On Wasserstein Geometry of the Space of Gaussian Measures
نویسنده
چکیده
Abstract. The space which consists of measures having finite second moment is an infinite dimensional metric space endowed with Wasserstein distance, while the space of Gaussian measures on Euclidean space is parameterized by mean and covariance matrices, hence a finite dimensional manifold. By restricting to the space of Gaussian measures inside the space of probability measures, we manage to provide detailed descriptions of the Wasserstein geometry from a Riemannian geometric viewpoint. In particular, using the results from the Monge-Kantrovich transport theory, an explicit expression of geodesics interpolating two Gaussian measures. It follows that the space of Gaussian measures is geodesically convex in the space of probability measures. Also, a Riemannian metric which induces the Wasserstein distance is specified. Using the Riemannian metric, a formula for the sectional curvatures of the space of Gaussian measures on the plane is written out in terms of the eigenvalues of the covariance matrix.
منابع مشابه
Wasserstein Geometry of Gaussian Measures
This paper concerns the Riemannian/Alexandrov geometry of Gaussian measures, from the view point of the L2-Wasserstein geometry. The space of Gaussian measures is of finite dimension, which allows to write down the explicit Riemannian metric which in turn induces the L2-Wasserstein distance. Moreover, its completion as a metric space provides a complete picture of the singular behavior of the L...
متن کاملBehaviors of φ-Exponential Distributions in Wasserstein Geometry and an Evolution Equation
A φ-exponential distribution is a generalization of an exponential distribution associated to functions φ in an appropriate class, and the space of φ-exponential distributions has a dually flat structure. We study features of the space of φexponential distributions, such as the convexity in Wasserstein geometry and the stability under an evolution equation. From this study, we provide the new c...
متن کاملWasserstein distances for discrete measures and convergence in nonparametric mixture models
We consider Wasserstein distance functionals for comparing between and assessing the convergence of latent discrete measures, which serve as mixing distributions in hierarchical and nonparametric mixture models. We explore the space of discrete probability measures metrized by Wasserstein distances, clarify the relationships between Wasserstein distances of mixing distributions and f -divergenc...
متن کاملConvergence of latent mixing measures in nonparametric and mixture models
We consider Wasserstein distance functionals for assessing the convergence of latent discrete measures, which serve as mixing distributions in hierarchical and nonparametric mixture models. We clarify the relationships between Wasserstein distances of mixing distributions and f -divergence functionals such as Hellinger and Kullback-Leibler distances on the space of mixture distributions using v...
متن کاملLearning in Wasserstein Space
Learning from empirical probability measures is an emerging problem that has potential benefiting multiple domains. My research focuses on developing scalable and effective learning algorithms that handle large-scale data in form of measures. In particular, the Wasserstein space provides a powerful geometry that houses the compositions of data, which can be of great interest to domain experts t...
متن کامل