QR-DQN Quantile Midpoint Bin

Problem

Map a probability level p to the index of the quantile midpoint bin it falls in.

Split [0, 1] into n_quantiles equal bins and return floor(p * n_quantiles), clamped so p <= 0 gives 0 and p >= 1 gives n_quantiles - 1. A non-positive quantile count returns -1.

Hint

The bin index tells you which quantile head should receive the sampled tau.

Test cases

case 1

in: [0.3,4]

expected: 1

case 2

in: [0.75,4]

expected: 3

case 3

in: [0.999,4]

expected: 3

case 4

in: [0,5]

expected: 0

case 5

in: [1,5]

expected: 4

Starter codePython

def qr_midpoint_bin(p, n_quantiles):
    # Your code here
    pass

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