See full documentation for detailed info.
OTT is a JAX toolbox that bundles a few utilities to solve optimal transport problems. These tools can help you compare and match two weighted point clouds (or histograms, measures, etc.), given a cost (e.g. a distance) between single points.
Most of OTT is, for now, supported by a sturdy, versatile and efficient implementation of the Sinkhorn algorithm that takes advantage of JAX features, such as JIT, auto-vectorization and implicit differentiation.
A typical OT problem has two ingredients: a pair of weight vectors a
and b
(one for each measure), with a ground cost matrix that is either directly given, or derived as the pairwise evaluation of a cost function on pairs of points taken from two measures. The main design choice in OTT comes from encapsulating the cost in a Geometry
object, and bundle it with a few useful operations (notably kernel applications). The most common geometry is that of two clouds of vectors compared with the squared Euclidean distance, as illustrated in the example below:
import jax
import jax.numpy as jnp
from ott.tools import transport
# Samples two point clouds and their weights.
rngs = jax.random.split(jax.random.PRNGKey(0),4)
n, m, d = 12, 14, 2
x = jax.random.normal(rngs[0], (n,d)) + 1
y = jax.random.uniform(rngs[1], (m,d))
a = jax.random.uniform(rngs[2], (n,))
b = jax.random.uniform(rngs[3], (m,))
a, b = a / jnp.sum(a), b / jnp.sum(b)
# Computes the couplings via Sinkhorn algorithm.
ot = transport.Transport(x, y, a=a, b=b)
P = ot.matrix
The call to sinkhorn
above works out the optimal transport solution by storing its output. The transport matrix can be instantiated using those optimal solutions and the Geometry
again. That transoprt matrix links each point from the first point cloud to one or more points from the second, as illustrated below.
To be more precise, the sinkhorn
algorithm operates on the Geometry
,
taking into account weights a
and b
, to solve the OT problem, produce a named tuple that contains two optimal dual potentials f
and g
(vectors of the same size as a
and b
), the objective reg_ot_cost
and a log of the errors
of the algorithm as it converges, and a converged
flag.
Currently implements the following classes and functions:
In the geometry folder,
The CostFn
class in costs.py and its descendants define cost functions between points. Two simple costs are currently provided, Euclidean
between vectors, and Bures
, between a pair of mean vector and covariance (p.d.) matrix.
The Geometry
class in geometry.py and its descendants describe a cost structure between two measures. That cost structure is accessed through various member functions, either used when running the Sinkhorn algorithm (typically kernel multiplications, or log-sum-exp row/column-wise application) or after (to apply the OT matrix to a vector).
In its generic Geometry
implementation, as in geometry.py, an object can be initialized with either a cost_matrix
along with an epsilon
regularization parameter (or scheduler), or with a kernel_matrix
.
If one wishes to compute OT between two weighted point clouds
and endowed with a
given cost function (e.g. Euclidean) , the PointCloud
class in pointcloud.py can be used to define the corresponding kernel
. When the number of these points grows very large, this geometry can be instantiated with an online=True
parameter, to avoid storing the kernel matrix and choose instead to recompute the matrix on the fly at each application.
Simlarly, if all measures to be considered are supported on a
separable grid (e.g. ), and the cost is separable
along all axis, i.e. the cost between two points on that
grid is equal to the sum of (possibly different) cost
functions evaluated on each of the pairs of coordinates, then
the application of the kernel is much simplified, both in log space
or on the histograms themselves. This particular case is exploited in the Grid
geometry in grid.py which can be instantiated as a hypercube using a grid_size
parameter, or directly through grid locations in x
.
In the core folder,
The sinkhorn
function in sinkhorn.py runs the Sinkhorn algorithm, with the aim of solving approximately one or various optimal transport problems in parallel. An OT problem is defined by a Geometry
object, and a pair (or batch thereof) of histograms. The function's outputs are stored in a SinkhornOutput
named t-uple, containing potentials, regularized OT cost, sequence of errors and a convergence flag. Such outputs (with the exception of errors and convergence flag) can be differentiated w.r.t. any of the three inputs (Geometry, a, b)
either through backprop or implicit differentiation of the optimality conditions of the optimal potentials f
and g
.
In discrete_barycenter.py: implementation of discrete Wasserstein barycenters : given histograms all supported on the same Geometry
, compute a barycenter of theses measures, using an algorithm by Janati et al. (2020)
In gromov_wasserstein.py: implementation of the Gromov-Wasserstein metric between measured-metric spaces, here encoded as a pair of Geometry
objects along with weights a
and b
.
In the tools folder,
In soft_sort.py: implementation of soft-sorting operators, notably soft-quantile transforms
The sinkhorn_divergence
function in sinkhorn_divergence.py, implements the unbalanced formulation of the Sinkhorn divergence, a variant of the Wasserstein distance that uses regularization and is computed by centering the output of sinkhorn
when comparing two measures.
The Transport
class in sinkhorn_divergence.py, provides a simple wrapper to the sinkhorn
function defined above when the user is primarily interested in computing and storing an OT matrix.
Disclaimer: this is not an official Google product.