Scientists are rethinking the universe’s deepest mysteries using numerical relativity, complex computer simulations of Einstein’s equations in extreme conditions. This method could help explore what happened before the Big Bang, test theories of cosmic inflation, investigate multiverse collisions, and even model cyclic universes that endlessly bounce through creation and destruction.
With the help of innovative large-scale simulations on various supercomputers, physicists at Johannes Gutenberg University Mainz (JGU) have succeeded in gaining new insights into previously elusive aspects of the physics of strong interaction.
Associate Professor Dr. Georg von Hippel and Dr. Konstantin Ottnad from the Institute of Nuclear Physics and the PRISMA+ Cluster of Excellence have calculated the interaction of the pion with the Higgs field with unprecedented precision based on quantum chromodynamics. Their findings were recently published in Physical Review Letters.
Mitsui & Co. has formally launched a new quantum-enabled chemistry platform, QIDO, in collaboration with U.S.-based Quantinuum and QSimulate. The system, designed to accelerate the discovery of new materials and pharmaceuticals, blends classical and quantum computing resources to streamline complex chemical calculation, according to a story in Nikkei and a Quantinuum blog post.
Quantum computers hold promise for modeling chemical reactions beyond the reach of traditional supercomputers. But fully fault-tolerant systems remain years away, leaving companies searching for ways to extract value from today’s noisy, early-stage machines. QIDO, short for Quantum-Integrated Discovery Orchestrator, attempts to bridge that gap.
The platform runs most computations on powerful classical hardware while sending only the most computationally expensive steps — such as the modeling of strongly correlated electrons — to a quantum computer. This hybrid workflow allows companies to perform higher-precision chemical simulations today, without waiting for fully mature quantum systems, Nikkei reports.
Practical fusion power that can provide cheap, clean energy could be a step closer thanks to artificial intelligence. Scientists at Lawrence Livermore National Laboratory have developed a deep learning model that accurately predicted the results of a nuclear fusion experiment conducted in 2022. Accurate predictions can help speed up the design of new experiments and accelerate the quest for this virtually limitless energy source.
In a paper published in Science, researchers describe how their AI model predicted with a probability of 74% that ignition was the likely outcome of a small 2022 fusion experiment at the National Ignition Facility (NIF). This is a significant advance as the model was able to cover more parameters with greater precision than traditional supercomputers.
Currently, nuclear power comes from nuclear fission, which generates energy by splitting atoms. However, it can produce radioactive waste that remains dangerous for thousands of years. Fusion generates energy by fusing atoms, similar to what happens inside the sun. The process is safer and does not produce any long-term radioactive waste. While it is a promising energy source, it is still a long way from being a viable commercial technology.
There are high hopes for quantum computers: they are supposed to perform specific calculations much faster than current supercomputers and, therefore, solve scientific and practical problems that are insurmountable for ordinary computers. The centerpiece of a quantum computer is the quantum bit, qubit for short, which can be realized in different ways—for instance, using the energy levels of atoms or the spins of electrons.
When making such qubits, however, researchers face a dilemma. On the one hand, a qubit needs to be isolated from its environment as much as possible. Otherwise, its quantum superpositions decay in a short time and the quantum calculations are disturbed. On the other hand, one would like to drive qubits as fast as possible in analogy with the clocking of classical bits, which requires a strong interaction with the environment.
Normally, these two conditions cannot be fulfilled at the same time, as a higher driving speed automatically entails a faster decay of the superpositions and, therefore, a shorter coherence time.
What this means in real time is that researchers using these maps do not know if there are any errors or issues ahead of them, nor do they know if these errors are part of the research design. Nevertheless, this is all they have to work with, so they have to make a decision based on this limited information, and doing so will always have high costs in terms of the ignition attempt, which is expensive.
To overcome this, the team at the NIF created a new way to create these “maps” by merging past data with high-fidelity physics simulations and the knowledge of experts. This was then fed into a supercomputer that ran statistical assessments in the course of over 30 million CPU hours. Effectively, this allows the researchers to see all the ways that things can go wrong and to pre-emptively assess their experimental designs. This saves a lot of time and, more importantly, money.
The team tested this approach on an experiment they ran in 2022, and, after a few changes to the model’s physics, was able to predict the outcome with an accuracy above 70 percent.
Questions to inspire discussion.
🚗 Q: How will AI6 be used in Tesla vehicles? A: AI6 will be used for FSD inference, with two chips in every car, enabling advanced autonomous driving capabilities.
🤖 Q: What role will AI6 play in Optimus? A: AI6 will enable on-device learning and reinforced learning in Optimus, enhancing its AI capabilities.
🔋 Q: Will AI6 be used in other Tesla products? A: AI6 will be integrated into every edge device produced by Tesla, including Tesla Semi, Mega Pack, and security cameras.
Technical Specifications.
💻 Q: What is the architecture of AI6? A: AI6 will use a cluster model of individual chips with a software layer on top, similar to Dojo 3 for training.
Scientists have developed a new machine learning approach that accurately predicted critical and difficult-to-compute properties of molten salts, materials with diverse nuclear energy applications.
In a Chemical Science article, Oak Ridge National Laboratory researchers demonstrated the ability to rapidly model salts in liquid and solid state with quantum chemical accuracy. Specifically, they looked at thermodynamic properties, which control how molten salts function in high-temperature applications. These applications include dissolving nuclear fuels and improving reliability of long-term reactor operations. The AI-enabled approach was made possible by ORNL’s supercomputer Summit.
“The exciting part is the simplicity of the approach,” said ORNL’s Luke Gibson. “In fewer steps than existing approaches, machine learning gets us to higher accuracy at a faster rate.”
Scientists at Lawrence Livermore National Laboratory (LLNL) have helped develop an advanced, real-time tsunami forecasting system—powered by El Capitan, the world’s fastest supercomputer—that could dramatically improve early warning capabilities for coastal communities near earthquake zones.
The exascale El Capitan, which has a theoretical peak performance of 2.79 quintillion calculations per second, was developed at the National Nuclear Security Administration (NNSA). As described in a preprint paper selected as a finalist for the 2025 ACM Gordon Bell Prize, researchers at LLNL harnessed the machine’s full computing power in a one-time, offline precomputation step, prior to the system’s transition to classified national-security work. The goal: to generate an immense library of physics-based simulations, linking earthquake-induced seafloor motion to resulting tsunami waves.
The paper is published on the arXiv preprint server.
