You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks @marouenbg@blawney! @marouenbg no working example needed, I asked @blawney to raise this so we can correct the ambiguity of the exception.
As Marouen mentioned, the problem and fix(workaround) are described in detail in #298. The remaining issue is that we need to add an informative exception so that instead of seeing "f(a) and f(b) must have different signs", the user is given a description of the issue and suggested to use the MC p-value calculation rather than the parametric version.
Outside of the exception mentioned above, is the current recommended solution to catch the exception from the estimate_p_values_dragon and then subsequently fall back to estimate_p_values_mc with the same function arguments?
Currently I am not planning to implement the default as fall back to estimate_p_values_mc at this point because it is somewhat time-consuming. We could provide the user an estimated run time for the MC calculation in the exception, I think, based on the size of their dataset.
That said, we are working on algorithmic improvements and parallelization and once those are in place, I think it would make sense to redirect as you describe above.
The
estimate_p_values_dragon
function is raising an exception stating "f(a) and f(b) must have different signs"The text was updated successfully, but these errors were encountered: