Acadia/SFU/UBC Dynamic Computer Experiments, 2014/15 Term 1

News

2014.11.18: Will's office hour on THURS, NOV 20 is cancelled. Office hours will be held as usual on Tuesdays at 2:00 pm and Thursdays at 3:00 pm from Nov 25 through Dec 11.

2014.11.18: Assignment 3 posted (due Dec 2)

2014.11.15: Software links cleaned up

2014.10.24: Topics for student presentations/projects updated (see Assignments)

2014.10.21: Slides for today's class posted

2014.10.11: Reading list updated

2014.09.28: Properties of the normal and multivariate normal distributions posted under Course Materials

2014.09.25: Slides for today's class posted; schedule updated

2014.08.28: An interim course outline is available under Course Materials.

Course Organization

Three courses are being held jointly by video link: Math 4013A1 / Math 5843A1 at Acadia University, STAT 890-4 at SFU, and STAT 547L at UBC. The course will be co-taught by Professor Derek Bingham at SFU and Professor Will Welch at UBC, with some classes led by Professors Hugh Chipman and Pritam Ranjan at Acadia. So this is a great opportunity for students at the three universities to interact. Each instructor is responsible for grading students registered at their own institution.

The course is also a graduate-student training component of the Statistical Modeling of the World grant held by Acadia University, SFU, and UBC, funded by CANSSI (Canadian Statistical Sciences Institute). The grant focusses on the use of computer (numerical) models and physical data for applications in earth, ocean, and atmospheric sciences. Such processes are dynamic, i.e., they evolve over space and time. The course will include many other application areas too.

Course Materials

Class Schedule (revised Oct 21)

Properties of the normal and multivariate normal distributions. Not all suggested properties are included, just the main probability results. Please send errors and typos. One result is often a consequence of another, so ideas to reorganize the properties are also welcomed.

Course Outline
Introductions
Module 1, What is a Computer Model? Arctic sea-ice model output
Examples of dynamic computer models Tidal Energy (Hugh and Pritam)
Glacier Science (Derek)
Module 2, Gaussian Process Models
Module 3, GP Estimation
Design 1 (revised)
Design 2
Ofir's Design Slides
Module 5, Visualization
Sequential design
Calibration 1
Computer Experiments: What Matters?
Propagation of Variation
Multivariate calibration (Dr Dave Higdon)
Calibration 2


Other materials will be posted here.

References

Annotated reading list (revised Oct 11)


The following textbooks are available online at the UBC library:
- Rasmussen and Williams, "Gaussian Processes for Machine Learning"
- Santner, Williams, and Notz, "The Design and Analysis of Computer Experiments"
(see References - General Methodology below for the full citations)

Assignments

Assignment 1 (due Sep 30)


Assignment 2 (due Oct 23)


Assignment 3 (due Dec 2) Data: x.txt y.txt

All presentation/project topics (updated Oct 24 with a new Project 6 in Will's list)

Examples

Ocean (for Assignment 3)
x.txt (x design data)
y.mtx (y data)

Software

How to install R on your own PC
An Introduction to R (a fairly lengthy tutorial)
R web site

R packages for GPs:
DiceKriging
GPFit
mlegp
SAVE

MUCM Toolkit (Managing Uncertainty in Computer Models)
GEM-SA (Gaussian Emulation Machine - Sensitivity Analysis)
PErK (software to support the Santner, Williams, and Notz book; specification of the statistical model, etc. is very similar to GaSP, as PeRK borrowed much of the GaSP syntax)

Useful Links

CANSSI (Canadian Statistical Sciences Institute)
Department of Statistics and Actuarial Science, SFU
Department of Statistics, UBC