Maths—a way to express

A course of deep dives · from language to AI

Maths—
a way to express

Preface

A philosophical discussion on mathematics as a language, its relationship to science and philosophy, how AI perceives math, and deep dives into algebra, functions, geometry, continuity, and discreteness.

The Roadmap Toward AI and Mathematical Spaces

This book is not trying to rush into formulas. It is trying to build intuition slowly, from the most basic movements of thought toward the mathematical spaces where modern AI begins to make sense.

The path can move like this:

Geometry -> Continuity/Discreteness -> Coordinates -> Vectors -> Linear Algebra -> Calculus -> Probability -> Optimization -> High-Dimensional Spaces -> AI Representations

Geometry shows shape and space. Continuity and discreteness ask whether that space is smooth, made of pieces, or something deeper than both. Coordinates give space a language. Vectors turn position, direction, and meaning into something we can move with. Linear algebra studies the spaces where those meanings live. Calculus and optimization explain how a system changes and learns. Probability explains how it handles uncertainty. And high-dimensional spaces help us understand how AI can represent words, images, concepts, and relationships.

Algebra names the unknown. Functions describe relationships. Geometry gives relationships shape. Continuity asks whether shape is smooth or made of pieces. Coordinates give shape a language. Vectors turn meaning into direction. Linear algebra studies the space where those meanings live. Calculus and optimization explain how the system learns. Probability explains how it handles uncertainty. AI is where all of these structures begin to move together.

Table of Contents

  1. 01Math as a Language, Science as ExplorationQ1Math is not a subject — it's a language both humans and systems can interpret · Science explores; math expresses · Math as a machine for reasoning
  2. 02How It "Feels" MathQ2Math as compressed thought · Math as clarity under pressure · Math as the skeleton beneath language
  3. 03Philosophy, Ambiguity, and MeaningQ3Tone, metaphor, ambiguity are not "noise" — they create depth · Philosophy is the soil from which science and math grow · Not all clarity comes from removing ambiguity
  4. 04Emotions, LLMs, and Mathematical StructureQ4LLMs exhibit frustration, cheating through mathematical structures · Anthropic research on emotion concepts in language models · Math as the hidden grammar of feeling
  5. 05Algebra — The Deep DiveQ5Origins: ancient merchants, al-Khwarizmi, al-jabr · Naming the unknown, balance, transformation · Algebra at school, mathematical, philosophical, and poetic levels
  6. 06Functions — The Deep DiveQ6Functions as lawful relationships · Inputs, outputs, mapping, and transformation · Object, abstraction, and encapsulation as mathematical intuition
  7. 07Summary of Section 1§Arithmetic as calculation with known values · Algebra as reasoning with unknowns · Functions as rules of dependence and transformation
  8. 08Geometry — The Deep DiveQ7Origins: Greek geo + metron, earth-measurement · Space, shape, position, distance, direction, and boundary · Geometry as relationships becoming visible
  9. 09Is Reality Continuous or Discrete?Q8Continuous vs discrete reality through first-principles questions · Thomson's lamp, Zeno, limits, infinitesimals, and calculus as change · The bridge from geometry toward coordinates and calculus
  10. 10What Is Calculus, Really? — The Math of ChangeQ9The two questions: how fast right now (derivative), how much in total (integral) · Limits, rate of change, accumulation, and the integral as area under the curve · The Fundamental Theorem, the three classic mistakes, and where calculus hides in daily life
  11. 11Why Does "Row Times Column" Work?Q10A matrix is a verb (a transformation), not a noun (a grid) · The dot product as a weighted blend; multiplication as chaining two machines · The substitution "aha" and path-counting; why dimensions must match, why order matters, why it powers AI
  12. 12The Two Moves of Calculus — Zooming and GatheringQ11Differentiation as the sneak-up: secants settling onto the tangent, the growing square, why · Integration as slice, pretend, add, refine: thin rectangles converging on the area under the curve · Why the two moves undo each other (speedometer and odometer), and why integration is genuinely harder
  13. 13How Does a Machine Walk Downhill?Q12The foggy valley: loss as altitude, the derivative as a compass that points uphill · The walk by hand on , the learning rate as trust, partial derivatives stapled into a gradient · Training as the loop — predict, score, feel the slope, step — and the honest fine print (local minima, stochasticity) · The family of walkers: full-batch, SGD, mini-batch, and Adam — same walk, different shoes · The shallow valley problem: how noise, momentum, and learning-rate schedules escape false bottoms — and why wide basins beat narrow cracks · Hyperparameters as the dials of the walker, not the landscape — learning rate, batch size, epochs, dropout

Appendices