Objective
The goal of this in-person workshop is to make progress in tackling the most difficult* parameter inference problems using state-of-the-art simulation-based inference techniques as implemented in Swyft**.
*Difficult = slow simulator/model; dozens, thousands or millions of parameters; complex posterior; unknown optimal summary statistics; ...
**Swyft is a pytorch-lightning-based implementation of truncated marginal neural ratio estimation (TMNRE), which is now successfully used for a diverse set of inference problems.
Swyft in action
Further information
The workshop is planned to be in-person. If you are interested to join online, please let us know. Depending on demand, we might change part of the workshop to a hybrid version.