changelog

CHANGES.rst

@@ -9,8 +9,10 @@ Changes
 - Use a Cholesky decomposition to calculate values with uncertainty using
    `correlated_values` and `correlated_values_norm` when the provided covariance matrix
    or correlation matrix is positive definite. If the provided matrix is strictly
-   positive semi-definite an eigenvalue decomposition is still used. Refactor the code
-   and clean up the documentation for these functions. (#99)
+   positive semi-definite an eigenvalue decomposition is still used. Explicit exceptions
+   are raised if the user-provided matrix is non-real, non-symmetric, or non positive
+   semi-definite. The code and documentation for the two functions has been cleaned up.
+   (#99)
 - Changed how `numpy` is handled as an optional dependency. Previously,
    importing a `numpy`-dependent function, like `correlated_values`,
    without `numpy` installed would result in an `ImportError` at import