Fix rigid: Output Rigid Body Auxiliary File

Hello,

I have been looking through the documentation about restarting a rigid system, and I saw in the documentation about using the “reinit no” command, and then using the infile keyword to read the auxiliary file. It seems; however, that to create the rigid auxiliary file in the first place you need to use the infile keyword which requires you to already have a rigid auxiliary file.

Is there any way of recreating this file using compute or dump commands that I am missing? I want to make sure the data between the fixes reinitializes as exactly as possible when I run the restart file.

Thanks so much for your help,

Zeke

Hello,

I have been looking through the documentation about restarting a rigid
system, and I saw in the documentation about using the "reinit no" command,
and then using the infile keyword to read the auxiliary file. It seems;
however, that to create the rigid auxiliary file in the first place you
need to use the infile keyword which requires you to already have a rigid
auxiliary file.

Is there any way of recreating this file using compute or dump commands
that I am missing? I want to make sure the data between the fixes
reinitializes as exactly as possible when I run the restart file.

​your logic is inverted. you are looking at solutions for problems that you
don't have. why not try restarting normally first?
the infile option is meant to be used for systems where the automatic
computation of body properties is not easily possible, e.g. when building
object from overlapping volumes of extended particles.
the reinit no option is meant to help for these kind of setups in case a
lammps script uses the run command multiple times.

unless your system qualifies (and it is unlikely, or else you would have
generated the infile already), there is no obvious need to use them.

so before worrying about potential problems and complex features, you
should see if the default setup actually works.
also, the obsession of restarting "exactly" is ignoring the realities of
the methodology you are using. MD is chaotic, so even the tiniest
difference will lead to a divergence and with floating point math being not
associative, these tiny differences are guaranteed. if you want perfect
behavior on restarts vs. run continuation, you need to write an MD code
based on fixed point math.

axel.

Dr. Kohlmeyer,

Thanks for your quick response. I was just overthinking the problem!

Thanks,