How to reshape a dmatrix?

[DennisZhangOiler]

Hi,

I am a newbie to math and nalgebra, I have some code written in numpy and I want to translate into nalgebra, how to do np.reshape in nalgebra?

[fumagall]

Hey,

I'm not super familiar with nalgebra and maybe there exist a simpler method but if you use the generic interface you can write:

use nalgebra::{DMatrix, Dyn, SMatrix, Const};

fn main() {
    // for dynamical sized stuff
    let mut x = DMatrix::<f64>::zeros(10, 2);
    let mut x = x.reshape_generic(Dyn(5), Dyn(4)); // works
    let mut x = x.reshape_generic(Dyn(2), Dyn(4)); // does not work but will be checked at run time
    let (n_row, n_col) = (5, 4);
    let mut x = x.reshape_generic(Dyn(n_row), Dyn(n_col)); // works
    
    // for static sized stuff
    let mut y= SMatrix::<f64, 3, 4>::zeros();
    let mut y = y.reshape_generic(Const::<6>, Const::<2>); // works
    // let mut y = y.reshape_generic(Const::<6>, Const::<1>); // does not work and you compile can already catch it
    let (n_row, n_col) = (6, 2);
    let mut y = y.reshape_generic(Const::<n_row>, Const::<n_col>); // does not work, since variables are not constant. You could use const generics.
}

In the reshape_generic you first specify the row and then the column. The Const::<6> tells that the size is known at compile time and has size 6, while Dyn(6) would say I don't know the size at compile time and it has the size 6. Since the size does not have to be known at compile time for Dyn you can also enter variables as well in contrast to the Const.