One of our most popular labs in pchem teaches the energy minimization techniques that are used in computational chemistry. The students build a mock energy function that models the dihedral rotation of 1,2-dichloroethane. Then three methods (The Monte Carlo Method, Newton minimization, and Metropolis simulated annealing) are be employed to solve for the preferred (lowest energy) dihedral angle (a).   

A performance plot is generated that shows the lowest energy and its root mean squared deviation RMSD from the known minimum structure (a = 180 degrees).  This plot clearly shows Newton’s propensity to get stuck in local minima.  It also clearly shows that the Monte Carlo method will always find the global minimum, but with greatly reduced efficiency.  And finally, the Metropolis simulated annealing technique is found to be flexible enough to accurately locate the minimum energy structure every time provided that the step size and temperatures are “tuned”. 

You can participate!  Download the Rosetta@home screen saver, and solve protein folding problems in your sleep.  (I have no official connection to the Rosetta at home folks, but their work is great! http://boinc.bakerlab.org/rosetta/)