Physlib

PhyslibAlpha.ProbabilisticTheory.Measurement.BoundedIntegral

Integrating bounded functions against an effect-valued measure

i. Overview

If the observables are complete for the order-unit norm, the integral of simple functions extends to bounded measurable functions. A bounded function is uniformly approximated by simple functions on a mesh of width `1 / (n + 1)`, and simple functions within `ε` and `ε'` of `f` have integrals within `ε + ε'`. So the integrals of the approximations converge, to a limit that does not depend on the approximation or the bound. The integral is linear and positive, and agrees with the simple integral on simple functions.

ii. Key results

- `EffectValuedMeasure.meshPiece`, `EffectValuedMeasure.meshWeight` : the mesh approximation. - `EffectValuedMeasure.orderUnitNorm_simpleIntegral_sub_le` : nearby simple functions have nearby integrals. - `EffectValuedMeasure.integral` : the integral of a bounded measurable function. - `EffectValuedMeasure.tendsto_of_uniformly_approximating` : every uniform approximation gives the integral. - `EffectValuedMeasure.integral_add`, `EffectValuedMeasure.nonneg_integral` : linearity and positivity.

iii. Table of contents

  • A. The mesh approximation of a bounded measurable function
  • B. Linear operations on simple values
  • C. Comparing the simple integrals of two approximations
  • D. The integral, via completeness

A. The mesh approximation of a bounded measurable function

B. Linear operations on simple values

C. Comparing the simple integrals of two approximations

The key ingredient for both the Cauchy property and the independence of the eventual integral from the choice of approximating sequence: if two simple functions (over possibly different partitions) both stay within `ε`, resp. `ε'`, of the same `f` everywhere, their integrals against `μ` stay within `ε + ε'` of each other, in the order-unit norm.

D. The integral, via completeness

The order-unit norm makes `E` a normed group, assumed complete. Most order-unit spaces are not complete.

40 declarations

definition

Number of mesh cells ⌈(n+1)M⌉\lceil (n+1)M \rceil for bound MM and mesh parameter nn

Given a bound M∈RM \in \mathbb{R} and a natural number nn (where the mesh width is defined as 1/(n+1)1/(n+1)), this function returns the number of mesh cells required to cover the interval from 00 to MM. It is defined as the ceiling of the product of MM and n+1n+1, expressed as ⌈(n+1)M⌉\lceil (n+1)M \rceil.

abbrev

Index set for the mesh on [−M,M][-M, M] at scale nn

Given a bound M∈RM \in \mathbb{R} and a discretization parameter n∈Nn \in \mathbb{N}, the type `MeshIndex M n` is the set of integer labels for the mesh cells used to partition the interval [−M,M][-M, M]. It is defined as the set of integers kk such that −⌈(n+1)M⌉≤k≤⌈(n+1)M⌉-\lceil (n+1)M \rceil \le k \le \lceil (n+1)M \rceil, where ⌈(n+1)M⌉\lceil (n+1)M \rceil is the value determined by `meshBound M n`.

definition

Mesh cell weight k/(n+1)k / (n + 1) at scale nn

Given a bound M∈RM \in \mathbb{R}, a discretization parameter n∈Nn \in \mathbb{N}, and an index kk from the set of mesh indices MeshIndex(M,n)\text{MeshIndex}(M, n), the function returns the real value kn+1\frac{k}{n + 1}. This value represents the "weight" of the kk-th mesh cell, corresponding to the left endpoint of that cell.

definition

The kk-th mesh piece of ff at scale nn: {x∈Ω∣⌊(n+1)f(x)⌋=k}\{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\}

Given a function f:Ω→Rf : \Omega \to \mathbb{R}, a real bound MM, a discretization parameter n∈Nn \in \mathbb{N} (defining a mesh width of 1/(n+1)1/(n+1)), and a mesh index k∈MeshIndex M nk \in \text{MeshIndex } M \ n, the set `meshPiece f M n k` is the subset of Ω\Omega consisting of all points xx such that ⌊(n+1)f(x)⌋=k\lfloor (n+1)f(x) \rfloor = k. This set is the pullback along ff of the kk-th mesh interval, representing the points where f(x)f(x) lies in the half-open interval [kn+1,k+1n+1)[\frac{k}{n+1}, \frac{k+1}{n+1}).

theorem

x∈meshPiece(f,M,n,k)  ⟺  ⌊(n+1)f(x)⌋=kx \in \text{meshPiece}(f, M, n, k) \iff \lfloor (n + 1) f(x) \rfloor = k

For a function f:Ω→Rf: \Omega \to \mathbb{R}, a real bound MM, a discretization parameter n∈Nn \in \mathbb{N}, and a mesh index k∈MeshIndex(M,n)k \in \text{MeshIndex}(M, n), a point x∈Ωx \in \Omega belongs to the kk-th mesh piece meshPiece(f,M,n,k)\text{meshPiece}(f, M, n, k) if and only if the floor of (n+1)f(x)(n + 1) f(x) is equal to kk: x∈meshPiece(f,M,n,k)  ⟺  ⌊(n+1)f(x)⌋=kx \in \text{meshPiece}(f, M, n, k) \iff \lfloor (n + 1) f(x) \rfloor = k

theorem

Measurability of the Mesh Pieces of ff

Let f:Ω→Rf : \Omega \to \mathbb{R} be a measurable function. For any discretization parameter n∈Nn \in \mathbb{N} and any mesh index k∈MeshIndex M nk \in \text{MeshIndex } M \ n, the kk-th mesh piece defined by the set {x∈Ω∣⌊(n+1)f(x)⌋=k}\{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\} is a measurable set.

definition

Mesh index k=⌊(n+1)f(x)⌋k = \lfloor (n+1) f(x) \rfloor for point xx

Given a function f:Ω→Rf: \Omega \to \mathbb{R} bounded by MM (such that ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega) and a discretization parameter n∈Nn \in \mathbb{N} (where the mesh width is 1/(n+1)1/(n+1)), this function maps a point x∈Ωx \in \Omega to its corresponding mesh cell index k=⌊(n+1)f(x)⌋k = \lfloor (n+1) f(x) \rfloor. This index is guaranteed to lie within the range of the mesh index set [−⌈(n+1)M⌉,⌈(n+1)M⌉][- \lceil (n+1)M \rceil, \lceil (n+1)M \rceil].

theorem

The mesh index of xx is ⌊(n+1)f(x)⌋\lfloor (n + 1) f(x) \rfloor

For a function f:Ω→Rf: \Omega \to \mathbb{R} bounded by MM, a natural number nn, and a point x∈Ωx \in \Omega, the mesh index of xx (viewed as an integer) is equal to the floor of the value (n+1)f(x)(n + 1) f(x), expressed as: meshIndexOf(x)=⌊(n+1)f(x)⌋ \text{meshIndexOf}(x) = \lfloor (n + 1) f(x) \rfloor

theorem

x∈meshPiece(f,M,n,meshIndexOf(hM,n,x))x \in \text{meshPiece}(f, M, n, \text{meshIndexOf}(hM, n, x))

Let f:Ω→Rf: \Omega \to \mathbb{R} be a function and MM be a real number such that ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega. For any n∈Nn \in \mathbb{N} and any x∈Ωx \in \Omega, let k=⌊(n+1)f(x)⌋k = \lfloor (n+1)f(x) \rfloor be the mesh index of xx (denoted as `meshIndexOf hM n x`). Then xx is an element of the kk-th mesh piece of ff at scale nn, which is defined as the set {y∈Ω∣⌊(n+1)f(y)⌋=k}\{y \in \Omega \mid \lfloor (n+1)f(y) \rfloor = k\}.

theorem

The mesh pieces of ff form a partition of Ω\Omega at scale nn

Let f:Ω→Rf: \Omega \to \mathbb{R} be a function bounded by M∈RM \in \mathbb{R} (i.e., ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega). For any natural number nn, the collection of mesh pieces Sk={x∈Ω∣⌊(n+1)f(x)⌋=k}S_k = \{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\}, indexed by k∈Zk \in \mathbb{Z} such that −⌈(n+1)M⌉≤k≤⌈(n+1)M⌉-\lceil (n+1)M \rceil \le k \le \lceil (n+1)M \rceil, forms a partition of the space Ω\Omega.

theorem

The value of the mesh simple function at xx is ⌊(n+1)f(x)⌋n+1\frac{\lfloor (n+1)f(x) \rfloor}{n+1}

Given a function f:Ω→Rf : \Omega \to \mathbb{R}, a bound M∈RM \in \mathbb{R}, and a discretization parameter n∈Nn \in \mathbb{N}, let the kk-th mesh piece be sk={x∈Ω∣⌊(n+1)f(x)⌋=k}s_k = \{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\} and the corresponding mesh weight be ck=kn+1c_k = \frac{k}{n+1}. For any x∈Ωx \in \Omega, the value of the simple function ∑kck1sk\sum_k c_k \mathbb{1}_{s_k} at xx is given by ⌊(n+1)f(x)⌋n+1 \frac{\lfloor (n + 1)f(x) \rfloor}{n + 1} where ⌊⋅⌋\lfloor \cdot \rfloor denotes the floor function.

theorem

Uniform error bound ∣gn(x)−f(x)∣≤1n+1|g_n(x) - f(x)| \le \frac{1}{n+1} for the mesh approximation gng_n

For any function f:Ω→Rf : \Omega \to \mathbb{R}, a real bound MM, and a discretization parameter n∈Nn \in \mathbb{N}, the mesh-based simple function approximation (with mesh width 1/(n+1)1/(n+1)) evaluated at any point x∈Ωx \in \Omega satisfies the inequality: ∣simpleValue(meshWeight M n,meshPiece f M n,x)−f(x)∣≤1n+1 \left| \text{simpleValue}(\text{meshWeight } M \ n, \text{meshPiece } f \ M \ n, x) - f(x) \right| \le \frac{1}{n + 1} Here, `meshPiece f M n` partitions Ω\Omega into sets where the value of f(x)f(x) falls into the kk-th mesh interval, and `meshWeight M n` assigns the value kn+1\frac{k}{n+1} to the kk-th piece. In essence, the mesh approximation gn(x)g_n(x) stays within 1/(n+1)1/(n+1) of f(x)f(x) for all xx.

theorem

Additivity of `simpleValue` in the weights

For a finite index set ι\iota, real coefficients a,b:ι→Ra, b : \iota \to \mathbb{R}, a family of sets s:ι→P(Ω)s : \iota \to \mathcal{P}(\Omega), and a point x∈Ωx \in \Omega, the value of the simple function ∑i∈ιci1si(x)\sum_{i \in \iota} c_i \mathbb{1}_{s_i}(x) is additive with respect to its coefficients: ∑i∈ι(ai+bi)1si(x)=∑i∈ιai1si(x)+∑i∈ιbi1si(x) \sum_{i \in \iota} (a_i + b_i) \mathbb{1}_{s_i}(x) = \sum_{i \in \iota} a_i \mathbb{1}_{s_i}(x) + \sum_{i \in \iota} b_i \mathbb{1}_{s_i}(x) where 1si\mathbb{1}_{s_i} is the indicator function of the set sis_i.

theorem

Scalar Multiplication Homogeneity of the Value of a Simple Function

Let ι\iota be a finite index set, r∈Rr \in \mathbb{R} be a scalar, and c:ι→Rc : \iota \to \mathbb{R} be a collection of coefficients. For any family of sets s:ι→P(Ω)s : \iota \to \mathcal{P}(\Omega) and any point x∈Ωx \in \Omega, the value of the simple function with scaled coefficients r⋅cr \cdot c at xx is equal to rr times the value of the simple function with coefficients cc at xx. In terms of the indicator function 1si\mathbb{1}_{s_i}, this states: ∑i∈ι(rci)1si(x)=r∑i∈ιci1si(x)\sum_{i \in \iota} (r c_i) \mathbb{1}_{s_i}(x) = r \sum_{i \in \iota} c_i \mathbb{1}_{s_i}(x)

theorem

The Intersection of Two Partitions is a Partition

Let Ω\Omega be a set, and let ι\iota and ι′\iota' be finite index sets. Suppose {si}i∈ι\{s_i\}_{i \in \iota} and {sj′}j∈ι′\{s'_j\}_{j \in \iota'} are partitions of Ω\Omega. Then the collection of pairwise intersections {si∩sj′}\{s_i \cap s'_j\} indexed by the product set ι×ι′\iota \times \iota' is also a partition of Ω\Omega.

theorem

∑ici1si=∑i,jci1si∩sj′\sum_i c_i \mathbb{1}_{s_i} = \sum_{i,j} c_i \mathbb{1}_{s_i \cap s'_j} for partitions {si}\{s_i\} and {sj′}\{s'_j\}

Let ι\iota and ι′\iota' be finite index sets, and let {si}i∈ι\{s_i\}_{i \in \iota} and {sj′}j∈ι′\{s'_j\}_{j \in \iota'} be partitions of a space Ω\Omega. For any coefficients ci∈Rc_i \in \mathbb{R} and any point x∈Ωx \in \Omega, the value of the simple function defined by coefficients cic_i and the partition sis_i at xx is equal to the value of the simple function defined by the same coefficients cic_i and the refined partition {si∩sj′}(i,j)∈ι×ι′\{s_i \cap s'_j\}_{(i,j) \in \iota \times \iota'} at xx. That is: ∑i∈ιci1si(x)=∑(i,j)∈ι×ι′ci1si∩sj′(x),\sum_{i \in \iota} c_i \mathbb{1}_{s_i}(x) = \sum_{(i, j) \in \iota \times \iota'} c_i \mathbb{1}_{s_i \cap s'_j}(x), where 1A\mathbb{1}_A denotes the indicator function of the set AA.

theorem

Equality of a simple function's value on a partition and its product partition refinement

Let Ω\Omega be a space, and let {si}i∈ι\{s_i\}_{i \in \iota} and {sj′}j∈ι′\{s'_j\}_{j \in \iota'} be partitions of Ω\Omega indexed by finite sets ι\iota and ι′\iota'. For any point x∈Ωx \in \Omega and any collection of real coefficients {cj′}j∈ι′\{c'_j\}_{j \in \iota'}, the value of the simple function defined by the coefficients cj′c'_j and the partition sj′s'_j at xx is equal to the value of the simple function defined over the product partition {si∩sj′}(i,j)∈ι×ι′\{s_i \cap s'_j\}_{(i,j) \in \iota \times \iota'} using the coefficients from the second index (i.e., cj′c'_j). That is, ∑j∈ι′cj′1sj′(x)=∑(i,j)∈ι×ι′cj′1si∩sj′(x),\sum_{j \in \iota'} c'_j \mathbb{1}_{s'_j}(x) = \sum_{(i, j) \in \iota \times \iota'} c'_j \mathbb{1}_{s_i \cap s'_j}(x), where 1A\mathbb{1}_A denotes the indicator function of the set AA.

theorem

The sum of two simple values is the simple value on their common refinement partition

Let Ω\Omega be a set and let s={si}i∈ιs = \{s_i\}_{i \in \iota} and s′={sj′}j∈ι′s' = \{s'_j\}_{j \in \iota'} be two partitions of Ω\Omega indexed by finite sets ι\iota and ι′\iota'. Let a:ι→Ra: \iota \to \mathbb{R} and b:ι′→Rb: \iota' \to \mathbb{R} be coefficients. For any x∈Ωx \in \Omega, the sum of the values of the two simple functions at xx is equal to the value of the simple function defined by the common refinement partition {si∩sj′}(i,j)∈ι×ι′\{s_i \cap s'_j\}_{(i, j) \in \iota \times \iota'} with summed weights: (∑i∈ιai1si(x))+(∑j∈ι′bj1sj′(x))=∑(i,j)∈ι×ι′(ai+bj)1si∩sj′(x)\left(\sum_{i \in \iota} a_i \mathbb{1}_{s_i}(x)\right) + \left(\sum_{j \in \iota'} b_j \mathbb{1}_{s'_j}(x)\right) = \sum_{(i, j) \in \iota \times \iota'} (a_i + b_j) \mathbb{1}_{s_i \cap s'_j}(x) where 1S\mathbb{1}_S denotes the indicator function of a set SS.

theorem

Equality of a simple integral and its product partition refinement

Let Ω\Omega be a space and μ\mu be an effect-valued measure on Ω\Omega taking values in EE. Let {si}i∈ι\{s_i\}_{i \in \iota} and {sj′}j∈ι′\{s'_j\}_{j \in \iota'} be partitions of Ω\Omega indexed by finite sets ι\iota and ι′\iota'. For any collection of real coefficients {ci}i∈ι\{c_i\}_{i \in \iota}, the simple integral of the function defined by cc and ss is equal to the simple integral defined over the product partition {si∩sj′}(i,j)∈ι×ι′\{s_i \cap s'_j\}_{(i, j) \in \iota \times \iota'} with coefficients determined by the first index of the product. Specifically, ∑i∈ιciμ(si)=∑(i,j)∈ι×ι′ciμ(si∩sj′).\sum_{i \in \iota} c_i \mu(s_i) = \sum_{(i, j) \in \iota \times \iota'} c_i \mu(s_i \cap s'_j).

theorem

The simple integral on a partition is equal to its integral on a refined partition (second component)

Let Ω\Omega be a measurable space and EE be a space of effects. Let μ\mu be an effect-valued measure on Ω\Omega taking values in EE. Let ι\iota and ι′\iota' be finite index sets. Suppose {si}i∈ι\{s_i\}_{i \in \iota} and {sj′}j∈ι′\{s'_j\}_{j \in \iota'} are two partitions of Ω\Omega. For any collection of real coefficients {cj′}j∈ι′\{c'_j\}_{j \in \iota'}, the simple integral of the function defined by c′c' on the partition s′s' is equal to the simple integral defined on the product refinement partition {si∩sj′}(i,j)∈ι×ι′\{s_i \cap s'_j\}_{(i, j) \in \iota \times \iota'} using the coefficients from the second index (cj′c'_j): ∑j∈ι′cj′μ(sj′)=∑(i,j)∈ι×ι′cj′μ(si∩sj′)\sum_{j \in \iota'} c'_j \mu(s'_j) = \sum_{(i, j) \in \iota \times \iota'} c'_j \mu(s_i \cap s'_j)

theorem

Difference of simple integrals over a common refinement

Let Ω\Omega be a measurable space and EE be an order-unit space. Let μ\mu be an effect-valued measure on Ω\Omega taking values in EE. Suppose {si}i∈I\{s_i\}_{i \in I} and {sj′}j∈J\{s'_j\}_{j \in J} are two finite partitions of Ω\Omega indexed by finite types II and JJ. For any collections of real coefficients {ci}i∈I\{c_i\}_{i \in I} and {cj′}j∈J\{c'_j\}_{j \in J}, the difference between their respective simple integrals is equal to the simple integral of the difference of the coefficients over the common refinement partition {si∩sj′}(i,j)∈I×J\{s_i \cap s'_j\}_{(i,j) \in I \times J}: ∑i∈Iciμ(si)−∑j∈Jcj′μ(sj′)=∑(i,j)∈I×J(ci−cj′)μ(si∩sj′) \sum_{i \in I} c_i \mu(s_i) - \sum_{j \in J} c'_j \mu(s'_j) = \sum_{(i, j) \in I \times J} (c_i - c'_j) \mu(s_i \cap s'_j)

theorem

The sum of two simple integrals is a simple integral over their common refinement

Let Ω\Omega be a measurable space and EE be an order-unit space. Let μ\mu be an effect-valued measure on Ω\Omega taking values in EE. Suppose {si}i∈I\{s_i\}_{i \in I} and {sj′}j∈J\{s'_j\}_{j \in J} are two finite partitions of Ω\Omega with corresponding real coefficients cic_i and cj′c'_j. The sum of their simple integrals is equal to the simple integral over the common refinement of the partitions {si∩sj′}(i,j)∈I×J\{s_i \cap s'_j\}_{(i, j) \in I \times J} with weights ci+cj′c_i + c'_j: ∑i∈Iciμ(si)+∑j∈Jcj′μ(sj′)=∑(i,j)∈I×J(ci+cj′)μ(si∩sj′)\sum_{i \in I} c_i \mu(s_i) + \sum_{j \in J} c'_j \mu(s'_j) = \sum_{(i, j) \in I \times J} (c_i + c'_j) \mu(s_i \cap s'_j)

theorem

The sum of an effect-valued measure over a partition is 11

Let μ\mu be an effect-valued measure on a space Ω\Omega with values in an order-unit space EE. For any finite partition {si}i∈ι\{s_i\}_{i \in \iota} of Ω\Omega into measurable sets, the sum of the measures of the sets in the partition is equal to the unit element (the certain outcome) 1∈E1 \in E: ∑i∈ιμ(si)=1\sum_{i \in \iota} \mu(s_i) = 1

theorem

The order-unit norm of a simple integral is bounded by the maximum absolute value of its coefficients.

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega with values in an order-unit space EE. Let {si}i∈I\{s_i\}_{i \in I} be a finite partition of Ω\Omega into measurable sets and ci∈Rc_i \in \mathbb{R} be coefficients for each i∈Ii \in I. If there exists a constant δ≥0\delta \geq 0 such that ∣ci∣≤δ|c_i| \leq \delta for every ii where the set sis_i is non-empty, then the order-unit norm of the simple integral of the function ∑ici1si\sum_i c_i \mathbf{1}_{s_i} against μ\mu is also bounded by δ\delta: ∥∑i∈Iciμ(si)∥≤δ\left\| \sum_{i \in I} c_i \mu(s_i) \right\| \leq \delta where ∥⋅∥\|\cdot\| denotes the order-unit norm in EE.

theorem

Non-negativity of simple integral for locally non-negative coefficients

Let μ\mu be an effect-valued measure on a space Ω\Omega taking values in an order-unit space EE. For a finite partition {si}i∈I\{s_i\}_{i \in I} of Ω\Omega and real coefficients ci∈Rc_i \in \mathbb{R}, if ci≥0c_i \geq 0 for all indices ii such that sis_i is non-empty, then the simple integral of the function ∑i∈Ici1si\sum_{i \in I} c_i \mathbf{1}_{s_i} against μ\mu is non-negative: ∑i∈Iciμ(si)≥0\sum_{i \in I} c_i \mu(s_i) \geq 0

theorem

∥∫g dμ−∫g′ dμ∥≤ε+ε′\|\int g \, d\mu - \int g' \, d\mu\| \le \varepsilon + \varepsilon' for simple functions g,g′g, g' uniformly approximating ff within ε,ε′\varepsilon, \varepsilon'

Let μ\mu be an effect-valued measure on a space Ω\Omega with values in an order-unit space EE equipped with the order-unit norm ∥⋅∥\|\cdot\|. Let f:Ω→Rf: \Omega \to \mathbb{R} be a function. Suppose gg and g′g' are two simple functions defined by finite partitions {si}i∈I\{s_i\}_{i \in I} and {sj′}j∈J\{s'_j\}_{j \in J} of Ω\Omega with real coefficients {ci}i∈I\{c_i\}_{i \in I} and {cj′}j∈J\{c'_j\}_{j \in J} respectively, such that g(x)=∑i∈Ici1si(x)g(x) = \sum_{i \in I} c_i \mathbf{1}_{s_i}(x) and g′(x)=∑j∈Jcj′1sj′(x)g'(x) = \sum_{j \in J} c'_j \mathbf{1}_{s'_j}(x). If there exist non-negative constants ε,ε′≥0\varepsilon, \varepsilon' \geq 0 such that ∣g(x)−f(x)∣≤ε|g(x) - f(x)| \le \varepsilon and ∣g′(x)−f(x)∣≤ε′|g'(x) - f(x)| \le \varepsilon' for all x∈Ωx \in \Omega, then the difference between their simple integrals against μ\mu satisfies: ∥∑i∈Iciμ(si)−∑j∈Jcj′μ(sj′)∥≤ε+ε′\left\| \sum_{i \in I} c_i \mu(s_i) - \sum_{j \in J} c'_j \mu(s'_j) \right\| \le \varepsilon + \varepsilon'

theorem

∥y∥<r  ⟹  y≤r⋅1\|y\| < r \implies y \leq r \cdot \mathbf{1}

Let EE be an order-unit space with order unit 1\mathbf{1} and order-unit norm ∥⋅∥\|\cdot\|. For any element y∈Ey \in E and any real number rr, if the order-unit norm of yy is strictly less than rr (i.e., ∥y∥<r\|y\| < r), then yy is bounded above by r⋅1r \cdot \mathbf{1} (i.e., y≤r⋅1y \leq r \cdot \mathbf{1}).

instance

EE is a Normed Additive Commutative Group under the order-unit norm

Let EE be an ordered additive commutative group equipped with an Archimedean order unit 1\mathbf{1}. The space EE is a normed additive commutative group under the order-unit norm ∥⋅∥\|\cdot\|, where for any x∈Ex \in E, the norm is defined as ∥x∥=inf⁡{r∈R∣−r⋅1≤x≤r⋅1}\|x\| = \inf \{ r \in \mathbb{R} \mid -r \cdot \mathbf{1} \le x \le r \cdot \mathbf{1} \}.

instance

EE is complete under the order-unit norm

Let EE be an ordered additive commutative group with an Archimedean order unit 1\mathbf{1}. The space EE is a complete metric space with respect to the order-unit norm ∥⋅∥\|\cdot\|, which is defined for any x∈Ex \in E as ∥x∥=inf⁡{r∈R∣−r⋅1≤x≤r⋅1}\|x\| = \inf \{ r \in \mathbb{R} \mid -r \cdot \mathbf{1} \le x \le r \cdot \mathbf{1} \}.

theorem

dist⁡(x,y)=∥x−y∥\operatorname{dist}(x, y) = \|x - y\| for the order-unit norm

Let EE be an ordered additive commutative group equipped with an Archimedean order unit. For any two elements x,y∈Ex, y \in E, the metric distance between them is equal to the order-unit norm of their difference, that is, dist⁡(x,y)=∥x−y∥\operatorname{dist}(x, y) = \|x - y\|.

theorem

If an→La_n \to L, then c⋅an→c⋅Lc \cdot a_n \to c \cdot L in the order-unit norm space EE

Let EE be an ordered additive commutative group with an Archimedean order unit 1\mathbf{1}, equipped with the order-unit norm ∥⋅∥\|\cdot\|. If a sequence (an)n∈N(a_n)_{n \in \mathbb{N}} in EE converges to a limit L∈EL \in E (with respect to the topology induced by the norm), then for any scalar c∈Rc \in \mathbb{R}, the sequence of scalar multiples (c⋅an)n∈N(c \cdot a_n)_{n \in \mathbb{N}} converges to c⋅Lc \cdot L.

theorem

The Mesh Approximation Sequence of Simple Integrals is Cauchy

Let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in a space EE equipped with an order-unit norm ∥⋅∥\|\cdot\|. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function bounded by M∈RM \in \mathbb{R} (i.e., ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega). For each n∈Nn \in \mathbb{N}, define the mesh pieces Sn,k={x∈Ω∣⌊(n+1)f(x)⌋=k}S_{n,k} = \{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\} and the weights wn,k=kn+1w_{n,k} = \frac{k}{n+1}. The sequence of simple integrals (In)n∈N(I_n)_{n \in \mathbb{N}} defined by In=∑kkn+1μ(Sn,k)I_n = \sum_k \frac{k}{n+1} \mu(S_{n,k}) is a Cauchy sequence in EE with respect to the order-unit norm.

definition

Integral ∫f dμ\int f \, d\mu of a bounded measurable function against an effect-valued measure μ\mu

Let EE be a complete Archimedean order-unit space equipped with the order-unit norm ∥⋅∥1\|\cdot\|_{\mathbf{1}}, and let μ\mu be an effect-valued measure on a measurable space Ω\Omega with values in EE. Given a measurable function f:Ω→Rf: \Omega \to \mathbb{R} and a real bound MM such that ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega, the integral of ff against μ\mu is defined as the limit in EE of the sequence of simple integrals (In)n∈N(I_n)_{n \in \mathbb{N}} as n→∞n \to \infty. This sequence is defined by a uniform mesh approximation: In=∑kkn+1μ({x∈Ω∣⌊(n+1)f(x)⌋=k})I_n = \sum_k \frac{k}{n+1} \mu(\{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\}) where the sum is taken over the finite set of indices kk such that the resulting sets partition Ω\Omega. The limit exists because the sequence InI_n is Cauchy and EE is assumed to be complete.

theorem

The Mesh Approximation Sequence Converges to the Integral

Let EE be a complete Archimedean order-unit space and μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function bounded by M∈RM \in \mathbb{R} (such that ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega). For each n∈Nn \in \mathbb{N}, define the mesh pieces Sn,k={x∈Ω∣⌊(n+1)f(x)⌋=k}S_{n,k} = \{x \in \Omega \mid \lfloor (n+1)f(x) \rfloor = k\} and the mesh weights wn,k=kn+1w_{n,k} = \frac{k}{n+1}. Then the sequence of simple integrals In=∑kkn+1μ(Sn,k)I_n = \sum_k \frac{k}{n+1} \mu(S_{n,k}) converges to the integral ∫f dμ\int f \, d\mu in the order-unit norm of EE as n→∞n \to \infty.

theorem

Uniform Convergence of Simple Functions Implies Convergence of Their Integrality against an Effect-Valued Measure

Let Ω\Omega be a measurable space and EE be a complete Archimedean order-unit space equipped with the order-unit norm ∥⋅∥1\|\cdot\|_{\mathbf{1}}. Let μ\mu be an effect-valued measure on Ω\Omega with values in EE. Suppose f:Ω→Rf: \Omega \to \mathbb{R} is a bounded measurable function. Consider a sequence of simple functions gn:Ω→Rg_n: \Omega \to \mathbb{R} defined by finite partitions {sn,i}i∈ιn\{s_{n,i}\}_{i \in \iota_n} and coefficients cn,ic_{n,i}, such that gn(x)=∑i∈ιncn,i1sn,i(x)g_n(x) = \sum_{i \in \iota_n} c_{n,i} \mathbf{1}_{s_{n,i}}(x). If the sequence gng_n converges uniformly to ff—that is, there exists a sequence of non-negative real numbers ϵn\epsilon_n such that lim⁡n→∞ϵn=0\lim_{n \to \infty} \epsilon_n = 0 and ∣gn(x)−f(x)∣≤ϵn|g_n(x) - f(x)| \le \epsilon_n for all x∈Ωx \in \Omega and n∈Nn \in \mathbb{N}—then the sequence of simple integrals ∫gn dμ=∑i∈ιncn,iμ(sn,i)\int g_n \, d\mu = \sum_{i \in \iota_n} c_{n,i} \mu(s_{n,i}) converges to the integral ∫f dμ\int f \, d\mu in EE as n→∞n \to \infty.

theorem

Independence of the integral ∫f dμ\int f \, d\mu from the choice of bound MM

Let EE be a complete Archimedean order-unit space and μ\mu be an effect-valued measure on a measurable space Ω\Omega with values in EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function. If MM and M′M' are two real numbers 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 integral of ff against μ\mu defined using the bound MM is equal to the integral defined using the bound M′M'.

theorem

The Integral of a Simple Function equals its Simple Integral

Let EE be a complete Archimedean order-unit space and μ\mu be an effect-valued measure on a measurable space Ω\Omega with values in EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function such that ∣f(x)∣≤M|f(x)| \le M for some real MM and all x∈Ωx \in \Omega. Suppose ff is a simple function represented by a finite partition {si}i∈I\{s_i\}_{i \in I} of Ω\Omega and coefficients ci∈Rc_i \in \mathbb{R} such that f(x)=∑i∈Ici1si(x)f(x) = \sum_{i \in I} c_i \mathbb{1}_{s_i}(x) for all x∈Ωx \in \Omega. Then the integral of ff against μ\mu coincides with the simple integral: ∫f dμ=∑i∈Iciμ(si)\int f \, d\mu = \sum_{i \in I} c_i \mu(s_i) where 1si\mathbb{1}_{s_i} denotes the indicator function of the set sis_i.

theorem

f≥0  ⟹  ∫f dμ≥0f \ge 0 \implies \int f \, d\mu \ge 0

Let Ω\Omega be a measurable space and EE be a complete Archimedean order-unit space. Let μ\mu be an effect-valued measure on Ω\Omega with values in EE. For any measurable function f:Ω→Rf: \Omega \to \mathbb{R} that is bounded by some M∈RM \in \mathbb{R} (i.e., ∣f(x)∣≤M|f(x)| \le M for all x∈Ωx \in \Omega), if ff is non-negative (f(x)≥0f(x) \ge 0 for all x∈Ωx \in \Omega), then its integral against μ\mu is also non-negative: 0≤∫f dμ0 \le \int f \, d\mu

theorem

∫(f+g) dμ=∫f dμ+∫g dμ\int (f + g) \, d\mu = \int f \, d\mu + \int g \, d\mu

Let EE be a complete Archimedean order-unit space and μ\mu be an effect-valued measure on a measurable space Ω\Omega with values in EE. For any two measurable functions f,g:Ω→Rf, g: \Omega \to \mathbb{R} such that ∣f(x)∣≤M|f(x)| \le M and ∣g(x)∣≤M′|g(x)| \le M' for all x∈Ωx \in \Omega, the integral of their sum f+gf + g against μ\mu is equal to the sum of their individual integrals: ∫(f+g) dμ=∫f dμ+∫g dμ\int (f + g) \, d\mu = \int f \, d\mu + \int g \, d\mu where the integral is defined as the limit of simple functions approximating the bounded measurable functions in the order-unit norm.

theorem

Homogeneity of the effect-valued integral: ∫(c⋅f) dμ=c⋅∫f dμ\int (c \cdot f) \, d\mu = c \cdot \int f \, d\mu

Let EE be a complete Archimedean order-unit space and let μ\mu be an effect-valued measure on a measurable space Ω\Omega taking values in EE. Let f:Ω→Rf: \Omega \to \mathbb{R} be a measurable function such that ∣f(x)∣≤M|f(x)| \leq M for all x∈Ωx \in \Omega. Then for any scalar c∈Rc \in \mathbb{R}, the integral of the scaled function c⋅fc \cdot f against μ\mu satisfies the homogeneity property: ∫(c⋅f) dμ=c⋅∫f dμ\int (c \cdot f) \, d\mu = c \cdot \int f \, d\mu where the integral is defined as the limit of a sequence of simple functions constructed via uniform mesh approximations.