It could hardly be more complicated: tiny particles swirl wildly with extremely high energy, countless interactions occur in the entanglement of quantum particles, and this results in a state of matter known as “quark-plasma”. gluon”. Immediately after the Big Bang, the whole universe was in this state; today it is produced by collisions of atomic nuclei at high energy, for example at CERN.
Such processes can only be studied using powerful computers and very complex computer simulations whose results are difficult to evaluate. Therefore, using artificial intelligence or machine learning for this purpose seems like an obvious idea. However, ordinary machine learning algorithms are not suitable for this task. The mathematical properties of particle physics require a very particular structure of neural networks. At TU Wien (Vienna), it has now been shown how neural networks can be successfully used for these difficult tasks in particle physics.
“Simulating a quark-gluon plasma as realistically as possible requires an extremely large amount of computing time,” says Dr. Andreas Ipp from the Institute for Theoretical Physics at TU Wien. “Even the largest supercomputers in the world are overwhelmed by this.” It would therefore be desirable not to precisely calculate every detail, but to recognize and predict certain plasma properties using artificial intelligence.
Therefore, neural networks are used, similar to those used for image recognition: artificial “neurons” are linked together on the computer in the same way as neurons in the brain – and this creates a network that can recognize, for example, whether or not a cat is visible in a certain image.
When applying this technique to quark-gluon plasma, however, there is a serious problem: the quantum fields used to mathematically describe particles and the forces between them can be represented in different ways. “It’s called gauge symmetries,” says Ipp. “The basic principle behind this is something we know: if I calibrate a measuring device differently, for example if I use the Kelvin scale instead of the Celsius scale for my thermometer, I get numbers completely different, even though I’m describing the same physical state. It’s similar with quantum theories – except that here the allowed changes are mathematically much more complicated.” Mathematical objects that look completely different at first glance may actually describe the same physical state.
Gauge symmetries built into the network structure
“If you ignore these gauge symmetries, you cannot meaningfully interpret the results of computer simulations,” says Dr. David I. Müller. “Teaching a neural network to understand these gauge symmetries on its own would be extremely difficult. It is much better to start by designing the structure of the neural network in such a way that the gauge symmetry is automatically taken into account – so that different representations of the same physical state also produce the same signals in the neural network,” says Müller. “That’s exactly what we managed to do: we developed completely new network layers that automatically take into ‘gauge invariance.’ In some test applications, it has been shown that these networks can actually learn much better how to process quark-gluon plasma simulation data.
“With such neural networks, it becomes possible to make predictions about the system – for example, to estimate what the quark-gluon plasma will look like at a later time without really having to calculate every intermediate step in time in detail. .,” says Andreas Ipp. “And at the same time, it’s guaranteed that the system only produces results that don’t contradict gauge symmetry — in other words, results that make sense at least in principle.”
It will be some time before it will be possible to fully simulate the collisions of atomic nuclei at CERN with such methods, but the new type of neural networks provide a completely new and promising tool for describing physical phenomena for which all other calculation methods may never be powerful enough. .
Source of the story:
Material provided by Vienna University of Technology. Note: Content may be edited for style and length.