What is the correct (stable, efficient) way to use matrix inversion in numpy? [duplicate]

Question

This question already has answers here:

Left inverse in numpy or scipy? (7 answers)

Closed 8 years ago.

In Matlab, using the inv() function is often discouraged due to numerical instability (see description section in http://www.mathworks.com/help/matlab/ref/inv.html). It is suggested to replace an expression like:

inv(A)*B 

(where both A and B are matrices), with:

A\B

This becomes critical when the inverted matrix A is close to singular.

Is there a nice way to write this in numpy / scipy? (would solve() work?)

Answer 1

As mentioned in the comments, you need to use the left inverse.

This is described in this question.

To summarize (imitatio, aemulatio):

• Use linalg.lstsq(A,y) in general.
• You can use linalg.solve(A,y) if you know A meets the right conditions.