About

Contains the reproduction details for the publication on the GPR-dimer and its performance compared to Sella.

Reference

If you use this repository or its parts please cite the corresponding publication or data source.

Preprint

A more accessible form of the same publication is:

R. Goswami, M. Masterov, S. Kamath, A. Peña-Torres, and H. Jónsson, “Efficient implementation of gaussian process regression accelerated saddle point searches with application to molecular reactions,” May 18, 2025, arXiv: arXiv:2505.12519. doi: 10.48550/arXiv.2505.12519.

Data source

Rohit Goswami, Maxim Masterov, Satish Kamath, Alejandro Peña-Torres, Hannes Jónsson, Efficient implementation of gaussian process regression accelerated saddle point searches with application to molecular reactions, Materials Cloud Archive 2025.87 (2025), https://doi.org/10.24435/materialscloud:j6-zz

Usage

Remember to inflate the data using the materialscloud source before using the scripts in the repository. This can be done by running the following–assuming that the .xz files are in data/runs relative to the repository root:

# EON Dimer runs (w/ & w/o GP)
cp $GITROOT/data/runs/eon_runs.tar.xz $GITROOT/bench_runs/eon
cd $GITROOT/bench_runs/eon
tar -xf eon_runs.tar.xz && rm -rf eon_runs.tar.xz
# Sella
cp $GITROOT/data/runs/sella_runs.tar.xz bench_runs/sella
cd $GITROOT/bench_runs/sella
tar -xf sella_runs.tar.xz && rm -rf sella_runs.tar.xz

Structure

The repository itself is structured into code archives, benchmark runs, and scripts for analysis.

➜ tree -L 2
.
├── bench_runs
│   ├── calc_rundata.py
│   ├── check_dist_sim.py
│   ├── check_sim.py
│   ├── eon
│   ├── gen_allgpdat.sh
│   ├── gen_paired_res.py
│   ├── plot_gprdruns.py
│   ├── profiles
│   ├── readme.org
│   ├── rundata
│   └── sella
├── data
│   ├── sella_si
│   └── sella_si_data.zip
├── LICENSE
├── pixi.toml
├── readme.org
├── readme.txt
└── subrepos
    ├── ase
    ├── chemparseplot
    ├── eOn
    ├── IterativeRotationsAssignments
    ├── nwchem
    ├── pychumpchem
    ├── rgpycrumbs
    ├── sella
    └── spack

Where the data in the archives expands to locations within the benchmarks.

Each of the benchmarks consists of the following structure:

.
├── doublets
│   ├── 000
# .....
│   └── 234
└── singlets
│   ├── 000
# .....
    └── 264

Comprising of 500 systems.

EON Dimer runs with and without GP acceleration

# eon_runs.tar.xz
# $GITROOT/bench_runs/eon/runs
➜ tree -L 4 .
.
└── 1e8m
    ├── gprd
    │   └── final_gprd_wparam
    │       ├── doublets
    │       └── singlets
    └── idimer
        └── final_cgrot_lbfgs_dimer
            ├── doublets
            ├── readme.org
            └── singlets

Sella runs

# sella_runs.tar.xz
# $GITROOT/bench_runs/sella/runs
➜ tree -L 3 .
.
└── 1e8m
    └── final_sella
        ├── doublets
        └── singlets

License

MIT. Sub-packages have their own licenses.