|Date||Topics & Readings||Note|
|| We discuss the "SAMSI/MUCM
Mid-Program Workshop" on April 2-3. The idea
with this workshop is to get an update of the progress of people in the
different on-going projects, so it is a very 'internal' workshop.
Also, Jim Crooks presented a problem encountered by the Terrestrial working group, namely, interpolating (emulating) non-Gaussian distributions. This is the motivating slide. An issue to be answered is whether quantile regression could be a possible solution. In this regard, Peter Riechert provided some suggestions for future readings: a summary, an introduction, and an R implementation. Please have a look before next meeting (January 22).
|1-15-2007||No meeting. Martin Lurther King's day.|
||Jim Crooks discussed the
quantile regression, and the problem he encountered with in his project.
will continue to address the issue of developing emulators for outputs
that are densities, not necessarily normal (Jim Crooks); also
Cari Kauffman will discuss a method for constructing emulators when the
dataset is too large to carry out the usual Gasp calculations, using
compactly supported covariance functions. Here
is the smart board note during the meeting. Cari's note on using
Covariance Tapering method to build efficient
emulators for computer experiments is available here.
More references are provided in the section of suggested readings.