mvdist_student_t
mvdist_student_t(df, location, scale=None)Construct a multivariate Student-t distribution.
Mirrors the C# MultivariateStudentT class of the Numerics library.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| df | float | Degrees of freedom. | required |
| location | array-like of float | Location parameters, length d. |
required |
| scale | array - like | d x d scale matrix; None (the default) uses the identity. |
None |
Returns
| Name | Type | Description |
|---|---|---|
| MultivariateDistribution | Family "MultivariateStudentT". |
Notes
MultivariateStudentT has no seed parameter here (unlike :func:mvdist_normal). Its CDF at dimension three and above draws 200 inner MultivariateNormal.cdf() evaluations from a clock-seeded stream that this class never lets the caller set, faithfully mirroring the upstream C# _MVNUNI = new MersenneTwister() default; that CDF is therefore not reproducible run to run, and R and Python cannot agree on a value there. :meth:MultivariateDistribution.random draws from a separate, genuinely seedable stream and is unaffected.
Examples
>>> mvdist_student_t(5, [0, 0]).dimension()
2