A Map of Mathematics: Twenty-Five Areas and What Depends on What

Mathematics is not a ladder with one top rung. It is a landscape of connected regions, some of which you must cross before others. This page walks through twenty-five areas in a sensible learning order, says what each depends on, and notes where philosophy and engineering echo it. For the shorter orientation, see Mathematics: Start Here.

How the ordering works

Plato has Socrates argue in the Republic (Book VII) that arithmetic and the sciences built on it turn the mind from changing things toward what is stable, and prepare it for dialectic. That is a philosophical claim about education, not a mathematical result. We borrow the spirit: mathematics is the study of structure, and structure is what lets thought travel between domains.

The order below ranks areas by leverage for rigorous thought and technical work, not by prestige. A later position does not mean lesser mathematics. Several areas can be studied in parallel, and the dependencies matter more than the numbers. Provisional This is one defensible ordering, not the unique correct one.

The foundation chain

Arithmetic and algebra (arithmetic, Algebra 1, Algebra 2). Reliable symbolic manipulation. See Algebra: The Skills That Carry Everything.
Geometry and trigonometry. Geometry plus algebra leads to trigonometry. See Geometry, Euclid and Trigonometry and Euclid's postulates.
Precalculus and calculus. Functions, limits, rates and accumulation. See Calculus and Beyond: Change, Structure and Uncertainty.

What calculus is not required for

A common mistake is to treat calculus as the gate to everything. It is not. Established Elementary statistics, discrete mathematics, introductory linear algebra and elementary proof need algebra and careful reading, not calculus. Calculus-based probability, mathematical statistics, real analysis, differential equations and most of optimization do need it. Check each text's own stated prerequisites.

The trunk: reasoning, structure, computation

Proof and logic. The grammar of valid inference; needs only comfortable algebra. See Proof and Precise Reasoning: From Arguments to Theorems.
Discrete mathematics. Sets, relations, counting, induction, recurrences. Needs proof habits. See Discrete Mathematics and Graphs: Counting, Relations and Networks.
Algorithms and data structures. The cost of a method before you run it. Needs discrete maths and some programming.
Probability. Uncertainty as a model. Elementary probability needs counting; the calculus-based version needs calculus.

Uncertainty, data and representation

Statistics. Separating signal from noise; elementary courses need algebra and graph reading.
Linear algebra. Vectors, matrices, transformations. Needs algebra and proof familiarity.
Relational algebra and data modelling. Sets and functions applied to tables, keys and joins. Needs set basics.

Networks and decisions

Graph theory. Vertices, edges, paths, cuts. Needs proof and discrete maths.
Optimization. Objectives and constraints. Linear programming needs linear algebra; convex optimization needs multivariable calculus as well.
Information theory. Entropy, compression, noise. Needs probability.
Queueing and stochastic processes. Waiting lines, bursts, why averages mislead. Needs probability and some calculus.
Time series. Trends, seasonality, noise. Needs statistics.

Change and feedback

Numerical methods. What computers actually approximate, and where error enters.
Ordinary differential equations. Continuous change. Needs calculus and, for systems, linear algebra.
Control theory. Feedback, stability, regulation. Needs ODEs and linear algebra.
Dynamical systems and complex adaptive systems. Thresholds, attractors, sensitivity to initial conditions, agent-based models.

Judgment under uncertainty

Bayesian decision theory. Joining evidence, prior belief and action.
Causal inference. What can and cannot be concluded from observed data.
Game theory. Strategic interaction and incentives.

Advanced bridges

Formal languages and computability. What machines can and cannot compute.
Cryptography and number theory. Modular arithmetic and hardness assumptions.
Spectral methods. Eigenvalues applied to graphs and data.

For areas 12 to 14 and 16 to 19, see Systems Thinking: Feedback, Queues, Information and Dynamics; for 20 to 22 in practical form, see Mathematics for Better Decisions and Judgment Under Uncertainty: Probability, Causation and Forecasting.

Philosophical parallels (analogy, not proof)

These are meant to enrich understanding. Provisional at best; none is a theorem.

Eigenvectors and Platonic form. An eigenvector is a direction a transformation only stretches. It is tempting to call it an "essence" of the transformation. The mathematics is exact; the philosophical reading is a metaphor.
Topology and essence. Topology asks what survives bending and stretching. What survives is a precise set of invariants, not an essence in the philosophical sense.
Bayes and epistemology. Bayes' rule is a theorem about conditional probability. As a model of rational belief revision it is influential and contested; see Judgment Under Uncertainty: Probability, Causation and Forecasting.

Engineering parallels

These are literal, historical connections. Established

Boolean algebra and switching. Boole published his algebra of logic in 1854; Shannon showed in 1937-38 that it describes relay and switching circuits.
Graphs and routing. Euler's 1735-36 treatment (published 1741) of the Konigsberg bridges founded graph theory; shortest-path and spanning-tree algorithms now underlie routing.
Queues and load. Waiting-line mathematics, begun by A. K. Erlang for telephone exchanges in the early 1900s, explains why delays rise sharply as utilization approaches capacity.
Eigenvectors and ranking. The original PageRank method of Brin and Page (1998) is a principal-eigenvector computation on the web's link graph.

Common traps

Network topology is not topological spaces. The word "topology" in computer networking means the arrangement of devices and links, modelled with graphs. Topology in mathematics studies spaces and continuity. They share a name and a loose history, not a method.
Simulation is not proof. A simulation is evidence about a specified model under chosen parameters. A proof covers every case in a stated domain. Simulations suggest conjectures and test models; they do not establish theorems.
There is no final level. Graduate analysis, algebra, topology and research mathematics lie beyond any introductory sequence. Category theory is an optional later field, not a required summit.

Choosing your next domain without collecting a list

Before adding anything, answer four questions in writing: What concrete question do I want to answer? Which prerequisites does the best text for it state, and have I demonstrated them? What will I produce at the end? What will I drop to make room? Then follow a structured pace such as Twelve-Week Arcs and the Capstone Project Method. The wider aims of the library, and how mathematics sits with reading and making, are set out in Liberal Arts and Reasoning: Start Here; the classical seven arts are in The Seven Liberal Arts, Hands-On: Logic, Proof and the Quadrivium.

Try this

Pick three areas from the list. For each, name the one prerequisite you are least sure of and write a five-question check for it.
Take any real system you know (a queue at a shop, a commute, a household budget) and name which three areas on this map would describe it best, and what each would capture or miss.
Write one sentence distinguishing "network topology" from "topological space", then check it against a textbook definition of each.