Reputation: 1234
I'm trying to write a C++ mex function to iterate through a cell array of matrices, each matrix of which is a different size. In Matlab, I could do this using the following code:
function Z = myFunction(X, Z, B)
for i = 1:size(X, 1)
for j = 1:size(X, 2)
for k = 1:size(X, 3)
temp = X{i, j, k};
for m = 1:size(temp, 1)
Z{temp(m, 1)}(temp(m, 2)) = Z{temp(m, 1)}(temp(m, 2)) + B(i, j, k);
end
end
end
end
Here X is a 3-dimensional cell array, where each cell contains a matrix with a variable number of rows and 2 columns. These two columns allow me to index another cell array of vectors Z, where each vector is of different length. Elements of vectors in Z that are indexed are incremented by elements from a 3-dimensional matrix B.
So far, I have the following code in C++ (I have never coded in C++ before):
#include "mex.h"
void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
{
const mxArray* X = prhs[0];
const mxArray* Z = prhs[1];
const mxArray* B = prhs[2];
const int* pDims = mxGetDimensions(X);
mwSize nsubs = mxGetNumberOfDimensions(X);
for (size_t i = 0; i < pDims[0]; i++) {
for (size_t j = 0; j < pDims[1]; j++) {
for (size_t k = 0; k < pDims[2]; k++) {
int subs [] = {i, j, k};
mxArray* temp = mxGetCell(X, mxCalcSingleSubscript(X, nsubs, subs));
const int* matDims = mxGetDimensions(temp);
for (size_t m = 0; m < matDims[0]; m++) {
}
}
}
}
}
Questions:
temp
and perform the indexing as I have done in the Matlab code?Upvotes: 2
Views: 1143
Reputation: 15837
-Since all input arrays are const
you should make a copy of Z
.
-mxCalcSingleSubscript
can be used for any type of array including cell arrays. Here I renamed it to sub2ind
.
-mxGetPr
is used to access elements of an array .
Here is an implementation (hasn't been tested on real data):
#include "mex.h"
#define sub2ind mxCalcSingleSubscript
void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
{
const mxArray* X = prhs[0];
const mxArray* Z = prhs[1];
const mxArray* B = prhs[2];
mxArray * out = mxDuplicateArray (Z);
const int* pDims = mxGetDimensions(X);
mwSize nsubs = mxGetNumberOfDimensions(X);
double* B_arr = mxGetPr(B);
for (size_t i = 0; i < pDims[0]; i++) {
for (size_t j = 0; j < pDims[1]; j++) {
for (size_t k = 0; k < pDims[2]; k++) {
int subs [] = {i, j, k};
mwIndex idx = sub2ind(X, nsubs, subs);
mxArray* temp = mxGetCell(X, idx);
double* temp_arr = mxGetPr(temp);
const int* matDims = mxGetDimensions(temp);
mwSize nsubs_temp = mxGetNumberOfDimensions(temp);
for (size_t m = 0; m < matDims[0]; m++) {
int subs_out_1 [] = {m,0};
int subs_out_2 [] = {m,1};
mwIndex temp_m_1 = temp_arr[sub2ind(temp, nsubs_temp, subs_out_1)]-1;
mwIndex temp_m_2 = temp_arr[sub2ind(temp, nsubs_temp, subs_out_2)]-1;
double* Z_out = mxGetPr (mxGetCell(out,temp_m_1));
Z_out[temp_m_2] += B_arr[idx];
}
}
}
}
plhs[0] = out;
}
However both MATLAB and c implementations can be changed to use linear indexing:
function Z = myFunction(X, Z, B)
for k = 1:numel(X)
for m = 1:size(X{k}, 1)
Z{X{k}(m, 1)}(X{k}(m, 2)) = Z{X{k}(m, 1)}(X{k}(m, 2)) + B(k);
end
end
end
#include "mex.h"
void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
{
const mxArray* X = prhs[0];
const mxArray* Z = prhs[1];
const mxArray* B = prhs[2];
mxArray * out = mxDuplicateArray (Z);
mwSize n_X = mxGetNumberOfElements(X);
double* B_arr = mxGetPr(B);
for (size_t k = 0; k < n_X; k++) {
mxArray* temp = mxGetCell(X, k);
double* temp_arr = mxGetPr(temp);
const int* matDims = mxGetDimensions(temp);
size_t rows = matDims[0];
for (size_t m = 0; m < rows; m++) {
mwIndex temp_m_1 = temp_arr[m]-1;
mwIndex temp_m_2 = temp_arr[m+rows]-1;
double* Z_out = mxGetPr (mxGetCell(out,temp_m_1));
Z_out[temp_m_2] += B_arr[k];
}
}
plhs[0] = out;
}
Upvotes: 3