Output a matrix with c++ and mex

问题:

I have a problem with my c++ code. I want to return a matrix of k-dimensions from my cpp program to Matlab.

The matrix I want to pass is stored in all_data, and is a matrix of size (npoints+1) x ndims.

I have been looking how to do that, and I have come up with:

    //send back points
    vector< vector <double> > indexes = mxGetPr(plhs[0]);
    for (int i=0; i < (npoints1+1); i++)
            for (int j=0; j < ndims1; j++)
                indexes[ i ][ j ] = all_data[ i ][ j ];

But it does not work, as all_data is a vector<vector<double>> variable, and matlab says:

error: conversion from 'double*' to non-scalar type 
'std::vector<std::vector<double, std::allocator<double> >, 
std::allocator<std::vector<double, 
std::allocator<double> > > >' requested

Can someone help me out? Thanks a lot!

回答1:

It looks like mxGetPr is returning you a pointer to an array of doubles and you are assigning it to a vector of vectors.

This should work:

double* indexes = mxGetPr(plhs[0]);

回答2:

mxGetPr does not return a vector<vector<double> >. It returns a double *. MATLAB arrays are stored contiguously in memory, column-major. Assuming you've created plhs[0] with the correct dimensions, then all you need to do is this:

double *indexes = mxGetPr(plhs[0]);
for (int i=0; i < (npoints1+1); i++)
    for (int j=0; j < ndims1; j++)
        indexes[i + ndims1*j] = all_data[ i ][ j ];

Note the conversion of the 2 indices to a linear offset.