Physlib

PhyslibAlpha.ProbabilisticTheory.Measurement.Integral

Integrating simple functions against an effect-valued measure

i. Overview

An effect-valued measure `μ` assigns an effect to every event. Integrating a simple function `∑ᵢ cᵢ 𝟙_{sᵢ}` against it gives `∑ᵢ cᵢ μ(sᵢ)`. The effect-valued measure is finitely additive, so this does not depend on how the function is written: two partitions representing the same function give the same value, by passing to their common refinement. The integral is linear and positive.

ii. Key results

- `EffectValuedMeasure.apply_iUnion_of_disjoint` : finite additivity. - `EffectValuedMeasure.IsPartition` : finite measurable partitions. - `EffectValuedMeasure.simpleIntegral` : the integral `∑ᵢ cᵢ μ(sᵢ)` of a simple function. - `EffectValuedMeasure.simpleIntegral_eq_of_pointwise_eq` : the integral depends only on the function. - `EffectValuedMeasure.simpleIntegral_nonneg` : the integral is positive.

iii. Table of contents

  • A. Finite measurable partitions and simple functions
  • B. Finite additivity
  • C. The integral of a simple function

A. Finite measurable partitions and simple functions

Nothing here refers to `E` or to a POVM at all yet: a partition and the simple function it carries are facts about `Ω` alone.

B. Finite additivity

`EffectValuedMeasure` only bundles *countable* additivity (`countably_additive`). Finite additivity is the special case of a family that is eventually `∅`, and the least upper bound of an eventually-constant, nondecreasing sequence (nondecreasing since every value of `μ` is a nonnegative effect) is just its eventual value — which is what lets us read finite sums off `countably_additive` directly.

C. The integral of a simple function

14 declarations

definition

Value of the simple function ∑ici1si\sum_i c_i \mathbb{1}_{s_i} at xx

Given a finite index set ι\iota, a collection of real coefficients cic_i, and a family of sets si⊆Ωs_i \subseteq \Omega for i∈ιi \in \iota, this function calculates the value of the simple function ∑i∈ιci1si\sum_{i \in \iota} c_i \mathbb{1}_{s_i} at a point x∈Ωx \in \Omega, where 1si\mathbb{1}_{s_i} denotes the indicator function of the set sis_i.

theorem

The value of the simple function ∑jcj1sj\sum_j c_j \mathbb{1}_{s_j} at x∈six \in s_i is cic_i

Let ι\iota be a finite index set and {sj}j∈ι\{s_j\}_{j \in \iota} be a partition of the space Ω\Omega. For a family of real coefficients cjc_j and the simple function defined by f(x)=∑j∈ιcj1sj(x)f(x) = \sum_{j \in \iota} c_j \mathbb{1}_{s_j}(x), if a point x∈Ωx \in \Omega lies in the piece sis_i for some i∈ιi \in \iota, then the function evaluates to cic_i. This holds because, since the pieces of the partition are pairwise disjoint, the indicator functions 1sj(x)\mathbb{1}_{s_j}(x) vanish for all j≠ij \neq i.

theorem

μ(s)=μ(t)\mu(s) = \mu(t) if s=ts = t for effect-valued measures

Let μ\mu be an effect-valued measure. For any two sets ss and tt such that s=ts = t, the value of the measure applied to ss is equal to the value of the measure applied to tt (i.e., μ(s)=μ(t)\mu(s) = \mu(t)). This equality holds regardless of the specific proofs of measurability provided for ss and tt.

theorem

Measure of Union equals Finite Sum for Eventually Empty Disjoint Sets

Let Ω\Omega be a measurable space and EE be a type representing effects. Let μ\mu be an effect-valued measure on Ω\Omega taking values in EE. Suppose {tn}n∈N\{t_n\}_{n \in \mathbb{N}} is a sequence of pairwise disjoint measurable sets in Ω\Omega. If there exists a natural number NN such that tn=∅t_n = \emptyset for all n≥Nn \geq N, then the effect-valued measure of their union is equal to the finite sum of their individual measures: μ(⋃n=0∞tn)=∑n=0N−1μ(tn)\mu\left(\bigcup_{n=0}^\infty t_n\right) = \sum_{n=0}^{N-1} \mu(t_n)

theorem

Binary additivity of effect-valued measures: μ(A∪B)=μ(A)+μ(B)\mu(A \cup B) = \mu(A) + \mu(B) for disjoint A,BA, B

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type EE. For any two measurable sets A,B⊆ΩA, B \subseteq \Omega that are disjoint (A∩B=∅A \cap B = \emptyset), the measure of their union is the sum of their individual measures: μ(A∪B)=μ(A)+μ(B)\mu(A \cup B) = \mu(A) + \mu(B) where the addition is performed in the space of effects EE.

theorem

μ(⋃i∈sfi)=∑i∈sμ(fi)\mu(\bigcup_{i \in s} f_i) = \sum_{i \in s} \mu(f_i) for finite disjoint collections

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type EE of effects. Let {fi}i∈ι\{f_i\}_{i \in \iota} be a family of measurable sets in Ω\Omega. For any finite set of indices ss, if the sets are pairwise disjoint for all i,j∈si, j \in s (i.e., fi∩fj=∅f_i \cap f_j = \emptyset whenever i≠ji \neq j), then the measure of their union over ss is the sum of their individual measures: μ(⋃i∈sfi)=∑i∈sμ(fi)\mu\left( \bigcup_{i \in s} f_i \right) = \sum_{i \in s} \mu(f_i)

theorem

Finite additivity of effect-valued measures: μ(⋃ifi)=∑iμ(fi)\mu(\bigcup_i f_i) = \sum_i \mu(f_i)

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type EE of effects. Let II be a finite index set and let {fi}i∈I\{f_i\}_{i \in I} be a family of measurable sets in Ω\Omega that are pairwise disjoint (i.e., fi∩fj=∅f_i \cap f_j = \emptyset for all i≠ji \neq j). Then the measure of the union of these sets is equal to the sum of the measures of the individual sets: μ(⋃i∈Ifi)=∑i∈Iμ(fi). \mu\left(\bigcup_{i \in I} f_i\right) = \sum_{i \in I} \mu(f_i).

definition

Simple integral ∑iciμ(si)\sum_i c_i \mu(s_i) of a simple function against an effect-valued measure

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type of effects EE. Given a finite index set II, a collection of real coefficients cic_i for i∈Ii \in I, and a collection of sets {si}i∈I\{s_i\}_{i \in I} that form a finite measurable partition of Ω\Omega, the integral of the simple function f=∑i∈Ici1sif = \sum_{i \in I} c_i \mathbf{1}_{s_i} against μ\mu is defined as the sum: ∑i∈Ici⋅μ(si)\sum_{i \in I} c_i \cdot \mu(s_i) where 1si\mathbf{1}_{s_i} is the indicator function of sis_i, and ci⋅μ(si)c_i \cdot \mu(s_i) denotes the scalar multiplication of the effect μ(si)\mu(s_i) by the real number cic_i. This construction is well-defined independently of the specific partition used to represent the simple function.

theorem

μ(si)=∑j∈Jμ(si∩sj′)\mu(s_i) = \sum_{j \in J} \mu(s_i \cap s'_j) for effect-valued measures and finite partitions {si}\{s_i\} and {sj′}\{s'_j\}

Let μ\mu be an effect-valued measure on a space Ω\Omega taking values in EE (where EE is an additive commutative group). Let {si}i∈I\{s_i\}_{i \in I} and {sj′}j∈J\{s'_j\}_{j \in J} be two finite measurable partitions of Ω\Omega. For any index i∈Ii \in I, the measure of the set sis_i is equal to the sum of the measures of its intersections with each piece of the second partition: μ(si)=∑j∈Jμ(si∩sj′). \mu(s_i) = \sum_{j \in J} \mu(s_i \cap s'_j). This result follows from the finite additivity of μ\mu, as the set sis_i is the disjoint union of the sets {si∩sj′}j∈J\{s_i \cap s'_j\}_{j \in J}.

theorem

Decomposition of an effect-valued measure element μ(sj′)\mu(s'_j) over a partition {si}\{s_i\}

Let Ω\Omega be a space and EE be a type of effects. Let μ\mu be an effect-valued measure on Ω\Omega taking values in EE. Suppose {si}i∈ι\{s_i\}_{i \in \iota} and {sj′}j∈ι′\{s'_j\}_{j \in \iota'} are two finite measurable partitions of Ω\Omega. For any index j∈ι′j \in \iota', the measure of the set sj′s'_j is equal to the sum of the measures of its intersections with every set in the partition {si}\{s_i\}: μ(sj′)=∑i∈ιμ(si∩sj′)\mu(s'_j) = \sum_{i \in \iota} \mu(s_i \cap s'_j)

theorem

Pointwise equality of simple functions implies equality of integrals against an effect-valued measure μ\mu

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type of effects EE. Suppose we have two finite measurable partitions of Ω\Omega, denoted by {si}i∈I\{s_i\}_{i \in I} and {sj′}j∈J\{s'_j\}_{j \in J}, with corresponding real coefficients cic_i and cj′c'_j. If the simple functions defined by these partitions are pointwise equal, that is, for all x∈Ωx \in \Omega: ∑i∈Ici1si(x)=∑j∈Jcj′1sj′(x)\sum_{i \in I} c_i \mathbb{1}_{s_i}(x) = \sum_{j \in J} c'_j \mathbb{1}_{s'_j}(x) where 1s\mathbb{1}_s is the indicator function of the set ss, then their integrals against μ\mu are equal: ∑i∈Iciμ(si)=∑j∈Jcj′μ(sj′)\sum_{i \in I} c_i \mu(s_i) = \sum_{j \in J} c'_j \mu(s'_j)

theorem

Additivity of the Simple Integral in Weights over a Fixed Partition

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type of effects EE. Let {si}i∈ι\{s_i\}_{i \in \iota} be a finite measurable partition of Ω\Omega. For any two weight functions c1,c2:ι→Rc_1, c_2: \iota \to \mathbb{R}, the integral of the simple function against μ\mu is additive with respect to the weights over the fixed partition: ∑i∈ι(c1(i)+c2(i))⋅μ(si)=∑i∈ιc1(i)⋅μ(si)+∑i∈ιc2(i)⋅μ(si)\sum_{i \in \iota} (c_1(i) + c_2(i)) \cdot \mu(s_i) = \sum_{i \in \iota} c_1(i) \cdot \mu(s_i) + \sum_{i \in \iota} c_2(i) \cdot \mu(s_i) where c⋅μ(s)c \cdot \mu(s) denotes the scalar multiplication of the effect μ(s)\mu(s) by the real number cc.

theorem

Homogeneity of the simple integral: simpleIntegral(μ,r⋅c,s)=r⋅simpleIntegral(μ,c,s)\text{simpleIntegral}(\mu, r \cdot c, s) = r \cdot \text{simpleIntegral}(\mu, c, s)

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a type of effects EE. Let {si}i∈I\{s_i\}_{i \in I} be a finite measurable partition of Ω\Omega indexed by a finite set II, c:I→Rc: I \to \mathbb{R} be a collection of weights, and r∈Rr \in \mathbb{R} be a scalar. The integral of the simple function with weights (r⋅c)i=r⋅ci(r \cdot c)_i = r \cdot c_i against μ\mu is equal to rr times the integral of the simple function with weights cic_i: ∑i∈I(r⋅ci)μ(si)=r⋅∑i∈Iciμ(si)\sum_{i \in I} (r \cdot c_i) \mu(s_i) = r \cdot \sum_{i \in I} c_i \mu(s_i) where the left-hand side represents the integral of the weighted function r⋅cr \cdot c and the right-hand side represents scalar multiplication of the resulting effect in EE by rr.

theorem

Non-negativity of the Simple Integral for Effect-Valued Measures

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a space of effects EE. Given a finite collection of non-negative real coefficients ci≥0c_i \ge 0 for i∈Ii \in I and a collection of sets {si}i∈I\{s_i\}_{i \in I} that form a finite measurable partition of Ω\Omega, the simple integral of the function ∑i∈Ici1si\sum_{i \in I} c_i \mathbf{1}_{s_i} against μ\mu is non-negative, such that: 0≤∑i∈Ici⋅μ(si)0 \le \sum_{i \in I} c_i \cdot \mu(s_i) where the inequality refers to the partial order on the effect space EE.