Molecular and Materials Modelling#
A Computational Materials Science Notebook Companion — by Raymond Amador.
This is a companion to the graduate course Molecular and Materials Modelling, taught at ETH Zürich. It teaches the methods of computational materials science — Monte Carlo and molecular dynamics, electronic structure, reaction paths, and machine-learning descriptors — by computation and visualisation: every idea is something you run, plot, and check, not just read.
How these notebooks are built#
Each notebook is self-contained and follows the same shape, so the series reads as one coherent work:
a short theory in brief — a genuine review of the governing equations, not a textbook chapter, with full derivations deferred to the literature;
a sequence of guided exercises, each stating exactly what to compute and why, with a labelled setup figure where a physical configuration is described;
an independent validation at the end of every exercise. This is the rule that matters most: a result is not finished until it is checked against something the computation did not assume — an exact limit, a conservation law, a closed-form result (Onsager’s \(T_c\)), or an exact enumeration. The checks print a ✓/✗ line and fail the build if the physics is wrong, so the published site is trustworthy by construction.
A word on reading a check: a ✗ means the output did not match what the validation expected. That can be a real error, a different-but-valid convention, or simply Monte Carlo noise against a tight tolerance — so treat it as a prompt to locate the discrepancy, never an automatic verdict. We also validate only what the code honestly reproduces; where a phenomenon needs physics beyond a clean, fast implementation, we show it qualitatively and say what is missing rather than overclaim.
Animations appear only where motion or evolution is the point; figures and equations are numbered continuously and cross-referenced (“Fig. 7”, “Eq. 12”). Where the physics is dimensionless we use natural units and say so.
How to use them#
Every page has a download button and launch buttons (Binder/Colab) in the toolbar — take a notebook and run it yourself. The published notebooks ship without the worked solutions: you see each problem, the resulting figures, and the validation outcomes, but not the answer code. If you would like the reference solutions — to teach from, or to check your own work — the contact is in the footer of any page.
Organisation, and continuity with the course#
The notebooks are grouped thematically into volumes rather than in strict
lecture order, because related methods (say, Monte Carlo in equilibrium and in
real time) belong together. Continuity with the original course is preserved a
different way: every notebook states the lecture it develops, with the number
explicit, both in its banner and in its overview. The FS 2023 course numbered
its lectures and exercises together (Lecture/Exercise 1 … 13); the full map lives
in manifest.yml, and each volume’s landing page lists the lectures it draws on.
Acknowledgement and provenance#
These materials are based on the lecture and exercise materials of Molecular and Materials Modelling at ETH Zürich and Empa (FS 2023), developed by Prof. Dr. Daniele Passerone (lectures), Dr. Carlo Pignedoli, and the author (exercises) — the author having led the exercise sessions for three years. Each notebook is largely, though not always verbatim, based on that material: where a topic was Dr. Pignedoli’s, the author adopts, reimplements, and rephrases it; elsewhere the author’s exercises develop Prof. Passerone’s lectures. Here it is all synthesised, expanded, and restyled into a single companion. The same credit appears in the footer of every page.
To cite this companion, use its DOI,
10.5281/zenodo.21418786 (all versions;
full metadata in the repository’s CITATION.cff).
A note on how this companion was written#
These notebooks were written in collaboration with a generative AI assistant (Anthropic’s Claude), and the fact belongs in the open, in a series whose central rule is that nothing is trusted unchecked. The scientific content and its provenance are exactly as credited above. The design of the companion is the author’s: the thematic organisation, the fidelity-first rule that the course’s real simulation inputs and outputs are shown and analysed rather than reinvented, the validation discipline, and the per-notebook specifications from which every page was built. The assistant drafted prose and code to those specifications; the author reviewed, corrected, and approved every page, and the final word on each is a human’s. And the rule that matters most applies to the authors too: neither was taken on trust — every exercise ends by checking its result against something the computation did not assume, so the reader never has to take anyone’s word, human or machine, for a single number.