recurrence in numpy / numexpr python

Question

I have an array x=np.ones(1000) and would like to perform operations on this based on recurrence formulae:

    x[i]= f(x[i-1], x[i-2])

What is the fastest way to do it in a vectorize way ?

I know that xx= ne.evaluate(xx1*xx2) can only manage already filled array.

Could you share more details about the func `f`? – Divakar Jan 24 '16 at 05:28 — Divakar, Jan 24 '16 at 05:28
f is using standard math function like sin, cos, exp,.. – Jan 24 '16 at 06:44 — , Jan 24 '16 at 06:44

score 0 · Answer 1 · answered Feb 10 '16 at 22:00

0

I found a partial way, at least for some functions:

list[:,1:]= np.exp(hlist)  
list = np.cumprod(list, axis = 1)

with f(x) = f(x-1)exp(ax)

I think generic cum would be useful

answered Feb 10 '16 at 22:00

1 Answers1