Skip to Main Content U.S. Department of Energy
mathematical sciences, Computational Sciences & Mathematics

PNNL’s Computing Research portfolio—spanning from basic to applied—includes data and computational engineering, high-performance computing, applied mathematics, semantic and human language technologies, machine learning, data and computing architectures, systems integration, and software and application development. At PNNL, scientists, engineers, programmers, and researchers work together to apply advanced theories, methods, algorithms, models, evaluation tools and testbeds, and computational-based solutions to address complex scientific challenges affecting energy, biological sciences, the environment, and national security.


Automating Expertise

A lack of labeled data is a bottleneck for deep learning that impacts many domains. To remedy this, PNNL scientists developed ChemNet, which learns expert knowledge from large unlabeled databases and outperforms current deep learning methods. ChemNet also has potential applications in other research areas with similar data challenges. This month, the team will showcase ChemNet during the 24th ACM SIGKDD Conference on Knowledge Discovery and Data Mining in London.


Examining the Irregular

PNNL scientists Antonino Tumeo (High Performance Computing), Mahantesh Halappanavar (Data Sciences), and John Feo (Northwest Institute for Advanced Computing) with Intel’s Fabrizio Petrini, are teaming up to guest edit the first-ever thematic special issue of ACM Transactions on Parallel Computing, which will focus on topics concerning irregular applications. A Call for Papers for the special issue has been released with submissions due by Oct. 31, 2018.


Xantheas Named to the Washington State Academy of Sciences

Congratulations to Sotiris Xantheas, a PNNL Laboratory Fellow and UW-PNNL Distinguished Faculty Fellow, who is a newly elected member of the Washington State Academy of Sciences. As an academy member, he will join 286 scientists and engineers from across the state, including 22 from PNNL, who share their expertise to address issues impacting the state and its residents.


Deep Learning Could Help Detect Nuclear Events Worldwide

Scientists Emily Mace and Jesse Ward, both from the National Security Directorate, teamed to explore the promise of deep learning to interpret signals from radioactive decay events, which could indicate underground nuclear testing. Their results showed the deep learning techniques separated signal events from instrument “noise” with nearly 100 percent accuracy and performed 25 times better than current computational methods. Mace presented their work at the 11th Methods and Applications of Radioanalytical Chemistry conference in Hawaii.


PNNL Thought Leaders Quoted in PC Magazine

Jonathan Cree and William Moeglein, both with the National Security Directorate, were interviewed for an article, “A Guide to Using BI Apps With Edge Computing,” recently featured in PC Magazine. The duo were asked for their take on the interconnections between deep neural networks, the cloud, and edge computing.

View Additional Highlights

Computing Research

Research Areas