Week of February 23 - 27, 1998

Monday, February 23

11:30 MATHEMATICS EDUCATION SEMINAR  - Room 447 MSC
      David Meel, Mathematics and Statistics, BGSU. 
      "Information-Processing as a mathematical learning theory"

 2:30 ALGEBRA SEMINAR  - Room 447 MSC
      Curt Bennett, Mathematics and Statistics, BGSU. 
      "Buildings and BN-pairs"

 2:30 ANALYSIS SEMINAR  - Room 459 MSC
      Kit Chan, Mathematics and Statistics, BGSU. 
      "Hypercyclicity"

 3:30 STATISTICS SEMINAR SERIES  - Room 459 MSC
      Jiahua Chen, University of Waterloo, visiting BGSU this semester
      "Empirical Likelihood Methods"

Tuesday, February 24

 2:30 SCIENTIFIC COMPUTATION SEMINAR  - Room 459 MSC
      Gordon Wade, Mathematics and Statistics, BGSU.
      "Traveling Waves in Reaction-Diffusion Models"

 2:30 MAPLE WORKSHOP  - Scientific Computing Lab, MSC
      John Gresser, Mathematics and Statistics, BGSU. 

Wednesday, February 25

11:30 MATHEMATICS EDUCATION SEMINAR  - Room 447 MSC
      David Meel, Mathematics and Statistics, BGSU. 
      "Information-Processing as a mathematical learning theory"
      
Thursday, February 26

 1:00 STATISTICAL COMPUTING SEMINAR  - Room 459 MSC
      Arthur Yeh, Applied Statistics and Operations Research, BGSU
      "Introduction to the Bootstrap"

Friday, February 27
      
 3:30 Coffee
 3:45 COLLOQUIUM  - Room 459 MSC
      Jane Harvill, Applied Statistics and Operations Research, BGSU
      "Testing time series linearity via goodness of fit methods"
      Abstract: Arguably, one of the most crucial aspects of
        statistically analyzing a dataset is proper model
        identification.  This is especially true in time series
        analysis where the statistical model selected must describe
        the deterministic relationship between the past, present, and
        future, and must also describe the randomness inherent in the
        data.  The field of linear time series is well-developed with
        a rich history in application and theory.  Recently, great
        strides in non-linear time series analysis have been made.
        With these advancements, it becomes desirable to develop
        reliable tests for the linearity of a time series.  Strengths
        and weaknesses of existing tests are discussed, and a new
        method for testing time series linearity which makes use of
        the distributional properties of the normalized bispectrum
        will be introduced.  Simulation studies on a general
        application of goodness of fit tests compared to existing
        methods will be presented.  In general, these studies
        indicated the proposed procedure will be more powerful than
        existing techniques.