0

I've read here that creating a list of lists is much faster than creating a NumPy array. However, I also read (somewhere on Stack Overflow but I'm not sure where exactly) that, to retrieve a value from an array, [i, j] is faster than [i][j] - I could use the former in a NumPy array, but I'd have to use the latter in a list of lists. So it seems that storing the data is quicker in a nested list, but retrieving the data is quicker in a NumPy array.

I'm guessing that one method is overall quicker than the other, so which one is it, and why?

(I want to do calculations with the stored data later, so the speed at which I retrieve the data matters, but they'll all be elementwise, so an array and a list of lists would both be ok.)

nevermindoxymorons
  • 133
  • 1
  • 8
  • NumPy is much faster. Most of it is actually C code so it avoids type-checking and other things that make Python a rather slow language. – Carl Shiles Jun 15 '17 at 15:14
  • Arrays are great if you use the compiled methods that operate on the whole array (or significant chunks). If you are operating element by element, stick with the nested list. The claim that `x[i,j]` is faster than `x[i][j]` is true only if it compares the same `x` array. It's not a general statement about indexing. – hpaulj Jun 15 '17 at 15:56

0 Answers0