Dear Steve/developers,
The recent patch with the “short cut” notation in fix ave/time has prompted me to send this mail.
Is it feasible to have a dump style “molecule” wherein per molecule quantities could be dumped
to a file? I am not aware of how dump is implemented currently, and so not sure if
this would be feasible/make sense.
Alternatively, if the thermo_style custom could have a similar short-cut as ave/time,
even that would be quite helpful (Though it might conflict with the present meaning of C_ID).
Thanks,
Mario
dear mario,
disclaimer: i don't speak for steve, this is just my personal opinion,
but steve is
calling the shots around here about what goes into the code and what not.
Dear Steve/developers,
The recent patch with the "short cut" notation in fix ave/time has prompted
me to send this mail.
Is it feasible to have a dump style "molecule" wherein per molecule
quantities could be dumped
to a file? I am not aware of how dump is implemented currently, and so not
sure if
this would be feasible/make sense.
i don't think it makes too much sense. what kind of per molecule properties
are you looking for and how do you want to further process them?
the clean way to deal with this would be to write a compute.
the problems with molecules is that they may span many domains,
so it might be quite a hassle (and thus detrimental to the performance)
to collect this information and then potentiall broadcast it back around,
so that each atom has the per molecule property data available.
in other words, unless you have a good case where this would be a
compute intense and well parallelizable property calculation that cannot
be easily done in post-processing, i believe you are better off doing
this after the fact.
be it as it may, if you can provide a compelling usage case where
there would be a big benefit to implement what you suggest, then
chances of somebody taking it into serious consideration.
i have a new "wouldn't it be great if"-type of idea every other day,
but that doesn't mean that they get implemented, or that it would
be worth doing so. 
cheers,
axel.
Dear Axel,
Thanks for you inputs.
I do not have a god case to present. :o)
In your words it was more of the “wouldn’t it be great if” category.
I am currently implementing the functionality I desire through post-processing itself…
The intention was: LAMMPS may add many more computes for molecule based properties in times to come… printing out these quantities is currently not too convenient, since one needs to specify the index of each element in the thermo_style. Since I haven’t checked how the dumping is done, I was trying my luck. ;o) But I get the message…
Thanks,
Mario
Dumps (like atom and local and custom) are designed to work
where each atom has one (or more) quantities to output. They're
not really designed to work with per-molecule output. But you
can you use fix ave/time to dump out molecular info since it
is a list of global quantities. See if you can't make it do what
you want. If not, post an example of something you want to
output in a particular way, and we can see what's needed.
Steve
Dear Steve,
Thanks for the suggestion. It is perfect!
Sorry, I had not looked at this line in the documentation for fix ave/time:
Note that if no time averaging is done, this command can be used as a convenient way to simply output one or more global values to a file.
It would be helpful if a line could be added in the “dump” doc page,
stating that fix ave/time can be used to output global quantities.
Thanks once again,
Mario
It would be helpful if a line could be added in the "dump" doc page,
stating that fix ave/time can be used to output global quantities.
Section 4.15 of the doc pages summarizes all this.
Steve