Simulation vs fitting

VineCopulas.jl v0.1 is a construction and evaluation package. It assumes that the user already knows the vine structure and the pair-copula parameters.

Simulation and evaluation

The package currently supports:

• constructing explicit C-vines, D-vines, and supported R-vines;

• evaluating pdf and logpdf;

• simulating with rand and simulate_qmc;

• computing Rosenblatt and inverse Rosenblatt transforms;

• computing loglikelihood, aic, and bic for explicit models.

Fitting and selection

The package does not yet estimate pair-copula parameters, select families, select structures, or choose truncation levels automatically. A future fitting layer will likely need:

1. pseudo-observation handling or integration with Copulas.pseudos,

2. pair-copula parameter estimation,

3. family selection using log-likelihood, AIC, BIC or other criteria,

4. tree selection using dependence scores such as Kendall's or Spearman's ,

5. truncation selection,

6. fitted wrapper types that preserve estimation metadata.

For now, users should transform raw data to the copula scale and construct a vine explicitly.