Chapter 18: Category Theory & Quantum Field Theory
1945–present
Category Theory: Abstract Nonsense Becomes Profound Structure
Samuel Eilenberg and Saunders Mac Lane invented category theory in 1945 as a language for describing natural transformations between functors in algebraic topology. At first it was dismissed by many mathematicians as “abstract nonsense” — a collection of definitions with no content. This judgment proved spectacularly wrong.
Category theory asks: rather than studying mathematical objects by their internal structure, what can you learn from how they relate to each other? A category\(\mathcal{C}\) consists of:
- • Objects: the things being studied (sets, groups, topological spaces, Hilbert spaces, ...)
- • Morphisms: structure-preserving maps between objects (\(f: A \to B\))
- • Composition: morphisms compose associatively (\(g \circ f: A \to C\))
- • Identity: each object has an identity morphism \(\text{id}_A: A \to A\)
The power of this abstraction is that the same categorical structure appears in wildly different mathematical contexts — and when it does, theorems transfer automatically. Category theory is the mathematics of mathematical structure itself.
Functors and Natural Transformations
A functor \(F: \mathcal{C} \to \mathcal{D}\) is a structure-preserving map between categories: it sends objects to objects, morphisms to morphisms, preserves composition and identities. Functors are the morphisms of the category of (small) categories.
A natural transformation \(\eta: F \Rightarrow G\) between two functors \(F, G: \mathcal{C} \to \mathcal{D}\) is a coherent family of morphisms \(\eta_X: F(X) \to G(X)\) for each object \(X\), such that for every morphism \(f: X \to Y\):
\( G(f) \circ \eta_X = \eta_Y \circ F(f) \)
Eilenberg and Mac Lane invented categories specifically to define natural transformations: they needed to say precisely what it means for a construction in topology to be “natural” (independent of arbitrary choices). The concept turned out to be one of the most generative in all of mathematics.
TQFT: Atiyah's Axioms (1988)
Michael Atiyah (1929–2019) gave the first mathematical definition of a topological quantum field theory (TQFT) in 1988. His axioms express TQFT in purely categorical language: a TQFT is a symmetric monoidal functor:
\( Z : \mathbf{Cob}_n \longrightarrow \mathbf{Vect}_\mathbb{C} \)
from the category of cobordisms to the category of complex vector spaces. In detail:
- • Objects of \(\mathbf{Cob}_n\): closed oriented \((n{-}1)\)-dimensional manifolds
- • Morphisms: compact oriented \(n\)-dimensional cobordisms between them
- • The functor assigns a vector space \(Z(\Sigma)\) to each manifold \(\Sigma\)
- • And a linear map \(Z(M): Z(\partial_{\text{in}} M) \to Z(\partial_{\text{out}} M)\) to each cobordism \(M\)
This single sentence encodes all the axioms of a quantum field theory in a purely geometric language: gluing cobordisms corresponds to composing linear maps, the empty manifold maps to \(\mathbb{C}\), and the partition function is a complex number. Atiyah's axioms transformed TQFT from a collection of calculations into a precise mathematical structure.
Witten's Chern-Simons Theory (1989)
In 1989, Edward Witten showed that three-dimensional Chern-Simons gauge theory — a quantum field theory with action:
\( S_{\text{CS}}[A] = \frac{k}{4\pi} \int_M \text{tr}\!\left(A \wedge dA + \frac{2}{3} A \wedge A \wedge A\right) \)
is a TQFT in Atiyah's sense. Its partition function is a topological invariant of 3-manifolds, and the expectation values of Wilson loop operators give the Jones polynomial invariant of knots. This was the first deep connection between quantum field theory and knot theory, and it earned Witten the Fields Medal in 1990 — the only physicist to receive that honor.
What Witten Showed
A physicist's path integral — a formal integral over all gauge field configurations weighted by \(e^{iS_{\text{CS}}}\) — computes a mathematical invariant of the ambient 3-manifold. Physics produces topology. The invariant doesn't depend on a metric, only on the smooth structure — hence “topological” field theory.
Higher Categories and Extended Field Theories
Atiyah's axioms describe TQFTs that assign data only to top-dimensional manifolds and their codimension-1 boundaries. But a deeper structure exists: extendedTQFTs assign data to manifolds in all codimensions simultaneously.
This requires higher category theory. An \(n\)-category has:
- • Objects (0-morphisms)
- • Morphisms between objects (1-morphisms)
- • Morphisms between morphisms (2-morphisms)
- • ⋮ up to \(n\)-morphisms, all with coherent composition laws
An extended \(n\)-dimensional TQFT is a functor from the\((\infty,n)\)-category of cobordisms to a symmetric monoidal\((\infty,n)\)-category. The coherence conditions — ensuring all ways of composing are compatible — become increasingly complex but are precisely governed by higher categorical axioms.
The Cobordism Hypothesis (Baez-Dolan, proved by Lurie)
In 1995, John Baez and James Dolan conjectured the cobordism hypothesis: a fully extended framed TQFT in dimension \(n\) is completely determined by the value it assigns to a single point — that value must be a fully dualizable object in the target \((\infty,n)\)-category:
\( \left\{ \text{Framed } n\text{-TQFTs} \right\} \;\simeq\; \left\{ \text{Fully dualizable objects in } \mathcal{C} \right\} \)
The hypothesis was proved by Jacob Lurie in 2009. It is a profound classification theorem: it says that to specify an entire quantum field theory in all dimensions simultaneously, you need only specify one piece of data — an object with particularly strong duality properties. The entire field theory is then reconstructed from this single object by the universal property of the cobordism category.
Why This Is Deep
The cobordism hypothesis says that the cobordism \((\infty,n)\)-category is the free symmetric monoidal \((\infty,n)\)-category with dualsgenerated by a single fully dualizable object (the point). This is a universal property — the most powerful kind of statement in category theory. It turns the classification of TQFTs into a purely algebraic problem.
Monoidal Categories & Tensor Networks
A monoidal category is a category \(\mathcal{C}\) equipped with a tensor product \(\otimes: \mathcal{C} \times \mathcal{C} \to \mathcal{C}\)that is associative and unital up to coherent isomorphism. The prime example is\(\mathbf{Vect}_k\) with the usual tensor product of vector spaces.
In quantum information, tensor networks are graphical calculi for composing linear maps between tensor products of Hilbert spaces. The graphical language for monoidal categories — where morphisms are boxes and objects are wires — makes complex tensor contractions visually transparent. Matrix product states, MERA, and PEPS (the tensor network ansatze for many-body quantum systems) are all examples of morphisms in a monoidal category of Hilbert spaces.
The string diagram calculus for monoidal categories, developed by Penrose, Joyal, and Street, has become an indispensable tool in quantum computing, condensed matter, and quantum gravity. Category theory provides the precise rules for a diagrammatic language that physicists had been using informally for decades.
The Bridge: Category Theory as the Language of Quantum Physics
Category theory provides the most natural language for expressing the deep structure of quantum physics at every level:
When Eilenberg and Mac Lane invented categories in 1945 to organize algebraic topology, they could not have foreseen that their “abstract nonsense” would become the organizing principle of quantum physics. Category theory does not merely describe physics — it reveals the hidden unity beneath apparently different physical theories, showing them to be different manifestations of the same categorical structures.