Supercomputers push neural quantum simulation beyond previous limits
Supercomputers push neural quantum simulation beyond previous limits
previous arrow
previous arrow
next arrow
next arrow
Shadow
Featured

Supercomputers push neural quantum simulation beyond previous limits

JAIST researchers combine artificial intelligence, Bayesian physics, and high-performance computing to make quantum Monte Carlo practical for larger molecular systems.

For decades, quantum chemists have grappled with a significant hurdle: the most precise methods for simulating molecular electronic behavior are also the most computationally demanding, limiting their use to small systems. Researchers from the Japan Advanced Institute of Science and Technology (JAIST), in collaboration with China’s ByteDance Seed and other institutions, have developed a solution.
 
As detailed in Nature Computational Science, their new framework integrates neural-network quantum Monte Carlo with a novel Bayesian localization technique. This innovation significantly lowers computational costs while maintaining the high accuracy required for first-principles simulations. Beyond the AI application, this work highlights the growing synergy between machine learning and high-performance computing, demonstrating how hybrid workflows can solve complex scientific problems that neither approach could effectively address in isolation.

Quantum Monte Carlo meets artificial intelligence

Quantum Monte Carlo (QMC) methods are widely regarded as among the most accurate computational techniques for solving the Schrödinger equation governing interacting electrons.
 
Unlike conventional density functional theory, QMC explicitly samples the quantum behavior of electrons using stochastic methods, often producing benchmark-quality predictions for molecular energies and material properties. The tradeoff has always been computational expense.
 
In recent years, neural-network wavefunctions have dramatically improved the expressive power of QMC calculations, allowing machine learning models to represent extremely complex electronic structures. However, training and evaluating these neural networks has introduced a new bottleneck: enormous computational requirements that restricted practical simulations to relatively modest molecular systems.
 
The JAIST-led team set out to remove that bottleneck.

A Bayesian shortcut for quantum physics

The researchers developed what they call Bayesian Localization of the Pseudo Hamiltonian, a mathematical framework that replaces computationally expensive nonlocal pseudopotential evaluations with localized approximations while maintaining high physical fidelity.
 
Rather than sacrificing accuracy for speed, the Bayesian framework intelligently estimates the localized interactions needed during quantum Monte Carlo sampling.
 
The result is a neural-network quantum simulation workflow that remains highly accurate while requiring substantially fewer computational resources. According to the researchers, the approach enables high-precision simulations of significantly larger molecular and materials systems than were previously practical.
 
For computational scientists, this represents the kind of algorithmic innovation that often produces larger performance gains than incremental hardware improvements alone.

Supercomputers still do the heavy lifting

Although artificial intelligence plays a central role, the research is fundamentally an HPC achievement.
 
The study relies on large-scale numerical simulation rather than replacing physics with machine learning. Neural networks become one component inside a much larger quantum computational pipeline that still demands substantial parallel computing resources.
 
The authors note that some of the calculations were performed using the facilities of the Center for Advanced Scientific Computing at JAIST, underscoring that state-of-the-art AI models continue to depend on advanced scientific computing infrastructure for both development and validation.
 
This reflects a growing trend across computational science: AI increasingly accelerates scientific simulation, but supercomputers remain the engines that make those simulations possible.

The rise of AI-augmented scientific computing

The new methodology belongs to a rapidly expanding class of hybrid computational techniques.
 
Rather than asking AI to replace traditional numerical simulation, researchers are embedding machine learning directly into established scientific algorithms.
 
In this study, neural networks improve the representation of electronic wavefunctions while Bayesian inference reduces the computational burden of evaluating pseudopotentials. The surrounding quantum Monte Carlo framework continues to enforce the underlying laws of quantum mechanics.
 
This philosophy differs fundamentally from purely data-driven AI.
 
Instead of learning chemistry from experimental databases alone, the algorithm performs physics-based simulations whose efficiency is enhanced by modern machine learning.
 
That distinction is increasingly defining next-generation scientific computing.

From molecules to materials

Reducing computational cost has implications far beyond faster benchmark calculations.
 
Many technologically important systems, including battery materials, heterogeneous catalysts, superconductors, semiconductor defects, and complex biomolecules, remain difficult to model accurately because of their electronic complexity.
 
The authors suggest their framework opens opportunities for investigating larger materials systems, more complicated chemical reactions, and biological phenomena that have previously remained beyond the practical reach of neural-network quantum Monte Carlo methods. Future extensions are expected to include broader elemental coverage, solid-state physics, and excited-state calculations.
 
For materials discovery, each increase in computational efficiency translates directly into larger searchable design spaces and more realistic simulations.

Algorithmic innovation as a performance multiplier

The history of supercomputing has often been told through faster processors and larger machines.
Yet many of the greatest advances have come from mathematics rather than hardware.
 
Multigrid solvers transformed computational fluid dynamics.
 
Fast Fourier Transforms revolutionized signal processing.
 
Sparse linear algebra enabled simulations that once seemed impossible.
 
The Bayesian localization strategy introduced in this work belongs to that same tradition.
 
Instead of waiting for future hardware generations, the researchers redesigned part of the quantum simulation itself, allowing existing HPC systems to solve substantially larger scientific problems.

Curiosity at the intersection of AI and HPC

As exaflops supercomputing continues to mature, researchers increasingly recognize that scientific progress will depend on both larger machines and smarter algorithms. The JAIST collaboration offers a compelling example of that convergence. Artificial intelligence contributes expressive neural representations. Bayesian statistics streamline quantum calculations. High-performance computing provides the computational foundation on which both operate. Together, they form a workflow capable of pushing neural-network quantum computation into scientific regimes that were previously impractical. For the HPC community, that may be the study’s most important lesson.
 
The next breakthroughs in computational chemistry are unlikely to come from AI alone or from faster supercomputers alone; they will emerge from carefully engineered collaborations between advanced algorithms and advanced computing infrastructure, where every improvement in mathematics unlocks more science from every available processor.
Featured

Supercomputers replace ‘bathtub’ flood maps with physics-based digital twins of Britain’s coastline

National-scale hydrodynamic simulations reveal how advances in high-performance computing are transforming centuries-long coastal flood forecasting.

For decades, global coastal flood projections have relied on the simplifying assumption that any land below projected sea level will naturally flood. Researchers often term these "bathtub models" because they treat the landscape as a basin being filled with water. However, a new study published in Nature Communications signals a shift toward far more sophisticated methodologies.
 
By leveraging advances in high-performance computing (HPC), researchers have developed a physics-based simulation capable of modeling coastal flooding across the entire United Kingdom at an unprecedented spatial resolution, with projections extending three centuries into the future. Ultimately, this research highlights how modern supercomputing has fundamentally expanded the computational boundaries of environmental simulation.

Computing the movement of water

The computational challenge facing flood modelers is enormous.
 
Water does not simply spread evenly across landscapes. It accelerates, slows, changes direction, interacts with rivers, follows terrain, overtops barriers, and responds dynamically to tides and storm surges. Capturing these processes requires solving the shallow-water equations across millions of computational cells while simultaneously modeling river networks and coastal boundaries.
 
Only recently have advances in numerical methods, high-resolution terrain datasets, and HPC resources made these simulations practical on national scales. As the authors explain, improvements in numerical schemes and High Performance Computing now allow hydrodynamic flood modeling to be applied at scales ranging from national to global, something that was previously impractical. Instead of assuming every low-lying area floods, the new model computes where water can physically travel, how quickly it moves, and how hydraulic connectivity influences inundation.

Building a digital twin of the United Kingdom

At the heart of the study is one of the most sophisticated national flood models yet constructed.
 
The researchers simulate the entire United Kingdom using a coupled hydrodynamic model operating at approximately 20–25-meter spatial resolution. The system combines two-dimensional shallow-water flow across floodplains with one-dimensional river channel simulations, allowing rivers of every size to interact realistically with coastal flooding.
 
Terrain elevations are derived primarily from airborne LiDAR surveys with approximately 10-centimeter vertical accuracy, while river geometries, coastal boundaries, storm surge profiles, tidal cycles, and wave setup are incorporated into a unified computational framework.
 
The result is effectively a high-resolution digital twin of Britain’s coastline capable of responding dynamically to changing sea levels and extreme coastal events.

Millions of calculations before the flooding even begins

The flood simulations represent only one stage of the computational workflow.
 
Before any water is modeled, the research constructs physically consistent sea-level rise storylines from 450,000 Monte Carlo simulations for each emissions scenario. These simulations combine multiple interacting components, including:
  • Ocean thermal expansion
  • Antarctic ice-sheet loss
  • Greenland ice-sheet melt
  • Mountain glacier contributions
  • Land-water storage changes
Researchers then filter these enormous ensembles to identify internally consistent future trajectories before coupling them into the hydrodynamic flood simulations.
 
This layered modeling strategy highlights how modern environmental science increasingly depends on large ensembles and computational statistics long before the primary physical simulations begin.

From static maps to dynamic physics

Traditional flood maps often assume that any land below a projected water elevation becomes inundated.
 
The authors argue this approach can substantially overestimate flooding because it ignores hydraulic connectivity and the actual physics governing water movement.
 
Their hydrodynamic approach instead solves the governing equations of fluid motion, enabling simulations that account for terrain, river channels, coastal geometry, tides, storm surges, and evolving water depths throughout an event.
 
The difference is analogous to replacing a static elevation map with a fully interactive fluid simulation.
 
For engineers, planners, and emergency managers, that distinction can significantly improve confidence in identifying which infrastructure is genuinely vulnerable.

Simulating centuries instead of storms

The study explores flooding under five physically consistent sea-level storylines extending through the years 2100, 2200, and 2300.
 
Each scenario requires repeated national-scale hydrodynamic simulations based on 1-in-200-year coastal storm events, allowing researchers to examine how changing boundary conditions alter flood behavior over centuries rather than days.
 
While the scientific conclusions concern long-term coastal exposure, the computational achievement is equally notable: running repeated high-resolution simulations across an entire nation using physically based hydrodynamics.

A broader trend in scientific computing

Flood modeling joins an expanding list of scientific disciplines transforming HPC.
 
Fields ranging from molecular dynamics and astrophysics to weather forecasting and materials science have increasingly abandoned simplified approximations in favor of direct numerical simulation as computational resources have expanded.
 
This study illustrates that coastal science is following the same trajectory.
 
Instead of asking where water might accumulate based solely on elevation, researchers can now simulate how water actually behaves.
 
That shift produces richer scientific insight while creating more realistic digital representations of complex natural systems.

Supercomputing as the enabling technology

Perhaps the paper’s most important contribution to computational science is not any individual flood projection but the demonstration that national-scale, physics-based environmental digital twins have become practical.
 
The authors explicitly acknowledge that advances in high performance computing were instrumental in making these simulations possible.
 
As exaflops supercomputing matures and increasingly detailed terrain datasets become available worldwide, similar computational frameworks could eventually model coastlines across entire continents with greater spatial resolution, larger ensembles, and more comprehensive representations of physical processes.
 
For the HPC community, that represents the real story.
 
The future of flood prediction is no longer built on filling digital bathtubs, it is built on solving the equations of motion across millions of grid cells, transforming coastlines into living computational systems that evolve under the laws of physics.
Featured

Melting icebergs may be reshaping Earth’s greatest ocean current

New climate simulations reveal how Pacific iceberg melt may weaken the Atlantic Meridional Overturning Circulation, offering fresh insight into one of Earth’s most important climate systems

For decades, climate scientists have primarily attributed potential slowdowns in the Atlantic Meridional Overturning Circulation (AMOC), the global conveyor belt of ocean currents responsible for heat redistribution, to melting ice in Greenland and the North Atlantic.
 
However, recent findings published in Nature Communications, supported by research from the University of California, Davis, suggest that the narrative may begin thousands of miles away. The study reveals that melting icebergs in the North Pacific played a critical role in weakening the AMOC during Earth’s last deglaciation. By leveraging the unprecedented fidelity of modern supercomputers, researchers have uncovered long-distance connections between ocean basins that were previously impossible to observe.
 
While this research centers on events from 19,000 years ago, its implications are modern and urgent. By tracing how freshwater pulses travel through global currents, scientists are gaining a clearer understanding of how contemporary ice loss could influence Earth’s climate in the centuries to come.

Turning supercomputers into time machines

Unlike many scientific disciplines, climate researchers cannot perform controlled experiments on Earth’s oceans.
 
Instead, they recreate Earth’s past inside supercomputers.
 
The international research team employed the isotope-enabled Community Earth System Model (iCESM1.3), one of the world’s most sophisticated coupled climate models. The model integrates atmospheric physics, land processes, ocean circulation, and sea ice dynamics into a single simulation capable of reproducing interactions across the entire Earth system.
 
To ensure computational reliability, the researchers compared simulations originally performed on the Yellowstone supercomputer with extended calculations run on Derecho, the National Science Foundation’s newest NCAR supercomputer. The close agreement between the two systems demonstrated that the simulated climate remained stable and reproducible across generations of HPC hardware.
 
This validation step may sound routine, but it highlights a cornerstone of computational science: scientific discoveries increasingly depend not only on powerful models, but also on confidence that those models produce consistent results on evolving supercomputing architectures.

Following freshwater across the planet

The researchers simulated enormous pulses of freshwater entering the northeastern Pacific as the Cordilleran Ice Sheet rapidly melted during the last Ice Age.
 
Rather than remaining confined to the Pacific, the simulations showed that freshwater gradually traveled through the Pacific, Indian, and Southern Oceans before reaching the Atlantic via the Agulhas Leakage, a major ocean gateway near southern Africa.
 
As this freshwater spread through the global ocean, it reduced the salinity of North Atlantic waters.
 
Because salty water is denser than fresh water, this freshening weakened the sinking motion that helps power the Atlantic Meridional Overturning Circulation.
 
In essence, iceberg melt in one ocean basin influenced the stability of another half a world away.
 
The University of California, Davis summarized the finding succinctly: melting icebergs can weaken a massive far-ocean current system by altering the global movement of freshwater rather than acting only where the ice melts.

A new perspective on ancient climate change

For years, many paleoclimate studies emphasized massive iceberg discharges into the North Atlantic, known as Heinrich Events, as the principal trigger for abrupt climate shifts.
 
This study proposes that earlier Pacific “Siku” meltwater events may have preconditioned the Atlantic, making it far more vulnerable when additional meltwater later entered from Europe and North America.
 
The simulations suggest a two-stage process:
  • Pacific ice-sheet melting first weakened the Atlantic circulation through long-distance freshwater transport.
  • Later meltwater entering directly into the North Atlantic amplified that weakening, producing the dramatic climate changes recorded in geological archives.
Rather than viewing the Atlantic in isolation, the work presents Earth’s oceans as a tightly connected planetary system.

Why high-resolution climate modeling matters

None of these conclusions could have been reached through field observations alone.
 
The researchers tracked freshwater using passive dye tracers, monitored evolving ocean salinity, measured changes in water density, and simulated hundreds of years of climate evolution under multiple freshwater-forcing scenarios.
 
Each experiment represented billions of numerical calculations describing fluid dynamics, thermodynamics, atmospheric circulation, and sea-ice interactions.
 
Such simulations require sustained access to leadership-class supercomputing facilities capable of integrating enormous systems of nonlinear equations over centuries of simulated time.
 
Climate science has become one of the defining workloads for modern high-performance computing.

Supercomputers continue to push climate science forward

The study also illustrates another important trend in computational research.
 
Climate models are becoming increasingly detailed.
 
The authors note that future progress will depend upon next-generation eddy-resolving, high-resolution ocean models, which can better represent narrow boundary currents, ocean convection, and turbulent eddies, features that strongly influence freshwater transport and the strength of the AMOC.
 
Those advances will demand even more computational power.
 
As exaflops supercomputers become more widely available, scientists expect to simulate smaller physical processes over larger portions of the globe while incorporating richer observational datasets.
 
Every increase in computing capability expands the realism of Earth’s digital twin.

Looking toward the future

The researchers emphasize that future changes in the AMOC may be driven not only by melting Greenland and Antarctic ice sheets, but also by changes in evaporation, precipitation, river runoff, and salinity transported from distant regions of the world. They argue that improving representations of the global hydrological cycle and ocean circulation will be essential for reducing uncertainty in future climate projections.
 
That perspective aligns with a growing realization across Earth system science: climate cannot be understood as a collection of isolated regional events. Every ocean basin, atmosphere, ice sheet, and continent participates in a deeply interconnected system whose behavior often emerges only through large-scale numerical simulation.

Inspiration through computation

The most compelling takeaway from this research is not merely its scientific findings, but the insight it provides into the transformative power of modern supercomputing. Today’s high-performance systems do far more than process raw data; they serve as time machines that reconstruct lost ice sheets, map invisible ocean currents, and refine our ability to forecast the planet’s future. The same computational infrastructure driving breakthroughs in artificial intelligence and astrophysics is now enabling us to decode the complexities of Earth's climate system. As exaflops computing advances, these models will offer even greater precision in addressing critical challenges like sea-level rise and ocean circulation. Ultimately, this study demonstrates a hopeful synergy: by pairing geological evidence with cutting-edge simulation, we are uncovering the hidden connections of our world and using that knowledge to build a more resilient future.
POPULAR RIGHT NOW
THIS YEAR'S MOST READ
MOST READ OF ALL-TIME