Differential Forms in Algebraic Topology
Cohomology through differential forms
Description
There's a moment, early in any topology course, when someone draws a coffee cup morphing into a doughnut and announces that the two are, for the mathematician, the same object. One hole each. The gag lands, everyone nods, and then the trouble starts. Because the next question — how do we actually count the holes in something we can't see, a four-dimensional space, a knotted surface, a manifold given only by equations — is where the friendly cartoon runs out of road. Counting holes turns out to be one of the hardest bookkeeping problems in mathematics, and the machinery built to do it, algebraic topology, has a reputation for being beautiful from a distance and merciless up close.
In 1982 Raoul Bott, a topologist at Harvard, and his former student Loring Tu published a book that took a stubbornly different route into that machinery. It grew out of a first-year graduate course, and its wager was simple: instead of starting with the abstract definitions and hoping intuition catches up, start with something students already half-know from calculus — the differential form, the thing you integrate over a surface — and show that it quietly does the job of cohomology. Follow that one concrete object far enough, the book argues, and the whole apparatus assembles itself in front of you, motivated at every step.
The result, Differential Forms in Algebraic Topology, became one of the most-recommended graduate texts of its generation, precisely because it refuses to be forbidding. It is a book about holes, invariants, and the surprisingly deep fact that calculus on a space remembers the shape of the space. But it is also a book with an opinion about how hard ideas should be taught.
The question we’re asking : How does something as familiar as a differential form end up computing the deep shape of a space — and why build a whole textbook around that one bridge?What we’ll see : How calculus quietly becomes topology, and what that path reveals about learning a hard theory from the inside out.
Table of contents
01Chapter 1 — The problem with counting holes
The intuition is easy and the execution is brutal. A sphere has a kind of hole a disk doesn't; a doughnut has a different kind still. We feel this. The task algebraic topology sets itself is to turn that feeling into a number, or better, into an algebraic gadget attached to every space — one that stays the same when the space is bent, stretched, or deformed, and changes only when the shape genuinely changes. Such a gadget is called an invariant, and the whole game is finding invariants strong enough to tell spaces apart and computable enough to actually use.
The classical route is homology: you chop a space into simplices — triangles, tetrahedra, their higher cousins — and do careful bookkeeping about which pieces are boundaries of other pieces and which aren't. A loop that isn't the boundary of any disk inside the space is detecting a hole. It works, and it is one of the great constructions of twentieth-century mathematics. It is also, as anyone who has tried it knows, a grind of index-chasing that can obscure why any of it means what it means.
02Chapter 2 — Differential forms as a first draft of cohomology
A differential form, stripped of ceremony, is a thing you can integrate. On the plane, an expression like the one you integrate to get area or flux; on a manifold, the same idea done carefully so it makes sense no matter which coordinates you happen to use. What makes forms into a topological instrument is a single operation called the exterior derivative, written d, which generalizes gradient, curl, and divergence all at once. Apply d and you get a new form; apply it twice and you always get zero. That last fact, d-squared equals zero, looks like a technicality and is in fact the entire engine.
Because d applied twice vanishes, the forms split naturally into two families that matter. A form is closed if applying d to it gives zero — nothing is changing, in the relevant sense. A form is exact if it is itself the d of something else. Every exact form is automatically closed, thanks to d-squared being zero. The interesting question, the one that carries all the information, is the reverse: are there closed forms that are not exact? On a space with no holes, no — every closed form is exact. On a space with a hole, some closed forms escape, and each one is a witness to a hole the loop can't contract past.
03Chapter 3 — When one chart isn't enough
The trouble with forms is that they are local creatures. They live in coordinate patches, and most interesting spaces cannot be covered by a single patch — a sphere needs at least two, and things only get worse from there. A form defined nicely on one piece may disagree with the form on the overlapping piece next door. To do topology globally, we need a disciplined way to glue local data into global conclusions, tracking exactly how the pieces fail to match on their overlaps.
This is where the book introduces its second core machine, the Cech-de Rham complex. The idea, which goes back to Jean Leray and Henri Cartan in the mid-twentieth century, is to cover the space with well-chosen open sets and keep books not only on each set but on every intersection of pairs, triples, and so on. You now have two directions of bookkeeping running at once — the calculus direction, from the exterior derivative, and the combinatorial direction, from the pattern of overlaps. Weaving them together produces a double complex, a grid of vector spaces linked in two ways, and out of that grid the global cohomology emerges.
04Chapter 4 — The machinery that computes the invisible
The last stretch of the book confronts the two devices that give algebraic topology its reputation for difficulty, and it earns the right to them by having built everything concretely first. The spectral sequence is the more intimidating of the pair — a bookkeeping scheme for double complexes that computes cohomology in successive rounds, each round a correction of the last, until the answer stabilizes. Described cold, it is one of the most opaque objects a graduate student meets. Introduced as the natural organizing tool for the double complexes already in hand, it becomes almost inevitable, a calculating engine rather than a magic trick.
Characteristic classes are the other summit. These are invariants attached not to a space alone but to the bundles of extra structure that live over it — the way tangent directions twist as you move around a surface, for instance. The Euler class, the Chern classes, the Pontryagin classes: each is a cohomology class that measures a specific kind of twisting, and each can be pinned down, in Bott and Tu's telling, using the very differential forms the book started with. The abstract obstruction becomes something you can integrate.
05Conclusion
The coffee cup and the doughnut are still the same, one hole each, but by the end of Bott and Tu's book that fact has a full apparatus behind it. A differential form, the humble thing from a calculus course, turns out to remember the shape of the space it lives on; the gap between closed and exact forms counts the holes; and when a single coordinate patch won't do, the Cech-de Rham machinery stitches the local pieces into a global truth. The heavy tools that follow — spectral sequences, characteristic classes — arrive not as decrees but as answers to questions the earlier chapters have already made unavoidable.
