The Setting
Split the rows and the columns of a random array into three blocks each. Given the array of parameters of the entries and the cells of one block, the cells of a row block that lie in the two other column blocks form a joining, and so do the cells of a column block that lie in the two other row blocks; the Markov operators of these joinings make a square of four operator families between four spaces. The question is whether fixed identifications bring the unitary parts of their polar decompositions into one commutative group of unitaries.
Fix a measurable automorphism of a standard Borel space and a finite index set . An array of order over a space is a point of , an entry of for every index tuple in ✦; a plane of an array of order is the array of order obtained by holding one of its indices at a slot .✦ The initial array laws of are probability measures on , one for each order , invariant under applied entrywise, consistent and proper in the sense defined below for the laws of an array object.✦ An array object over the initial array laws of is a standard Borel space of states with an automorphism , an equivariant factor map to and an equivariant parameter map to a standard Borel space with an automorphism , together with probability measures on , invariant under applied entrywise, carried to by the factor map, and consistent, proper and conditionally local in the sense defined below.✦ The laws of an array object are consistent if every plane of an array of order has law under , and proper if, under , the planes along the last axis at the slots of a proper subset of are independent. The laws are conditionally local if, given the parameter array, obtained by applying entrywise, the planes along the last axis at the slots of a proper subset of stay independent, each with the conditional law of an array of order given its own parameters. Under , whose arrays are points of , the factor map and are independent.
An extension of is an array object over the same initial laws with measurable equivariant maps from its states and its parameters to those of , compatible with the factor and parameter maps, whose state map, applied entrywise, carries the law of to at every order.✦ For an extension of , write for composition with the entrywise state map, the pullback along it; this is a linear isometry from to .✦ The data fields of an array object are -algebras of arrays of order : the parameter field , the outside fields and the enlarged fields , with the index of the field. An array object is saturated if, for every extension of and every order , on for each data field, where and are the orthogonal projections onto the functions measurable for that data field of and of .✦
For a parameter array of order , write for the conditional law under of an array given that its parameter array is , and for the law of the parameter array under .
A triple partition of is an ordered triple of nonempty, pairwise disjoint subsets of with union .✦ Call the first index of an array of order its row index and the last its column index, and fix triple partitions of the rows and of the columns, and an order . In an array of order , the cell in row and column is the array of order obtained by holding the first index at and the last at ✦, and is the family of cells with row in and column in .✦ The square reads through four blocks: (1)
Fix a parameter array of order . The cell law is the law of when the rows , , arrays of order obtained by holding the first index at , are drawn independently, each from with the row of at .✦ The row joining operator is the Markov operator of the joining of and given by their conditional law under , given that the parameter array is and that the block is ; it takes a function of the second member to its conditional expectation given the first. The row joining operator is the Markov operator of the joining of and given by their conditional law under , given that the parameter array is and that the block is ; it takes a function of the second member to its conditional expectation given the first.
Along the columns, given and , the columns , , arrays of order obtained by holding the last index at , are drawn independently, each from the conditional law under given that its cells in the rows of are those of , with the column of at .✦ The column joining operator is the Markov operator of the resulting joining of and . The column joining operator is the Markov operator of the joining of and obtained from the columns , , drawn independently, each from the conditional law under given that its cells in the rows of are those of .
Polar Rectangles
With , , and , the four families form a square: (2)
The essential kernel of a family of bounded operators from to on a measure space is the subspace of vectors with for almost every ✦, and its essential range is the orthogonal complement in of the essential kernel of the adjoint family .✦ Write , , , for the families , , , . Under the cell laws , , and of the four blocks, the rectangle spaces are (3)
The compression of a bounded operator to closed subspaces and is the operator that applies and projects orthogonally onto ; compressing gives families , , and .✦
Write for the unitary operators from a Hilbert space onto a Hilbert space , and , a topological group in the strong operator topology. A measurable map into is a measurable polar decomposition of a family of operators from to if, for almost every , the operator is injective with dense range and , where .✦
Let be a measure on a space , let be maps from to spaces with measures , and let be operator families indexed by these spaces and placed between Hilbert spaces as in (2). A measurable polar rectangle over these data is a choice of measurable polar decompositions of , of , of and of , under the respective measures, with (4)✦ The grouped square at is the square of the compressed families between for the fixed triple partitions. A polar rectangle of the grouped square at is a measurable polar rectangle over the compressed families, with the space of arrays of order under , the blocks (1) and the cell laws .✦
Haar Normalization
In a group, is the left translation . For Hilbert spaces and fixed unitaries , and , the normalization maps carry the four unitary spaces of the square into : (5) for , , and .
Probability measures on , , and have a Haar normalization if there are unitaries as above and a closed abelian subgroup of with (6)✦ A measurable polar rectangle has a Haar normalization when the laws , , and of its four maps, its polar variables, have a Haar normalization.✦
The Claim
Theorem 1. (i) Let be a second-countable Hausdorff topological group in which the indicator of every closed set is a pointwise limit of continuous functions equal to on that set. Let be a measure on a measurable space , let and be s-finite kernels from to , and let and be probability measures on . Assume that for -almost every both marginals of are , both marginals of are , and (7) that and ; and that at some point the first marginals of and are and , the pair lies in the supports of and of , and (7) holds at . Then there is a closed abelian subgroup of with (8)✦
(ii) Let be complex Hilbert spaces, a probability measure on , a Markov kernel from to and one from to , and probability measures on these four unitary spaces. Assume that for -almost every the marginals of are and , those of are and , and (7) holds with products read as compositions, and that (9)✦ Then have a Haar normalization.✦
(iii) Let be a saturated array object, with triple partitions of the rows and of the columns and an order fixed as above. For -almost every parameter array of order , (10)✦
(iv) Let be a complex Hilbert space, a closed abelian subgroup of , and a probability measure on with support and for every . Then there are a countable set , a Hilbert basis of and continuous homomorphisms from to the unit circle of , with its Borel -algebra, such that (11)✦
(v) Let , and be as in (iv), and let be a probability measure on with support . Let be a measure on a space and maps from to spaces . Let on and on be measurable maps into carrying measures and to , and on and on measurable maps into carrying and to . Let and be Markov kernels from to , the conditional laws under of and of given . Assume that for almost every under the law of both marginals of are , both marginals of are , and (7) holds with for . Then there are a countable set and measurable maps from to such that the square-integrable functions of and those of form the same closed subspace of , those of and of the same closed subspace of , and, -almost everywhere, (12)✦
In the paper, the result reads: after fixed unitary identifications of the four spaces with one Hilbert space , the four polar variables take values in one closed abelian subgroup of with common support ; and have a common left-invariant law ; , where is a function of the variable ; has a simultaneous orthonormal eigenbasis; and the monomials in its circle coordinates at span the square-integrable functions of and are exact for the row and column three-cell laws, the laws of three cells along a row and along a column.
What Is Checked
The claim is stated over saturated array objects with their laws and their conditional laws given parameters, triple partitions of the rows and the columns, the cell laws and the row and column joining operators of the grouped square, their compressions to the rectangle spaces, measurable polar decompositions and polar rectangles, and Haar normalization; its general parts are stated over kernels into a topological group or into spaces of unitaries, and over closed abelian groups of unitaries with a left-invariant law.
Beyond the claim, the library proves that the entrywise state map of an extension is measure-preserving from the law of the extension to the law of at every order✦, and that the essential kernel of an operator family is a linear subspace. It also proves that the essential range of an operator family is a closed subspace, so that the four rectangle spaces are closed subspaces.