Skip to content

Releases: mathematicalmichael/mud

readthedocs release

12 Jul 18:18
01ef9d6
Compare
Choose a tag to compare

automatic building of documentation using readthedocs. creating tag so that stable points to a good looking document.

update packaging + other quality of life improvements

11 Jul 18:03
77cc5e8
Compare
Choose a tag to compare

:shipit: packaging now compliant with PEP 517 + PEP 518
🌟 linting with black + flake8
🔧 fixed codecov history + added coveralls
🥇 updated workflows
📛 new badges in README

maps for repeated observations

25 Apr 01:15
2daea51
Compare
Choose a tag to compare

more efficient computation for maps which represented repeated measurements on the same quantity of interest

stable MAP solution

16 Mar 20:51
13946c8
Compare
Choose a tag to compare

verified by playing with up to 500 measurements over a bunch of examples.

stability bugfixes with map solution for high N, low sd

15 Mar 06:46
13946c8
Compare
Choose a tag to compare

fix for v0.0.22, which was published from the wrong commit. this version removes division by the evidence when computing the posterior distribution. since it is a constant, there is no benefit in doing this when using it for parameter estimation.

adds runtime

21 Feb 07:28
82d29f6
Compare
Choose a tag to compare

a wip, but in the spirit of main reflecting the latest versions... publishing.

package cleanup

20 Feb 22:09
fd0749f
Compare
Choose a tag to compare

removing some dead weight.

optional log-posterior for numerical stability

15 Feb 09:22
Compare
Choose a tag to compare

experimenting with how to compute the map point, particularly with small standard deviations (0.1) and many measurements (over 500), revealed stability problems (overflow errors). this release gives the option to pass log=True to map_problem

handle multiple QoI

08 Feb 08:56
dd280fa
Compare
Choose a tag to compare

support for vector-valued MUD solutions. no checks on max number of output dimensions, allowed to exceed input dimension, may cause uncertain behavior. may change in future to restrict to max equal to input dimension.

observed distribution already has .prod(axis=1) so support should be handled by default. needs tests.

bugfix for initial / prior densities

08 Feb 08:07
f8d3f9c
Compare
Choose a tag to compare

now uniforms are set as the default, not normals.