%% Simulation
%% Notes
% This is a very simple set of exercises designed to provide some familiarity
% with random number generation
%% Setup
% Clear and reset the workspace and load required data
% Clean up everything
clear all
close all
clc
% Reset rng to make runs the same
rng('default')
%% Simulate standard normal random variables
% |randn| is the main function to simulate normal random variables.
x = randn(100,1);
%% Simulate random variables with mean and variance
% To get a particular mean and variance, multiple by the standard deviation
% and add the mean. Alternatively you can use |normrnd| which takes the mean and
% standard deviation as parameters.
y = .08 + 0.2 * randn(100,1);
y2 = normrnd(0.08, 0.2, 100, 1);
%% Simulate Students t random variables
% |trnd| can be used to simulate Student's t random variables.
z = trnd(8,100,1);
%% Simulate Students t random variables with mean and variance
% |trnd| generates Student's t random variables. The variance of a Student's
% t with v degrees of freedom is v/(v-2) and so it is nevessary to standardize
% them to have unit variance.
v = 8;
w = trnd(v,10000,1);
w = w/sqrt(v/(v-2));
w = 0.08 + 0.20 * w;
disp('mean')
mean(w)
disp('std')
std(w)
%% Simulate identical sets random variables
% Pesudo random numbers are very long, but ultimatley deterministic, sequences.
% Since tehey are deterministic, it is possible to reset to the same condition
% which will produce the same values. Here the generator is reset using the command
% |rng| to first get the stats and then calling with the saved state to reset
% the random number generator to the same point.
state = rng;
u1 = randn(100,1);
rng(state)
u2 = randn(100,1);
max(abs(u1-u2))
%% Simulate Students t using only randn
% Since Student's t random variables are defined as a normal divided by the
% square root of a an independent, standardized chi square and a chi square is
% just the sum of independent normal random variables squared, it is easy to construct
% these from only standard normals.
n = 10000;
z2 = randn(n,1);
v = 8;
c = sum(randn(n,v).^2,2);
z2 = z2./sqrt(c/v);
%% Verify similarity to Student T random variables
% These should be close to the 45 degree line when plotted against Student T
% 8 random variables. The extremes will be noisier than the center.
clf;
z3 = trnd(8,n, 1);
diag_line = linspace(min(min([z2,z3])),max(max([z2,z3])));
plot(sort(z2), sort(z3), 'r.', diag_line, diag_line, 'k:')
hold on
legend('Random numbers','45^o line')
title('Plot against Student T 8 RVS')