Changes - Return to using the SciPy multivariate normal distribution.
0.3.5
Changes - Remove the condition number control from the matrix module
0.3.4
Changes - Improve stability of likelihood sampling by using hand rolled likelihood function. - Reverted the removal the `rtol` parameter from the `bootstrap` method.
0.3.4a
See [kinisi-0.3.4](https://github.com/bjmorgan/kinisi/releases/tag/0.3.4). This release is to trigger the PyPI distribution release.
0.3.3
Changes - Improve stability of covariance matrix - Remove the `rtol` parameter from the `bootstrap` method.
0.3.2
Changes - Reverted to using likelihood sampling, which with the fixes to the covariance matrix gives the same results as the bootstrap approach.