Russian artificial intelligence helped to discover the elusive particle at the Large hadron Collider

Neural network, developed by Russian specialists, while processing the data with the largest accelerator in the world did what was considered impossible. With it was opened, the particle that physicists have never been experimentally and was hoping to fix on the TANK.

Achievement is described in a scientific article submitted to the journal Physical Review Letters. Until then, you can read about their Preprint.

we are Talking about omega-baryons in the excited state. Omega baryon consists of one lovely (beauty, or b) quark and two strange (strange, or s) quarks. So the new particle, we can say, joined strangeness and charm. Its energy is more than the minimum possible for the omega-baryon, so physics and talking about the excited state.

Recall that the Large hadron Collider pushes between the protons accelerated to nearly the speed of light. In these collisions are born the different particles. Those that are born much more often, has been studied thoroughly. So interesting to physicists, the particles that are formed rarely.

What in this case, “rarely” and “often”? For every ten billion are running towards each other protons collide a few particles. It should not be surprising, because to get a proton in flying towards him proton – a task more difficult than to shoot down a bullet with a bullet counter. Most likely, the “shells” just raskinetsya.

fortunately, physicists run into the tunnel as many protons to watch tens of millions (!) collisions per second. And even the birth of some rare particles have to wait literally years.

the Search of such a precious event among the many background and uninteresting task, after which finding a needle in a haystack seem like child’s play. Of course, this is impossible to handle manually.

Therefore, scientists use special computer programs – triggers. These algorithms sift the huge flow of raw data and selected the most interesting for analysis of the event, which is not tens of millions and thousands in the second. Then the torch picked up by other programs. In the end, you get the result that a large team of specialists can handle in a reasonable time.

Excited omega-baryons are born so rarely that physicists did not expect to catch them from the mass of background events with at least some of the trigger. However, the creation of experts from the Higher school of Economics and the company “Yandex” has exceeded all expectations.

“the trigger is based on machine learning, – says Denis Derkach from the Higher school of Economics. Andrey Ustyuzhanin (head of the laboratory of big data analysis HSE – approx. ed.) and his graduate students took [the algorithm] Matriksnet, which is based on the Yandex search, and it proposed a new solution which gave the opportunity to record this data more effectively.”

Almost all used on the TANK triggers based on artificial intelligence. However, the development of our compatriots stands out from this series.

“It [the decision] records not specific decays, not a specific physics, as the band decays, which can be [we] are interesting. Advantage of this approach is in the fact that we have these decays can search a broad search. After all, if we seek pentaquark, we find only pentaquark. And then we can look for many different new particles and phenomena,” explains Derkach.

after Analyzing data collected over several years, scientists have identified four events in which was born of the elusive particle. The accuracy of detection is five Sigma (“Conduct.Science” (nauka.vesti.ru) explained that it is). Recall that this level of confidence is the gold standard for particle physics and allows to speak with confidence about the opening.

In all four cases, the mass of the particle coincides with the prediction of the Standard model (about 6.3 GeV). So a breakthrough into the new physics and this time did not happen. But scientists have once again convinced thatcvetelina a good understanding of how the world of elementary particles.

Text: To.Science