Laboratory for Research in Statistics and Probability

LRSP Probability and Statistics Seminar
  FALL 2006

This seminar runs under the umbrella of the Laboratory for Research in Statistics and Probability, alternating with a similar seminar at the University of Ottawa.

Coordinator: Antal A. Járai

Note: The seminar usually meets every second Friday, alternating with the seminar at the University of Ottawa. Occasionally, additional talks are scheduled for other days.

Past seminars: Fall 2004, Winter 2005


  1.   Date and time:  Friday, October 13, 1:30pm
      Location:  Room 4351 Herzberg Building

      Speaker:  Mohamedou Ould Haye, Carleton University
      Title:  Linear regression with long memory errors
      Abstract:  We study several nonrobust and robust estimators such as the LSE, BLUE, M, S-estimators for parameters in a multiple linear regression when errors are strongly dependent. We focus on the variances and the asymptotic behaviors of these estimators. We will also discuss how the strong dependence affects preliminary test estimators types.


  2.   Date and time:  Friday, October 27, 1:30 pm
      Location:  Room 4351 Herzberg Building

      Speaker:  Antal A. Járai (Carleton University)
      Title:  Random walk on the incipient infinite cluster for oriented percolation in high dimensions
      Abstract:  We consider simple random walk on the incipient infinite cluster for the spread-out model of oriented percolation in d+1 dimensions. For d > 6, we obtain bounds on exit times, transition probabilities, and the range of the random walk, which establish that the spectral dimension of the incipient infinite cluster is 4/3, and thereby prove a version of the Alexander-Orbach conjecture in this setting. The proof divides into two parts. One part establishes general estimates for simple random walk on an arbitrary infinite random graph, given suitable bounds on volume and effective resistance for the random graph. A second part then provides these bounds on volume and effective resistance for the incipient infinite cluster in dimensions d > 6, by extending results about critical oriented percolation obtained previously via the lace expansion.


  3.   Date and time:  Friday, November 3, 11:30 am
      Location:  Room 4351 Herzberg Building

      Speaker:  Yuliya Martsynyuk (Carleton University and University of Ottawa)
      Title:  On the domain of attraction of the normal law and its use in regression models
      Abstract:  We discuss some analytic and stochastic characterizations for the domain of attraction of the normal law (DAN). Examples of regression models where DAN has been used are given. We introduce DAN for regression with errors in variables (error-in-variables models) the first time around. The rationale behind using DAN, and consequent benefits for asymptotic theory in such models are illustrated.

    Note: as usual, there is a seminar at the University of Ottawa this week, starting at 1:30 pm.


  4.   Date and time:  Friday, November 10, 1:30 pm
      Location:  Room 4351 Herzberg Building

      Speaker:  Qiying Wang (University of Sydney)
      Title:  Long-range dependent time series specification
      Abstract:  Model specification of short-range dependent stationary time series has become a very active research field in both econometrics and statistics since about two decades ago. In the meantime, estimation of long-range dependent stationary time series models has also been quite active. To the best of our knowledge, however, model specification of stationary time series with long-range dependence (LRD) has not been discussed in the literature. This is probably due to unavailability of certain central limit theorems for weighted quadratic forms of stationary time series with LRD. In this paper we try to tackle such difficult issues by establishing a nonparametric model specification test for parametric time series with LRD. In order to establish asymptotic distributions of the proposed test statistic, we develop new central limit theorems for certain weighted quadratic forms of stationary time series with LRD. In order to implement the proposed test in practice, we develop a computer-intensive parametric bootstrap simulation procedure for finding simulated critical values. As a result, our finite-sample studies show that both the proposed theory and the simulation procedure work well and that the proposed test has little size distortion and reasonable power.


  5.   Date and time:  Friday, November 24, 1:30 pm
      Location:  Room 4351 Herzberg Building

      Speaker:  Emad-Eldin A. A. Aly (Kuwait University)
      Title:  On some measures of income inequality
      Abstract:  We consider the problem of developing appropriate measures of income inequality. The most well-known of these measures are the Lorenz curve, the Gini index and the Bonferroni concentration index. In this talk we present some important aspects of certain measures of income inequality and explain their relationship with some stochastic orders and record values.


  6.   Note the change of date and time!
      Date and time:  Tuesday, December 5, 12:30 pm
      Location:  Room 4351 Herzberg Building

      Speaker:  Peter Hooper (University of Alberta)
      Title:  Period analysis of variable stars: temporal dependence and local optima
      Abstract:  A number of methods have been proposed to estimate the period of a variable star; e.g., a recent approach uses smoothing spline regression to fit tentative periodic functions (light curves) and selects the period minimizing a robust goodness-of-fit criterion. These methods assume that measurement errors vary independently over time. Empirical evidence, however, indicates substantial temporal dependence, possibly related to changes in observing conditions. Dependence complicates the period analysis in several respects: selection of a "best" period among several local optima, estimation of the light curve, and evaluation of uncertainty about period and light curve estimates. In this talk I will describe some new methods designed to accommodate dependent errors. The new methods employ adaptive logistic basis (ALB) regression models. An analysis of several data sets shows that the proposed approach can produce substantially different and arguably better results compared with other methods.