Connect with us

TOP SCEINCE

Scientists use generative AI to answer complex questions in physics

Published

on

Scientists use generative AI to answer complex questions in physics


When water freezes, it transitions from a liquid phase to a solid phase, resulting in a drastic change in properties like density and volume. Phase transitions in water are so common most of us probably don’t even think about them, but phase transitions in novel materials or complex physical systems are an important area of study.

To fully understand these systems, scientists must be able to recognize phases and detect the transitions between. But how to quantify phase changes in an unknown system is often unclear, especially when data are scarce.

Researchers from MIT and the University of Basel in Switzerland applied generative artificial intelligence models to this problem, developing a new machine-learning framework that can automatically map out phase diagrams for novel physical systems.

Their physics-informed machine-learning approach is more efficient than laborious, manual techniques which rely on theoretical expertise. Importantly, because their approach leverages generative models, it does not require huge, labeled training datasets used in other machine-learning techniques.

Such a framework could help scientists investigate the thermodynamic properties of novel materials or detect entanglement in quantum systems, for instance. Ultimately, this technique could make it possible for scientists to discover unknown phases of matter autonomously.

“If you have a new system with fully unknown properties, how would you choose which observable quantity to study? The hope, at least with data-driven tools, is that you could scan large new systems in an automated way, and it will point you to important changes in the system. This might be a tool in the pipeline of automated scientific discovery of new, exotic properties of phases,” says Frank Schäfer, a postdoc in the Julia Lab in the Computer Science and Artificial Intelligence Laboratory (CSAIL) and co-author of a paper on this approach.

Joining Schäfer on the paper are first author Julian Arnold, a graduate student at the University of Basel; Alan Edelman, applied mathematics professor in the Department of Mathematics and leader of the Julia Lab; and senior author Christoph Bruder, professor in the Department of Physics at the University of Basel. The research is published today in Physical Review Letters.

Detecting phase transitions using AI

While water transitioning to ice might be among the most obvious examples of a phase change, more exotic phase changes, like when a material transitions from being a normal conductor to a superconductor, are of keen interest to scientists.

These transitions can be detected by identifying an “order parameter,” a quantity that is important and expected to change. For instance, water freezes and transitions to a solid phase (ice) when its temperature drops below 0 degrees Celsius. In this case, an appropriate order parameter could be defined in terms of the proportion of water molecules that are part of the crystalline lattice versus those that remain in a disordered state.

In the past, researchers have relied on physics expertise to build phase diagrams manually, drawing on theoretical understanding to know which order parameters are important. Not only is this tedious for complex systems, and perhaps impossible for unknown systems with new behaviors, but it also introduces human bias into the solution.

More recently, researchers have begun using machine learning to build discriminative classifiers that can solve this task by learning to classify a measurement statistic as coming from a particular phase of the physical system, the same way such models classify an image as a cat or dog.

The MIT researchers demonstrated how generative models can be used to solve this classification task much more efficiently, and in a physics-informed manner.

The Julia Programming Language, a popular language for scientific computing that is also used in MIT’s introductory linear algebra classes, offers many tools that make it invaluable for constructing such generative models, Schäfer adds.

Generative models, like those that underlie ChatGPT and Dall-E, typically work by estimating the probability distribution of some data, which they use to generate new data points that fit the distribution (such as new cat images that are similar to existing cat images).

However, when simulations of a physical system using tried-and-true scientific techniques are available, researchers get a model of its probability distribution for free. This distribution describes the measurement statistics of the physical system.

A more knowledgeable model

The MIT team’s insight is that this probability distribution also defines a generative model upon which a classifier can be constructed. They plug the generative model into standard statistical formulas to directly construct a classifier instead of learning it from samples, as was done with discriminative approaches.

“This is a really nice way of incorporating something you know about your physical system deep inside your machine-learning scheme. It goes far beyond just performing feature engineering on your data samples or simple inductive biases,” Schäfer says.

This generative classifier can determine what phase the system is in given some parameter, like temperature or pressure. And because the researchers directly approximate the probability distributions underlying measurements from the physical system, the classifier has system knowledge.

This enables their method to perform better than other machine-learning techniques. And because it can work automatically without the need for extensive training, their approach significantly enhances the computational efficiency of identifying phase transitions.

At the end of the day, similar to how one might ask ChatGPT to solve a math problem, the researchers can ask the generative classifier questions like “does this sample belong to phase I or phase II?” or “was this sample generated at high temperature or low temperature?”

Scientists could also use this approach to solve different binary classification tasks in physical systems, possibly to detect entanglement in quantum systems (Is the state entangled or not?) or determine whether theory A or B is best suited to solve a particular problem. They could also use this approach to better understand and improve large language models like ChatGPT by identifying how certain parameters should be tuned so the chatbot gives the best outputs.

In the future, the researchers also want to study theoretical guarantees regarding how many measurements they would need to effectively detect phase transitions and estimate the amount of computation that would require.

This work was funded, in part, by the Swiss National Science Foundation, the MIT-Switzerland Lockheed Martin Seed Fund, and MIT International Science and Technology Initiatives.



Source link

Continue Reading
Click to comment

Leave a Reply

TOP SCEINCE

Ancient ocean slowdown warns of future climate chaos

Published

on

By

Ancient ocean slowdown warns of future climate chaos


When it comes to the ocean’s response to global warming, we’re not in entirely uncharted waters. A UC Riverside study shows that episodes of extreme heat in Earth’s past caused the exchange of waters from the surface to the deep ocean to decline.

This system has been described as the “global conveyer belt,” because it redistributes heat around the globe through the movement of the ocean waters, making large portions of the planet habitable.

Using tiny, fossilized shells recovered from ancient deep-sea sediments, the study in the Proceedings of the National Academy of Sciences demonstrates how the conveyor belt responded around 50 million years ago. At that time, Earth’s climate resembled conditions predicted by the end of this century, if significant action is not taken to reduce carbon emissions.

Oceans play a crucial role in regulating Earth’s climate. They move warm water from the equator toward the north and south poles, balancing the planet’s temperatures. Without this circulation system, the tropics would be much hotter and the poles much colder. Changes in this system are linked to significant and abrupt climate change.

Furthermore, the oceans serve a critical role in removing anthropogenic carbon dioxide from the atmosphere. “The oceans are by far the largest standing pool of carbon on Earth’s surface today,” said Sandra Kirtland Turner, vice-chair of UCR’s Department of Earth and Planetary Sciences and first author of the study.

“Today, the oceans contain nearly 40,000 billion tons of carbon — more than 40 times the amount of carbon in the atmosphere. Oceans also take up about a quarter of anthropogenic CO2 emissions,” Kirtland Turner said. “If ocean circulation slows, absorption of carbon into the ocean may also slow, amplifying the amount of CO2 that stays in the atmosphere.”

Previous studies have measured changes in ocean circulation in Earth’s more recent geologic past, such as coming out of the last ice age; however, those do not approximate the levels of atmospheric CO2 or warming happening to the planet today. Other studies provide the first evidence that deep ocean circulation, particularly in the North Atlantic, is already starting to slow.

To better predict how ocean circulation responds to greenhouse gas-driven global warming, the research team looked to the early Eocene epoch, between roughly 49 and 53 million years ago. Earth then was much warmer than today, and that high-heat baseline was punctuated by spikes in CO2 and temperature called hyperthermals.

During that period, the deep ocean was up to 12 degrees Celsius warmer than it is today. During the hyperthermals, the oceans warmed an additional 3 degrees Celsius.

“Though the exact cause of the hyperthermal events is debated, and they occurred long before the existence of humans, these hyperthermals are the best analogs we have for future climate change,” Kirtland Turner said.

By analyzing tiny fossil shells from different sea floor locations around the globe, the researchers reconstructed patterns of deep ocean circulation during these hyperthermal events. The shells are from microorganisms called foraminifera, which can be found living throughout the world’s oceans, both on the surface and on the sea floor. They are about the size of a period at the end of a sentence.

“As the creatures are building their shells, they incorporate elements from the oceans, and we can measure the differences in the chemistry of these shells to broadly reconstruct information about ancient ocean temperatures and circulation patterns,” Kirtland Turner said.

The shells themselves are made of calcium carbonate. Oxygen isotopes in the calcium carbonate are indicators of temperatures in the water the organisms grew in, and the amount of ice on the planet at the time.

The researchers also examined carbon isotopes in the shells, which reflect the age of the water where the shells were collected, or how long water has been isolated from the ocean surface. In this way, they can reconstruct patterns of deep ocean water movement.

Foraminifera can’t photosynthesize, but their shells indicate the impact of photosynthesis of other organisms nearby, like phytoplankton. “Photosynthesis occurs in the surface ocean only, so water that has recently been at the surface has a carbon-13 rich signal that is reflected in the shells when that water sinks to the deep ocean,” Kirtland Turner said.

“Conversely, water that has been isolated from the surface for a long time has built up relatively more carbon-12 as the remains of photosynthetic organisms sink and decay. So, older water has relatively more carbon-12 compared to ‘young’ water.”

Scientists often make predictions about ocean circulation today using computer climate models. They use these models to answer the question: ‘how is the ocean going to change as the planet keeps warming?’ This team similarly used models to simulate the ancient ocean’s response to warming. They then used the foraminifera shell analysis to help test results from their climate models.

During the Eocene, there were about 1,000 parts per million (ppm) of carbon dioxide in the atmosphere, which contributed to that era’s high temperatures. Today, the atmosphere holds about 425 ppm.

However, humans emit nearly 37 billion tons of CO2 into the atmosphere each year; if these emission levels continue, similar conditions to the Early Eocene could occur by the end of this century.

Therefore, Kirtland Turner argues it is imperative to make every effort to reduce emissions.

“It’s not an all-or-nothing situation,” she said. “Every incremental bit of change is important when it comes to carbon emissions. Even small reductions of CO2 correlate to less impacts, less loss of life, and less change to the natural world.”



Source link

Continue Reading

TOP SCEINCE

Pacific coast gray whales have gotten 13% shorter in the past 20-30 years, Oregon State study finds

Published

on

By

Pacific coast gray whales have gotten 13% shorter in the past 20-30 years, Oregon State study finds


Gray whales that spend their summers feeding in the shallow waters off the Pacific Northwest coast have undergone a significant decline in body length since around the year 2000, a new Oregon State University study found.

The smaller size could have major consequences for the health and reproductive success of the affected whales, and also raises alarm bells about the state of the food web in which they coexist, researchers say.

“This could be an early warning sign that the abundance of this population is starting to decline, or is not healthy,” said K.C. Bierlich, co-author on the study and an assistant professor at OSU’s Marine Mammal Institute in Newport. “And whales are considered ecosystem sentinels, so if the whale population isn’t doing well, that might say a lot about the environment itself.”

The study, published in Global Change Biology, looked at the Pacific Coast Feeding Group (PCFG), a small subset of about 200 gray whales within the larger Eastern North Pacific (ENP) population of around 14,500. This subgroup stays closer to shore along the Oregon coast, feeding in shallower, warmer waters than the Arctic seas where the bulk of the gray whale population spends most of the year.

Recent studies from OSU have shown that whales in this subgroup are smaller and in overall worse body condition than their ENP counterparts. The current study reveals that they’ve been getting smaller in recent decades.

The Marine Mammal Institute’s Geospatial Ecology of Marine Megafauna (GEMM) Lab has been studying this subgroup of gray whales since 2016, including flying drones over the whales to measure their size. Using images from 2016-2022 of 130 individual whales with known or estimated age, researchers determined that a full-grown gray whale born in 2020 is expected to reach an adult body length that is 1.65 meters (about 5 feet, 5 inches) shorter than a gray whale born prior to 2000. For PCFG gray whales that grow to be 38-41 feet long at full maturity, that accounts for a loss of more than 13% of their total length.

If the same trend were to happen in humans, that would be like the height of the average American woman shrinking from 5 feet, 4 inches to 4 feet, 8 inches tall over the course of 20 years.

“In general, size is critical for animals,” said Enrico Pirotta, lead author on the study and a researcher at the University of St. Andrews in Scotland. “It affects their behavior, their physiology, their life history, and it has cascading effects for the animals and for the community they’re a part of.”

Whale calves that are smaller at weaning age may be unable to cope with the uncertainty that comes with being newly independent, which can affect survival rates, Pirotta said.

For adult gray whales, one of the biggest concerns is reproductive success.

“With them being smaller, there are questions of how effectively these PCFG gray whales can store and allocate energy toward growing and maintaining their health. Importantly, are they able to put enough energy toward reproduction and keep the population growing?” Bierlich said.

Scarring on PCFG whales from boat strikes and fishing gear entanglement also makes the team concerned that smaller body size with lower energy reserves may make the whales less resilient to injuries.

The study also examined the patterns of the ocean environment that likely regulate food availability for these gray whales off the Pacific coast by tracking cycles of “upwelling” and “relaxation” in the ocean. Upwelling sweeps nutrients from deeper to shallower regions, while relaxation periods then allow those nutrients to remain in shallower areas where light allows for growth of plankton and other tiny organisms, including the prey of gray whales.

“Without a balance between upwelling and relaxation, the ecosystem may not be able to produce enough prey to support the large size of these gray whales,” said co-author Leigh Torres, associate professor and director of the GEMM Lab at OSU.

The data show that whale size declined concurrently with changes in the balance between upwelling and relaxation, Pirotta said.

“We haven’t looked specifically at how climate change is affecting these patterns, but in general we know that climate change is affecting the oceanography of the Northeast Pacific through changes in wind patterns and water temperature,” he said. “And these factors and others affect the dynamics of upwelling and relaxation in the area.”

Now that they know the PCFG gray whales’ body size is declining, researchers say they have a lot of new questions about downstream consequences of that decline and the factors that could be contributing to it.

“We’re heading into our ninth field season studying this PCFG subgroup,” Bierlich said. “This is a powerful dataset that allows us to detect changes in body condition each year, so now we’re examining the environmental drivers of those changes.”

The other co-authors on the paper were Lisa Hildebrand, Clara Bird and Alejandro Ajó at OSU and Leslie New at Ursinus College in Pennsylvania.



Source link

Continue Reading

TOP SCEINCE

Western agricultural communities need water conservation strategies to adapt to future shortages

Published

on

By

Western agricultural communities need water conservation strategies to adapt to future shortages


The Western U.S. is heavily reliant on mountain snowpacks and their gradual melt for water storage and supply, and climate change is expected to upend the reliability of this natural process. Many agricultural communities in this part of the country are examining ways to adapt to a future with less water, and new research shows that a focus on supplementing water supply by expanding reservoir capacity won’t be enough to avert future water crises.

Led by scientists at the Desert Research Institute (DRI), the study published June 11 in Earth’s Future. By identifying agricultural communities considered at-risk from looming changes in snowfall and snowmelt patterns, the researchers found that water conservation measures like changes in crop type and extent were more stable adaptive strategies than changes to reservoir capacity. By the end of the century, many areas could have less than half the water they have historically relied on to refill their reservoirs, but changing the types and extent of their crops could help by restoring an average of about 20% of reservoir capacity.

The research team included scientists with the diversity of expertise needed to capture the complexities of water systems while balancing concerns for locally focused adaptation. Beatrice Gordon, lead author of the study and sociohydrologist and postdoctoral researcher at DRI, says the research is needed to inform water management at the local level, where most decisions are made. Gordon herself grew up on a ranch in Wyoming, where she learned firsthand the challenges that face water-insecure communities — an experience that helped lead to her research focus on agriculture and water in the Western U.S.

“A lot of decisions about water are made at the local level, but there’s this big disconnect between that reality and the macro-scale level of most research on this topic,” Gordon says. “We really wanted to understand what the future could look like at the scale that most communities manage their water resources. What are the levers that folks in these communities have when it comes to a future with less snow?”

Mountain snowpacks have historically acted as nature’s water towers across much of the region by storing winter precipitation and releasing it downstream during drier months. Water management systems were designed with this process in mind, but climate change is altering snowmelt patterns in ways that will make it difficult for existing systems to meet the needs of downstream water users. As the world’s largest user of freshwater, irrigated agriculture is at particularly high risk from these changes.

Strategies for addressing water shortages that focus on augmenting supply include expanding reservoirs and replenishing groundwater with surplus water, but these approaches become less effective as the timing and availability of precipitation become more unpredictable. In contrast, water conservation strategies such as reducing total crop acreage, periodic crop fallowing, and shifting toward higher value crops can help manage these risks.

To find out how risk management practices could work on a community-level scale, the researchers built a comprehensive risk assessment framework based on guidance from the Intergovernmental Panel on Climate Change (IPCC). For each of 13 communities, they gathered historical data on irrigation water supply, agricultural water demand, snow storage and snowmelt patterns, and more. They then used projections for the future climate through 2100 to understand how supply and demand dynamics may change in the near future.

“We gathered all these data together and looked at the picture of risk, and then also the ways that adaptation could reduce risk,” Gordon says. “Our goal was really to make this as relevant as possible for the people who are actually making decisions on the ground.”

“Dr. Gordon assembled a very impressive and unprecedented dataset for this paper linking agricultural water supply and demand across the Western United States,” says study co-author Gabrielle Boisramé, assistant research professor at DRI.

The Western agricultural communities the researchers selected are located in headwaters areas, making them both subject to significant changes in future climate and sentinels for the future of the West. Several of them are located in the Upper Colorado River Basin, which feeds into the main stem of the river — a water system that supports more than 40 million people.

“A lot of these areas are providing downstream water to other communities,” Gordon says. “So, if they have an increase in demand and a decrease in supply, it impacts not only that area, but also the areas that rely on that water downstream.”

The study results show that there will be a stark decline in how much many of these communities will be able to refill their reservoirs in just a few decades, with some seeing declines to about half of the water they were historically able to store. A drop that significant is particularly acute in many of the smaller reservoirs that can only hold about a year’s worth of water.

“It shows how important it is to dedicate effort — now, not in 20 to 50 years — to figuring out how we, as scientists, can provide better information about water conservation,” Gordon says. “And I think that there’s an opportunity to really think about how we support communities in these efforts, especially small communities in headwaters regions that might be fully dependent on agriculture.”

“Our results indicate the importance of water conservation as an adaptive strategy in a warmer future with less snow,” she continues. “And that’s broadly true across a lot of different places in the Western U.S.”



Source link

Continue Reading

Trending

Copyright © 2017 Zox News Theme. Theme by MVP Themes, powered by WordPress.