Segmentation fault (core dumped) - CCSD(T) calculation

1

Dear friends, I hope you are fine.

I am trying to calculate the energy of a methane dimer at the CCSD(T)/aug-pVTZ level, but psi4 is giving me the error: “Segmentation fault (core dumped)”.

Below is my script:

memory 16gb

molecule ch4_dimer {
   C 0.00000000 0.00000000 0.00000000
   H 0.00000000 1.08452710 0.00000000
   H -0.51125100 -0.36150900 0.88551270
   H -0.51125100 -0.36150900 -0.88551270
   H 1.02250190 -0.36150900 0.00000000
   --
   C 0.00000000 3.70000000 -0.00000000
   H 0.00000000 2.61547290 -0.00000000
   H -0.51125100 4.06150900 -0.88551270
   H -0.51125100 4.06150900 0.88551270
   H 1.02250190 4.06150900 0.00000000
}

set basis aug-cc-pVTZ

energy('ccsd(t)')

My output file ends here:

image
image1095×417 28.9 KB

I get this error when I run this script on my pc, however on a cluster it is running normally.
I’m running this script with the following command: “psi4 -i test.dat -o test.out -n 1”

My version of psi4 is 1.1. and apparently I’m creating the SCRATCH directory correctly.

If you can kindly help me with this issue, I would be grateful. Thanks in advance.

2

1.1 is a very old version of Psi4. Without looking at any details, my immediate recommendation is to try out 1.5 and see if you still have the problem there.

3

Thanks for your fast reply, @jmisiewicz.

I think the problem was my version of psi4, the calculations in version 1.3 on the cluster I use didn’t give any errors.

Can you kindly help me with another related question? When I run the same methane dimer at the CCSD(T)/aug-pVQZ level I am getting the following error:

RuntimeError:
Fatal Error: PSIO Error
Error occurred in file: /scratch/psilocaluser/conda-builds/psi4-multiout_1557940846948/work/psi4/src/psi4/libpsio/error.cc on line: 128
The most recent 5 function calls were:

psi::PSIO::rw(unsigned long, char*, psi::psio_address, unsigned long, int)
psi::IntegralTransform::transform_tei_first_half(std::shared_ptrpsi::MOSpace, std::shared_ptrpsi::MOSpace)

I set the scratch directory where my script is with the command:

export PSI_SCRATCH=/home/users/emerson/my_directory

Doing this has a problem or is the error related to memory?

Below is my script:

memory 16gb

molecule ch4_dimer {
   C 0.00000000 0.00000000 0.00000000
   H 0.00000000 1.08452710 0.00000000
   H -0.51125100 -0.36150900 0.88551270
   H -0.51125100 -0.36150900 -0.88551270
   H 1.02250190 -0.36150900 0.00000000
   --
   C 0.00000000 3.70000000 -0.00000000
   H 0.00000000 2.61547290 -0.00000000
   H -0.51125100 4.06150900 -0.88551270
   H -0.51125100 4.06150900 0.88551270
   H 1.02250190 4.06150900 0.00000000
}

set basis aug-cc-pVQZ

energy('ccsd(t)')

Below is my output:

teste1.txt (23.8 KB)

Thanks in advance for your time and help.

4

What gets written to stderr when you attempt this job?

5

Thanks for the reply, @jmisiewicz

The error file is filled with this message:

PSIO_ERROR: unit = 41, errval = 12
PSIO_ERROR: 12 (error writing to file)

Thanks in advance for your time and help.

6

That failure is occurring in the early part of the integral transformation and suggests that you’re running out of disk space.

7

@crawdad, many thanks for the answer!

Is there any way to know what disk space would be used in an estimated way?

In the directory I ran this script that gave an error there was ~170gb of space available.

Thanks in advance for your time and help.

8

Ignoring symmetry, the integral pre-sort requires (N*(N+1)/2)^2 double-precision words, which would be 145 GB for your calculation (for which N = 528 basis functions). It also requires you to hold the original integral file on disk, which would be another 9 GB. If there are other files taking up space on that disk, then you could be running out of room.