Non-Negative Least Squares Step

Problem

Take one projected gradient step for non-negative least squares with a scalar weight.

Compute grad = sum(xi * (w * xi - yi)), set w = w - lr * grad, then project back onto w >= 0. Return the new weight.

Hint

Projection onto the non-negative orthant is just max(0, w).

case 1

in: [0,[1,2],[2,4],0.1]

expected: 1

case 2

in: [1,[1,1],[0,0],0.5]

expected: 0

case 3

in: [-0.5,[1],[2],0.1]

expected: 0

case 4

in: [2,[],[],0.1]

expected: 2

Starter codePython

def nnls_step(w, X, y, lr):
    # Your code here
    pass

4 test cases · run your code against them in the editor