Memoize a NumPy vectorized function

Question

I have a function is_prime(n) which returns True if n is prime and False otherwise. In NumPy I am looping, checking if an array contains primes, and the start of the array will be identical through every iteration, so I want to memoize the is_prime(n) function to avoid a lot of unnecessary calculations.

Since I have an array, I want to vectorize is_prime(n) so I can apply it on arrays element by element, NumPy style. I do this with one line from the NumPy tutorial (shown later)

I also use a memoization template I found on the net:

def memoize(function):
    cache = {}
    def decorated_function(*args):
        if args in cache:
            return cache[args]
        else:
            val = function(*args)
            cache[args] = val
            return val
    return decorated_function

Then:

is_prime = memoize(is_prime)

BUT, is V_prime now correctly memoized if i now vectorize the memoized is_prime function?:

V_prime = np.vectorize(is_prime)

Thank you

Samy Vilar · Accepted Answer · 2012-06-14 20:25:57Z

5

Well lets test it.

import numpy as np

def test(input):
    return input

def memoize(function):
    cache = {}
    def decorated_function(*args):
        if args in cache:
            print 'cached'
            return cache[args]
        else:
            print 'not cached'
            val = function(*args)
            cache[args] = val
            return val
    return decorated_function

test = memoize(test)
print test(9)
print test(9)
test = np.vectorize(test)
print test(9)
print test(10)
print test(10)

I get this on my machine.

not cached
9
cached
9
cached
cached
9
not cached
10
cached
10

so yes, it is memoize, on my machine using numpy 1.6.1

answered Jun 14, 2012 at 20:25

Samy Vilar

11.2k2 gold badges42 silver badges34 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

luffe Over a year ago

Oh well, that was clever! Thank you!

Samy Vilar Over a year ago

no problem, memoization is quite cool, though be careful since it can consume quite a bit of memory depending on what types of objects you are working with... people tend to use dedicated objects instead of dictionaries to control/monitor memory and destroy/remove objects when memory is low or they haven't accessed them in a while.

Collectives™ on Stack Overflow

Memoize a NumPy vectorized function

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related