Question

What is an efficient way to multliply two numpy arrays together? For example, given

A = [1, 2, 3, 4]
B = [2, 3, 5, 7]

I want to calculate the dot product between A and B, which is

  A.B/|A||B| = (1*2 + 2*3 + .. 4*7)/sqrt(1^2 + 2^2... +4^2) * sqrt(.....)

How can I do this efficiently and fast?

Answer 1

If you are using numpy, numpy.dot would do the Job for you

numpy.dot(A,B)
51

---

The Fastest norm for a vector would be

n = math.sqrt(numpy.dot(A,A.conj()))

and here is the comparison with other methods

>>> t1=timeit.Timer("n = math.sqrt(numpy.dot(A,A.conj()))","from __main__ import A,math,numpy")
>>> t2=timeit.Timer("n = math.sqrt(sum(abs(A)**2))","from __main__ import A,math")
>>> t3=timeit.Timer("numpy.linalg.norm(A)","from __main__ import A,numpy")
>>> print "%.2f usec/pass" % (1000000 * t1.timeit(number=100000)/100000)
2.82 usec/pass
>>> print "%.2f usec/pass" % (1000000 * t2.timeit(number=100000)/100000)
13.16 usec/pass
>>> print "%.2f usec/pass" % (1000000 * t3.timeit(number=100000)/100000)
15.68 usec/pass
>>>