This pull requests improves conditioning of the outlier component, using a small multiple of the identity . The documentation is also updated. An inconsequential typo was also corrected in the prior mean calculation.
Note machine precision is used rather than 0 to improve catching of numerical issues.
This pull requests improves conditioning of the outlier component, using a small multiple of the identity . The documentation is also updated. An inconsequential typo was also corrected in the prior mean calculation.
Note machine precision is used rather than 0 to improve catching of numerical issues.