Skip to content

About

A Lean 4 formalization of quantum systems from an operator-algebraic perspective.

Topics

Resources

Stars

4 stars

Watchers

0 watching

Forks

Repository files navigation

QuantumSystem

A Lean 4 formalization of quantum systems through reconstruction theorems and the operator-algebraic and cohomological arguments they share.

Warning

This project is a work in progress. Breaking API changes — renamed or removed declarations, changed signatures, and reorganized modules — happen frequently without deprecation.

Note

This project is currently not accepting contributors because the following areas still need to be addressed:

  • Contributing guidelines (including an AI policy)
  • Documentation (including Lean Blueprint)

Scope

QuantumSystem aims to formalize reconstruction: recovering an algebraic or group-theoretic structure from data that carry less of it. A functor sends structures to data, contravariantly for groups and covariantly for algebras: a compact group G to the symmetric tensor C⋆-category Rep G of its finite-dimensional unitary representations, a unital C⋆-algebra to its underlying operator system. Each reconstruction asks how far this functor is full and faithful in the 2-categorical sense, matching morphisms up to the isomorphisms its target allows, and what its essential image is:

  • Uniqueness. Every morphism of the data comes from a morphism of the structures, unique up to the appropriate isomorphism: every symmetric tensor ⋆-functor Rep H → Rep G is, up to unitary tensor isomorphism, restriction along a continuous homomorphism G → H, unique up to conjugation in H. Where the data have more morphisms than the structures, as a unital completely positive map between C⋆-algebras need not be a ⋆-homomorphism, the functor is faithful and is asked to be full on isomorphisms only. In particular, isomorphic data come from isomorphic structures.
  • Existence. The data that come from a structure are characterized intrinsically, and the structure is rebuilt from them: a compact group, with composition as its group law, as the unitary tensor automorphisms of a fiber functor; the multiplication of a unital C⋆-algebra, on an operator system that coincides with its C⋆-envelope.

Uniqueness and existence rest on shared arguments. Three recur across the reconstructions above, M1–M3 below; the first two also have approximate versions, where the axioms hold up to a small defect, and asymptotic ones, along inductive systems of finite-dimensional stages.

  1. M1. Multiplication from complete positivity. The product is not part of the data; it is recovered from the matrix order. An idempotent morphism of the data, a unital completely positive projection, transports the ambient product to its image, compressed by the projection, and makes the image a C⋆-algebra in its own right, with the ambient involution and norm, even when it is not a subalgebra; when it is one, the projection is a conditional expectation and the product is the restricted one. Applied to the injective envelope of an operator system, the construction leads to its C⋆-envelope, and the operator systems that come from unital C⋆-algebras are exactly those that coincide with it. Positivity on matrices, not only on the algebra itself, is what carries the product: a unital order isomorphism need only preserve the Jordan product.
  2. M2. Rigidity. The structure cannot move without the data moving: an isomorphism of the data is an isomorphism of the structures. The argument is an equality case. Every morphism of the data satisfies an inequality that measures how far it is from preserving the structure, and equality singles out where it does: for the product, a Schwarz inequality whose two-sided equality set is a C⋆-subalgebra on which the morphism is a ⋆-homomorphism; for a divergence such as the relative entropy, the monotonicity inequality, whose equality case, for faithful normal states of finite divergence under a unital 2-positive normal morphism, can be reversibility by another morphism of the data on the states compared. An isomorphism of the data attains equality everywhere, since its inverse is again a morphism of the data, so the matrix order determines the product. In approximate versions near-equality gives near-preservation or approximate reversibility, and rigidity becomes stability, which may need an input of its own, such as amenability, a vanishing of cohomology that averaging supplies: without it an approximately multiplicative map need not lie near a multiplicative one.
  3. M3. Duality. A group is recovered from its representations as the automorphisms of a forgetful functor. The unitary tensor automorphisms of a fiber functor on a symmetric tensor C⋆-category form a compact group, with composition as group law, and the ⋆-structure is what singles it out among all tensor automorphisms, as positivity on matrices singles out the product in M1. The fiber functor is not an input: a symmetric tensor C⋆-category with conjugates, direct sums, subobjects and simple unit determines one, symmetric once the symmetry is twisted by the sign of the statistics and unique up to unitary tensor isomorphism among such, and with it the compact group, together with a central element of order at most two that records that sign. Together with M1, duality recovers not only the group but what it acts on: the fixed points of a continuous compact group action are the image of a conditional expectation, and, under suitable irreducibility conditions, the fixed-point algebra together with the symmetric category of endomorphisms induced by the group's representations inside the algebra determines both the algebra and the group.

Roadmap

Abstraction first. Each phase builds the primitives the next one states its theorems in, and the design is revisited at each step. The Scope is meant to be stable; this roadmap is not, and Phase III in particular will be set as the foundations take shape. What is open for each item before it can be stated is kept in docs/roadmap.md.

Phase I — Foundations

  1. Tensor products. Minimal tensor product of C⋆-algebras and nuclearity; the projective tensor product of Banach algebras, which holds bounded approximate diagonals (M1, M2).
  2. Dynamical systems. C⋆- and W⋆-dynamical systems: a group acting by ⋆-automorphisms, continuously in the point-norm (W⋆: point-σ-weak) topology; as an example, the action of the covariance group of a faithful local net by ⋆-automorphisms of its quasi-local algebra, already formalized, through which a discrete symmetry group acts by a homomorphism into it (M1, M3).
  3. Tensor C⋆-categories. Symmetric tensor C⋆-categories with conjugates, direct sums, subobjects and simple unit; fiber functors and their unitary tensor automorphisms (M3).

Phase II — Exact theory

  1. Von Neumann algebras. Beyond Kaplansky density for the unit ball and finite-dimensional factors as matrix algebras, already formalized: Kaplansky density for unitaries and in the strong* form, normal ⋆-homomorphisms, the predual; Sakai's characterization of von Neumann algebras as the C⋆-algebras with a predual; and the uniqueness up to order of the matrix sizes in the structure theorem of finite-dimensional C⋆-algebras (M2).
  2. Tomita–Takesaki theory. Beyond the relative modular operators already formalized: JMJ = M′ and Δ^{it}MΔ^{-it} = M for a cyclic and separating vector, the modular automorphism group σ^φ of a faithful normal state, the KMS condition characterizing it (Takesaki), and the Connes cocycle [Dψ : Dφ] of two faithful normal states, a unitary 1-cocycle for σ^φ (M2).
  3. Standard form. The natural cone, the unique vector in it representing each normal positive functional, and the unitary implementation of ⋆-automorphisms (M2).
  4. Reversibility. Beyond the monotonicity of the relative entropy already formalized, its equality case — the Petz recovery of faithful normal states at finite divergence under a unital 2-positive normal map, built on the items Tomita–Takesaki theory and Standard form (M2).
  5. Averaging. Peter–Weyl; for a compact group acting continuously by ⋆-automorphisms, Haar averaging as the conditional expectation onto the fixed points, and the density of the spectral subspaces; for a finite-dimensional C⋆-algebra, the diagonal ∫ U⋆ ⊗ U dU over its unitary group, of projective norm one whatever the dimension (M1, M2, M3).
  6. Duality. Tannaka–Krein for compact groups, the group law being composition and the topology pointwise convergence, extending Mathlib's Tannaka duality for finite groups; the Doplicher–Roberts duality theorem on the categories of Tensor C⋆-categories: existence and uniqueness of the fiber functor that is symmetric for the twisted symmetry, and the equivalence with Rep G twisted by a central element k, k² = e; its form for a compact group acting faithfully on a unital C⋆-algebra F, A = F^G, with full Hilbert spectrum and A′ ∩ F = ℂ1: the endomorphisms of A induced by the G-invariant Hilbert spaces in F determine F and G (M3).

Phase III — To be determined

Candidates, which no theorem of Phase II is stated in:

  • Inductive systems of operator systems along completely positive contractions, with limits in ∏Aₙ/⊕Aₙ, the ambient algebra of inductive limits (M1, M2).
  • Approximate structures: ε-C⋆-algebras (an exact norm and involution; associativity, submultiplicativity, unit and C⋆-identity up to ε), δ-homomorphisms (⋆-preserving linear maps preserving the unit and the product up to δ) and δ-isomorphisms (Kitaev, arXiv:2405.02434) (M1, M2).
  • Hochschild cohomology in low degree, after Mathlib's group cohomology in low degree: H² classifying the infinitesimal deformations of a multiplication, whose vanishing makes it rigid (Gerstenhaber), and H³ holding the obstructions to extending them; its bounded version for Banach algebras, amenability and virtual diagonals, in the bidual of the projective tensor product; for C⋆-algebras, amenability is nuclearity (Connes, Haagerup) (M1, M2).
  • The Lefschetz–Hopf fixed-point theorem and Hopf's structure theorem for the real cohomology of connected H-spaces of finite type, the topological input to the stability theorem for ε-C⋆-algebras (Kitaev, arXiv:2405.02434) (M2).

Usage

To depend on this library, add to your project's lakefile.toml:

[[require]]
name = "QuantumSystem"
git = "https://github.com/kencyke/quantum-system"
rev = "v4.x.y" # the release tag matching your lean-toolchain

Claude Code skills for other Lean projects

The /math-review, /citation-check and /lint-fix skills used here are published from plugins/lean-formalization/ as the lean-formalization Claude Code plugin; its README describes them, their requirements and how to install them. Do not install the plugin inside this repository: each entry of .claude/skills and .claude/agents already links to it, so every skill would be listed twice (/math-review and /lean-formalization:math-review) and the math-reviewer agent, which Claude Code uses proactively, would run twice.

About

A Lean 4 formalization of quantum systems from an operator-algebraic perspective.

Topics

Resources

Stars

4 stars

Watchers

0 watching

Forks

Releases

Contributors

Languages

Generated from kencyke/lean-project