2
* Revision 1.10 2004/05/03 04:32:40 crawdad
3
* Major mods based on merge with stable psi-3-2-1 release. Note that this
4
* version has not been fully tested and some scf-optn test cases do not run
5
* correctly beccause of changes in mid-March 2004 to optking.
8
/* Revision 1.9.4.1 2004/04/10 19:41:32 crawdad
9
/* Fixed the DIIS code for UHF cases. The new version uses the Pulay scheme of
10
/* building the error vector in the AO basis as FDS-SDF, followed by xformation
11
/* into the orthogonal AO basis. This code converges faster for test cases
12
/* like cc8, but fails for linearly dependent basis sets for unknown reasons.
15
/* Revision 1.9 2003/04/14 17:25:47 sherrill
16
/* Change "total energy" to "SCF total energy" to make more explicit for
17
/* new users. Yeah, this will probably break some test case perl scripts
20
/* Revision 1.8 2001/01/04 14:13:35 sbrown
21
/* Fixed the problem with iconv: The new versions of linux had iconv already
22
/* assigned to something else so I changed all references of it to scf_conv.
24
/* Revision 1.7 2000/12/05 19:40:03 sbrown
25
/* Added Unrestricted Kohn-Sham DFT.
27
/* Revision 1.6 2000/09/02 20:48:51 evaleev
28
/* Print out one- and two-electron energies every iteration if iprint&2 .
30
/* Revision 1.5 2000/08/23 17:15:16 sbrown
31
/* Added portions to separate out the correlation and exchange energy at the
32
/* end the calculation as well as do the consistency check on the integrated
35
/* Revision 1.4 2000/06/26 19:04:09 sbrown
36
/* Added DFT capapbilities to interface with cints using direct scf
38
/* Revision 1.3 2000/06/22 22:15:00 evaleev
39
/* Modifications for KS DFT. Reading in XC Fock matrices and XC energy in formg_direct need to be uncommented (at present those are not produced by CINTS yet).
41
/* Revision 1.2 2000/06/02 13:32:16 kenny
44
/* Added dynamic integral accuracy cutoffs for direct scf. Added a few global
45
/* variables. Added keyword 'dyn_acc'; true--use dynamic cutoffs. Use of
46
/* 'dconv' and 'delta' to keep track of density convergence somewhat awkward,
47
/* but avoids problems when accuracy is switched and we have to wipe out density
48
/* matrices. Also added error message and exit if direct rohf singlet is
49
/* attempted since it doesn't work.
52
/* Revision 1.1.1.1 2000/02/04 22:52:30 evaleev
53
/* Started PSI 3 repository
55
/* Revision 1.3 1999/11/02 23:55:56 localpsi
56
/* Shawn Brown - (11/2/99) Modified to the code in a few major ways.
58
/* 1. Added the capability to do UHF. All of the features available with the
59
/* other refrences have been added for UHF.
61
/* 2. For UHF, I had to alter the structure of file30. (See cleanup.c for a
62
/* map) This entailed adding a pointer array right after the header in the SCF
63
/* section of file30 that pointed to all of the data for the SCF caclulation.
64
/* Functions were added to libfile30 to account for this and they are
65
/* incorporated in this code.
67
/* 3. Updated and fixed all of the problems associated with my previous
68
/* guessing code. The code no longer uses OPENTYPE to specify the type of
69
/* occupation. The keword REFERENCE and MULTP can now be used to indicate any
70
/* type of calculation. (e.g. ROHF with MULTP of 1 is an open shell singlet
71
/* ROHF calculation) This code was moved to occ_fun.c. The code can also
72
/* guess at any multplicity in a highspin case, provided enough electrons.
74
/* Revision 1.2 1999/08/17 19:04:14 evaleev
75
/* Changed the default symmetric orthogonalization to the canonical
76
/* orthogonalization. Now, if near-linear dependencies in the basis are found,
77
/* eigenvectors of the overlap matrix with eigenvalues less than 1E-6 will be
78
/* left out. This will lead to num_mo != num_so, i.e. SCF eigenvector is no
79
/* longer a square matrix. Had to rework some routines in libfile30, and add some.
80
/* The progrem prints out a warning if near-linear dependencies are found. TRANSQT
81
/* and a whole bunch of other codes has to be fixed to work with such basis sets.
83
/* Revision 1.1.1.1 1999/04/12 16:59:26 evaleev
84
/* Added a version of CSCF that can work with CINTS.
88
static char *rcsid = "$Id: ecalc.c 2455 2004-05-03 04:32:41Z crawdad $";
94
static double twocut=1.0;
95
static double eelec; /*--- elec. energy from the previous iteration ---*/
104
double plimit = pow(10.0,(double) -scf_conv);
106
double oe_energy, te_energy, dtmp, dtmp1;
111
oe_energy = te_energy = 0.0;
112
for (k=0; k < num_ir ; k++) {
116
for (i=ij=0; i < nn ; i++) {
117
for (j = 0 ; j <= i ; j++,ij++) {
118
oe_energy += 0.5*s->pmat[ij]*s->hmat[ij];
120
te_energy += 0.5*((spin_info[0].scf_spin[k].pmat[ij]
121
*spin_info[0].scf_spin[k].fock_pac[ij])
122
+(spin_info[1].scf_spin[k].pmat[ij]
123
*spin_info[1].scf_spin[k].fock_pac[ij]));
126
te_energy += 0.5*s->pmat[ij]*s->fock_pac[ij];
129
te_energy += 0.5*s->pmat[ij]*s->fock_pac[ij]
130
- 0.5*s->pmato[ij]*s->gmato[ij];
136
for (i = 0; i < ioff[nn] ; i++) {
137
dtmp = spin_info[0].scf_spin[k].dpmat[i];
138
dtmp1 = spin_info[1].scf_spin[k].dpmat[i];
140
delta += dtmp1*dtmp1;
144
for (i = 0; i < ioff[nn] ; i++) {
152
neelec = oe_energy + te_energy;
154
/*JPK(6/1/00) dynamic integral accuracy modifications*/
155
dconv = sqrt(delta)/mxcoef2;
157
if(acc_switch==1 || iter==0) {
161
coulomb_energy = neelec;
164
/*printf("XC_energy = %10.10lf",exc);*/
167
etot = repnuc + neelec;
168
edif = eelec - neelec;
171
if (!iter) fprintf(outfile,"\n iter total energy delta E delta P diiser\n");
172
fprintf(outfile, "%5d %20.10f %15.6e %15.6e %15.6e\n",
173
++iter, etot, edif, dconv, diiser);
175
fprintf(outfile, "one-electron energy = %25.15f\n", oe_energy);
176
fprintf(outfile, "two-electron energy = %25.15f\n", te_energy);
177
fprintf(outfile, "coulomb energy = %25.15f\n",coulomb_energy);
178
fprintf(outfile, "SCF total energy = %25.15f\n", etot);
183
if ( delta < plimit && iter > 1) {
185
if(!iopen || iopen && fock_typ >= 2) cleanup();
190
cinext = pow(10.0,-twocut);
191
if (delta < cinext && delta && !converged) {
added
removed
src/bin/cscf/ecalc.c
