François Perron

Overview

François Perron is a statistician in the Department of Mathematics and Statistics at the Université de Montréal. His research includes Bayesian nonparametric methods, approximation theory for distribution functions, and MCMC computation. He is not to be confused with Pierre Perron (Boston University), the econometrician known for structural break testing and unit root inference.

Key Contributions / Features

• Perron and Mengersen (2001) — "Bayesian Nonparametric Modeling Using Mixtures of Triangular Distributions", Biometrics 57(2): 518–528: proved that a variable-partition mixture of triangular CDFs (quadratic spline) achieves uniform approximation error $\|F-H\|\leq 3/2r$ over all monotone $F:[0,1]\to[0,1]$ — strictly better than beta-mixture Bernstein polynomial approximations ($O(r^{-1/2})$); developed a reversible jump MCMC algorithm (birth/death/move) for joint estimation of the partition and dimension.

Related

• Kerrie Mengersen
• Reversible Jump MCMC
• Perron-Mengersen (2001)
• Markov Chain Monte Carlo