source: molecuilder/src/molecules.cpp@ d50d2a

Last change on this file since d50d2a was d50d2a, checked in by Frederik Heber <heber@…>, 17 years ago

AtomStackClass -> template <typename T> StackClass<T> change in new file stackclass.hpp, other templates to helpers.hpp

StackClass was changed to a template and as template code may only be present in header file was moved to a new
header file stackclass.hpp. New file was added to Makefile.am
Other templates (list management et al) were moved to helpers.hpp as they are more sensibly placed there.

  • Property mode set to 100644
File size: 179.1 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 int NumberCells, divisor[NDIM], n[NDIM], N[NDIM], index, Index, j;
1252 molecule **CellList;
1253 double distance, MinDistance, MaxDistance;
1254 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
1255 vector x;
1256
1257 BondDistance = bonddistance;
1258 *out << Verbose(0) << "Begin of CreateAdjacencyList." << endl;
1259 // remove every bond from the list
1260 if ((first->next != last) && (last->previous != first)) { // there are bonds present
1261 cleanup(first,last);
1262 }
1263
1264 // count atoms in molecule = dimension of matrix (also give each unique name and continuous numbering)
1265 CountAtoms(out);
1266 *out << Verbose(1) << "AtomCount " << AtomCount << "." << endl;
1267
1268 if (AtomCount != 0) {
1269 // 1. find divisor for each axis, such that a sphere with radius of at least bonddistance can be placed into each cell
1270 j=-1;
1271 for (int i=0;i<NDIM;i++) {
1272 j += i+1;
1273 divisor[i] = (int)floor(cell_size[j]/bonddistance); // take smaller value such that size of linked cell is at least bonddistance
1274 *out << Verbose(1) << "divisor[" << i << "] = " << divisor[i] << "." << endl;
1275 }
1276 // 2a. allocate memory for the cell list
1277 NumberCells = divisor[0]*divisor[1]*divisor[2];
1278 *out << Verbose(1) << "Allocating " << NumberCells << " cells." << endl;
1279 CellList = (molecule **) Malloc(sizeof(molecule *)*NumberCells, "molecule::CreateAdjacencyList - ** CellList");
1280 for (int i=0;i<NumberCells;i++)
1281 CellList[i] = NULL;
1282
1283 // 2b. put all atoms into its corresponding list
1284 Walker = start;
1285 while(Walker->next != end) {
1286 Walker = Walker->next;
1287 //*out << Verbose(1) << "Current atom is " << *Walker << " with coordinates ";
1288 //Walker->x.Output(out);
1289 //*out << "." << endl;
1290 // compute the cell by the atom's coordinates
1291 j=-1;
1292 for (int i=0;i<NDIM;i++) {
1293 j += i+1;
1294 x.CopyVector(&(Walker->x));
1295 x.KeepPeriodic(out, matrix);
1296 n[i] = (int)floor(x.x[i]/cell_size[j]*(double)divisor[i]);
1297 }
1298 index = n[2] + (n[1] + n[0] * divisor[1]) * divisor[2];
1299 *out << Verbose(1) << "Atom " << *Walker << " goes into cell number [" << n[0] << "," << n[1] << "," << n[2] << "] = " << index << "." << endl;
1300 // add copy atom to this cell
1301 if (CellList[index] == NULL) // allocate molecule if not done
1302 CellList[index] = new molecule(elemente);
1303 OtherWalker = CellList[index]->AddCopyAtom(Walker); // add a copy of walker to this atom, father will be walker for later reference
1304 //*out << Verbose(1) << "Copy Atom is " << *OtherWalker << "." << endl;
1305 }
1306 //for (int i=0;i<NumberCells;i++)
1307 //*out << Verbose(1) << "Cell number " << i << ": " << CellList[i] << "." << endl;
1308
1309 // 3a. go through every cell
1310 for (N[0]=0;N[0]<divisor[0];N[0]++)
1311 for (N[1]=0;N[1]<divisor[1];N[1]++)
1312 for (N[2]=0;N[2]<divisor[2];N[2]++) {
1313 Index = N[2] + (N[1] + N[0] * divisor[1]) * divisor[2];
1314 if (CellList[Index] != NULL) { // if there atoms in this cell
1315 //*out << Verbose(1) << "Current cell is " << Index << "." << endl;
1316 // 3b. for every atom therein
1317 Walker = CellList[Index]->start;
1318 while (Walker->next != CellList[Index]->end) { // go through every atom
1319 Walker = Walker->next;
1320 //*out << Verbose(0) << "Current Atom is " << *Walker << "." << endl;
1321 // 3c. check for possible bond between each atom in this and every one in the 27 cells
1322 for (n[0]=-1;n[0]<=1;n[0]++)
1323 for (n[1]=-1;n[1]<=1;n[1]++)
1324 for (n[2]=-1;n[2]<=1;n[2]++) {
1325 // compute the index of this comparison cell and make it periodic
1326 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];
1327 //*out << Verbose(1) << "Number of comparison cell is " << index << "." << endl;
1328 if (CellList[index] != NULL) { // if there are any atoms in this cell
1329 OtherWalker = CellList[index]->start;
1330 while(OtherWalker->next != CellList[index]->end) { // go through every atom in this cell
1331 OtherWalker = OtherWalker->next;
1332 //*out << Verbose(0) << "Current comparison atom is " << *OtherWalker << "." << endl;
1333 /// \todo periodic check is missing here!
1334 //*out << Verbose(1) << "Checking distance " << OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size) << " against typical bond length of " << bonddistance*bonddistance << "." << endl;
1335 MinDistance = OtherWalker->type->CovalentRadius + Walker->type->CovalentRadius;
1336 MaxDistance = MinDistance + BONDTHRESHOLD;
1337 MinDistance -= BONDTHRESHOLD;
1338 distance = OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size);
1339 if ((OtherWalker->father->nr > Walker->father->nr) && (distance <= MaxDistance*MaxDistance) && (distance >= MinDistance*MinDistance)) { // create bond if distance is smaller
1340 *out << Verbose(0) << "Adding Bond between " << *Walker << " and " << *OtherWalker << "." << endl;
1341 AddBond(Walker->father, OtherWalker->father, 1); // also increases molecule::BondCount
1342 } else {
1343 //*out << Verbose(1) << "Not Adding: Wrong label order or distance too great." << endl;
1344 }
1345 }
1346 }
1347 }
1348 }
1349 }
1350 }
1351 // 4. free the cell again
1352 for (int i=0;i<NumberCells;i++)
1353 if (CellList[i] != NULL) {
1354 delete(CellList[i]);
1355 }
1356 Free((void **)&CellList, "molecule::CreateAdjacencyList - ** CellList");
1357
1358 // create the adjacency list per atom
1359 CreateListOfBondsPerAtom(out);
1360
1361 // correct Bond degree of each bond by checking of updated(!) sum of bond degree for an atom match its valence count
1362 // bond degrres are correctled iteratively by one, so that 2-2 instead of 1-3 or 3-1 corrections are favoured: We want
1363 // a rather symmetric distribution of higher bond degrees
1364 if (BondCount != 0) {
1365 NoCyclicBonds = 0;
1366 *out << Verbose(1) << "Correcting Bond degree of each bond ... ";
1367 do {
1368 No = 0; // No acts as breakup flag (if 1 we still continue)
1369 Walker = start;
1370 while (Walker->next != end) { // go through every atom
1371 Walker = Walker->next;
1372 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through each of its bond partners
1373 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
1374 // count valence of first partner (updated!), might have changed during last bond partner
1375 NoBonds = 0;
1376 for(j=0;j<NumberOfBondsPerAtom[Walker->nr];j++)
1377 NoBonds += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
1378 //*out << Verbose(3) << "Walker: " << (int)Walker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
1379 if ((int)(Walker->type->NoValenceOrbitals) > NoBonds) { // we have a mismatch, check NoBonds of other atom
1380 // count valence of second partner
1381 NoBonds = 0;
1382 for(j=0;j<NumberOfBondsPerAtom[OtherWalker->nr];j++)
1383 NoBonds += ListOfBondsPerAtom[OtherWalker->nr][j]->BondDegree;
1384 //*out << Verbose(3) << "OtherWalker: " << (int)OtherWalker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
1385 if ((int)(OtherWalker->type->NoValenceOrbitals) > NoBonds) // increase bond degree by one
1386 ListOfBondsPerAtom[Walker->nr][i]->BondDegree++;
1387 }
1388 }
1389 }
1390 } while (No);
1391 *out << " done." << endl;
1392 } else
1393 *out << Verbose(1) << "BondCount is " << BondCount << ", no bonds between any of the " << AtomCount << " atoms." << endl;
1394 *out << Verbose(1) << "I detected " << BondCount << " bonds in the molecule with distance " << bonddistance << "." << endl;
1395
1396// // output bonds for debugging (if bond chain list was correctly installed)
1397// *out << Verbose(1) << endl << "From contents of bond chain list:";
1398// bond *Binder = first;
1399// while(Binder->next != last) {
1400// Binder = Binder->next;
1401// *out << *Binder << "\t" << endl;
1402// }
1403// *out << endl;
1404 } else
1405 *out << Verbose(1) << "AtomCount is " << AtomCount << ", thus no bonds, no connections!." << endl;
1406 *out << Verbose(0) << "End of CreateAdjacencyList." << endl;
1407 Free((void **)&matrix, "molecule::CreateAdjacencyList: *matrix");
1408};
1409
1410/** Performs a Depth-First search on this molecule.
1411 * Marks bonds in molecule as cyclic, bridge, ... and atoms as
1412 * articulations points, ...
1413 * We use the algorithm from [Even, Graph Algorithms, p.62].
1414 * \param *out output stream for debugging
1415 * \param ReturnStack true - return pointer to atom stack of separable components, false - return NULL
1416 * \param MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found
1417 * \return list of each disconnected subgraph as an individual molecule class structure
1418 */
1419MoleculeLeafClass * molecule::DepthFirstSearchAnalysis(ofstream *out, bool ReturnStack, int &MinimumRingSize)
1420{
1421 class StackClass<atom *> *AtomStack;
1422 AtomStack = new StackClass<atom *>(AtomCount);
1423 class StackClass<bond *> *BackEdgeStack = new StackClass<bond *> (BondCount);
1424 MoleculeLeafClass *SubGraphs = new MoleculeLeafClass(NULL);
1425 MoleculeLeafClass *LeafWalker = SubGraphs;
1426 int CurrentGraphNr = 0, OldGraphNr;
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 BackEdgeStack->ClearStack();
1438 while (Root != end) { // if there any atoms at all
1439 // (1) mark all edges unused, empty stack, set atom->GraphNr = 0 for all
1440 AtomStack->ClearStack();
1441
1442 // put into new subgraph molecule and add this to list of subgraphs
1443 LeafWalker = new MoleculeLeafClass(LeafWalker);
1444 LeafWalker->Leaf = new molecule(elemente);
1445 LeafWalker->Leaf->AddCopyAtom(Root);
1446
1447 OldGraphNr = CurrentGraphNr;
1448 Walker = Root;
1449 do { // (10)
1450 do { // (2) set number and Lowpoint of Atom to i, increase i, push current atom
1451 if (!BackStepping) { // if we don't just return from (8)
1452 Walker->GraphNr = CurrentGraphNr;
1453 Walker->LowpointNr = CurrentGraphNr;
1454 *out << Verbose(1) << "Setting Walker[" << Walker->Name << "]'s number to " << Walker->GraphNr << " with Lowpoint " << Walker->LowpointNr << "." << endl;
1455 AtomStack->Push(Walker);
1456 CurrentGraphNr++;
1457 }
1458 do { // (3) if Walker has no unused egdes, go to (5)
1459 BackStepping = false; // reset backstepping flag for (8)
1460 if (Binder == NULL) // if we don't just return from (11), Binder is already set to next unused
1461 Binder = FindNextUnused(Walker);
1462 if (Binder == NULL)
1463 break;
1464 *out << Verbose(2) << "Current Unused Bond is " << *Binder << "." << endl;
1465 // (4) Mark Binder used, ...
1466 Binder->MarkUsed(black);
1467 OtherAtom = Binder->GetOtherAtom(Walker);
1468 *out << Verbose(2) << "(4) OtherAtom is " << OtherAtom->Name << "." << endl;
1469 if (OtherAtom->GraphNr != -1) {
1470 // (4a) ... if "other" atom has been visited (GraphNr != 0), set lowpoint to minimum of both, go to (3)
1471 Binder->Type = BackEdge;
1472 BackEdgeStack->Push(Binder);
1473 Walker->LowpointNr = ( Walker->LowpointNr < OtherAtom->GraphNr ) ? Walker->LowpointNr : OtherAtom->GraphNr;
1474 *out << Verbose(3) << "(4a) Visited: Setting Lowpoint of Walker[" << Walker->Name << "] to " << Walker->LowpointNr << "." << endl;
1475 } else {
1476 // (4b) ... otherwise set OtherAtom as Ancestor of Walker and Walker as OtherAtom, go to (2)
1477 Binder->Type = TreeEdge;
1478 OtherAtom->Ancestor = Walker;
1479 Walker = OtherAtom;
1480 *out << Verbose(3) << "(4b) Not Visited: OtherAtom[" << OtherAtom->Name << "]'s Ancestor is now " << OtherAtom->Ancestor->Name << ", Walker is OtherAtom " << OtherAtom->Name << "." << endl;
1481 break;
1482 }
1483 Binder = NULL;
1484 } while (1); // (3)
1485 if (Binder == NULL) {
1486 *out << Verbose(2) << "No more Unused Bonds." << endl;
1487 break;
1488 } else
1489 Binder = NULL;
1490 } while (1); // (2)
1491
1492 // 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!
1493 if ((Walker == Root) && (Binder == NULL))
1494 break;
1495
1496 // (5) if Ancestor of Walker is ...
1497 *out << Verbose(1) << "(5) Number of Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "] is " << Walker->Ancestor->GraphNr << "." << endl;
1498 if (Walker->Ancestor->GraphNr != Root->GraphNr) {
1499 // (6) (Ancestor of Walker is not Root)
1500 if (Walker->LowpointNr < Walker->Ancestor->GraphNr) {
1501 // (6a) set Ancestor's Lowpoint number to minimum of of its Ancestor and itself, go to Step(8)
1502 Walker->Ancestor->LowpointNr = (Walker->Ancestor->LowpointNr < Walker->LowpointNr) ? Walker->Ancestor->LowpointNr : Walker->LowpointNr;
1503 *out << Verbose(2) << "(6) Setting Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s Lowpoint to " << Walker->Ancestor->LowpointNr << "." << endl;
1504 } else {
1505 // (7) (Ancestor of Walker is a separating vertex, remove all from stack till Walker (including), these and Ancestor form a component
1506 Walker->Ancestor->SeparationVertex = true;
1507 *out << Verbose(2) << "(7) Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s is a separating vertex, creating component." << endl;
1508 SetNextComponentNumber(Walker->Ancestor, ComponentNumber);
1509 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Ancestor's Compont is " << ComponentNumber << "." << endl;
1510 SetNextComponentNumber(Walker, ComponentNumber);
1511 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1512 do {
1513 OtherAtom = AtomStack->PopLast();
1514 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
1515 SetNextComponentNumber(OtherAtom, ComponentNumber);
1516 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1517 } while (OtherAtom != Walker);
1518 ComponentNumber++;
1519 }
1520 // (8) Walker becomes its Ancestor, go to (3)
1521 *out << Verbose(2) << "(8) Walker[" << Walker->Name << "] is now its Ancestor " << Walker->Ancestor->Name << ", backstepping. " << endl;
1522 Walker = Walker->Ancestor;
1523 BackStepping = true;
1524 }
1525 if (!BackStepping) { // coming from (8) want to go to (3)
1526 // (9) remove all from stack till Walker (including), these and Root form a component
1527 AtomStack->Output(out);
1528 SetNextComponentNumber(Root, ComponentNumber);
1529 *out << Verbose(3) << "(9) Root[" << Root->Name << "]'s Component is " << ComponentNumber << "." << endl;
1530 SetNextComponentNumber(Walker, ComponentNumber);
1531 *out << Verbose(3) << "(9) Walker[" << Walker->Name << "]'s Component is " << ComponentNumber << "." << endl;
1532 do {
1533 OtherAtom = AtomStack->PopLast();
1534 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
1535 SetNextComponentNumber(OtherAtom, ComponentNumber);
1536 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1537 } while (OtherAtom != Walker);
1538 ComponentNumber++;
1539
1540 // (11) Root is separation vertex, set Walker to Root and go to (4)
1541 Walker = Root;
1542 Binder = FindNextUnused(Walker);
1543 *out << Verbose(1) << "(10) Walker is Root[" << Root->Name << "], next Unused Bond is " << Binder << "." << endl;
1544 if (Binder != NULL) { // Root is separation vertex
1545 *out << Verbose(1) << "(11) Root is a separation vertex." << endl;
1546 Walker->SeparationVertex = true;
1547 }
1548 }
1549 } while ((BackStepping) || (Binder != NULL)); // (10) halt only if Root has no unused edges
1550
1551 // From OldGraphNr to CurrentGraphNr ranges an disconnected subgraph
1552 *out << Verbose(0) << "Disconnected subgraph ranges from " << OldGraphNr << " to " << CurrentGraphNr << "." << endl;
1553 LeafWalker->Leaf->Output(out);
1554 *out << endl;
1555
1556 // step on to next root
1557 while ((Root != end) && (Root->GraphNr != -1)) {
1558 //*out << Verbose(1) << "Current next subgraph root candidate is " << Root->Name << "." << endl;
1559 if (Root->GraphNr != -1) // if already discovered, step on
1560 Root = Root->next;
1561 }
1562 }
1563 // set cyclic bond criterium on "same LP" basis
1564 Binder = first;
1565 while(Binder->next != last) {
1566 Binder = Binder->next;
1567 if (Binder->rightatom->LowpointNr == Binder->leftatom->LowpointNr) { // cyclic ??
1568 Binder->Cyclic = true;
1569 NoCyclicBonds++;
1570 }
1571 }
1572
1573 // analysis of the cycles (print rings, get minimum cycle length)
1574 CyclicStructureAnalysis(out, MinimumRingSize);
1575
1576 *out << Verbose(1) << "Final graph info for each atom is:" << endl;
1577 Walker = start;
1578 while (Walker->next != end) {
1579 Walker = Walker->next;
1580 *out << Verbose(2) << "Atom " << Walker->Name << " is " << ((Walker->SeparationVertex) ? "a" : "not a") << " separation vertex, components are ";
1581 OutputComponentNumber(out, Walker);
1582 *out << " with Lowpoint " << Walker->LowpointNr << " and Graph Nr. " << Walker->GraphNr << "." << endl;
1583 }
1584
1585 *out << Verbose(1) << "Final graph info for each bond is:" << endl;
1586 Binder = first;
1587 while(Binder->next != last) {
1588 Binder = Binder->next;
1589 *out << Verbose(2) << ((Binder->Type == TreeEdge) ? "TreeEdge " : "BackEdge ") << *Binder << ": <";
1590 *out << ((Binder->leftatom->SeparationVertex) ? "SP," : "") << "L" << Binder->leftatom->LowpointNr << " G" << Binder->leftatom->GraphNr << " Comp.";
1591 OutputComponentNumber(out, Binder->leftatom);
1592 *out << " === ";
1593 *out << ((Binder->rightatom->SeparationVertex) ? "SP," : "") << "L" << Binder->rightatom->LowpointNr << " G" << Binder->rightatom->GraphNr << " Comp.";
1594 OutputComponentNumber(out, Binder->rightatom);
1595 *out << ">." << endl;
1596 if (Binder->Cyclic) // cyclic ??
1597 *out << Verbose(3) << "Lowpoint at each side are equal: CYCLIC!" << endl;
1598 }
1599
1600 // free all and exit
1601 delete(AtomStack);
1602 *out << Verbose(0) << "End of DepthFirstSearchAnalysis" << endl;
1603 return SubGraphs;
1604};
1605
1606/** Analyses the cycles found and returns minimum of all cycle lengths.
1607 * \param *out output stream for debugging
1608 * \param MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found
1609 * \todo BFS from the not-same-LP to find back to starting point of tributary cycle over more than one bond
1610 */
1611void molecule::CyclicStructureAnalysis(ofstream *out, int &MinimumRingSize)
1612{
1613 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
1614 int LP;
1615 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Runner = NULL;
1616 bond *Binder = NULL;
1617 int RingSize, NumCycles;
1618
1619 // count number of cyclic bonds per atom and push all with 3 cyclic bonds onto stack
1620 AtomStack->ClearStack();
1621 int *NoCyclicBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CyclicStructureAnalysis: *NoCyclicBondsPerAtom");
1622 Walker = start;
1623 while (Walker->next != end) {
1624 Walker = Walker->next;
1625 NoCyclicBondsPerAtom[Walker->nr] = 0;
1626 // check whether it's connected to cyclic bonds and count per atom
1627 // 0 means not part of a cycle, 2 means in a cycle, 3 or more means interconnection site of cycles
1628 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
1629 Binder = ListOfBondsPerAtom[Walker->nr][i];
1630 NoCyclicBondsPerAtom[Walker->nr] += (int) Binder->Cyclic;
1631 if (NoCyclicBondsPerAtom[Walker->nr] == 3) //push all intersections
1632 AtomStack->Push(Walker);
1633 }
1634 }
1635 *out << Verbose(1) << "NoCyclicBondsPerAtom: ";
1636 for(int i=0;i<AtomCount;i++) {
1637 *out << NoCyclicBondsPerAtom[i] << " ";
1638 }
1639 *out << endl;
1640 *out << Verbose(1) << "Analysing cycles ... " << endl;
1641 MinimumRingSize = -1;
1642 NumCycles = 0;
1643 while (!AtomStack->IsEmpty()) {
1644 Walker = AtomStack->PopFirst();
1645 if (NoCyclicBondsPerAtom[Walker->nr] > 1) {
1646 NoCyclicBondsPerAtom[Walker->nr]--; // remove one for being intersection
1647 RingSize = 0;
1648 *out << Verbose(2) << "Current intersection is " << *Walker << ", expect to find " << NoCyclicBondsPerAtom[Walker->nr] << " cycles." << endl;
1649 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
1650 Binder = ListOfBondsPerAtom[Walker->nr][i];
1651 OtherAtom = Binder->GetOtherAtom(Walker);
1652 // note down the LowPoint number of this cycle
1653 if (NoCyclicBondsPerAtom[OtherAtom->nr] > 1) {
1654 LP = OtherAtom->LowpointNr;
1655 NoCyclicBondsPerAtom[Walker->nr]--; // walker is start of cycle
1656 if (LP != Walker->LowpointNr)
1657 *out << Verbose(2) << "Tributary cycle: ... <-> " << Walker->Name;
1658 else
1659 *out << Verbose(2) << "Main cycle: ... <-> " << Walker->Name;
1660 Root = Walker; // root acts as predecessor marker so that we don't step back accidentally
1661 RingSize = 1;
1662 do {
1663 for(int j=0;j<NumberOfBondsPerAtom[OtherAtom->nr];j++) { // search among its bonds for next in cycle (same lowpoint nr)
1664 Runner = ListOfBondsPerAtom[OtherAtom->nr][j]->GetOtherAtom(OtherAtom);
1665 if (((Runner->LowpointNr == LP) || (Runner->LowpointNr == Walker->LowpointNr)) && (Runner != Root)) {
1666 // first check is to stay in the cycle
1667 // middle check is allow for getting back into main cycle briefly from tributary cycle (just one step, then while further down stops)
1668 // last check is not step back
1669 *out << " <-> " << OtherAtom->Name;
1670 NoCyclicBondsPerAtom[OtherAtom->nr]--;
1671 Root = OtherAtom;
1672 OtherAtom = Runner;
1673 NoCyclicBondsPerAtom[Root->nr]--;
1674 RingSize++;
1675 break;
1676 }
1677 }
1678 } while ((OtherAtom->LowpointNr == LP) && (Walker != OtherAtom) && (Root->LowpointNr == OtherAtom->LowpointNr));
1679 // now check if the LP is different from Walker's, as then there is one more bond to follow
1680 if (LP != Walker->LowpointNr) {
1681 // find last bond to home base
1682 for(int j=0;j<NumberOfBondsPerAtom[OtherAtom->nr];j++)
1683 if (ListOfBondsPerAtom[OtherAtom->nr][j]->GetOtherAtom(OtherAtom) == Root) {
1684 *out << " <-> " << OtherAtom->Name;
1685 RingSize++;
1686 NoCyclicBondsPerAtom[OtherAtom->nr]--;
1687 }
1688 } else {
1689 // we have made the complete cycle
1690 }
1691 *out << " <-> ... with cycle length of " << RingSize << "." << endl;
1692 NumCycles++;
1693 if ((RingSize < MinimumRingSize) || (MinimumRingSize == -1))
1694 MinimumRingSize = RingSize;
1695 }
1696 }
1697 }
1698 }
1699
1700 // print NoCyclicBondsPerAtom to visually check of all are zero
1701 *out << Verbose(1) << "NoCyclicBondsPerAtom: ";
1702 for(int i=0;i<AtomCount;i++) {
1703 if (NoCyclicBondsPerAtom[i] > 0)
1704 cerr << "There was an error in molecule::CyclicStructureAnalysis!" << endl;
1705 *out << NoCyclicBondsPerAtom[i] << " ";
1706 }
1707 *out << endl;
1708
1709 if (MinimumRingSize != -1)
1710 *out << Verbose(1) << "Minimum ring size is " << MinimumRingSize << ", over " << NumCycles << " cycles total." << endl;
1711 else
1712 *out << Verbose(1) << "No rings were detected in the molecular structure." << endl;
1713
1714 Free((void **)&NoCyclicBondsPerAtom, "molecule::CyclicStructureAnalysis: *NoCyclicBondsPerAtom");
1715 delete(AtomStack);
1716};
1717
1718/** Sets the next component number.
1719 * This is O(N) as the number of bonds per atom is bound.
1720 * \param *vertex atom whose next atom::*ComponentNr is to be set
1721 * \param nr number to use
1722 */
1723void molecule::SetNextComponentNumber(atom *vertex, int nr)
1724{
1725 int i=0;
1726 if (vertex != NULL) {
1727 for(;i<NumberOfBondsPerAtom[vertex->nr];i++) {
1728 if (vertex->ComponentNr[i] == -1) { // check if not yet used
1729 vertex->ComponentNr[i] = nr;
1730 break;
1731 }
1732 else if (vertex->ComponentNr[i] == nr) // if number is already present, don't add another time
1733 break; // breaking here will not cause error!
1734 }
1735 if (i == NumberOfBondsPerAtom[vertex->nr])
1736 cerr << "Error: All Component entries are already occupied!" << endl;
1737 } else
1738 cerr << "Error: Given vertex is NULL!" << endl;
1739};
1740
1741/** Output a list of flags, stating whether the bond was visited or not.
1742 * \param *out output stream for debugging
1743 */
1744void molecule::OutputComponentNumber(ofstream *out, atom *vertex)
1745{
1746 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
1747 *out << vertex->ComponentNr[i] << " ";
1748};
1749
1750/** Allocates memory for all atom::*ComponentNr in this molecule and sets each entry to -1.
1751 */
1752void molecule::InitComponentNumbers()
1753{
1754 atom *Walker = start;
1755 while(Walker->next != end) {
1756 Walker = Walker->next;
1757 if (Walker->ComponentNr != NULL)
1758 Free((void **)&Walker->ComponentNr, "molecule::InitComponentNumbers: **Walker->ComponentNr");
1759 Walker->ComponentNr = (int *) Malloc(sizeof(int)*NumberOfBondsPerAtom[Walker->nr], "molecule::InitComponentNumbers: *Walker->ComponentNr");
1760 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
1761 Walker->ComponentNr[i] = -1;
1762 }
1763};
1764
1765/** Returns next unused bond for this atom \a *vertex or NULL of none exists.
1766 * \param *vertex atom to regard
1767 * \return bond class or NULL
1768 */
1769bond * molecule::FindNextUnused(atom *vertex)
1770{
1771 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
1772 if (ListOfBondsPerAtom[vertex->nr][i]->IsUsed() == white)
1773 return(ListOfBondsPerAtom[vertex->nr][i]);
1774 return NULL;
1775};
1776
1777/** Resets bond::Used flag of all bonds in this molecule.
1778 * \return true - success, false - -failure
1779 */
1780void molecule::ResetAllBondsToUnused()
1781{
1782 bond *Binder = first;
1783 while (Binder->next != last) {
1784 Binder = Binder->next;
1785 Binder->ResetUsed();
1786 }
1787};
1788
1789/** Resets atom::nr to -1 of all atoms in this molecule.
1790 */
1791void molecule::ResetAllAtomNumbers()
1792{
1793 atom *Walker = start;
1794 while (Walker->next != end) {
1795 Walker = Walker->next;
1796 Walker->GraphNr = -1;
1797 }
1798};
1799
1800/** Output a list of flags, stating whether the bond was visited or not.
1801 * \param *out output stream for debugging
1802 * \param *list
1803 */
1804void OutputAlreadyVisited(ofstream *out, int *list)
1805{
1806 *out << Verbose(4) << "Already Visited Bonds:\t";
1807 for(int i=1;i<=list[0];i++) *out << Verbose(0) << list[i] << " ";
1808 *out << endl;
1809};
1810
1811/** Estimates by educated guessing (using upper limit) the expected number of fragments.
1812 * The upper limit is
1813 * \f[
1814 * n = N \cdot C^k
1815 * \f]
1816 * where \f$C=2^c\f$ and c is the maximum bond degree over N number of atoms.
1817 * \param *out output stream for debugging
1818 * \param order bond order k
1819 * \return number n of fragments
1820 */
1821int molecule::GuesstimateFragmentCount(ofstream *out, int order)
1822{
1823 int c = 0;
1824 int FragmentCount;
1825 // get maximum bond degree
1826 atom *Walker = start;
1827 while (Walker->next != end) {
1828 Walker = Walker->next;
1829 c = (NumberOfBondsPerAtom[Walker->nr] > c) ? NumberOfBondsPerAtom[Walker->nr] : c;
1830 }
1831 FragmentCount = NoNonHydrogen*(1 << (c*order));
1832 *out << Verbose(1) << "Upper limit for this subgraph is " << FragmentCount << " for " << NoNonHydrogen << " non-H atoms with maximum bond degree of " << c << "." << endl;
1833 return FragmentCount;
1834};
1835
1836/** Scans a single line for number and puts them into \a KeySet.
1837 * \param *out output stream for debugging
1838 * \param *buffer buffer to scan
1839 * \param &CurrentSet filled KeySet on return
1840 * \return true - at least one valid atom id parsed, false - CurrentSet is empty
1841 */
1842bool molecule::ScanBufferIntoKeySet(ofstream *out, char *buffer, KeySet &CurrentSet)
1843{
1844 stringstream line;
1845 int AtomNr;
1846 int status = 0;
1847
1848 line.str(buffer);
1849 while (!line.eof()) {
1850 line >> AtomNr;
1851 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
1852 CurrentSet.insert(AtomNr); // insert at end, hence in same order as in file!
1853 status++;
1854 } // else it's "-1" or else and thus must not be added
1855 }
1856 *out << Verbose(1) << "The scanned KeySet is ";
1857 for(KeySet::iterator runner = CurrentSet.begin(); runner != CurrentSet.end(); runner++) {
1858 *out << (*runner) << "\t";
1859 }
1860 *out << endl;
1861 return (status != 0);
1862};
1863
1864/** Parses the KeySet file and fills \a *FragmentList from the known molecule structure.
1865 * Does two-pass scanning:
1866 * -# Scans the keyset file and initialises a temporary graph
1867 * -# Scans TEFactors file and sets the TEFactor of each key set in the temporary graph accordingly
1868 * Finally, the temporary graph is inserted into the given \a FragmentList for return.
1869 * \param *out output stream for debugging
1870 * \param *path path to file
1871 * \param *FragmentList empty, filled on return
1872 * \param IsAngstroem whether we have Ansgtroem or bohrradius
1873 * \return true - parsing successfully, false - failure on parsing (FragmentList will be NULL)
1874 */
1875bool molecule::ParseKeySetFile(ofstream *out, char *path, Graph *&FragmentList, bool IsAngstroem)
1876{
1877 bool status = true;
1878 ifstream InputFile;
1879 stringstream line;
1880 GraphTestPair testGraphInsert;
1881 int NumberOfFragments = 0;
1882 double TEFactor;
1883 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - filename");
1884
1885 if (FragmentList == NULL) { // check list pointer
1886 FragmentList = new Graph;
1887 }
1888
1889 // 1st pass: open file and read
1890 *out << Verbose(1) << "Parsing the KeySet file ... " << endl;
1891 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, KEYSETFILE);
1892 InputFile.open(filename);
1893 if (InputFile != NULL) {
1894 // each line represents a new fragment
1895 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - *buffer");
1896 // 1. parse keysets and insert into temp. graph
1897 while (!InputFile.eof()) {
1898 InputFile.getline(buffer, MAXSTRINGSIZE);
1899 KeySet CurrentSet;
1900 if ((strlen(buffer) > 0) && (ScanBufferIntoKeySet(out, buffer, CurrentSet))) { // if at least one valid atom was added, write config
1901 testGraphInsert = FragmentList->insert(GraphPair (CurrentSet,pair<int,double>(NumberOfFragments++,1))); // store fragment number and current factor
1902 if (!testGraphInsert.second) {
1903 cerr << "KeySet file must be corrupt as there are two equal key sets therein!" << endl;
1904 }
1905 //FragmentList->ListOfMolecules[NumberOfFragments++] = StoreFragmentFromKeySet(out, CurrentSet, IsAngstroem);
1906 }
1907 }
1908 // 2. Free and done
1909 InputFile.close();
1910 InputFile.clear();
1911 Free((void **)&buffer, "molecule::ParseKeySetFile - *buffer");
1912 *out << Verbose(1) << "done." << endl;
1913 } else {
1914 *out << Verbose(1) << "File " << filename << " not found." << endl;
1915 status = false;
1916 }
1917
1918 // 2nd pass: open TEFactors file and read
1919 *out << Verbose(1) << "Parsing the TEFactors file ... " << endl;
1920 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, TEFACTORSFILE);
1921 InputFile.open(filename);
1922 if (InputFile != NULL) {
1923 // 3. add found TEFactors to each keyset
1924 NumberOfFragments = 0;
1925 for(Graph::iterator runner = FragmentList->begin();runner != FragmentList->end(); runner++) {
1926 if (!InputFile.eof()) {
1927 InputFile >> TEFactor;
1928 (*runner).second.second = TEFactor;
1929 *out << Verbose(2) << "Setting " << ++NumberOfFragments << " fragment's TEFactor to " << (*runner).second.second << "." << endl;
1930 } else {
1931 status = false;
1932 break;
1933 }
1934 }
1935 // 4. Free and done
1936 InputFile.close();
1937 *out << Verbose(1) << "done." << endl;
1938 } else {
1939 *out << Verbose(1) << "File " << filename << " not found." << endl;
1940 status = false;
1941 }
1942
1943 Free((void **)&filename, "molecule::ParseKeySetFile - filename");
1944
1945 return status;
1946};
1947
1948/** Stores keysets and TEFactors to file.
1949 * \param *out output stream for debugging
1950 * \param KeySetList Graph with Keysets and factors
1951 * \param *path path to file
1952 * \return true - file written successfully, false - writing failed
1953 */
1954bool molecule::StoreKeySetFile(ofstream *out, Graph &KeySetList, char *path)
1955{
1956 ofstream output;
1957 bool status = true;
1958 string line;
1959 string::iterator ende;
1960
1961 // open KeySet file
1962 line = path;
1963 line.append("/");
1964 line += FRAGMENTPREFIX;
1965 ende = line.end();
1966 line += KEYSETFILE;
1967 output.open(line.c_str(), ios::out);
1968 *out << Verbose(1) << "Saving key sets of the total graph ... ";
1969 if(output != NULL) {
1970 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++) {
1971 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
1972 if (sprinter != (*runner).first.begin())
1973 output << "\t";
1974 output << *sprinter;
1975 }
1976 output << endl;
1977 }
1978 *out << "done." << endl;
1979 } else {
1980 cerr << "Unable to open " << line << " for writing keysets!" << endl;
1981 status = false;
1982 }
1983 output.close();
1984 output.clear();
1985
1986 // open TEFactors file
1987 line.erase(ende, line.end());
1988 line += TEFACTORSFILE;
1989 output.open(line.c_str(), ios::out);
1990 *out << Verbose(1) << "Saving TEFactors of the total graph ... ";
1991 if(output != NULL) {
1992 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++)
1993 output << (*runner).second.second << endl;
1994 *out << Verbose(1) << "done." << endl;
1995 } else {
1996 *out << Verbose(1) << "failed to open file" << line << "." << endl;
1997 status = false;
1998 }
1999 output.close();
2000
2001 return status;
2002};
2003
2004/** Storing the bond structure of a molecule to file.
2005 * Simply stores Atom::nr and then the Atom::nr of all bond partners per line.
2006 * \param *out output stream for debugging
2007 * \param *path path to file
2008 * \return true - file written successfully, false - writing failed
2009 */
2010bool molecule::StoreAdjacencyToFile(ofstream *out, char *path)
2011{
2012 ofstream AdjacencyFile;
2013 atom *Walker = NULL;
2014 stringstream line;
2015 bool status = true;
2016
2017 line << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
2018 AdjacencyFile.open(line.str().c_str(), ios::out);
2019 *out << Verbose(1) << "Saving adjacency list ... ";
2020 if (AdjacencyFile != NULL) {
2021 Walker = start;
2022 while(Walker->next != end) {
2023 Walker = Walker->next;
2024 AdjacencyFile << Walker->nr << "\t";
2025 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
2026 AdjacencyFile << ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker)->nr << "\t";
2027 AdjacencyFile << endl;
2028 }
2029 AdjacencyFile.close();
2030 *out << Verbose(1) << "done." << endl;
2031 } else {
2032 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
2033 status = false;
2034 }
2035
2036 return status;
2037};
2038
2039/** Checks contents of adjacency file against bond structure in structure molecule.
2040 * \param *out output stream for debugging
2041 * \param *path path to file
2042 * \param **ListOfAtoms allocated (molecule::AtomCount) and filled lookup table for ids (Atom::nr) to *Atom
2043 * \return true - structure is equal, false - not equivalence
2044 */
2045bool molecule::CheckAdjacencyFileAgainstMolecule(ofstream *out, char *path, atom **ListOfAtoms)
2046{
2047 ifstream File;
2048 stringstream line;
2049 bool status = true;
2050 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
2051
2052 line << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
2053 File.open(line.str().c_str(), ios::out);
2054 *out << Verbose(1) << "Looking at bond structure stored in adjacency file and comparing to present one ... ";
2055 if (File != NULL) {
2056 // allocate storage structure
2057 int NonMatchNumber = 0; // will number of atoms with differing bond structure
2058 int *CurrentBonds = (int *) Malloc(sizeof(int)*8, "molecule::CheckAdjacencyFileAgainstMolecule - CurrentBonds"); // contains parsed bonds of current atom
2059 int CurrentBondsOfAtom;
2060
2061 // Parse the file line by line and count the bonds
2062 while (!File.eof()) {
2063 File.getline(buffer, MAXSTRINGSIZE);
2064 stringstream line;
2065 line.str(buffer);
2066 int AtomNr = -1;
2067 line >> AtomNr;
2068 CurrentBondsOfAtom = -1; // we count one too far due to line end
2069 // parse into structure
2070 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
2071 while (!line.eof())
2072 line >> CurrentBonds[ ++CurrentBondsOfAtom ];
2073 // compare against present bonds
2074 //cout << Verbose(2) << "Walker is " << *Walker << ", bond partners: ";
2075 if (CurrentBondsOfAtom == NumberOfBondsPerAtom[AtomNr]) {
2076 for(int i=0;i<NumberOfBondsPerAtom[AtomNr];i++) {
2077 int id = ListOfBondsPerAtom[AtomNr][i]->GetOtherAtom(ListOfAtoms[AtomNr])->nr;
2078 int j = 0;
2079 for (;(j<CurrentBondsOfAtom) && (CurrentBonds[j++] != id);); // check against all parsed bonds
2080 if (CurrentBonds[j-1] != id) { // no match ? Then mark in ListOfAtoms
2081 ListOfAtoms[AtomNr] = NULL;
2082 NonMatchNumber++;
2083 status = false;
2084 //out << "[" << id << "]\t";
2085 } else {
2086 //out << id << "\t";
2087 }
2088 }
2089 //out << endl;
2090 } else {
2091 *out << "Number of bonds for Atom " << *ListOfAtoms[AtomNr] << " does not match, parsed " << CurrentBondsOfAtom << " against " << NumberOfBondsPerAtom[AtomNr] << "." << endl;
2092 status = false;
2093 }
2094 }
2095 }
2096 File.close();
2097 File.clear();
2098 if (status) { // if equal we parse the KeySetFile
2099 *out << Verbose(1) << "done: Equal." << endl;
2100 status = true;
2101 } else
2102 *out << Verbose(1) << "done: Not equal by " << NonMatchNumber << " atoms." << endl;
2103 Free((void **)&CurrentBonds, "molecule::CheckAdjacencyFileAgainstMolecule - **CurrentBonds");
2104 } else {
2105 *out << Verbose(1) << "Adjacency file not found." << endl;
2106 status = false;
2107 }
2108 *out << endl;
2109 Free((void **)&buffer, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
2110
2111 return status;
2112};
2113
2114/** Performs a many-body bond order analysis for a given bond order.
2115 * -# parses adjacency, keysets and orderatsite files
2116 * -# performs DFS to find connected subgraphs (to leave this in was a design decision: might be useful later)
2117 * -# RootStack is created for every subgraph (here, later we implement the "update 10 sites with highest energ
2118y contribution", and that's why this consciously not done in the following loop)
2119 * -# in a loop over all subgraphs
2120 * -# calls FragmentBOSSANOVA with this RootStack and within the subgraph molecule structure
2121 * -# creates molecule (fragment)s from the returned keysets (StoreFragmentFromKeySet)
2122 * -# combines the generated molecule lists from all subgraphs
2123 * -# saves to disk: fragment configs, adjacency, orderatsite, keyset files
2124 * Note that as we split "this" molecule up into a list of subgraphs, i.e. a MoleculeListClass, we have two sets
2125 * of vertex indices: Global always means the index in "this" molecule, whereas local refers to the molecule or
2126 * subgraph in the MoleculeListClass.
2127 * \param *out output stream for debugging
2128 * \param Order up to how many neighbouring bonds a fragment contains in BondOrderScheme::BottumUp scheme
2129 * \param *configuration configuration for writing config files for each fragment
2130 */
2131void molecule::FragmentMolecule(ofstream *out, int Order, config *configuration)
2132{
2133 MoleculeListClass *BondFragments = NULL;
2134 atom *Walker = NULL;
2135 atom *OtherWalker = NULL;
2136 bond *Binder = NULL;
2137 int *SortIndex = NULL;
2138 element *runner = NULL;
2139 int AtomNo;
2140 int MinimumRingSize;
2141 int FragmentCounter;
2142 MoleculeLeafClass *MolecularWalker = NULL;
2143 MoleculeLeafClass *Subgraphs = NULL; // list of subgraphs from DFS analysis
2144 fstream File;
2145 bool FragmentationToDo = true;
2146 Graph **FragmentList = NULL;
2147 Graph *TempFragmentList = NULL;
2148 Graph TotalGraph; // graph with all keysets however local numbers
2149 KeySet *TempSet = NULL;
2150 int TotalNumberOfKeySets = 0;
2151 int KeySetCounter = 0;
2152 atom ***ListOfLocalAtoms = NULL;
2153
2154 *out << endl;
2155#ifdef ADDHYDROGEN
2156 *out << Verbose(0) << "I will treat hydrogen special and saturate dangling bonds with it." << endl;
2157#else
2158 *out << Verbose(0) << "Hydrogen is treated just like the rest of the lot." << endl;
2159#endif
2160
2161 // ===== 1. Check whether bond structure is same as stored in files ====
2162
2163 // fill the adjacency list
2164 CreateListOfBondsPerAtom(out);
2165
2166 // create lookup table for Atom::nr
2167 atom **ListOfAtoms = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::FragmentMolecule - **ListOfAtoms");
2168 Walker = start;
2169 while (Walker->next != end) { // create a lookup table (Atom::nr -> atom) used as a marker table lateron
2170 Walker = Walker->next;
2171 if ((Walker->nr >= 0) && (Walker->nr < AtomCount)) {
2172 ListOfAtoms[Walker->nr] = Walker;
2173 } else
2174 break;
2175 }
2176 if (Walker->next != end) { // everything went alright
2177 *out << " range of nuclear ids exceeded [0, AtomCount)." << endl;
2178 FragmentationToDo = false;
2179 }
2180 // === compare it with adjacency file ===
2181 FragmentationToDo = FragmentationToDo && CheckAdjacencyFileAgainstMolecule(out, configuration->configpath, ListOfAtoms);
2182 Free((void **)&ListOfAtoms, "molecule::FragmentMolecule - **ListOfAtoms");
2183
2184 // ===== 2. perform a DFS analysis to gather info on cyclic structure and a list of disconnected subgraphs =====
2185 Subgraphs = DepthFirstSearchAnalysis((ofstream *)&*out, false, MinimumRingSize);
2186
2187 // fill index lookup list for each subgraph from global nr to local pointer
2188 ListOfLocalAtoms = (atom ***) Malloc(sizeof(atom **)*FragmentCounter, "molecule::FragmentMolecule - **ListOfLocalAtoms");
2189 for (int i=0;i<FragmentCounter;i++) {
2190 ListOfLocalAtoms[i] = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::FragmentMolecule - *ListOfLocalAtoms[]");
2191 for(int j=0;j<AtomCount;j++)
2192 ListOfLocalAtoms[i][j] = NULL;
2193 }
2194 FragmentCounter = 0;
2195 MolecularWalker = Subgraphs;
2196 while (MolecularWalker->next != NULL) {
2197 MolecularWalker = MolecularWalker->next;
2198 Walker = MolecularWalker->Leaf->start;
2199 while (Walker->next != MolecularWalker->Leaf->end) {
2200 Walker = Walker->next;
2201#ifdef ADDHYDROGEN
2202 if (Walker->type->Z != 1) // skip hydrogen
2203#endif
2204 ListOfLocalAtoms[FragmentCounter][Walker->GetTrueFather()->nr] = Walker; // set present global id index to the local pointer
2205 }
2206 FragmentCounter++;
2207 }
2208
2209 // fill the bond structure of the individually stored subgraphs
2210 FragmentCounter = 0;
2211 MolecularWalker = Subgraphs;
2212 while (MolecularWalker->next != NULL) {
2213 MolecularWalker = MolecularWalker->next;
2214 *out << Verbose(1) << "Creating adjacency list for subgraph " << MolecularWalker << "." << endl;
2215 Walker = MolecularWalker->Leaf->start;
2216 while (Walker->next != MolecularWalker->Leaf->end) {
2217 Walker = Walker->next;
2218 AtomNo = Walker->father->nr; // global id of the current walker
2219 for(int i=0;i<NumberOfBondsPerAtom[AtomNo];i++) { // go through father's bonds and copy them all
2220 Binder = ListOfBondsPerAtom[AtomNo][i];
2221 OtherWalker = ListOfLocalAtoms[FragmentCounter][Binder->GetOtherAtom(Walker->father)->nr]; // local copy of current bond partner of walker
2222 if ((OtherWalker != NULL) && (OtherWalker->nr > Walker->nr))
2223 MolecularWalker->Leaf->AddBond(Walker, OtherWalker, Binder->BondDegree);
2224 }
2225 }
2226 //MolecularWalker->Leaf->CreateAdjacencyList(out, BondDistance);
2227 MolecularWalker->Leaf->CreateListOfBondsPerAtom(out);
2228 FragmentCounter++;
2229 }
2230
2231 // ===== 3. if structure still valid, parse key set file and others =====
2232 if (FragmentationToDo) {
2233 // parse key sets into new graph
2234 TempFragmentList = new Graph;
2235 FragmentationToDo = FragmentationToDo && ParseKeySetFile(out, configuration->configpath, TempFragmentList, configuration->GetIsAngstroem());
2236 } else
2237 *out << Verbose(1) << "Creation of lookup table Atom::nr <-> Atom failed!" << endl;
2238 if (FragmentationToDo) // parse the adaptive order per atom/site/vertex
2239 FragmentationToDo = FragmentationToDo && ParseOrderAtSiteFromFile(out, configuration->configpath);
2240 else
2241 *out << Verbose(1) << "Parsing keyset file failed" << endl;
2242
2243 // ===== 4. check globally whether there's something to do actually (first adaptivity check)
2244 if (FragmentationToDo) { // check whether OrderAtSite is above Order everywhere
2245 FragmentationToDo = false;
2246 Walker = start;
2247 while (Walker->next != end) {
2248 Walker = Walker->next;
2249#ifdef ADDHYDROGEN
2250 if (Walker->type->Z != 1) // skip hydrogen
2251#endif
2252 if (Walker->AdaptiveOrder < Order)
2253 FragmentationToDo = true;
2254 }
2255 } else
2256 *out << Verbose(1) << "Parsing order at site file failed" << endl;
2257
2258 if (!FragmentationToDo)
2259 *out << Verbose(0) << "Order at every site is already equal or above desired order " << Order << "." << endl;
2260
2261 // =================================== Begin of FRAGMENTATION ===============================
2262 // check cyclic lengths
2263 if ((MinimumRingSize != -1) && (Order >= MinimumRingSize)) {
2264 *out << Verbose(0) << "Bond order " << Order << " greater than or equal to Minimum Ring size of " << MinimumRingSize << " found is not allowed." << endl;
2265 } else {
2266 FragmentCounter = 0;
2267 MolecularWalker = Subgraphs;
2268 // count subgraphs and allocate fragments
2269 while (MolecularWalker->next != NULL) {
2270 MolecularWalker = MolecularWalker->next;
2271 FragmentCounter++;
2272 }
2273 FragmentList = (Graph **) Malloc(sizeof(Graph *)*FragmentCounter, "molecule::FragmentMolecule - **BondFragments");
2274
2275 // ===== 6a. fill RootStack for each subgraph (second adaptivity check) =====
2276 // NOTE: (keep this extern of following while loop, as lateron we may here look for which site to add to which subgraph)
2277 // fill the root stack
2278 KeyStack RootStack[FragmentCounter];
2279
2280 if (Order == -1) { // means we want to increase order adaptively
2281 cerr << "Adaptive chosing of sites not yet implemented!" << endl;
2282 } else { // means we just want to increase it at every site by one
2283 FragmentCounter = 0;
2284 MolecularWalker = Subgraphs;
2285 // count subgraphs and allocate fragments
2286 while (MolecularWalker->next != NULL) {
2287 MolecularWalker = MolecularWalker->next;
2288 // find first root candidates
2289 RootStack[FragmentCounter].clear();
2290 Walker = MolecularWalker->Leaf->start;
2291 while (Walker->next != MolecularWalker->Leaf->end) { // go through all (non-hydrogen) atoms
2292 Walker = Walker->next;
2293 #ifdef ADDHYDROGEN
2294 if (Walker->type->Z != 1) // skip hydrogen
2295 #endif
2296 if (Walker->GetTrueFather()->AdaptiveOrder < Order) // only if Order is still greater
2297 RootStack[FragmentCounter].push_front(Walker->nr);
2298 }
2299 FragmentCounter++;
2300 }
2301 }
2302
2303 // ===== 6b. assign each keyset to its respective subgraph =====
2304 if ((TempFragmentList != NULL) && (TempFragmentList->size() != 0)) { // if there are some scanned keysets at all
2305 // spread the keysets
2306 FragmentCounter = 0;
2307 MolecularWalker = Subgraphs;
2308 while (MolecularWalker->next != NULL) {
2309 MolecularWalker = MolecularWalker->next;
2310 // assign scanned keysets
2311 FragmentList[FragmentCounter] = new Graph;
2312 TempSet = new KeySet;
2313 KeySetCounter = 0;
2314 for(Graph::iterator runner = TempFragmentList->begin();runner != TempFragmentList->end(); runner++) { // key sets contain global numbers!
2315 if ( ListOfLocalAtoms[FragmentCounter][FindAtom(*((*runner).first.begin()))->nr]->nr != -1) {// as we may assume that that bond structure is unchanged, we only test the first key in each set
2316 // translate keyset to local numbers
2317 for(KeySet::iterator sprinter = (*runner).first.begin(); sprinter != (*runner).first.end(); sprinter++)
2318 TempSet->insert(ListOfLocalAtoms[FragmentCounter][FindAtom(*sprinter)->nr]->nr);
2319 // insert into FragmentList
2320 FragmentList[FragmentCounter]->insert(GraphPair (*TempSet, pair<int,double>(KeySetCounter++, (*runner).second.second)));
2321 }
2322 TempSet->clear();
2323 }
2324 delete(TempSet);
2325 if (KeySetCounter == 0)
2326 delete(FragmentList[FragmentCounter]);
2327 else
2328 *out << Verbose(1) << KeySetCounter << " atoms were put into subgraph " << FragmentCounter << "." << endl;
2329 FragmentCounter++;
2330 }
2331 } else // otherwise make sure all lists are initialised to NULL
2332 for(int i=0;i<FragmentCounter;i++)
2333 FragmentList[i] = NULL;
2334
2335
2336 // ===== 7. fill the bond fragment list =====
2337 FragmentCounter = 0;
2338 TotalNumberOfKeySets = 0;
2339 MolecularWalker = Subgraphs;
2340 while (MolecularWalker->next != NULL) {
2341 MolecularWalker = MolecularWalker->next;
2342 *out << Verbose(1) << "Fragmenting subgraph " << MolecularWalker << "." << endl;
2343 if (MolecularWalker->Leaf->first->next != MolecularWalker->Leaf->last) {
2344 // output ListOfBondsPerAtom for debugging
2345 *out << Verbose(0) << endl << "From Contents of ListOfBondsPerAtom, all non-hydrogen atoms:" << endl;
2346 Walker = MolecularWalker->Leaf->start;
2347 while (Walker->next != MolecularWalker->Leaf->end) {
2348 Walker = Walker->next;
2349#ifdef ADDHYDROGEN
2350 if (Walker->type->Z != 1) { // regard only non-hydrogen
2351#endif
2352 *out << Verbose(0) << "Atom " << Walker->Name << " has Bonds: "<<endl;
2353 for(int j=0;j<MolecularWalker->Leaf->NumberOfBondsPerAtom[Walker->nr];j++) {
2354 *out << Verbose(1) << *(MolecularWalker->Leaf->ListOfBondsPerAtom)[Walker->nr][j] << endl;
2355 }
2356#ifdef ADDHYDROGEN
2357 }
2358#endif
2359 }
2360 *out << endl;
2361
2362 *out << Verbose(0) << endl << " ========== BOND ENERGY of subgraph " << FragmentCounter << " ========================= " << endl;
2363 *out << Verbose(0) << "Begin of bond fragmentation." << endl;
2364
2365 // call BOSSANOVA method
2366 MolecularWalker->Leaf->FragmentBOSSANOVA(out, FragmentList[FragmentCounter], RootStack[FragmentCounter]);
2367
2368 // translate list into global numbers (i.e. valid in "this" molecule, not in MolecularWalker->Leaf)
2369 KeySet *TempSet = new KeySet;
2370 for(Graph::iterator runner = FragmentList[FragmentCounter]->begin(); runner != FragmentList[FragmentCounter]->end(); runner++) {
2371 for(KeySet::iterator sprinter = (*runner).first.begin(); sprinter != (*runner).first.end(); sprinter++)
2372 TempSet->insert((MolecularWalker->Leaf->FindAtom(*sprinter))->GetTrueFather()->nr);
2373 TotalGraph.insert(GraphPair(*TempSet, pair<int,double>(TotalNumberOfKeySets++, (*runner).second.second)));
2374 TempSet->clear();
2375 }
2376 delete(TempSet);
2377 delete(FragmentList[FragmentCounter]);
2378 } else {
2379 cerr << "Subgraph " << MolecularWalker << " has no atoms!" << endl;
2380 }
2381 FragmentCounter++; // next fragment list
2382 }
2383 }
2384 Free((void **)&FragmentList, "molecule::FragmentMolecule - **FragmentList");
2385
2386 // ===== 8. gather keyset lists (graphs) from all subgraphs and transform into MoleculeListClass =====
2387 // allocate memory for the pointer array and transmorph graphs into full molecular fragments
2388 BondFragments = new MoleculeListClass(TotalGraph.size(), AtomCount);
2389 int k=0;
2390 for(Graph::iterator runner = TotalGraph.begin(); runner != TotalGraph.end(); runner++) {
2391 KeySet test = (*runner).first;
2392 *out << "Fragment No." << (*runner).second.first << " with TEFactor " << (*runner).second.second << "." << endl;
2393 BondFragments->ListOfMolecules[k] = StoreFragmentFromKeySet(out, test, configuration);
2394 k++;
2395 }
2396 *out << k << "/" << BondFragments->NumberOfMolecules << " fragments generated from the keysets." << endl;
2397
2398 // ==================================== End of FRAGMENTATION ================================
2399
2400 // ===== 10. Save fragments' configuration and keyset files et al to disk ===
2401 if (BondFragments->NumberOfMolecules != 0) {
2402 // create a SortIndex to map from BFS labels to the sequence in which the atoms are given in the config file
2403 SortIndex = (int *) Malloc(sizeof(int)*AtomCount, "molecule::FragmentMolecule: *SortIndex");
2404 for(int i=0;i<AtomCount;i++)
2405 SortIndex[i] = -1;
2406 runner = elemente->start;
2407 AtomNo = 0;
2408 while (runner->next != elemente->end) { // go through every element
2409 runner = runner->next;
2410 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
2411 Walker = start;
2412 while (Walker->next != end) { // go through every atom of this element
2413 Walker = Walker->next;
2414 if (Walker->type->Z == runner->Z) // if this atom fits to element
2415 SortIndex[Walker->nr] = AtomNo++;
2416 }
2417 }
2418 }
2419 *out << Verbose(1) << "Writing " << BondFragments->NumberOfMolecules << " possible bond fragmentation configs" << endl;
2420 if (BondFragments->OutputConfigForListOfFragments(out, configuration, SortIndex))
2421 *out << Verbose(1) << "All configs written." << endl;
2422 else
2423 *out << Verbose(1) << "Some config writing failed." << endl;
2424
2425 // store force index reference file
2426 BondFragments->StoreForcesFile(out, configuration->configpath, SortIndex);
2427
2428 // store keysets file
2429 StoreKeySetFile(out, TotalGraph, configuration->configpath);
2430
2431 // store Adjacency file
2432 StoreAdjacencyToFile(out, configuration->configpath);
2433
2434 // store adaptive orders into file
2435 StoreOrderAtSiteFile(out, configuration->configpath);
2436
2437 // restore orbital and Stop values
2438 CalculateOrbitals(*configuration);
2439
2440 // free memory for bond part
2441 *out << Verbose(1) << "Freeing bond memory" << endl;
2442 delete(FragmentList); // remove bond molecule from memory
2443 FragmentList = NULL;
2444 Free((void **)&SortIndex, "molecule::FragmentMolecule: *SortIndex");
2445 } else
2446 *out << Verbose(1) << "FragmentList is zero on return, splitting failed." << endl;
2447
2448 // free the index lookup list
2449 for (int i=0;i<FragmentCounter;i++)
2450 Free((void **)&ListOfLocalAtoms[i], "molecule::FragmentMolecule - *ListOfLocalAtoms[]");
2451 Free((void **)&ListOfLocalAtoms, "molecule::FragmentMolecule - **ListOfLocalAtoms");
2452 // free subgraph memory again
2453 delete(TempFragmentList);
2454 if (Subgraphs != NULL) {
2455 while (Subgraphs->next != NULL) {
2456 Subgraphs = Subgraphs->next;
2457 delete(Subgraphs->previous);
2458 }
2459 delete(Subgraphs);
2460 }
2461
2462 *out << Verbose(0) << "End of bond fragmentation." << endl;
2463};
2464
2465/** Stores pairs (Atom::nr, Atom::AdaptiveOrder) into file.
2466 * Atoms not present in the file get "-1".
2467 * \param *out output stream for debugging
2468 * \param *path path to file ORDERATSITEFILE
2469 * \return true - file writable, false - not writable
2470 */
2471bool molecule::StoreOrderAtSiteFile(ofstream *out, char *path)
2472{
2473 stringstream line;
2474 ofstream file;
2475
2476 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
2477 file.open(line.str().c_str());
2478 *out << Verbose(1) << "Writing OrderAtSite " << ORDERATSITEFILE << " ... " << endl;
2479 if (file != NULL) {
2480 atom *Walker = start;
2481 while (Walker->next != end) {
2482 Walker = Walker->next;
2483 file << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << endl;
2484 *out << Verbose(2) << "Storing: " << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "." << endl;
2485 }
2486 file.close();
2487 *out << Verbose(1) << "done." << endl;
2488 return true;
2489 } else {
2490 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
2491 return false;
2492 }
2493};
2494
2495/** Parses pairs(Atom::nr, Atom::AdaptiveOrder) from file and stores in molecule's Atom's.
2496 * Atoms not present in the file get "0".
2497 * \param *out output stream for debugging
2498 * \param *path path to file ORDERATSITEFILEe
2499 * \return true - file found and scanned, false - file not found
2500 * \sa ParseKeySetFile() and CheckAdjacencyFileAgainstMolecule() as this is meant to be used in conjunction with the two
2501 */
2502bool molecule::ParseOrderAtSiteFromFile(ofstream *out, char *path)
2503{
2504 unsigned char *OrderArray = (unsigned char *) Malloc(sizeof(unsigned char)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
2505 bool status;
2506 int AtomNr;
2507 stringstream line;
2508 ifstream file;
2509 int Order;
2510
2511 *out << Verbose(1) << "Begin of ParseOrderAtSiteFromFile" << endl;
2512 for(int i=0;i<AtomCount;i++)
2513 OrderArray[i] = 0;
2514 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
2515 file.open(line.str().c_str());
2516 if (file != NULL) {
2517 for (int i=0;i<AtomCount;i++) // initialise with 0
2518 OrderArray[i] = 0;
2519 while (!file.eof()) { // parse from file
2520 file >> AtomNr;
2521 file >> Order;
2522 OrderArray[AtomNr] = (unsigned char) Order;
2523 //*out << Verbose(2) << "AtomNr " << AtomNr << " with order " << (int)OrderArray[AtomNr] << "." << endl;
2524 }
2525 atom *Walker = start;
2526 while (Walker->next != end) { // fill into atom classes
2527 Walker = Walker->next;
2528 Walker->AdaptiveOrder = OrderArray[Walker->nr];
2529 *out << Verbose(2) << *Walker << " gets order " << (int)Walker->AdaptiveOrder << "." << endl;
2530 }
2531 file.close();
2532 *out << Verbose(1) << "done." << endl;
2533 status = true;
2534 } else {
2535 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
2536 status = false;
2537 }
2538 Free((void **)&OrderArray, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
2539
2540 *out << Verbose(1) << "End of ParseOrderAtSiteFromFile" << endl;
2541 return status;
2542};
2543
2544/** Creates an 2d array of pointer with an entry for each atom and each bond it has.
2545 * Updates molecule::ListOfBondsPerAtom, molecule::NumberOfBondsPerAtom by parsing through
2546 * bond chain list, using molecule::AtomCount and molecule::BondCount.
2547 * Allocates memory, fills the array and exits
2548 * \param *out output stream for debugging
2549 */
2550void molecule::CreateListOfBondsPerAtom(ofstream *out)
2551{
2552 bond *Binder = NULL;
2553 atom *Walker = NULL;
2554 int TotalDegree;
2555 *out << Verbose(1) << "Begin of Creating ListOfBondsPerAtom: AtomCount = " << AtomCount << "\tBondCount = " << BondCount << "\tNoNonBonds = " << NoNonBonds << "." << endl;
2556
2557 // re-allocate memory
2558 *out << Verbose(2) << "(Re-)Allocating memory." << endl;
2559 if (ListOfBondsPerAtom != NULL) {
2560 for(int i=0;i<AtomCount;i++)
2561 Free((void **)&ListOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom[i]");
2562 Free((void **)&ListOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom");
2563 }
2564 if (NumberOfBondsPerAtom != NULL)
2565 Free((void **)&NumberOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: NumberOfBondsPerAtom");
2566 ListOfBondsPerAtom = (bond ***) Malloc(sizeof(bond **)*AtomCount, "molecule::CreateListOfBondsPerAtom: ***ListOfBondsPerAtom");
2567 NumberOfBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfBondsPerAtom: *NumberOfBondsPerAtom");
2568
2569 // reset bond counts per atom
2570 for(int i=0;i<AtomCount;i++)
2571 NumberOfBondsPerAtom[i] = 0;
2572 // count bonds per atom
2573 Binder = first;
2574 while (Binder->next != last) {
2575 Binder = Binder->next;
2576 NumberOfBondsPerAtom[Binder->leftatom->nr]++;
2577 NumberOfBondsPerAtom[Binder->rightatom->nr]++;
2578 }
2579 // allocate list of bonds per atom
2580 for(int i=0;i<AtomCount;i++)
2581 ListOfBondsPerAtom[i] = (bond **) Malloc(sizeof(bond *)*NumberOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: **ListOfBondsPerAtom[]");
2582 // clear the list again, now each NumberOfBondsPerAtom marks current free field
2583 for(int i=0;i<AtomCount;i++)
2584 NumberOfBondsPerAtom[i] = 0;
2585 // fill the list
2586 Binder = first;
2587 while (Binder->next != last) {
2588 Binder = Binder->next;
2589 ListOfBondsPerAtom[Binder->leftatom->nr][NumberOfBondsPerAtom[Binder->leftatom->nr]++] = Binder;
2590 ListOfBondsPerAtom[Binder->rightatom->nr][NumberOfBondsPerAtom[Binder->rightatom->nr]++] = Binder;
2591 }
2592
2593 // output list for debugging
2594 *out << Verbose(3) << "ListOfBondsPerAtom for each atom:" << endl;
2595 Walker = start;
2596 while (Walker->next != end) {
2597 Walker = Walker->next;
2598 *out << Verbose(4) << "Atom " << Walker->Name << " with " << NumberOfBondsPerAtom[Walker->nr] << " bonds: ";
2599 TotalDegree = 0;
2600 for (int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
2601 *out << *ListOfBondsPerAtom[Walker->nr][j] << "\t";
2602 TotalDegree += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
2603 }
2604 *out << " -- TotalDegree: " << TotalDegree << endl;
2605 }
2606 *out << Verbose(1) << "End of Creating ListOfBondsPerAtom." << endl << endl;
2607};
2608
2609/** Adds atoms up to \a BondCount distance from \a *Root and notes them down in \a **AddedAtomList.
2610 * Gray vertices are always enqueued in an StackClass<atom *> FIFO queue, the rest is usual BFS with adding vertices found was
2611 * white and putting into queue.
2612 * \param *out output stream for debugging
2613 * \param *Mol Molecule class to add atoms to
2614 * \param **AddedAtomList list with added atom pointers, index is atom father's number
2615 * \param **AddedBondList list with added bond pointers, index is bond father's number
2616 * \param *Root root vertex for BFS
2617 * \param *Bond bond not to look beyond
2618 * \param BondOrder maximum distance for vertices to add
2619 * \param IsAngstroem lengths are in angstroem or bohrradii
2620 */
2621void molecule::BreadthFirstSearchAdd(ofstream *out, molecule *Mol, atom **&AddedAtomList, bond **&AddedBondList, atom *Root, bond *Bond, int BondOrder, bool IsAngstroem)
2622{
2623 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::BreadthFirstSearchAdd: **PredecessorList");
2624 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::BreadthFirstSearchAdd: *ShortestPathList");
2625 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::BreadthFirstSearchAdd: *ColorList");
2626 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
2627 atom *Walker = NULL, *OtherAtom = NULL;
2628 bond *Binder = NULL;
2629
2630 // add Root if not done yet
2631 AtomStack->ClearStack();
2632 if (AddedAtomList[Root->nr] == NULL) // add Root if not yet present
2633 AddedAtomList[Root->nr] = Mol->AddCopyAtom(Root);
2634 AtomStack->Push(Root);
2635
2636 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
2637 for (int i=0;i<AtomCount;i++) {
2638 PredecessorList[i] = NULL;
2639 ShortestPathList[i] = -1;
2640 if (AddedAtomList[i] != NULL) // mark already present atoms (i.e. Root and maybe others) as visited
2641 ColorList[i] = lightgray;
2642 else
2643 ColorList[i] = white;
2644 }
2645 ShortestPathList[Root->nr] = 0;
2646
2647 // and go on ... Queue always contains all lightgray vertices
2648 while (!AtomStack->IsEmpty()) {
2649 // we have to pop the oldest atom from stack. This keeps the atoms on the stack always of the same ShortestPath distance.
2650 // 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
2651 // append length of 3 (their neighbours). Thus on stack we have always atoms of a certain length n at bottom of stack and
2652 // followed by n+1 till top of stack.
2653 Walker = AtomStack->PopFirst(); // pop oldest added
2654 *out << Verbose(1) << "Current Walker is: " << Walker->Name << ", and has " << NumberOfBondsPerAtom[Walker->nr] << " bonds." << endl;
2655 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2656 Binder = ListOfBondsPerAtom[Walker->nr][i];
2657 if (Binder != NULL) { // don't look at bond equal NULL
2658 OtherAtom = Binder->GetOtherAtom(Walker);
2659 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2660 if (ColorList[OtherAtom->nr] == white) {
2661 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)
2662 ColorList[OtherAtom->nr] = lightgray;
2663 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2664 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2665 *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;
2666 if ((((ShortestPathList[OtherAtom->nr] < BondOrder) && (Binder != Bond))) ) { // Check for maximum distance
2667 *out << Verbose(3);
2668 if (AddedAtomList[OtherAtom->nr] == NULL) { // add if it's not been so far
2669 AddedAtomList[OtherAtom->nr] = Mol->AddCopyAtom(OtherAtom);
2670 *out << "Added OtherAtom " << OtherAtom->Name;
2671 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2672 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2673 AddedBondList[Binder->nr]->Type = Binder->Type;
2674 *out << " and bond " << *(AddedBondList[Binder->nr]) << ", ";
2675 } 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)
2676 *out << "Not adding OtherAtom " << OtherAtom->Name;
2677 if (AddedBondList[Binder->nr] == NULL) {
2678 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2679 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2680 AddedBondList[Binder->nr]->Type = Binder->Type;
2681 *out << ", added Bond " << *(AddedBondList[Binder->nr]);
2682 } else
2683 *out << ", not added Bond ";
2684 }
2685 *out << ", putting OtherAtom into queue." << endl;
2686 AtomStack->Push(OtherAtom);
2687 } else { // out of bond order, then replace
2688 if ((AddedAtomList[OtherAtom->nr] == NULL) && (Binder->Cyclic))
2689 ColorList[OtherAtom->nr] = white; // unmark if it has not been queued/added, to make it available via its other bonds (cyclic)
2690 if (Binder == Bond)
2691 *out << Verbose(3) << "Not Queueing, is the Root bond";
2692 else if (ShortestPathList[OtherAtom->nr] >= BondOrder)
2693 *out << Verbose(3) << "Not Queueing, is out of Bond Count of " << BondOrder;
2694 if (!Binder->Cyclic)
2695 *out << ", is not part of a cyclic bond, saturating bond with Hydrogen." << endl;
2696 if (AddedBondList[Binder->nr] == NULL) {
2697 if ((AddedAtomList[OtherAtom->nr] != NULL)) { // .. whether we add or saturate
2698 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2699 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2700 AddedBondList[Binder->nr]->Type = Binder->Type;
2701 } else {
2702#ifdef ADDHYDROGEN
2703 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
2704#endif
2705 }
2706 }
2707 }
2708 } else {
2709 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
2710 // This has to be a cyclic bond, check whether it's present ...
2711 if (AddedBondList[Binder->nr] == NULL) {
2712 if ((Binder != Bond) && (Binder->Cyclic) && (((ShortestPathList[Walker->nr]+1) < BondOrder))) {
2713 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2714 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2715 AddedBondList[Binder->nr]->Type = Binder->Type;
2716 } else { // if it's root bond it has to broken (otherwise we would not create the fragments)
2717#ifdef ADDHYDROGEN
2718 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
2719#endif
2720 }
2721 }
2722 }
2723 }
2724 }
2725 ColorList[Walker->nr] = black;
2726 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2727 }
2728 Free((void **)&PredecessorList, "molecule::BreadthFirstSearchAdd: **PredecessorList");
2729 Free((void **)&ShortestPathList, "molecule::BreadthFirstSearchAdd: **ShortestPathList");
2730 Free((void **)&ColorList, "molecule::BreadthFirstSearchAdd: **ColorList");
2731 delete(AtomStack);
2732};
2733
2734/** Adds bond structure to this molecule from \a Father molecule.
2735 * This basically causes this molecule to become an induced subgraph of the \a Father, i.e. for every bond in Father
2736 * with end points present in this molecule, bond is created in this molecule.
2737 * Special care was taken to ensure that this is of complexity O(N), where N is the \a Father's molecule::AtomCount.
2738 * \param *out output stream for debugging
2739 * \param *Father father molecule
2740 * \return true - is induced subgraph, false - there are atoms with fathers not in \a Father
2741 * \todo not checked, not fully working probably
2742 */
2743bool molecule::BuildInducedSubgraph(ofstream *out, const molecule *Father)
2744{
2745 atom *Walker = NULL, *OtherAtom = NULL;
2746 bool status = true;
2747 atom **ParentList = (atom **) Malloc(sizeof(atom *)*Father->AtomCount, "molecule::BuildInducedSubgraph: **ParentList");
2748
2749 *out << Verbose(2) << "Begin of BuildInducedSubgraph." << endl;
2750
2751 // reset parent list
2752 *out << Verbose(3) << "Resetting ParentList." << endl;
2753 for (int i=0;i<Father->AtomCount;i++)
2754 ParentList[i] = NULL;
2755
2756 // fill parent list with sons
2757 *out << Verbose(3) << "Filling Parent List." << endl;
2758 Walker = start;
2759 while (Walker->next != end) {
2760 Walker = Walker->next;
2761 ParentList[Walker->father->nr] = Walker;
2762 // Outputting List for debugging
2763 *out << Verbose(4) << "Son["<< Walker->father->nr <<"] of " << Walker->father << " is " << ParentList[Walker->father->nr] << "." << endl;
2764 }
2765
2766 // check each entry of parent list and if ok (one-to-and-onto matching) create bonds
2767 *out << Verbose(3) << "Creating bonds." << endl;
2768 Walker = Father->start;
2769 while (Walker->next != Father->end) {
2770 Walker = Walker->next;
2771 if (ParentList[Walker->nr] != NULL) {
2772 if (ParentList[Walker->nr]->father != Walker) {
2773 status = false;
2774 } else {
2775 for (int i=0;i<Father->NumberOfBondsPerAtom[Walker->nr];i++) {
2776 OtherAtom = Father->ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
2777 if (ParentList[OtherAtom->nr] != NULL) { // if otheratom is also a father of an atom on this molecule, create the bond
2778 *out << Verbose(4) << "Endpoints of Bond " << Father->ListOfBondsPerAtom[Walker->nr][i] << " are both present: " << ParentList[Walker->nr]->Name << " and " << ParentList[OtherAtom->nr]->Name << "." << endl;
2779 AddBond(ParentList[Walker->nr], ParentList[OtherAtom->nr], Father->ListOfBondsPerAtom[Walker->nr][i]->BondDegree);
2780 }
2781 }
2782 }
2783 }
2784 }
2785
2786 Free((void **)&ParentList, "molecule::BuildInducedSubgraph: **ParentList");
2787 *out << Verbose(2) << "End of BuildInducedSubgraph." << endl;
2788 return status;
2789};
2790
2791
2792/** Looks through a StackClass<atom *> and returns the likeliest removal candiate.
2793 * \param *out output stream for debugging messages
2794 * \param *&Leaf KeySet to look through
2795 * \param *&ShortestPathList list of the shortest path to decide which atom to suggest as removal candidate in the end
2796 * \param index of the atom suggested for removal
2797 */
2798int molecule::LookForRemovalCandidate(ofstream *&out, KeySet *&Leaf, int *&ShortestPathList)
2799{
2800 atom *Runner = NULL;
2801 int SP, Removal;
2802
2803 *out << Verbose(2) << "Looking for removal candidate." << endl;
2804 SP = -1; //0; // not -1, so that Root is never removed
2805 Removal = -1;
2806 for (KeySet::iterator runner = Leaf->begin(); runner != Leaf->end(); runner++) {
2807 Runner = FindAtom((*runner));
2808 if (Runner->type->Z != 1) { // skip all those added hydrogens when re-filling snake stack
2809 if (ShortestPathList[(*runner)] > SP) { // remove the oldest one with longest shortest path
2810 SP = ShortestPathList[(*runner)];
2811 Removal = (*runner);
2812 }
2813 }
2814 }
2815 return Removal;
2816};
2817
2818/** Stores a fragment from \a KeySet into \a molecule.
2819 * First creates the minimal set of atoms from the KeySet, then creates the bond structure from the complete
2820 * molecule and adds missing hydrogen where bonds were cut.
2821 * \param *out output stream for debugging messages
2822 * \param &Leaflet pointer to KeySet structure
2823 * \param IsAngstroem whether we have Ansgtroem or bohrradius
2824 * \return pointer to constructed molecule
2825 */
2826molecule * molecule::StoreFragmentFromKeySet(ofstream *out, KeySet &Leaflet, bool IsAngstroem)
2827{
2828 atom *Runner = NULL, *FatherOfRunner = NULL, *OtherFather = NULL;
2829 atom **SonList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::StoreFragmentFromStack: **SonList");
2830 molecule *Leaf = new molecule(elemente);
2831
2832// *out << Verbose(1) << "Begin of StoreFragmentFromKeyset." << endl;
2833
2834 Leaf->BondDistance = BondDistance;
2835 for(int i=0;i<NDIM*2;i++)
2836 Leaf->cell_size[i] = cell_size[i];
2837
2838 // initialise SonList (indicates when we need to replace a bond with hydrogen instead)
2839 for(int i=0;i<AtomCount;i++)
2840 SonList[i] = NULL;
2841
2842 // first create the minimal set of atoms from the KeySet
2843 for(KeySet::iterator runner = Leaflet.begin(); runner != Leaflet.end(); runner++) {
2844 FatherOfRunner = FindAtom((*runner)); // find the id
2845 SonList[FatherOfRunner->nr] = Leaf->AddCopyAtom(FatherOfRunner);
2846 }
2847
2848 // create the bonds between all: Make it an induced subgraph and add hydrogen
2849// *out << Verbose(2) << "Creating bonds from father graph (i.e. induced subgraph creation)." << endl;
2850 Runner = Leaf->start;
2851 while (Runner->next != Leaf->end) {
2852 Runner = Runner->next;
2853 FatherOfRunner = Runner->father;
2854 if (SonList[FatherOfRunner->nr] != NULL) { // check if this, our father, is present in list
2855 // create all bonds
2856 for (int i=0;i<NumberOfBondsPerAtom[FatherOfRunner->nr];i++) { // go through every bond of father
2857 OtherFather = ListOfBondsPerAtom[FatherOfRunner->nr][i]->GetOtherAtom(FatherOfRunner);
2858// *out << Verbose(2) << "Father " << *FatherOfRunner << " of son " << *SonList[FatherOfRunner->nr] << " is bound to " << *OtherFather;
2859 if (SonList[OtherFather->nr] != NULL) {
2860// *out << ", whose son is " << *SonList[OtherFather->nr] << "." << endl;
2861 if (OtherFather->nr > FatherOfRunner->nr) { // add bond (nr check is for adding only one of both variants: ab, ba)
2862// *out << Verbose(3) << "Adding Bond: " << Leaf->AddBond(Runner, SonList[OtherFather->nr], ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree) << "." << endl;
2863 //NumBonds[Runner->nr]++;
2864 } else {
2865// *out << Verbose(3) << "Not adding bond, labels in wrong order." << endl;
2866 }
2867 } else {
2868// *out << ", who has no son in this fragment molecule." << endl;
2869#ifdef ADDHYDROGEN
2870// *out << Verbose(3) << "Adding Hydrogen to " << Runner->Name << " and a bond in between." << endl;
2871 Leaf->AddHydrogenReplacementAtom(out, ListOfBondsPerAtom[FatherOfRunner->nr][i], Runner, FatherOfRunner, OtherFather, ListOfBondsPerAtom[FatherOfRunner->nr],NumberOfBondsPerAtom[FatherOfRunner->nr], IsAngstroem);
2872#endif
2873 //NumBonds[Runner->nr] += ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree;
2874 }
2875 }
2876 } else {
2877 *out << Verbose(0) << "ERROR: Son " << Runner->Name << " has father " << FatherOfRunner->Name << " but its entry in SonList is " << SonList[FatherOfRunner->nr] << "!" << endl;
2878 }
2879#ifdef ADDHYDROGEN
2880 while ((Runner->next != Leaf->end) && (Runner->next->type->Z == 1)) // skip added hydrogen
2881 Runner = Runner->next;
2882#endif
2883 }
2884 Leaf->CreateListOfBondsPerAtom(out);
2885 //Leaflet->Leaf->ScanForPeriodicCorrection(out);
2886 Free((void **)&SonList, "molecule::StoreFragmentFromStack: **SonList");
2887// *out << Verbose(1) << "End of StoreFragmentFromKeyset." << endl;
2888 return Leaf;
2889};
2890
2891/** Creates \a MoleculeListClass of all unique fragments of the \a molecule containing \a Order atoms or vertices.
2892 * The picture to have in mind is that of a DFS "snake" of a certain length \a Order, i.e. as in the infamous
2893 * computer game, that winds through the connected graph representing the molecule. Color (white,
2894 * lightgray, darkgray, black) indicates whether a vertex has been discovered so far or not. Labels will help in
2895 * creating only unique fragments and not additional ones with vertices simply in different sequence.
2896 * The Predecessor is always the one that came before in discovering, needed on backstepping. And
2897 * finally, the ShortestPath is needed for removing vertices from the snake stack during the back-
2898 * stepping.
2899 * \param *out output stream for debugging
2900 * \param Order number of atoms in each fragment
2901 * \param *configuration configuration for writing config files for each fragment
2902 * \return List of all unique fragments with \a Order atoms
2903 */
2904/*
2905MoleculeListClass * molecule::CreateListOfUniqueFragmentsOfOrder(ofstream *out, int Order, config *configuration)
2906{
2907 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
2908 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
2909 int *Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
2910 enum Shading *ColorVertexList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
2911 enum Shading *ColorEdgeList = (enum Shading *) Malloc(sizeof(enum Shading)*BondCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorBondList");
2912 StackClass<atom *> *RootStack = new StackClass<atom *>(AtomCount);
2913 StackClass<atom *> *TouchedStack = new StackClass<atom *>((int)pow(4,Order)+2); // number of atoms reached from one with maximal 4 bonds plus Root itself
2914 StackClass<atom *> *SnakeStack = new StackClass<atom *>(Order+1); // equal to Order is not possible, as then the StackClass<atom *> cannot discern between full and empty stack!
2915 MoleculeLeafClass *Leaflet = NULL, *TempLeaf = NULL;
2916 MoleculeListClass *FragmentList = NULL;
2917 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Removal = NULL;
2918 bond *Binder = NULL;
2919 int RunningIndex = 0, FragmentCounter = 0;
2920
2921 *out << Verbose(1) << "Begin of CreateListOfUniqueFragmentsOfOrder." << endl;
2922
2923 // reset parent list
2924 *out << Verbose(3) << "Resetting labels, parent, predecessor, color and shortest path lists." << endl;
2925 for (int i=0;i<AtomCount;i++) { // reset all atom labels
2926 // initialise each vertex as white with no predecessor, empty queue, color lightgray, not labelled, no sons
2927 Labels[i] = -1;
2928 SonList[i] = NULL;
2929 PredecessorList[i] = NULL;
2930 ColorVertexList[i] = white;
2931 ShortestPathList[i] = -1;
2932 }
2933 for (int i=0;i<BondCount;i++)
2934 ColorEdgeList[i] = white;
2935 RootStack->ClearStack(); // clearstack and push first atom if exists
2936 TouchedStack->ClearStack();
2937 Walker = start->next;
2938 while ((Walker != end)
2939#ifdef ADDHYDROGEN
2940 && (Walker->type->Z == 1)
2941#endif
2942 ) { // search for first non-hydrogen atom
2943 *out << Verbose(4) << "Current Root candidate is " << Walker->Name << "." << endl;
2944 Walker = Walker->next;
2945 }
2946 if (Walker != end)
2947 RootStack->Push(Walker);
2948 else
2949 *out << Verbose(0) << "ERROR: Could not find an appropriate Root atom!" << endl;
2950 *out << Verbose(3) << "Root " << Walker->Name << " is on AtomStack, beginning loop through all vertices ..." << endl;
2951
2952 ///// OUTER LOOP ////////////
2953 while (!RootStack->IsEmpty()) {
2954 // get new root vertex from atom stack
2955 Root = RootStack->PopFirst();
2956 ShortestPathList[Root->nr] = 0;
2957 if (Labels[Root->nr] == -1)
2958 Labels[Root->nr] = RunningIndex++; // prevent it from getting again on AtomStack
2959 PredecessorList[Root->nr] = Root;
2960 TouchedStack->Push(Root);
2961 *out << Verbose(0) << "Root for this loop is: " << Root->Name << ".\n";
2962
2963 // clear snake stack
2964 SnakeStack->ClearStack();
2965 //SnakeStack->TestImplementation(out, start->next);
2966
2967 ///// INNER LOOP ////////////
2968 // Problems:
2969 // - what about cyclic bonds?
2970 Walker = Root;
2971 do {
2972 *out << Verbose(1) << "Current Walker is: " << Walker->Name;
2973 // initial setting of the new Walker: label, color, shortest path and put on stacks
2974 if (Labels[Walker->nr] == -1) { // give atom a unique, monotonely increasing number
2975 Labels[Walker->nr] = RunningIndex++;
2976 RootStack->Push(Walker);
2977 }
2978 *out << ", has label " << Labels[Walker->nr];
2979 if ((ColorVertexList[Walker->nr] == white) || ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white))) { // color it if newly discovered and push on stacks (and if within reach!)
2980 if ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white)) {
2981 // Binder ought to be set still from last neighbour search
2982 *out << ", coloring bond " << *Binder << " black";
2983 ColorEdgeList[Binder->nr] = black; // mark this bond as used
2984 }
2985 if (ShortestPathList[Walker->nr] == -1) {
2986 ShortestPathList[Walker->nr] = ShortestPathList[PredecessorList[Walker->nr]->nr]+1;
2987 TouchedStack->Push(Walker); // mark every atom for lists cleanup later, whose shortest path has been changed
2988 }
2989 if ((ShortestPathList[Walker->nr] < Order) && (ColorVertexList[Walker->nr] != darkgray)) { // if not already on snake stack
2990 SnakeStack->Push(Walker);
2991 ColorVertexList[Walker->nr] = darkgray; // mark as dark gray of on snake stack
2992 }
2993 }
2994 *out << ", SP of " << ShortestPathList[Walker->nr] << " and its color is " << GetColor(ColorVertexList[Walker->nr]) << "." << endl;
2995
2996 // then check the stack for a newly stumbled upon fragment
2997 if (SnakeStack->ItemCount() == Order) { // is stack full?
2998 // store the fragment if it is one and get a removal candidate
2999 Removal = StoreFragmentFromStack(out, Root, Walker, Leaflet, SnakeStack, ShortestPathList, SonList, Labels, &FragmentCounter, configuration);
3000 // remove the candidate if one was found
3001 if (Removal != NULL) {
3002 *out << Verbose(2) << "Removing item " << Removal->Name << " with SP of " << ShortestPathList[Removal->nr] << " from snake stack." << endl;
3003 SnakeStack->RemoveItem(Removal);
3004 ColorVertexList[Removal->nr] = lightgray; // return back to not on snake stack but explored marking
3005 if (Walker == Removal) { // if the current atom is to be removed, we also have to take a step back
3006 Walker = PredecessorList[Removal->nr];
3007 *out << Verbose(2) << "Stepping back to " << Walker->Name << "." << endl;
3008 }
3009 }
3010 } else
3011 Removal = NULL;
3012
3013 // finally, look for a white neighbour as the next Walker
3014 Binder = NULL;
3015 if ((Removal == NULL) || (Walker != PredecessorList[Removal->nr])) { // don't look, if a new walker has been set above
3016 *out << Verbose(2) << "Snake has currently " << SnakeStack->ItemCount() << " item(s)." << endl;
3017 OtherAtom = NULL; // this is actually not needed, every atom has at least one neighbour
3018 if (ShortestPathList[Walker->nr] < Order) {
3019 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
3020 Binder = ListOfBondsPerAtom[Walker->nr][i];
3021 *out << Verbose(2) << "Current bond is " << *Binder << ": ";
3022 OtherAtom = Binder->GetOtherAtom(Walker);
3023 if ((Labels[OtherAtom->nr] != -1) && (Labels[OtherAtom->nr] < Labels[Root->nr])) { // we don't step up to labels bigger than us
3024 *out << "Label " << Labels[OtherAtom->nr] << " is smaller than Root's " << Labels[Root->nr] << "." << endl;
3025 //ColorVertexList[OtherAtom->nr] = lightgray; // mark as explored
3026 } else { // otherwise check its colour and element
3027 if (
3028#ifdef ADDHYDROGEN
3029 (OtherAtom->type->Z != 1) &&
3030#endif
3031 (ColorEdgeList[Binder->nr] == white)) { // skip hydrogen, look for unexplored vertices
3032 *out << "Moving along " << GetColor(ColorEdgeList[Binder->nr]) << " bond " << Binder << " to " << ((ColorVertexList[OtherAtom->nr] == white) ? "unexplored" : "explored") << " item: " << OtherAtom->Name << "." << endl;
3033 // i find it currently rather sensible to always set the predecessor in order to find one's way back
3034 //if (PredecessorList[OtherAtom->nr] == NULL) {
3035 PredecessorList[OtherAtom->nr] = Walker;
3036 *out << Verbose(3) << "Setting Predecessor of " << OtherAtom->Name << " to " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
3037 //} else {
3038 // *out << Verbose(3) << "Predecessor of " << OtherAtom->Name << " is " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
3039 //}
3040 Walker = OtherAtom;
3041 break;
3042 } else {
3043 if (OtherAtom->type->Z == 1)
3044 *out << "Links to a hydrogen atom." << endl;
3045 else
3046 *out << "Bond has not white but " << GetColor(ColorEdgeList[Binder->nr]) << " color." << endl;
3047 }
3048 }
3049 }
3050 } else { // means we have stepped beyond the horizon: Return!
3051 Walker = PredecessorList[Walker->nr];
3052 OtherAtom = Walker;
3053 *out << Verbose(3) << "We have gone too far, stepping back to " << Walker->Name << "." << endl;
3054 }
3055 if (Walker != OtherAtom) { // if no white neighbours anymore, color it black
3056 *out << Verbose(2) << "Coloring " << Walker->Name << " black." << endl;
3057 ColorVertexList[Walker->nr] = black;
3058 Walker = PredecessorList[Walker->nr];
3059 }
3060 }
3061 } while ((Walker != Root) || (ColorVertexList[Root->nr] != black));
3062 *out << Verbose(2) << "Inner Looping is finished." << endl;
3063
3064 // if we reset all AtomCount atoms, we have again technically O(N^2) ...
3065 *out << Verbose(2) << "Resetting lists." << endl;
3066 Walker = NULL;
3067 Binder = NULL;
3068 while (!TouchedStack->IsEmpty()) {
3069 Walker = TouchedStack->PopLast();
3070 *out << Verbose(3) << "Re-initialising entries of " << *Walker << "." << endl;
3071 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
3072 ColorEdgeList[ListOfBondsPerAtom[Walker->nr][i]->nr] = white;
3073 PredecessorList[Walker->nr] = NULL;
3074 ColorVertexList[Walker->nr] = white;
3075 ShortestPathList[Walker->nr] = -1;
3076 }
3077 }
3078 *out << Verbose(1) << "Outer Looping over all vertices is done." << endl;
3079
3080 // copy together
3081 *out << Verbose(1) << "Copying all fragments into MoleculeList structure." << endl;
3082 FragmentList = new MoleculeListClass(FragmentCounter, AtomCount);
3083 RunningIndex = 0;
3084 while ((Leaflet != NULL) && (RunningIndex < FragmentCounter)) {
3085 FragmentList->ListOfMolecules[RunningIndex++] = Leaflet->Leaf;
3086 Leaflet->Leaf = NULL; // prevent molecule from being removed
3087 TempLeaf = Leaflet;
3088 Leaflet = Leaflet->previous;
3089 delete(TempLeaf);
3090 };
3091
3092 // free memory and exit
3093 Free((void **)&PredecessorList, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
3094 Free((void **)&ShortestPathList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
3095 Free((void **)&Labels, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
3096 Free((void **)&ColorVertexList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
3097 delete(RootStack);
3098 delete(TouchedStack);
3099 delete(SnakeStack);
3100
3101 *out << Verbose(1) << "End of CreateListOfUniqueFragmentsOfOrder." << endl;
3102 return FragmentList;
3103};
3104*/
3105
3106/** Structure containing all values in power set combination generation.
3107 */
3108struct UniqueFragments {
3109 config *configuration;
3110 atom *Root;
3111 Graph *Leaflet;
3112 KeySet *FragmentSet;
3113 int ANOVAOrder;
3114 int FragmentCounter;
3115 int CurrentIndex;
3116 int *Labels;
3117 double TEFactor;
3118 int *ShortestPathList;
3119 bool **UsedList;
3120 bond **BondsPerSPList;
3121 int *BondsPerSPCount;
3122};
3123
3124/** From a given set of Bond sorted by Shortest Path distance, create all possible fragments of size \a SetDimension.
3125 * -# loops over every possible combination (2^dimension of edge set)
3126 * -# inserts current set, if there's still space left
3127 * -# yes: calls SPFragmentGenerator with structure, created new edge list and size respective to root dist
3128ance+1
3129 * -# no: stores fragment into keyset list by calling InsertFragmentIntoGraph
3130 * -# removes all items added into the snake stack (in UniqueFragments structure) added during level (root
3131distance) and current set
3132 * \param *out output stream for debugging
3133 * \param FragmentSearch UniqueFragments structure with all values needed
3134 * \param RootDistance current shortest path level, whose set of edges is represented by **BondsSet
3135 * \param SetDimension Number of possible bonds on this level (i.e. size of the array BondsSet[])
3136 * \param SubOrder remaining number of allowed vertices to add
3137 */
3138void molecule::SPFragmentGenerator(ofstream *out, struct UniqueFragments *FragmentSearch, int RootDistance, bond **BondsSet, int SetDimension, int SubOrder)
3139{
3140 atom *OtherWalker = NULL;
3141 int verbosity = 0; //FragmentSearch->ANOVAOrder-SubOrder;
3142 int NumCombinations;
3143 bool bit;
3144 int bits, TouchedIndex, SubSetDimension, SP;
3145 int Removal;
3146 int *TouchedList = (int *) Malloc(sizeof(int)*(SubOrder+1), "molecule::SPFragmentGenerator: *TouchedList");
3147 bond *Binder = NULL;
3148 bond **BondsList = NULL;
3149
3150 NumCombinations = 1 << SetDimension;
3151
3152 // Hier muessen von 1 bis NumberOfBondsPerAtom[Walker->nr] alle Kombinationen
3153 // von Endstuecken (aus den Bonds) hinzugefÃŒgt werden und fÃŒr verbleibende ANOVAOrder
3154 // rekursiv GraphCrawler in der nÀchsten Ebene aufgerufen werden
3155
3156 *out << Verbose(1+verbosity) << "Begin of SPFragmentGenerator." << endl;
3157 *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;
3158
3159 // initialised touched list (stores added atoms on this level)
3160 *out << Verbose(1+verbosity) << "Clearing touched list." << endl;
3161 for (TouchedIndex=0;TouchedIndex<=SubOrder;TouchedIndex++) // empty touched list
3162 TouchedList[TouchedIndex] = -1;
3163 TouchedIndex = 0;
3164
3165 // create every possible combination of the endpieces
3166 *out << Verbose(1+verbosity) << "Going through all combinations of the power set." << endl;
3167 for (int i=1;i<NumCombinations;i++) { // sweep through all power set combinations (skip empty set!)
3168 // count the set bit of i
3169 bits = 0;
3170 for (int j=0;j<SetDimension;j++)
3171 bits += (i & (1 << j)) >> j;
3172
3173 *out << Verbose(1+verbosity) << "Current set is " << Binary(i | (1 << SetDimension)) << ", number of bits is " << bits << "." << endl;
3174 if (bits <= SubOrder) { // if not greater than additional atoms allowed on stack, continue
3175 // --1-- add this set of the power set of bond partners to the snake stack
3176 for (int j=0;j<SetDimension;j++) { // pull out every bit by shifting
3177 bit = ((i & (1 << j)) != 0); // mask the bit for the j-th bond
3178 if (bit) { // if bit is set, we add this bond partner
3179 OtherWalker = BondsSet[j]->rightatom; // rightatom is always the one more distant, i.e. the one to add
3180 //*out << Verbose(1+verbosity) << "Current Bond is " << ListOfBondsPerAtom[Walker->nr][i] << ", checking on " << *OtherWalker << "." << endl;
3181 //if ((!FragmentSearch->UsedList[OtherWalker->nr][i]) && (FragmentSearch->Labels[OtherWalker->nr] > FragmentSearch->Labels[FragmentSearch->Root->nr])) {
3182 //*out << Verbose(2+verbosity) << "Not used so far, label " << FragmentSearch->Labels[OtherWalker->nr] << " is bigger than Root's " << FragmentSearch->Labels[FragmentSearch->Root->nr] << "." << endl;
3183 *out << Verbose(2+verbosity) << "Adding " << *OtherWalker << " with nr " << OtherWalker->nr << "." << endl;
3184 TouchedList[TouchedIndex++] = OtherWalker->nr; // note as added
3185 FragmentSearch->FragmentSet->insert(OtherWalker->nr);
3186 //FragmentSearch->UsedList[OtherWalker->nr][i] = true;
3187 //}
3188 } else {
3189 *out << Verbose(2+verbosity) << "Not adding." << endl;
3190 }
3191 }
3192
3193 if (bits < SubOrder) {
3194 *out << Verbose(1+verbosity) << "There's still some space left on stack: " << (SubOrder - bits) << "." << endl;
3195 // --2-- look at all added end pieces of this combination, construct bond subsets and sweep through a power set of these by recursion
3196 SP = RootDistance+1; // this is the next level
3197 // first count the members in the subset
3198 SubSetDimension = 0;
3199 Binder = FragmentSearch->BondsPerSPList[2*SP]; // start node for this level
3200 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) { // compare to end node of this level
3201 Binder = Binder->next;
3202 for (int k=0;k<TouchedIndex;k++) {
3203 if (Binder->Contains(TouchedList[k])) // if we added this very endpiece
3204 SubSetDimension++;
3205 }
3206 }
3207 // then allocate and fill the list
3208 BondsList = (bond **) Malloc(sizeof(bond *)*SubSetDimension, "molecule::SPFragmentGenerator: **BondsList");
3209 SubSetDimension = 0;
3210 Binder = FragmentSearch->BondsPerSPList[2*SP];
3211 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) {
3212 Binder = Binder->next;
3213 for (int k=0;k<TouchedIndex;k++) {
3214 if (Binder->leftatom->nr == TouchedList[k]) // leftatom is always the close one
3215 BondsList[SubSetDimension++] = Binder;
3216 }
3217 }
3218 *out << Verbose(2+verbosity) << "Calling subset generator " << SP << " away from root " << *FragmentSearch->Root << " with sub set dimension " << SubSetDimension << "." << endl;
3219 SPFragmentGenerator(out, FragmentSearch, SP, BondsList, SubSetDimension, SubOrder-bits);
3220 Free((void **)&BondsList, "molecule::SPFragmentGenerator: **BondsList");
3221 } else {
3222 // --2-- otherwise store the complete fragment
3223 *out << Verbose(1+verbosity) << "Enough items on stack for a fragment!" << endl;
3224 // store fragment as a KeySet
3225 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch->FragmentCounter << "], local nr.s are: ";
3226 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
3227 *out << (*runner) << " ";
3228 InsertFragmentIntoGraph(out, FragmentSearch);
3229 //Removal = LookForRemovalCandidate(out, FragmentSearch->FragmentSet, FragmentSearch->ShortestPathList);
3230 //Removal = StoreFragmentFromStack(out, FragmentSearch->Root, FragmentSearch->Leaflet, FragmentSearch->FragmentStack, FragmentSearch->ShortestPathList,FragmentSearch->Labels, &FragmentSearch->FragmentCounter, FragmentSearch->configuration);
3231 }
3232
3233 // --3-- remove all added items in this level from snake stack
3234 *out << Verbose(1+verbosity) << "Removing all items that were added on this SP level " << RootDistance << "." << endl;
3235 for(int j=0;j<TouchedIndex;j++) {
3236 Removal = TouchedList[j];
3237 *out << Verbose(2+verbosity) << "Removing item nr. " << Removal << " from snake stack." << endl;
3238 FragmentSearch->FragmentSet->erase(Removal);
3239 TouchedList[j] = -1;
3240 }
3241 *out << Verbose(2) << "Remaining local nr.s on snake stack are: ";
3242 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
3243 *out << (*runner) << " ";
3244 *out << endl;
3245 TouchedIndex = 0; // set Index to 0 for list of atoms added on this level
3246 } else {
3247 *out << Verbose(2+verbosity) << "More atoms to add for this set (" << bits << ") than space left on stack " << SubOrder << ", skipping this set." << endl;
3248 }
3249 }
3250 Free((void **)&TouchedList, "molecule::SPFragmentGenerator: *TouchedList");
3251 *out << Verbose(1+verbosity) << "End of SPFragmentGenerator, " << RootDistance << " away from Root " << *FragmentSearch->Root << " and SubOrder is " << SubOrder << "." << endl;
3252};
3253
3254/** 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.
3255 * -# initialises UniqueFragments structure
3256 * -# fills edge list via BFS
3257 * -# creates the fragment by calling recursive function SPFragmentGenerator with UniqueFragments structure, 0 as
3258 root distance, the edge set, its dimension and the current suborder
3259 * -# Free'ing structure
3260 * 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
3261 * with SP of 2, then those with SP of 3, then those with SP of 4 and so on.
3262 * \param *out output stream for debugging
3263 * \param Order bond order (limits BFS exploration and "number of digits" in power set generation
3264 * \param FragmentSearch UniqueFragments structure containing TEFactor, root atom and so on
3265 * \param RestrictedKeySet Restricted vertex set to use in context of molecule
3266 * \return number of inserted fragments
3267 * \note ShortestPathList in FragmentSearch structure is probably due to NumberOfAtomsSPLevel and SP not needed anymore
3268 */
3269int molecule::PowerSetGenerator(ofstream *out, int Order, struct UniqueFragments &FragmentSearch, KeySet RestrictedKeySet)
3270{
3271 int SP, UniqueIndex, AtomKeyNr;
3272 int *NumberOfAtomsSPLevel = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *SPLevelCount");
3273 atom *Walker = NULL, *OtherWalker = NULL;
3274 bond *Binder = NULL;
3275 bond **BondsList = NULL;
3276 KeyStack AtomStack;
3277 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::PowerSetGenerator: **PredecessorList");
3278 KeySet::iterator runner;
3279 int RootKeyNr = FragmentSearch.Root->nr;
3280 int Counter = FragmentSearch.FragmentCounter;
3281
3282 for (int i=0;i<AtomCount;i++) {
3283 PredecessorList[i] = NULL;
3284 }
3285
3286 *out << endl;
3287 *out << Verbose(0) << "Begin of PowerSetGenerator with order " << Order << " at Root " << *FragmentSearch.Root << "." << endl;
3288
3289 UniqueIndex = 0;
3290 if (FragmentSearch.Labels[RootKeyNr] == -1)
3291 FragmentSearch.Labels[RootKeyNr] = UniqueIndex++;
3292 FragmentSearch.ShortestPathList[RootKeyNr] = 0;
3293 // prepare the atom stack counters (number of atoms with certain SP on stack)
3294 for (int i=0;i<Order;i++)
3295 NumberOfAtomsSPLevel[i] = 0;
3296 NumberOfAtomsSPLevel[0] = 1; // for root
3297 SP = -1;
3298 *out << endl;
3299 *out << Verbose(0) << "Starting BFS analysis ..." << endl;
3300 // push as first on atom stack and goooo ...
3301 AtomStack.clear();
3302 AtomStack.push_back(RootKeyNr);
3303 *out << Verbose(0) << "Cleared AtomStack and pushed root as first item onto it." << endl;
3304 // do a BFS search to fill the SP lists and label the found vertices
3305 while (!AtomStack.empty()) {
3306 // pop next atom
3307 AtomKeyNr = AtomStack.front();
3308 AtomStack.pop_front();
3309 if (SP != -1)
3310 NumberOfAtomsSPLevel[SP]--;
3311 if ((SP == -1) || (NumberOfAtomsSPLevel[SP] == -1)) {
3312 SP++;
3313 NumberOfAtomsSPLevel[SP]--; // carry over "-1" to next level
3314 *out << Verbose(1) << "New SP level reached: " << SP << ", creating new SP list with 0 item(s)";
3315 if (SP > 0)
3316 *out << ", old level closed with " << FragmentSearch.BondsPerSPCount[SP-1] << " item(s)." << endl;
3317 else
3318 *out << "." << endl;
3319 FragmentSearch.BondsPerSPCount[SP] = 0;
3320 } else {
3321 *out << Verbose(1) << "Still " << NumberOfAtomsSPLevel[SP]+1 << " on this SP (" << SP << ") level, currently having " << FragmentSearch.BondsPerSPCount[SP] << " item(s)." << endl;
3322 }
3323 Walker = FindAtom(AtomKeyNr);
3324 *out << Verbose(0) << "Current Walker is: " << *Walker << " with nr " << Walker->nr << " and label " << FragmentSearch.Labels[AtomKeyNr] << " and SP of " << SP << "." << endl;
3325 // check for new sp level
3326 // go through all its bonds
3327 *out << Verbose(1) << "Going through all bonds of Walker." << endl;
3328 for (int i=0;i<NumberOfBondsPerAtom[AtomKeyNr];i++) {
3329 Binder = ListOfBondsPerAtom[AtomKeyNr][i];
3330 OtherWalker = Binder->GetOtherAtom(Walker);
3331 if ((RestrictedKeySet.find(OtherWalker->nr) != RestrictedKeySet.end())
3332#ifdef ADDHYDROGEN
3333 && (OtherWalker->type->Z != 1)
3334#endif
3335 ) { // skip hydrogens and restrict to fragment
3336 *out << Verbose(2) << "Current partner is " << *OtherWalker << " with nr " << OtherWalker->nr << " in bond " << *Binder << "." << endl;
3337 // set the label if not set (and push on root stack as well)
3338 if (FragmentSearch.Labels[OtherWalker->nr] == -1) {
3339 FragmentSearch.Labels[OtherWalker->nr] = UniqueIndex++;
3340 *out << Verbose(3) << "Set label to " << FragmentSearch.Labels[OtherWalker->nr] << "." << endl;
3341 } else {
3342 *out << Verbose(3) << "Label is already " << FragmentSearch.Labels[OtherWalker->nr] << "." << endl;
3343 }
3344 if ((OtherWalker != PredecessorList[AtomKeyNr]) && (OtherWalker->nr > RootKeyNr)) { // only pass through those with label bigger than Root's
3345 FragmentSearch.ShortestPathList[OtherWalker->nr] = SP+1;
3346 *out << Verbose(3) << "Set Shortest Path to " << FragmentSearch.ShortestPathList[OtherWalker->nr] << "." << endl;
3347 if (FragmentSearch.ShortestPathList[OtherWalker->nr] < Order) { // only pass through those within reach of Order
3348 // push for exploration on stack (only if the SP of OtherWalker is longer than of Walker! (otherwise it has been added already!)
3349 if (FragmentSearch.ShortestPathList[OtherWalker->nr] > SP) {
3350 if (FragmentSearch.ShortestPathList[OtherWalker->nr] < (Order-1)) {
3351 *out << Verbose(3) << "Putting on atom stack for further exploration." << endl;
3352 PredecessorList[OtherWalker->nr] = Walker; // note down the one further up the exploration tree
3353 AtomStack.push_back(OtherWalker->nr);
3354 NumberOfAtomsSPLevel[FragmentSearch.ShortestPathList[OtherWalker->nr]]++;
3355 } else {
3356 *out << Verbose(3) << "Not putting on atom stack, is at end of reach." << endl;
3357 }
3358 // add the bond in between to the SP list
3359 Binder = new bond(Walker, OtherWalker); // create a new bond in such a manner, that bond::rightatom is always the one more distant
3360 add(Binder, FragmentSearch.BondsPerSPList[2*SP+1]);
3361 FragmentSearch.BondsPerSPCount[SP]++;
3362 *out << Verbose(3) << "Added its bond to SP list, having now " << FragmentSearch.BondsPerSPCount[SP] << " item(s)." << endl;
3363 } else *out << Verbose(3) << "Not putting on atom stack, nor adding bond, as " << *OtherWalker << "'s SP is shorter than Walker's." << endl;
3364 } else *out << Verbose(3) << "Not continuing, as " << *OtherWalker << " is out of reach of order " << Order << "." << endl;
3365 } else *out << Verbose(3) << "Not passing on, as index of " << *OtherWalker << " " << OtherWalker->nr << " is smaller than that of Root " << RootKeyNr << " or this is my predecessor." << endl;
3366 } else *out << Verbose(2) << "Is not in the restricted keyset or skipping hydrogen " << *OtherWalker << "." << endl;
3367 }
3368 }
3369 // reset predecessor list
3370 for(int i=0;i<Order;i++) {
3371 Binder = FragmentSearch.BondsPerSPList[2*i];
3372 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
3373 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3374 Binder = Binder->next;
3375 PredecessorList[Binder->rightatom->nr] = NULL; // By construction "OtherAtom" is always Bond::rightatom!
3376 }
3377 }
3378 *out << endl;
3379
3380 // outputting all list for debugging
3381 *out << Verbose(0) << "Printing all found lists." << endl;
3382 for(int i=0;i<Order;i++) {
3383 Binder = FragmentSearch.BondsPerSPList[2*i];
3384 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
3385 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3386 Binder = Binder->next;
3387 *out << Verbose(2) << *Binder << endl;
3388 }
3389 }
3390
3391 // creating fragments with the found edge sets
3392 SP = 0;
3393 for(int i=0;i<Order;i++) { // sum up all found edges
3394 Binder = FragmentSearch.BondsPerSPList[2*i];
3395 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3396 Binder = Binder->next;
3397 SP ++;
3398 }
3399 }
3400 *out << Verbose(0) << "Total number of edges is " << SP << "." << endl;
3401 if (SP >= (Order-1)) {
3402 // start with root (push on fragment stack)
3403 *out << Verbose(0) << "Starting fragment generation with " << *FragmentSearch.Root << ", local nr is " << FragmentSearch.Root->nr << "." << endl;
3404 FragmentSearch.FragmentSet->clear();
3405 FragmentSearch.FragmentSet->insert(FragmentSearch.Root->nr);
3406
3407 if (FragmentSearch.FragmentSet->size() == (unsigned int) Order) {
3408 *out << Verbose(0) << "Enough items on stack already for a fragment!" << endl;
3409 // store fragment as a KeySet
3410 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch.FragmentCounter << "], local nr.s are: ";
3411 for(KeySet::iterator runner = FragmentSearch.FragmentSet->begin(); runner != FragmentSearch.FragmentSet->end(); runner++) {
3412 *out << (*runner) << " ";
3413 }
3414 *out << endl;
3415 InsertFragmentIntoGraph(out, &FragmentSearch);
3416 } else {
3417 *out << Verbose(0) << "Preparing subset for this root and calling generator." << endl;
3418 // prepare the subset and call the generator
3419 BondsList = (bond **) Malloc(sizeof(bond *)*FragmentSearch.BondsPerSPCount[0], "molecule::PowerSetGenerator: **BondsList");
3420 Binder = FragmentSearch.BondsPerSPList[0];
3421 for(int i=0;i<FragmentSearch.BondsPerSPCount[0];i++) {
3422 Binder = Binder->next;
3423 BondsList[i] = Binder;
3424 }
3425 SPFragmentGenerator(out, &FragmentSearch, 0, BondsList, FragmentSearch.BondsPerSPCount[0], Order-1);
3426 Free((void **)&BondsList, "molecule::PowerSetGenerator: **BondsList");
3427 }
3428 } else {
3429 *out << Verbose(0) << "Not enough total number of edges to build " << Order << "-body fragments." << endl;
3430 }
3431
3432 // as FragmentSearch structure is used only once, we don't have to clean it anymore
3433 // remove root from stack
3434 *out << Verbose(0) << "Removing root again from stack." << endl;
3435 FragmentSearch.FragmentSet->erase(FragmentSearch.Root->nr);
3436
3437 // free'ing the bonds lists
3438 *out << Verbose(0) << "Free'ing all found lists. and resetting index lists" << endl;
3439 for(int i=0;i<Order;i++) {
3440 *out << Verbose(1) << "Current SP level is " << i << ": ";
3441 Binder = FragmentSearch.BondsPerSPList[2*i];
3442 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3443 Binder = Binder->next;
3444 // *out << "Removing atom " << Binder->leftatom->nr << " and " << Binder->rightatom->nr << "." << endl; // make sure numbers are local
3445 FragmentSearch.ShortestPathList[Binder->leftatom->nr] = -1;
3446 FragmentSearch.ShortestPathList[Binder->rightatom->nr] = -1;
3447 }
3448 // delete added bonds
3449 cleanup(FragmentSearch.BondsPerSPList[2*i], FragmentSearch.BondsPerSPList[2*i+1]);
3450 // also start and end node
3451 *out << "cleaned." << endl;
3452 }
3453
3454 // free allocated memory
3455 Free((void **)&NumberOfAtomsSPLevel, "molecule::PowerSetGenerator: *SPLevelCount");
3456 Free((void **)&PredecessorList, "molecule::PowerSetGenerator: **PredecessorList");
3457
3458 // return list
3459 *out << Verbose(0) << "End of PowerSetGenerator." << endl;
3460 return (FragmentSearch.FragmentCounter - Counter);
3461};
3462
3463/** Corrects the nuclei position if the fragment was created over the cell borders.
3464 * Scans all bonds, checks the distance, if greater than typical, we have a candidate for the correction.
3465 * We remove the bond whereafter the graph probably separates. Then, we translate the one component periodically
3466 * and re-add the bond. Looping on the distance check.
3467 * \param *out ofstream for debugging messages
3468 */
3469void molecule::ScanForPeriodicCorrection(ofstream *out)
3470{
3471 bond *Binder = NULL;
3472 bond *OtherBinder = NULL;
3473 atom *Walker = NULL;
3474 atom *OtherWalker = NULL;
3475 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
3476 enum Shading *ColorList = NULL;
3477 double tmp;
3478 vector TranslationVector;
3479 //class StackClass<atom *> *CompStack = NULL;
3480 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
3481 bool flag = true;
3482
3483// *out << Verbose(1) << "Begin of ScanForPeriodicCorrection." << endl;
3484
3485 ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::ScanForPeriodicCorrection: *ColorList");
3486 while (flag) {
3487 // remove bonds that are beyond bonddistance
3488 for(int i=0;i<NDIM;i++)
3489 TranslationVector.x[i] = 0.;
3490 // scan all bonds
3491 Binder = first;
3492 flag = false;
3493 while ((!flag) && (Binder->next != last)) {
3494 Binder = Binder->next;
3495 for (int i=0;i<NDIM;i++) {
3496 tmp = fabs(Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i]);
3497 //*out << Verbose(3) << "Checking " << i << "th distance of " << *Binder->leftatom << " to " << *Binder->rightatom << ": " << tmp << "." << endl;
3498 if (tmp > BondDistance) {
3499 OtherBinder = Binder->next; // note down binding partner for later re-insertion
3500 unlink(Binder); // unlink bond
3501// *out << Verbose(2) << "Correcting at bond " << *Binder << "." << endl;
3502 flag = true;
3503 break;
3504 }
3505 }
3506 }
3507 if (flag) {
3508 // create translation vector from their periodically modified distance
3509 for (int i=0;i<NDIM;i++) {
3510 tmp = Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i];
3511 if (fabs(tmp) > BondDistance)
3512 TranslationVector.x[i] = (tmp < 0) ? +1. : -1.;
3513 }
3514 TranslationVector.MatrixMultiplication(matrix);
3515 //*out << "Translation vector is ";
3516 //TranslationVector.Output(out);
3517 //*out << endl;
3518 // apply to all atoms of first component via BFS
3519 for (int i=0;i<AtomCount;i++)
3520 ColorList[i] = white;
3521 AtomStack->Push(Binder->leftatom);
3522 while (!AtomStack->IsEmpty()) {
3523 Walker = AtomStack->PopFirst();
3524 //*out << Verbose (3) << "Current Walker is: " << *Walker << "." << endl;
3525 ColorList[Walker->nr] = black; // mark as explored
3526 Walker->x.AddVector(&TranslationVector); // translate
3527 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through all binding partners
3528 if (ListOfBondsPerAtom[Walker->nr][i] != Binder) {
3529 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
3530 if (ColorList[OtherWalker->nr] == white) {
3531 AtomStack->Push(OtherWalker); // push if yet unexplored
3532 }
3533 }
3534 }
3535 }
3536 // re-add bond
3537 link(Binder, OtherBinder);
3538 } else {
3539// *out << Verbose(2) << "No corrections for this fragment." << endl;
3540 }
3541 //delete(CompStack);
3542 }
3543
3544 // free allocated space from ReturnFullMatrixforSymmetric()
3545 delete(AtomStack);
3546 Free((void **)&ColorList, "molecule::ScanForPeriodicCorrection: *ColorList");
3547 Free((void **)&matrix, "molecule::ScanForPeriodicCorrection: *matrix");
3548// *out << Verbose(1) << "End of ScanForPeriodicCorrection." << endl;
3549};
3550
3551/** Blows the 6-dimensional \a cell_size array up to a full NDIM by NDIM matrix.
3552 * \param *symm 6-dim array of unique symmetric matrix components
3553 * \return allocated NDIM*NDIM array with the symmetric matrix
3554 */
3555double * molecule::ReturnFullMatrixforSymmetric(double *symm)
3556{
3557 double *matrix = (double *) Malloc(sizeof(double)*NDIM*NDIM, "molecule::ReturnFullMatrixforSymmetric: *matrix");
3558 matrix[0] = symm[0];
3559 matrix[1] = symm[1];
3560 matrix[2] = symm[3];
3561 matrix[3] = symm[1];
3562 matrix[4] = symm[2];
3563 matrix[5] = symm[4];
3564 matrix[6] = symm[3];
3565 matrix[7] = symm[4];
3566 matrix[8] = symm[5];
3567 return matrix;
3568};
3569
3570bool KeyCompare::operator() (const KeySet SubgraphA, const KeySet SubgraphB) const
3571{
3572 //cout << "my check is used." << endl;
3573 if (SubgraphA.size() < SubgraphB.size()) {
3574 return true;
3575 } else {
3576 if (SubgraphA.size() > SubgraphB.size()) {
3577 return false;
3578 } else {
3579 KeySet::iterator IteratorA = SubgraphA.begin();
3580 KeySet::iterator IteratorB = SubgraphB.begin();
3581 while ((IteratorA != SubgraphA.end()) && (IteratorB != SubgraphB.end())) {
3582 if ((*IteratorA) < (*IteratorB))
3583 return true;
3584 else if ((*IteratorA) > (*IteratorB)) {
3585 return false;
3586 } // else, go on to next index
3587 IteratorA++;
3588 IteratorB++;
3589 } // end of while loop
3590 }// end of check in case of equal sizes
3591 }
3592 return false; // if we reach this point, they are equal
3593};
3594
3595//bool operator < (KeySet SubgraphA, KeySet SubgraphB)
3596//{
3597// return KeyCompare(SubgraphA, SubgraphB);
3598//};
3599
3600/** Checking whether KeySet is not already present in Graph, if so just adds factor.
3601 * \param *out output stream for debugging
3602 * \param &set KeySet to insert
3603 * \param &graph Graph to insert into
3604 * \param *counter pointer to unique fragment count
3605 * \param factor energy factor for the fragment
3606 */
3607inline void InsertFragmentIntoGraph(ofstream *out, struct UniqueFragments *Fragment)
3608{
3609 GraphTestPair testGraphInsert;
3610
3611 testGraphInsert = Fragment->Leaflet->insert(GraphPair (*Fragment->FragmentSet,pair<int,double>(Fragment->FragmentCounter,Fragment->TEFactor))); // store fragment number and current factor
3612 if (testGraphInsert.second) {
3613 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " successfully inserted." << endl;
3614 Fragment->FragmentCounter++;
3615 } else {
3616 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
3617 ((*(testGraphInsert.first)).second).second += Fragment->TEFactor; // increase the "created" counter
3618 *out << Verbose(2) << "New factor is " << ((*(testGraphInsert.first)).second).second << "." << endl;
3619 }
3620};
3621//void inline InsertIntoGraph(ofstream *out, KeyStack &stack, Graph &graph, int *counter, double factor)
3622//{
3623// // copy stack contents to set and call overloaded function again
3624// KeySet set;
3625// for(KeyStack::iterator runner = stack.begin(); runner != stack.begin(); runner++)
3626// set.insert((*runner));
3627// InsertIntoGraph(out, set, graph, counter, factor);
3628//};
3629
3630/** Inserts each KeySet in \a graph2 into \a graph1.
3631 * \param *out output stream for debugging
3632 * \param graph1 first (dest) graph
3633 * \param graph2 second (source) graph
3634 * \param *counter keyset counter that gets increased
3635 */
3636inline void InsertGraphIntoGraph(ofstream *out, Graph &graph1, Graph &graph2, int *counter)
3637{
3638 GraphTestPair testGraphInsert;
3639
3640 for(Graph::iterator runner = graph2.begin(); runner != graph2.end(); runner++) {
3641 testGraphInsert = graph1.insert(GraphPair ((*runner).first,pair<int,double>((*counter)++,((*runner).second).second))); // store fragment number and current factor
3642 if (testGraphInsert.second) {
3643 *out << Verbose(2) << "KeySet " << (*counter)-1 << " successfully inserted." << endl;
3644 } else {
3645 *out << Verbose(2) << "KeySet " << (*counter)-1 << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
3646 ((*(testGraphInsert.first)).second).second += (*runner).second.second;
3647 *out << Verbose(2) << "New factor is " << (*(testGraphInsert.first)).second.second << "." << endl;
3648 }
3649 }
3650};
3651
3652
3653/** Performs BOSSANOVA decomposition at selected sites, increasing the cutoff by one at these sites.
3654 * -# constructs a complete keyset of the molecule
3655 * -# In a loop over all possible roots from the given rootstack
3656 * -# increases order of root site
3657 * -# calls PowerSetGenerator with this order, the complete keyset and the rootkeynr
3658 * -# for all consecutive lower levels PowerSetGenerator is called with the suborder, the higher order keyset
3659as the restricted one and each site in the set as the root)
3660 * -# these are merged into a fragment list of keysets
3661 * -# All fragment lists (for all orders, i.e. from all destination fields) are merged into one list for return
3662 * Important only is that we create all fragments, it is not important if we create them more than once
3663 * as these copies are filtered out via use of the hash table (KeySet).
3664 * \param *out output stream for debugging
3665 * \param Fragment&*List list of already present keystacks (adaptive scheme) or empty list
3666 * \param &RootStack stack with all root candidates (unequal to each atom in complete molecule if adaptive scheme is applied)
3667 * \return pointer to Graph list
3668 */
3669void molecule::FragmentBOSSANOVA(ofstream *out, Graph *&FragmentList, KeyStack &RootStack)
3670{
3671 Graph ***FragmentLowerOrdersList = NULL;
3672 int Order, NumLevels, NumMolecules, TotalNumMolecules = 0, *NumMoleculesOfOrder = NULL;
3673 int counter = 0;
3674 int UpgradeCount = RootStack.size();
3675 KeyStack FragmentRootStack;
3676 int RootKeyNr, RootNr;
3677 struct UniqueFragments FragmentSearch;
3678
3679 *out << Verbose(0) << "Begin of FragmentBOSSANOVA." << endl;
3680
3681 // FragmentLowerOrdersList is a 2D-array of pointer to MoleculeListClass objects, one dimension represents the ANOVA expansion of a single order (i.e. 5)
3682 // with all needed lower orders that are subtracted, the other dimension is the BondOrder (i.e. from 1 to 5)
3683 NumMoleculesOfOrder = (int *) Malloc(sizeof(int)*UpgradeCount, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
3684 FragmentLowerOrdersList = (Graph ***) Malloc(sizeof(Graph **)*UpgradeCount, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
3685
3686 // initialise the fragments structure
3687 FragmentSearch.BondsPerSPList = (bond **) Malloc(sizeof(bond *)*Order*2, "molecule::PowerSetGenerator: ***BondsPerSPList");
3688 FragmentSearch.BondsPerSPCount = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *BondsPerSPCount");
3689 FragmentSearch.ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *ShortestPathList");
3690 FragmentSearch.Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *Labels");
3691 FragmentSearch.FragmentCounter = 0;
3692 FragmentSearch.FragmentSet = new KeySet;
3693 FragmentSearch.Root = FindAtom(RootKeyNr);
3694 for (int i=0;i<AtomCount;i++) {
3695 FragmentSearch.Labels[i] = -1;
3696 FragmentSearch.ShortestPathList[i] = -1;
3697 }
3698
3699 // Construct the complete KeySet which we need for topmost level only (but for all Roots)
3700 atom *Walker = start;
3701 KeySet CompleteMolecule;
3702 while (Walker->next != end) {
3703 Walker = Walker->next;
3704 CompleteMolecule.insert(Walker->GetTrueFather()->nr);
3705 }
3706
3707 // 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
3708 // 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),
3709 // hence we have overall four 2th order levels for splitting. This also allows for putting all into a single array (FragmentLowerOrdersList[])
3710 // 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)
3711 RootNr = 0; // counts through the roots in RootStack
3712 while (RootNr < UpgradeCount) {
3713 RootKeyNr = RootStack.front();
3714 RootStack.pop_front();
3715 // increase adaptive order by one
3716 Walker = FindAtom(RootKeyNr);
3717 Walker->GetTrueFather()->AdaptiveOrder++;
3718 Order = Walker->AdaptiveOrder = Walker->GetTrueFather()->AdaptiveOrder;
3719
3720 // initialise Order-dependent entries of UniqueFragments structure
3721 FragmentSearch.BondsPerSPList = (bond **) Malloc(sizeof(bond *)*Order*2, "molecule::PowerSetGenerator: ***BondsPerSPList");
3722 FragmentSearch.BondsPerSPCount = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *BondsPerSPCount");
3723 for (int i=0;i<Order;i++) {
3724 FragmentSearch.BondsPerSPList[2*i] = new bond(); // start node
3725 FragmentSearch.BondsPerSPList[2*i+1] = new bond(); // end node
3726 FragmentSearch.BondsPerSPList[2*i]->next = FragmentSearch.BondsPerSPList[2*i+1]; // intertwine these two
3727 FragmentSearch.BondsPerSPList[2*i+1]->previous = FragmentSearch.BondsPerSPList[2*i];
3728 FragmentSearch.BondsPerSPCount[i] = 0;
3729 }
3730
3731 // allocate memory for all lower level orders in this 1D-array of ptrs
3732 NumLevels = 1 << (Order-1); // (int)pow(2,Order);
3733 FragmentLowerOrdersList[RootNr] = (Graph **) Malloc(sizeof(Graph *)*NumLevels, "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
3734
3735 // create top order where nothing is reduced
3736 *out << Verbose(0) << "==============================================================================================================" << endl;
3737 *out << Verbose(0) << "Creating KeySets of Bond Order " << Order << " for " << *Walker << ", NumLevels is " << NumLevels << ", " << (RootStack.size()-RootNr-1) << " Roots remaining." << endl;
3738
3739 // Create list of Graphs of current Bond Order (i.e. F_{ij})
3740 FragmentLowerOrdersList[RootNr][0] = new Graph;
3741 FragmentSearch.TEFactor = 1.;
3742 FragmentSearch.Leaflet = FragmentLowerOrdersList[RootNr][0]; // set to insertion graph
3743 FragmentSearch.Root = Walker;
3744 NumMoleculesOfOrder[RootNr] = PowerSetGenerator(out, Walker->AdaptiveOrder, FragmentSearch, CompleteMolecule);
3745 *out << Verbose(1) << "Number of resulting KeySets is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
3746 NumMolecules = 0;
3747
3748 if ((NumLevels >> 1) > 0) {
3749 // create lower order fragments
3750 *out << Verbose(0) << "Creating list of unique fragments of lower Bond Order terms to be subtracted." << endl;
3751 Order = Walker->AdaptiveOrder;
3752 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)
3753 // step down to next order at (virtual) boundary of powers of 2 in array
3754 while (source >= (1 << (Walker->AdaptiveOrder-Order))) // (int)pow(2,Walker->AdaptiveOrder-Order))
3755 Order--;
3756 *out << Verbose(0) << "Current Order is: " << Order << "." << endl;
3757 for (int SubOrder=Order-1;SubOrder>0;SubOrder--) {
3758 int dest = source + (1 << (Walker->AdaptiveOrder-(SubOrder+1)));
3759 *out << Verbose(0) << "--------------------------------------------------------------------------------------------------------------" << endl;
3760 *out << Verbose(0) << "Current SubOrder is: " << SubOrder << " with source " << source << " to destination " << dest << "." << endl;
3761
3762 // every molecule is split into a list of again (Order - 1) molecules, while counting all molecules
3763 //*out << Verbose(1) << "Splitting the " << (*FragmentLowerOrdersList[RootNr][source]).size() << " molecules of the " << source << "th cell in the array." << endl;
3764 //NumMolecules = 0;
3765 FragmentLowerOrdersList[RootNr][dest] = new Graph;
3766 for(Graph::iterator runner = (*FragmentLowerOrdersList[RootNr][source]).begin();runner != (*FragmentLowerOrdersList[RootNr][source]).end(); runner++) {
3767 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
3768 Graph TempFragmentList;
3769 FragmentSearch.TEFactor = -(*runner).second.second;
3770 FragmentSearch.Leaflet = &TempFragmentList; // set to insertion graph
3771 FragmentSearch.Root = FindAtom(*sprinter);
3772 NumMoleculesOfOrder[RootNr] += PowerSetGenerator(out, SubOrder, FragmentSearch, (*runner).first);
3773 // insert new keysets FragmentList into FragmentLowerOrdersList[Walker->AdaptiveOrder-1][dest]
3774 *out << Verbose(1) << "Merging resulting key sets with those present in destination " << dest << "." << endl;
3775 InsertGraphIntoGraph(out, *FragmentLowerOrdersList[RootNr][dest], TempFragmentList, &NumMolecules);
3776 }
3777 }
3778 *out << Verbose(1) << "Number of resulting molecules for SubOrder " << SubOrder << " is: " << NumMolecules << "." << endl;
3779 }
3780 }
3781 }
3782 // now, we have completely filled each cell of FragmentLowerOrdersList[] for the current Walker->AdaptiveOrder
3783 //NumMoleculesOfOrder[Walker->AdaptiveOrder-1] = NumMolecules;
3784 TotalNumMolecules += NumMoleculesOfOrder[RootNr];
3785 *out << Verbose(1) << "Number of resulting molecules for Order " << (int)Walker->GetTrueFather()->AdaptiveOrder << " is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
3786 RootStack.push_back(RootKeyNr); // put back on stack
3787 RootNr++;
3788
3789 // free Order-dependent entries of UniqueFragments structure for next loop cycle
3790 Free((void **)&FragmentSearch.BondsPerSPCount, "molecule::PowerSetGenerator: *BondsPerSPCount");
3791 for (int i=0;i<Order;i++) {
3792 delete(FragmentSearch.BondsPerSPList[2*i]);
3793 delete(FragmentSearch.BondsPerSPList[2*i+1]);
3794 }
3795 Free((void **)&FragmentSearch.BondsPerSPList, "molecule::PowerSetGenerator: ***BondsPerSPList");
3796 }
3797 *out << Verbose(0) << "==============================================================================================================" << endl;
3798 *out << Verbose(1) << "Total number of resulting molecules is: " << TotalNumMolecules << "." << endl;
3799 *out << Verbose(0) << "==============================================================================================================" << endl;
3800
3801 // cleanup FragmentSearch structure
3802 Free((void **)&FragmentSearch.ShortestPathList, "molecule::PowerSetGenerator: *ShortestPathList");
3803 Free((void **)&FragmentSearch.Labels, "molecule::PowerSetGenerator: *Labels");
3804 delete(FragmentSearch.FragmentSet);
3805
3806 // now, FragmentLowerOrdersList is complete, it looks - for BondOrder 5 - as this (number is the ANOVA Order of the terms therein)
3807 // 5433222211111111
3808 // 43221111
3809 // 3211
3810 // 21
3811 // 1
3812
3813 // Subsequently, we combine all into a single list (FragmentList)
3814
3815 *out << Verbose(0) << "Combining the lists of all orders per order and finally into a single one." << endl;
3816 if (FragmentList == NULL) {
3817 FragmentList = new Graph;
3818 counter = 0;
3819 } else {
3820 counter = FragmentList->size();
3821 }
3822 RootNr = 0;
3823 while (!RootStack.empty()) {
3824 RootKeyNr = RootStack.front();
3825 RootStack.pop_front();
3826 Walker = FindAtom(RootKeyNr);
3827 NumLevels = 1 << (Walker->AdaptiveOrder - 1);
3828 for(int i=0;i<NumLevels;i++) {
3829 InsertGraphIntoGraph(out, *FragmentList, (*FragmentLowerOrdersList[RootNr][i]), &counter);
3830 delete(FragmentLowerOrdersList[RootNr][i]);
3831 }
3832 Free((void **)&FragmentLowerOrdersList[RootNr], "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
3833 RootNr++;
3834 }
3835 Free((void **)&FragmentLowerOrdersList, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
3836 Free((void **)&NumMoleculesOfOrder, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
3837
3838 *out << Verbose(0) << "End of FragmentBOSSANOVA." << endl;
3839};
3840
3841/** Comparison function for GSL heapsort on distances in two molecules.
3842 * \param *a
3843 * \param *b
3844 * \return <0, \a *a less than \a *b, ==0 if equal, >0 \a *a greater than \a *b
3845 */
3846int CompareDoubles (const void * a, const void * b)
3847{
3848 if (*(double *)a > *(double *)b)
3849 return -1;
3850 else if (*(double *)a < *(double *)b)
3851 return 1;
3852 else
3853 return 0;
3854};
3855
3856/** Determines whether two molecules actually contain the same atoms and coordination.
3857 * \param *out output stream for debugging
3858 * \param *OtherMolecule the molecule to compare this one to
3859 * \param threshold upper limit of difference when comparing the coordination.
3860 * \return NULL - not equal, otherwise an allocated (molecule::AtomCount) permutation map of the atom numbers (which corresponds to which)
3861 */
3862int * molecule::IsEqualToWithinThreshold(ofstream *out, molecule *OtherMolecule, double threshold)
3863{
3864 int flag;
3865 double *Distances = NULL, *OtherDistances = NULL;
3866 vector CenterOfGravity, OtherCenterOfGravity;
3867 size_t *PermMap = NULL, *OtherPermMap = NULL;
3868 int *PermutationMap = NULL;
3869 atom *Walker = NULL;
3870 bool result = true; // status of comparison
3871
3872 *out << Verbose(3) << "Begin of IsEqualToWithinThreshold." << endl;
3873 /// first count both their atoms and elements and update lists thereby ...
3874 //*out << Verbose(0) << "Counting atoms, updating list" << endl;
3875 CountAtoms(out);
3876 OtherMolecule->CountAtoms(out);
3877 CountElements();
3878 OtherMolecule->CountElements();
3879
3880 /// ... and compare:
3881 /// -# AtomCount
3882 if (result) {
3883 if (AtomCount != OtherMolecule->AtomCount) {
3884 *out << Verbose(4) << "AtomCounts don't match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
3885 result = false;
3886 } else *out << Verbose(4) << "AtomCounts match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
3887 }
3888 /// -# ElementCount
3889 if (result) {
3890 if (ElementCount != OtherMolecule->ElementCount) {
3891 *out << Verbose(4) << "ElementCount don't match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
3892 result = false;
3893 } else *out << Verbose(4) << "ElementCount match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
3894 }
3895 /// -# ElementsInMolecule
3896 if (result) {
3897 for (flag=0;flag<MAX_ELEMENTS;flag++) {
3898 //*out << Verbose(5) << "Element " << flag << ": " << ElementsInMolecule[flag] << " <-> " << OtherMolecule->ElementsInMolecule[flag] << "." << endl;
3899 if (ElementsInMolecule[flag] != OtherMolecule->ElementsInMolecule[flag])
3900 break;
3901 }
3902 if (flag < MAX_ELEMENTS) {
3903 *out << Verbose(4) << "ElementsInMolecule don't match." << endl;
3904 result = false;
3905 } else *out << Verbose(4) << "ElementsInMolecule match." << endl;
3906 }
3907 /// then determine and compare center of gravity for each molecule ...
3908 if (result) {
3909 *out << Verbose(5) << "Calculating Centers of Gravity" << endl;
3910 DetermineCenterOfGravity(CenterOfGravity);
3911 OtherMolecule->DetermineCenterOfGravity(OtherCenterOfGravity);
3912 *out << Verbose(5) << "Center of Gravity: ";
3913 CenterOfGravity.Output(out);
3914 *out << endl << Verbose(5) << "Other Center of Gravity: ";
3915 OtherCenterOfGravity.Output(out);
3916 *out << endl;
3917 if (CenterOfGravity.Distance(&OtherCenterOfGravity) > threshold) {
3918 *out << Verbose(4) << "Centers of gravity don't match." << endl;
3919 result = false;
3920 }
3921 }
3922
3923 /// ... then make a list with the euclidian distance to this center for each atom of both molecules
3924 if (result) {
3925 *out << Verbose(5) << "Calculating distances" << endl;
3926 Distances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: Distances");
3927 OtherDistances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: OtherDistances");
3928 Walker = start;
3929 while (Walker->next != end) {
3930 Walker = Walker->next;
3931 //for (i=0;i<AtomCount;i++) {
3932 Distances[Walker->nr] = CenterOfGravity.Distance(&Walker->x);
3933 }
3934 Walker = OtherMolecule->start;
3935 while (Walker->next != OtherMolecule->end) {
3936 Walker = Walker->next;
3937 OtherDistances[Walker->nr] = OtherCenterOfGravity.Distance(&Walker->x);
3938 }
3939
3940 /// ... sort each list (using heapsort (o(N log N)) from GSL)
3941 *out << Verbose(5) << "Sorting distances" << endl;
3942 PermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermMap");
3943 OtherPermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
3944 gsl_heapsort_index (PermMap, Distances, AtomCount, sizeof(double), CompareDoubles);
3945 gsl_heapsort_index (OtherPermMap, OtherDistances, AtomCount, sizeof(double), CompareDoubles);
3946 PermutationMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermutationMap");
3947 *out << Verbose(5) << "Combining Permutation Maps" << endl;
3948 for(int i=0;i<AtomCount;i++)
3949 PermutationMap[PermMap[i]] = (int) OtherPermMap[i];
3950
3951 /// ... and compare them step by step, whether the difference is individiually(!) below \a threshold for all
3952 *out << Verbose(4) << "Comparing distances" << endl;
3953 flag = 0;
3954 for (int i=0;i<AtomCount;i++) {
3955 *out << Verbose(5) << "Distances: |" << Distances[PermMap[i]] << " - " << OtherDistances[OtherPermMap[i]] << "| = " << fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) << " ?<? " << threshold << endl;
3956 if (fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) > threshold)
3957 flag = 1;
3958 }
3959 Free((void **)&PermMap, "molecule::IsEqualToWithinThreshold: *PermMap");
3960 Free((void **)&OtherPermMap, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
3961
3962 /// free memory
3963 Free((void **)&Distances, "molecule::IsEqualToWithinThreshold: Distances");
3964 Free((void **)&OtherDistances, "molecule::IsEqualToWithinThreshold: OtherDistances");
3965 if (flag) { // if not equal
3966 Free((void **)&PermutationMap, "molecule::IsEqualToWithinThreshold: *PermutationMap");
3967 result = false;
3968 }
3969 }
3970 /// return pointer to map if all distances were below \a threshold
3971 *out << Verbose(3) << "End of IsEqualToWithinThreshold." << endl;
3972 if (result) {
3973 *out << Verbose(3) << "Result: Equal." << endl;
3974 return PermutationMap;
3975 } else {
3976 *out << Verbose(3) << "Result: Not equal." << endl;
3977 return NULL;
3978 }
3979};
3980
3981/** Returns an index map for two father-son-molecules.
3982 * The map tells which atom in this molecule corresponds to which one in the other molecul with their fathers.
3983 * \param *out output stream for debugging
3984 * \param *OtherMolecule corresponding molecule with fathers
3985 * \return allocated map of size molecule::AtomCount with map
3986 * \todo make this with a good sort O(n), not O(n^2)
3987 */
3988int * molecule::GetFatherSonAtomicMap(ofstream *out, molecule *OtherMolecule)
3989{
3990 atom *Walker = NULL, *OtherWalker = NULL;
3991 *out << Verbose(3) << "Begin of GetFatherAtomicMap." << endl;
3992 int *AtomicMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::GetAtomicMap: *AtomicMap"); //Calloc
3993 for (int i=0;i<AtomCount;i++)
3994 AtomicMap[i] = -1;
3995 if (OtherMolecule == this) { // same molecule
3996 for (int i=0;i<AtomCount;i++) // no need as -1 means already that there is trivial correspondence
3997 AtomicMap[i] = i;
3998 *out << Verbose(4) << "Map is trivial." << endl;
3999 } else {
4000 *out << Verbose(4) << "Map is ";
4001 Walker = start;
4002 while (Walker->next != end) {
4003 Walker = Walker->next;
4004 if (Walker->father == NULL) {
4005 AtomicMap[Walker->nr] = -2;
4006 } else {
4007 OtherWalker = OtherMolecule->start;
4008 while (OtherWalker->next != OtherMolecule->end) {
4009 OtherWalker = OtherWalker->next;
4010 //for (int i=0;i<AtomCount;i++) { // search atom
4011 //for (int j=0;j<OtherMolecule->AtomCount;j++) {
4012 //*out << Verbose(4) << "Comparing father " << Walker->father << " with the other one " << OtherWalker->father << "." << endl;
4013 if (Walker->father == OtherWalker)
4014 AtomicMap[Walker->nr] = OtherWalker->nr;
4015 }
4016 }
4017 *out << AtomicMap[Walker->nr] << "\t";
4018 }
4019 *out << endl;
4020 }
4021 *out << Verbose(3) << "End of GetFatherAtomicMap." << endl;
4022 return AtomicMap;
4023};
4024
Note: See TracBrowser for help on using the repository browser.