The measures noise-connected to the reference measure have finite second moment and are pairwise noise-connected, noise-optimal couplings exist, and the noise Wasserstein distance is a metric on them that dominates the quadratic Wasserstein distance up to a constant and is lower semicontinuous under weak convergence.
In the setting of Probability Measures on a Hilbert Space Transported in the Noise Norm: Standing Notation, with the reference measure of Probability Measures on a Hilbert Space Transported in the Noise Norm: Standing Notation §reference, the sets of couplings of finite noise cost and noise-connectedness are those of Couplings of Finite Noise Cost and Their Noise Cost; is the noise Wasserstein distance, a nonnegative real-valued function on the noise-connected ordered pairs of members of ; noise-optimal couplings are as defined there; and is the set of measures noise-connected to . Then claims 1 to 3 hold, claims 4 to 6 hold for all , and claims 7 to 9 hold.
1. (The reference measure and finite second moments) . .
2. (Noise-connectedness) For any two members of , distinct or not, the ordered pair is noise-connected, so that is defined.
3. (Noise-optimal couplings) For all such that is noise-connected there is a noise-optimal coupling .
4. (Symmetry) .
5. (Separation) if and only if .
6. (Triangle inequality) .
7. (The noise Wasserstein space) The restriction of to pairs of members of is a metric on ; the metric space is called the noise Wasserstein space based at .
8. (Comparison) For all , which belong to by claim 1, the quadratic Wasserstein distance satisfies , with the positive bound on the noise weights of Probability Measures on a Hilbert Space Transported in the Noise Norm: Standing Notation §weights.
9. (Lower semicontinuity) Let for and satisfy and , and suppose that the sequence of real numbers is bounded. Then is noise-connected and , the limit inferior of that bounded sequence. If moreover , then .
Loading…
No relations recorded yet.