Compute momentum

[rupertnash]

Summary

Add a compute for the total momentum of a group on atoms, in the style of compute KE

For now, I have added this to the main LAMMPS src/ directory but you may prefer that I contribute this to USER-MISC. If this is the case, please let me know and I will alter the PR to make it so.

This compute is useful for a larger suite of fixes (for coupling between LAMMPS an other codes using MUI https://github.com/MxUI/MUI) I plan to contribute back as a user package in the coming months.

Related Issues

Author(s)

Rupert Nash, EPCC, The University of Edinburgh, r.nash@epcc.ed.ac.uk

Licensing

By submitting this pull request, I agree, that my contribution will be included in LAMMPS and redistributed under either the GNU General Public License version 2 (GPL v2) or the GNU Lesser General Public License version 2.1 (LGPL v2.1).

Backward Compatibility

No

Implementation Notes

I cribbed the class definition from compute_ke.h/.cpp, changing from scalar to vector

This was tested on a variety of input scripts by also dumping atom velocities masses.

No other features are affected.

Post Submission Checklist

Please check the fields below as they are completed after the pull request has been submitted. Delete lines that don't apply

• The feature or features in this pull request is complete
• Licensing information is complete
• Corresponding author information is complete
• The source code follows the LAMMPS formatting guidelines
• Suitable new documentation files and/or updates to the existing docs are included
• The added/updated documentation is integrated and tested with the documentation build system
• The feature has been verified to work with the conventional build system
• The feature has been verified to work with the CMake based build system
• A package specific README file has been included or updated
• One or more example input decks are included

Further Information, Files, and Links

[akohlmey]

Thanks for putting in the effort to submit a pull request. But why not simply define the following atom style variables:

variable mx atom mass*vx
variable my atom mass*vy
variable mz atom mass*vz

and then access v_mx, v_my, and v_mz instead of using this compute

compute c1 all momentum

with access through c_c1[1], c_c1[2], and c_c1[3]?
Is there any use case, where using this compute offers significant advantages the atom style variables?

[rupertnash]

Thanks for looking at this! However I think you missed that this compute will calc the total momentum of all the atoms in a group whereas (as I understand it and I'm new to LAMMPS...) an atom style variable is per-atom. Is that right?

[akohlmey]

Thanks for looking at this! However I think you missed that this compute will calc the total momentum of all the atoms in a group whereas (as I understand it and I'm new to LAMMPS...) an atom style variable is per-atom. Is that right?

you are correct, I missed that. however, that would just be one more line of input.

variable mx atom mass*vx
variable my atom mass*vy
variable mz atom mass*vz
compute c1 all reduce sum v_mx v_my v_mz

[rupertnash]

OK! Thanks for the comments Axel. I guess this is a risk with LAMMPS being so big.

However since the code now exists, and would save users some effort/learning, I'm still willing to tick off the remaining items above, if you think it's of value to the users. If you think on balance it's better to use the variable + reduce compute in script then I'm happy if you just close the PR.

[akohlmey]

OK! Thanks for the comments Axel. I guess this is a risk with LAMMPS being so big.

However since the code now exists, and would save users some effort/learning, I'm still willing to tick off the remaining items above, if you think it's of value to the users. If you think on balance it's better to use the variable + reduce compute in script then I'm happy if you just close the PR.

I am coming out neutral on this one. On one hand, this command allows to do with one line of input script, what otherwise requires four, on the other hand, it is additional code (albeit simple code) to maintain. I was asking for a particular use case for this command versus input script to tip the scales toward including it. If added, it should definitely go into the USER-MISC package, though.

I'll assign the PR to @sjplimp for a second opinion.

[sjplimp]

I think it's fine to add to USER-MISC as a new compute.

[sjplimp]

just move to USER-MISC, and state that in command doc page,
also edit the USER-MISC/README

[rupertnash]

In the last few commits I've moved the compute to USER-MISC, updated the docs, and added a test input script. I think this is now ready to go.

[sjplimp]

agreed: ready to go when you make the doc page edits. Thanks

[rupertnash]

Sorry @sjplimp - I don't follow which doc page needs updated. I've done

• doc/src/compute.txt
• doc/src/Commands_compute.txt
• doc/src/compute_momentum.txt (note about USER-MISC in restrictions similar to e.g. compute_entropy_atom)
• doc/src/computes.txt
• src/USER-MISC/README
  Let me know what I've missed!
  Cheers

[sjplimp]

I just meant the paragraph at the bottom of compute_momentum.txt that I flagged
in an earlier comment. It has some wrong grammar.

[rupertnash]

OK - I think I've got the grammar error- afraid I didn't see anything explicitly flagged. Also resolved the merge conflicts. Thanks.

[akohlmey]

@sjplimp this also seems to be ready and waiting to be approved and assigned to me

[sjplimp]

on -> of

[akohlmey]

made some cosmetic changes