Scores on the Configuration Space: the Score of a Tensor Power is the Product Field of the Score, and the Score of the One-Particle Marginal is the Projection of the Score
lemmaAnalysisProbabilitylem:score-tensor-marginal-wasserstein-2026aA tensor power of a measure with finite Fisher information has finite Fisher information, with the product field of the score as its score and N times the Fisher information; and the one-particle marginal of a measure on with finite Fisher information has as score the one-particle projection of its score, with Fisher information at most 1/N times.
In the setting of Wasserstein Spaces, Random Vectors, Vector Fields and Symmetric Matrices in Every Dimension: Standing Notation, let . Tensor powers are those of The Tensor Power of a Probability Measure on Euclidean Space §tensor, one-particle marginals those of The One-Particle Marginal of a Probability Measure on the Configuration Space §marginal (with ), and product fields and projections those of Product Fields and the Projection onto One-Particle Tangent Fields. For and , the tangent space , finite Fisher information, the score and the Fisher information are those definitions read with in place of : they are stated for the arbitrary dimension of The Wasserstein Space and Its Lift to Square-Integrable Random Vectors: Standing Notation §data and use nothing of it beyond , so they apply in every dimension . is the set of measures in with finite Fisher information. For , by Tensor Powers and One-Particle Marginals: Particle Laws, Product Integrals, Push-Forwards, Moments, Product Maps and Diagonal Shifts §moments. The letter denotes a probability measure on ; the probability space of the setting is not used.
1. (Tensor powers)¶ For , , and .
2. (One-particle marginals)¶ For , , and . In particular for every .
Loading…
Prerequisites
No prerequisites tracked.
Dependents
No dependents yet.
Dependent proofs
No dependent proofs yet.
No relations recorded yet.