Restart QMMM frequency job on QM region?
Clicked A Few Times
11:24:37 AM PDT - Mon, Apr 23rd 2012
Hi,

Is it possible to restart a frequency job for the QMMM module? If so, what file(s) is/are needed to be present? What changes if any should be made to the input deck? I can see a file with a ".fdrst" extension, I assume that has something to do with restarting the finite difference calculation; is it enough to have that file present?

Thanks

Clicked A Few Times
2:42:37 PM PDT - Tue, Apr 24th 2012
Update, calculate with .fdrst present
UPDATE: OK, I tried, and did in fact get more gradient information, but the intervals make it seem like everything was being recalculated. For example, from the initial calculation (which got through atom 32 1(+) )
atom:   1 xyz: 1(+) wall time:   16005.8      date:  Fri Apr 20 19:49:54 2012
atom:   1 xyz: 1(-) wall time:   16177.1      date:  Fri Apr 20 19:52:46 2012
atom:   1 xyz: 2(+) wall time:   16339.6      date:  Fri Apr 20 19:55:28 2012
atom:   1 xyz: 2(-) wall time:   16560.5      date:  Fri Apr 20 19:59:09 2012
atom:   1 xyz: 3(+) wall time:   16726.7      date:  Fri Apr 20 20:01:55 2012
atom:   1 xyz: 3(-) wall time:   16893.0      date:  Fri Apr 20 20:04:42 2012

and from the “restart” (which got through atom 41 2(-) ):
atom:   1 xyz: 1(+) wall time:     268.4      date:  Mon Apr 23 20:23:13 2012
atom:   1 xyz: 1(-) wall time:     560.5      date:  Mon Apr 23 20:28:05 2012
atom:   1 xyz: 2(+) wall time:     777.3      date:  Mon Apr 23 20:31:42 2012
atom:   1 xyz: 2(-) wall time:    1181.9      date:  Mon Apr 23 20:38:26 2012
atom:   1 xyz: 3(+) wall time:    1477.0      date:  Mon Apr 23 20:43:22 2012
atom:   1 xyz: 3(-) wall time:    1890.4      date:  Mon Apr 23 20:50:15 2012

Is there an RTDB setting to avoid recalculation? Any other changes that should be made to the input deck?

Clicked A Few Times
2:49:22 PM PDT - Tue, Apr 24th 2012
Update, start --> restart
UPDATE: I changed the "start" directive to "restart" (I didn't think this was needed for the QMMM module, but perhaps that's only relevant to geometry optimizations and SCF), and things seem to proceed OK, then I see the error
forrtl: severe (104): incorrect STATUS= specifier value for connected file,
unit 69, file /scratch/chchang/8Puke_4SbnDuCSwE/perm/8Puke_4SbnDuCSwE.fdrst

The file in question is present before the calculation starts, but disappears, so perhaps there is a STATUS being set to NEW, when it should be OLD? What NWChem source file would contain the reading of a numerical Hessian restart file for QMMM?

nwchem 6.0, compiled with Intel 11.1-064.


Thanks,

Chris
--Cchang Apr 24th 2:48 pm Local Time

Clicked A Few Times
1:06:53 PM PDT - Fri, May 25th 2012
Restart capability is not supported for QM/MM calculations.

Clicked A Few Times
6:51:17 PM PDT - Mon, Jul 1st 2019
Restart QM/MM calculations
The post from 2012 indicates that QM/MM calculations may not be restarted. Is this still the case?


Forum >> NWChem's corner >> QMMM