kd383 / networkdos Goto Github PK

This is a great tool! I'm working on a project where we need to estimate the density of eigenvalues of a large Laplacian matrix. Reading your paper, this looks like a useful way to speed up our computations. If it works, eventually we'd like to include this in a package my lab is developing for biomedical data analysis over at https://github.com/krishnaswamylab/MELD.

To test this though, it would be really helpful if you could please provide example usage for the Python implementation. I am currently going through the tests files, but example usage for a simple case would be helpful.

Also, before we can include this in our package, we really will need a Python 3 implementation. We can cross that bridge later, but with Python 2 losing support this year, we need to stick to Python 3.

Thanks for your great work!
-Dan

The paper has mentioned that this approximation method can be easily extended to compute the Laplacian matrix's eigenvalues. Could you please give me some intuition on how to modify the original algorithm, i.e. how to scale the Laplacian matrix spectrum(to let its spectrum lies in [-1, 1])