Can't calculate more states using eom-sf-ccsd calculation in Q-Chem 5.4.0

[Mandira]

Hello, I am trying to calculate some excited states using eom-sf-ccsd calculation in Q-Chem 5.4.0. But I can’t calcute more then 5 states. Can anybody give some hints?

rem part of my input file:

$rem
basis = 6-31g
method = eom-ccsd
symmetry = false
sym_ignore = true
sf_states = [6]
scf_guess = core
CC_OSFNO = true
CC_FNO_THRESH = 9900
cc_eom_prop = 1
cc_eom_prop_te = 1
scf_convergence = 7
max_scf_cycles = 200
MEM_TOTAL = 480000
MEM_STATIC = 10000
cc_memory = 300000
$end

Output file:

and at last the job exits with “Error: in the serial run”.
Kindly help.

[epif]

From this snippet it looks like Q-Chem is solving for 6, not 5, states. Does this problem occur every time you try this job? Does it crash at the same exact spot? Have you tries using a different number of threads (qchem -nt N)? I would also suggest sending your full output to support@q-chem.com for further investigation.

[Mandira]

Yes, this problem occurs every time when I am submitting the job and exits at the same spot. I have also tried with different number of threads too but the result is same. I am sending the full output file at the given mail id. Please check.

[epif]

We received the files you sent, thank you. I don’t know what may be causing the crash, but in the input file you are asking for 480GB of memory for a very small job. It shouldn’t be a problem in itself (unless your workstation doesn’t in fact have that much memory), but replacing your memory settings with defaults makes the job complete in seconds:

!MEM_TOTAL = 480000
!MEM_STATIC = 10000
!CC_MEMORY = 300000
cc_backend = incore

Note that I am also adding CC_BACKEND=INCORE which improves efficiency if the job fits in memory such as in this case.

[Mandira]

Thanks for your response. Actually I thought that problem may occur due to memory so I had used maximum available memory. I have also tried according to your suggestions but the problem didn’t resolved.