2022
DOI: 10.1112/jlms.12676
|View full text |Cite
|
Sign up to set email alerts
|

The isometry group of Wasserstein spaces: the Hilbertian case

Abstract: Motivated by Kloeckner's result on the isometry group of the quadratic Wasserstein space W2(R n ), we describe the isometry group Isom(Wp(E)) for all parameters 0 < p < ∞ and for all separable real Hilbert spaces E. In particular, we show that Wp(X) is isometrically rigid for all Polish space X whenever 0 < p < 1. This is a consequence of our more general result: we prove that W1(X) is isometrically rigid if X is a complete separable metric space that satisfies the strict triangle inequality. Furthermore, we s… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1

Citation Types

0
4
0

Year Published

2022
2022
2025
2025

Publication Types

Select...
6

Relationship

2
4

Authors

Journals

citations
Cited by 7 publications
(4 citation statements)
references
References 29 publications
0
4
0
Order By: Relevance
“…In fact, this method works in the 0<p<1$0&lt;p&lt;1$ case as well, but we decided to not include it in the main body. On the one hand, we have already proved in [10] that p ‐Wasserstein spaces are all isometrically rigid if 0<p<1$0&lt;p&lt;1$, regardless of what the underlying space is. On the other hand, as the definition of the p ‐Wasserstein distance is slightly different in the 0<p<1$0&lt;p&lt;1$ case, we should add one more branch to all proofs, without any serious novelty.…”
Section: The Wasserstein Potential Of Measuresmentioning
confidence: 99%
See 1 more Smart Citation
“…In fact, this method works in the 0<p<1$0&lt;p&lt;1$ case as well, but we decided to not include it in the main body. On the one hand, we have already proved in [10] that p ‐Wasserstein spaces are all isometrically rigid if 0<p<1$0&lt;p&lt;1$, regardless of what the underlying space is. On the other hand, as the definition of the p ‐Wasserstein distance is slightly different in the 0<p<1$0&lt;p&lt;1$ case, we should add one more branch to all proofs, without any serious novelty.…”
Section: The Wasserstein Potential Of Measuresmentioning
confidence: 99%
“…In this case, false(X,ϱfalse)$(X,\varrho )$ is a complete separable metric space and scriptS=scriptPfalse(Xfalse)$\mathcal {S}=\mathcal {P}(X)$. ‐The quadratic Wasserstein metric dW2$d_{\mathcal {W}_2}$ turned out to be very effective in a wide range of AI applications including pattern recognition and image processing problems. In these applications, false(X,rfalse)$(X,r)$ is typically the n ‐dimensional Euclidean space and S$\mathcal {S}$ is the collection of all Borel probability measures with finite second moment. In recent years, there has been a considerable interest in the characterisation of the above‐mentioned (and many other) metric spaces of measures, see, for example, [1, 3, 4, 6–12, 15, 17, 19]. In most cases, it turned out that isometries of S$\mathcal {S}$ are strongly related to self‐maps of the underlying space X .…”
Section: Introductionmentioning
confidence: 99%
“…[1, 3, 18, 19, 25, 26, 30-37, 46, 50-52, 57, 64]. We highlight three papers which deals with the structure of Wasserstein isometries over Euclidean spaces [33,35,46]. In [46] Kloeckner described the isometry group of the quadratic Wasserstein space W 2 (R n ).…”
Section: Introductionmentioning
confidence: 99%
“…In [46] Kloeckner described the isometry group of the quadratic Wasserstein space W 2 (R n ). Later in [33] and [35] we gave a complete characterisation of isometries of p-Wasserstein spaces over real and separable Hilbert spaces for all parameters 1 ≤ p < ∞.…”
Section: Introductionmentioning
confidence: 99%