = false;
3692 }
3693 }
3694
3695 /// ... then make a list with the euclidian distance to this center for each atom of both molecules
3696 if (result) {
3697 *out << Verbose(5) << "Calculating distances" << endl;
3698 Distances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: Distances");
3699 OtherDistances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: OtherDistances");
3700 Walker = start;
3701 while (Walker->next != end) {
3702 Walker = Walker->next;
3703 //for (i=0;i<AtomCount;i++) {
3704 Distances[Walker->nr] = CenterOfGravity.Distance(&Walker->x);
3705 }
3706 Walker = OtherMolecule->start;
3707 while (Walker->next != OtherMolecule->end) {
3708 Walker = Walker->next;
3709 OtherDistances[Walker->nr] = OtherCenterOfGravity.Distance(&Walker->x);
3710 }
3711
3712 /// ... sort each list (using heapsort (o(N log N)) from GSL)
3713 *out << Verbose(5) << "Sorting distances" << endl;
3714 PermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermMap");
3715 OtherPermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
3716 gsl_heapsort_index (PermMap, Distances, AtomCount, sizeof(double), CompareDoubles);
3717 gsl_heapsort_index (OtherPermMap, OtherDistances, AtomCount, sizeof(double), CompareDoubles);
3718 PermutationMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermutationMap");
3719 *out << Verbose(5) << "Combining Permutation Maps" << endl;
3720 for(int i=0;i<AtomCount;i++)
3721 PermutationMap[PermMap[i]] = (int) OtherPermMap[i];
3722
3723 /// ... and compare them step by step, whether the difference is individiually(!) below \a threshold for all
3724 *out << Verbose(4) << "Comparing distances" << endl;
3725 flag = 0;
3726 for (int i=0;i<AtomCount;i++) {
3727 *out << Verbose(5) << "Distances: |" << Distances[PermMap[i]] << " - " << OtherDistances[OtherPermMap[i]] << "| = " << fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) << " ?<? " << threshold << endl;
3728 if (fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) > threshold)
3729 flag = 1;
3730 }
3731 Free((void **)&PermMap, "molecule::IsEqualToWithinThreshold: *PermMap");
3732 Free((void **)&OtherPermMap, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
3733
3734 /// free memory
3735 Free((void **)&Distances, "molecule::IsEqualToWithinThreshold: Distances");
3736 Free((void **)&OtherDistances, "molecule::IsEqualToWithinThreshold: OtherDistances");
3737 if (flag) { // if not equal
3738 Free((void **)&PermutationMap, "molecule::IsEqualToWithinThreshold: *PermutationMap");
3739 result = false;
3740 }
3741 }
3742 /// return pointer to map if all distances were below \a threshold
3743 *out << Verbose(3) << "End of IsEqualToWithinThreshold." << endl;
3744 if (result) {
3745 *out << Verbose(3) << "Result: Equal." << endl;
3746 return PermutationMap;
3747 } else {
3748 *out << Verbose(3) << "Result: Not equal." << endl;
3749 return NULL;
3750 }
3751};
3752
3753/** Returns an index map for two father-son-molecules.
3754 * The map tells which atom in this molecule corresponds to which one in the other molecul with their fathers.
3755 * \param *out output stream for debugging
3756 * \param *OtherMolecule corresponding molecule with fathers
3757 * \return allocated map of size molecule::AtomCount with map
3758 * \todo make this with a good sort O(n), not O(n^2)
3759 */
3760int * molecule::GetFatherSonAtomicMap(ofstream *out, molecule *OtherMolecule)
3761{
3762 atom *Walker = NULL, *OtherWalker = NULL;
3763 *out << Verbose(3) << "Begin of GetFatherAtomicMap." << endl;
3764 int *AtomicMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::GetAtomicMap: *AtomicMap"); //Calloc
3765 for (int i=0;i<AtomCount;i++)
3766 AtomicMap[i] = -1;
3767 if (OtherMolecule == this) { // same molecule
3768 for (int i=0;i<AtomCount;i++) // no need as -1 means already that there is trivial correspondence
3769 AtomicMap[i] = i;
3770 *out << Verbose(4) << "Map is trivial." << endl;
3771 } else {
3772 *out << Verbose(4) << "Map is ";
3773 Walker = start;
3774 while (Walker->next != end) {
3775 Walker = Walker->next;
3776 if (Walker->father == NULL) {
3777 AtomicMap[Walker->nr] = -2;
3778 } else {
3779 OtherWalker = OtherMolecule->start;
3780 while (OtherWalker->next != OtherMolecule->end) {
3781 OtherWalker = OtherWalker->next;
3782 //for (int i=0;i<AtomCount;i++) { // search atom
3783 //for (int j=0;j<OtherMolecule->AtomCount;j++) {
3784 //*out << Verbose(4) << "Comparing father " << Walker->father << " with the other one " << OtherWalker->father << "." << endl;
3785 if (Walker->father == OtherWalker)
3786 AtomicMap[Walker->nr] = OtherWalker->nr;
3787 }
3788 }
3789 *out << AtomicMap[Walker->nr] << "\t";
3790 }
3791 *out << endl;
3792 }
3793 *out << Verbose(3) << "End of GetFatherAtomicMap." << endl;
3794 return AtomicMap;
3795};
3796
Note: See TracBrowser for help on using the repository browser.