In September this year the first trainees of Tomsk Polytechnic University, three young scientists, two have been from the Institute of Cybernetics, went on a long-term internship to the European Nuclear Research Center (CERN) situated near Geneva (Switzerland). They took an active part in the experiment “ATLAS” at the Large Hadron Collider. One of the fellow staff, Maksim Gubin, the programmer of the Department of control system optimization, told about the work there and how the experience gained at TPU helped him.
His work there so far has been to analyze the statistics of CERN processing capacity and to search for the bottlenecks. He has already presented the interim results of his work at the weekly conference ADC Weekly, where about 200 representatives of the ATLAS collaboration participating organizations took part from all over the world, and his report was well appreciated. In addition, subject to this work, Maksim was invited to participate at the conference on machine learning in January 2016. So the work is progressing very well.
However of course there are some nuances. So, at CERN they prefer free software and they even have (and widely use) their own distribution package Linux - Scientific Linux CERN. It doesn’t cause any problems, but for those who are left for work to CERN, it is recommended to be prepared to work with Linux. Also it should be mentioned that CERN consists of the Meyren site in Switzerland and the Prevessin site in France, and during working hours daily it is sometimes necessary to cross the border several times.
Now there are two tasks that our programmer has to solve. The first one is to analyze the statistics of CERN processing capacity, while searching for the bottlenecks. The second task is to cooperate with the Kurchatov Institute as an employee of the TPU Big Data laboratory on the development of Data Knowledge Catalog, i. e. the project that is focused on the data storage about what results of the Collider operation are used by the physicists and their analysis using the machine learning methods.