Skip to content

[TCopula] Multivariate CDF implementation and validation #406

Description

@lrnv

Motivation

The multivariate Student copula CDF is currently a distinct numerical problem from Spearman rho and moment matching tracked in #316. PR #330 contains an exploratory QMC implementation, but it should be reviewed independently because it changes the public numerical behavior of cdf.

Proposed scope

  • Specify the numerical algorithm and cite its primary references.
  • Define accuracy, error-estimation, RNG and reproducibility semantics.
  • Cover central values, extreme tails, singular or nearly singular correlation matrices, and Float32/Float64 behavior.
  • Compare against trusted external reference values and the current implementation where applicable.
  • Add focused Tachometer benchmarks across representative dimensions.
  • Keep the implementation independent from the Student rho/moment-matching work.

Historical work

The exploratory implementation and reference fixtures currently live in #330, which should be split before further review.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    new featureNew feature or request

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions