Physlib

PhyslibAlpha.ProbabilisticTheory.Classical.SeparableObservables

Classical systems with separable observables

i. Overview

Most systems in physics can be described by countably many observables: the observables are separable. For such systems every state is a mixture of pure states. This is Choquet's theorem.

The idea is simple. The states form a compact convex set. Its corners are the pure states. A state that is not pure is the midpoint of two other states, so its weight can be pushed outwards to them. Pushing weight outwards as far as possible leaves a probability measure that lives on the pure states. Countably many observables are needed to make this limit exist.

Combined with the uniqueness theorem, this gives a clean criterion. A system with separable observables is classical exactly when its ensembles refine.

ii. Key results

- `UnitalPositiveLinearMap.exists_pure_representingMeasure` proves that every state is the barycenter of a probability measure living on the pure states. - `PureState.hasPureDecomposition_of_separable` proves that every state has a pure decomposition. - `isSimplexStateSpace_iff_ensemblesRefine` proves that the system is classical exactly when ensembles refine. This is the **Choquet–Meyer theorem**.

iii. Table of contents

  • A. A dense sequence of observables separates states
  • B. Every state is a mixture of pure states
  • C. Pure decompositions
  • D. Classical systems are those whose ensembles refine

iv. References

- G. Choquet and P.-A. Meyer, *Existence et unicité des représentations intégrales dans les convexes compacts quelconques*, Ann. Inst. Fourier 13 (1963), 139–154.

A. A dense sequence of observables separates states

B. Every state is a mixture of pure states

C. Pure decompositions

D. Classical systems are those whose ensembles refine

9 declarations

definition

Evaluation of the nn-th observable in a dense sequence

Let EE be an Archimedean order-unit space and let (An)n∈N(A_n)_{n \in \mathbb{N}} be a countable dense sequence of observables in EE. For a given index n∈Nn \in \mathbb{N}, this definition represents the continuous linear functional on the weak dual space E∗E^* of EE that maps an element ω∈E∗\omega \in E^* to the value ω(An)\omega(A_n). In the context of physical systems, this corresponds to the expectation value of the nn-th observable in the dense sequence for a given state ω\omega.

theorem

A dense sequence of observables separates states

Let EE be an Archimedean order-unit space and let (An)n∈N(A_n)_{n \in \mathbb{N}} be a countable dense sequence of observables in EE. For any two states xx and yy in the state space of EE, if the expectation values of all observables in the sequence are equal for both states—that is, x(An)=y(An)x(A_n) = y(A_n) for all n∈Nn \in \mathbb{N}—then the states are identical (x=yx = y).

theorem

Choquet's Theorem: Every state is the barycenter of a probability measure on pure states

Let EE be an Archimedean order-unit space. For every state ω∈S(R,E)\omega \in \mathcal{S}(\mathbb{R}, E), there exists a probability measure μ\mu on the state space stateSpace E\text{stateSpace } E such that the barycenter of μ\mu is ω\omega and μ\mu is concentrated on the pure states. That is, barycenter μ=ω\text{barycenter } \mu = \omega and for μ\mu-almost every ϕ\phi in the state space, the state ϕ\phi is pure.

theorem

The set of pure states is measurable

Let EE be an Archimedean order-unit space. The set of pure states {ω∈stateSpace E∣ω is pure}\{\omega \in \text{stateSpace } E \mid \omega \text{ is pure}\} is a measurable set with respect to the Borel σ\sigma-algebra of the weak-star topology on the state space. A state is defined as pure if it is an extreme point of the convex state space.

theorem

The inclusion PureState E↪stateSpace E\text{PureState } E \hookrightarrow \text{stateSpace } E is a measurable embedding

Let EE be an Archimedean order-unit space. Let stateSpace E\text{stateSpace } E be the set of normalized positive linear functionals on EE equipped with the Borel σ\sigma-algebra of the weak-star topology, and let PureState E\text{PureState } E be the subset of pure states. The canonical inclusion map ι:PureState E→stateSpace E\iota : \text{PureState } E \to \text{stateSpace } E is a measurable embedding.

theorem

Measures concentrated on pure states are recovered by push-forward of their restriction

Let EE be an Archimedean order-unit space, and let stateSpace E\text{stateSpace } E denote its state space (the set of normalized positive continuous linear functionals). Let P⊆stateSpace EP \subseteq \text{stateSpace } E be the set of pure states, and let ι:P↪stateSpace E\iota: P \hookrightarrow \text{stateSpace } E be the inclusion map. If a measure μ\mu on stateSpace E\text{stateSpace } E is concentrated on the pure states—meaning that for μ\mu-almost every ϕ\phi, ϕ\phi is a pure state—then the push-forward of the restriction (comap) of μ\mu to PP along ι\iota is equal to μ\mu: ι∗(ι∗μ)=μ \iota_* (\iota^* \mu) = \mu

theorem

A probability measure on the state space supported on pure states induces a probability measure on the space of pure states.

Let EE be an Archimedean order-unit space and let S(E)\mathcal{S}(E) be its state space (the set of normalized positive continuous linear functionals in the weak dual E∗E^*). Let μ\mu be a probability measure on S(E)\mathcal{S}(E). If μ\mu is supported on the set of pure states (i.e., for μ\mu-almost every ϕ∈S(E)\phi \in \mathcal{S}(E), the state ϕ\phi is pure), then the pullback measure (comap) of μ\mu under the inclusion map ι:PureState E→S(E)\iota : \text{PureState } E \to \mathcal{S}(E) is a probability measure on the space of pure states.

theorem

Every state in a system with separable observables has a pure decomposition

Let EE be an Archimedean order-unit space with separable observables. For every state ω∈S(E)\omega \in \mathcal{S}(E) (the set of unital positive linear maps from EE to R\mathbb{R}), there exists a pure decomposition of ω\omega. Specifically, there exists a regular probability measure μ\mu on the set of pure states of EE such that for every f∈Ef \in E, the value of the state is given by the integral: ω(f)=∫PureState Eϕ(f) dμ(ϕ)\omega(f) = \int_{\text{PureState } E} \phi(f) \, d\mu(\phi)

theorem

Choquet–Meyer Theorem: Simplex State Space   ⟺  \iff Ensemble Refinement

For an Archimedean order-unit space EE representing a system with separable observables, the state space S(E)\mathcal{S}(E) is a simplex if and only if the system satisfies the ensemble refinement property. This property, known as the **Choquet–Meyer theorem**, asserts that the uniqueness of representing a state as a mixture of pure states is equivalent to the ability to find a common refinement for any two finite ensembles of weighted states that have the same sum.