|Date||Topics & Readings||Notes|
|10-02-2006||We devote this meeting to the description of a couple of
computer models, followed by a discussion of what we could do, which
problems we could address, tentative approaches, etc. For useful
discussions, you should be somewhat familiar with the basic issues in
model calibration, prediction, emulators, etc. A short and basic
reference list appears in the readings for date 10-9-06.
a. Bruce Pitman will talk about granular avalanche modeling. A first version of the slides can be downloaded here; he also has provided a brief overview of the main issues they would like to address.
b. Pierre Gremaud will present models of Cerebral Blood Flows. See slide here.
|10-09-2006||This meeting will be devoted to discussion
of the main issues of the approaches we could take. Basic reading is
giving bellow. Please get familiar with them (tutorials will not be
given) and prepare lists of questions (basic and/or complex) to be
discussed in the meeting. (More extended list of readings will be
a. Emulators. A basic reading is the paper by Sacks et al. (1989), with the accompanying discussions here and here.
b. `Traditional’ Bayesian approach. Basic readings:
i. Kennedy and O’Hagan (2001), along with supplementary material that had to be cut off from the paper, which can be downloaded from here and here.
ii. A very detailed, but yet basic, technical report Bayarri et al. (2002).
c. ‘Linear Bayes’ approach. Michael Goldstein will be presenting the basics of the Linear Bayes approach, via Webex. Download the slide here. Do not be shy and have questions prepared for him!
i. Craig et al. (2001) describes a general Bayesian approach for using the simulator to forecast system outcome. Download here.
ii. Goldstein (1998) presents a basic description of the linear Bayes approach, and Craig et al. (1997) discusses its implementation to CompMode calibration. Download the files here and here.
iii. A shorter version Craig et al. (1996) can be downloaded here.
|10-16-2006||We keep exploring different simulators and associated
problems. In this session we will be presented with new models, some of
them somewhat simple (useful for methodology discussions) followed by
discussion of the possibilities for the interests of the Methodology
a. Peter Riehert will be presenting hydrology models. Slide is here.
b. Ken Reckhow will be presenting water quality models. Slide is here.
|References to the treed GP model:
a. Link to the main reference for treed Gaussian Processes.
b. Link to the main reference for adaptive sampling .
c. Link to the R package tgp.
this point, we will be
confronted with new and very challenging models
a. Bernt Mueller will introduce relativistic heavy ion collider model, needed for studying what the universe may have looked like in the first few moments after its creation.
b. Jim Clark will present a challenging (for emulator purposes) dynamical terrestrial model, with complex interactions between statistical (hierarchical) spatial models, and several simulators. This is a complex dynamical model.
WebEx and Teleconferencing
|10-27-2006||Joint Engineering and Methodology Subprograms Workshop||
WebEx and Teleconferencing
|Bayesian Random Matrix Workshop.|