The software on this page has been developed by or significantly contributed by the researchers in the Environmental and Ecological research group. The license information, rights of use and additional information can be found within the program download.
The GPstuff toolbox is a versatile collection of Gaussian process models and computational tools required for inference with these models. The tools include, among others, various inference methods, sparse approximations and model assessment methods. The toolbox is the most versatile toolbox for GP modeling.
The GPstuff toolbox works (at least) with Matlab versions r2009b (7.9) or newer and most of the functionalities work also with Octave (3.6.4 or newer, see release notes for details). GPstuff can also be called from R with RcppOctave package. Most of the code is written in m-files but some of the most computationally critical parts have been coded in C.
The newest version of GPstuff is available at GitHub. We encourage to use the developer version since it includes the latest updates.
Jarno Vanhatalo, Jaakko Riihimäki, Jouni Hartikainen, Pasi Jylänki, Ville Tolvanen, Aki Vehtari (2013). GPstuff: Bayesian Modeling with Gaussian Processes. Journal of Machine Learning Research, 14:1175-1179. [Link]
GPQTLmapping is a code package to do Gaussian process (GP) modeling and Bayesian variable selection for mapping function-valued quantitative traits with incomplete phenotype data. It uses Gaussian processes (GPs) to model the continuously varying coefficients which describe how the effects of molecular markers on the quantitative trait are changing over time.
The code is available through GitHub: https://github.com/jpvanhat/GPQTLmapping
Jarno Vanhatalo, Zitong Li and Mikko Sillanpää (2019). A Gaussian process model and Bayesian variable selection for mapping function-valued quantitative traits with incomplete phenotype data. Bioinformatics, 35(19):3684-3692. [Link]
Spatiotemporal Clustering using Gaussian Processes Embedded in a Mixture Model
SpatClustMixtures is a Matlab code package to do Gaussian process (GP) spatiotemporal smoothing for cluster components in mixture modeling of multidimensional data. Many applications of clustering, including the majority of tasks in ecology, use data that are inherently spatial and often also temporal. However, spatiotemporal dependence is typically ignored when clustering multivariate data. SpatClustMixtures implements a finite mixture model for spatial and spatiotemporal clustering that incorporates spatial and spatiotemporal autocorrelation by including appropriate GPs into a model for the mixing proportions. We also allow for flexible and semi-parametric dependence on environmental covariates, once again using Gaussian processes. The package employs Bayesian inference through three tiers of approximate methods: a Laplace approximation that allows efficient analysis of large data sets, and both partial and full Markov chain Monte Carlo approaches that improve accuracy at the cost of increased computational time.
The code is available through GitHub: https://github.com/jpvanhat/SpatClustMixtures
Jarno Vanhatalo, Scott D. Foster and Geoffrey R. Hosack (2021). Spatiotemporal Clustering using Gaussian Processes Embedded in a Mixture Model. Environmetrics,32:e2681. [Link]
An R software to quantify ecological memory functions for a subset of covariates within a linear or generalized linear model using penalized splines. Ecological memory functions indicate the length of persistent responses and the relative importance of past environmental conditions of current ecosystem function. Memory functions are used to generate weighted covariate values reflecting the cumulative effect of environmental conditions over time.
The software is available at https://github.com/msitter/EcoMem
Malcolm S. Itter, Jarno Vanhatalo and Andrew O. Finley (in press). EcoMem: An R package for quantifying ecological memory. Environmental Modelling & Software, 119:305-308. [Link]
Demo code for the inversion of differential mobility particle sizer (DMPS) data and its extension SMPS inversion are available from here: DMPS and SMPS
Bjarke Mølgaard, Jarno Vanhatalo, Pasi Aalto, Nönne L. Prisle and Kaarle Hämeri (2016). Notably improved inversion of Differential Mobility Particle Sizer data obtained under conditions of fluctuating particle number concentrations.. Atmospheric Measurement Techniques, 9:741-751. [Link]
A demo code for GP priors for heterogeneous Student-t model with Laplace approximation and natural gradients.
Available at https://github.com/mahaa2/LP-approximation-and-NG-for-GPs-with-heteroscedastic-Student-t-model
Marcelo Hartmann and Jarno Vanhatalo (2019). Laplace approximation and natural gradient for Gaussian process regression with heteroscedastic Student-t model. Statistics and Computing, 29:753-773. [Link]