Scientists from the University of New Mexico and Los Alamos National Laboratory presented the THOR artificial intelligence system, capable of solving one of the most difficult problems in statistical physics, xrust reported. What once took weeks of calculations is now done in seconds, opening up new possibilities for science and industry.
What happened and why it is important
Researchers have developed a new computing approach that helps understand how materials behave at the atomic level. This is important for creating:
- strong metals
- new alloys
- materials for electronics
- substances operating at extreme temperatures and pressures
The key feature of the THOR system is the ability to quickly calculate the so-called configuration integrals . In its simplest form, it is a mathematical way of describing how particles in matter interact.
Previously, such calculations were so complex that:
- took weeks or months
- required supercomputers
- gave only approximate results
Now THOR gets the job done in seconds.
Why was this previously considered almost impossible
The main problem is the so-called “curse of dimensionality”. The more particles there are in the system, the faster the computational complexity increases.
In simple words:
- each new particle increases the number of possible interactions
- calculations grow not linearly, but like an avalanche
- even powerful computers could not cope with accurate calculations
According to Professor Dimiter Petsev, traditional methods could require time exceeding the age of the Universe.
Scientists used workaround methods:
- molecular dynamics
- Monte Carlo simulation
But they gave only approximate results.
How THOR works
The THOR system uses a combination of artificial intelligence and a special mathematical technique — tensor networks.
To explain as simply as possible, THOR:
- Divides a huge complex problem into many small parts
- Analyzes them separately
class=»notranslate»>__GTAG5__ Combines the result into an exact solution
Additional system:
- reveals hidden symmetries in materials
- reduces the amount of calculations
- maintains high accuracy
Los Alamos senior researcher Boyan Alexandrov noted:
“Accurate determination of thermodynamic behavior deepens our understanding of physics and opens up new possibilities for metallurgy.”
Test results
Scientists tested THOR on different types of materials:
- copper
- high pressure argon
- solid phase transitions of tin
The results showed:
- coincidence with classical methods
- acceleration of calculations more than 400 times
- maintaining accuracy
- materials science
- solid state physics
- chemistry
- energy
- creation of ultra-strong materials
- modeling of conditions in the interior of planets
- improvement of industrial processes
class=»notranslate»>__GTAG20__
This means that complex scientific calculations can now be performed in near real time.
What does this change for science and technology
class=»notranslate»>__GTAG27__ The emergence of THOR can accelerate the development of several areas at once:
Practical applications include:
- class=»notranslate»>__GTAG10__ development of new batteries
In addition, the system easily integrates with modern machine learning models, which makes it a universal research tool.
Expert opinion
According to the researchers, THOR sets a new standard for computing in physics.
Experts emphasize that the main achievement is not just acceleration, but the ability to directly solve problems that were previously considered practically unattainable.
Telegraph about the innovation
class=»notranslate»>__GTAG35__ The problem of calculating configuration integrals has remained one of the key ones in statistical mechanics for more than 100 years. It underlies the understanding of the properties of substances at the atomic level.
Before THOR, scientists were forced to use approximate methods because direct calculations were considered too complex even for modern supercomputers.
Based on materials from : Official research material: https://www.lanl.gov. University of New Mexico: https://www.unm.edu
Xrust THOR AI solved one of the most difficult physics problems in seconds
- Если Вам понравилась статья, рекомендуем почитать
- Greg Kroah-Hartman tests AI tool "Clanker" to find bugs in the Linux kernel
- Anthropic Announces Breakthrough: AI Mythos Changes Cybersecurity
