Closed Virtakuono closed 8 years ago
Actually for MLMC, mods=[-1, 1] For MIMC, things are a bit more complicated. I just wanted a general mechanism for both and for Stochastic Collocation in the future.
Okay so inds gives a list of multi-indices and mods gives the weights of those? As in for MLMC I should be looking at mods = [-1 1] and inds = [n-1,n] for some integer n?
inds = [[n-1], [n]] weird for MLMC, but makes sense for MIMC
Ah, so inds is a list of not multi-indices, but list of lists of actual indices?
inds is a list of multi-indices a multi-index is just another list
I am somewhat puzzled about the meaning of the parameter mods in https://github.com/StochasticNumerics/mimclib/blob/693653b4868a74a9db50ea4f98cd776441a0e3e4/tests/gbm/run.py#L104
This is supposed to be an array of doubles, I presume? I don't quite get why anyone would want to call this function with anything but np.ones(1) . What am I missing here?