If you have problems during the execution of MRCC, please attach the output with an adequate description of your case as well as the followings:
  • the way mrcc was invoked
  • the way build.mrcc was invoked
  • the output of build.mrcc
  • compiler version (for example: ifort -V, gfortran -v)
  • blas/lapack versions
  • as well as gcc and glibc versions

This information really helps us during troubleshooting :)

No SCF convergence -> CC energies + mem for LRCC

  • rayultine
  • Topic Author
  • Offline
  • Junior Member
  • Junior Member
More
6 years 7 months ago #399 by rayultine
Dear Mihály,
A couple issues with the MRCC standalone:
Is there a flag I can set or code that can be modified to make sure that MRCC does not proceed to the CC energy calculation if the SCF energy does not converge?

Also, sometimes for larger jobs I have allocated enough memory to satisfy goldstone/xmrcc, but when I get to the cc density module, I get an error that I have not allocated enough memory. The program does not tell me how much more memory I need. When computing gradients/frequencies/properties/etc., can MRCC provide an extra memory requirement assessment for the energy derivatives before the CC energy is computed?

Nate

Please Log in or Create an account to join the conversation.

  • kallay
  • Offline
  • Administrator
  • Administrator
  • Mihaly Kallay
More
6 years 7 months ago #400 by kallay
Dear Nate,
1) Unfortunately there is no such flag. You are right, it is not convenient. We should fix it.
2) What do you mean by CC density module? That is, which executable is running then, mrcc or prog?

Best regards,
Mihaly Kallay

Please Log in or Create an account to join the conversation.

  • rayultine
  • Topic Author
  • Offline
  • Junior Member
  • Junior Member
More
6 years 7 months ago #401 by rayultine
Replied by rayultine on topic No SCF convergence -> CC energies + mem for LRCC
Mihály,
Sorry I didn't try very hard to find an example of this error before posting.

The error "Insufficient memory for gradient calculation" occurs in the prop binary.

xmrcc prints memory requirements for this job as Minimal=3116MB, Optimal=40743MB (!)
We are allocating 24GB of memory, CCSD and left-hand LR-CCSD energies converge normally. prop is running out of memory.

Thanks for the help!
Nate

Please Log in or Create an account to join the conversation.

  • kallay
  • Offline
  • Administrator
  • Administrator
  • Mihaly Kallay
More
6 years 7 months ago #402 by kallay
Dear Nate,
Yes, xmrcc calculates the memory requirement for the mrcc executable, not for prop. We should also fix this.
You should allocate more memory. For an RHF (UHF) calculation you need roughly 8*6*n^3 (8*12*n^3) bytes, where n is the basis set size. Unfortunately it was designed for higher-order CC calculations, where this part of the calculation is not a bottleneck.

Best regards,
Mihaly Kallay
The following user(s) said Thank You: rayultine

Please Log in or Create an account to join the conversation.

  • rayultine
  • Topic Author
  • Offline
  • Junior Member
  • Junior Member
More
6 years 7 months ago #403 by rayultine
Replied by rayultine on topic No SCF convergence -> CC energies + mem for LRCC
Dear Mihály,
Thank you! That is very helpful info. If you have time to let me know how to patch the code to run these extra checks, that would be appreciated.

Nate

Please Log in or Create an account to join the conversation.

Time to create page: 0.040 seconds
Powered by Kunena Forum