OMD Regret Term

Problem

Compute the linearized regret accumulated by an online mirror descent player.

Given per-round gradient vectors grads, the played weights weights, and a fixed comparator u, return the sum over rounds t and coordinates i of grads[t][i] * (weights[t][i] - u[i]).

Hint

This is the linearized regret, where each round charges the gradient at the played point.

Test cases

case 1

in: [[[1,0],[0,1]],[[0.5,0.5],[0.4,0.6]],[0.5,0.5]]

expected: 0.09999999999999998

case 2

in: [[[1,2]],[[0.2,0.8]],[0.5,0.5]]

expected: 0.3000000000000001

case 3

in: [[[0,0],[0,0]],[[0.1,0.9],[0.3,0.7]],[0.5,0.5]]

expected: 0

Starter codePython

def omd_regret_term(grads, weights, u):
    # Your code here
    pass

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