Winograd Transform Lite

Problem

Apply the F(2,3) Winograd input transform to a 4-element tile: B^T d with rows computed as d0 - d2, d1 + d2, -d1 + d2, d1 - d3.

The signature is winograd_transform_1d(d). Return the transformed 4-vector.

Hint

Winograd turns 3x3 convolutions into cheap element-wise multiplies in the transform domain.

case 1

in: [[1,2,3,4]]

expected: [-2,5,1,-2]

case 2

in: [[0,0,0,0]]

expected: [0,0,0,0]

case 3

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

expected: [2,1,-1,0]

case 4

in: [[1,1,1,1]]

expected: [0,2,0,0]

Starter codePython

def winograd_transform_1d(d):
    # Your code here
    pass

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