GPU-Accelerated Hierarchical Bayesian Inference with Application to Modeling Cosmic Populations: CUDAHM

by   János M. Szalai-Gindl, et al.

We describe a computational framework for hierarchical Bayesian inference with simple (typically single-plate) parametric graphical models that uses graphics processing units (GPUs) to accelerate computations, enabling deployment on very large datasets. Its C++ implementation, CUDAHM (CUDA for Hierarchical Models) exploits conditional independence between instances of a plate, facilitating massively parallel exploration of the replication parameter space using the single instruction, multiple data architecture of GPUs. It provides support for constructing Metropolis-within-Gibbs samplers that iterate between GPU-accelerated robust adaptive Metropolis sampling of plate-level parameters conditional on upper-level parameters, and Metropolis-Hastings sampling of upper-level parameters on the host processor conditional on the GPU results. CUDAHM is motivated by demographic problems in astronomy, where density estimation and linear and nonlinear regression problems must be addressed for populations of thousands to millions of objects whose features are measured with possibly complex uncertainties. We describe a thinned latent point process framework for modeling such demographic data. We demonstrate accurate GPU-accelerated parametric conditional density deconvolution for simulated populations of up to 300,000 objects in  1 hour using a single NVIDIA Tesla K40c GPU. Supplementary material provides details about the CUDAHM API and the demonstration problem.


page 1

page 2

page 3

page 4


RLT2-based Parallel Algorithms for Solving Large Quadratic Assignment Problems on Graphics Processing Unit Clusters

This paper discusses efficient parallel algorithms for obtaining strong ...

GIGA-Lens: Fast Bayesian Inference for Strong Gravitational Lens Modeling

We present GIGA-Lens: a gradient-informed, GPU-accelerated Bayesian fram...

Merlin HugeCTR: GPU-accelerated Recommender System Training and Inference

In this talk, we introduce Merlin HugeCTR. Merlin HugeCTR is an open sou...

Finite Population Survey Sampling: An Unapologetic Bayesian Perspective

This article attempts to offer some perspectives on Bayesian inference f...

Montblanc: GPU accelerated Radio Interferometer Measurement Equations in support of Bayesian Inference for Radio Observations

We present Montblanc, a GPU implementation of the Radio interferometer m...

Machine Learning based Autotuning of a GPU-accelerated Computational Fluid Dynamics Code

A machine learning-based autotuning technique is employed to optimize 14...

CosmoPower-JAX: high-dimensional Bayesian inference with differentiable cosmological emulators

We present CosmoPower-JAX, a JAX-based implementation of the CosmoPower ...

Please sign up or login with your details

Forgot password? Click here to reset