New features to analyse Lammps outputs

[JaviFdezT]

LAMMPS output files provide information about CPU usage, MPI & OPENMP tasks and MPI task timing. This information can be useful to predict the resources necessary to run future simulations or run performance tests.

This new feature will be implemented in a new branch (dev). The branch has to be created by the owner or we need additional permissions. At the moment, we cannot create a pull request for this purpose.

[lmhale99]

I added a "newlog" branch and updated the Log file with the one I sent you.

Are you able to fork atomman to your account by clicking on the "fork" button at the top right? The typical way to contribute is

1. Fork to your account.
2. Use git clone to download your fork. This should set the "origin" remote to be your fork.
3. If you use gitlab, create a new repository there. Set a different remote name to point to that repository and git push to it.
4. Forking should also be possible in gitlab allowing all users to make their own changes and have their own copies.
5. After making changes to the local copy, git commit saves and marks the changes.
6. git push sends committed changes in one branch to one remote location. The github fork does not have to be updated until you want/have permission.
7. Once the changes are pushed to a github/gitlab fork, a pull request can be made to send the changes to a different fork.

[JaviFdezT]

Thanks, Lucas.

We pushed these changes into this new branch. The pull request is made now.

[lmhale99]

Thanks, Javier. I'll take a look at it today.

[lmhale99]

I merged the new Log.py file in. I then did a quick read through of the code mostly for clarity and cleanup. The actual parsing of the performance data should be untouched.

Changes:

• I renamed all parameters using perf or perfo to performance.
• The headers and footers parameters were renamed to start with thermo or performance accordingly.
• Comment lines and the unused Log.__performance_data list were removed.
• model() was removed. I haven't used it in years and I don't think the documentation has even mentioned it for almost as long.
• read_simulations() was split into __read_thermo() and read_performance(). Simulations are created with __read_thermo and performance data is then added afterwards.

The last point allows for the thermo data of incomplete runs when LAMMPS is externally stopped to still be collected. In that case, the final simulation won't have performance data because it won't be in the log output.

Please check if my changes inadvertently broke anything.

[JaviFdezT]

Thanks, Lucas. It works. Please, feel free to create a new version.

[JaviFdezT]

Hi Lucas. Would it be possible to upload this new version to PyPi so that it can be downloaded with pip?

[lmhale99]

Hi Javier,

I'm going through and updating the documentation for all the new additions. The new version should release next week.

[lmhale99]

Version 1.3.7 is out.

[JaviFdezT]

Thanks a lot, Lucas. It has been a great pleasure to collaborate with you. I am closing this thread now.