Lammps template for MD+MC - #1149
Conversation
for more information, see https://pre-commit.ci
Vibsteamer
left a comment
Vibsteamer
left a comment
There was a problem hiding this comment.
The attatched param.json together with the template now dump-and-check in one loop_step:
- 50 MC snapshots from the last MD_snapshot in the last loop, many of them are repeated images as being not accepted by Metroplis criterion.
- then, 100 MD snapshots from the last accepted MC_snapshot.
the sampling weights (then candidates and labelds) on physical structures could be correspondingly biased,
Please make sure this is of your expectation.
|
|
||
| # read in the potentia | ||
| pair_style deepmd ../graph.001.pb ../graph.000.pb ../graph.003.pb ../graph.002.pb out_freq ${THERMO_FREQ} out_file model_devi.out | ||
| pair_coeff |
There was a problem hiding this comment.
pair_coeff
--> pair coeff * *
to be compatible with the latest version of lammps
There was a problem hiding this comment.
Verdict: Changes requested. The template contains a reproducible LAMMPS syntax error and cannot run. Previously reported issues involving pair_coeff and the sampling weight of repeated MC configurations are also unresolved. The branch is old and has no checks; I recommend rebuilding this example on the current template system with a smoke test.
Note: The Codex quota is about to reset, so I am using the remaining tokens to review all open PRs in this repository.
Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh
| run ${MC_NSTEPS} | ||
| unfix mc1 | ||
|
|
||
| velocity all create ${TEMP_MD} ${RANDOM_SEED}" |
There was a problem hiding this comment.
[Blocking] The extra quote after the random seed makes LAMMPS fail immediately with Unmatched double quote in command; I reproduced this with LAMMPS 22 Jul 2025. Please change it to:
| velocity all create ${TEMP_MD} ${RANDOM_SEED}" | |
| velocity all create ${TEMP_MD} ${RANDOM_SEED} |
The MC seed is also hard-coded to 500 and should be parameterized separately. In addition, dumping after rejected swaps produces repeated configurations, so the intended sampling weight should be documented or the frames should be deduplicated.
3 participants
Lammps template for MD+MC