Neighbours in an array

[Mekire]

My previous reply was basically the answer.  

for y, x in inp2:
    # set maximum to the current value 
    for dy, dx in ADJACENTS:
        if 0 <= y+dy < P and 0 <= x+dx < Q:
            # set maximum to adjacent value if higher than current maximum
    # set the location in the new array to the maximum

Just replace the lines.  You still index into inp1 with the coordinates (x, y, x+dx, y+dy).

Set the initial maximum to inp1[y,x]
Then on line four set maximum to the adjacent you are checking inp1[y+dy, x+dx] if that is higher than current maximum.
On line five set inp_new[y,x] to the maximum.