CS 525, Simulation

Quiz 5, 6 April 2005

  1. A colleague of yours doesn't understand why simulations should be run with repetition or batch means. Why not just run the simulation for a really long time and be done with it? What explanation would you give your colleague?

    Confidence intervals need a range of sample means to be able to develop reasonable estimators of the oringal population's mean and variance. Also, a single run is more sensative to dependencies among the samples gathered; batch means and repetition produce sample runs with greater independence than does a single run.

  2. The Central Limit Theorem (CLT) involves three populations. Describe the importance of these three populations when using the CLT to bound mean estimates to a particular confidence. Extensive (or even any) mathematics is not required; just describe each population and its importance in a sentence or two.

    The three populations are:

    1. The orginal population P, which is from an arbitrary distribution but has a known (theoretically, at least) mean and variance.

    2. The population of sample means, with the samples taken from P. This population has a normal-like distribution, with the likeness growing greater as the size and number of sample means grows.

    3. The population of deviations of the sample means from the P population mean. This population has a standard normal distribution.

This page last modified on 12 April 2005.