Gradient of L2-norm using dlgradient

Is it possible to compute the gradient of n = norm(X,"fro") using dlgradient() ?

function y = f(x,theta)

y = norm(x-theta, "fro");

end

function [y, dy] = fun_and_deriv(x,theta)

y = f(x,theta);

dy = dlgradient(y,theta);

end

Answer 1

[Matt J]

Why do that, when the gradient is analytically known?

function [y, dy] = fun_and_deriv(x,theta)

     delta=theta-x;

    y = norm(delta, "fro");
    
    dy = delta./y;
end