On Fri, Aug 26, 2016, Hannes Loeffler wrote:
>
> I have just noticed that there seems to have been a change in how the
> begin time is recorded in the mdout file of pmemd16. In the output
> files of a pmemd14 run I see that the begin time is initially taken from
> the restart file and then any restart based on this will increment the
> begin time with the run time of the previous run.
>
> Pmemd16 does record the first begin time as well but then resets the
> counter to zero e.g. if the begin time was initially 1000.0 ps and I
> have 500.0 ps runs the output of successive runs will be: 1000.0 ps,
> 500.0 ps, 1000.0 ps, 1500.0 ps etc.
I don't see this behavior, but I'm having trouble following your description
of the problem. Try this:
cd $AMBERHOME/test/dhfr
export TESTsander=pmemd
Run.dhfr
The test should PASS, and the output file removed, but this just means that
the mdout.dhfr.save file is the same as mdout.dhfr. You could modify
the Run.dhfr script to exit before the "dacdif" command.)
This is a restart calculation, and the times look OK to me. I checked this
with a netCDF restart file as well, and saw good behavior. (That is, I took
the (netCDF) restrt file from the Run.dhfr output, and used it as input
for a second short calculation.
So: can you try to see what is different between your calculation and this
test case? Maybe I am not understanding your description of the problem...can
you make a small test case that illustrates the bad behavior?
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Aug 26 2016 - 06:30:03 PDT
