How can I install deepmd-kit with all PLUMED modules functional using conda?

[Dustglaxy]

Dear developers,

When I use the old installation command for an A800 GPU, all PLUMED modules are enabled. However, when I use the latest conda installation command for an H100 GPU, some modules are turned off. How can I install deepmd-kit with conda while ensuring that all PLUMED functions are enabled?

old command:
conda create -n deepmd deepmd-kit=2.2.9=*cpu libdeepmd=2.2.9=*cpu lammps horovod -c https://conda.deepmodeling.com -c defaults

new command:
conda create -n deepmd deepmd-kit lammps horovod -c conda-forge

Thanks for your time.
Xiangyang

[njzjz-bot]

The enabled PLUMED modules are determined when the PLUMED binary package is built, not by the deepmd-kit package. A conda command cannot enable modules that were disabled in the conda-forge plumed build. The current conda-forge recipe runs PLUMED's normal configure without --enable-modules=all, so optional modules remain at their recipe defaults; conda-forge's LAMMPS package then links against that fixed PLUMED build.

If the conda-forge module set is insufficient, the reliable route is:

1. Create a conda environment containing DeePMD-kit and the build/runtime dependencies you need.
2. Build PLUMED from source into that environment with --enable-modules=all (or, preferably, explicitly enable only the required optional modules). Modules with external dependencies still require those dependencies to be installed.
3. Build LAMMPS against that PLUMED installation and add DeePMD-kit through plugin mode, the precompiled C library, or a source build.

Alternatively, maintain a custom conda PLUMED/LAMMPS recipe with the requested module set. Reinstalling deepmd-kit alone cannot change the PLUMED feature list. You can inspect the actual installation with plumed info --configuration and the module-specific commands before rebuilding.

Coding agent: Codex
Codex version: codex-cli 0.149.0
Model: gpt-5.6-sol
Reasoning effort: xhigh