CoDe-Mat Lab @ CSIR-IICT

handbook / Resources

Books

Some useful books/resources in relevant fields (personal favorites / things that seemed interesting):

Atomistic Simulations

  • The Atomic Simulation Environment (ASE) documentation — the standard Python library for setting up, running, and analyzing atomistic simulations; see also the reference paper, Larsen et al., J. Phys.: Condens. Matter 29, 273002 (2017).
  • dft-book — John Kitchin — a free, worked-example-driven book on modeling materials using DFT with VASP and ASE; good for going from “I know the theory” to “I can actually run and interpret a calculation.”

Machine / Deep Learning

Electronic Structure Theory

  • Electronic Structure: Basic Theory and Practical Methods — Richard M. Martin (Cambridge University Press) — the standard graduate-level reference; covers DFT foundations, the Kohn–Sham approach, and practical methods in depth. Not free, but very likely available through your institution’s library.
  • A Mathematical Introduction to Electronic Structure Theory — Lin Lin & Jianfeng Lu (SIAM) — a more mathematically-oriented, compact treatment of Kohn–Sham DFT and linear response theory; good complement to Martin’s book.

Important papers

A short, non-exhaustive list of papers worth reading early if you’re working at the intersection of atomistic simulation and machine learning:

  • Machine learning force fields (review) — Unke, Chmiela, Sauceda, Gastegger, Poltavsky, Schütt, Tkatchenko, Müller, Chemical Reviews 121(16), 10142–10186 (2021). DOI: 10.1021/acs.chemrev.0c01111 — a thorough, step-by-step overview of how ML force fields are built, tested, and where the field is headed. Good starting point for anyone new to MLFFs.
  • Machine learning for molecular simulation (review) — Noé, Tkatchenko, Müller, Clementi, Annu. Rev. Phys. Chem. 71, 361–390 (2020). DOI: 10.1146/annurev-physchem-042018-052331
  • The Atomic Simulation Environment — Larsen et al., J. Phys.: Condens. Matter 29, 273002 (2017). DOI: 10.1088/1361-648X/aa680e
  • MACE (equivariant message-passing MLIPs) — Batatia et al., NeurIPS 2022. arXiv:2206.07697 — one of the more widely-used modern architectures for interatomic potentials; see also the MACE GitHub repo and tutorials.
  • NequIP (E(3)-equivariant GNN potentials) — Batzner et al., Nature Communications 13, 2453 (2022). DOI: 10.1038/s41467-022-29939-5

(add more landmark/must-read papers here as your group identifies them, organized by subtopic)

YouTube channels

(add recommended channels here — e.g., for tutorials, seminar recordings, conference talks. A few generally useful starting points to consider adding: DFT/electronic-structure code tutorial channels from your group’s software of choice — VASP, Quantum ESPRESSO, GPAW, etc. — and conference recordings from Psi-k, CECAM, or APS March Meeting sessions.)

Coding practices

(add group conventions/resources here — e.g., version control habits, code review norms, style guides, use of Jupyter notebooks vs. scripts, environment management.)

GitHub repos

Curated/aggregator lists that are useful starting points for discovering tools in this space:

  • best-of-atomistic-machine-learning — a large, actively maintained, ranked list of atomistic ML projects (potentials, datasets, DFT-ML surrogates, educational resources, community resources), organized into ~22 categories.
  • awesome-ml_iap — curated list specifically of machine-learned interatomic potential packages (NequIP, MACE, PyXtal FF, REANN, etc.).
  • ASE ecosystem list — packages that integrate with or extend ASE.

Specific tools/packages worth knowing:

(add your group’s own repos, forks, and commonly used internal tools here.) ~

Travel & Funding →