Prioritized Exponent Alpha

Problem

Recover the prioritization exponent from a desired max/min sampling-probability ratio.

alpha = ln(prob_ratio) / ln(p_max / p_min). Return 0.0 when the inputs are degenerate: p_min <= 0, p_max <= p_min, or prob_ratio <= 1.

Hint

Alpha of 0 samples uniformly; alpha of 1 samples proportional to priority.

case 1

in: [1,10,2]

expected: 0.30102999566398114

case 2

in: [1,4,2]

expected: 0.5

case 3

in: [2,2,3]

expected: 0

case 4

in: [1,10,1]

expected: 0

Starter codePython

import math
def prioritized_exponent(p_min, p_max, prob_ratio):
    # Your code here
    pass

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