0.17.10: Parameterize starting mean (#100)

* Handles getting parameters deeper than one level

* Improvements to local linear trend

* Ups version

* Passes kwargs

* Black formatting

* Minor improvements

* Bug fix for prediction

* Adds test for prediction

* Adds new model and tests, and fixes SemiLocalLinear

* Applies black

* Bug fix

* Bug fix again

* Removes todo

* Improvements to local_linear_trend.py

* Improves testing suite and adds test for smoothing

* Record states is now on filter

* Fixes smoothing

* Correct initialization

* Fixes smoothing for parameterized models and adds test

0.17.9: Attempt joint series (#99)

* Uses monte carlo samples for estimating the noise variance instead, and increase number of samples

* Adds more steps

* Moves `num_steps` to base class instead as not unique to diffusion

* Ups version

* Correct slicing and event_shape

* Adds first iteration of JointState

* Adds first iteration of JointStochasticProcess and Affine version

* `functional_parameters` now take function as input

* Adds filtering tests

* Concat along -1

* TODO and check

* Adds a couple of models

* Adds arg_constraints as property instead

* Do not validate args

* Fix for batched

* Up version

* Adapt `VariationalBayes` after `num_steps` changes

0.17.8

Fixes and improves logging for inference module

0.16.3

Hotfix

0.15.1

Changes from module to mixin

0.14.2

Bug fix

0.14.0: Modularization (#90)

* Initial rewrite to use torch.nn.Module

* Minor changes

* Dunno

* Dunno

* Fix

* Dunno what happened

* More changes

* Changes and cleanup

* PMMH fix

* Some changes

* Minor changes

* NB updates

* Black

* Doc

0.13.2: Pmmh (#87)

* Adds PMMH

* Lowers step size

* Test

* Include all states

* Only pass state

* Passes observation as well, and also adds first iteration of gradient based proposal

* Updates README

* Version update

Co-authored-by: Victor <[email protected]>

0.13.1

Update NB

0.11.1: To functionality (#84)

* Result is not saved on filter

* Improves VB somewhat

* Type checking

* Small improvement

* Exception

* Type check

* Improvements to VariationalBayes

* Removes PMMH, need better implementation

* Better inheritance?

* NB updates

Co-authored-by: Victor <[email protected]>