Improve Checkpointing #69
Conversation
|
Since the patch introduces new features, we will need to update the documentation accordingly (added to the todo list above), including this section and this section. Changes for these places can be made in |
|
BTW, you can check the changes made in the documentation by running |
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #69 +/- ##
==========================================
+ Coverage 96.00% 96.01% +0.01%
==========================================
Files 11 11
Lines 2576 2583 +7
==========================================
+ Hits 2473 2480 +7
Misses 103 103 ☔ View full report in Codecov by Sentry. |
| @@ -322,16 +332,20 @@ def main(): | |||
| if (i + 1) % REXEE.n_ckpt == 0: | |||
| if len(REXEE.g_vecs) != 0: | |||
| # Save g_vec as a function of time if weight combination was used. | |||
| np.save('g_vecs.npy', REXEE.g_vecs) | |||
| np.save(args.g_vecs, REXEE.g_vecs) | |||
There was a problem hiding this comment.
So the input argument g_vecs was intended to be for specifying an input file like g_vecs.npy for extending the simulation (same for the rep_trajs), but I do think it makes sense to have them specify the output files to be saved as well. I'll tweak the help message of the argument parser.
There was a problem hiding this comment.
Hmm actually I think the current help message is fine.
This PR adds a new checkpoint file (default=equil.npy) which is utilized for weight updating REXEE and MT-REXEE simulations to allow for simulations to restart when one or more simulations have equilibrated weights, but not all simulations. Additionally, this PR fixes a minor bug in computing g_vec which only affects MT-REXEE simulations.