Add nix and conda packages

[knc6]

Add nix packages here

Most of these packages have conda

• [ ] jarvis-tools
• [ ] alignn
• [ ] quantum-espresso
• [ ] gpaw
• [ ] atomvision
• [ ] chemnlp
• [ ] tb3py
• [ ] atomqc
• [ ] jarvis_leaderboard
• [ ] qmcpack
• [ ] kgcnn
• [ ] elemnet

and others.

For complete list grep "software_used" jarvis_leaderboard/contributions/*/metadata.json

[john-bonini]

I checked the status of the packages on this list which are not developed at nist:

Nixpkgs already has quantum-espresso and gpaw.

The NixOS-QChem overlay has qmcpack.

kgcnn is in pypi, so maybe we could get away with using poetry2nix, but that should be tested.

elemnet only has a requirements.txt so probably needs to be packaged.

[john-bonini]

While I'm at it I hacked together this one-liner: grep "software_used" jarvis_leaderboard/contributions/*/metadata.json | sed -E 's|.*"([^"]*)"[^"]*$|\1|' | sed "s|,\s*|\n|g" | sort | uniq | sed -E "s|(.*)|- [ ] \1|", which outputs the list below (where I've manually started checking things off):

• [ ] abinit
• [-] abipy*
• [-] aflow*
• [-] alignn*
• [ ] allegro (too many things are called this)
• [X] atomman
• [ ] cdvae
• [ ] ChatGPT_May24_version
• [X] ev.x (assuming it is in quantum-espresso package)
• [X] gromacs
• [-] iprPy*
• [-] jarvis-tools*
• [-] jarvis-tools==2022.9.16*
• [-] kgcnn*
• [X] lammps
• [X] LAMMPS
• [-] m3gnet*
• [-] matminer*
• [X] matplotlib
• [ ] nexus (too many things are called this)
• [X] numpy
• [-] oqmd*
• [X] periodictable
• [X] plumed
• [X] pymatgen
• [-] py-xgboost-gpu*
• [X] qmc-fit† (assuming it is in qmcpack)
• [X] qmcpack†
• [ ] qmcpack v.3.14.0 (I see overlay has 3.12.0, 3.16.0, 3.17.1 with minimal changes so likely easy)
• [X] quantum espresso
• [X] quantum espresso v.7.0†
• [X] scipy
• [X] tensorflow
• [X] ThreeBodyTB.jl‡
• [X] torch
• [-] torch-cluster*
• [-] torch-geometric*
• [-] torch-scatter*
• [-] torch-sparse*
• [ ] VASP 6.0

Above I've done the following: † indicates that nix derivations are in NixOS-QChem overlay. ‡ is for a julia package which should be able to be installed with julia.withPackages in a nix derivation.

• [-] package* indicates that package is on pypi so something like poetry2nix may work.1