Second-order models for optimal transport and cubic splines on the Wasserstein Space

DOI10.1007/S10208-019-09425-ZMaRDI QIDQ2329043zbMATH OpenOpenAlexFDO

Authors Jean-David Benamou, Thomas O. Gallouët, François-Xavier Vialard

Publication date 17 October 2019

Published in Foundations of Computational Mathematics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1801.04144

zbMATH Keywords

cubic splinesmultimarginal optimal transportWasserstein geodesics

Mathematics Subject Classification ID

Numerical methods in optimal control (49M99) Optimal transportation (49Q22) Numerical approximation and computational geometry (primarily algorithms) (65D99) Transportation, logistics and supply chain management (90B06)

Abstract: On the space of probability densities, we extend the Wasserstein geodesics to the case of higher-order interpolation such as cubic spline interpolation. After presenting the natural extension of cubic splines to the Wasserstein space, we propose a simpler approach based on the relaxation of the variational problem on the path space. We explore two different numerical approaches, one based on multi-marginal optimal transport and entropic regularization and the other based on semi-discrete optimal transport.

Review text

The authors give a new variational method to introduce a notion of cubic splines on the space of probability densities. The starting point is the definition of cubic splines on Riemannian manifolds. The authors reformulate it in a way which admits a generalization to the Wasserstein space, i.e. space of probability densities with finite second moment equipped with the 2-Wasserstein metric. The cubic interpolants are defined as follows: for a given sequence \(x_n\) of points and \(t_n\) of times, we take the uniquely defined curve \(x(t)\) such that \(x(t_i) = x_i\) which minimizes the acceleration (to be exact, the integral of the square of the second derivative). This theoretical introduction is complemented with two numerical techniques. The first one is based on multimarginal optimal transport and the second one is based on semi-discrete optimal transport.

Recommendations

• Measure-Valued Spline Curves: An Optimal Transport Viewpoint
• Second-order in time schemes for gradient flows in Wasserstein and geodesic metric spaces
• Existence and uniqueness of optimal transport maps obtained by the secondary variational method
• Second order derivative of a functional associated to an optimal transport map
• Optimal transport and curvature
• Optimal Transport Approximation of 2-Dimensional Measures
• Partial \(W^{2,p}\) regularity for optimal transport maps
• Sobolev estimates for optimal transport maps on Gaussian spaces
• Quadratically regularized optimal transport
• Multi-marginal optimal transport on Riemannian manifolds

Cites work

• A general condition for Monge solutions in the multi-marginal optimal transport problem
• A numerical algorithm for \(C^2\)-splines on symmetric spaces
• A numerical algorithm for \(L_2\) semi-discrete optimal transport in 3D
• A numerical method to solve multi-marginal optimal transport problems with Coulomb cost
• An interpolating distance between optimal transport and Fisher-Rao metrics
• Cubic Splines on Curved Spaces
• Diagonal Equivalence to Matrices with Prescribed Row and Column Sums
• Generalized incompressible flows, multi-marginal transport and Sinkhorn algorithm
• Invariant higher-order variational problems
• Invariant higher-order variational problems. II
• Iterative Bregman projections for regularized transportation problems
• Measure-Valued Spline Curves: An Optimal Transport Viewpoint
• Minimal Geodesics Along Volume-Preserving Maps, Through Semidiscrete Optimal Transport
• Multi-marginal optimal transport: theory and applications
• Optimal Transport
• Optimal transport for applied mathematicians. Calculus of variations, PDEs, and modeling
• Second-order models for optimal transport and cubic splines on the Wasserstein Space
• Shape splines and stochastic shape evolutions: a second order point of view
• Some geometric calculations on Wasserstein space
• Splines of class C^k on non-euclidean spaces
• The dynamic interpolation problem: on Riemannian manifolds, Lie groups, and symmetric spaces
• THE GEOMETRY OF DISSIPATIVE EVOLUTION EQUATIONS: THE POROUS MEDIUM EQUATION
• The Geometry of Infinite-Dimensional Groups
• The Least Action Principle and the Related Concept of Generalized Flows for Incompressible Perfect Fluids
• Variational time discretization of Riemannian splines

Cited in

(14)

• Second-order models for optimal transport and cubic splines on the Wasserstein Space
• A structural model on a hypercube represented by optimal transport
• Measure-Valued Spline Curves: An Optimal Transport Viewpoint
• Optimal transportation, modelling and numerical simulation
• The GenCol Algorithm for High-Dimensional Optimal Transport: General Formulation and Application to Barycenters and Wasserstein Splines
• Toward a mathematical theory of trajectory inference
• Subdivision scheme for discrete probability measure-valued data
• Approximation of splines in Wasserstein spaces
• Computing regularized splines in the Riemannian manifold of probability measures
• Cubic Hermite splines on the Hilbert manifold of probability measures
• Metric extrapolation in the Wasserstein space
• Cubic Hermite interpolators on the space of probability measures
• From kinetic theory to AI: a rediscovery of high-dimensional divergences and their properties
• Flows on measure spaces and applications in machine learning. Abstracts from the workshop held March 22--27, 2026

Describes a project that uses

Uses Software

• GitHub

This page was built for publication: Second-order models for optimal transport and cubic splines on the Wasserstein Space

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2329043)