sharc-md / sharc

The SHARC molecular dynamics (MD) program suite is an ab initio MD software package developed to study the excited-state dynamics of molecules.
https://www.sharc-md.org
GNU General Public License v3.0
59 stars 31 forks source link

Optimization and Frequency Calculation Convergence Issue #99

Open lukhman9020 opened 4 months ago

lukhman9020 commented 4 months ago

Dear Sir,

I hope this message finds you well. I am currently facing convergence challenges with my Molcas optimization and frequency calculations using CASPT2/cc-pVDZ levels of theory on our cluster infrastructure. Despite a substantial computational setup featuring 88 compute nodes with 112 cores each and a peak performance is RPeak 205 Teraflops, RMax 141 Teraflops, CPU type: Intel(R) Xeon(R) Gold 6132 the calculations fail to converge within the allotted 5-day timeframe. I'm using OpenMolcas standard v23.06 for the simulations. I'm uncertain whether the convergence problem stems from the software version or if it's related to our cluster's performance specifications. Would modifying the basis set, reducing the MAXITER value from 120(in molcas.input file) , or adjusting the active space size potentially help with convergence? Could you please provide guidance on how to correct this?. I here attach my files for your reference https://drive.google.com/drive/folders/1vz9d7W8QS4H5AqCcRT1M4qGvyxz_DU-S

Looking forward to your guidance.

Best regards, LUKHMANUL HAKEEM K.

maisebastian commented 4 months ago

Dear lukhman9020, again, this is a pure OpenMolcas issue. It is not our responsibility as SHARC developers to provide you support about OpenMolcas issues. You will be better served in the OpenMolcas forum.

After briefly checking the log file, there are several problems:

Best regards, Sebastian Mai

lukhman9020 commented 4 months ago

Dear lukhman9020, again, this is a pure OpenMolcas issue. It is not our responsibility as SHARC developers to provide you support about OpenMolcas issues. You will be better served in the OpenMolcas forum.

After briefly checking the log file, there are several problems:

  • you are not adding the keywords for CASPT2 analytical gradients. Instead, you are using numerical gradients, which are expensive. Note that sharc_molcas.py cannot write OpenMolcas inputs for all possible jobs (as is written in the Manual), so you still need to know for yourself how to write OpenMolcas inputs.
  • You are only using a single CPU core, so your large computer cluster does not help you here at all.
  • There are no "convergence issues". Please look into the log file and see for yourself.

Best regards, Sebastian Mai

OK, Thank you sir.

lukhman9020 commented 4 months ago

Dear sir, I hope this message finds you well. I am writing to seek assistance regarding an issue I encountered while performing an optimization and frequency calculation in MOLCAS version 23.06 using the CASPT2/cc-pVDZ level of theory.

During the calculation process, I encountered a premature abort error with the following message in the molcas.log file: ############################################################################### ###############################################################################

Location: AixRd

File: CHVEC1

Premature abort while reading buffer from disk

End of file reached

############################################################################### ############################################################################### [ process 0]: xquit (rc = 162): _IO_ERRORREAD

Program aborted. Backtrace:

0 0x55643e49282b in ???

1 0x55643e3fa822 in ???

2 0x55643e4c0b8e in ???

3 0x55643e4648d9 in ???

4 0x55643e46457c in ???

5 0x55643e4643cb in ???

6 0x55643e3ad104 in ???

7 0x55643e379399 in ???

8 0x55643e379872 in ???

9 0x55643e31507f in ???

10 0x55643e31b896 in ???

11 0x55643e2d4897 in ???

12 0x55643e287552 in ???

13 0x55643e2840ba in ???

14 0x7ff47a1fed84 in ???

15 0x55643e28410c in ???

--- Stop Module: caspt2 at Fri Feb 23 00:29:45 2024 /rc=-6 --- --- Module caspt2 spent 15 minutes 51 seconds ---

END DO

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

Timing: Wall=1036.87 User=901.08 System=108.63

I have attempted to troubleshoot the issue by change MOLCAS disk space availability and also add RICD keyword but the problem persists. I know this problem related with MOLCAS So I had already posted my question in Molcas forum but I did not get any reply. Could you please provide guidance on how to resolve this error?. Looking forward to your guidance. I here attach my files https://drive.google.com/drive/folders/1vz9d7W8QS4H5AqCcRT1M4qGvyxz_DU-S?usp=sharing Best regards, LUKHMANUL HAKEEM K.

maisebastian commented 4 months ago

Dear lukhman9020, did you check whether there is enough disk space/quota for your calculation? According to https://molcas.gitlab.io/OpenMolcas/sphinx/tutorials/tut_errors.html, such errors can appear if there is insufficient disk space. Alternatively, I guess such errors could happen if a network-mounted disk is unavailable due to some reasons. By the way, did you try to simply not set any MOLCASDISK? Typically, I never set this variable. See https://molcas.gitlab.io/OpenMolcas/sphinx/users.guide/environment.html#system-variables for an explanation.

If none of these ideas help, it might be due to a bug. You could try to run a different version of OpenMolcas to check that. Otherwise, there is nothing else that I know.

Best, Sebastian

lukhman9020 commented 4 months ago

Dear lukhman9020, did you check whether there is enough disk space/quota for your calculation? According to https://molcas.gitlab.io/OpenMolcas/sphinx/tutorials/tut_errors.html, such errors can appear if there is insufficient disk space. Alternatively, I guess such errors could happen if a network-mounted disk is unavailable due to some reasons. By the way, did you try to simply not set any MOLCASDISK? Typically, I never set this variable. See https://molcas.gitlab.io/OpenMolcas/sphinx/users.guide/environment.html#system-variables for an explanation.

If none of these ideas help, it might be due to a bug. You could try to run a different version of OpenMolcas to check that. Otherwise, there is nothing else that I know.

Best, Sebastian

Dear Sir, Thank you very much for your help.

Best regards, LUKHMANUL HAKEEM K.