Renamed entropy.* to thermochemistry.*
[gromacs.git] / src / gromacs / gmxana / gmx_densmap.cpp
blobde5e182276f201911bfd38ad8e00a6a7dc5a5142
1 /*
2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2004, The GROMACS development team.
6 * Copyright (c) 2013,2014,2015,2016,2017, by the GROMACS development team, led by
7 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
8 * and including many others, as listed in the AUTHORS file in the
9 * top-level source directory and at http://www.gromacs.org.
11 * GROMACS is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public License
13 * as published by the Free Software Foundation; either version 2.1
14 * of the License, or (at your option) any later version.
16 * GROMACS is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with GROMACS; if not, see
23 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
24 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
26 * If you want to redistribute modifications to GROMACS, please
27 * consider that scientific software is very special. Version
28 * control is crucial - bugs must be traceable. We will be happy to
29 * consider code for inclusion in the official distribution, but
30 * derived work must not be called official GROMACS. Details are found
31 * in the README & COPYING files - if they are missing, get the
32 * official version at http://www.gromacs.org.
34 * To help us fund GROMACS development, we humbly ask that you cite
35 * the research papers on the package. Check out http://www.gromacs.org.
37 #include "gmxpre.h"
39 #include <cmath>
40 #include <cstring>
42 #include "gromacs/commandline/pargs.h"
43 #include "gromacs/commandline/viewit.h"
44 #include "gromacs/fileio/confio.h"
45 #include "gromacs/fileio/matio.h"
46 #include "gromacs/fileio/trxio.h"
47 #include "gromacs/gmxana/gmx_ana.h"
48 #include "gromacs/gmxana/gstat.h"
49 #include "gromacs/math/utilities.h"
50 #include "gromacs/math/vec.h"
51 #include "gromacs/pbcutil/pbc.h"
52 #include "gromacs/topology/index.h"
53 #include "gromacs/topology/topology.h"
54 #include "gromacs/utility/arraysize.h"
55 #include "gromacs/utility/cstringutil.h"
56 #include "gromacs/utility/fatalerror.h"
57 #include "gromacs/utility/futil.h"
58 #include "gromacs/utility/gmxassert.h"
59 #include "gromacs/utility/smalloc.h"
61 int gmx_densmap(int argc, char *argv[])
63 const char *desc[] = {
64 "[THISMODULE] computes 2D number-density maps.",
65 "It can make planar and axial-radial density maps.",
66 "The output [REF].xpm[ref] file can be visualized with for instance xv",
67 "and can be converted to postscript with [TT]xpm2ps[tt].",
68 "Optionally, output can be in text form to a [REF].dat[ref] file with [TT]-od[tt], instead of the usual [REF].xpm[ref] file with [TT]-o[tt].",
69 "[PAR]",
70 "The default analysis is a 2-D number-density map for a selected",
71 "group of atoms in the x-y plane.",
72 "The averaging direction can be changed with the option [TT]-aver[tt].",
73 "When [TT]-xmin[tt] and/or [TT]-xmax[tt] are set only atoms that are",
74 "within the limit(s) in the averaging direction are taken into account.",
75 "The grid spacing is set with the option [TT]-bin[tt].",
76 "When [TT]-n1[tt] or [TT]-n2[tt] is non-zero, the grid",
77 "size is set by this option.",
78 "Box size fluctuations are properly taken into account.",
79 "[PAR]",
80 "When options [TT]-amax[tt] and [TT]-rmax[tt] are set, an axial-radial",
81 "number-density map is made. Three groups should be supplied, the centers",
82 "of mass of the first two groups define the axis, the third defines the",
83 "analysis group. The axial direction goes from -amax to +amax, where",
84 "the center is defined as the midpoint between the centers of mass and",
85 "the positive direction goes from the first to the second center of mass.",
86 "The radial direction goes from 0 to rmax or from -rmax to +rmax",
87 "when the [TT]-mirror[tt] option has been set.",
88 "[PAR]",
89 "The normalization of the output is set with the [TT]-unit[tt] option.",
90 "The default produces a true number density. Unit [TT]nm-2[tt] leaves out",
91 "the normalization for the averaging or the angular direction.",
92 "Option [TT]count[tt] produces the count for each grid cell.",
93 "When you do not want the scale in the output to go",
94 "from zero to the maximum density, you can set the maximum",
95 "with the option [TT]-dmax[tt]."
97 static int n1 = 0, n2 = 0;
98 static real xmin = -1, xmax = -1, bin = 0.02, dmin = 0, dmax = 0, amax = 0, rmax = 0;
99 static gmx_bool bMirror = FALSE, bSums = FALSE;
100 static const char *eaver[] = { nullptr, "z", "y", "x", nullptr };
101 static const char *eunit[] = { nullptr, "nm-3", "nm-2", "count", nullptr };
103 t_pargs pa[] = {
104 { "-bin", FALSE, etREAL, {&bin},
105 "Grid size (nm)" },
106 { "-aver", FALSE, etENUM, {eaver},
107 "The direction to average over" },
108 { "-xmin", FALSE, etREAL, {&xmin},
109 "Minimum coordinate for averaging" },
110 { "-xmax", FALSE, etREAL, {&xmax},
111 "Maximum coordinate for averaging" },
112 { "-n1", FALSE, etINT, {&n1},
113 "Number of grid cells in the first direction" },
114 { "-n2", FALSE, etINT, {&n2},
115 "Number of grid cells in the second direction" },
116 { "-amax", FALSE, etREAL, {&amax},
117 "Maximum axial distance from the center"},
118 { "-rmax", FALSE, etREAL, {&rmax},
119 "Maximum radial distance" },
120 { "-mirror", FALSE, etBOOL, {&bMirror},
121 "Add the mirror image below the axial axis" },
122 { "-sums", FALSE, etBOOL, {&bSums},
123 "Print density sums (1D map) to stdout" },
124 { "-unit", FALSE, etENUM, {eunit},
125 "Unit for the output" },
126 { "-dmin", FALSE, etREAL, {&dmin},
127 "Minimum density in output"},
128 { "-dmax", FALSE, etREAL, {&dmax},
129 "Maximum density in output (0 means calculate it)"},
131 gmx_bool bXmin, bXmax, bRadial;
132 FILE *fp;
133 t_trxstatus *status;
134 t_topology top;
135 int ePBC = -1;
136 rvec *x, xcom[2], direction, center, dx;
137 matrix box;
138 real t, m, mtot;
139 t_pbc pbc;
140 int cav = 0, c1 = 0, c2 = 0;
141 char **grpname, buf[STRLEN];
142 const char *unit;
143 int i, j, k, l, ngrps, anagrp, *gnx = nullptr, nindex, nradial = 0, nfr, nmpower;
144 int **ind = nullptr, *index;
145 real **grid, maxgrid, m1, m2, box1, box2, *tickx, *tickz, invcellvol;
146 real invspa = 0, invspz = 0, axial, r, vol_old, vol, rowsum;
147 int nlev = 51;
148 t_rgb rlo = {1, 1, 1}, rhi = {0, 0, 0};
149 gmx_output_env_t *oenv;
150 const char *label[] = { "x (nm)", "y (nm)", "z (nm)" };
151 t_filenm fnm[] = {
152 { efTRX, "-f", nullptr, ffREAD },
153 { efTPS, nullptr, nullptr, ffOPTRD },
154 { efNDX, nullptr, nullptr, ffOPTRD },
155 { efDAT, "-od", "densmap", ffOPTWR },
156 { efXPM, "-o", "densmap", ffWRITE }
158 #define NFILE asize(fnm)
159 int npargs;
161 npargs = asize(pa);
163 if (!parse_common_args(&argc, argv, PCA_CAN_TIME | PCA_CAN_VIEW,
164 NFILE, fnm, npargs, pa, asize(desc), desc, 0, nullptr, &oenv))
166 return 0;
169 bXmin = opt2parg_bSet("-xmin", npargs, pa);
170 bXmax = opt2parg_bSet("-xmax", npargs, pa);
171 bRadial = (amax > 0 || rmax > 0);
172 if (bRadial)
174 if (amax <= 0 || rmax <= 0)
176 gmx_fatal(FARGS, "Both amax and rmax should be larger than zero");
180 GMX_RELEASE_ASSERT(eunit[0] != nullptr, "Option setting inconsistency; eunit[0] is NULL");
182 if (std::strcmp(eunit[0], "nm-3") == 0)
184 nmpower = -3;
185 unit = "(nm^-3)";
187 else if (std::strcmp(eunit[0], "nm-2") == 0)
189 nmpower = -2;
190 unit = "(nm^-2)";
192 else
194 nmpower = 0;
195 unit = "count";
198 if (ftp2bSet(efTPS, NFILE, fnm) || !ftp2bSet(efNDX, NFILE, fnm))
200 read_tps_conf(ftp2fn(efTPS, NFILE, fnm), &top, &ePBC, &x, nullptr, box,
201 bRadial);
203 if (!bRadial)
205 ngrps = 1;
206 fprintf(stderr, "\nSelect an analysis group\n");
208 else
210 ngrps = 3;
211 fprintf(stderr,
212 "\nSelect two groups to define the axis and an analysis group\n");
214 snew(gnx, ngrps);
215 snew(grpname, ngrps);
216 snew(ind, ngrps);
217 get_index(&top.atoms, ftp2fn_null(efNDX, NFILE, fnm), ngrps, gnx, ind, grpname);
218 anagrp = ngrps - 1;
219 nindex = gnx[anagrp];
220 index = ind[anagrp];
221 if (bRadial)
223 if ((gnx[0] > 1 || gnx[1] > 1) && !ftp2bSet(efTPS, NFILE, fnm))
225 gmx_fatal(FARGS, "No run input file was supplied (option -s), this is required for the center of mass calculation");
229 GMX_RELEASE_ASSERT(eaver[0] != nullptr, "Option setting inconsistency; eaver[0] is NULL");
231 switch (eaver[0][0])
233 case 'x': cav = XX; c1 = YY; c2 = ZZ; break;
234 case 'y': cav = YY; c1 = XX; c2 = ZZ; break;
235 case 'z': cav = ZZ; c1 = XX; c2 = YY; break;
238 read_first_x(oenv, &status, ftp2fn(efTRX, NFILE, fnm), &t, &x, box);
240 if (!bRadial)
242 if (n1 == 0)
244 n1 = static_cast<int>(box[c1][c1]/bin + 0.5);
246 if (n2 == 0)
248 n2 = static_cast<int>(box[c2][c2]/bin + 0.5);
251 else
253 n1 = static_cast<int>(2*amax/bin + 0.5);
254 nradial = static_cast<int>(rmax/bin + 0.5);
255 invspa = n1/(2*amax);
256 invspz = nradial/rmax;
257 if (bMirror)
259 n2 = 2*nradial;
261 else
263 n2 = nradial;
267 snew(grid, n1);
268 for (i = 0; i < n1; i++)
270 snew(grid[i], n2);
273 box1 = 0;
274 box2 = 0;
275 nfr = 0;
278 if (!bRadial)
280 box1 += box[c1][c1];
281 box2 += box[c2][c2];
282 invcellvol = n1*n2;
283 if (nmpower == -3)
285 invcellvol /= det(box);
287 else if (nmpower == -2)
289 invcellvol /= box[c1][c1]*box[c2][c2];
291 for (i = 0; i < nindex; i++)
293 j = index[i];
294 if ((!bXmin || x[j][cav] >= xmin) &&
295 (!bXmax || x[j][cav] <= xmax))
297 m1 = x[j][c1]/box[c1][c1];
298 if (m1 >= 1)
300 m1 -= 1;
302 if (m1 < 0)
304 m1 += 1;
306 m2 = x[j][c2]/box[c2][c2];
307 if (m2 >= 1)
309 m2 -= 1;
311 if (m2 < 0)
313 m2 += 1;
315 grid[static_cast<int>(m1*n1)][static_cast<int>(m2*n2)] += invcellvol;
319 else
321 set_pbc(&pbc, ePBC, box);
322 for (i = 0; i < 2; i++)
324 if (gnx[i] == 1)
326 /* One atom, just copy the coordinates */
327 copy_rvec(x[ind[i][0]], xcom[i]);
329 else
331 /* Calculate the center of mass */
332 clear_rvec(xcom[i]);
333 mtot = 0;
334 for (j = 0; j < gnx[i]; j++)
336 k = ind[i][j];
337 m = top.atoms.atom[k].m;
338 for (l = 0; l < DIM; l++)
340 xcom[i][l] += m*x[k][l];
342 mtot += m;
344 svmul(1/mtot, xcom[i], xcom[i]);
347 pbc_dx(&pbc, xcom[1], xcom[0], direction);
348 for (i = 0; i < DIM; i++)
350 center[i] = xcom[0][i] + 0.5*direction[i];
352 unitv(direction, direction);
353 for (i = 0; i < nindex; i++)
355 j = index[i];
356 pbc_dx(&pbc, x[j], center, dx);
357 axial = iprod(dx, direction);
358 r = std::sqrt(norm2(dx) - axial*axial);
359 if (axial >= -amax && axial < amax && r < rmax)
361 if (bMirror)
363 r += rmax;
365 grid[static_cast<int>((axial + amax)*invspa)][static_cast<int>(r*invspz)] += 1;
369 nfr++;
371 while (read_next_x(oenv, status, &t, x, box));
372 close_trx(status);
374 /* normalize gridpoints */
375 maxgrid = 0;
376 if (!bRadial)
378 for (i = 0; i < n1; i++)
380 for (j = 0; j < n2; j++)
382 grid[i][j] /= nfr;
383 if (grid[i][j] > maxgrid)
385 maxgrid = grid[i][j];
390 else
392 for (i = 0; i < n1; i++)
394 vol_old = 0;
395 for (j = 0; j < nradial; j++)
397 switch (nmpower)
399 case -3: vol = M_PI*(j+1)*(j+1)/(invspz*invspz*invspa); break;
400 case -2: vol = (j+1)/(invspz*invspa); break;
401 default: vol = j+1; break;
403 if (bMirror)
405 k = j + nradial;
407 else
409 k = j;
411 grid[i][k] /= nfr*(vol - vol_old);
412 if (bMirror)
414 grid[i][nradial-1-j] = grid[i][k];
416 vol_old = vol;
417 if (grid[i][k] > maxgrid)
419 maxgrid = grid[i][k];
424 fprintf(stdout, "\n The maximum density is %f %s\n", maxgrid, unit);
425 if (dmax > 0)
427 maxgrid = dmax;
430 snew(tickx, n1+1);
431 snew(tickz, n2+1);
432 if (!bRadial)
434 /* normalize box-axes */
435 box1 /= nfr;
436 box2 /= nfr;
437 for (i = 0; i <= n1; i++)
439 tickx[i] = i*box1/n1;
441 for (i = 0; i <= n2; i++)
443 tickz[i] = i*box2/n2;
446 else
448 for (i = 0; i <= n1; i++)
450 tickx[i] = i/invspa - amax;
452 if (bMirror)
454 for (i = 0; i <= n2; i++)
456 tickz[i] = i/invspz - rmax;
459 else
461 for (i = 0; i <= n2; i++)
463 tickz[i] = i/invspz;
468 if (bSums)
470 for (i = 0; i < n1; ++i)
472 fprintf(stdout, "Density sums:\n");
473 rowsum = 0;
474 for (j = 0; j < n2; ++j)
476 rowsum += grid[i][j];
478 fprintf(stdout, "%g\t", rowsum);
480 fprintf(stdout, "\n");
483 sprintf(buf, "%s number density", grpname[anagrp]);
484 if (!bRadial && (bXmin || bXmax))
486 if (!bXmax)
488 sprintf(buf+std::strlen(buf), ", %c > %g nm", eaver[0][0], xmin);
490 else if (!bXmin)
492 sprintf(buf+std::strlen(buf), ", %c < %g nm", eaver[0][0], xmax);
494 else
496 sprintf(buf+std::strlen(buf), ", %c: %g - %g nm", eaver[0][0], xmin, xmax);
499 if (ftp2bSet(efDAT, NFILE, fnm))
501 fp = gmx_ffopen(ftp2fn(efDAT, NFILE, fnm), "w");
502 /*optional text form output: first row is tickz; first col is tickx */
503 fprintf(fp, "0\t");
504 for (j = 0; j < n2; ++j)
506 fprintf(fp, "%g\t", tickz[j]);
508 fprintf(fp, "\n");
510 for (i = 0; i < n1; ++i)
512 fprintf(fp, "%g\t", tickx[i]);
513 for (j = 0; j < n2; ++j)
515 fprintf(fp, "%g\t", grid[i][j]);
517 fprintf(fp, "\n");
519 gmx_ffclose(fp);
521 else
523 fp = gmx_ffopen(ftp2fn(efXPM, NFILE, fnm), "w");
524 write_xpm(fp, MAT_SPATIAL_X | MAT_SPATIAL_Y, buf, unit,
525 bRadial ? "axial (nm)" : label[c1], bRadial ? "r (nm)" : label[c2],
526 n1, n2, tickx, tickz, grid, dmin, maxgrid, rlo, rhi, &nlev);
527 gmx_ffclose(fp);
530 do_view(oenv, opt2fn("-o", NFILE, fnm), nullptr);
532 return 0;