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, 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.
42 #include "gromacs/math/utilities.h"
43 #include "gromacs/legacyheaders/typedefs.h"
44 #include "gromacs/legacyheaders/names.h"
45 #include "gromacs/utility/smalloc.h"
46 #include "gromacs/utility/fatalerror.h"
47 #include "gromacs/utility/futil.h"
48 #include "gromacs/fileio/xvgr.h"
49 #include "gromacs/math/vec.h"
50 #include "gromacs/legacyheaders/network.h"
51 #include "gromacs/math/units.h"
52 #include "gromacs/legacyheaders/force.h"
53 #include "gromacs/fileio/gmxfio.h"
54 #include "gromacs/legacyheaders/macros.h"
55 #include "gromacs/legacyheaders/tables.h"
57 /* All the possible (implemented) table functions */
83 /** Evaluates to true if the table type contains user data. */
84 #define ETAB_USER(e) ((e) == etabUSER || \
85 (e) == etabEwaldUser || (e) == etabEwaldUserSwitch)
92 /* This structure holds name and a flag that tells whether
93 this is a Coulomb type funtion */
94 static const t_tab_props tprops[etabNR] = {
97 { "LJ6Shift", FALSE },
98 { "LJ12Shift", FALSE },
104 { "Ewald-Switch", TRUE },
105 { "Ewald-User", TRUE },
106 { "Ewald-User-Switch", TRUE },
107 { "LJ6Ewald", FALSE },
108 { "LJ6Switch", FALSE },
109 { "LJ12Switch", FALSE },
110 { "COULSwitch", TRUE },
111 { "LJ6-Encad shift", FALSE },
112 { "LJ12-Encad shift", FALSE },
113 { "COUL-Encad shift", TRUE },
118 /* Index in the table that says which function to use */
120 etiCOUL, etiLJ6, etiLJ12, etiNR
129 #define pow2(x) ((x)*(x))
130 #define pow3(x) ((x)*(x)*(x))
131 #define pow4(x) ((x)*(x)*(x)*(x))
132 #define pow5(x) ((x)*(x)*(x)*(x)*(x))
134 double v_q_ewald_lr(double beta, double r)
138 return beta*2/sqrt(M_PI);
142 return gmx_erfd(beta*r)/r;
146 double v_lj_ewald_lr(double beta, double r)
148 double br, br2, br4, r6, factor;
151 return pow(beta, 6)/6;
159 factor = (1.0 - exp(-br2)*(1 + br2 + 0.5*br4))/r6;
164 void table_spline3_fill_ewald_lr(real *table_f,
170 real_space_grid_contribution_computer v_lr)
175 gmx_bool bOutOfRange;
176 double v_r0, v_r1, v_inrange, vi, a0, a1, a2dx;
179 /* This function is called using either v_ewald_lr or v_lj_ewald_lr as a function argument
180 * depending on wether we should create electrostatic or Lennard-Jones Ewald tables.
185 gmx_fatal(FARGS, "Can not make a spline table with less than 2 points");
188 /* We need some margin to be able to divide table values by r
189 * in the kernel and also to do the integration arithmetics
190 * without going out of range. Furthemore, we divide by dx below.
192 tab_max = GMX_REAL_MAX*0.0001;
194 /* This function produces a table with:
195 * maximum energy error: V'''/(6*12*sqrt(3))*dx^3
196 * maximum force error: V'''/(6*4)*dx^2
197 * The rms force error is the max error times 1/sqrt(5)=0.45.
204 for (i = ntab-1; i >= 0; i--)
208 v_r0 = (*v_lr)(beta, x_r0);
219 /* Linear continuation for the last point in range */
220 vi = v_inrange - dc*(i - i_inrange)*dx;
233 /* Get the potential at table point i-1 */
234 v_r1 = (*v_lr)(beta, (i-1)*dx);
236 if (v_r1 != v_r1 || v_r1 < -tab_max || v_r1 > tab_max)
243 /* Calculate the average second derivative times dx over interval i-1 to i.
244 * Using the function values at the end points and in the middle.
246 a2dx = (v_r0 + v_r1 - 2*(*v_lr)(beta, x_r0-0.5*dx))/(0.25*dx);
247 /* Set the derivative of the spline to match the difference in potential
248 * over the interval plus the average effect of the quadratic term.
249 * This is the essential step for minimizing the error in the force.
251 dc = (v_r0 - v_r1)/dx + 0.5*a2dx;
256 /* Fill the table with the force, minus the derivative of the spline */
261 /* tab[i] will contain the average of the splines over the two intervals */
262 table_f[i] += -0.5*dc;
267 /* Make spline s(x) = a0 + a1*(x - xr) + 0.5*a2*(x - xr)^2
268 * matching the potential at the two end points
269 * and the derivative dc at the end point xr.
273 a2dx = (a1*dx + v_r1 - a0)*2/dx;
275 /* Set dc to the derivative at the next point */
278 if (dc_new != dc_new || dc_new < -tab_max || dc_new > tab_max)
288 table_f[(i-1)] = -0.5*dc;
290 /* Currently the last value only contains half the force: double it */
293 if (table_v != NULL && table_fdv0 != NULL)
295 /* Copy to FDV0 table too. Allocation occurs in forcerec.c,
296 * init_ewald_f_table().
298 for (i = 0; i < ntab-1; i++)
300 table_fdv0[4*i] = table_f[i];
301 table_fdv0[4*i+1] = table_f[i+1]-table_f[i];
302 table_fdv0[4*i+2] = table_v[i];
303 table_fdv0[4*i+3] = 0.0;
305 table_fdv0[4*(ntab-1)] = table_f[(ntab-1)];
306 table_fdv0[4*(ntab-1)+1] = -table_f[(ntab-1)];
307 table_fdv0[4*(ntab-1)+2] = table_v[(ntab-1)];
308 table_fdv0[4*(ntab-1)+3] = 0.0;
312 /* Returns the spacing for a function using the maximum of
313 * the third derivative, x_scale (unit 1/length)
314 * and function tolerance.
316 static double spline3_table_scale(double third_deriv_max,
321 double sc_deriv, sc_func;
323 /* Force tolerance: single precision accuracy */
324 deriv_tol = GMX_FLOAT_EPS;
325 sc_deriv = sqrt(third_deriv_max/(6*4*deriv_tol*x_scale))*x_scale;
327 /* Don't try to be more accurate on energy than the precision */
328 func_tol = max(func_tol, GMX_REAL_EPS);
329 sc_func = pow(third_deriv_max/(6*12*sqrt(3)*func_tol), 1.0/3.0)*x_scale;
331 return max(sc_deriv, sc_func);
334 /* The scale (1/spacing) for third order spline interpolation
335 * of the Ewald mesh contribution which needs to be subtracted
336 * from the non-bonded interactions.
337 * Since there is currently only one spacing for Coulomb and LJ,
338 * the finest spacing is used if both Ewald types are present.
340 * Note that we could also implement completely separate tables
341 * for Coulomb and LJ Ewald, each with their own spacing.
342 * The current setup with the same spacing can provide slightly
343 * faster kernels with both Coulomb and LJ Ewald, especially
344 * when interleaving both tables (currently not implemented).
346 real ewald_spline3_table_scale(const interaction_const_t *ic)
352 if (ic->eeltype == eelEWALD || EEL_PME(ic->eeltype))
354 double erf_x_d3 = 1.0522; /* max of (erf(x)/x)''' */
358 /* Energy tolerance: 0.1 times the cut-off jump */
359 etol = 0.1*gmx_erfc(ic->ewaldcoeff_q*ic->rcoulomb);
361 sc_q = spline3_table_scale(erf_x_d3, ic->ewaldcoeff_q, etol);
365 fprintf(debug, "Ewald Coulomb quadratic spline table spacing: %f 1/nm\n", 1/sc_q);
371 if (EVDW_PME(ic->vdwtype))
373 double func_d3 = 0.42888; /* max of (x^-6 (1 - exp(-x^2)(1+x^2+x^4/2)))''' */
377 /* Energy tolerance: 0.1 times the cut-off jump */
378 xrc2 = sqr(ic->ewaldcoeff_lj*ic->rvdw);
379 etol = 0.1*exp(-xrc2)*(1 + xrc2 + xrc2*xrc2/2.0);
381 sc_lj = spline3_table_scale(func_d3, ic->ewaldcoeff_lj, etol);
385 fprintf(debug, "Ewald LJ quadratic spline table spacing: %f 1/nm\n", 1/sc_lj);
394 /* Calculate the potential and force for an r value
395 * in exactly the same way it is done in the inner loop.
396 * VFtab is a pointer to the table data, offset is
397 * the point where we should begin and stride is
398 * 4 if we have a buckingham table, 3 otherwise.
399 * If you want to evaluate table no N, set offset to 4*N.
401 * We use normal precision here, since that is what we
402 * will use in the inner loops.
404 static void evaluate_table(real VFtab[], int offset, int stride,
405 real tabscale, real r, real *y, real *yp)
409 real Y, F, Geps, Heps2, Fp;
418 Geps = eps*VFtab[n+2];
419 Heps2 = eps2*VFtab[n+3];
422 *yp = (Fp+Geps+2.0*Heps2)*tabscale;
425 static void copy2table(int n, int offset, int stride,
426 double x[], double Vtab[], double Ftab[], real scalefactor,
429 /* Use double prec. for the intermediary variables
430 * and temporary x/vtab/vtab2 data to avoid unnecessary
437 for (i = 0; (i < n); i++)
443 G = 3*(Vtab[i+1] - Vtab[i]) + (Ftab[i+1] + 2*Ftab[i])*h;
444 H = -2*(Vtab[i+1] - Vtab[i]) - (Ftab[i+1] + Ftab[i])*h;
448 /* Fill the last entry with a linear potential,
449 * this is mainly for rounding issues with angle and dihedral potentials.
455 nn0 = offset + i*stride;
456 dest[nn0] = scalefactor*Vtab[i];
457 dest[nn0+1] = scalefactor*F;
458 dest[nn0+2] = scalefactor*G;
459 dest[nn0+3] = scalefactor*H;
463 static void init_table(int n, int nx0,
464 double tabscale, t_tabledata *td, gmx_bool bAlloc)
470 td->tabscale = tabscale;
477 for (i = 0; (i < td->nx); i++)
479 td->x[i] = i/tabscale;
483 static void spline_forces(int nx, double h, double v[], gmx_bool bS3, gmx_bool bE3,
487 double v3, b_s, b_e, b;
490 /* Formulas can be found in:
491 * H.J.C. Berendsen, Simulating the Physical World, Cambridge 2007
494 if (nx < 4 && (bS3 || bE3))
496 gmx_fatal(FARGS, "Can not generate splines with third derivative boundary conditions with less than 4 (%d) points", nx);
499 /* To make life easy we initially set the spacing to 1
500 * and correct for this at the end.
505 /* Fit V''' at the start */
506 v3 = v[3] - 3*v[2] + 3*v[1] - v[0];
509 fprintf(debug, "The left third derivative is %g\n", v3/(h*h*h));
511 b_s = 2*(v[1] - v[0]) + v3/6;
516 /* Fit V'' at the start */
519 v2 = -v[3] + 4*v[2] - 5*v[1] + 2*v[0];
520 /* v2 = v[2] - 2*v[1] + v[0]; */
523 fprintf(debug, "The left second derivative is %g\n", v2/(h*h));
525 b_s = 3*(v[1] - v[0]) - v2/2;
531 b_s = 3*(v[2] - v[0]) + f[0]*h;
536 /* Fit V''' at the end */
537 v3 = v[nx-1] - 3*v[nx-2] + 3*v[nx-3] - v[nx-4];
540 fprintf(debug, "The right third derivative is %g\n", v3/(h*h*h));
542 b_e = 2*(v[nx-1] - v[nx-2]) + v3/6;
547 /* V'=0 at the end */
548 b_e = 3*(v[nx-1] - v[nx-3]) + f[nx-1]*h;
553 beta = (bS3 ? 1 : 4);
555 /* For V'' fitting */
556 /* beta = (bS3 ? 2 : 4); */
559 for (i = start+1; i < end; i++)
563 b = 3*(v[i+1] - v[i-1]);
564 f[i] = (b - f[i-1])/beta;
566 gamma[end-1] = 1/beta;
567 beta = (bE3 ? 1 : 4) - gamma[end-1];
568 f[end-1] = (b_e - f[end-2])/beta;
570 for (i = end-2; i >= start; i--)
572 f[i] -= gamma[i+1]*f[i+1];
576 /* Correct for the minus sign and the spacing */
577 for (i = start; i < end; i++)
583 static void set_forces(FILE *fp, int angle,
584 int nx, double h, double v[], double f[],
592 "Force generation for dihedral tables is not (yet) implemented");
596 while (v[start] == 0)
602 while (v[end-1] == 0)
617 fprintf(fp, "Generating forces for table %d, boundary conditions: V''' at %g, %s at %g\n",
618 table+1, start*h, end == nx ? "V'''" : "V'=0", (end-1)*h);
620 spline_forces(end-start, h, v+start, TRUE, end == nx, f+start);
623 static void read_tables(FILE *fp, const char *fn,
624 int ntab, int angle, t_tabledata td[])
628 double **yy = NULL, start, end, dx0, dx1, ssd, vm, vp, f, numf;
629 int k, i, nx, nx0 = 0, ny, nny, ns;
630 gmx_bool bAllZero, bZeroV, bZeroF;
634 libfn = gmxlibfn(fn);
635 nx = read_xvg(libfn, &yy, &ny);
638 gmx_fatal(FARGS, "Trying to read file %s, but nr columns = %d, should be %d",
646 "The first distance in file %s is %f nm instead of %f nm",
647 libfn, yy[0][0], 0.0);
661 if (yy[0][0] != start || yy[0][nx-1] != end)
663 gmx_fatal(FARGS, "The angles in file %s should go from %f to %f instead of %f to %f\n",
664 libfn, start, end, yy[0][0], yy[0][nx-1]);
668 tabscale = (nx-1)/(yy[0][nx-1] - yy[0][0]);
672 fprintf(fp, "Read user tables from %s with %d data points.\n", libfn, nx);
675 fprintf(fp, "Tabscale = %g points/nm\n", tabscale);
680 for (k = 0; k < ntab; k++)
684 for (i = 0; (i < nx); i++)
688 dx0 = yy[0][i-1] - yy[0][i-2];
689 dx1 = yy[0][i] - yy[0][i-1];
690 /* Check for 1% deviation in spacing */
691 if (fabs(dx1 - dx0) >= 0.005*(fabs(dx0) + fabs(dx1)))
693 gmx_fatal(FARGS, "In table file '%s' the x values are not equally spaced: %f %f %f", fn, yy[0][i-2], yy[0][i-1], yy[0][i]);
696 if (yy[1+k*2][i] != 0)
704 if (yy[1+k*2][i] > 0.01*GMX_REAL_MAX ||
705 yy[1+k*2][i] < -0.01*GMX_REAL_MAX)
707 gmx_fatal(FARGS, "Out of range potential value %g in file '%s'",
711 if (yy[1+k*2+1][i] != 0)
719 if (yy[1+k*2+1][i] > 0.01*GMX_REAL_MAX ||
720 yy[1+k*2+1][i] < -0.01*GMX_REAL_MAX)
722 gmx_fatal(FARGS, "Out of range force value %g in file '%s'",
728 if (!bZeroV && bZeroF)
730 set_forces(fp, angle, nx, 1/tabscale, yy[1+k*2], yy[1+k*2+1], k);
734 /* Check if the second column is close to minus the numerical
735 * derivative of the first column.
739 for (i = 1; (i < nx-1); i++)
744 if (vm != 0 && vp != 0 && f != 0)
746 /* Take the centered difference */
747 numf = -(vp - vm)*0.5*tabscale;
748 ssd += fabs(2*(f - numf)/(f + numf));
755 sprintf(buf, "For the %d non-zero entries for table %d in %s the forces deviate on average %d%% from minus the numerical derivative of the potential\n", ns, k, libfn, (int)(100*ssd+0.5));
758 fprintf(debug, "%s", buf);
764 fprintf(fp, "\nWARNING: %s\n", buf);
766 fprintf(stderr, "\nWARNING: %s\n", buf);
773 fprintf(fp, "\nNOTE: All elements in table %s are zero\n\n", libfn);
776 for (k = 0; (k < ntab); k++)
778 init_table(nx, nx0, tabscale, &(td[k]), TRUE);
779 for (i = 0; (i < nx); i++)
781 td[k].x[i] = yy[0][i];
782 td[k].v[i] = yy[2*k+1][i];
783 td[k].f[i] = yy[2*k+2][i];
786 for (i = 0; (i < ny); i++)
794 static void done_tabledata(t_tabledata *td)
808 static void fill_table(t_tabledata *td, int tp, const t_forcerec *fr,
811 /* Fill the table according to the formulas in the manual.
812 * In principle, we only need the potential and the second
813 * derivative, but then we would have to do lots of calculations
814 * in the inner loop. By precalculating some terms (see manual)
815 * we get better eventual performance, despite a larger table.
817 * Since some of these higher-order terms are very small,
818 * we always use double precision to calculate them here, in order
819 * to avoid unnecessary loss of precision.
826 double r1, rc, r12, r13;
827 double r, r2, r6, rc2, rc6, rc12;
828 double expr, Vtab, Ftab;
829 /* Parameters for David's function */
830 double A = 0, B = 0, C = 0, A_3 = 0, B_4 = 0;
831 /* Parameters for the switching function */
832 double ksw, swi, swi1;
833 /* Temporary parameters */
834 gmx_bool bPotentialSwitch, bForceSwitch, bPotentialShift;
835 double ewc = fr->ewaldcoeff_q;
836 double ewclj = fr->ewaldcoeff_lj;
841 bPotentialSwitch = FALSE;
842 bForceSwitch = FALSE;
843 bPotentialShift = FALSE;
847 bPotentialSwitch = ((tp == etabLJ6Switch) || (tp == etabLJ12Switch) ||
848 (tp == etabCOULSwitch) ||
849 (tp == etabEwaldSwitch) || (tp == etabEwaldUserSwitch) ||
850 (tprops[tp].bCoulomb && (fr->coulomb_modifier == eintmodPOTSWITCH)) ||
851 (!tprops[tp].bCoulomb && (fr->vdw_modifier == eintmodPOTSWITCH)));
852 bForceSwitch = ((tp == etabLJ6Shift) || (tp == etabLJ12Shift) ||
854 (tprops[tp].bCoulomb && (fr->coulomb_modifier == eintmodFORCESWITCH)) ||
855 (!tprops[tp].bCoulomb && (fr->vdw_modifier == eintmodFORCESWITCH)));
856 bPotentialShift = ((tprops[tp].bCoulomb && (fr->coulomb_modifier == eintmodPOTSHIFT)) ||
857 (!tprops[tp].bCoulomb && (fr->vdw_modifier == eintmodPOTSHIFT)));
862 if (tprops[tp].bCoulomb)
864 r1 = fr->rcoulomb_switch;
869 r1 = fr->rvdw_switch;
872 if (bPotentialSwitch)
874 ksw = 1.0/(pow5(rc-r1));
886 else if (tp == etabLJ6Shift)
895 A = p * ((p+1)*r1-(p+4)*rc)/(pow(rc, p+2)*pow2(rc-r1));
896 B = -p * ((p+1)*r1-(p+3)*rc)/(pow(rc, p+2)*pow3(rc-r1));
897 C = 1.0/pow(rc, p)-A/3.0*pow3(rc-r1)-B/4.0*pow4(rc-r1);
898 if (tp == etabLJ6Shift)
909 fprintf(debug, "Setting up tables\n"); fflush(debug);
913 fp = xvgropen("switch.xvg", "switch", "r", "s");
919 rc6 = 1.0/(rc2*rc2*rc2);
920 if (gmx_within_tol(reppow, 12.0, 10*GMX_DOUBLE_EPS))
926 rc12 = pow(rc, -reppow);
936 Vcut = -rc6*exp(-ewclj*ewclj*rc2)*(1 + ewclj*ewclj*rc2 + pow4(ewclj)*rc2*rc2/2);
946 case etabEwaldSwitch:
947 Vtab = gmx_erfc(ewc*rc)/rc;
950 /* Only calculate minus the reciprocal space contribution */
951 Vtab = -gmx_erf(ewc*rc)/rc;
955 /* No need for preventing the usage of modifiers with RF */
962 gmx_fatal(FARGS, "Cannot apply new potential-shift modifier to interaction type '%s' yet. (%s,%d)",
963 tprops[tp].name, __FILE__, __LINE__);
967 for (i = td->nx0; (i < td->nx); i++)
972 if (gmx_within_tol(reppow, 12.0, 10*GMX_DOUBLE_EPS))
978 r12 = pow(r, -reppow);
982 if (bPotentialSwitch)
984 /* swi is function, swi1 1st derivative and swi2 2nd derivative */
985 /* The switch function is 1 for r<r1, 0 for r>rc, and smooth for
986 * r1<=r<=rc. The 1st and 2nd derivatives are both zero at
988 * ksw is just the constant 1/(rc-r1)^5, to save some calculations...
1002 swi = 1 - 10*pow3(r-r1)*ksw*pow2(rc-r1)
1003 + 15*pow4(r-r1)*ksw*(rc-r1) - 6*pow5(r-r1)*ksw;
1004 swi1 = -30*pow2(r-r1)*ksw*pow2(rc-r1)
1005 + 60*pow3(r-r1)*ksw*(rc-r1) - 30*pow4(r-r1)*ksw;
1008 else /* not really needed, but avoids compiler warnings... */
1014 fprintf(fp, "%10g %10g %10g %10g\n", r, swi, swi1, swi2);
1018 rc6 = 1.0/(rc6*rc6);
1040 Ftab = reppow*Vtab/r;
1042 case etabLJ12Switch:
1048 Ftab = reppow*Vtab/r;
1054 Vtab = -(r6-6.0*(rc-r)*rc6/rc-rc6);
1055 Ftab = -(6.0*r6/r-6.0*rc6/rc);
1066 Vtab = -(r6-6.0*(rc-r)*rc6/rc-rc6);
1067 Ftab = -(6.0*r6/r-6.0*rc6/rc);
1079 case etabCOULSwitch:
1088 case etabEwaldSwitch:
1089 Vtab = gmx_erfc(ewc*r)/r;
1090 Ftab = gmx_erfc(ewc*r)/r2+exp(-(ewc*ewc*r2))*ewc*M_2_SQRTPI/r;
1093 case etabEwaldUserSwitch:
1094 /* Only calculate the negative of the reciprocal space contribution */
1095 Vtab = -gmx_erf(ewc*r)/r;
1096 Ftab = -gmx_erf(ewc*r)/r2+exp(-(ewc*ewc*r2))*ewc*M_2_SQRTPI/r;
1099 Vtab = -r6*exp(-ewclj*ewclj*r2)*(1 + ewclj*ewclj*r2 + pow4(ewclj)*r2*r2/2);
1100 Ftab = 6.0*Vtab/r - r6*exp(-ewclj*ewclj*r2)*pow5(ewclj)*ewclj*r2*r2*r;
1104 Vtab = 1.0/r + fr->k_rf*r2 - fr->c_rf;
1105 Ftab = 1.0/r2 - 2*fr->k_rf*r;
1106 if (tp == etabRF_ZERO && r >= rc)
1120 Vtab = 1.0/r-(rc-r)/(rc*rc)-1.0/rc;
1121 Ftab = 1.0/r2-1.0/(rc*rc);
1130 gmx_fatal(FARGS, "Table type %d not implemented yet. (%s,%d)",
1131 tp, __FILE__, __LINE__);
1135 /* Normal coulomb with cut-off correction for potential */
1139 /* If in Shifting range add something to it */
1142 r12 = (r-r1)*(r-r1);
1144 Vtab += -A_3*r13 - B_4*r12*r12;
1145 Ftab += A*r12 + B*r13;
1150 /* Make sure interactions are zero outside cutoff with modifiers */
1155 if (bPotentialShift)
1163 /* Make sure interactions are zero outside cutoff with modifiers */
1175 if (bPotentialSwitch)
1179 /* Make sure interactions are zero outside cutoff with modifiers */
1185 Ftab = Ftab*swi - Vtab*swi1;
1189 /* Convert to single precision when we store to mem */
1194 /* Continue the table linearly from nx0 to 0.
1195 * These values are only required for energy minimization with overlap or TPI.
1197 for (i = td->nx0-1; i >= 0; i--)
1199 td->v[i] = td->v[i+1] + td->f[i+1]*(td->x[i+1] - td->x[i]);
1200 td->f[i] = td->f[i+1];
1208 static void set_table_type(int tabsel[], const t_forcerec *fr, gmx_bool b14only)
1210 int eltype, vdwtype;
1212 /* Set the different table indices.
1219 switch (fr->eeltype)
1226 case eelPMEUSERSWITCH:
1235 eltype = fr->eeltype;
1241 tabsel[etiCOUL] = etabCOUL;
1244 tabsel[etiCOUL] = etabShift;
1247 if (fr->rcoulomb > fr->rcoulomb_switch)
1249 tabsel[etiCOUL] = etabShift;
1253 tabsel[etiCOUL] = etabCOUL;
1259 tabsel[etiCOUL] = etabEwald;
1262 tabsel[etiCOUL] = etabEwaldSwitch;
1265 tabsel[etiCOUL] = etabEwaldUser;
1267 case eelPMEUSERSWITCH:
1268 tabsel[etiCOUL] = etabEwaldUserSwitch;
1273 tabsel[etiCOUL] = etabRF;
1276 tabsel[etiCOUL] = etabRF_ZERO;
1279 tabsel[etiCOUL] = etabCOULSwitch;
1282 tabsel[etiCOUL] = etabUSER;
1285 tabsel[etiCOUL] = etabCOULEncad;
1288 gmx_fatal(FARGS, "Invalid eeltype %d", eltype);
1291 /* Van der Waals time */
1292 if (fr->bBHAM && !b14only)
1294 tabsel[etiLJ6] = etabLJ6;
1295 tabsel[etiLJ12] = etabEXPMIN;
1299 if (b14only && fr->vdwtype != evdwUSER)
1305 vdwtype = fr->vdwtype;
1311 tabsel[etiLJ6] = etabLJ6Switch;
1312 tabsel[etiLJ12] = etabLJ12Switch;
1315 tabsel[etiLJ6] = etabLJ6Shift;
1316 tabsel[etiLJ12] = etabLJ12Shift;
1319 tabsel[etiLJ6] = etabUSER;
1320 tabsel[etiLJ12] = etabUSER;
1323 tabsel[etiLJ6] = etabLJ6;
1324 tabsel[etiLJ12] = etabLJ12;
1326 case evdwENCADSHIFT:
1327 tabsel[etiLJ6] = etabLJ6Encad;
1328 tabsel[etiLJ12] = etabLJ12Encad;
1331 tabsel[etiLJ6] = etabLJ6Ewald;
1332 tabsel[etiLJ12] = etabLJ12;
1335 gmx_fatal(FARGS, "Invalid vdwtype %d in %s line %d", vdwtype,
1336 __FILE__, __LINE__);
1339 if (!b14only && fr->vdw_modifier != eintmodNONE)
1341 if (fr->vdw_modifier != eintmodPOTSHIFT &&
1342 fr->vdwtype != evdwCUT)
1344 gmx_incons("Potential modifiers other than potential-shift are only implemented for LJ cut-off");
1347 /* LJ-PME and other (shift-only) modifiers are handled by applying the modifiers
1348 * to the original interaction forms when we fill the table, so we only check cutoffs here.
1350 if (fr->vdwtype == evdwCUT)
1352 switch (fr->vdw_modifier)
1355 case eintmodPOTSHIFT:
1356 case eintmodEXACTCUTOFF:
1357 /* No modification */
1359 case eintmodPOTSWITCH:
1360 tabsel[etiLJ6] = etabLJ6Switch;
1361 tabsel[etiLJ12] = etabLJ12Switch;
1363 case eintmodFORCESWITCH:
1364 tabsel[etiLJ6] = etabLJ6Shift;
1365 tabsel[etiLJ12] = etabLJ12Shift;
1368 gmx_incons("Unsupported vdw_modifier");
1375 t_forcetable make_tables(FILE *out, const output_env_t oenv,
1376 const t_forcerec *fr,
1377 gmx_bool bVerbose, const char *fn,
1378 real rtab, int flags)
1380 const char *fns[3] = { "ctab.xvg", "dtab.xvg", "rtab.xvg" };
1381 const char *fns14[3] = { "ctab14.xvg", "dtab14.xvg", "rtab14.xvg" };
1384 gmx_bool b14only, bReadTab, bGenTab;
1386 int i, j, k, nx, nx0, tabsel[etiNR];
1391 b14only = (flags & GMX_MAKETABLES_14ONLY);
1393 if (flags & GMX_MAKETABLES_FORCEUSER)
1395 tabsel[etiCOUL] = etabUSER;
1396 tabsel[etiLJ6] = etabUSER;
1397 tabsel[etiLJ12] = etabUSER;
1401 set_table_type(tabsel, fr, b14only);
1407 table.scale_exp = 0;
1411 table.interaction = GMX_TABLE_INTERACTION_ELEC_VDWREP_VDWDISP;
1412 table.format = GMX_TABLE_FORMAT_CUBICSPLINE_YFGH;
1413 table.formatsize = 4;
1414 table.ninteractions = 3;
1415 table.stride = table.formatsize*table.ninteractions;
1417 /* Check whether we have to read or generate */
1420 for (i = 0; (i < etiNR); i++)
1422 if (ETAB_USER(tabsel[i]))
1426 if (tabsel[i] != etabUSER)
1433 read_tables(out, fn, etiNR, 0, td);
1434 if (rtab == 0 || (flags & GMX_MAKETABLES_14ONLY))
1436 rtab = td[0].x[td[0].nx-1];
1442 if (td[0].x[td[0].nx-1] < rtab)
1444 gmx_fatal(FARGS, "Tables in file %s not long enough for cut-off:\n"
1445 "\tshould be at least %f nm\n", fn, rtab);
1447 nx = table.n = (int)(rtab*td[0].tabscale + 0.5);
1449 table.scale = td[0].tabscale;
1457 table.scale = 2000.0;
1459 table.scale = 500.0;
1461 nx = table.n = rtab*table.scale;
1466 if (fr->bham_b_max != 0)
1468 table.scale_exp = table.scale/fr->bham_b_max;
1472 table.scale_exp = table.scale;
1476 /* Each table type (e.g. coul,lj6,lj12) requires four
1477 * numbers per nx+1 data points. For performance reasons we want
1478 * the table data to be aligned to 16-byte.
1480 snew_aligned(table.data, 12*(nx+1)*sizeof(real), 32);
1482 for (k = 0; (k < etiNR); k++)
1484 if (tabsel[k] != etabUSER)
1487 (tabsel[k] == etabEXPMIN) ? table.scale_exp : table.scale,
1488 &(td[k]), !bReadTab);
1489 fill_table(&(td[k]), tabsel[k], fr, b14only);
1492 fprintf(out, "%s table with %d data points for %s%s.\n"
1493 "Tabscale = %g points/nm\n",
1494 ETAB_USER(tabsel[k]) ? "Modified" : "Generated",
1495 td[k].nx, b14only ? "1-4 " : "", tprops[tabsel[k]].name,
1500 /* Set scalefactor for c6/c12 tables. This is because we save flops in the non-table kernels
1501 * by including the derivative constants (6.0 or 12.0) in the parameters, since
1502 * we no longer calculate force in most steps. This means the c6/c12 parameters
1503 * have been scaled up, so we need to scale down the table interactions too.
1504 * It comes here since we need to scale user tables too.
1508 scalefactor = 1.0/6.0;
1510 else if (k == etiLJ12 && tabsel[k] != etabEXPMIN)
1512 scalefactor = 1.0/12.0;
1519 copy2table(table.n, k*4, 12, td[k].x, td[k].v, td[k].f, scalefactor, table.data);
1521 if (bDebugMode() && bVerbose)
1525 fp = xvgropen(fns14[k], fns14[k], "r", "V", oenv);
1529 fp = xvgropen(fns[k], fns[k], "r", "V", oenv);
1531 /* plot the output 5 times denser than the table data */
1532 for (i = 5*((nx0+1)/2); i < 5*table.n; i++)
1534 x0 = i*table.r/(5*(table.n-1));
1535 evaluate_table(table.data, 4*k, 12, table.scale, x0, &y0, &yp);
1536 fprintf(fp, "%15.10e %15.10e %15.10e\n", x0, y0, yp);
1540 done_tabledata(&(td[k]));
1547 t_forcetable make_gb_table(const output_env_t oenv,
1548 const t_forcerec *fr)
1550 const char *fns[3] = { "gbctab.xvg", "gbdtab.xvg", "gbrtab.xvg" };
1551 const char *fns14[3] = { "gbctab14.xvg", "gbdtab14.xvg", "gbrtab14.xvg" };
1554 gmx_bool bReadTab, bGenTab;
1556 int i, j, k, nx, nx0, tabsel[etiNR];
1557 double r, r2, Vtab, Ftab, expterm;
1561 double abs_error_r, abs_error_r2;
1562 double rel_error_r, rel_error_r2;
1563 double rel_error_r_old = 0, rel_error_r2_old = 0;
1564 double x0_r_error, x0_r2_error;
1567 /* Only set a Coulomb table for GB */
1574 /* Set the table dimensions for GB, not really necessary to
1575 * use etiNR (since we only have one table, but ...)
1578 table.interaction = GMX_TABLE_INTERACTION_ELEC;
1579 table.format = GMX_TABLE_FORMAT_CUBICSPLINE_YFGH;
1580 table.r = fr->gbtabr;
1581 table.scale = fr->gbtabscale;
1582 table.scale_exp = 0;
1583 table.n = table.scale*table.r;
1584 table.formatsize = 4;
1585 table.ninteractions = 1;
1586 table.stride = table.formatsize*table.ninteractions;
1588 nx = table.scale*table.r;
1590 /* Check whether we have to read or generate
1591 * We will always generate a table, so remove the read code
1592 * (Compare with original make_table function
1597 /* Each table type (e.g. coul,lj6,lj12) requires four
1598 * numbers per datapoint. For performance reasons we want
1599 * the table data to be aligned to 16-byte. This is accomplished
1600 * by allocating 16 bytes extra to a temporary pointer, and then
1601 * calculating an aligned pointer. This new pointer must not be
1602 * used in a free() call, but thankfully we're sloppy enough not
1606 snew_aligned(table.data, 4*nx, 32);
1608 init_table(nx, nx0, table.scale, &(td[0]), !bReadTab);
1610 /* Local implementation so we don't have to use the etabGB
1611 * enum above, which will cause problems later when
1612 * making the other tables (right now even though we are using
1613 * GB, the normal Coulomb tables will be created, but this
1614 * will cause a problem since fr->eeltype==etabGB which will not
1615 * be defined in fill_table and set_table_type
1618 for (i = nx0; i < nx; i++)
1622 expterm = exp(-0.25*r2);
1624 Vtab = 1/sqrt(r2+expterm);
1625 Ftab = (r-0.25*r*expterm)/((r2+expterm)*sqrt(r2+expterm));
1627 /* Convert to single precision when we store to mem */
1633 copy2table(table.n, 0, 4, td[0].x, td[0].v, td[0].f, 1.0, table.data);
1637 fp = xvgropen(fns[0], fns[0], "r", "V", oenv);
1638 /* plot the output 5 times denser than the table data */
1639 /* for(i=5*nx0;i<5*table.n;i++) */
1640 for (i = nx0; i < table.n; i++)
1642 /* x0=i*table.r/(5*table.n); */
1643 x0 = i*table.r/table.n;
1644 evaluate_table(table.data, 0, 4, table.scale, x0, &y0, &yp);
1645 fprintf(fp, "%15.10e %15.10e %15.10e\n", x0, y0, yp);
1652 for(i=100*nx0;i<99.81*table.n;i++)
1654 r = i*table.r/(100*table.n);
1656 expterm = exp(-0.25*r2);
1658 Vtab = 1/sqrt(r2+expterm);
1659 Ftab = (r-0.25*r*expterm)/((r2+expterm)*sqrt(r2+expterm));
1662 evaluate_table(table.data,0,4,table.scale,r,&y0,&yp);
1663 printf("gb: i=%d, x0=%g, y0=%15.15f, Vtab=%15.15f, yp=%15.15f, Ftab=%15.15f\n",i,r, y0, Vtab, yp, Ftab);
1665 abs_error_r=fabs(y0-Vtab);
1666 abs_error_r2=fabs(yp-(-1)*Ftab);
1668 rel_error_r=abs_error_r/y0;
1669 rel_error_r2=fabs(abs_error_r2/yp);
1672 if(rel_error_r>rel_error_r_old)
1674 rel_error_r_old=rel_error_r;
1678 if(rel_error_r2>rel_error_r2_old)
1680 rel_error_r2_old=rel_error_r2;
1685 printf("gb: MAX REL ERROR IN R=%15.15f, MAX REL ERROR IN R2=%15.15f\n",rel_error_r_old, rel_error_r2_old);
1686 printf("gb: XO_R=%g, X0_R2=%g\n",x0_r_error, x0_r2_error);
1689 done_tabledata(&(td[0]));
1697 t_forcetable make_atf_table(FILE *out, const output_env_t oenv,
1698 const t_forcerec *fr,
1702 const char *fns[3] = { "tf_tab.xvg", "atfdtab.xvg", "atfrtab.xvg" };
1705 real x0, y0, yp, rtab;
1707 real rx, ry, rz, box_r;
1712 /* Set the table dimensions for ATF, not really necessary to
1713 * use etiNR (since we only have one table, but ...)
1717 if (fr->adress_type == eAdressSphere)
1719 /* take half box diagonal direction as tab range */
1720 rx = 0.5*box[0][0]+0.5*box[1][0]+0.5*box[2][0];
1721 ry = 0.5*box[0][1]+0.5*box[1][1]+0.5*box[2][1];
1722 rz = 0.5*box[0][2]+0.5*box[1][2]+0.5*box[2][2];
1723 box_r = sqrt(rx*rx+ry*ry+rz*rz);
1728 /* xsplit: take half box x direction as tab range */
1729 box_r = box[0][0]/2;
1734 table.scale_exp = 0;
1738 read_tables(out, fn, 1, 0, td);
1739 rtab = td[0].x[td[0].nx-1];
1741 if (fr->adress_type == eAdressXSplit && (rtab < box[0][0]/2))
1743 gmx_fatal(FARGS, "AdResS full box therm force table in file %s extends to %f:\n"
1744 "\tshould extend to at least half the length of the box in x-direction"
1745 "%f\n", fn, rtab, box[0][0]/2);
1749 gmx_fatal(FARGS, "AdResS full box therm force table in file %s extends to %f:\n"
1750 "\tshould extend to at least for spherical adress"
1751 "%f (=distance from center to furthermost point in box \n", fn, rtab, box_r);
1757 table.scale = td[0].tabscale;
1760 /* Each table type (e.g. coul,lj6,lj12) requires four
1761 * numbers per datapoint. For performance reasons we want
1762 * the table data to be aligned to 16-byte. This is accomplished
1763 * by allocating 16 bytes extra to a temporary pointer, and then
1764 * calculating an aligned pointer. This new pointer must not be
1765 * used in a free() call, but thankfully we're sloppy enough not
1769 snew_aligned(table.data, 4*nx, 32);
1771 copy2table(table.n, 0, 4, td[0].x, td[0].v, td[0].f, 1.0, table.data);
1775 fp = xvgropen(fns[0], fns[0], "r", "V", oenv);
1776 /* plot the output 5 times denser than the table data */
1777 /* for(i=5*nx0;i<5*table.n;i++) */
1779 for (i = 5*((nx0+1)/2); i < 5*table.n; i++)
1781 /* x0=i*table.r/(5*table.n); */
1782 x0 = i*table.r/(5*(table.n-1));
1783 evaluate_table(table.data, 0, 4, table.scale, x0, &y0, &yp);
1784 fprintf(fp, "%15.10e %15.10e %15.10e\n", x0, y0, yp);
1790 done_tabledata(&(td[0]));
1793 table.interaction = GMX_TABLE_INTERACTION_ELEC_VDWREP_VDWDISP;
1794 table.format = GMX_TABLE_FORMAT_CUBICSPLINE_YFGH;
1795 table.formatsize = 4;
1796 table.ninteractions = 3;
1797 table.stride = table.formatsize*table.ninteractions;
1803 bondedtable_t make_bonded_table(FILE *fplog, char *fn, int angle)
1818 read_tables(fplog, fn, 1, angle, &td);
1821 /* Convert the table from degrees to radians */
1822 for (i = 0; i < td.nx; i++)
1827 td.tabscale *= RAD2DEG;
1830 tab.scale = td.tabscale;
1831 snew(tab.data, tab.n*4);
1832 copy2table(tab.n, 0, 4, td.x, td.v, td.f, 1.0, tab.data);
1833 done_tabledata(&td);