Matlab

General

Armadillo provides a wrapper for Matlab integration in .../mex_interface/armaMex.hpp. Your function must first be compiled to a mex file by the Matlab compiler.

The way forward is :

• create your c++ wrapper e.g myFunc.cpp (the file name will also be the function name in Matlab)
• compile it using mex in Matlab
• put the mex file where Matlab can find it

Compilation

Compilation is done in the Matlab environment. Compile parameters/flags is basically the same as when you build the "normal" SigPack applications, so if you want FFTW support you need to add e.g -DHAVE_FFTW -lfftw3

Windows

mex -I<Armadillo install dir>\include ...
    -I<Armadillo install dir>\mex_interface ...
    -I<Sigpack install dir>\include ...    
    -L<Armadillo install dir>\examples\lib_win64 ...
    -lblas_win64_MT.lib -llapack_win64_MT.lib ...
    myFunc.cpp

Linux

mex -I<Armadillo install dir>/include ...
    -I<Armadillo install dir>/mex_interface ...
    -I<Sigpack install dir>/include ...  
    -larmadillo ...
    myFunc.cpp

Matlab code

% Demo 
X = 0.5+randn(10000,1); % Generate random signal
 
Y = myFunc(X);          % Run the demo using your c++ code
 
% Plot spectrum
figure;
plot(10*log10(abs(Y)),'b')
Y2 = pwelch(filter(fir1(17,0.25),1,X),512,256,'twosided','power');
hold on
plot(10*log10(abs(Y2)),'+r')
legend('SigPack','Matlab')

C++ code

#include "armaMex.hpp"
#include "sigpack.h"
 
void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
{
    // Check input parameters
    int M = mxGetM(prhs[0]);
    int N = mxGetN(prhs[0]);
    if(nrhs != 1)               mexErrMsgTxt("One input required.");
    if(nlhs != 1)               mexErrMsgTxt("One output required.");     
    if( M == 1 &&  N >= 1 )     mexErrMsgTxt("Column vector required.");
    if( mxIsComplex(prhs[0]))   mexErrMsgTxt("Real input data required.");
    
    // Convert each input argument to Armadillo
    vec x = conv_to<vec>::from(armaGetPr(prhs[0],true));
 
    //===================================================================
    // START: Your stuff
    
    int Nfft = 512;
    vec y(x.size());
    vec Pyy(Nfft);
    sp::FIR_filt<double,double,double> fir_filt;
    int K = 17;
    vec b(K);
    
    // Create a FIR filter
    b = sp::fir1(K,0.25);
    fir_filt.set_coeffs(b);
    
    // Filter the signal
    y = fir_filt.filter(x);
    
    // Calculate spectrum
    Pyy = sp::pwelch(y,Nfft,Nfft/2);
 
    // END: Your stuff
    //===================================================================
 
 
    // Convert outputs back to Matlab
    plhs[0] = armaCreateMxMatrix(Pyy.size(), 1, mxDOUBLE_CLASS, mxREAL); 
    armaSetPr(plhs[0], conv_to<mat>::from(Pyy));
    return;
}