It’s possible that I shall make an ass of myself. But in that case one can always get out of it with a little dialectic. I have, of course, so worded my proposition as to be right either way (K.Marx, Letter to F.Engels on the Indian Mutiny)
Wits researchers use light’s topology to beat atmospheric distortion
World-first experiment shows information encoded in a beam of light can survive real-world atmospheric turbulence, opening new possibilities for long-distance optical and quantum communication.
Researchers at the University of the Witwatersrand (Wits) and the University of Bordeaux in France have demonstrated a new way of sending information through the atmosphere without having to correct for the distortions that normally disrupt optical communication.
In a world-first experiment conducted across the Wits West Campus in Johannesburg, the researchers showed that information encoded in a special property of light known as topology remains intact even when the laser beam carrying it is strongly distorted by atmospheric turbulence.
The findings, published in Science Advances, could help pave the way for more reliable long-distance optical communication, including links to satellites and spacecraft and, potentially, improved connectivity in remote and underserved areas.
Information that survives distortion
Modern communication systems use different properties of light, including colour, intensity and polarisation, to carry information. Fibre-optic cables protect these signals from many outside disturbances, but sending light through open air is more difficult.
Changes in air temperature and pressure create atmospheric turbulence, which can distort a laser beam as it travels. This can damage the information it carries.
The Wits-led research takes a different approach by encoding information in topology — a mathematical idea describing properties that remain unchanged even when an object is stretched or deformed.
“To explain its benefits, we can liken it to how a coffee mug can be reshaped into the form of a doughnut,” says Prof Andrew Forbes, Head of the Structured Light Lab in the Wits School of Physics.
“Despite their very different shapes, both have a single hole. You can stretch or distort them without changing that fundamental property. In the same way, the light beam can become badly distorted while its topological information remains unchanged.”
The researchers used an optical structure known as a skyrmion to encode this information into laser beams.
They then transmitted the beams between two buildings on the Wits campus over a distance of hundreds of metres, exposing them to naturally occurring atmospheric turbulence.
Although the shape of the light arriving at the receiver was significantly different from the transmitted beam, the information contained in its topology remained intact.
“Small laboratory experiments have previously suggested that topology could provide a robust way of carrying information,” says lead author Cade Peters. “This is the first time we have demonstrated that robustness across a real-world optical link and under naturally occurring atmospheric conditions.”
No need to correct the beam
Conventional free-space optical communication systems often need to measure atmospheric distortion and then compensate for it using specialised hardware and complex calculations.
The new approach does not require the distortion to be measured or corrected before the information can be recovered.
This could make future systems simpler and reduce the technical and computing demands involved in transmitting information through difficult environments.
“The ability to encode and transmit information using light has transformed the way the world communicates,” says Prof Mitchel Cox of the Optical Communications Laboratory in the Wits School of Electrical and Information Engineering.
“As our demand for data continues to grow, we need new ways to increase the capacity, reliability and performance of communication systems. This work shows that topology is a largely untapped resource that could contribute to the next generation of optical communications.”
The researchers also explored applications for both conventional and quantum communication, where maintaining the integrity of information is particularly important.
For South Africa, Peters says the research could eventually have wider social benefits.
“South Africa still faces major challenges in providing reliable access to information and communication technologies, particularly in remote and underserved communities,” he says.
“By developing new approaches to long-range optical communication, we hope this work can contribute to the broader effort to bridge the digital divide.”
Journal
Science Advances
Even protected Florida wetlands can’t escape microplastic pollution
Credit: Erik Johanson, Florida Atlantic University
Microplastics are turning up almost everywhere – from oceans and rivers to soils and sediments – yet far less is known about what happens when they enter freshwater wetlands. Often overlooked in favor of marine and coastal environments, wetlands can act as both pathways and traps for microplastics, potentially capturing and retaining pollution.
That knowledge gap is particularly important in small, urban-adjacent wetlands throughout Florida, where stormwater runoff, recreational activity and connections to rivers and other waterways can introduce microplastics. Yet little is known about how these particles move and accumulate within these ecosystems.
To help fill these gaps, Florida Atlantic University researchers examined microplastics at Spruce Bluff Preserve, a 97-acre freshwater wetland along the St. Lucie River in Port St. Lucie. The study provides the first detailed look at how microplastics are distributed across the preserve and how vegetation, sediment composition and proximity to water influence where particles accumulate.
Researchers collected 40 surface sediment samples from eight transects spanning different ecological zones. Samples were collected every 5 meters in December 2024 during the dry season, and researchers recorded each site’s distance from water. They also used infrared spectroscopy to confirm a subset of particles, providing a more accurate estimate of microplastic abundance.
Results of the study, published in the journal Evolving Earth, reveal that microplastics were found throughout Spruce Bluff Preserve, and detected in 38 of 40 sediment samples (95%) across every ecological zone. Most of the confirmed particles were tiny – about 88% were smaller than 2.5 millimeters – and nearly half were fibers, suggesting that synthetic textiles and other everyday materials may be contributing to plastic pollution in the preserve.
But the particles were not distributed randomly. The highest concentrations were found along the St. Lucie River and an outlet canal used for flood control, suggesting that water movement is helping transport and concentrate microplastics within the wetland.
“Perhaps the most striking finding is that the microplastics weren’t simply scattered randomly across the preserve – they appeared to follow the pathways of water,” said Erik N. Johanson, Ph.D., senior author and an associate professor in the Department of Geosciences within FAU’s Charles E. Schmidt College of Science. “The river edge and flood-control canal had some of the highest concentrations, suggesting that these areas can act as conduits and collection points for plastic moving through the landscape. It shows that even in a protected wetland, what happens upstream can ultimately show up in the sediment.”
The findings indicate that water-connected areas may act as pathways and collection points for microplastics entering the preserve through runoff and other waterborne sources, while interior areas farther from active water flow generally had lower concentrations.
“The way water moves through the landscape may be more important in determining where microplastics accumulate than simply how close a location is to water,” Johanson said.
The types of particles also offer clues about their origins. Fibers were the most common type, consistent with inputs from synthetic textiles and urban runoff, while the predominance of small particles suggests that plastics are breaking down and weathering over time. The mix of particle types and colors points to multiple, diffuse sources rather than a single source of contamination.
The findings underscore that protecting a wetland from development does not necessarily protect it from pollution. Spruce Bluff Preserve reflects a broader Florida landscape in which natural wetlands are interconnected with rivers, stormwater systems, flood-control canals and other engineered waterways.
These systems can act as both filters and conduits, trapping microplastics in sediments while also transporting them downstream.
The researchers say the findings point to the need to consider plastic pollution alongside the nutrients, sediment and other contaminants already monitored in Florida’s wetlands and stormwater systems. Incorporating microplastic monitoring into existing environmental management programs could help identify where plastics are entering waterways and where they are accumulating.
“Wetlands are often viewed as places that protect us by filtering what moves through the landscape, but they can also capture and redistribute pollutants,” said Johanson. “If we want to protect these ecosystems for the long term, we need to understand not only what is entering them, but how water moves that pollution through the system. A protected wetland is not an isolated wetland – it is connected to everything upstream and downstream through the movement of water.”
Study co-authors are Juana Baudrix, who earned her master’s degree at FAU in geosciences; and Julie Buchanich, a Ph.D. student in FAU’s Department of Geosciences.
Florida Atlantic University is one of the nation’s fastest-rising public research universities, serving more than 32,000 students in South Florida. Ranked among the Top 100 Public Universities by U.S. News & World Report, recognized as a Top 25 Best-In-Class College, and cited by Washington Monthly as one of the nation’s most effective engines of upward mobility, Florida Atlantic is also one of only 13 institutions nationwide to hold Carnegie Foundation designations for R1 research, opportunity and community engagement. Guided by its strategic plan, “2031FAU: Where Tomorrow Begins,” the university is focused on delivering career-ready education and experiential learning, driving scholarly inquiry that creates healthier, safer and more prosperous communities, strengthening institutional excellence, and elevating its impact across South Florida and beyond. Florida Atlantic continues to advance its position as Florida’s first quantum university, integrating research, education and industry partnerships around next-generation computing technologies. Through other signature strengths in neuroscience and healthy aging, environmental, ocean and coastal innovation, and national defense and autonomous systems, Florida Atlantic expands knowledge, fuels economic opportunity and fulfills its mission as South Florida’s hometown university.
Copper market is on fire. Altonorte plant in Chile. Credit: Glencore
Copper hit a record in New York and closed in on all-time highs in London on Tuesday, as the threat of US import tariffs keeps pulling metal into American warehouses and turns what was supposed to be a comfortable global surplus into scarcity everywhere else.
Comex copper for September delivery rose as much as 1.8% to $6.7270 a pound (about $14,830 a tonne), topping the previous record of $6.7140 set on August 12. The contract was last up 1.6% at $6.7125, some $550 a tonne above the London price, a premium of close to 4%.
On the LME, the three-month contract added 0.4% to $14,251 a tonne after finishing Monday at $14,201, its highest ever close, and traded as high as $14,343 during the session, within 1.3% of January’s all-time peak of $14,527.50.
Click on chart for live prices.
Copper had touched $14,396 a tonne on August 17, its highest since late January, BMI noted on Friday, before a rush of returning metal knocked it back towards $14,000. Stocks on warrant in LME warehouses jumped 74.5% in a week, including the largest daily inflow since 2024, and the cash premium over three-month metal collapsed to $248 from the five-year high of $434 reached during this month’s squeeze.
The reprieve barely lasted a week. Orders to withdraw 51,400 tonnes hit the warehouse system on Monday, part of some 65,400 tonnes earmarked for departure in recent days, as metal resumed its march across the Atlantic to Comex, where inventories have risen for 46 straight days to a record 675,185 tonnes.
‘A deficit market in reality’
The US imported 885,000 tonnes of refined copper in the first half, up 3% from a year earlier and on pace to approach 2025’s record 1.64 million tonnes, as traders position for another round of tariff roulette: duties on refined copper of 15% from January 2027, stepping up to 30% from 2028, remain on the table in Washington.
CRU projected a 639,000-tonne global surplus for 2026 but now regards the market as at best balanced. “If imports keep coming in as they have been, then it’s going to look like a deficit market in reality,” principal copper analyst Robert Edwards told Reuters.
“Based on our numbers, you’re looking at years for that metal to get consumed,” Macquarie strategist Alice Fox said of the record Comex stockpile. Bank of China International’s Amelia Fu expects “new record highs in copper prices in coming weeks or months”, though Glencore chief executive Gary Nagle has argued a tariff announcement, whichever way it falls, would take the heat out of prices simply by ending the uncertainty.
Wall Street is struggling to keep up, with the consensus price target near $166, more than 20% below the market, and CICC, the Beijing-based state-backed investment bank, cutting the stock to market perform on valuation late last week even after record second-quarter sales of $4.3 billion and a dividend hike.
The rally has redrawn the top of the mining league table: at about $183 billion, Southern Copper is now worth more than Rio Tinto, at just under $180 billion, and trails only BHP’s $246 billion, despite Rio generating nearly four times the revenue and more than twice the profit over the past year. In MINING.COM’s TOP 50 ranking, where the stock stands in for parent Grupo Mexico, that would mark the copper producer’s first appearance at no. 2.
Freeport has fared even better over the week, adding almost 19% to trade within cents of its own all-time high, while First Quantum, Ivanhoe Mines and Teck Resources are all up about 11%.
Zijin Mining is the laggard, falling 2.5% in New York on Tuesday after the Chinese group warned that its 1.2 million tonne mined copper target is under pressure, with flooding at the Kamoa-Kakula mine in Congo expected to cut its attributable output by up to 57,000 tonnes this year.
McGill University researchers have developed a more energy-efficient method of building AI systems that are better at measuring – and indicating – their own uncertainty. This will help users determine when human oversight is needed, when additional data should be collected and when a model is being asked to work beyond the conditions it was trained for, the researchers said.
The research was supervised by David A. Stephens, Professor in the Department of Mathematics and Statistics.
More reliable estimates
Standard neural networks learn patterns from data and make predictions, but they typically provide a single answer without clearly indicating how confident they are in that response. Bayesian neural networks address this limitation by representing their internal settings as probabilities rather than fixed values, enabling them to estimate uncertainty, particularly when faced with unfamiliar data. However, this capability often requires significant computational and memory resources, making these networks difficult to deploy at the scale of modern AI systems.
The researchers found a way to make Bayesian neural networks substantially more efficient while maintaining strong predictive performance. In one experiment, their approach used about 33 times fewer parameters than a commonly used method for estimating uncertainty in AI systems.
The researchers are now exploring ways to automate the process of identifying which parts of a neural network are most important for a given task. This could help the approach work more effectively across different kinds of data and AI tools, they said.
NB: Machine learning research follows a different publication model from many other scientific fields. The top peer-reviewed conferences are often the main archival publication venues, rather than a preliminary step before journal submission.
ICML papers are reviewed through a double-blind peer-review process, and accepted papers are published through the Proceedings of Machine Learning Research.
Credit: Mochammad Ariyanto et al., 2026, Biohybrid Navigation through Real-Time Terrain Recognition and Natural Climbing in Cyborg Insect, Device
Osaka, Japan - Cyborg insects combine the mobility of living organisms with miniature electronic devices, offering potential applications in search-and-rescue operations, infrastructure inspection, and exploration of environments that are difficult for conventional robots. However, most autonomous navigation systems are designed to avoid obstacles, even when an insect could naturally climb over them. This can result in longer routes and reduced exploration efficiency.
An international research team from the University of Osaka and Universitas Diponegoro has developed a new navigation system for cyborg insects that combines the cockroach’s natural climbing ability with AI-based real-time terrain recognition, enabling the insects to traverse obstacles faster and more efficiently than with conventional navigation methods.
Conventional systems direct cyborg insects around obstacles, even when the insects can climb over them. The team therefore developed a reactive-climbing strategy combining goal-seeking, obstacle avoidance, wall-following, and innate climbing behavior. However, because the controller could not identify terrain, it issued steering commands during climbing, causing hesitation and inefficient movement.
To address this problem, the researchers incorporated a multilayer-perceptron-based AI module that used onboard sensor data to recognize flat surfaces, ascents, descents, and holes in real time. The classifier achieved 92% accuracy in offline evaluation, enabling the controller to adjust stimulation according to the terrain, reduced unnecessary steering, and support sustained forward movement across challenging surfaces.
“The main challenge was to develop a system capable of recognizing terrain in real time without compromising the insect’s natural locomotor abilities,” explains Professor Keisuke Morishima of the University of Osaka. “In this study, we propose ‘biohybrid physical AI’ to enable efficient autonomous navigation. We hope these findings will inspire the development of robotic systems capable of operating in complex environments, including search-and-rescue sites.”
###
The article, “Biohybrid Navigation through Real-Time Terrain Recognition and Natural Climbing in Cyborg Insect,” will be published in Device at DOI: https://doi.org/10.1016/j.device.2026.101277
About The University of Osaka
The University of Osaka was founded in 1931 as one of the seven imperial universities of Japan and is now one of Japan's leading comprehensive universities with a broad disciplinary spectrum. This strength is coupled with a singular drive for innovation that extends throughout the scientific process, from fundamental research to the creation of applied technology with positive economic impacts. Its commitment to innovation has been recognized in Japan and around the world. Now, The University of Osaka is leveraging its role as a Designated National University Corporation selected by the Ministry of Education, Culture, Sports, Science and Technology to contribute to innovation for human welfare, sustainable development of society, and social transformation.
Biohybrid Navigation through Real-Time Terrain Recognition and Natural Climbing in Cyborg Insect
Article Publication Date
20-Aug-2026
Advanced locomotion control strategy based on terrain recognition. (A) MLP-based terrain classifier for identifying flat ground, uphill, downhill, and holes. (B) Adaptive stimulation strategy that adjusts locomotion control based on the recognized terrain to enhance locomotion and climbing efficiency.
Credit
Mochammad Ariyanto et al., 2026, Biohybrid Navigation through Real-Time Terrain Recognition and Natural Climbing in Cyborg Insect, Device
Climbing cyborg insect
Credit
Mochammad Ariyanto et al., 2026, Biohybrid Navigation through Real-Time Terrain Recognition and Natural Climbing in Cyborg Insect, Device
Monell researchers use machine learning to begin mapping complex scents
PHILADELPHIA, PA – Aug. 20, 2026 – If you want to describe a particular color, you could look to the Pantone color wheel to find its exact hue, saturation and brightness, and how it compares to other colors. But nothing like that has existed for complex odors.
Now, research co-authored by scientists at the Monell Chemical Senses Center has gotten closer to that work, creating a means of using machine learning to discriminate among scents and how they relate to each other. A description of the work was published online Aug. 4 in the Proceedings of the National Academy of Sciences.
The work provides a validated metric and benchmark for comparing smells, laying a foundation for technologies such as digital olfaction, the ability to digitize scents, said study co-author Joel Mainland, Ph.D., a Member of the Monell Center.
“We’ve been interested for a long time in trying to digitize odors, to mathematically represent these in some way similar to what we have done with color vision and for hearing,” Mainland said.
In 2015, IBM ran an open investigator DREAM challenge to take a single odor molecule and predict what it smells like based on its chemical structure, he said. That drove a lot of science. But most odors we encounter in day-to-day life are complex mixtures of dozens or hundreds of molecules. “We have to understand how mixtures work if we want to digitize anything,” Mainland said.
Being able to quantitatively map odor mixtures has numerous applications, he continued. For example, some conditions like diabetes and liver failure have olfactory signatures, distinct scents that could be helpful in diagnosis. Mapping scents also could be helpful in quantifying flavors of foods, or trademarking particular aromas like the scents of brand-name laundry detergents. “The companies that make smells are doing a lot of trial and error, so the thought process is that if you could fix that part where it’s more mathematical, they could make products more efficiently,” he said.
Mainland and colleagues released their own DREAM challenge, inviting international teams to use machine learning to develop models to predict the amount of similarity between two scent mixtures. First, they standardized and compiled six datasets of odor-similarity measurements from three different studies into a new dataset, comprising 168 unique single molecules, 731 unique mixtures, and 507 mixture-pair measurements. The mixture pair distances were mapped onto a continuous perceptual scale from 0 (indistinguishable) to 1 (most distinct).
Next, over a three-month period, 26 teams competed to predict how similar the paired scents would be on a hidden test-set of 46 mixture pairs. The competition resulted in a four-way tie. Mainland and colleagues then built an ensemble model by averaging the predictions from the four winning teams with those from two additional high-performing models. Following the challenge, they validated the model on an independent set of 50 olfactory mixture pairs.
Their final model was found to be quite accurate. It achieved a median RMSE (root mean squared error) of 0.08, with 0 being a perfect score, meaning the model had good accuracy in predicting scent similarities. It also achieved a Pearson correlation of 0.57 on the test set, meaning the model had a moderate to strong positive ability to predict the characteristics of the dataset.
While many in the sensory science field believed that using science to predict similarity among mixtures was going to be much more difficult than in single molecules, that didn’t turn out to be the case, Mainland said: “This paper shows that if you’re already able to predict what a single component smells like, then you can make a pretty good prediction of mixtures out of that.”
Interestingly, he said, the machine learning models were more likely to use semantic language like “fruity and sweet” to describe scent mixtures rather than chemistry terms like ester or molecular weights. “Working with these semantic labels is a huge jump forward in being able to make predictions of what mixtures smell like,” he said. When a journal reviewer asked them to eliminate the semantic features to see how the models performed without that information, it was much more difficult to make predictions.
Investigators are now writing up results from a third DREAM Olfaction challenge to submit for journal publication. In that exercise, teams were given people’s impressions of what individual components of mixtures smelled like and were asked to use that information to predict how the complete mixture would smell.
“What we’re seeing is that, again, if you know what the components smell like, you know what the mixture smells like — it’s basically just an average of the components, which I think is really surprising to a lot of people in the field,” he said.
Mainland serves on the scientific advisory board of Osmo Labs, PBC, and receives compensation for these activities.
# # #
About the Monell Chemical Senses Center
The Monell Chemical Senses Center is an independent nonprofit research institute in Philadelphia, Pennsylvania. It was founded in 1968 to advance and share discoveries in the science of the chemical senses of smell, taste, chemesthesis, and interoception to solve the world’s health, societal, and environmental challenges.
In two Perspectives, authors highlight the strengths and limitations of “self-driving” laboratories, spaces that combine robotics, high-throughput experiments, artificial intelligence (AI), and automated analysis to transform scientific discovery. They were envisioned to help discovery go from a slow process of trial and error into a continuous, adaptive learning cycle. These systems are already being used to discover molecules and optimize materials, sometimes exploring enormous experimental spaces while using far less material and time than conventional approaches.
In one Perspective, Milad Abolhasani highlights a major challenge in using AI tools in laboratories. These tools can generate hypotheses far faster than physical laboratories can test them. Linking instruments, robots, software, and shared datasets could allow laboratories to learn from one another, uncover patterns hidden in complex systems, and reduce redundant experiments. At the same time, greater autonomy will require rigorous safety measures, transparent decision-making, standardized data, and broader access to prevent these technologies from becoming concentrated among a small number of institutions. Ultimately, Abolhasani envisions autonomous laboratories augmenting scientists rather than replacing them, with human researchers setting goals and interpreting evidence while intelligent systems handle much of the experimental exploration.
In another Perspective, Martin Burke and colleagues highlight “blocc” chemistry. Blocc chemistry is a modular approach to building small molecules from standardized chemical building blocks that could make organic synthesis faster, more automated, and accessible to nonspecialists. By enabling robots to repeatedly assemble carbon-carbon bonds, Burke suggests that the method could be used to generate large, standardized datasets that AI systems could use to predict and optimize the properties of new molecules, creating a feedback loop between automated synthesis, testing, and machine learning. According to the author, the approach has already produced promising materials for applications including organic electronics and solar cells. What’s more, blocc chemistry could also democratize molecular discovery and transform chemistry education. However, broader access will require standardized methods as well as safeguards to ensure that automated chemical innovation is used safely and responsibly.
The exchange-correlation (XC) energy density for a benzene molecule was calculated using the Skala model. It contributes to the system’s total energy density. The inner ring shows the molecule’s six carbon atoms, while the smaller outer ring shows the six hydrogen atoms. The highest values of the XC energy density are found in a ring-shaped pattern around the individual atoms of the molecule, particularly around the carbon atoms.
The Skala AI model, developed by Microsoft Research AI for Science, is now available through the CP2K software ecosystem. Thanks to Skala, researchers can perform quantum mechanical simulations of larger molecular systems while achieving a higher level of accuracy. In early 2026, Microsoft’s research division and the CP2K team at the Center for Advanced Systems Understanding (CASUS) at the Helmholtz-Zentrum Dresden-Rossendorf started a collaboration to integrate the Skala AI model into CP2K. The global CP2K community is expected to test the AI model’s applicability to a wide variety of problems in chemistry, physics, materials science, and engineering.
For many simulation applications, density functional theory (DFT)—which was awarded the Nobel Prize in 1998—has proven to be crucial. “The Achilles’ heel of DFT is the so-called exchange-correlation functional,” says CASUS Director Prof. Thomas D. Kühne. In science, functionals refer to the search for the function best suited to a given problem. Although the exchange-correlation functionals developed for DFT in recent years have become increasingly sophisticated, the most accurate functionals require so much computation time that they can only be used for systems with a small number of particles. Skala is a new exchange-correlation functional developed by Microsoft Research AI for Science. Instead of introducing yet another layer of mathematical formulas, Microsoft took a new approach with Skala: Here, a neural network has learned how electron densities in different regions influence one another. This makes Skala one of the first AI-based exchange-correlation functionals available.
“The results presented by Microsoft Research in 2025 were impressive, particularly the combination of accuracy and computational efficiency for certain DFT calculations,” reports Kühne. "We saw an opportunity to evaluate the approach within CP2K and better understand its applicability to problems relevant to our community. Of course we’re excited to be the first group outside Microsoft that has confirmed the benefits of Skala.” In mid-August, scientists from CASUS and Microsoft Research presented the first results of the collaboration. They demonstrate that the approach taken is indeed promising. “I can confirm that with Skala we’ve achieved a noticeable leap in the accuracy of our simulations for our specific test case,” says lead author Franz Pöschel of CASUS’ Scientific Computing Core.
High level of interest in response to announcement
Microsoft Research AI for Science had announced the upcoming integration of Skala into CP2K in the spring of 2026 at a conference. Since then, the community has been regularly checking in on the progress. “Even though we were pleased by the interest, we felt a certain pressure to deliver,” says Pöschel. “I’m therefore glad to confirm that Skala is now available within CP2K for simulations of molecular systems.”
Dr. Sebastian Ehlert, Senior Researcher at Microsoft Research AI for Science, explains why CP2K is so meaningful to the Skala team: “To increase adoption, Skala should be available where the community already works. CP2K has been a cornerstone of computational chemistry research for many years, making it a natural priority for integration.”
The CP2K software ecosystem is a powerful tool for performing DFT calculations, particularly in the context of dynamic simulations of large systems over long time periods. On the open-source platform, molecules, liquids, solids, and biological systems can be simulated using quantum mechanical methods such as density functional theory and classical molecular dynamics. Thanks to its efficient algorithms and the ability to execute individual computational steps in parallel on specialized computer architectures, CP2K can compute even very large systems with thousands to tens of thousands of atoms. This makes the software particularly well-suited for research into battery materials, catalysts, semiconductors, proteins, and other complex materials. Recently, CP2K was enhanced with features that enable the use of AI-based models. Using training data generated by CP2K, these models can predict molecular energies and forces with high accuracy. This expands the time and length scales of the simulations.
More updates in the works
Prior to release, the teams at Microsoft Research AI for Science and CASUS extensively tested the Skala integration in CP2K. “We want to ensure that Skala delivers consistent accuracy and speed across different programs and settings,” says Ehlert. “Together with the CP2K team, we created a set of integration tests to ensure that Skala delivers numerically correct results. We are especially grateful for the support of the CASUS team with their in-depth experience on numerical verification of computational methods to design this suite of integration tests.”
Skala is not a static approximation: The AI model is constantly being improved and expanded with new capabilities. It will soon enable simulations of periodic solids—such as metals and semiconductors—as well as liquids. CP2K can serve as a platform for rapid and widespread adoption: the recent integration will make it easier to have the latest Skala release in CP2K. “Microsoft Research AI for Science puts high standards on its research partners,” explains Kühne. “We are confident that, following this initial success, our collaboration will lead to many more achievements.”
