Frank Ritter (fer2@psu.edu)

17 June 2000, revised 12 nov 2000

There are multiple ways that numbers can be tested to see if they are not random. The simplest ways are to look at the distributions and to look at the order of the numbers.


Create two distributions. For example, take one from a handbook of truely random numbers, and the other is get from a computer's random number generator.


Week 1. Choosing distributions. Generate two distributions each of 50 numbers.

One should be truely random. You should document how you created it and why you believe it is random. The other should come from a psuedo-random distribution, such as a computer program or programming language.You should note how its non-random nature might be noticed. You are essentially setting up a homework problem for another group. You will be marked based on how fair and appropriate the two distributions are. You should note what you expect the other team to do, and how they might tell the distributions apart.

Email the TA your distributions, and hand in paper copies in class.

Week 2. Take the two distributions given to you and test them. Generate an answer showing which distribution you think is the psuedo random distribution, and which is the real random distribution. You will be marked less on your correctness (this year anyhow), than on your ablilty to carry through the anlysis and present your case.