When Computers Flip Coins

Ian Barland
Department of Computer Science, Rice University

Abstract: If a computer is allowed access to random coin-flips, can it solve problems more efficiently than before? While randomization at first seems antithetical to the notion of a program (the epitome of regularity), we will consider some problems where randomness *does* seem to help. The question then becomes, can we quantify how much randomness gains us, and how much it costs? We'll look at probabilistic algorithms for problems including sorting lists, testing primality, and even just remembering how to rotate a mattress. (No previous background in computer science is presumed.)