Sparse solvers provide essential functionality for a wide variety of scientific applications. Highly parallel sparse solvers are essential for continuing advances in high-fidelity, multi-physics and multi-scale simulations, especially as we target exascale platforms. This paper describes the challenges, strategies and progress of the US Department of Energy Exascale Computing project towards providing sparse solvers for exascale computing platforms. We address the demands of systems with thousands of high-performance node devices where exposing concurrency, hiding latency and creating alternative algorithms become essential. The efforts described here are works in progress, highlighting current success and upcoming challenges. This article is part of a discussion meeting issue 'Numerical algorithms for high-performance computational science'.
Download full-text PDF |
Source |
|---|---|
| http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7015295 | PMC |
| http://dx.doi.org/10.1098/rsta.2019.0053 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
A fully linearized ADMM algorithm for optimization based image reconstruction.
J Xray Sci Technol
December 2024
Department of Radiation and Cellular Oncology, University of Chicago, Chicago, IL, USA.
Background And Objective: Optimization based image reconstruction algorithm is an advanced algorithm in medical imaging. However, the corresponding solving algorithm is challenging because the model is usually large-scale and non-smooth. This work aims to devise a simple and convergent solver for optimization model.
View Article and Find Full Text PDFMegavariate Methods Capture Complex Genotype-by-Environment Interactions.
Genetics
November 2024
Corteva Agrisciences, 8305 NW 62nd Ave, Johnston, IA, USA 50131.
Article Synopsis
- Genomic prediction models help forecast how certain genotypes will perform in different environments, but they can be hard to compute effectively.
- This study presents three advanced algorithms (MegaLMM, MegaSEM, and PEGS) designed to tackle genotype-by-environment interactions and assesses their accuracy and runtime efficiency through simulated scenarios.
- Results showed that MegaLMM and PEGS models achieved high accuracy with many testing environments, and PEGS was significantly faster than traditional methods, with MegaSEM excelling in speed when handling large datasets.
Solving Zero-Shot Sparse-View CT Reconstruction With Variational Score Solver.
IEEE Trans Med Imaging
October 2024
Article Synopsis
- Computed tomography (CT) is a crucial diagnostic tool in medicine, but concerns about radiation exposure have led to the need for lower doses, which typically reduce image quality.
- The study introduces a new method called Variational Score Solver (VSS) that improves CT reconstruction from limited data without requiring paired datasets, thus avoiding traditional constraints in deep learning techniques.
- VSS uses a novel distribution-based method to achieve high-quality reconstructions and shows better performance than existing unsupervised methods and results similar to advanced supervised techniques, with code available online.
Sparse-spike seismic inversion with semismooth newton algorithm solver.
Sci Rep
August 2024
School of Mathematics and Information, China West Normal University, Nanchong, 637009, China.
Seismic prospecting has been widely used in the exploration and development of underground geological resources, such as mineral products (e.x., coal, uranium deposit), oil and gas, groundwater, and so forth.
View Article and Find Full Text PDFComputing Smooth and Integrable Cross Fields via Iterative Singularity Adjustment.
IEEE Trans Vis Comput Graph
June 2024
We propose a new method for computing smooth and integrable cross fields on 2D and 3D surfaces. We first compute smooth cross fields by minimizing the Dirichlet energy. Unlike the existing optimization based approaches, our method determines the singularity configuration, i.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!