I observed the following with the recent 2020 R1 version: Trying to find vibrational frequencies for a state optimized with MCSCF, the CPMCHF equations do not converge in 300 iterations when optimization and Hessian calculation are done in the same run. i.e.
$CONTRL RUNTYP=OPTIMIZE...
$STATPT HSSEND=.T. ...
However, if the two steps are done in separate runs, i.e.
$CONTRL RUNTYP=HESSIAN ... in the second run,
convergence is obtained in ca 30 iterations. I attach the two input and output files for the combined run, and for the separate HESSIAN run. (The OPTIMIZE-run starts already with the optimized structure).
I see that a convergence problem was previously reported (11.Feb.2019) for SA-MCSCF, which seems so far unsolved. My observation is for MCSCF of a single state.
I compiled on two different Ubuntu versions (16.04 LTS and 20.04 LTS) with gfortran 5 and gfortran 9.3
I used either openblas or MKL math library
The error occurs for parallel and serial run.
Hence it looks to me like some problem in the source.
best regards,
Bernhard
Make sure these boxes are checked before submitting your issue - thank you!
[x] Summary of the issue and expected results.
[x] Description of the run environment:
operating system
Ubuntu 16.04 LTS and Ubuntu 20.04 LTS
compiler (if using a pre-compiled binary indicate with pre-compiled binary)
gfortran 5 and gfortran 9
math library
openblas and MKL
communication library
sockets
[x] An input file (if applicable) to reproduce the issue.
(Optional) A copy of your install.info file.
If the input file is more than 50 lines then please send it as an attachment to the following email address (gamess@iastate.edu) with the subject being the title of the issue.
I observed the following with the recent 2020 R1 version: Trying to find vibrational frequencies for a state optimized with MCSCF, the CPMCHF equations do not converge in 300 iterations when optimization and Hessian calculation are done in the same run. i.e. $CONTRL RUNTYP=OPTIMIZE... $STATPT HSSEND=.T. ... However, if the two steps are done in separate runs, i.e. $CONTRL RUNTYP=HESSIAN ... in the second run, convergence is obtained in ca 30 iterations. I attach the two input and output files for the combined run, and for the separate HESSIAN run. (The OPTIMIZE-run starts already with the optimized structure).
I see that a convergence problem was previously reported (11.Feb.2019) for SA-MCSCF, which seems so far unsolved. My observation is for MCSCF of a single state.
I compiled on two different Ubuntu versions (16.04 LTS and 20.04 LTS) with gfortran 5 and gfortran 9.3 I used either openblas or MKL math library The error occurs for parallel and serial run.
Hence it looks to me like some problem in the source. best regards, Bernhard
Hessian-Problem.zip
Make sure these boxes are checked before submitting your issue - thank you!
[x] Summary of the issue and expected results.
[x] Description of the run environment:
pre-compiled binary
) gfortran 5 and gfortran 9[x] An input file (if applicable) to reproduce the issue.
(Optional) A copy of your install.info file.
If the input file is more than 50 lines then please send it as an attachment to the following email address (gamess@iastate.edu) with the subject being the title of the issue.