Physlib

PhyslibAlpha.ProbabilisticTheory.LocalNet.Maps

Maps between observable systems

Transformations that respect what can be measured: restriction to a subsystem and reversible changes of description.

i. Overview

The measurable quantities of a physical system form a real vector space with an order: an observable is positive if every measurement of it gives a nonnegative number. The certain outcome is a distinguished observable `1`. A map between two such systems that is physically meaningful sends positive observables to positive observables and the certain outcome to the certain outcome. This is a `Channel`: the Heisenberg-picture form of a transformation of states.

Three kinds of channel occur when a system is assembled from its parts:

* an *embedding*, which regards a measurement made in a small region as a measurement made in a larger one. It loses nothing: it also reflects the order, so observables that compare as `≤` in the larger system already compared so in the smaller one; * an *isomorphism*, a reversible change of description such as moving a region by a translation, whose inverse is again a channel; * a *symmetry*, an isomorphism of a system with itself.

No lattice structure is assumed on the observables, since the observables of a quantum system do not have one.

ii. Key results

- `Channel.IsOrderEmbedding` : the channel loses no information about the order of observables. - `Channel.IsOrderEmbedding.injective` : an embedding distinguishes all the observables it embeds. - `OrderUnitIso` : a reversible change of description of a system. - `OrderUnitIso.refl`, `.symm`, `.trans` : doing nothing, undoing, and composing changes of description. - `OrderUnitIso.toSymmetry` : a reversible change of a system into itself is a symmetry.

iii. Table of contents

  • A. Order embeddings
  • B. Order-unit isomorphisms
  • C. Automorphisms and symmetries

iv. References

* O. Bratteli & D. Robinson, Operator Algebras and Quantum Statistical Mechanics 1, 2nd ed., Chapter 2. * R. Haag, Local Quantum Physics, 2nd ed., Chapter III.

A. Order embeddings

B. Order-unit isomorphisms

C. Automorphisms and symmetries

36 declarations

definition

ϕ\phi is an order embedding (ϕ(a)≤ϕ(b)  ⟹  a≤b\phi(a) \le \phi(b) \implies a \le b)

Let EE and FF be spaces of observables (order-unit spaces). A channel ϕ:E→F\phi : E \to F is an **order embedding** if it reflects the partial order of the observables: for any a,b∈Ea, b \in E, the condition ϕ(a)≤ϕ(b)\phi(a) \le \phi(b) implies a≤ba \le b. Since channels are monotonic by definition, this property ensures that ϕ\phi preserves and reflects the order, identifying EE as an ordered subsystem of FF.

theorem

ϕ(a)≤ϕ(b)  ⟺  a≤b\phi(a) \le \phi(b) \iff a \le b for order embeddings

Let EE and FF be order-unit spaces representing systems of observables. For a channel ϕ:E→F\phi : E \to F that is an order embedding, the order between any two observables a,b∈Ea, b \in E is preserved and reflected by ϕ\phi. That is, ϕ(a)≤ϕ(b)\phi(a) \le \phi(b) if and only if a≤ba \le b.

theorem

ϕ\phi is an order embedding   ⟺  \iff (ϕ(a)≥0  ⟹  a≥0\phi(a) \ge 0 \implies a \ge 0)

Let EE and FF be order-unit spaces representing systems of physical observables. A channel ϕ:E→F\phi : E \to F is an order embedding (meaning ϕ(a)≤ϕ(b)  ⟹  a≤b\phi(a) \le \phi(b) \implies a \le b for all a,b∈Ea, b \in E) if and only if it reflects positivity: for any observable a∈Ea \in E, if 0≤ϕ(a)0 \le \phi(a), then 0≤a0 \le a.

theorem

Order Embeddings are Injective

Let EE and FF be order-unit spaces representing physical systems of observables. If a channel ϕ:E→F\phi : E \to F is an order embedding (meaning that for any a,b∈Ea, b \in E, ϕ(a)≤ϕ(b)\phi(a) \le \phi(b) implies a≤ba \le b), then ϕ\phi is injective.

theorem

The identity channel is an order embedding

Let EE be an order-unit space representing a physical system of observables. The identity channel id:E→E\text{id} : E \to E, which maps every observable to itself, is an order embedding. That is, for any observables a,b∈Ea, b \in E, the condition id(a)≤id(b)\text{id}(a) \le \text{id}(b) implies a≤ba \le b.

theorem

Composition of Order Embeddings is an Order Embedding

Let E,FE, F, and GG be spaces of observables. If ϕ:E→F\phi : E \to F and ψ:F→G\psi : F \to G are channels that are order embeddings (meaning ϕ(a)≤ϕ(b)  ⟹  a≤b\phi(a) \le \phi(b) \implies a \le b and ψ(a)≤ψ(b)  ⟹  a≤b\psi(a) \le \psi(b) \implies a \le b for all observables in their respective domains), then their composition ψ∘ϕ:E→G\psi \circ \phi : E \to G is also an order embedding.

instance

Order-unit isomorphisms act as functions

An order-unit isomorphism ee between two systems of observables EE and FF can be treated as a function. This allows for the direct application of ee to an observable x∈Ex \in E to produce a value e(x)∈Fe(x) \in F, where the underlying mapping is the isomorphism's forward channel.

theorem

Extensionality of order-unit isomorphisms (e(x)=e′(x)  ⟹  e=e′e(x) = e'(x) \implies e = e')

Let EE and FF be systems of observables. For any two order-unit isomorphisms e,e′:E→Fe, e' : E \to F, if e(x)=e′(x)e(x) = e'(x) for all x∈Ex \in E, then e=e′e = e'.

theorem

e.toChannel(x)=e(x)e.\text{toChannel}(x) = e(x) for Order-Unit Isomorphisms

Let EE and FF be systems of observables (modeled as order-unit spaces). For any order-unit isomorphism e:E→Fe : E \to F and any observable x∈Ex \in E, the application of the underlying channel (the forward unital positive linear map) of ee to xx is equal to the application of ee to xx when treated as a function. That is, e.toChannel(x)=e(x)e.\text{toChannel}(x) = e(x).

theorem

e−1(e(x))=xe^{-1}(e(x)) = x for order-unit isomorphisms

Let EE and FF be order-unit spaces representing systems of observables. For any order-unit isomorphism e:E→Fe : E \to F and any observable x∈Ex \in E, applying the isomorphism ee and then applying its inverse channel e−1e^{-1} returns the original observable xx. Mathematically, this is expressed as e−1(e(x))=xe^{-1}(e(x)) = x.

theorem

e(e−1(y))=ye(e^{-1}(y)) = y for Order-Unit Isomorphisms

Let EE and FF be systems of observables (order-unit spaces). For any order-unit isomorphism e:E→Fe: E \to F and any observable y∈Fy \in F, applying the inverse channel e−1e^{-1} of the isomorphism to yy and then applying ee to the result returns the original observable yy. That is, e(e−1(y))=ye(e^{-1}(y)) = y.

definition

Identity order-unit isomorphism on EE

Let EE be an order-unit space, representing a physical system of observables. The identity order-unit isomorphism is the reversible transformation from EE to itself where both the forward mapping and its inverse are the identity map. It maps every observable x∈Ex \in E to itself while preserving the ordering of observables and the unit element 11.

theorem

The identity order-unit isomorphism maps xx to xx

Let EE be an order-unit space representing a physical system of observables. For any observable x∈Ex \in E, applying the identity order-unit isomorphism reflE\text{refl}_E to xx results in the observable itself: reflE(x)=x\text{refl}_E(x) = x

definition

Inverse of an order-unit isomorphism ee

Given an order-unit isomorphism e:E≅Fe: E \cong F between two order-unit spaces EE and FF, this construction defines the inverse isomorphism e−1:F≅Ee^{-1}: F \cong E by interchanging the roles of the underlying channel and its inverse.

theorem

e−1(y)=e.invChannel(y)e^{-1}(y) = e.\text{invChannel}(y)

Let EE and FF be systems of observables. For any order-unit isomorphism e:E≅Fe : E \cong F and any observable y∈Fy \in F, the application of the inverse isomorphism e−1e^{-1} (denoted `e.symm`) to yy is equal to the evaluation of the underlying inverse channel of ee at yy: e−1(y)=e.invChannel(y)e^{-1}(y) = e.\text{invChannel}(y)

theorem

(e−1)−1=e(e^{-1})^{-1} = e for order-unit isomorphisms

Let EE and FF be order-unit spaces. For any order-unit isomorphism e:E≅Fe : E \cong F, the inverse of its inverse is equal to the original isomorphism, i.e., (e−1)−1=e(e^{-1})^{-1} = e.

definition

Composition of order-unit isomorphisms e′∘ee' \circ e

Let EE, FF, and GG be systems of observables. Given an order-unit isomorphism e:E≅Fe: E \cong F and an order-unit isomorphism e′:F≅Ge': F \cong G, their composition e′∘ee' \circ e (denoted as e.trans e′e.\text{trans } e') is the order-unit isomorphism from EE to GG whose underlying forward map is the composition of the forward maps of e′e' and ee, and whose inverse map is the composition of the inverse maps of ee and e′e'.

theorem

(e′∘e)(x)=e′(e(x))(e' \circ e)(x) = e'(e(x)) for order-unit isomorphisms

Let EE, FF, and GG be systems of observables. For any order-unit isomorphisms e:E≅Fe: E \cong F and e′:F≅Ge': F \cong G, and any observable x∈Ex \in E, the application of the composed isomorphism e′∘ee' \circ e (denoted in Lean as `e.trans e'`) to xx satisfies (e′∘e)(x)=e′(e(x))(e' \circ e)(x) = e'(e(x)).

theorem

reflE∘e=e\text{refl}_E \circ e = e

Let EE and FF be order-unit spaces representing physical systems of observables. For any order-unit isomorphism e:E≅Fe: E \cong F, the composition of the identity isomorphism reflE\text{refl}_E with ee is equal to ee.

theorem

e∘reflF=ee \circ \text{refl}_F = e for order-unit isomorphisms

Let EE and FF be order-unit spaces representing systems of physical observables. For any order-unit isomorphism e:E≅Fe: E \cong F (a reversible change of description), let reflF\text{refl}_F be the identity order-unit isomorphism on FF. The composition of ee with reflF\text{refl}_F is equal to ee. In mathematical notation, this is expressed as e∘idF=ee \circ \text{id}_F = e.

theorem

Composition of order-unit isomorphisms is associative

Let E,F,G,E, F, G, and HH be systems of observables. For any order-unit isomorphisms e:E≅Fe: E \cong F, e′:F≅Ge': F \cong G, and e′′:G≅He'': G \cong H, the composition of these isomorphisms is associative: (e.trans e′).trans e′′=e.trans (e′.trans e′′)(e.\text{trans } e').\text{trans } e'' = e.\text{trans } (e'.\text{trans } e'')

theorem

e∘e−1=idFe \circ e^{-1} = \text{id}_F

Let EE and FF be order-unit spaces representing physical systems of observables. For any order-unit isomorphism e:E≅Fe: E \cong F, the composition of its inverse e−1:F≅Ee^{-1}: F \cong E and the original isomorphism ee (applying the inverse first) is the identity isomorphism on FF, denoted as e∘e−1=idFe \circ e^{-1} = \text{id}_F.

theorem

The composition of an order-unit isomorphism with its inverse is the identity, e∘e−1=idEe \circ e^{-1} = \text{id}_E

Let EE and FF be order-unit spaces representing physical systems of observables. For any order-unit isomorphism e:E≅Fe : E \cong F, the composition of ee and its inverse e−1e^{-1} (denoted as e.symme.\text{symm}) is the identity order-unit isomorphism on EE. Mathematically, this is expressed as e∘e−1=idEe \circ e^{-1} = \text{id}_E.

theorem

Left-cancellativity of order-unit isomorphism composition (a∘e=b∘e  ⟹  a=ba \circ e = b \circ e \implies a = b)

Let EE, FF, and GG be systems of observables. For any order-unit isomorphisms e:E≅Fe: E \cong F, a:F≅Ga: F \cong G, and b:F≅Gb: F \cong G, if the composition a∘ea \circ e is equal to b∘eb \circ e, then aa is equal to bb. Here, an order-unit isomorphism represents a reversible transformation between systems that preserves the order of observables and the unit (the certain outcome).

theorem

Right cancellation of order-unit isomorphisms: a∘e=b∘e  ⟹  a=ba \circ e = b \circ e \implies a = b

Let E,F,E, F, and GG be systems of observables (order-unit spaces). For any order-unit isomorphisms a,b:E≅Fa, b : E \cong F and e:F≅Ge : F \cong G, if the composition of aa with ee is equal to the composition of bb with ee (denoted a.trans e=b.trans ea.\text{trans } e = b.\text{trans } e), then a=ba = b. This indicates that order-unit isomorphisms are right-cancellative with respect to composition.

theorem

An Order-Unit Isomorphism is Bijective

Given an order-unit isomorphism e:E→Fe : E \to F between two systems of observables EE and FF, the map ee is a bijective function. Here, an order-unit isomorphism represents a reversible transformation between systems that preserves the order of observables and the certain outcome (the order unit).

theorem

Order-unit isomorphisms are order embeddings

Let EE and FF be order-unit spaces representing systems of observables. For any order-unit isomorphism e:E→Fe : E \to F, the corresponding channel is an order embedding. That is, for any observables a,b∈Ea, b \in E, the condition e(a)≤e(b)e(a) \le e(b) implies a≤ba \le b.

definition

Pullback of a state ω\omega along an isomorphism ee

Given an order-unit isomorphism e:E→Fe : E \to F between two systems of observables and a state ω\omega on the system FF, the pullback of ω\omega along ee is the state on system EE defined by the composition ω∘e\omega \circ e. For any observable x∈Ex \in E, the value of the pullback state is given by ω(e(x))\omega(e(x)).

theorem

(e.pullback ω)(x)=ω(e(x))(e.\text{pullback } \omega)(x) = \omega(e(x))

Given an order-unit isomorphism e:E→Fe: E \to F between two systems of observables EE and FF, for any state ω∈S[R,F]\omega \in \mathcal{S}[\mathbb{R}, F] and any observable x∈Ex \in E, the evaluation of the pullback state e.pullback ωe.\text{pullback } \omega at xx is equal to the evaluation of the state ω\omega at the transformed observable e(x)e(x), i.e., (e.pullback ω)(x)=ω(e(x))(e.\text{pullback } \omega)(x) = \omega(e(x)).

theorem

(e.trans e′).pullback ω=e.pullback (e′.pullback ω)(e.\text{trans } e').\text{pullback } \omega = e.\text{pullback } (e'.\text{pullback } \omega)

Let E,F,E, F, and GG be systems of observables. Given order-unit isomorphisms e:E→Fe : E \to F and e′:F→Ge' : F \to G, and a state ω∈S[R,G]\omega \in \mathcal{S}[\mathbb{R}, G] on the system GG, the pullback of ω\omega along the composition e′∘ee' \circ e (denoted as e.trans e′e.\text{trans } e') is equal to the pullback along ee of the state obtained by pulling back ω\omega along e′e'. That is, (e.trans e′).pullback ω=e.pullback (e′.pullback ω)(e.\text{trans } e').\text{pullback } \omega = e.\text{pullback } (e'.\text{pullback } \omega)

definition

Order-unit isomorphism e:E≅Ee: E \cong E as a symmetry

This definition interprets an order-unit isomorphism e:E≅Ee: E \cong E from an order unit space EE to itself as a symmetry of the system. A symmetry is a channel (a positive, unital linear map) that possesses a two-sided inverse channel, thereby acting as an order-automorphism of EE.

definition

Symmetry as an order-unit isomorphism ϕ:E≅E\phi: E \cong E

Given a symmetry ϕ\phi of an order unit space EE, this definition constructs the corresponding order-unit isomorphism from EE to itself. It uses the channel associated with ϕ\phi as the forward map and the inverse channel ϕ−1\phi^{-1} (provided by the fact that ϕ\phi is an order-automorphism) as the inverse map, satisfying ϕ−1∘ϕ=idE\phi^{-1} \circ \phi = \text{id}_E and ϕ∘ϕ−1=idE\phi \circ \phi^{-1} = \text{id}_E.

theorem

(e.trans e′).toSymmetry=e′.toSymmetry⋅e.toSymmetry(e.\text{trans } e').\text{toSymmetry} = e'.\text{toSymmetry} \cdot e.\text{toSymmetry}

Let EE be an order unit space. For any two order-unit isomorphisms e,e′:E≅Ee, e' : E \cong E, the symmetry associated with their composition e.trans e′e.{\text{trans }} e' is equal to the product of the individual symmetries of e′e' and ee in the symmetry group of EE. That is, (e.trans e′).toSymmetry=e′.toSymmetry⋅e.toSymmetry (e.{\text{trans }} e').\text{toSymmetry} = e'.\text{toSymmetry} \cdot e.\text{toSymmetry} where e.trans e′e.{\text{trans }} e' denotes the composition of isomorphisms (applying ee then e′e', or e′∘ee' \circ e), and the multiplication ⋅\cdot in the symmetry group corresponds to the composition of the underlying channels.

theorem

(refl E).toSymmetry=1(\text{refl } E).\text{toSymmetry} = 1

Let EE be an order-unit space. The symmetry of EE obtained from the identity order-unit isomorphism reflE\text{refl}_E is the identity element 11 of the symmetry group Symmetry(E)\text{Symmetry}(E).

theorem

(ofSymmetry ϕ).toSymmetry=ϕ(\text{ofSymmetry } \phi).\text{toSymmetry} = \phi

For any symmetry ϕ\phi of an order unit space EE, converting ϕ\phi into an order-unit isomorphism and then back into a symmetry returns the original symmetry ϕ\phi. That is, (ofSymmetry ϕ).toSymmetry=ϕ(\text{ofSymmetry } \phi).\text{toSymmetry} = \phi.

theorem

ofSymmetry(ϕ)(x)=ϕ(x)\text{ofSymmetry}(\phi)(x) = \phi(x)

Let EE be an order unit space representing a physical system. For any symmetry ϕ\phi of EE and any observable x∈Ex \in E, applying the order-unit isomorphism induced by ϕ\phi to xx is equivalent to applying the underlying channel of ϕ\phi to xx. That is: ofSymmetry(ϕ)(x)=ϕ(x)\text{ofSymmetry}(\phi)(x) = \phi(x)