TDDFT problem using NWCHEM 6.1


Just Got Here
Dear all, I tried a simple test calculation using the TDDFT module
as implemented in NWChem 6.1. I report the complete input deck used for the test:

echo
START h2o
TITLE "TDDFT B3LYP/3-21G* H2O"
memory total 400 mb
geometry units au
O 0 0 0
H 1.430 0.0 1.1
H -1.430 0.0 1.1
end
BASIS
* library 3-21G*
END
DFT
XC B3LYP
grid fine
END
TDDFT
rpa
NROOTS 20
notriplet
thresh 1.0d-05
END
task tddft energy
permanent_dir /media/DSK_1/pernwc
scratch_dir /media/DSK_2/scrnwc


The calculation aborts, after the SCF procedure, with the following error:

0:0:Too small a subspace:: 40
1:1:Too small a subspace:: 40
(rank:1 hostname:antares2 pid:9906):ARMCI DASSERT fail. ../../ga-5-1/armci/src/common/armci.c:ARMCI_Error():208 cond:0
2:2:Too small a subspace:: 40
(rank:2 hostname:antares2 pid:9907):ARMCI DASSERT fail. ../../ga-5-1/armci/src/common/armci.c:ARMCI_Error():208 cond:0
(rank:0 hostname:antares2 pid:9905):ARMCI DASSERT fail. ../../ga-5-1/armci/src/common/armci.c:ARMCI_Error():208 cond:0
3:3:Too small a subspace:: 40
(rank:3 hostname:antares2 pid:9908):ARMCI DASSERT fail. ../../ga-5-1/armci/src/common/armci.c:ARMCI_Error():208 cond:0

Interestingly, using NWChem 6.0 (and the same input file), the job ends normally.
The compilation, of both the 6.1 and 6.0 releases,was performed considering the settings reported below:

export NWCHEM_TOP=$HOME/nwchem-6.1
export LARGE_FILES=TRUE
export TCGRSH=/usr/bin/ssh
export NWCHEM_TARGET=LINUX64
export NWCHEM_MODULES=all
export ENABLE_COMPONENT=yes

export USE_MPI=yes
export USE_MPIF=yes
export USE_MPIF4=yes
export MPI_HOME=/opt/mpich2-1.4.1p1
export MPI_LOC=$MPI_HOME
export MPI_LIB=$MPI_LOC/lib
export MPI_INCLUDE=$MPI_LOC/include
export LIBMPI='-lfmpich -lmpich -lmpichf90 -lopa -lmpl -lrt -lpthread'

export HAS_BLAS=yes
export BLASOPT='-L/opt/intel/Compiler/11.1/072/mkl/lib/em64t -lmkl_intel_ilp64 -lmkl_solver_ilp64_sequential -lmkl_sequential -lmkl_core -lpthread'

export LIB_DEFINES=-DDFLT_TOT_MEM=16777216
export FOPTIMIZE='-O3 -xSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 -no-prec-div -funroll-loops -unroll-aggressive'
export COPTIMIZE='-O3 -xSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 -no-prec-div -funroll-loops'

Do you have any ideas/suggestions about the origin of the problem when NWChem 6.1 is used
for the TDDFT test?
Thank you very much in advance for your kind reply.
Best regards,

Davide Vanossi

Forum Regular
Hi Davide,

Looks like you have requested too many roots. You probably don't have 20 excited states with 3-21G*.
Increase the basis set size or reduce the number of roots. I'll make a note to add a clearer error messge.

Best,
-Niri

niri.govind@pnnl.gov

Just Got Here
Quote:Niri Feb 29th 8:19 pm
Hi Davide,

Looks like you have requested too many roots. You probably don't have 20 excited states with 3-21G*.
Increase the basis set size or reduce the number of roots. I'll make a note to add a clearer error messge.

Best,
-Niri

niri.govind@pnnl.gov



Dear Niri,
the problem is not related to the numer of roots requested or to the use of a small basis set:
using NWChem 6.0 everything was fine and the test ended normally (as I reported in my previous post).
The TDDFT calculation aborts only using the 6.1 release of the code.
Thank you very much for your kind reply.

All the best,
Davide

Forum Regular
Dear Davide,

I will look into this. It's possible the Davidson restarts that we implemented in this release has affected something. But I assure you the calculated excitations are fine if you don't encounter this problem.

Best regards,
-Niri


Forum >> NWChem's corner >> Running NWChem