Researchers combine GPU acceleration, machine-learning potentials, quantum calculations, and human-supervised AI agents to tackle a longstanding challenge in computational materials science.

The next evolution of artificial intelligence in scientific computing may not come from building larger language models or simply adding more GPUs to existing supercomputers. It may come from teaching AI systems how to coordinate the complex computational workflows that make scientific discovery possible.

A research team led by scientists at Tohoku University in Japan has demonstrated a promising approach, combining multiple AI agents, machine-learning interatomic potentials, GPU-accelerated computing, and quantum-mechanical calculations to investigate one of the longstanding challenges in materials science: understanding how aluminum nanoparticles transition from relatively stable, oxide-protected particles into highly reactive materials.

The study, published in the Proceedings of the National Academy of Sciences (PNAS) DOI: https://doi.org/10.1073/pnas.2537780123 introduces a human-audited multi-agent framework called MLP Copilot (https://github.com/flarecentury/mlpcopilot). Rather than treating AI as an isolated computational assistant, the researchers use it as a coordination layer connecting scientific task planning, active learning, model development, validation, and computational evidence.

This workflow improves the scalability, reproducibility, and auditability of atomic simulations while keeping humans in control. For HPC, it demonstrates how AI agents can become essential infrastructure for managing complex scientific research.

Breaking through the computational limits of atomic simulation

The accurate simulation of chemical reactions at the atomic scale poses a formidable computational challenge. While density functional theory (DFT) and other first-principles quantum-mechanical approaches provide high-fidelity descriptions of atomic interactions, their significant computational demands inherently limit the scale and duration of simulations. Conventional ab initio methods are generally restricted to systems comprising fewer than 500 atoms and picosecond-scale observations. Conversely, while classical reactive force fields such as ReaxFF facilitate simulations of larger systems over extended periods, they often lack the precision required to capture complex charge-transfer dynamics, evolving chemical bonds, and unconventional atomic configurations. This discrepancy creates a notable gap between quantum-level accuracy and the computational capacity necessary to model realistic materials behavior.

The research team addressed this challenge by utilizing machine-learning potentials (MLPs) to derive atomic interaction behavior from quantum-mechanical reference calculations. By leveraging the attention-based DPA-2 architecture implemented via DeepMD-kit, the model successfully approximates atomic energies and forces with significantly greater computational efficiency than conventional first-principles methods. However, developing a robust reactive potential extends beyond simple neural network training; it necessitates an extensive exploration of atomic configurations, the identification of rare reaction states, rigorous error evaluation, and the validation of model reliability beyond the initial training dataset. To navigate these complexities, the researchers introduced the MLP Copilot framework.

MLP Copilot: Turning AI agents into scientific workflow coordinators

The framework organizes the development and validation of machine-learning potentials through a coordinated, human-supervised process.

Instead of relying on disconnected scripts and manually managed computational tasks, MLP Copilot connects several stages of the scientific computing lifecycle.

These include scientific task planning, literature and project-context organization, active-learning iterations, model evaluation, validation evidence collection, dataset management, checkpoint tracking, and approval-controlled model deployment.

Language-model agents organize project knowledge, retrieve relevant evidence, assist with validation planning, and prepare information for researchers.

Computational tools produce the actual calculations and validation measurements. They retain results as identifiable artifacts, including datasets, model checkpoints, error reports, and validation records. This distinction is important.

The AI agents are not treated as independent authorities capable of declaring that a scientific model is correct. Researchers retain control over model acceptance, scientific interpretation, and decisions about where a trained model can safely be used.

The approach establishes a framework in which AI can coordinate complicated computational research without removing scientific accountability.

Active Learning and GPU Acceleration Drive Model Development

The researchers used an iterative active-learning strategy to construct a comprehensive training dataset covering aluminum-oxygen interactions.

The workflow began with foundational quantum-mechanical molecular dynamics calculations and progressively expanded its knowledge of the atomic potential-energy landscape.

Four concurrently trained neural networks, initialized with different random seeds, were used in a Query-by-Committee approach. Differences between their predictions helped identify uncertain atomic configurations requiring additional investigation.

Candidate structures were explored using molecular dynamics and time-stamped force-bias Monte Carlo sampling. Configurations exceeding predefined uncertainty thresholds were selected for additional DFT calculations using CP2K.

The resulting reference data were incorporated into subsequent training iterations.

After approximately 80 iterations, the researchers had assembled a dataset containing 88,833 atomic configurations covering temperatures from 200 to 3,000 kelvin.

GPU acceleration also played an important role in managing the growing dataset.

The team employed smooth overlap of atomic positions (SOAP) descriptors to characterize local atomic environments and identify structural similarities. A furthest-point sampling strategy helped eliminate redundant configurations while preserving representative examples of chemically important structures.

According to the researchers, GPU acceleration of descriptor calculations and structural clustering delivered more than a 100-fold performance improvement compared with CPU implementations.

That improvement is particularly relevant for HPC users working with large scientific datasets. Accelerating the identification and selection of informative training configurations can reduce the computational overhead associated with building increasingly sophisticated machine-learning models.

Validation beyond the training dataset

A machine-learning potential is only useful when its predictions remain reliable across the physical conditions for which it is intended.

The researchers reported an energy root-mean-square error (RMSE) of 1.2 millielectronvolts per atom and a force RMSE of 0.126 electronvolts per angstrom on their testing data.

They also conducted independent validation using aluminum-oxide clusters ranging from Al13 to Al309, examining how model accuracy changed with system size and atomic coordination.

The smallest, highly undercoordinated oxygen-rich clusters presented the greatest prediction errors. Larger nanoparticle-like structures generally produced lower errors, reflecting their closer correspondence to the local atomic environments represented in the training dataset.

The researchers further evaluated molecular-dynamics inference performance across systems containing approximately 100 to 3,000 atoms.

Both simulation time per step and GPU-memory requirements exhibited near-linear scaling across the tested range.

This is an important finding for computational materials science because it suggests that the trained potential can extend simulations beyond the practical scale of direct ab initio molecular dynamics while retaining a defined validation boundary.

The results do not establish unlimited scalability or eliminate the need for quantum-mechanical reference calculations. Instead, they demonstrate a practical route for extending accurate atomistic modeling to larger systems.

The scientific discovery: An oxide shell that acts as a gatekeeper

The computational framework also enabled the researchers to investigate the oxidation behavior of aluminum nanoparticles.

Aluminum nanoparticles are attractive materials for high-energy applications because of their substantial energy density. However, their behavior is strongly influenced by the thin aluminum-oxide layer that naturally forms around their surfaces.

The precise mechanisms controlling the movement of aluminum and oxygen through this protective layer have remained an important scientific question.

The simulations revealed that aluminum outward diffusion, rather than oxygen transport, dominates mass transfer across the temperature regimes examined.

The researchers reported aluminum diffusion coefficients consistently exceeding those of oxygen by approximately two to three orders of magnitude.

More importantly, they identified two distinct behaviors in nanoparticles with an existing oxide shell.

The breathing mode: controlled oxidation

At approximately 1,500 K, the oxide shell undergoes dynamic structural fluctuations.

These temporary changes create nanoscale pathways through which atomic transport can occur. The shell remains sufficiently intact to constrain the reaction, effectively regulating access to the aluminum core.

The researchers describe this behavior as a breathing mode, in which transient transport channels permit continued oxidation without immediately exposing the entire reactive interior.

The rupture mode: rapid oxidation

At approximately 2,000 K, the simulations show a different behavior.

Thermomechanical stress causes the oxide shell to lose structural integrity and rupture, exposing the molten aluminum core to oxygen.

This removes an important transport barrier and enables rapid oxidation, including direct aluminum vaporization and gas-phase reactions.

The distinction provides a unified explanation for how temperature can change the behavior of aluminum nanoparticles from controlled oxidation to much more rapid combustion.

The researchers also identify oxide-shell thickness and particle morphology as potential parameters for controlling the transition.

Although the findings are based on computational modeling, the authors discuss supporting experimental observations from previous in situ transmission electron microscopy research.

Supercomputing infrastructure remains central to the discovery

While the research highlights the role of AI agents, the underlying computational infrastructure remains essential.

The study acknowledges the use of MASAMUNE-IMR at Tohoku University's Institute for Materials Research, as well as supercomputing facilities at the University of Tokyo.

The computational workflow also relies on established scientific software and methods, including CP2K, DeepMD-kit, molecular dynamics, density functional theory, and GPU-accelerated structural analysis.

This combination illustrates an increasingly important direction for HPC: AI is not necessarily replacing traditional scientific computing. Instead, it is becoming an additional layer that helps researchers organize, accelerate, and validate computational work.

For supercomputing centers, this creates opportunities as well as challenges.

Future scientific workflows may require closer integration between GPU-accelerated simulation engines, machine-learning training environments, scientific data repositories, workflow orchestration systems, and AI-agent platforms.

The ability to maintain model provenance, track computational artifacts, reproduce previous experiments, and establish reliable validation checkpoints could become just as important as raw floating-point performance.

The research also highlights the importance of matching computational resources to individual stages of scientific discovery. Quantum calculations provide reference data, GPU acceleration helps process and organize large collections of atomic configurations, and trained machine-learning potentials extend the reach of molecular-dynamics simulations.

The result is a heterogeneous computational workflow in which different technologies contribute to different parts of the research process.

A new direction for AI-driven scientific computing

The most significant contribution of the MLP Copilot framework lies in its comprehensive approach to scientific automation. Rather than employing AI solely for generating code or answering isolated queries, this framework demonstrates how multiple agents can orchestrate a rigorous research process that integrates computational tools, persistent evidence, systematic validation, and human oversight. This distinction is vital as research organizations increasingly investigate agentic AI systems capable of interacting directly with scientific software and infrastructure. While an AI agent capable of assisting with simulation planning is beneficial, an integrated system that manages datasets, identifies predictive uncertainties, coordinates concurrent calculations, tracks model versioning, and assembles validation evidence represents a transformative advancement. Although the framework currently addresses the specific challenges of aluminum nanoparticle oxidation, its coordination model provides a scalable foundation for future scientific discovery. Ultimately, this approach signals a shift in high-performance computing, where researchers define objectives, AI agents coordinate complex workflows, and validated machine-learning models extend the reach of established scientific methods.

An unusually powerful Antarctic polar vortex helped produce the seventh largest ozone hole in nearly five decades of monitoring. Behind the latest NASA findings lies a much larger scientific challenge: understanding how atmospheric chemistry, climate dynamics, satellite observations, and the computational systems used to model a changing planet interact.

The recent seasonal maximum of the Antarctic ozone hole serves as a poignant reminder of the inherent complexity of Earth's atmospheric systems. According to NASA’s Earth Observatory report titled "Ozone Hole Reaches Its 2026 Maximum," released on November 9, the ozone hole was ranked as the seventh largest by average area within the 48-year monitoring record. This finding is particularly significant, as it demonstrates the intricate interplay between atmospheric circulation, chemical reactions, and persistent human influences in driving year-over-year variability.

During the peak depletion period from September 7 to October 13, the ozone hole spanned an average of approximately 25 million square kilometers, reaching a maximum one-day extent of 27.1 million square kilometers on September 18, an area roughly triple the size of the contiguous United States. However, these figures warrant nuanced interpretation; while the hole was unusually expansive, its depth did not reach the severity observed in previous decades. For atmospheric scientists, this distinction is critical, as the spatial extent of depletion and the absolute concentration of ozone are distinct metrics that necessitate rigorous, sustained observation and sophisticated analytical methods.

A powerful polar vortex changes the equation

The principal factor behind this year's unusually large ozone hole was an exceptionally strong and extensive Antarctic polar vortex.

The polar vortex is a circulation of powerful winds that forms around Antarctica during the Southern Hemisphere winter. It effectively isolates extremely cold air over the continent, helping create the conditions necessary for polar stratospheric clouds to form.

These clouds provide surfaces on which chemical reactions involving chlorine and bromine compounds become especially effective at destroying ozone.

During August and September, atmospheric wave activity that normally helps disrupt the polar vortex was unusually weak. As a result, the vortex remained stronger and extended farther from the South Pole, allowing ozone-depleting conditions to persist over a larger region.

This is a critical distinction in interpreting the 2026 measurements. The unusually large ozone hole was not evidence that international efforts to protect the ozone layer had failed. Rather, atmospheric conditions temporarily amplified the effects of ozone-depleting chemicals that remain in the stratosphere.

NOAA measurements at the South Pole found that ozone concentrations directly overhead reached a seasonal low of 105 Dobson units on October 17. The historical low was 92 Dobson units, recorded in October 2006.

The Dobson unit measures the total amount of ozone in a vertical column of atmosphere. Scientists use 220 Dobson units as the threshold for defining the Antarctic ozone hole.

The word hole, however, is not literal. The atmosphere is not empty of ozone. It describes a region where ozone concentrations have fallen below that established threshold.

The chemistry behind decades of ozone depletion

The origins of the ozone crisis can be traced to chemicals that became widely used during the twentieth century.

Chlorofluorocarbons, or CFCs, were common in refrigeration systems, aerosol products and industrial applications. Their chemical stability allowed them to remain in the atmosphere for decades, eventually reaching the stratosphere.

There, ultraviolet radiation breaks down these compounds, releasing chlorine atoms capable of repeatedly destroying ozone molecules through catalytic chemical reactions.

Bromine-containing compounds also contribute to ozone depletion and can be particularly destructive.

The discovery of the Antarctic ozone hole in the 1980s fundamentally changed scientific understanding of how industrial chemicals could alter the global atmosphere.

The international response produced the 1987 Montreal Protocol, which established a global framework for phasing out the production and consumption of major ozone-depleting substances.

The results have been substantial. NASA and NOAA scientists report that chlorine levels in the Antarctic stratosphere attributable to ozone-depleting substances have declined by approximately one-third from their 1980s levels.

NASA estimates that the Antarctic ozone layer could return to its 1980 condition during the 2070s.

However, the chemicals responsible for ozone depletion have long atmospheric lifetimes. Even after production restrictions, legacy compounds continue circulating through the atmosphere, making recovery a process measured in decades rather than years.

The 2026 measurements demonstrate why scientists cannot simply declare the problem solved.

Ozone depletion and climate change are connected, but they are not the same problem

The relationship between ozone depletion and climate change is frequently misunderstood.

Both involve changes to atmospheric chemistry and circulation, but they are driven by different primary mechanisms.

Ozone depletion is principally associated with chlorine- and bromine-containing chemicals that destroy stratospheric ozone. Climate change is primarily driven by the accumulation of greenhouse gases, including carbon dioxide and methane, which alter Earth's energy balance.

However, the two processes interact.

Increasing greenhouse gas concentrations warm much of the lower atmosphere while generally cooling the stratosphere. Changes in atmospheric temperature and circulation can influence ozone chemistry, the formation of polar stratospheric clouds, and the timing of seasonal ozone depletion.

Ozone itself also influences atmospheric temperatures because it absorbs ultraviolet radiation.

Consequently, understanding the future of the ozone layer requires scientists to consider more than the concentration of ozone-destroying chemicals. They must also examine atmospheric circulation, temperature profiles, seasonal variability and interactions between different layers of the atmosphere.

Importantly, NASA's explanation of the 2026 ozone hole identifies the unusually strong polar vortex and weak atmospheric wave activity as the principal factors behind this year's large extent. It does not establish climate change as the cause of this particular year's result.

That distinction is essential to understanding the science accurately.

The supercomputing challenge: Modeling an atmosphere that never stops moving

Behind the satellite images and atmospheric measurements is a computational challenge that has become increasingly important to modern Earth science.

The atmosphere is a constantly changing three-dimensional system. Temperature, pressure, humidity, chemical concentrations, wind patterns, and solar radiation interact across enormous geographic distances and multiple timescales.

Understanding these interactions requires scientists to combine observations with mathematical models that simulate atmospheric physics and chemistry.

NASA's Earth Modeling Nexus is part of that effort. The initiative brings together Earth-system modeling, scientific computing, and data assimilation capabilities to improve the agency's understanding of interconnected planetary processes.

At the center of its computing infrastructure is NASA's Discover supercomputing cluster at Goddard Space Flight Center.

The system provides:

•  213,288 compute cores.
• Approximately 8.28 petaflops of computing performance.
• 49.5 petabytes of managed data storage.

Discover supports large-scale simulations involving atmospheric, oceanic, land, and ice processes. NASA describes the infrastructure as a means of transforming decades of Earth observations into high-resolution computational representations of planetary systems.

For atmospheric research, this type of infrastructure is particularly valuable.

Scientists must work with enormous observational datasets, reconstruct atmospheric conditions across time and space, and run simulations that account for interactions between physical processes and chemical reactions.

Numerical models can examine how atmospheric conditions evolve, test different assumptions, and investigate how changes in chemical concentrations or circulation patterns influence ozone depletion.

The computational requirements grow as researchers increase spatial resolution, incorporate additional chemical processes, and explore larger ensembles of possible atmospheric conditions.

More detailed simulations can help reveal regional variations that broad global averages may conceal.

It is important to distinguish this broader computational capability from the specific 2026 ozone measurement. NASA has not identified Discover as the system that directly calculated the September 18 ozone-hole map. Satellite instruments and ground-based measurements established the observed conditions. Supercomputing provides the broader modeling and analytical infrastructure used to understand atmospheric behavior and investigate future scenarios.

Satellites provide the evidence; computing provides the context

NASA and NOAA have spent decades developing the observational infrastructure needed to monitor the ozone layer.

The historical record includes measurements from NASA's Total Ozone Mapping Spectrometer instruments, the Ozone Monitoring Instrument aboard the Aura satellite, and the Ozone Mapping and Profiler Suite instruments aboard successive polar-orbiting satellites.

For 2026, the OMPS instrument aboard NOAA-20 contributed to the ozone monitoring record.

These observations are complemented by ground-based instruments and balloon-borne ozonesondes deployed from Antarctic research stations.

The combination is important because satellites provide broad geographic coverage, while ground-based instruments offer independent measurements that help validate and interpret the satellite record.

The resulting datasets allow researchers to compare current conditions against decades of observations.

That historical perspective is indispensable. A single year's ozone-hole measurements cannot establish whether the atmosphere is recovering or deteriorating. Scientists need consistent observations, validated datasets, and computational models capable of separating long-term trends from natural variability.

The same principles apply to climate research, where researchers must distinguish long-term changes from seasonal fluctuations and short-lived atmospheric events.

The next generation of Earth science will depend on computational power

The 2026 ozone hole highlights why high-performance computing is vital for environmental science. As data volume grows, researchers must integrate satellite observations with physics-based simulations to model atmospheric complexity. The future of Earth science relies on combining numerical modeling, advanced data assimilation, and AI, as seen in initiatives like NASA’s Earth Modeling Nexus. Ultimately, accurately monitoring our changing planet requires the computational power to synthesize vast datasets and simulate dynamic atmospheric processes.