SAPT_DFT segmentation error

SAPT
1

Dear All,

I have recently started using psi4 to carry out SAPT calculations. I ran a couple of dimers given in examples and they worked fine. When I try the molecule of my interest, I get an error stating the following.

“forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
libgdma.so 00002B69F87D40DC for__signal_handl Unknown Unknown
libpthread-2.17.s 00002B69ED2365D0 Unknown Unknown Unknown
libchemps2.so.3 00002B69F8605D7A __intel_avx_rep_m Unknown Unknown
core.cpython-36m- 00002B69F7113A8A _ZN3psi6linalg6de Unknown Unknown
core.cpython-36m- 00002B69F70D8EFF _ZN3psi6Matrix5al Unknown Unknown
core.cpython-36m- 00002B69F70D99F8 _ZN3psi6MatrixC1E Unknown Unknown
core.cpython-36m- 00002B69F6A7709B _ZNSt12__shared_p Unknown Unknown
core.cpython-36m- 00002B69F6A72D25 _ZN3psi8DiskDFJK1 Unknown Unknown
core.cpython-36m- 00002B69F6A3F3A6 _ZN3psi8DiskDFJK1 Unknown Unknown
core.cpython-36m- 00002B69F4A975A7 Unknown Unknown Unknown
core.cpython-36m- 00002B69F4A300D6 Unknown Unknown Unknown
python3.6 0000558F25579A14 _PyCFunction_Fast Unknown Unknown
python3.6 0000558F25601BAE Unknown Unknown Unknown
python3.6 0000558F2562425A _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FAFD4 Unknown Unknown Unknown
python3.6 0000558F255FBE51 Unknown Unknown Unknown
python3.6 0000558F25601B35 Unknown Unknown Unknown
python3.6 0000558F25625019 _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FBC1B Unknown Unknown Unknown
python3.6 0000558F25601B35 Unknown Unknown Unknown
python3.6 0000558F2562425A _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FBC1B Unknown Unknown Unknown
python3.6 0000558F25601B35 Unknown Unknown Unknown
python3.6 0000558F2562425A _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FC969 PyEval_EvalCodeEx Unknown Unknown
python3.6 0000558F255FD896 Unknown Unknown Unknown
python3.6 0000558F2557981E PyObject_Call Unknown Unknown
python3.6 0000558F256258BB _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FC969 PyEval_EvalCodeEx Unknown Unknown
python3.6 0000558F255FD896 Unknown Unknown Unknown
python3.6 0000558F2557981E PyObject_Call Unknown Unknown
python3.6 0000558F256258BB _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FAFD4 Unknown Unknown Unknown
python3.6 0000558F255FBE51 Unknown Unknown Unknown
python3.6 0000558F25601B35 Unknown Unknown Unknown
python3.6 0000558F25625019 _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FC969 PyEval_EvalCodeEx Unknown Unknown
python3.6 0000558F255FD70C PyEval_EvalCode Unknown Unknown
python3.6 0000558F2567D574 Unknown Unknown Unknown
python3.6 0000558F2567D60D PyRun_StringFlags Unknown Unknown
python3.6 0000558F25621BFE Unknown Unknown Unknown
python3.6 0000558F25579951 _PyCFunction_Fast Unknown Unknown
python3.6 0000558F25601A5C Unknown Unknown Unknown
python3.6 0000558F2562425A _PyEval_EvalFrame Unknown Unknown
python3.6 0000558F255FC969 PyEval_EvalCodeEx Unknown Unknown
python3.6 0000558F255FD70C PyEval_EvalCode Unknown Unknown
python3.6 0000558F2567D574 Unknown Unknown Unknown
python3.6 0000558F2567D971 PyRun_FileExFlags Unknown Unknown
python3.6 0000558F2567DB73 PyRun_SimpleFileE Unknown Unknown
python3.6 0000558F25681663 Py_Main Unknown Unknown
python3.6 0000558F2554B43E main Unknown Unknown
libc-2.17.so 00002B69ED4653D5 __libc_start_main Unknown Unknown
python3.6 0000558F2562AD0B Unknown Unknown Unknown”

I went through a few threads which has similar errors. One of which says, have to build from the scratch as there was ‘integer overflow’ problem. Since this was in 2018, I assume the latest version comes with the problem fixed. Am I right in assuming it??. I am also new to the linux environment.

I have attached the input and output files for your kind reference. I am also not completely sure if my input is completely correct. Prof. Sherill mentioned in a thread that SAPT-DFT is under testing. (About SAPT (DFT)). I am not sure how to go about my problem. It would helpful if someone can kindly help me troubleshoot the error. Please let me know if you need any further details.

Thank you.Znsapt2dft.txt (6.2 KB)Znsapt2dft_out.dat (23.5 KB)

2

Hi,

Certainly a big molecule but it already crashes at the SCF part which should not happen.
Did you specify a scratch directory? Default is /tmp/ and generally not suitable for production calculations.

export PSI_SCRATCH=/path/to/existing/writable/local-not-network/disk/for/scratch/files
OR
psi4 -s /path/to/existing/writable/local-not-network/disk/for/scratch/files

3

Dear Dr. Holger,

Thank you for the kind reply.

Yes, I have changed the scratch directory, I think you had mentioned it in one of the posts. So I changed it. And also, when the job was running, I monitored the scratch folder, to see if it populates with files and it does. Please let me know if you need any further information.

4

Dear Dr. Holger,

I had run three calculations so far, including the ones I had already attached along with the question. I ran the calculations assuming the memory corresponds to storage. Is it correct? Kindly let me know.

The first one : memory = 300GB
The second one : memory = 800GB
The third one : memory = 360GB

The output of the third one is already attached with the question. Now, I am attaching the input and output files of first and second ones with this message for your kind reference.

1st_Znsapt2dft.log.txt (4.1 KB) 1st_Znsapt2dft.out.txt (23.5 KB) 1st_Znsapt2dft.txt (6.2 KB) 2nd_Znsapt2dft.log.txt (983 Bytes) 2nd_Znsapt2dft.out.txt (250.3 KB) 2nd_Znsapt2dft.txt (6.2 KB)

Please let me know if I am not clear or you need any further information. Thank you.

5

The memory sets the available RAM memory, not disk storage. Disk storage is not controllable in PSI4.
set memory to 70-80% of the node. That is generally safe.

6

@hokru

Thank you for the suggestion. In one 4th post, I have given the input , output and log files where I have used the memory to be 300 GB out of available 396 GB RAM, and this corresponds to 75% of the RAM. I still got error ,which said the following.
“forrtl: error (78): process killed (SIGTERM)”

I have reattached the files for your convenience. Can you please check the error and let me know how I can troubleshoot the error.

Thank you.1st_Znsapt2dft.txt (6.2 KB) 1st_Znsapt2dft.log.txt (4.1 KB) 1st_Znsapt2dft.out.txt (23.5 KB)

7

I dont know how to troubleshoot the error. The integral step should be without problem. But I dont have a machine where I can request 300 GB of memory.
With latest psi4 version, and using only 100 GiB, I dont seem to have issues. It’s too big of a job to finish it, though.

BTW, the aug-DZ basis leads to linear dependency issues for your system, so I’d think about adjusting my basis to something less diffuse, too. Likely the fullerenes is causing this.

Something like below could work better:

basis {
assign aug-cc-pVDZ
assign h cc-pVDZ
assign c cc-pVDZ
}

Also make sure you have your charges right, the Zn-complex is really neutral?

And lastly, remove the D3 from the SAPT functional, it does nothing to SAPT and might only lead to issues.

Good luck!