v1.2

Flatten the state before transport to handle multidimensional data

v1.1

What's Changed

• Fix an edge case (1-D Gaussians) by @AdrienCorenflos in #5

Full Changelog: v1...v1.1

v1

Gaussian tails experiment (#4)

* c(x) and right solver implmemented

* left exp exp solver

* fixed equation solvers and implemented various other parts

* tp and tq sampling

* Gamma_hat and other yet untested parts added

* p and q sampling

* Converted while loops to while_loops

* First version that appears to work

* Unit tests and fixes

* Experiment draft

* coupled truncated exponentials in functional paradigm + logspace

* Made gauss tails residual using quick solver.

* I think this works just fine. Let's just design a bunch of unittest, document and merge.

* Make experiment use the "new" code.

* Visibly some issue with the second marginal of the exponential coupling.

* Finished refactoring and changed experiment

* Finalise Gaussian tails experiment.

* Some fixes on sampling tails + better test using p-value.

* truncnorm was failing.

* Fixed the Gaussian tails.

* Started the reflected max example

* Reflected mh (#3)

* need to commit before holidays

* This should be the reflected mh implemented.

* Finalised MMALA example

* add some replication

* Update README.md

Co-authored-by: Simo Särkkä <[email protected]>