3 * This source code is part of
7 * GROningen MAchine for Chemical Simulations
10 * Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
11 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
12 * Copyright (c) 2001-2004, The GROMACS development team,
13 * check out http://www.gromacs.org for more information.
15 * This program is free software; you can redistribute it and/or
16 * modify it under the terms of the GNU General Public License
17 * as published by the Free Software Foundation; either version 2
18 * of the License, or (at your option) any later version.
20 * If you want to redistribute modifications, please consider that
21 * scientific software is very special. Version control is crucial -
22 * bugs must be traceable. We will be happy to consider code for
23 * inclusion in the official distribution, but derived work must not
24 * be called official GROMACS. Details are found in the README & COPYING
25 * files - if they are missing, get the official version at www.gromacs.org.
27 * To help us fund GROMACS development, we humbly ask that you cite
28 * the papers on the package - you can find them in the top README file.
30 * For more info, check our website at http://www.gromacs.org
33 * Green Red Orange Magenta Azure Cyan Skyblue
57 #include "eigensolver.h"
60 #include "sparsematrix.h"
67 nma_full_hessian(real * hess,
80 natoms = top->atoms.nr;
82 /* divide elements hess[i][j] by sqrt(mas[i])*sqrt(mas[j]) when required */
86 for (i=0; (i<natoms); i++)
88 for (j=0; (j<DIM); j++)
90 for (k=0; (k<natoms); k++)
92 mass_fac=gmx_invsqrt(top->atoms.atom[i].m*top->atoms.atom[k].m);
93 for (l=0; (l<DIM); l++)
94 hess[(i*DIM+j)*ndim+k*DIM+l]*=mass_fac;
100 /* call diagonalization routine. */
102 fprintf(stderr,"\nDiagonalizing to find vectors %d through %d...\n",begin,end);
105 eigensolver(hess,ndim,begin-1,end-1,eigenvalues,eigenvectors);
107 /* And scale the output eigenvectors */
108 if (bM && eigenvectors!=NULL)
110 for(i=0;i<(end-begin+1);i++)
112 for(j=0;j<natoms;j++)
114 mass_fac = gmx_invsqrt(top->atoms.atom[j].m);
115 for (k=0; (k<DIM); k++)
117 eigenvectors[i*ndim+j*DIM+k] *= mass_fac;
127 nma_sparse_hessian(gmx_sparsematrix_t * sparse_hessian,
141 natoms = top->atoms.nr;
144 /* Cannot check symmetry since we only store half matrix */
145 /* divide elements hess[i][j] by sqrt(mas[i])*sqrt(mas[j]) when required */
149 for (iatom=0; (iatom<natoms); iatom++)
151 for (j=0; (j<DIM); j++)
154 for(k=0;k<sparse_hessian->ndata[row];k++)
156 col = sparse_hessian->data[row][k].col;
158 mass_fac=gmx_invsqrt(top->atoms.atom[iatom].m*top->atoms.atom[katom].m);
159 sparse_hessian->data[row][k].value *=mass_fac;
164 fprintf(stderr,"\nDiagonalizing to find eigenvectors 1 through %d...\n",neig);
167 sparse_eigensolver(sparse_hessian,neig,eigenvalues,eigenvectors,10000000);
169 /* Scale output eigenvectors */
170 if (bM && eigenvectors!=NULL)
174 for(j=0;j<natoms;j++)
176 mass_fac = gmx_invsqrt(top->atoms.atom[j].m);
177 for (k=0; (k<DIM); k++)
179 eigenvectors[i*ndim+j*DIM+k] *= mass_fac;
188 int gmx_nmeig(int argc,char *argv[])
190 const char *desc[] = {
191 "[TT]g_nmeig[tt] calculates the eigenvectors/values of a (Hessian) matrix,",
192 "which can be calculated with [TT]mdrun[tt].",
193 "The eigenvectors are written to a trajectory file ([TT]-v[tt]).",
194 "The structure is written first with t=0. The eigenvectors",
195 "are written as frames with the eigenvector number as timestamp.",
196 "The eigenvectors can be analyzed with [TT]g_anaeig[tt].",
197 "An ensemble of structures can be generated from the eigenvectors with",
198 "[TT]g_nmens[tt]. When mass weighting is used, the generated eigenvectors",
199 "will be scaled back to plain Cartesian coordinates before generating the",
200 "output. In this case, they will no longer be exactly orthogonal in the",
201 "standard Cartesian norm, but in the mass-weighted norm they would be."
204 static gmx_bool bM=TRUE;
205 static int begin=1,end=50;
208 { "-m", FALSE, etBOOL, {&bM},
209 "Divide elements of Hessian by product of sqrt(mass) of involved "
210 "atoms prior to diagonalization. This should be used for 'Normal Modes' "
212 { "-first", FALSE, etINT, {&begin},
213 "First eigenvector to write away" },
214 { "-last", FALSE, etINT, {&end},
215 "Last eigenvector to write away" }
226 int natoms,ndim,nrow,ncol,count;
227 char *grpname,title[256];
232 real factor_gmx_to_omega2;
233 real factor_omega_to_wavenumber;
237 real * full_hessian = NULL;
238 gmx_sparsematrix_t * sparse_hessian = NULL;
241 { efMTX, "-f", "hessian", ffREAD },
242 { efTPS, NULL, NULL, ffREAD },
243 { efXVG, "-of", "eigenfreq", ffWRITE },
244 { efXVG, "-ol", "eigenval", ffWRITE },
245 { efTRN, "-v", "eigenvec", ffWRITE }
247 #define NFILE asize(fnm)
249 cr = init_par(&argc,&argv);
252 CopyRight(stderr,argv[0]);
254 parse_common_args(&argc,argv,PCA_BE_NICE | (MASTER(cr) ? 0 : PCA_QUIET),
255 NFILE,fnm,asize(pa),pa,asize(desc),desc,0,NULL,&oenv);
257 read_tps_conf(ftp2fn(efTPS,NFILE,fnm),title,&top,&ePBC,&top_x,NULL,box,bM);
259 natoms = top.atoms.nr;
267 /*open Hessian matrix */
268 gmx_mtxio_read(ftp2fn(efMTX,NFILE,fnm),&nrow,&ncol,&full_hessian,&sparse_hessian);
270 /* Memory for eigenvalues and eigenvectors (begin..end) */
271 snew(eigenvalues,nrow);
272 snew(eigenvectors,nrow*(end-begin+1));
274 /* If the Hessian is in sparse format we can calculate max (ndim-1) eigenvectors,
275 * and they must start at the first one. If this is not valid we convert to full matrix
276 * storage, but warn the user that we might run out of memory...
278 if((sparse_hessian != NULL) && (begin!=1 || end==ndim))
282 fprintf(stderr,"Cannot use sparse Hessian with first eigenvector != 1.\n");
286 fprintf(stderr,"Cannot use sparse Hessian to calculate all eigenvectors.\n");
289 fprintf(stderr,"Will try to allocate memory and convert to full matrix representation...\n");
291 snew(full_hessian,nrow*ncol);
292 for(i=0;i<nrow*ncol;i++)
295 for(i=0;i<sparse_hessian->nrow;i++)
297 for(j=0;j<sparse_hessian->ndata[i];j++)
299 k = sparse_hessian->data[i][j].col;
300 value = sparse_hessian->data[i][j].value;
301 full_hessian[i*ndim+k] = value;
302 full_hessian[k*ndim+i] = value;
305 gmx_sparsematrix_destroy(sparse_hessian);
306 sparse_hessian = NULL;
307 fprintf(stderr,"Converted sparse to full matrix storage.\n");
310 if(full_hessian != NULL)
312 /* Using full matrix storage */
313 nma_full_hessian(full_hessian,nrow,bM,&top,begin,end,eigenvalues,eigenvectors);
317 /* Sparse memory storage, allocate memory for eigenvectors */
318 snew(eigenvectors,ncol*end);
319 nma_sparse_hessian(sparse_hessian,bM,&top,end,eigenvalues,eigenvectors);
323 /* check the output, first 6 eigenvalues should be reasonably small */
325 for (i=begin-1; (i<6); i++)
327 if (fabs(eigenvalues[i]) > 1.0e-3)
332 fprintf(stderr,"\nOne of the lowest 6 eigenvalues has a non-zero value.\n");
333 fprintf(stderr,"This could mean that the reference structure was not\n");
334 fprintf(stderr,"properly energy minimized.\n");
338 /* now write the output */
339 fprintf (stderr,"Writing eigenvalues...\n");
340 out=xvgropen(opt2fn("-ol",NFILE,fnm),
341 "Eigenvalues","Eigenvalue index","Eigenvalue [Gromacs units]",
343 if (output_env_get_print_xvgr_codes(oenv)) {
345 fprintf(out,"@ subtitle \"mass weighted\"\n");
347 fprintf(out,"@ subtitle \"not mass weighted\"\n");
350 for (i=0; i<=(end-begin); i++)
351 fprintf (out,"%6d %15g\n",begin+i,eigenvalues[i]);
356 fprintf(stderr,"Writing eigenfrequencies - negative eigenvalues will be set to zero.\n");
358 out=xvgropen(opt2fn("-of",NFILE,fnm),
359 "Eigenfrequencies","Eigenvector index","Wavenumber [cm\\S-1\\N]",
361 if (output_env_get_print_xvgr_codes(oenv)) {
363 fprintf(out,"@ subtitle \"mass weighted\"\n");
365 fprintf(out,"@ subtitle \"not mass weighted\"\n");
368 /* Gromacs units are kJ/(mol*nm*nm*amu),
369 * where amu is the atomic mass unit.
371 * For the eigenfrequencies we want to convert this to spectroscopic absorption
372 * wavenumbers given in cm^(-1), which is the frequency divided by the speed of
373 * light. Do this by first converting to omega^2 (units 1/s), take the square
374 * root, and finally divide by the speed of light (nm/ps in gromacs).
376 factor_gmx_to_omega2 = 1.0E21/(AVOGADRO*AMU);
377 factor_omega_to_wavenumber = 1.0E-5/(2.0*M_PI*SPEED_OF_LIGHT);
379 for (i=0; i<=(end-begin); i++)
381 value = eigenvalues[i];
384 value=sqrt(value*factor_gmx_to_omega2)*factor_omega_to_wavenumber;
385 fprintf (out,"%6d %15g\n",begin+i,value);
389 /* Writing eigenvectors. Note that if mass scaling was used, the eigenvectors
390 * were scaled back from mass weighted cartesian to plain cartesian in the
391 * nma_full_hessian() or nma_sparse_hessian() routines. Mass scaled vectors
392 * will not be strictly orthogonal in plain cartesian scalar products.
394 write_eigenvectors(opt2fn("-v",NFILE,fnm),natoms,eigenvectors,FALSE,begin,end,
395 eWXR_NO,NULL,FALSE,top_x,bM,eigenvalues);