Travis: Segment 8

From Computational Statistics (CSE383M and CS395T)
Revision as of 23:18, 4 February 2013 by Tsanders (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

To Calculate

1. In Segment 6 (slide 8) we used the improper prior <math>1/r</math>. Show that this is just a limiting case of a (completely proper) Lognormal prior.

2. Prove that <math>{\rm Gamma}(\alpha,\beta)</math> has a single mode at <math>(\alpha-1)/\beta</math> when <math>\alpha \ge 1</math>.

<math> p(x) = \frac{\beta^\alpha}{\Gamma (\alpha)}x^{\alpha-1}e^{-\beta x} </math>

<math> \frac{d (p(x))}{dx} = 0 = \frac{\beta^\alpha}{\Gamma (\alpha)}(\alpha -1)x^{\alpha-2}e^{-\beta x} - \frac{\beta^\alpha}{\Gamma (\alpha)}x^{\alpha-1} (-\beta) e^{-\beta x} = \frac{\beta^\alpha}{\Gamma (\alpha)}x^{\alpha-2}e^{-\beta x} \bigg[ (\alpha -1) - \beta x \bigg] = 0</math>

<math> \therefore x = \frac{\alpha -1}{\beta} </math> for an optimal point with <math> \alpha \ge 1 </math>.

3. Show that the limiting case of the Student distribution as <math>\nu\rightarrow\infty</math> is the Normal distribution.

To Think About

1. Suppose you have an algorithm that can compute a CDF, <math>P(x)</math>. How would you design an algorithm to compute its inverse (see slide 9) <math>x(P)</math>?

2. The lifetime t of a radioactive nucleus (say Uranium 238) is distributed as the Exponential distribution. Do you know why? (Hint: What is the distribution of an Exponential<math>(\beta)</math> random variable conditioned on its being greater than some given value?)

Class Activity

Here's the code group 1 did for class, but we can't figure out what's wrong with our code. Help please :-) On the first iteration, the performance index is 2025.86. but then it converges to 918 which is way off.

clear all
close all
X0 = [2,1.2,5]; % mu, sigma, nu 
ops = optimset('Display','iter');
[x,fval,exitflag,output] = fminsearch(@opt, X0,ops)
function [prob] = opt(X)
m = load('stacks.txt');
newm = (m-(X(1)))./X(2);
T = tpdf(newm,X(3)); 
prob = sum(log(T))*-1;