Preview

PDF (In-press) 129174 - prepri...pdf | Download (1MB) | Preview

Abstract

Probability Density Approximation (PDA) is a non-parametric method of calculating probabilitydensities. When integrated into Bayesian estimation, it allows researchers to fit psychologicalprocesses for which analytic probability functions are unavailable, significantly expanding thescope of theories that can be quantitatively tested. PDA is, however, computationally intensive,requiring large numbers of Monte Carlo simulations to attain good precision. We introduceParallel PDA (pPDA), a highly efficient implementation of this method utilizing Armadillo C++and CUDA C libraries to conduct millions of model simulations simultaneously in graphicsprocessing units (GPUs). This approach provides a practical solution for rapidly approximatingprobability densities with high precision. In addition to demonstrating this method, we fit a Piecewise Linear Ballistic Accumulator model (Holmes, Trueblood & Heathcote, 2016) toempirical data. Finally, we conduct simulation studies to investigate various issues associatedwith the PDA and provide guidelines for pPDA applications to other complex cognitive models.

Item Type:	Article
Authors/Creators:	Lin, YS and Heathcote, A and Holmes, WR
Keywords:	R, C++, CUDA, GPU, kernel density estimate, Markov Chain Monte Carlo
Journal or Publication Title:	Behavior Research Methods
Publisher:	Springer New York LLC
ISSN:	1554-3528
DOI / ID Number:	https://doi.org/10.3758/s13428-018-1153-1
Copyright Information:	Copyright The Psychonomic Society, Inc. 2019Post-prints are subject to Springer Nature re-use terms
Related URLs:	Google Scholar: Heathcote, A UTAS Profile: Heathcote, A
Item Statistics:	View statistics for this item

Actions (login required)

Item Control Page