Hi!
I'd like to do a rowwise max reduction as in the matlab command

Code: Select all

[val ind] = max(mat, [], 2)

I tried changing the eigen example on the website:

Code: Select all

#include <iostream>
#include <Eigen/Dense>

using namespace std;
using namespace Eigen;
int main()
{
    MatrixXf mat(2,4);
    mat << 1, 2, 6, 9,
           3, 1, 7, 2;
    MatrixXf::Index   maxIndex;
    VectorXf maxVal = mat.rowwise().maxCoeff(&maxIndex);
    std::cout << "Maxima at positions " << endl;
    std::cout << maxIndex << std::endl;
    std::cout << "maxVal " << maxVal << endl;
}

where my change was to alter from

Code: Select all

float maxNorm = mat.rowwise().sum().maxCoeff(&maxIndex);

to

Code: Select all

VectorXf maxVal = mat.rowwise().maxCoeff(&maxIndex);

this gives an error. is there any way to do this operation? i.e. take the max over each row and return maximal value and it's index?
thanks!

Currently that's not possible. You can only do:

int i;
float x = mat.row(3).maxCoeff(&i);

or

VectorXf mat.rowwise().maxCoeff();

but not both. Feel free to add a feature request there: http://eigen.tuxfamily.org/bz and even try to propose a patch!! The relevant file is Eigen/src/Core/VectorwiseOp.h

thanks! i'll check it out.

ggael wrote:Feel free to add a feature request there: http://eigen.tuxfamily.org/bz

There already is request for this feature there: http://eigen.tuxfamily.org/bz/show_bug.cgi?id=401

A patch would still be very welcome!