Wrong answer in nc::linalg::inv

[tatianyi]

Describe the bug
The wrong result is obtained when calculating the inverse matrix.

To Reproduce
Steps to reproduce the behavior:

1. Let m be a 2*2 specific matrix: nc::NdArray<int> m = { {0, 1}, {2, 3} };
2. Let n be the inverse matrix of m: auto n = nc::linalg::inv(m);
3. Output the matrix n: std::cout << n << std::endl;
4. See error

Expected behavior
The correct answer of n should be: [ [-1.5, 0.5], [1, 0] ]

Screenshots

The cause of the problem should be here:

where t=0 (the first element in the matrix m)

Additional context
Test platform: Microsoft Visual Studio Community 2022 (x64) 17.0.5 Debug x64
defined the macro "NUMCPP_NO_USE_BOOST" at the very beginning.

This problem does not recur in NumPy.

[jpalves]

It's because pivoting is not implemented correctly, I have made my own inv and replace, take a look. https://github.com/jpalves/NumCpp-Base/blob/main/include/NumCpp/Linalg/inv.hpp

[dpilger26]

Fixed in Version 2.8.0 release. Thank you @jpalves.

[jpalves]

You are welcome dpilger26