A field guide for those heading to the frontier

The Research Scientist's Math of AI

At research level, the math isn't under the hood — it is the product. New architectures, losses and training regimes are mathematical objects first and code second. This guide maps the five mathematical territories the 100% is made of: what each field is, which AI breakthroughs it produced, and what mastering it lets you do.

Where the line is

The difference between the 50% and the 100% is not more chapters — it's a change of role.

What 100% means

Math as the instrument of invention:

  • Invent objectives and prove they behave (convergence, stability)
  • Read a paper's theory section as the main course, not the garnish
  • Move between fields fluently — a diffusion paper is stochastic calculus + score matching + numerics in one
  • Recognize which mathematical tool a research problem is really asking for

Built on the other tracks

The scientist track presumes fluency in everything below it:

Each territory page goes deeper in three levels — core objects, working theory, and the frontier where it powers current research.

The map: five territories, and the breakthroughs they produced

Every landmark result in AI is a flag planted in one of these fields.

Fields ↔ breakthroughs

1 · Analysis & matrix calculus Jacobians · Hessians · smoothness Backpropagation itself autodiff = Jacobian products NTK · sharpness (SAM) training dynamics as analysis 2 · Stochastic processes Markov chains · SDEs · MCMC Diffusion models generation = a reversed SDE Why SGD works · RL stochastic approximation, MDPs 3 · Statistical learning theory generalization · capacity Scaling laws · Chinchilla how big, how much data Double descent · grokking the modern generalization puzzles 4 · Optimization & game theory convexity · minimax · transport Adam · WGAN optimal transport fixed GANs TRPO / PPO (RLHF) trust regions, policy games 5 · Information theory entropy · KL · mutual information VAE · CLIP (InfoNCE) variational bounds, contrastive MI DPO · compression view alignment as KL-regularized RL
Read the map right-to-left too: pick any famous result, and its "theory section" is a guided tour of exactly one or two of these territories.

The five territories — each with its own deep-dive page

1 · Analysis & Matrix Calculus

Derivatives grown up: Jacobians, Hessians, curvature and smoothness — the field in which backprop is a theorem and training dynamics become equations you can solve.

  • Backprop as products of Jacobians
  • Curvature, conditioning, sharp vs flat minima
  • NTK: infinitely wide networks, solved
Enter the territory →

2 · Probability Theory & Stochastic Processes

Randomness with a time axis: Markov chains, Langevin dynamics and SDEs — the mathematics that turned "add noise, then learn to remove it" into the diffusion revolution.

  • Forward / reverse SDEs — diffusion, derived
  • MCMC & Langevin sampling
  • SGD as a stochastic process
Enter the territory →

3 · Statistical Learning Theory

The deepest question in the field: why does minimizing training loss work on data the model has never seen? Classical bounds, and the modern puzzles that broke them.

  • Capacity, VC dimension, uniform convergence
  • Double descent & benign overfitting
  • Neural scaling laws & compute-optimal training
Enter the territory →

4 · Optimization & Game Theory

What convergence even means: rates, conditioning, non-convex landscapes — and what happens when two networks optimize against each other.

  • Convexity, rates, acceleration, Adam's theory
  • Minimax dynamics: why GAN training orbits
  • Optimal transport — the WGAN story in full
Enter the territory →

5 · Information Theory & the Geometry of Learning

Bits as the currency of learning: entropy, KL and mutual information — the field where compression, generation and alignment turn out to be the same subject.

  • KL asymmetry: mode-covering vs mode-seeking
  • ELBO: the variational workhorse (VAE, diffusion)
  • InfoNCE, information bottleneck, DPO
Enter the territory →

Quick orientation: territory → landmarks → what mastery buys

TerritoryLandmark resultsWhat mastering it lets you do
Analysis & matrix calculusBackprop, NTK, SAMDerive gradients for anything; analyze training dynamics
Stochastic processesDDPM, score-based SDEs, DPM-SolverDesign generative processes and samplers
Statistical learning theoryScaling laws, Chinchilla, double descentPredict what more data/params/compute will buy
Optimization & game theoryAdam, Nesterov, WGAN, PPOInvent optimizers; stabilize adversarial & RL training
Information theoryVAE/ELBO, InfoNCE, DPODerive objectives instead of guessing them
How the territories connect: a single modern paper routinely crosses three of them — DPO is information theory (KL) applied to a game-theoretic objective (preferences) with an analysis-grade derivation. The 100% isn't five separate subjects; it's the fluency to move between them mid-thought.
← All field guidesArchitect · Engineer · Scientist