BlockSeries should support scalars
Right now BlockOperatorSeries
does not support scalars, where shape=()
.
We'd like to implement the fibonacci series as an example, but right now it needs to be implemented with 1-dimensional arrays like:
F = BlockOperatorSeries(shape=(1,), n_infinite=1, data={(0,0):np.array([0]), (0,1):np.array([1])})
def eval(index):
return F.evaluated[0, index[1]-2] + F.evaluated[0, index[1]-1]
F.eval = eval
F.evaluated[0, 5]