Grant: Discovery Grants Program - Individual

I am planning to develop efficient semiparametric approaches to univariate and multivariate regression and density estimation and to study related convergence issues. A recently proposed differentiated logdensity approximant, which turns out to possess numerous desirable properties, requires further investigation. I shall extend my results on the distribution of the distance between random points located in rectangles, polygons and parallelepipeds to random points located in polygonal pyramids, and consider certain applications to transportation logistics, such as those associated with intercity traffic and the positioning of airtankers. Further developments on generalized distributions should lead to more efficient techniques in reliability analysis and quality control. The Bayesian predictive approach will be used to make inferences in connection with problems arising in the biomedical fields, and in particular, in statistical genetics. Extensions of the published distributional results in Time Series to general ARMA models are also being investigated. I intend to explore an alternative structural representation of elliptically contoured distributions; promising applications include the modelling of time series with thick-tailed distributed innovations as well as the development of efficient alternative techniques for simulating functions involving elliptically contoured vectors, among others. I shall develop efficient algorithms for determining the distribution function of indefinite quadratic forms in singular elliptically contoured or complex random vectors. I shall make the code associated with several of my results available on Wolfram's website. I am also preparing monographs on density estimation and Hermitian quadratic forms. Finally, I am aiming to complete a textbook on multivariate statistical analysis with the addition of chapters on classification, multivariate analysis of variance and generalized linear models.