Julien Schueller
Julien Schueller
you need to call the function on a sample, else you're just evaluating sequentially: ``` import openturns as ot import time def pyf(x): time.sleep(2) print(x[0]) return [1] if __name__ ==...
it boils down to Normal zero quantile returning inf ```ot.Normal().computeScalarQuantile(0.0)``` as we changed the quantiles to return infinity instead of the numerical bounds: https://github.com/openturns/openturns/pull/2516 It seems restoring it for DistFunc.qNormal...
in fact thats the same as #2723
the nan comes from the composition with a LinearFunction, where it goes through blas (gemv)
yes, maybe with sparse product
yes, its transposed in LinearFunction
Anne: The two classes LinearFunction and LinearEvaluation are note coherent. **LinearEvaluation**: this class implements the evaluation of the function: $f : \mathbb{R}^d \rightarrow \mathbb{R}^p$ such that: $$ f(X) = A^t...
I dont see it either, could you rebase just in case ? then I will try to rebuild it sequentially
I reabsed your branch and know how to fix it, do you want me to fix & force push ?
dont worry its fixed in master, it will go away after a rebase