The Complex Geometry and Representation Theory of Statistical Transformation Models
Shuhao Li

TL;DR
This paper explores the geometric and representation-theoretic structures of statistical transformation models, linking Kähler geometry, Lie group actions, and unitary representations within the context of probability measures and exponential families.
Contribution
It establishes a connection between the geometry of tangent bundles of exponential models and unitary representations derived from Lie group actions, extending the orbit method to statistical models.
Findings
Tangent bundles of exponential models have Kähler structures similar to Fubini-Study.
The group actions on tangent bundles are equivariant with those on $L^2$ spaces.
Coadjoint orbits induce irreducible unitary representations related to the $L^2$ representation.
Abstract
Given a measure space , we can construct a number of induced structures: eg. its space, the space of probability distributions on . If, in addition, admits a transitive measure-preserving Lie group action, natural actions are induced on those structures. We expect relationships between these induced structures and actions. We study, in particular, the relations between and exponential transformation models on , which are special "submanifolds" of closed under the induced action, whose tangent bundles are K\"ahler manifolds (given by Molitor). Geometrically, we show the tangent bundle has, locally, the "same" K\"ahler metric with the Fubini-Study metric on the projectivization of . Moreover we show the action on the tangent bundle…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGeometry and complex manifolds · Geometric Analysis and Curvature Flows · Black Holes and Theoretical Physics
