Math goldmines: the free books that carry CS

Math resources have the worst discovery problem: the famous books are gates, and the actual on-ramps are free PDFs passed around like samizdat. Here are the ones worth the hype they never got.

The on-ramps

• An Infinitely Large Napkin — Evan Chen. A free ~1000-page tour of university math (groups through Galois theory, topology, measure, category theory) written for people who want the ideas fast, with honest signposts about what’s being skipped. There is nothing else like it: it optimizes for “what is this field and why,” which is exactly what an engineer raiding math needs.
• Mathematics for Computer Science — Lehman, Leighton, Meyer (MIT 6.042, free book + lectures). Proofs, induction, number theory, graphs, counting, probability — the discrete foundation under algorithms, taught with CS examples throughout. The proofs chapters alone fix the “I can code it but can’t prove it” gap.
• Linear Algebra Done Wrong — Sergei Treil, free. The title jokes at Axler’s famous book: this one keeps determinants and matrices front and center, which is the version computer people actually use. Rigorous, terse, and free forever.

Interactive and reference

• Seeing Theory — Brown’s interactive probability visualizations. Twenty minutes here fixes more intuition about distributions and inference than a week of slides.
• The Matrix Cookbook and matrixcalculus.org — the identity reference and the derivative calculator. Between them, matrix calculus stops being a hazing ritual.

Applied math in disguise: cryptography

• A Graduate Course in Applied Cryptography — Dan Boneh and Victor Shoup, free. The definitive modern crypto text: security definitions and proofs done properly, from ciphers through zero-knowledge. It’s the book Stanford’s famous course grew into, and it costs nothing.
• Cryptopals — the gym for the book above: 48+ challenges where you break real crypto (ECB cut-and-paste, CBC padding oracles, nonce reuse) with nothing but code. The single most effective “learn by attacking” curriculum in any field; finishing set 3 changes how you read every security adv isory forever.

How to raid math as an engineer

Don’t read linearly; raid with a purpose. Napkin to scout a field, 6.042 when proofs feel shaky, Treil when linear algebra needs rebuilding under ML, Boneh–Shoup plus cryptopals when you want the math to fight back. A theorem you used beats ten you highlighted.