Physlib

PhyslibAlpha.ProbabilisticTheory.JordanOrderUnit.ProjectionResolution

Projection resolutions

i. Overview

A projection resolution is an effect-valued measure whose values are Jordan projections and whose value on an intersection is the Jordan product of the values. Integrating bounded measurable functions against it gives a bounded Borel functional calculus.

ii. Key results

- `MeasurableProjectionResolution` : projection resolutions. - `MeasurableProjectionResolution.jordanOrthogonal_of_disjoint` : disjoint sets give orthogonal projections. - `MeasurableProjectionResolution.boundedBorel` : the bounded Borel functional calculus.

iii. Table of contents

  • A. Bounded Borel calculus

A. Bounded Borel calculus

The resolution does not duplicate the order-unit integral: in a JB algebra, which is complete in its order-unit norm, this section gives that integral its spectral-calculus name.

19 declarations

instance

Projection resolution as an effect-valued measure

Every measurable projection resolution PP on a measurable space Ω\Omega with values in a space EE can be viewed as an effect-valued measure. This allows a projection resolution to be used in any context where an effect-valued measure is expected.

instance

Projection resolution PP as a function s↦P(s)s \mapsto P(s)

For a measurable projection resolution PP over a sample space Ω\Omega with values in an effect space EE, this instance allows PP to be treated as a function that maps any measurable set s⊆Ωs \subseteq \Omega to its corresponding effect P(s)∈EP(s) \in E.

theorem

Extensionality of Measurable Projection Resolutions

Let Ω\Omega be a measurable space and EE be a type representing a system of effects (typically within a JB algebra). Let PP and QQ be measurable projection resolutions on Ω\Omega taking values in EE. If for every measurable set s⊆Ωs \subseteq \Omega, the projections P(s)P(s) and Q(s)Q(s) are equal as elements of EE, then the projection resolutions PP and QQ are equal.

theorem

P(∅)=0P(\emptyset) = 0 for a measurable projection resolution

For any measurable projection resolution PP on a measurable space Ω\Omega with values in the effects EE, the value of PP on the empty set ∅\emptyset (representing the impossible event) is the zero effect 00.

theorem

P(Ω)=1P(\Omega) = 1 for measurable projection resolutions

For any measurable projection resolution PP on a space Ω\Omega taking values in a Jordan algebra EE, the value assigned to the universal set Ω\Omega is the unit element 11 of EE.

theorem

Values of a Measurable Projection Resolution are Jordan Projections

Let Ω\Omega be a measurable space and EE be a Jordan algebra. If PP is a measurable projection resolution from Ω\Omega to EE, then for any measurable set s⊆Ωs \subseteq \Omega, the value P(s)∈EP(s) \in E is a Jordan projection.

theorem

P(s)⋅P(t)=P(s∩t)P(s) \cdot P(t) = P(s \cap t) for measurable projection resolutions

Let PP be a measurable projection resolution on a measurable space Ω\Omega taking values in a Jordan algebra EE. For any measurable sets s,t⊆Ωs, t \subseteq \Omega, the Jordan product of the projections P(s)P(s) and P(t)P(t) is equal to the projection of their intersection s∩ts \cap t: P(s)⋅P(t)=P(s∩t)P(s) \cdot P(t) = P(s \cap t)

theorem

Disjoint measurable sets have Jordan-orthogonal projections

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. For any two measurable sets s,t⊆Ωs, t \subseteq \Omega, if ss and tt are disjoint (i.e., s∩t=∅s \cap t = \emptyset), then the corresponding projections P(s)P(s) and P(t)P(t) are Jordan-orthogonal in EE.

theorem

Countable additivity of measurable projection resolutions

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in an ordered Jordan algebra EE. For any sequence of pairwise disjoint measurable sets {sn}n∈N\{s_n\}_{n \in \mathbb{N}} in Ω\Omega, the projection of their union P(⋃n∈Nsn)P\left(\bigcup_{n \in \mathbb{N}} s_n\right) is the least upper bound (supremum) of the set of finite partial sums {∑n=0N−1P(sn)∣N∈N}\left\{ \sum_{n=0}^{N-1} P(s_n) \mid N \in \mathbb{N} \right\}.

definition

Bounded Borel functional calculus ∫f dP\int f \, dP of a projection resolution

Given a measurable projection resolution PP on a measurable space Ω\Omega with values in a Jordan algebra EE, and a measurable function f:Ω→Rf: \Omega \to \mathbb{R} that is bounded by a constant MM (i.e., ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega), the bounded Borel calculus is the element in EE defined by the integral of ff with respect to the projection resolution PP, denoted as ∫Ωf dP\int_\Omega f \, dP.

theorem

The bounded Borel functional calculus of a non-negative function is non-negative (f≥0  ⟹  ∫f dP≥0f \ge 0 \implies \int f \, dP \ge 0)

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function that is bounded by a constant MM (i.e., ∣f(x)∣≤M|f(x)| \leq M for all x∈Ωx \in \Omega). If ff is non-negative (f(x)≥0f(x) \geq 0 for all x∈Ωx \in \Omega), then its bounded Borel functional calculus under PP, denoted as ∫Ωf dP\int_\Omega f \, dP, is a non-negative element in EE (i.e., 0≤∫Ωf dP0 \leq \int_\Omega f \, dP).

theorem

The bounded Borel calculus ∫f dP\int f \, dP is independent of the bound MM

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function. Suppose MM and M′M' are two real constants such that ∣f(x)∣≤M|f(x)| \le M and ∣f(x)∣≤M′|f(x)| \le M' for all x∈Ωx \in \Omega. Then the value of the bounded Borel functional calculus ∫Ωf dP\int_{\Omega} f \, dP is independent of the choice of bound, i.e., the integral calculated using MM is equal to the integral calculated using M′M'.

theorem

∫(f+g) dP=∫f dP+∫g dP\int (f + g) \, dP = \int f \, dP + \int g \, dP

Let PP be a measurable projection resolution on a measurable space Ω\Omega taking values in a Jordan algebra EE. Let f,g:Ω→Rf, g: \Omega \to \mathbb{R} be measurable functions such that ff is bounded by MM (∣f(x)∣≤M|f(x)| \le M) and gg is bounded by M′M' (∣g(x)∣≤M′|g(x)| \le M'). If the sum f+gf + g is measurable and bounded by M+M′M + M', then the bounded Borel functional calculus is additive: ∫Ω(f+g) dP=∫Ωf dP+∫Ωg dP\int_{\Omega} (f + g) \, dP = \int_{\Omega} f \, dP + \int_{\Omega} g \, dP

theorem

Real-Homogeneity of the Bounded Borel Functional Calculus: ∫(cf) dP=c∫f dP\int (c f) \, dP = c \int f \, dP

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function such that ∣f(x)∣≤M|f(x)| \le M for some M∈RM \in \mathbb{R}. For any scalar c∈Rc \in \mathbb{R}, the bounded Borel functional calculus (denoted by ∫f dP\int f \, dP) is real-homogeneous, satisfying: ∫Ω(c⋅f) dP=c⋅∫Ωf dP\int_{\Omega} (c \cdot f) \, dP = c \cdot \int_{\Omega} f \, dP

theorem

The Bounded Borel Calculus of an Indicator Function 1s\mathbf{1}_s Equals P(s)P(s)

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. For any measurable set s⊆Ωs \subseteq \Omega, let 1s:Ω→R\mathbf{1}_s: \Omega \to \mathbb{R} denote the indicator function of ss (where 1s(x)=1\mathbf{1}_s(x) = 1 if x∈sx \in s and 00 otherwise). The bounded Borel functional calculus of 1s\mathbf{1}_s with respect to PP is equal to the projection P(s)P(s), i.e., ∫Ω1s dP=P(s). \int_\Omega \mathbf{1}_s \, dP = P(s).

theorem

∫Ω1 dP=1\int_{\Omega} 1 \, dP = 1

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. The bounded Borel functional calculus of the constant function 11 (defined as f(x)=1f(x) = 1 for all x∈Ωx \in \Omega) is equal to the order unit (identity) 11 of the algebra EE. This represents the normalization of the projection resolution, expressed as: ∫Ω1 dP=1\int_{\Omega} 1 \, dP = 1

definition

Effect-valued bounded Borel calculus for 0≤f≤10 \le f \le 1

Given a measurable projection resolution PP on a measurable space Ω\Omega with values in a Jordan algebra EE, and a measurable function f:Ω→Rf: \Omega \to \mathbb{R} such that 0≤f(x)≤10 \le f(x) \le 1 for all x∈Ωx \in \Omega, this definition represents the bounded Borel functional calculus ∫Ωf dP\int_\Omega f \, dP as an **effect** in EE. An effect is an element of EE belonging to the order interval [0,1][0, 1]. This definition packages the properties that the integral of a [0,1][0, 1]-valued function is positive (0≤∫f dP0 \le \int f \, dP) and normalized (∫f dP≤1\int f \, dP \le 1).

theorem

The effect-valued bounded Borel calculus ∫Ωf dP\int_\Omega f \, dP equals its value in EE

Let PP be a measurable projection resolution on a measurable space Ω\Omega with values in a Jordan algebra EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function such that 0≤f(x)≤10 \le f(x) \le 1 for all x∈Ωx \in \Omega. Then the effect-valued bounded Borel functional calculus (the integral ∫Ωf dP\int_\Omega f \, dP viewed as an element of the order interval [0,1]⊆E[0, 1] \subseteq E) is equal to the standard bounded Borel functional calculus ∫Ωf dP\int_\Omega f \, dP viewed as an element of EE. In this context, coercing the effect-valued integral into the algebra EE simply forgets the established interval bounds [0,1][0, 1].

theorem

Extensionality of projection resolutions via bounded Borel calculus

Let Ω\Omega be a measurable space and EE be a Jordan algebra. Let PP and QQ be measurable projection resolutions on Ω\Omega with values in EE. If for every measurable function f:Ω→Rf: \Omega \to \mathbb{R} that is bounded by some constant M∈RM \in \mathbb{R} (i.e., ∣f(x)∣≤M|f(x)| \leq M for all x∈Ωx \in \Omega), the bounded Borel functional calculus results are equal, ∫Ωf dP=∫Ωf dQ\int_\Omega f \, dP = \int_\Omega f \, dQ, then the projection resolutions themselves are equal, P=QP = Q.