Intel MKL bug with Gaussian fitting in Python?

I am doing a Monte Carlo simulation in Python, in which I get a set of intensities in certain 2D coordinates, and then apply 2D Gaussian to them. I am using the scipy.optimize.leastsq function, and all this works well, except for the following error:
Intel MKL error: parameter 6 was invalid when entering DGELSD.
The problem occurs several times in the simulation. I looked around and realized that this was due to a bug in the Intel MKL library. I cannot find a solution to the problem, and so I looked at an alternative fitting function that I could use. If someone knows how to get rid of a problem that will also be good.

+5
source share
1 answer

You can try the Intel python distribution. It includes a prebuilt scipy optimized using MKL.

-1
source

Source: https://habr.com/ru/post/1209806/


All Articles