mathurinm/celer

BUG: fitting small X and y raises an index out of bound exception

Closed this issue · 2 comments

sklearn example:
X = [[0,0], [1, 1], [2, 2]]
y = [0, 1, 2]

celer(X, y, alpha=0.1) raises an error

It seems related to p0 = 10 being the default. If n_features < p0 something crashes.

np.argsort(a, k=len(a)) caused the crashes. Fixed in #6