PhyslibAlpha.AlgebraicFramework.OrderUnit.Effect.BoundedIntegral
Integrating bounded measurable functions against an effect-valued measure
i. Overview
`Effect/Integral.lean` defines `EffectValuedMeasure.simpleIntegral`. This file extends it to bounded measurable `f : Ω → ℝ`, assuming `E` is complete for its order-unit norm.
ii. Construction
For a bound `M` with `|f x| ≤ M` and scale `n`, `meshPiece f M n` and `meshWeight M n` define a finite simple approximation with uniform error at most `1/(n+1)`. The comparison lemma for two simple approximations makes these integrals Cauchy and proves that every uniformly approximating sequence has the same limit. The resulting integral is independent of the bound, agrees with `simpleIntegral`, and is linear and positive.
iii. Key definitions and results
- `EffectValuedMeasure.meshBound`, `MeshIndex`, `meshWeight`, `meshPiece` : the width-`1/(n+1)` mesh simple function approximating a bounded measurable `f` with `|f x| ≤ M`, and its uniform convergence to `f` (`abs_simpleValue_meshWeight_meshPiece_sub_le`). - `EffectValuedMeasure.orderUnitNorm_simpleIntegral_sub_le` : the key comparison lemma — two simple functions within `ε`, `ε'` of `f` give simple integrals within `ε + ε'` of each other. - `EffectValuedMeasure.cauchySeq_meshSimpleIntegral`, `EffectValuedMeasure.integral` : the mesh sequence is Cauchy, and (under `[CompleteSpace E]`, fixed via `letI`/`local instance` from `IsArchimedeanOrderUnit.orderUnitNormedAddCommGroup`) `integral hf hM μ` is its limit. - `EffectValuedMeasure.tendsto_of_uniformly_approximating` : independence from the approximating sequence — any uniformly-approximating simple-function sequence has the same limit. - `EffectValuedMeasure.integral_indep_of_bound`, `integral_eq_simpleIntegral` : independence from the bound `M`, and agreement with `simpleIntegral` on functions that already are simple. - `EffectValuedMeasure.integral_add`, `integral_smul`, `nonneg_integral` : linearity and positivity.
iv. Table of contents
- A. Mesh approximation
- B. Linear operations on simple values
- C. Comparison of simple integrals
- D. Construction by 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
`E`'s order-unit norm (`IsArchimedeanOrderUnit.orderUnitNorm`) makes it a normed group via `IsArchimedeanOrderUnit.orderUnitNormedAddCommGroup`, deliberately not a registered instance at this level of generality (see `OrderUnit/Norm.lean`); we fix it as a `local instance` for this section (the term-mode `letI` the file's plan mentions is the tactic-mode spelling of the same thing; at the section/command level `local instance` is what registers it for the elaborator), and additionally assume `[CompleteSpace E]` under *that* instance — a genuine extra hypothesis, since most order-unit spaces are not complete.
40 declarations
Mesh bound
For a given real-valued bound and a natural number , the mesh bound is the natural number . This value represents the number of mesh cells, each of width , required to cover the distance from up to .
Mesh index set
For a real-valued bound and a natural number representing the scale, the mesh index set is the type of integers such that . This set provides the labels for the mesh cells of width used to partition the range during the approximation of a bounded measurable function.
Mesh weight
For a given real-valued bound , a scale , and a mesh index (where ), the mesh weight is defined as the real number . This value represents the left endpoint of the -th cell in a uniform partition of width used to approximate a bounded measurable function.
Mesh cell
For a function , a real-valued bound , a scale , and an index belonging to the mesh index set (where ), the mesh piece is the set of points such that the floor of is equal to . Mathematically, this is the pre-image , representing the subset of the domain where falls into the -th mesh interval .
Let be a function, be a bound, and be a scale parameter. For any point and any index in the mesh index set , belongs to the mesh cell if and only if the floor of is equal to , i.e., .
The mesh cell is a measurable set
Let be a measurable function. For any real-valued bound , any scale , and any index in the mesh index set , the mesh cell defined by the set is a measurable set.
Mesh index of a point
For a measurable function bounded by (such that for all ) and a scale , the function `meshIndexOf` assigns to each point an integer index . This index represents the mesh cell that contains the value and is proven to satisfy the bound , making it a valid element of the mesh index set .
The mesh index of equals
Let be a measurable function bounded by , such that for all . For any scale and any point , the integer value of the mesh index assigned to is equal to .
belongs to the mesh piece of its index
Let be a function bounded by (i.e., for all ). For any scale and any point , belongs to the mesh piece (cell) corresponding to the index .
The mesh pieces form a partition of
For a function bounded by (such that for all ) and a scale , the collection of mesh pieces , where ranges over the mesh index set , forms a partition of the domain .
Value of the mesh simple function at is
For a real-valued function , a bound , and a scale , the value of the mesh simple function at a point —constructed using the weights and the partition —is given by
Uniform Convergence of the Mesh Approximation
For a function and a real-valued bound , let be the mesh simple function approximation at scale , which is defined by the weights and the mesh pieces . For any point , the absolute difference between the simple function value and the function value satisfies: where represents the mesh resolution.
For an index set , a collection of sets , weight functions , and a point , the evaluation of the simple function at is additive with respect to the weights. That is, where denotes the pointwise addition of the weights.
Let be a scalar, be a function representing weights, be a collection of sets, and be a point. The value of the simple function (defined by the weights and the sets ) at the point is homogeneous with respect to the weights. That is, where denotes the pointwise scalar multiplication of the weights.
The Intersection of Two Partitions is a Partition
Let be a set, and let and be finite index sets. If and are both partitions of , then the collection of pairwise intersections is also a partition of .
Equality of a simple function's value and its value on a refined product partition
Let be a set, and let and be two finite partitions of . Given a set of coefficients , for any , the value of the simple function defined by at is equal to the value of the simple function defined on the product partition using the coefficients for each pair . That is, , where denotes the indicator function of a set .
Equality of simple function values under partition refinement (second component)
Let be a space, and let and be two finite partitions of . For any collection of coefficients and any point , the value of the simple function associated with coefficients and partition at is equal to the value of the simple function defined on the product partition where the coefficient for the index is . That is,
Pointwise Sum of Simple Functions via Common Refinement of Partitions
Let be a set, and let and be finite index sets. Let and be partitions of . For any weight functions and , and for any point , the sum of the simple functions defined by and at is equal to the simple function defined by the common refinement partition with weights . Mathematically: , where denotes the value at of the simple function associated with weights and partition .
Invariance of Simple Integral under Partition Refinement (First Component)
Let be a set and be an effect-valued measure on taking values in a space . Let and be finite index sets, and let and be two partitions of . For any collection of coefficients , the integral of the simple function defined by the weights and the partition is equal to the integral of the simple function defined on the product partition where the weight for the index is given by . That is,
Simple Integral over a Partition Equals its Integral over a Common Refinement (Second Component)
Let be an effect-valued measure on a space with values in . Let and be two finite partitions of . For any weight function , the simple integral of with respect to the partition is equal to the simple integral over the common refinement partition where the weights are given by the second component . Mathematically:
The Difference of Two Simple Integrals is a Simple Integral over the Common Refinement Partition
Let be an effect-valued measure on a space taking values in . Suppose and are two finite partitions of with corresponding weight functions and . The difference of the two simple integrals is equal to the simple integral over the common refinement partition with weights given by the pointwise difference . That is:
The Sum of Two Simple Integrals is a Simple Integral over the Common Refinement Partition
Let be an effect-valued measure on a space taking values in . Suppose and are two finite partitions of with corresponding weights and . The sum of the two simple integrals is equal to the simple integral over the common refinement partition with weights given by the pointwise sum . That is:
for a finite partition
Let be an effect-valued measure on a space with values in , where is an ordered space with a unit element . If is a finite partition of , then the sum of the measures of the sets is equal to the unit element of :
The order-unit norm of a simple integral is bounded by the maximum weight
Let be an effect-valued measure on a space with values in an ordered space possessing an Archimedean order unit. Let be a finite partition of and be a set of weights. If there exists a constant such that for every where is non-empty, then the order-unit norm of the simple integral of these weights against is bounded by : where denotes the order-unit norm.
The simple integral is non-negative if weights are non-negative on all non-empty pieces of the partition.
Let be an effect-valued measure on a space taking values in an ordered space . Let be a finite partition of and let be a collection of weights. If for every such that the set is non-empty, the weight is non-negative (), then the simple integral of these weights against is non-negative:
for simple functions approximating within
Let be an effect-valued measure on a space with values in an ordered space possessing an Archimedean order unit. Let be a function. Suppose and are two real-valued simple functions on defined by finite partitions and with weights and , respectively. If there exist non-negative real numbers such that the simple functions satisfy and for all , then the order-unit norm of the difference of their simple integrals satisfies: where and are the simple integrals of and against .
The order-unit norm of a scalar multiple satisfies
Let be an ordered space with an Archimedean order unit . For any real scalar and any element , the order-unit norm satisfies the inequality: where the order-unit norm is defined as the infimum of all such that .
Let be an ordered space with an Archimedean order unit . For any element and any real number , if the order-unit norm of (denoted ) satisfies , then .
Normed additive commutative group structure on induced by the order-unit norm
The space is equipped with the structure of a normed additive commutative group. This structure is induced by the order-unit norm associated with an Archimedean order unit , where the norm of an element is defined as . This structure provides the necessary metric and topological framework for the space, where the distance between two elements is given by .
For any elements in the space , the distance is equal to the order-unit norm of their difference, .
Scalar Multiplication Preserves Sequence Convergence:
Let be a space equipped with an Archimedean order-unit norm . For any sequence in and any limit , if converges to (with respect to the topology induced by the order-unit norm), then for any real scalar , the scaled sequence converges to .
The sequence of mesh simple integrals is Cauchy
Let be a measurable function such that for some bound and all . Let be an effect-valued measure on taking values in an Archimedean order-unit space equipped with the order-unit norm . For each , let be the integral of the simple function approximating at scale , defined as: where the sum ranges over the mesh index set . Then the sequence is a Cauchy sequence in .
Integral of a bounded measurable function against an effect-valued measure
Let be a measurable function bounded by a constant (such that for all ), and let be an effect-valued measure on taking values in a space that is complete with respect to its order-unit norm. The **integral** of against is defined as the limit of the integrals of a sequence of simple functions that approximate uniformly. Specifically, it is the limit as of the mesh simple integrals: where the sum is taken over integers such that . This limit exists because the sequence of simple integrals is Cauchy in the complete space .
Mesh simple integrals converge to the integral
Let be a measurable function bounded by , such that for all . Let be an effect-valued measure on taking values in a space that is complete with respect to its order-unit norm. For each , consider the mesh simple integral defined as: where the sum is taken over the mesh index set . Then the sequence converges to the integral of against as in the topology of the order-unit norm:
Simple integrals of uniform approximations of converge to
Let be a sample space and be an order-unit space that is complete with respect to its order-unit norm. Let be an effect-valued measure on taking values in , and let be a measurable function bounded by a constant (i.e., for all ). Consider a sequence of simple functions , each defined by a finite index set , real weights , and a measurable partition of , such that . If the sequence converges to uniformly—specifically, if there exists a sequence of non-negative real numbers such that as and for all and —then the sequence of simple integrals of against converges to the integral of against in the norm of : This result ensures that the integral of is independent of the specific choice of approximating simple functions, provided they converge uniformly to .
The Integral of a Bounded Function against an Effect-Valued Measure is Independent of the Bound
Let be a measurable function and be an effect-valued measure on taking values in a space that is complete with respect to its order-unit norm. Suppose and are real numbers such that and for all . Then the integral of against constructed using the bound is equal to the integral constructed using the bound , i.e., where is the proof of measurability of , and are the proofs of boundedness by and respectively.
The Integral of a Simple Function Equals its Simple Integral
Let be an effect-valued measure on a space taking values in a complete space equipped with an order-unit norm. Let be a measurable function such that for some . If is a simple function defined by a finite partition of and weights such that for all , then the integral of against (constructed via uniform limits) is equal to its simple integral:
If , then for effect-valued measures
Let be a measurable space and be an Archimedean order-unit space that is complete with respect to its order-unit norm. Let be an effect-valued measure on with values in . If is a measurable function bounded by a constant (such that for all ) and is non-negative (i.e., for all ), then its integral against is also non-negative:
Additivity of the effect-valued integral:
Let be a measurable space and be a space that is complete with respect to its order-unit norm. Let be measurable functions such that and for all and constants . For an effect-valued measure on taking values in , the integral of the sum is equal to the sum of the individual integrals: This property holds assuming the measurability and boundedness of (with bound ), which are derived from the properties of and .
Homogeneity of the effect-valued integral:
Let be a measurable function bounded by a constant (i.e., for all ), and let be an effect-valued measure taking values in a complete order-unit space . For any scalar , let be the function . If is measurable and bounded by , then the integral of against satisfies:
