OTT-JAX

Leverages JAX for automatic differentiation and GPU/TPU acceleration, enabling scalable OT computations on large datasets, as highlighted in the JAX integration feature.

Implements discrete solvers like Sinkhorn with scheduling and momentum, plus advanced problems like Gromov-Wasserstein and Wasserstein barycenters, covering a wide range of OT tasks.

Provides neural network approaches to approximate optimal transport maps between measures, bridging discrete and continuous methods for flexible modeling.

Supports diverse geometric domains and custom cost functions for comparing point clouds, allowing versatile applications beyond standard Euclidean spaces.