The Setting
A conditioned local purification need not have unit length. Its normalization cost is measured by the squared Hilbert-Schmidt distance to a unit vector, averaged over the original question distribution.
A finite probability law on a finite set assigns nonnegative weights with sum one.✦ The expectation of a vector-valued function under a finite probability law is the weighted sum in (1). (1) The mixture of laws with mixing law is the probability law assigning the weight . The point mass at is the probability law assigning weight one to and zero elsewhere.✦ The pushforward of a probability law under a function is the mixture of the point masses with mixing law .✦ For a finite coordinate set and , the coordinate marginal of a probability law on is the pushforward of by evaluation at .✦ The product law of probability laws is the probability law assigning the weight .✦
Let be finite nonempty question alphabets, a finite coordinate set, and a finite nonempty record set. A question profile on the coordinate set with question alphabets is a function ; write .✦ Fix a law on and its product law . Choose default questions , and an ordered list of distinct coordinates.
For a complex matrix , its squared Hilbert-Schmidt norm is .✦ A coefficient matrix on finite local bases determines the bipartite vector .✦ For local operators and on Alice’s and Bob’s spaces, respectively, the bipartite operator expectation and its Born probability are the scalars defined by (2): (2)
A local quantum event system on question profiles , with record set and local bases , consists of a coefficient matrix with , local effects and , and weights . Alice’s effect is independent of all Bob questions, and Bob’s effect is independent of all Alice questions. The weight is independent of both questions at each coordinate in . The event likelihood of a local quantum event system is the function in (3), with values at most one and positive mean ; the conditioned density of the system is the function defined there: (3)
Conditional Exposure
Normalized nonnegative weights are divided by their total mass when that mass is positive; at zero mass they give a point mass at the chosen default.✦ The conditional laws on and on are normalized slices of , with defaults and , respectively.
For a question profile , coordinate and question , the left coordinate replacement is the question profile obtained by changing Alice’s question at to , retaining Bob’s question and all other coordinates.✦ The right coordinate replacement of a question profile , at coordinate with question , changes Bob’s question at to and retains every other question.✦ A function on is left coordinate invariant at when its value is unchanged by every replacement with , and right coordinate invariant at when its value is unchanged by every replacement with . For and a function into a vector space , the left and right conditional averaging operators are given by (4): (4) Iterated conditional averaging along a list composes these operators in list order, with the last operator applied first and the empty composition equal to the identity. Alice’s and Bob’s exposure chains are the families in (5), for : (5) The exposure index space is .✦ Alice’s exposure effect family assigns the matrix ✦; Bob’s exposure effect family assigns it .✦
Common Purifications
For a finite family of positive effects, the resolvent of at is ✦ and its purification kernel is in (6). The joint purification kernel is the horizontal concatenation of the kernels of the family.✦ The common Gram matrix of the effect family is the matrix defined by the integral in (6): (6) The finite common purification at index is the th block column of .✦ The common resolvent purification of an effect in the family is the block column of corresponding to a chosen occurrence of in .✦ The local purification maps are defined by (7): (7)
The weighted block state with weights and coefficient matrices is the block diagonal matrix in (8): (8)✦ Write for replacement of the entire question pair at by . The conditional block state associated with local maps and question pair at coordinate is the coefficient matrix in (9), with block index : (9) For , the joint conditional state and the local marginal states are defined by (10): (10) These coefficient matrices have row index and column index .
For a unit coefficient matrix on the enlarged coefficient spaces, Hilbert-Schmidt normalization with fallback is the map in (11). Alice’s and Bob’s normalized purifications are and : (11)
Relative Entropy
A law is absolutely continuous with respect to , written , when entails .✦ Their finite relative entropy is the sum in (12), with terms having assigned zero. All logarithms are natural. (12)✦ For a nonnegative function with , the tilted law on is defined by (13): (13)✦
The Claim
Theorem 1. For a local quantum event system as in (3), the conditioned density satisfies (14) at every question profile: (14)✦
Let be a finite probability law and a nonnegative function with . If and for every , the tilted relative entropy satisfies (15): (15)✦
Let be a law on for a finite coordinate set , and let be laws on . If , its coordinate marginals obey the relative entropy tensorization inequality (16): (16)✦
For every unit fallback matrix on the enlarged spaces, the local normalization costs satisfy (17): (17)✦ The normalized approximation errors satisfy (18): (18)✦
What Is Checked
The library’s statements concern finite probability laws with the stated support conditions and local quantum event systems with distinct exposure coordinates, using common resolvent purifications and an arbitrary unit fallback matrix.