Stat 376: Advanced Modeling and Scientific Computing

Tues, Thurs 11:40-12:55pm 025 Old Chem Bldg

Instructor: Scott Schmidler
Email: schmidler@stat.duke.edu
Office: 223D Old Chem Bldg
Phone: 684-8064

Advanced topics and research areas in statistical computing with particular emphasis on Monte Carlo methods.

Topics will include:

Advanced Markov Chain Monte Carlo: Theory and algorithms, including tempering, auxiliary variable techniques, diffusions, evolutionary MC, multicanonical/histogram MC methods, and others
Sequential Monte Carlo:Particle filter and generalizations
Numerical Optimization: Stochastic optimization, quadratic and semi-definite programming, taboo search, other recent interesting algorithms
Variational methods
Smoothing and meshing: Principal curves and surfaces, manifold learning,

Course requirements: Grading will be based on in class discussion and presentation of readings and a final project.

References (background material):

Monte Carlo Statistical Methods, C. Robert & G. Casella (1999, Springer Verlag)
Stochastic Simulation, B. Ripley (1987, Wiley)
Markov Chain Monte Carlo in Practice, Gilks et al (1996, Chapman & Hall)
Bayesian Data Analysis, Gelman et al (1995, Chapman & Hall)
Monte Carlo Strategies in Scientific Computing J. Liu

Enrolling: Due to the low outside workload and interactive nature of the course, I encourage everyone to enroll for credit. Please see me individually if you feel you need to enroll as an auditor. All students who sit in the class must be enrolled either for credit or as an auditor.

© 2007 Scott C. Schmidler