Researchers at the Department of Energy's Lawrence Berkeley and Oak Ridge national laboratories are developing DELERIA, a new data streaming platform aimed at dramatically speeding up the processing time of nuclear physics experiments from hours or days to mere seconds. According to a statement obtained by the Lawrence Berkeley National Laboratory, the software is engineered to interface with the Gamma Ray Energy Tracking Array (GRETA), facilitating a direct data feed to the nation's leading computing centers for lightning-fast analysis.
Set to operate alongside GRETA at the Facility for Rare Isotope Beams at Michigan State University in 2026, the goal of DELERIA is to rapidly analyze large datasets and give researchers the ability to quickly make necessary adjustments to their experiments. Gustav Jansen, a computational nuclear physicist at ORNL, highlighted the broader vision: "Our primary goal is to establish a data pipeline for future experimental needs without having to build a local computing infrastructure." Jansen emphasized the platform's versatility, projecting to soon demonstrate DELERIA's potential scalability across diverse scientific applications, as detailed by ORNL.
In practice, GRETA will detect photon collisions emitted from charged particle interactions within its germanium crystal array, and relay this data through the Energy Sciences Network, or ESnet. Each recorded photon hit is a puzzle piece in uncovering the intricacies of matter at a nuclear level. Processing these events not only requires identifying the precise energy and coordinates of impacts but also converting that raw data into something meaningful for scientists.
Jansen provided a clearer picture of the process, stating, "When a photon hits one of the crystals, an electrical signal goes out on each contact point. We need to figure out where inside the crystal the photons hit — the X, Y, Z coordinates." The task for the pipeline is to rapidly process this information, a challenge considering the physical distances that the data needs to travel, according to ORNL. Working with collaborators, Jansen intends to considerably shorten latency issues by employing parallel processing during transfers to keep the computing cluster actively engaged.
The ambitious project is part of the DOE's efforts to establish a seamless integrated research framework, as noted by ORNL. Utilizing OLCF's Defiant computing cluster, the team has shrunk the volume of required data storage by a factor of 40 and successfully managed streams of data reaching 35 gigabits per second, surpassing GRETA's needs and highlighting the proficiency of ESnet6's infrastructure. "This is really pioneering work that's setting the standards for how we network the nation's leading centers of research," Tom Beck from ORNL's National Center for Computation Sciences remarked in acknowledgment of the importance of such advances in data processing for scientific discovery.
While the DELERIA software development is a Laboratory Directed Research and Development project led by Berkeley Lab, it also garners support from the DOE Office of Science’s Advanced Scientific Computing Research program and the Office of Nuclear Physics, illustrating the collaborative efforts underpinning this groundbreaking initiative. Ultimately, DELERIA anticipates ushering in a new era in nuclear physics experimentation, where the processing speed and analytical prowess redefine what is achievable in scientific exploration and discovery.
