An Intel Core i7 980 XE processor can run 100 billion floating point operations every second. But some data processing jobs require astronomically huge computing resources, which require tasks to be distributed over several machines. Often, this means using an algorithm called MapReduce, which deals with the fact that two pieces of data sent to two different machines may, in fact, depend on each other. In this course, we will explore some basics of distributed computing, and then talk about the MapReduce algorithm conceptually, before seeing a basic example and discussing some practical aspects of the algorithm and its open source implementation, Hadoop, and Amazon's MapReduce service, EMR.

Edge detection is a basic yet fundamental problem in image processing and computer vision. In this course, we will talk about some basic tools and concepts in image processing, then use this foundation to study the Canny Edge Detection algorithm and some of its optimizations. It is my goal for everyone in the class to be understand the Canny edge detector well enough to implement it on their own, although we will not do this during the class. If time permits, we will also take a brief high level look at some more modern approaches to edge detection, especially those involving machine learning.

In statistical mechanics, physicists develop equations describing the macroscopic behavior of complicated systems based on the dynamics of their microscopic constituents. But what if we want to study a system in which these microscopic dynamics are too complicated to lead to nice, useful macroscopic equations? Of course, computers will come to the rescue, but how? One possible answer is a numerical analysis technique called Monte Carlo simulation, in which a large number of simple interactions are simulated and their results aggregated and/or averaged. In this course, we will briefly introduce some of the basic concepts behind Monte Carlo simulations, and then explore a simple ideal gas simulator implemented in Python.