Researchers at HSE’s Laboratory of Methods for Big Data Analysis (LAMBDA) and the Yandex School of Data Analysis have significantly reduced the cost of CERN’s future SHiP detector. The detector will search for particles responsible for still unexplained phenomena in the Universe. With use of modern machine learning methods, LAMBDA and Yandex scientists came up with very effective configuration of magnets which protect the detector from background particles. Not only were scientists able to reduce the size of the magnet, they also reduced the number of particles causing interference. As a result, the cost of the entire structure was reduced by 25%. The results of the research have been published in the Journal of Physics: Conference Series.
The Higgs boson, discovered in 2012, was the final brick in the Standard Model (SM), the model which describes the structure and interaction of elementary particles. This model, however, can not explain such phenomena as the source of dark matter or the baryon asymmetry of the Universe. To approach these problems, the European Organization for Nuclear Research (CERN) plans to launch a new experiment named ‘SHiP’ after ‘Search for Hidden Particles’ after 2020, with the ultimate goal to discover very weakly interacting massive particles (vWIMP). Different theories suggests that this kind of particles may constitute dark matter.
Scientists discuss existence of other particles in the Universe, in addition to those already described by the Standard Model. Because of very weak interaction of such particles with SM ones, they are very rare and therefore difficult to observe. The ‘Hidden Valley’ is one of such models explaining the existence of those ‘hidden’ elementary particles. According to this model, some particles are not observed due to the existence of an energy barrier which prevents their interaction with regular SM particles. One paper on this subject, ‘Echoes of a hidden valley at hadron colliders’, was published in 2006.
The aim of the SHiP experiment is to find such particles. The data collected will provide new information on the possible structure of the hidden sector of the Universe. The experiment involves accelerating proton beams in a super proton synchrotron at CERN. When the beam interacts with a stationary target, many different particles are produced, including hidden sector particles. Most of produced particles are stopped in the absorber installed downstream of the target. Only particles that weakly interact with material can pass through it. This includes usual neutrinos and muons, but also the particles that scientists wish to investigate: vWIMP.
Muons, however, are a dangerous background for observing vWIMP. In order to suppress muons in the SHiP experiment, a unique technology of active muon shield has been used for the first time. It involves deflecting muons in a magnetic field and pushing them away from the sensitive region of the detector, rather than absorbing them in the common approach.
Magnets for this shield will be produced in Russia. Their total weight will be about 1000 tons and they will create a magnetic field of 1.8 Tesla, that is an extreme strength for warm magnets. The international team of researchers employed machine learning methods to optimize the shape and position of the magnets. The method uses Gaussian processes for stochastic optimization. Active shield enables 99.9999% of muons to avoid the sensitive area of the detector.
‘We optimize geometry of the magnetic system which drives the total mass of the system, and hence its cost, keeping low number of background muons that still get through the protection into the sensitive volume of the detector. The machine learning algorithm yielded a solution that is 25% lighter and also even reduces the number of muons in the detector area. Although the detector has not yet been built, we can predict now that the proposed configuration will decrease the cost of the shield by a quarter that would save more than one million Swiss francs,’ says Fedor Ratnikov, Senior Researcher at the HSE Laboratory of Methods for Big Data Analysis.
Neutral vWIMP will pass through the magnetic shield without deflection and decay in the vacuum vessel, located downstream of the shield, right in front of the detector. The detector consists of tracker, calorimeters and muon chambers. It also includes emulsion chamber which will detect traces of neutrinos. Such a configuration will enable scientists to explore new phenomena of the Nature, far beyond the ones merely described by the Standard Model.
The SHiP experiment is currently in the development stage and the first results are expected by 2025.