Eigen MatrixXf subtract a VectorXf

July 25, 2022

I have a Eigen::MatrixXf mat that has size of 113(rows) X 2009(cols); I am trying to subtract the max of each of its column from each of its column. Here is my code:

VectorXf minVal = mat.colwise().minCoeff();  // mat is MatrixXf
mat.colwise() -= minVal;

Here is the error message:

Assertion failed: dst.rows() == src.rows() && dst.cols() ==
src.cols(), file
C:\cui\Projects\eigen-3.4.0\Eigen\src\Core\AssignEvaluator.h, line 754

MEDevel.com: Open-source for Healthcare and Education

Collecting and validating open-source software for healthcare, education, enterprise, development, medical imaging, medical records, and digital pathology.
Visit Medevel

Can anyone give a pointer? Thanks a lot.

>Solution :

mat.colwise().minCoeff() returns a row vector (you can assign this to a column vector, because in most cases Eigen implicitly transposes row vectors to column vectors and vice versa).
You should store the result as a Eigen::RowVectorXf and then subtract that from each row, i.e., mat.rowwise() -= minVal;

This should work:

Eigen::RowVectorXf minVal = mat.colwise().minCoeff();

std::cout << mat << "\n\n" << minVal << "\n\n";

mat.rowwise() -= minVal;

std::cout << mat << "\n\n";