In Eigen, I can do a row-wise or column-wise "partial reduction" to get the maximum coefficients.
For example, this program:
#include <iostream>
#include <Eigen/Dense>
int main()
{
Eigen::MatrixXf mat(2,4);
mat << 1, 2, 6, 9,
3, 1, 7, 2;
std::cout << "Column's maximum: " << std::endl
<< mat.colwise().maxCoeff() << std::endl;
}
Outputs:
Column's maximum:
3 2 7 9
Instead of creating a row vector with the maximum coefficient in each column, I would like to construct a row vector with the index of the maximum coefficient of each column.
In other words, I would like to modify the program so that the output becomes:
Column's maximum:
1, 0, 1, 0
I know I can get the index one column at a time doing something like this:
Eigen::MatrixXf::Index max_index;
mat.col(i).maxCoeff(&max_index);
but I was hoping there was a better way that could do this all in one step instead of manually looping through each column. Is this possible? (I'm using Eigen v3.2.7)
I found a post on Eigen's user forums from 2012 that suggests this is not possible, and that looping over the rows/columns is indeed the best way.
There is no shorter way yet. Regarding vectorization,
vec.maxCoeff()is vectorized (standard reduction), but not the version returning the index:vec.maxCoeff(int&). It's not impossible but I'd not expect significant gain if any.
I briefly grepped through some of the 3.2.7 code base, and it doesn't appear that anything has changed since the post.
Collected from the Internet
Please contact [email protected] to delete if infringement.
Comments