[molpro-user] Converging CPL equations
Taube, Andrew
Andrew.Taube at DEShawResearch.com
Tue Jul 13 20:55:26 BST 2010
Hi,
In a geometry optimization using df-lmp2, I am running into a problem converging the coupled-perturbed localization equations. After a successful iteration (or several successful iterations, depending on system) the calculation hits the CPL part of the code and diverges. This actually doesn't yield an error (which it probably should) but does lead to the CPHF diverging as well (which does yield an error). Strangely, restarting the calculation at the same geometry leads to convergence -- but only for one geometry step, after which the calculation breaks down again.
Here is the input:
memory,481,M
thrdscf=1e-12,thrdscf_max=1e-9
gthresh,throvl=1e-9,throrth=1e-7,optstep=1e-4,optgrad=1e-4
gdirect, thrdscf=1e-13, thrdscf_max=1e-10
basis={
h=avtz
o=avtz
xe=aug-cc-pvtz-pp
}
geomtyp=xyz
symmetry,nosym
orient,noorient
geometry={
4
structure 1
O1 -0.7311853248 -0.9329885132 2.0323859062
H2 -0.0551706907 -0.2638753144 2.1780326567
H3 -0.2935012919 -1.5932928363 1.4864318358
XE4 0.8906235318 2.1608654092 -0.1147280126
}
core,1
set,charge=0
df-hf,accu=15,maxit=100
df-lmp2;local,thrbp=1,mergedom=1;pipek,delete=2,method=3
optg,maxit=50,method=slnr
put,xyz
and here is the relevant section of output:
Solving coupled perturbed localization equations
Number of orbital parameters: 136
Norm of right-hand side: 0.1592D-06
Deleting in PM localization the last 2 functions of each type
Gradient of localization functional: 0.2834D-07
WARNING: SMALL DIAGONAL ELEMENT OF CPL HESSIAN FOR ROTATION 16.1 - 9.1: 0.93D-08 GRAD= 0.11D-07 (ROTATION ELIMINATED)
Solving CPL equations. Requested accuracy= 0.10D-09
ITER GRAD VAR
1 0.28D-06 0.14D+01
2 0.44D-06 0.18D+01
3 0.15D-06 0.27D+01
4 0.54D-06 0.63D+01
5 0.56D-06 0.85D+01
6 0.16D-05 0.21D+02
7 0.19D-05 0.28D+02
8 0.53D-05 0.65D+02
9 0.67D-05 0.91D+02
10 0.18D-04 0.21D+03
11 0.23D-04 0.30D+03
12 0.57D-04 0.66D+03
13 0.72D-04 0.10D+04
14 0.17D-03 0.21D+04
15 0.22D-03 0.34D+04
16 0.51D-03 0.69D+04
17 0.68D-03 0.12D+05
18 0.15D-02 0.23D+05
19 0.22D-02 0.41D+05
20 0.46D-02 0.77D+05
21 0.74D-02 0.14D+06
22 0.15D-01 0.26D+06
23 0.25D-01 0.49D+06
24 0.49D-01 0.90D+06
25 0.88D-01 0.17D+07
26 0.16D+00 0.31D+07
27 0.31D+00 0.58D+07
28 0.56D+00 0.11D+08
29 0.11D+01 0.20D+08
30 0.19D+01 0.37D+08
31 0.36D+01 0.68D+08
32 0.66D+01 0.13D+09
33 0.12D+02 0.24D+09
34 0.23D+02 0.43D+09
35 0.43D+02 0.81D+09
36 0.79D+02 0.15D+10
37 0.15D+03 0.28D+10
38 0.27D+03 0.52D+10
39 0.51D+03 0.96D+10
40 0.93D+03 0.18D+11
41 0.17D+04 0.33D+11
42 0.32D+04 0.61D+11
43 0.60D+04 0.11D+12
44 0.11D+05 0.21D+12
45 0.21D+05 0.39D+12
46 0.38D+05 0.72D+12
47 0.71D+05 0.13D+13
48 0.13D+06 0.25D+13
49 0.24D+06 0.46D+13
50 0.45D+06 0.86D+13
51 0.84D+06 0.16D+14
52 0.16D+07 0.29D+14
53 0.29D+07 0.55D+14
54 0.54D+07 0.10D+15
55 0.99D+07 0.19D+15
56 0.18D+08 0.35D+15
57 0.34D+08 0.65D+15
58 0.63D+08 0.12D+16
59 0.12D+09 0.22D+16
60 0.22D+09 0.41D+16
61 0.40D+09 0.77D+16
62 0.75D+09 0.14D+17
63 0.14D+10 0.26D+17
64 0.26D+10 0.49D+17
65 0.48D+10 0.91D+17
66 0.89D+10 0.17D+18
67 0.16D+11 0.31D+18
68 0.31D+11 0.58D+18
69 0.57D+11 0.11D+19
70 0.11D+12 0.20D+19
71 0.20D+12 0.37D+19
72 0.36D+12 0.69D+19
73 0.67D+12 0.13D+20
74 0.12D+13 0.24D+20
75 0.23D+13 0.44D+20
76 0.43D+13 0.81D+20
77 0.79D+13 0.15D+21
78 0.15D+14 0.28D+21
79 0.27D+14 0.52D+21
80 0.51D+14 0.96D+21
81 0.94D+14 0.18D+22
82 0.17D+15 0.33D+22
83 0.32D+15 0.61D+22
84 0.60D+15 0.11D+23
85 0.11D+16 0.21D+23
86 0.21D+16 0.39D+23
87 0.38D+16 0.73D+23
88 0.71D+16 0.13D+24
89 0.13D+17 0.25D+24
90 0.24D+17 0.46D+24
91 0.45D+17 0.86D+24
92 0.84D+17 0.16D+25
93 0.16D+18 0.30D+25
94 0.29D+18 0.55D+25
95 0.54D+18 0.10D+26
96 0.10D+19 0.19D+26
97 0.18D+19 0.35D+26
98 0.34D+19 0.65D+26
99 0.64D+19 0.12D+27
100 0.12D+20 0.22D+27
Time for CPL C-matrix: 0.00 sec
Time for CPL solve: 0.00 sec
Time for CPL (total): 0.00 sec
Thresholds used in cphf: THRAO= 1.0D-10, THRMO= 1.0D-10, THROV= 1.0D-12, THRSW= 1.0D-10, THRPROD= 1.0D-10
Starting Coupled-perturbed Hartree-Fock, convergence threshold= 1.0D-06, integral threshold= 1.0D-13
ITER MIC DIIS VAR VARC CPU
1 0 1 0.28D+22 0.81D+17 6.98
2 0 2 0.64D+21 0.25D+17 8.10
3 0 3 0.20D+21 0.27D+17 9.21
4 0 4 0.53D+20 0.20D+17 10.31
5 0 5 0.91D+19 0.46D+16 11.36
6 0 6 0.20D+19 0.18D+16 12.41
7 0 6 0.66D+18 0.11D+16 13.46
8 0 6 0.19D+18 0.54D+15 14.51
9 0 6 0.36D+17 0.25D+15 15.56
10 0 6 0.16D+17 0.67D+14 16.60
11 0 6 0.38D+16 0.56D+13 17.63
12 0 6 0.75D+15 0.18D+13 18.65
13 0 6 0.25D+15 0.73D+12 19.68
14 0 6 0.94D+14 0.28D+12 20.70
15 0 6 0.36D+14 0.93D+11 21.73
16 0 6 0.79D+13 0.83D+10 22.78
17 0 6 0.15D+13 0.37D+10 23.81
18 0 6 0.46D+12 0.22D+10 24.85
19 0 6 0.18D+12 0.11D+10 25.89
20 0 6 0.62D+11 0.13D+09 26.94
21 0 6 0.11D+11 0.21D+08 27.98
22 0 6 0.39D+10 0.85D+07 29.00
23 0 6 0.33D+10 0.65D+07 30.04
24 0 6 0.18D+10 0.34D+07 31.06
25 0 6 0.11D+10 0.25D+07 32.10
26 0 6 0.72D+09 0.20D+07 33.13
27 0 6 0.70D+09 0.20D+07 34.16
28 0 6 0.63D+09 0.15D+07 35.19
29 0 6 0.60D+09 0.15D+07 36.23
30 0 6 0.56D+09 0.13D+07 37.27
31 0 6 0.49D+09 0.12D+07 38.30
32 0 6 0.35D+09 0.91D+06 39.34
33 0 6 0.25D+09 0.78D+06 40.36
34 0 6 0.22D+09 0.79D+06 41.40
35 0 6 0.20D+09 0.67D+06 42.43
36 0 6 0.20D+09 0.63D+06 43.46
37 0 6 0.20D+09 0.67D+06 44.51
38 0 6 0.20D+09 0.66D+06 45.55
39 0 6 0.19D+09 0.62D+06 46.59
40 0 6 0.20D+09 0.65D+06 47.62
41 0 6 0.20D+09 0.68D+06 48.65
42 0 6 0.20D+09 0.65D+06 49.68
43 0 6 0.19D+09 0.65D+06 50.71
44 0 6 0.20D+09 0.67D+06 51.75
45 0 6 0.20D+09 0.66D+06 52.78
46 0 6 0.20D+09 0.68D+06 53.81
47 0 6 0.20D+09 0.67D+06 54.85
48 0 6 0.20D+09 0.64D+06 55.89
49 0 6 0.20D+09 0.64D+06 56.92
No convergence of CPHF after 50 macroiterations and 0 microiterations. Accuracy: 1.9898D+08
?ERROR: NO CONVERGENCE. EXIT FROM CPHF THIS CAN BE AVOIDED USING THE NOCHECK OPTION.
Any help would be appreciated,
Thanks,
Andrew
---
Andrew G. Taube
D. E. Shaw Research
email: Andrew.Taube at DEShawResearch.com
Tel: +1-212-478-0118
Fax: +1-212-845-1118
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20100713/f0c908f1/attachment.html>
More information about the Molpro-user
mailing list