double truncated data sample

Hi,

I generate data x using the code :

clc;
clear;
eta=2;
beta=4;
theta=1;
T=1;
t1=0.1;
t2=0.9;
ni=[10 50 75 100 50];
for i=1:length(ni)
      n=ni(i);  
for t=1:T
x=generate_sample(n,eta,beta,theta);
f1=pdf_WP(sort(x),eta,beta,theta);
F=cdf_WP(sort(x),eta,beta,theta);
R_Real=1-F;
end
end

the function is:

function [x]=generate_sample(n,eta,beta,theta)
for i=1:n
      u=rand;
    x(i)=theta.*(1./eta.*(log(1/(1-u)))).^(1./beta);
end

How can I make doule truncate from t1 to t2 from x

with my best regurds

[Walter Roberson]

https://www.mathworks.com/help/stats/prob.normaldistribution.truncate.html maybe?

Answer 1

[Jeff Miller]

This is pretty ugly, but I think will do what you asked for:

function [x]=generate_sample(n,eta,beta,theta,t1,t2)
    
    % find the bounds on u corresponding to the desired
    % bounds on t1 and t2
    xFn = @(u) theta.*(1./eta.*(log(1/(1-u)))).^(1./beta);
    u1 = fzero(@(x) xFn(x)-t1,[eps 1-eps]);
    u2 = fzero(@(x) xFn(x)-t2,[eps 1-eps]);
    
    for i=1:n
        % Generate u's only within the bounds
        u=u1 + (u2-u1)*rand;
        x(i)=xFn(u);
    end
    
end

[Bashar AlHalaq]

thank you for response on my demand ,

but i neeed truncate the sample data range , the code above dont cutting the sample data range

with my best reguard.

[Jeff Miller]

Sorry, I guess I don't understand what you want. Maybe a numerical example would help?