We learn the score function for discretisations of infinite-dimensional stochastic differential equations. We use this to bridge the SDEs conditional on an end set.
To install the package, run the following command in the root directory of the repository (sdebridge):
pip install .Alternatively, to run examples and tests too, install everything with the following command:
pip install sdebridge[full]The package provides a class SDE, for defining SDEs.
Definitions for the class and specific SDEs including Brownian motion and Kunita SDEs are provided in the module sdebridge.sdes.
In order to define reverse bridges, we can either define a score function or learn it using the module sdebridge.diffusion_bridge.
Examples of how to use the package are provided in the notebooks directory.
Please cite our article as follows
@article{baker2024conditioning,
title={Conditioning non-linear and infinite-dimensional diffusion processes},
author={Elizabeth Louise Baker and Gefan Yang and Michael L. Severinsen and Christy Anna Hipsley and Stefan Sommer},
year={2024},
eprint={preprint arXiv:2402.01434},
}