Cataloging the visible universe through Bayesian inference in Julia at petascale

Regier, Jeffrey; Fischer, Keno; Pamnany, Kiran; Noack, Andreas; Revels, Jarrett; Lam, Maximilian; Howard, Steve; Giordano, Ryan; Schlegel, David J.; McAuliffe, Jon; Thomas, R. C.; Prabhat,

doi:10.1016/j.jpdc.2018.12.008

Cited by 13 publications

(7 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Remarkably, a recent project using Julia parallelism achieved peak performance over a petaflop per second. 63 Julia utilizes high-level semantics similar to interpreted Additionally, Julia has a sophisticated multiple dispatch system that, similar to function overloading, allows the definition of multiple versions (termed methods) of a function. For the previously defined function f, we might want to add a special behavior for the case where the x is an integer and y a float-point number.…”

Section: ■ Juliamentioning

confidence: 99%

See 1 more Smart Citation

Fermi.jl: A Modern Design for Quantum Chemistry

Aroeira

Davis

Turney

et al. 2022

J. Chem. Theory Comput.

View full text Add to dashboard Cite

Approximating molecular wave functions involves heavy numerical effort; therefore, codes for such tasks are written completely or partially in efficient languages such as C, C++, and Fortran. While these tools are dominant throughout quantum chemistry packages, the efficient development of new methods is often hindered by the complexity associated with code development. In order to ameliorate this scenario, some software packages take a dual approach where a simpler, higher-level language, such as Python, substitutes the traditional ones wherever performance is not critical. Julia is a novel, dynamically typed, programming language that aims to solve this two-language problem. It gained attention because of its modern and intuitive design, while still being highly optimized to compete with “low-level” languages. Recently, some chemistry-related projects have emerged exploring the capabilities of Julia. Herein, we introduce the quantum chemistry package Fermi.jl, which contains the first implementations of post-Hartree–Fock methods written in Julia. Its design makes use of many Julia core features, including multiple dispatch, metaprogramming, and interactive usage. Fermi.jl is a modular package, where new methods and implementations can be easily added to the existing code. Furthermore, it is designed to maximize code reusability by relying on general functions with specialized methods for particular cases. The feasibility of the project is explored through evaluating the performance of popular ab initio methods. It is our hope that this project motivates the usage of Julia within the community and brings new contributions into Fermi.jl.

show abstract

Section: ■ Juliamentioning

confidence: 99%

“…Moreover, unlike older languages such as Fortran and C, new processor and parallelism technologies have been considered from the ground-up in Julia. Remarkably, a recent project using Julia parallelism achieved peak performance over a petaflop per second …”

Section: Juliamentioning

confidence: 99%

Fermi.jl: A Modern Design for Quantum Chemistry

Aroeira

Davis

Turney

et al. 2022

J. Chem. Theory Comput.

View full text Add to dashboard Cite

show abstract

“…Julia, while being a relatively new language to the HPC scene, already has several case studies on implementing fullscale HPC applications. Regier et al experimented with implementing bayesian inference for large datasets using Julia that runs on a cluster of 8192 Xeon Phi nodes [5]. Based on the new work of Besard et al on bringing Julia to GPUs [6], we have seen several successful HPC application implementations in Julia.…”

Section: Related Workmentioning

confidence: 99%

Comparing Julia to Performance Portable Parallel Programming Models for HPC

Lin

McIntosh–Smith

2021

2021 International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS)

View full text Add to dashboard Cite

Julia is a general-purpose, managed, strongly and dynamically-typed programming language with emphasis on high performance scientific computing. Traditionally, HPC software development uses languages such as C, C++ and Fortran, which compile to unmanaged code. This offers the programmer near bare-metal performance at the expense of safety properties that a managed runtime would otherwise provide. Julia, on the other hand, combines novel programming language design approaches to achieve high levels of productivity without sacrificing performance while using a fully managed runtime.This study provides an evaluation of Julia's suitability for HPC applications from a performance point of view across a diverse range of CPU and GPU platforms. We select representative memory-bandwidth bound and compute bound mini-apps, port them to Julia, and conduct benchmarks across a wide range of current HPC CPUs and GPUs from vendors such as Intel ® , AMD ® , NVIDIA ® , Marvell ® , and Fujitsu ® . We then compare and characterise the results against existing parallel programming frameworks such as OpenMP ® , Kokkos, OpenCL ™ , and firstparty frameworks such as CUDA ® , HIP ™ , and oneAPI ™ SYCL ™ . Finally, we show that Julia's performance either matches the competition or is only a short way behind.

show abstract

“…The latter problem is particularly important since many downstream applications will yield misleading results unless sources are properly separated (Portillo et al, 2017), such as analyses which attempt to classify point sources based on, e.g., their intensity. These challenges have motivated the development of Bayesian models which can simultaneously capture uncertainty about both the number and locations of sources, including approaches which can scale to catalogue massive astronomical surveys (Regier et al, 2019;Liu et al, 2021b). However, extracting interpretable results from these models can be challenging (Feder et al, 2020).…”

Section: Application To Astronomical Point Source Detectionmentioning

confidence: 99%

Controlled Discovery and Localization of Signals via Bayesian Linear Programming

Asher¹,

Janson²

2022

Preprint

View full text Add to dashboard Cite

In many statistical problems, it is necessary to simultaneously discover signals and localize them as precisely as possible. For instance, genetic fine-mapping studies aim to discover causal genetic variants, but the strong local dependence structure of the genome makes it hard to identify the exact locations of those variants. So the statistical task is to output as many regions as possible and have those regions be as small as possible while controlling how many outputted regions contain no signal. The same problem arises in any application where signals cannot be perfectly localized, such as locating stars in astronomical sky surveys and change point detection in time series data. However, there are two competing objectives: maximizing the number of discoveries and minimizing the size of those discoveries (all while controlling false positives), so our first contribution is to propose a unified measure called resolutionadjusted power that formally trades off these two objectives and thus, in principle, can be maximized subject to a constraint on false positives. We take a Bayesian approach, but the resulting posterior optimization problem is non-convex and extremely high-dimensional. Thus our second contribution is Bayesian Linear Programming (BLiP), a method which overcomes this intractability to jointly detect and localize signals in a way that verifiably nearly maximizes the expected resolution-adjusted power while provably controlling false positives. BLiP is very computationally efficient and can wrap around nearly any Bayesian model and algorithm. Applying BLiP on top of existing state-of-the-art analyses of UK Biobank data (for genetic fine-mapping) and the Sloan Digital Sky Survey (for astronomical point source detection) increased resolution-adjusted power by 30-120% in just a few minutes of computation. BLiP is implemented in the new packages pyblip (Python) and blipr (R).

show abstract

Cataloging the visible universe through Bayesian inference in Julia at petascale

Cited by 13 publications

References 18 publications

Fermi.jl: A Modern Design for Quantum Chemistry

Fermi.jl: A Modern Design for Quantum Chemistry

Comparing Julia to Performance Portable Parallel Programming Models for HPC

Controlled Discovery and Localization of Signals via Bayesian Linear Programming

Contact Info

Product

Resources

About