PNNL researchers created a model that helps materials scientists not only predict what is, but what could be.

PNNL researchers design an artificial intelligence-guided electron microscope in a Laboratory Directed Research and Development study.

ICON science is a Department of Energy-developed framework to enhance scientific outcomes via more intentional design of research efforts across all domains of science.

New data analysis approach gives researchers deeper understanding of microbial ecology and evolution.

Quantum simulators help navigate noise on GPU-based HPC systems.

Improving predictability of water quality and nutrient cycling in large river systems.

Developing conceptual models for microbial-environmental–ecosystem interactions is key to enhancing the ability of models to predict future ecosystem function.

Research from the PNNL-led WHONDRS consortium reveals chemistry and spatial gradients of metabolomes from river water and sediments around the globe.

Researchers achieved the complete assignment of the infrared spectrum of a hydronium-water “magic number” cluster reported in Science in 2004.

A multi-institutional team has developed a deep learning framework to identify novel molecules as drug candidates.

PNNL researchers developed a way to produce accurate simulations of complex physical systems using renormalized approximations across multiple scales

Detailed characterization of molecules in, under, and along the Columbia River reveal potential mechanisms for biogeochemical cycling.

Scientists created a benchmark dataset and explored new algorithms to better predict changes in the intensity of North Atlantic tropical cyclones.

Hypergraphs can more faithfully identify important genes in complex immune responses.

The web-based tool, called PLATIPUS, is publicly available and interacts with publications available in the CovidScholar database.

Seeking a deeper understanding of when, where, and why large wildfires occur using explainable artificial intelligence.

Researchers used deep learning methods to estimate the subsurface permeability of a watershed from stream discharge measurements.

Rigorous mathematical methods can help avoid unnecessary simplifications in numerical models of atmospheric clouds.

More accurate methods provide better results in numerical simulations of atmospheric clouds by avoiding accidental contradictions.

A code comparison study demonstrates a leap in the accuracy and reliability of modeling gas hydrate for the discovery and production of natural gas.