Problems in NumForce: "STOP NumForce: There was a problem with point xm01 !"

[evgeniy]

Dear All,

I am trying to do vibrational analysis using NumForce at the ricc2 level.
For some reasons I get the following error after numerical calculations
of the second derivatives:

STOP NumForce: There was a problem with point xm01 !

 Please try to restart NumForce !
 NumForce will just redo the missing points.
 If you get this error message also after restart, check the log file xm01.log

I have checked the xm01.log file -- everything is fine there. But there is
no the xm01.grad file and this causes the crash of the calculation.
Restarting the calculation did't help; I got the same error message.

I will very appreciate any pointers to solve the problem!
(Btw, is there a way to curcumvent it by calculating the
xm01 point "by hands" and then creating the xm01.grad
file "by hands" and restarting the NumForce?) Thanks!

Best regards,
Evgeniy

[christof.haettig]

This can have several reasons, as e.g. a problem in the set up of you shell, etc.

But in of the output or log files or in printout or error output from your calculation there should be an error message with give more information about what is going wrong.

Anyway, you should contact the staff at COSMOLogic for support.

Christof

[Arnim]

Hi,

how did you specify the gradient calculation in the control file?

Arnim

[evgeniy]

Dear Christof and Arnim,

Many thanks for your replys.
Concerning the error message(s) which should be in some of the
files, I have checked everywhere and there is nothing like that.
Moreover, I noticed that when the job was running (not all distortions
were done) the file "xm01.grad" (!!!) was present in the numforce/Kraftwerk
directory. But after all points had been done this file disappeared - perhaps got deleted
by NumForce of some other script. This happened when I ran NumForce in
parallel. But recently I started it just with one processor and it recalculated
xm01.grad and kept it. So it passed throught that obstacle with the
xm01.grad file. But then it crashed again  at the aoforce step because it
didn't find the hessian file. Here is the error from aoforce.out:

 $hessian : could not find file <hessian> !

 MODTRACE: no modules on stack

  HSSRD : error reading $hessian !
 force ended abnormally



So, it looks like it didn't write the hessian in the hessian file.
(btw, there was a post concerning this same error: http://www.turbo-forum.com/index.php?topic=201.msg542#msg542
but the solution doesn't seem to have been found or was not posted).

Concerning Arnim's question (how did you specify the gradient calculation in the control file?),
here is the part of my control where it is sepcified:

$ricc2
 cc2
 geoopt model=CC2
$excitations
 irrep=a nexc=1 npre=3
 xgrad states=(a 1)
$freeze
 implicit core=   19 virt=    0
$cbas    file=auxbasis


This worked fine at the geometry optimization.

Best regards,
Evgeniy

[evgeniy]

Dear Christof and Arnim,

There was kind of an error message that dipoles from ricc2 were not available
(sorry that I ovelooked it first). Is it important for calculating Hessian to have
also dipoles and dipole derivatives? Could it be the culprit of the crash?

Thanks!

Best,
Evgeniy

[Arnim]

Hi,

indeed that seems to be the problem.
You would have to add
$excitations
  exprop  states=all  operators=diplen
to the control file.

I hope this helps,
Arnim

[evgeniy]

Dear Arnim,

Thanks for your reply. Your suggestion:

> You would have to add
> $excitations
>  exprop  states=all  operators=diplen
> to the control file.

worked perfectly out . Even in parallel the calculation ran smoothly.
Maybe one should add the corresponding note in the manual if it is not yet there.

Thanks again,
Best,
Evgeniy