* Change ac_noise_var into float, int or dict

* Add checks for acq_noise_var

* Move acq_noise_var tests to acquisition.py.

* Fix lint.

* Add and modify tests for new acquisition API.

* Add CHANGELOG.

* Fix test_bo.py

* Fix doctrings that refer to AcquisitionBase.

* Remove check for None.

* Store parameter_names in GPyRegression

* Fix a bug on acquisition-tests.

* Modify documents to follow new API

* Fix a bug in docstring.

* Fix a bug in tests.

* update array check

* remove obsolete float conversion

* update CHANGELOG

* Update CHANGELOG.rst

Co-authored-by: Henri Pesonen <[email protected]>

* Require a dictionary for sigma_proposals to be fed in metropolis.

* Fix a bug in type annotation

* initial commit

* update cost function

-standardise input and output shape
-include documentation

* add tests

* add type check

* fix style issues

* update CHANGELOG

* Minor style changes

Co-authored-by: Henri Pesonen <[email protected]>

* Update and fix tox.ini

* Add CHANGELOG

* Update MANIFEST.in

* Update CHANGELOG

* Use always surrogate parameter_names in BOLFI to avoid bugs.

* Add CHANGELOG.rst

* Use target_model.parameters_names instead of model.parameter_names.

* use batch system to run simulations

* clean

-rearranged new functions and added docstrings
-removed unused functions and updated unit tests

* update CHANGELOG

* fix typo

* fix random state setting

* update likelihood construction

* use negative log-posterior in minimisation

* make minimisation optional

* update CHANGELOG

* update tests

* fix style issue

Co-authored-by: Henri Pesonen <[email protected]>

* Add model details to ARCH-docstring

* Add CHANGELOG

* fix nodes type

* add new test

* update CHANGELOG

* add len check

* update CHANGELOG

Co-authored-by: Henri Pesonen <[email protected]>

* add optional input variable feature_names

* make training data size not optional

* update tests

* fix style issue

* update CHANGELOG

* add control over parameter order in model prior

add parameter names as optional input variable in model prior and use that to control parameter order in the priors used in BOLFI and BOLFIRE

* update CHANGELOG

* Add option to disable observation noise from lotka-volterra.

* Modify observation_noise-option logic.

* Simplify allowed true_parameters -option.

* Change logger.errors to ValueErrors.

* Add option to disable observation noise from lotka-volterra.

* Modify observation_noise-option logic.

* Simplify allowed true_parameters -option.

* Add summary statistics to LV

* Add sumstats to LV

* Add CHANGELOG.rst

* Enable batch evaluation.

* Add option to disable observation noise from lotka-volterra.

* Modify observation_noise-option logic.

* Simplify allowed true_parameters -option.

* Add summary statistics to LV

* Add sumstats to LV

* Add CHANGELOG.rst

* Fix linting in arch.py

* Reformat summary-output.

* Add CHANGELOG.rst

* Bring back sample_means_summary

* Remove double definition of weighted sample quantiles

* Revert to older definition of sample quantiles.

* fix plot_state

* update CHANGELOG

* Change LV priors from poisson to normal approximations.

* Approximate poisson distributions with gaussians.

* Add CHANGELOG.rst