Hi all,
I'm using rerun to post process a dump file for generating rdf. But I get "ERROR: Cannot reset timestep with a time-dependent fix defined" but I'm not resetting timestep!
I appreciate any help to resolve this issue. I put required files to regenerate the error and I'm using LAMMPS-ICMS ver. 26Feb2015
https://drive.google.com/folderview?id=0BzZfCUgRNwfYNHhTd0lfWDdSQW8&usp=sharing
Thank you,
Kasra.
Hi all,
I'm using rerun to post process a dump file for generating rdf. But I
get "ERROR: Cannot reset timestep with a time-dependent fix defined" but
I'm not resetting timestep!
you do! implicitly via read_dump which is called through rerun.
the error happens through changes in LAMMPS that were added with the
"chunk" commands.
not sure what a clean solution is here. steve would be the expert for
this. as a workaround, you may try changing the value of time_depend
from 1 to 0 in the fix_ave_time.cpp or similar fixes that expect a
continuous time step progression. or revert to a version before 10 Feb
2015.
axel.
Thanks Axel, then I'll try the hack for time_depend as you mentioned and I've Cced Steve to further comment on this.
Best,
Kasra.
The error check isn’t really part of the chunk commands. Rather
I realized that all the “ave” fixes should be considered time
dependent. Fix ave/time was probably just missing that flag
in previous versions.
The issue with read_dump would be if you read a series
of snapshots that are not consistent with the timesteps
on which fix ave/time needs its data, which depends
on the 3 values (Nevery, Nfreq, etc) that you specify
for fix ave/time. Then fix ave/time will be screwed up.
So the safe thing is the current error. It needs to be
made smarter to determine if it is actually OK. I think
that would probably mean that fix ave/time should check
if you missed a needed step, or if you backed up in time
(which reset_timestep would allow you to do).
Steve
Steve,
Sorry Steve I forgot to include the list so I’m resending it again.
But in the test files that I’ve sent the link, I’m using a dump file with snapshots 0 and 100 and I’m using ave/time with nevery=100, nrepeat=1 and nfreq=100 which are consistent with available snapshots, aren’t they?
But I still get the error.
Best,
Kasra.
Steve,
Sorry Steve I forgot to include the list so I’m resending it again.
But in the test files that I’ve sent the link, I’m using a dump file with snapshots 0 and 100 and I’m using ave/time with nevery=100, nrepeat=1 and nfreq=100 which are consistent with available snapshots, aren’t they?
But lammps currently doesn’t have a way to tell. So it assumes the worst.
Axel,
Then, it always will complain about that no matter if the fix ave/time timestep requirement are consistent with dump or not, right? In other words, can fix ave/time be used with rerun command in current version of lammps?
Thank you,
Kasra.
Axel,
Then, it always will complain about that no matter if the fix ave/time timestep requirement are consistent with dump or not, right? In other words, can
Right.
fix ave/time be used with rerun command in current version of lammps?
Not without hacking it the way I suggested.
This should be fixed now … will be in next patch.
So you can use fix ave/time with rerun, so long
as you ask fix ave/time to only use timesteps
that rerun provides.
Steve
Thanks a lot Steve. I actually was doing it by the Axel suggested hack. Does the patch works for all time-dependent fixes like fix ave/spatial also?
Best,
Kasra.
All the fix ave/* commands should now work with rerun.
Steve