Group Soft Threshold

Problem

Apply the group soft-thresholding (group lasso proximal) operator to a coefficient vector.

Compute norm = sqrt(sum(x ** 2)). If norm <= lam, return all zeros; otherwise return every component scaled by 1 - lam / norm.

Hint

Shrink the whole group by a single factor, or zero it out entirely.

case 1

in: [[3,4],2]

expected: [1.7999999999999998,2.4]

case 2

in: [[0.3,0.4],0.5]

expected: [0,0]

case 3

in: [[0,0],0.1]

expected: [0,0]

case 4

in: [[5],1]

expected: [4]

Starter codePython

def group_soft_threshold(v, lam):
    # Your code here
    pass

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