Created
June 14, 2019 11:52
-
-
Save AidanRocke/a4898097ce572bc8bc5a977fcbda6ed8 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
using Distributions | |
using Statistics | |
function monotone_approx(N::Int64,n_trials::Int64,n::Int64) | |
""" | |
inputs: | |
N: the range of U([-N,N]) | |
n_trials: the number of times we generate random vectors | |
n: the 'dimension' of the random vector | |
outputs: | |
Q: the quality of the approximation for ten values | |
monotone relation: a logical verification that the monotone relation holds | |
""" | |
X = rand(-1*N:N,n_trials,n); | |
Y = sum(X,dims=(2)); | |
## comparison with binomial: | |
Q = zeros(10); | |
D, D_hat = zeros(10), zeros(10); | |
for k =1:10 | |
## delta is approximately N/2 | |
delta = N/2 | |
## an approximation of n_hat which is 'n' in the n choose k formula | |
n_hat = Int(floor((2*N/(2*N+1))*n)); | |
## the fraction of S_n whose value is greater than 2*Delta*k: | |
D[k] = mean(Y.>2*delta*k); | |
## k in the n choose k formula: | |
D_hat[k] = Int(floor((2*k + n_hat)/2)); | |
## a CDF that should have a monotone relationship with W | |
## this may be considered an approximation | |
Z = 1-cdf(Binomial(n_hat),D_hat[k]); | |
## checking the quality of the approximation | |
Q[k] = min(D[k],D_hat[k])/max(D[k],D_hat[k]) | |
end | |
return Q, minimum((D[2:10].-D[1:9]).*(D_hat[2:10].-D_hat[1:9])) > 0 | |
end |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment