1.4.1

[afernandezdemarcos]

This version includes:

• New K-fold method unif_test_cv(), and its internal functions:
  • lambda_hat(): Obtains optimal parameters for each K-fold.
  • k_fold_stat(): Computes approximate optimal statistic (and the corresponding parameter with lambda_hat) for each K-fold.
  • p_value_hmp(): Aggregates multiple p-values using the Harmonic Mean P-value method.
  • k_fold_split(): Given a sample size n and a number of folds K, it returns a partition of the sample of roughly equal sizes.
• New null_var_Sobolev() function that computes the variance of the Sobolev statistic under the uniform distribution.
• Modification of weights_dfs_Sobolev, and transformation between Sobolev coefficients b_k, u_k, and v_k^2 to include the computation of weight k = 0, whenever K_start = 0.

Only "Poisson", "Softmax", and "Stereo" statistics are available for these functions.