Molcas Forum

Support and discussions for Molcas and OpenMolcas users and developers

You are not logged in.

Announcement

Welcome to the Molcas forum.

Please note: The forum's URL has changed. The new URL is: https://molcasforum.univie.ac.at. Please update your bookmarks!

You can choose an avatar and change the default style by going to "Profile" → "Personality" or "Display".

#1 2022-05-13 22:34:06

laumot
Member
Registered: 2019-07-26
Posts: 17

DMRG - MPSSI bug

Dear OpenMolcas users,

I am hoping I can get some help with what appears to be a bug in the MPSSI module in the latest OpenMolcas release.

Briefly, we are trying to calculate some properties (EPR and HFC) using DMRG and MPSSI with multiple states of different multiplicities. However, the job always crashes with a "Non-zero return code" error. The code passes all the qcmaquis tests, including 007, and it appears the problem is only triggered when we input multiple excited states into MPSSI.

This problem was not present in the OpenMolcas v18.09 release. It appears to start occurring from v20.10 up. Unfortunately, we cannot use v18.09 because it does not have all features we need.

I have attached below the standard 007 qcmaquis test output (which finishes) and the modified version with multiple states (which crashes) for reference. I have also checked with DMRG installations from colleagues, and they all seem to fail in the way.

I would greatly appreciate any information or guidance to solve this problem.

Thank you,
Laura


007-standard test input:

&GATEWAY
coord
3
 CH2 Triplet coordinates in Angstrom
 C 0.000000 0.000000 0.000000
 H 0.000000 0.000000 1.077500
 H 0.784304 0.000000 -0.738832
 basis=ANO-RCC-MB
Group=Nosym
 AMFI
 ANGMOM
 0.0 0.0 0.0

&SEWARD

&DMRGSCF
ActiveSpaceOptimizer=QCMaquis
DMRGSettings
 max_bond_dimension=1024
 nsweeps=10
EndDMRGSettings
OOptimizationSettings
 Spin=3
 Inactive=1
 Ras2=6
 NActEl=6,0,0
EndOOptimizationSettings

>>> COPY $Project.JobIph JOBOLD

>>> COPY $Project.dmrgscf.h5 $Project.trip.h5

>>> EXEC mv $CurrDir/$Project.checkpoint_state.0.h5 $CurrDir/$Project.trip.checkpoint_state.0.h5

>>> EXEC $MOLCAS/Tools/qcmaquis/qcm_checkpoint_rename.py $Project.trip.h5 -q

&DMRGSCF
ActiveSpaceOptimizer=QCMaquis
DMRGSettings
 max_bond_dimension=1024
 nsweeps=10
EndDMRGSettings
OOptimizationSettings
Spin=1
Inactive=1
Ras2=6
NActEl=6,0,0
JobIph
EndOOptimizationSettings

>>> COPY $Project.dmrgscf.h5 $Project.sing.h5

>>> EXEC mv $CurrDir/$Project.checkpoint_state.0.h5 $CurrDir/$Project.sing.checkpoint_state.0.h5

>>> EXEC $MOLCAS/Tools/qcmaquis/qcm_checkpoint_rename.py $Project.sing.h5 -q

&MPSSI
Nrof
2 1 1
1
1
FILE
2
$Project.trip.h5
$Project.sing.h5
EJOB
SOCOupling
0.0001
SPIN

007-standard test output

      ****************************************************************************************************
      *                                                                                                  *
      *                                    Special properties section                                    *
      *                                                                                                  *
      ****************************************************************************************************


++ Length and velocity gauge comparison (SO states):

 --------------------------------------------------
 A comparison between the dipole oscillator strengths in
 length and velocity gauge will be performed

     All dipole oscillator differences above the tolerance of   1.00000000E-01  will be printed

 Due to basis set deficiency these oscillator may be problematic

 The tolerance is defined as ABS(1-O_l/O_v)
 O_l : dipole oscillator strength in length gauge
 O_p : dipole oscillator strength in velocity gauge
 --------------------------------------------------

 No problematic oscillator strengths above the tolerance   0.10000000000000001       have been found

--

 Electric-dipole - electric-octupole not included
++ I/O STATISTICS

  I. General I/O information
  - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
  Unit  Name          Flsize      Write/Read            MBytes           Write/Read
                      (MBytes)       Calls              In/Out           Time, sec.
  - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
   1  SCRATCH           0.00 .       0/       0 .      0.0/      0.0 .       0/       0
   2  ANNI              0.00 .       0/       0 .      0.0/      0.0 .       0/       0
   3  RUNFILE           0.14 .     427/    1905 .      3.0/     14.0 .       0/       0
   4  ONEINT            1.37 .       0/     317 .      0.0/    296.0 .       0/       0
  - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
   *  TOTAL             1.51 .     427/    2222 .      3.0/    310.0 .       0/       0
  - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

  II. I/O Access Patterns
  - - - - - - - - - - - - - - - - - - - -
  Unit  Name               % of random
                         Write/Read calls
  - - - - - - - - - - - - - - - - - - - -
   1  SCRATCH              0.0/   0.0
   2  ANNI                 0.0/   0.0
   3  RUNFILE             28.6/   9.4
   4  ONEINT               0.0/  97.2
  - - - - - - - - - - - - - - - - - - - -
--
--- Stop Module: mpssi at Wed May 11 20:02:00 2022 /rc=_RC_ALL_IS_WELL_ ---
*** files: ch3-hfcc.mpssi.h5 xmldump
    saved to directory /projects/academic/lauramot/HFCC/TEST
--- Module mpssi spent 2 seconds ---

.##################.
.# Happy landing! #.
.##################.

007-modified input file

&GATEWAY
coord
3
 CH2 Triplet coordinates in Angstrom
 C 0.000000 0.000000 0.000000
 H 0.000000 0.000000 1.077500
 H 0.784304 0.000000 -0.738832
 basis=ANO-RCC-MB
Group=Nosym
 AMFI
 ANGMOM
 0.0 0.0 0.0
FLDG
0

&SEWARD
clight
137.036

&DMRGSCF
ActiveSpaceOptimizer=QCMaquis
DMRGSettings
 max_bond_dimension=1024
 nsweeps=10
EndDMRGSettings
OOptimizationSettings
 Spin=3
 Inactive=1
 Ras2=6
 NActEl=6,0,0
 CIROOT=2 2 1
EndOOptimizationSettings

>>> COPY $Project.JobIph JOBOLD

>>> COPY $Project.dmrgscf.h5 $Project.trip.h5

>>> EXEC mv $CurrDir/$Project.checkpoint_state.0.h5 $CurrDir/$Project.trip.checkpoint_state.0.h5

>>> EXEC $MOLCAS/Tools/qcmaquis/qcm_checkpoint_rename.py $Project.trip.h5 -q

>>> EXEC mv $CurrDir/$Project.checkpoint_state.1.h5 $CurrDir/$Project.trip.checkpoint_state.1.h5

>>> EXEC $MOLCAS/Tools/qcmaquis/qcm_checkpoint_rename.py $Project.trip.h5 -q

&DMRGSCF
ActiveSpaceOptimizer=QCMaquis
DMRGSettings
 max_bond_dimension=1024
 nsweeps=10
EndDMRGSettings
OOptimizationSettings
Spin=1
Inactive=1
Ras2=6
NActEl=6,0,0
CIROOT=2 2 1
JobIph
EndOOptimizationSettings

>>> COPY $Project.dmrgscf.h5 $Project.sing.h5

>>> EXEC mv $CurrDir/$Project.checkpoint_state.0.h5 $CurrDir/$Project.sing.checkpoint_state.0.h5

>>> EXEC $MOLCAS/Tools/qcmaquis/qcm_checkpoint_rename.py $Project.sing.h5 -q

>>> EXEC mv $CurrDir/$Project.checkpoint_state.1.h5 $CurrDir/$Project.sing.checkpoint_state.1.h5

>>> EXEC $MOLCAS/Tools/qcmaquis/qcm_checkpoint_rename.py $Project.sing.h5 -q

&MPSSI
Nrof
2 2 2
1 2
1 2
FILE
2
$Project.trip.h5
$Project.sing.h5
EJOB
SOCOupling
0.0001
SPIN
EPRG
 2.0

007-modified output (crashed)

This binary contains symmetries: 2u1pg su2u1pg
DMRG version 3.1.1-22a0f5706267a24c183ae1bd3831b54e5a70fcc4 (HEAD) release-3.1.1
Temporary storage enabled in ./tmp//storage_temp_2b5daa989e2b/
Will start again at site -1 in sweep 15
Loading checkpoint from /scratch/10491472/ch3-hfcc.sing.checkpoint_state.0.0.0.h5
Parameters:
               Sites: 6
   Spin-up electrons: 3
 Spin-down electrons: 3
               Irrep: 0
The hamiltonian will contain 1444 terms
MPO Bond 0: 87/34
MPO Bond 1: 98/40
MPO Bond 2: 125/54
MPO Bond 3: 136/60
MPO Bond 4: 17/6
MPO Bond 5: 1/0
MPS initialization has finished...
Measuring oneptdm_aa
 measuring in 2u1 version of tagged_nrank
Measuring oneptdm_bb
 measuring in 2u1 version of tagged_nrank
--- Stop Module: mpssi at Wed May 11 19:30:35 2022 /rc=-6 ---
*** files: ch3-hfcc.mpssi.h5 xmldump
    saved to directory /projects/academic/lauramot/HFCC/TEST
--- Module mpssi spent 6 seconds ---

.########################.
.# Non-zero return code #.
.########################.

Last edited by laumot (2022-05-13 22:37:18)

Offline

Board footer

Powered by FluxBB 1.5.11

Last refresh: Today 19:10:51