Saturday October 26, 2002
Denver Athletic Club
The fall meeting of the Colorado-Wyoming Chapter of the ASA will be held on Saturday October 26, 2002 at the Denver Athletic Club, 1325 Glenarm Place, in downtown Denver.
The format of this meeting will be somewhat different than recent meetings. The meeting schedule is as follows:
5:30 pm Social/cocktail hour (cash bar)
7:00 pm Dinner and round table discussions
7:45 pm Chapter announcements and introduction of Dr. Max Morris
8:00 pm Address by Dr. Morris
Meeting registration in advance is appreciated, and thus offered at a $15 discount. In order to foster student participation, the chapter is subsidizing the cost of student registration. Registration includes dinner (choices are listed on the registration form), non-alcoholic beverages, and dessert. Prices are as follows:
Postmarked before October 15th: $18(Students ) $35(Others)
Postmarked after October 15th or at the door: $33(Students ) $50(Others)
Click here for the registration form for the meeting.
Parking is not included in the price of the meeting, but is available at the Denver Athletic Club for $3. Carpooling is encouraged.
Abstract for Max Morris’ Keynote Address:
COMPUTER EXPERIMENTS AND STATISTICS
Computer models, programs written to simulate physical systems of interest, have become fundamental tools in scientific and engineering investigations. Because computer models are essentially representations of completely deterministic functions, the need for statistical experiments is not immediately obvious. However, with advances in theoretical knowledge and computational power, models from many disciplines (e.g., meteorology, physics, transportation, environment) have become enormously complex, often representing hundreds of major physical sub-processes with thousands of input and output variables. Such deterministic models, even though "known" to the team of programmers who create them, are so complex that even a subject-matter expert cannot accurately predict outputs from a given set of inputs. Answers to questions such as "Which inputs are most important?" and "Which conditions lead to meaningful patterns in the outputs?" may require the design, implementation, and analysis of an empirical study -- a ``computer experiment'' -- followed by an analysis of the resulting computed ``data''.
In this talk, I will present a brief overview of some statistical ideas and methods that have been used for studying computer models. Previous knowledge about computer experiments will not be assumed. Generally, earlier methods tend to treat computer models as ``black boxes'', while more recently introduced techniques incorporate more assumptions, information about the model, or data from the computer experiment. Projecting this trend, I see this topic as a prime opportunity for collaborative work among statisticians, applied mathematicians, and scientists willing to pool their expertise to create even more informative models, analysis techniques, and computer experiments.