Can I get the matrix determinant using Numpy?

Question

I read in the manual of Numpy that there is function det(M) that can calculate the determinant. However, I can't find the det() method in Numpy.

By the way, I use Python 2.5. There should be no compatibility problems with Numpy.

Answer 1

You can use numpy.linalg.det to compute the determinant of an array:

In [1]: import numpy

In [2]: M = [[1, 2], [3, 4]]

In [3]: print numpy.linalg.det(M)
Out[3]: -2.0000000000000004

Answer 2

For large arrays underflow/overflow may occur when using numpy.linalg.det, or you may get inf or -inf as an answer.

In many of these cases you can use numpy.linalg.slogdet (see documentation):

sign, logdet = np.linalg.slogdet(M)

where sign is the sign and logdet the logarithm of the determinant. You can calculate the determinant simply by:

det = np.exp(logdet)

For sparse matrices (2-D arrays), I highly recommend another approach based on LU decomposition.