matrix multiplication is strange in numpy(m*m equals m.dot(m)!!!)

Question

I find a strange problem in numpy: if m is a matrix, the results of m1*m2 is always the same as m1.dot(m2)!!! So how can I multipy two matrixes by elements?(such as m1.*m2 in matlab)

Answer 1

If you multiply matrices (of type numpy.matrix), NumPy assumes you want matrix multiplication, which doesn't really seem that strange to me. To multiply element-wise, either use arrays (numpy.array) instead of matrices, or use numpy.multiply().

Answer 2

This is by-design. http://wiki.scipy.org/NumPy_for_Matlab_Users

For matrix, '*' means matrix multiplication, and the multiply() function is used for element-wise multiplication.

e.g.

>>> import numpy
>>> numpy.multiply([[1, 2], [3, 4]], [[5, 6], [7, 8]])
array([[ 5, 12],
       [21, 32]])