python - Apply function to 2D numpy array elements -

i've seen post , want similar, not same.

i implementing little game of life game , using numpy arrays representing states of game. need check, how many alive neighbors cell has. got function getting window of neighbors given coordinate , row count , column count window size want have.

so windows of 3x3 size this:

t = true f = false [[t,t,t],  [f,t,t],  [f,f,f]]  # random truth values 

in representation true stands cell being alive. wrote code iterating on cells of state, counting true values , on using double loop, think there better numpy solution.

what i'd in naive approach:

1. iterate on cells of state (not window) (i'd formulate code to executed if cell meets criteria or (being alive , surviving or being dead , coming alive))
2. get window (wrapping or not wrapping) (function have)
3. check if current cell alive (could lookup in state's numpy array)
4. if alive start alive neighbors count of -1 otherwise start 0
5. count true values of window (np.sum) , add alive neighbors count (which -1 if cell alive, count neighbors not cell itself)
6. depending on whether count of alive neighbors in ranges (configurable), write in (new) state's array true values. (i'd start out array, created using: np.full((height, width), false, dtype=bool))
7. go on new array, keeping old 1 in list history or logging purposes

basically:

if cell meets criteria:     write true @ cell's position in new array 

however meeting criteria depends on multiple rows, because state's numpy array 2d array. that's why think linked post close not need.

how can in efficient numpy-y way, avoiding unnecessary looping?

clarification

i searching best way of implementing in python using numpy , scipy, aims readable , has performance.

perhaps did not understand trying do, stopping using numpy.sum function?

example - let state be:

import numpy np state = np.random.randint(1, 10, (9,9)) 

here using {0, 1} values state, 1 means "alive". can slice around cell being investigated, e.g. [2,3]

s = state[1:3,2:5] if s[1,1]:    val = -1 else    val = 0 val += s.sum() 

if put in loop , pay attention border cases, clamping or wrapping appropriate, should describe.

if looking short elegant implementation, can done efficiently python , numpy.