Perry de Valpine

Associate Professor, Environmental Science, Policy, and Management
BIDS Senior Fellow

Real name: 
Perry de Valpine

Perry de Valpine is a mathematical and statistical ecologist. He leads the development of a computational statistics programming system called NIMBLE. NIMBLE facilitates the implementation and sharing of algorithms, such as Markov chain Monte Carlo and sequential Monte Carlo, that can operate on general model structures. It combines a new implementation of the BUGS language for declaring hierarchical statistical models with a language for programming algorithms that use models generically. The system is embedded within R and includes a compiler that generates model-algorithm-specific C++. NIMBLE provides a new degree of programmability for these kinds of algorithms and has many potential applications, including models in many scientific fields and a variety of statistical methods for model estimation, assessment, and validation.

Perry's research interests span ecology and evolution as well as statistics. In ecology, his work has primarily addressed questions about ecological population dynamics with a focus on how to estimate realistic models from noisy data to estimate abundance, density feedbacks, species interactions, and climate effects. Example systems have included data on trees, butterflies, spider mites, kangaroos, fishers, bats, birds, and fish. Other work has included the development of theory on life history evolution and the role of individual heterogeneity in stage-structure population dynamics. His work in statistics is motivated by the numerous challenging problems that arise from complex ecological data and corresponding gaps in methods and software.

Projects

Nimble logo

NIMBLE: Numerical Inference for Hierarchical Models Using Bayesian and Likelihood Estimation