Currently, p-value computation is done via the comparison of each entry in the partial correlation matrix to the empirical distribution in a vanilla way that takes quite some time (I would guess infeasible for more than ~100 predictors).
We have some algorithmic improvements, including parallelization, that need to be added so that this function is of practical use.
In the meantime, it would be good to also add a warning message to the code for users who try to run estimate_p_values_mc so that they aren't waiting a long time for results that are not going to appear in a reasonable time frame.
Currently, p-value computation is done via the comparison of each entry in the partial correlation matrix to the empirical distribution in a vanilla way that takes quite some time (I would guess infeasible for more than ~100 predictors).
We have some algorithmic improvements, including parallelization, that need to be added so that this function is of practical use.
In the meantime, it would be good to also add a warning message to the code for users who try to run estimate_p_values_mc so that they aren't waiting a long time for results that are not going to appear in a reasonable time frame.