Speed up the wPLI calculation with C++

[yacineMahdid]

Right now MATLAB capabilities are not sufficient for having a fast wPLI, it takes too long. Implementing wPLI from C++ should give us at least 10X improvement.

[yacineMahdid]

I found this library Armadillo which can be used to do the linear algebra stuff we need to do.

[yacineMahdid]

• will need to cite the package if we are using it for research purposes

[yacineMahdid]

Had some issues with hdf5 will need to retry the installation from scratch: https://tutorialforlinux.com/2018/12/05/step-by-step-armadillo-ubuntu-18-04-installation-guide/

[yacineMahdid]

I was able to get it to work but it was a pain, it was a problem with conda and hdf5 I had to copy the file over to my lib folder

[yacineMahdid]

It takes some time to get use to the library, but I am already able to pass around the matrix from MATLAB to Armadillo

[yacineMahdid]

Here is one way of writting the matrix in a binary format:

m=10;
A = randn(m,m);
name = 'eeg.bin'

[F,err] = fopen(name,'w');
if F<0,error(err);end

fwrite(F,A,'double');
fclose(F);

[yacineMahdid]

I just need to reshape the matrix once I get it in armadillo

[yacineMahdid]

We can go even simpler by using numpy and scipy since these are wrapper around C++ implementation