[molpro-user] Strange limit on RASSCF sizes - uncovered bug?
Dr Seth OLSEN
s.olsen1 at uq.edu.au
Wed Jun 22 11:21:43 BST 2005
Hi Molpro-users,
I have been trying how much I can get into an RASSCF calculation with
MolPro. My system is 126 electrons on 26 atoms. I am interested in the
excited state surface, so I have been doing 2-state averaged calculations.
I have changed the mxshel parameter in cconf and ccconf1 to 25 and
mrslr=25 in cref. So far so good. With this compile I was able to get
up to an RASSCF(8(mxhole=2),2,5(mxparticle=2)). That is RASSCF(18,15)
which includes a core with only 2 excitations out of orbitals below
'LUMO' and only 2 excitations into orbitals above 'HOMO'. This seems to
work fine and the resulting energies & natural orbitals look good and
are quite informative. However, when I attempt to increase the number
of virtuals in this calculation to RASSCF(18,16) (with the identical
restrictions below/above the HOMO/LUMO pair), the job crashes with an
error. The error seems to indicate a programming bug, and I have not
seen the parameter referred to before (see below). It seems from the
multi source code to be connected with the p-space threshold, but I am
not certain. I have attatched the full output. Below is the output of
MULTI:
***
1PROGRAM * MULTI (Direct Multiconfiguration SCF) Authors: P.J.
Knowles, H.-J. Werner (1984) S.T. Elbert (1988)
Number of closed-shell orbitals: 54 ( 54 )
Number of active orbitals: 16 ( 16 )
Number of external orbitals: 198 (198 )
State symmetry 1
Orbital occupancy restrictions for symmetry 1
0 2 0000000000222222
14 16 2222222200000000
Reference space for state symmetry 1
------------------------------------
Sym Shell Conf. CSFs
1 1 1 - 112 1 - 112
1 2 113 - 1096 113 - 1096
1 3 1097 - 2848 1097 - 4600
1 4 2849 - 3268 4601 - 6700
Number of electrons: 18 Spin symmetry=Singlet Space symmetry=1
Number of states: 2
Number of CSFs: 6700
Storage for configuration lists 69005, coupling coefficients 25887,
address lists 501608
Molecular orbitals read from record 2140.2 Type=MATROP/MATROP
Wavefunction dump at record 2141.2
Convergence thresholds 0.10E-01 (gradient) 0.10E-05 (energy)
0.00E+00 (step length)
Weight factors for state symmetry 1: 0.50000 0.50000
Eigenvalues of redundancy matrix
1- 10 -0.000000 -0.000000 -0.000000 -0.000000 -0.000000
-0.000000 -0.000000 -0.000000 -0.000000 -0.000000
11- 20 -0.000000 -0.000000 -0.000000 -0.000000 -0.000000
-0.000000 -0.000000 -0.000000 -0.000000 -0.000000
21- 30 -0.000000 -0.000000 -0.000000 0.000000 0.000000
0.000000 0.000000 0.000000 0.000000 0.000000
31- 40 0.000000 0.000000 0.000000 0.000000 0.000000
0.000000 0.000000 0.000000 0.000000 0.000000
41- 50 0.000000 0.000000 0.000000 0.000000 0.245680
0.339661 0.354046 0.580233 0.598297 0.598680
51- 60 0.600180 0.602238 0.628619 0.651691 0.652339
0.653234 0.654234 0.655112 0.657345 0.692562
61- 70 0.693732 0.694992 0.696263 0.697557 0.700114
0.755021 0.850798 0.873288 0.989323 0.994008
71- 80 0.995801 0.997020 1.008652 1.044258 1.048045
1.052341 1.056994 1.061901 1.069910 1.205660
81- 90 1.206340 1.208259 1.214000 1.258507 1.398027
1.441443 1.456479 1.468920 1.471006 1.474704
91-100 1.479232 1.485733 1.486020 1.492871 1.493966
1.496204 1.767194 1.798176 1.807558 1.820422
101-110 1.827833 1.829082 1.836017 1.836184 1.840252
1.845778 1.862532 1.863722 1.869921 1.871433
111-120 1.879861 1.886974 1.889899 1.894577 1.894940
1.902603 1.903470 1.911190 1.913630 1.921397
Non-redundant active-active orbital rotations:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
2
3
4
5
6
7
8
9 + + + + + + + +
10 + + + + + + + +
11 + + + + + + + + + +
12 + + + + + + + + + +
13 + + + + + + + + + +
14 + + + + + + + + + +
15 + + + + + + + + + +
16 + + + + + + + + + +
Number of orbital rotations 14800 ( 864 Core/Active10692
Core/Virtual 76 Active/Active 3168 Active/Virtual)
Total number of variables = 28200
ITER. MIC NCI NEG ENERGY(VAR) ENERGY(PROJ) ENERGY CHANGE
GRAD(0) GRAD(ORB) GRAD(CI) STEP TIME
?PROGRAMMING ERROR: UNKNOWN PARAMETER THRPSP IN INPUT SET THRESH FOR
PROGRAM MULTI
KNOWN PARAMETERS:
? Error
? Unknown parameter
? The problem occurs in get_inpf
ERROR EXIT
CURRENT STACK: MAIN
*****
The parameter this error refers to, as above, seems connected to the
p-space threshold, but I am not certain. I have not found the code line
that invokes the error yet, so I am not clear on what condition has been
made true to invoke the error. In the relevant input I have not set the
p-space threshold, nor does it seem (from other tests) that the
threshold has any effect. Presumably, these conditions would result in
other, more familiar errors anyway. Is this a bug in the program? Is
there another parameter which I must alter in order to make it work?
Has my messing with the common variables mentioned above altered the
memory allocation scheme etc. in some fundamentally wrong way? Should I
increase maxatoms? I am running on a Sunfire v880 8cpu SMP enclosure.
The memory requirements should be well within my limits. Apart from the
parameters above, it is a standard tcgmsg parallel molpro install.
I will continue probing this to see what I can do to bring it around,
and will post to the list if I can work it out. However, if there is
someone on the list who is more familiar with the programming details of
molpro can lend some insight, that would be quite cool. It seems that
molpro was written by more savvy coders than myself.
Cheers,
Seth
P.S. I apologize for the brusqueness of yesterdays post: re the
geometry optimization error. That was out of line & due to late-night
frayed nerves. If the moderator wishes to strike it from the list, I
would not at all mind. Upon review it (the error) seems to have come up
before.
ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
Dr Seth Olsen, PhD
Postdoctoral Fellow, Computational Systems Biology Group
Centre for Computational Molecular Science
Chemistry Building,
The University of Queensland
Qld 4072, Brisbane, Australia
tel (617) 33653732
fax (617) 33654623
email: s.olsen1 at uq.edu.au
Web: www.ccms.uq.edu.au
ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
-------------- next part --------------
A non-text attachment was scrubbed...
Name: m0car1816s2s.out
Type: application/octet-stream
Size: 52090 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20050622/91c31dd7/attachment.obj>
More information about the Molpro-user
mailing list