by Grzegorz Kotkowski

Since the beginning of October I have been in secondment at CERN. I expressed my first feelings about this complex research center in my post “CERN at first sight”. Two months have passed and my former admiration has settled down. The common days allowed me to get accustom to it and eventually I could focus on my work.

Throughout this time I have participated in numerous seminars and speeches. As I am not a physicist some of them were difficult for me to follow, but eventually as I learned more and more about particle physics the presentations became understandable to me.

Before coming to CERN, particle physics had been presented to me as a data-mining tasks. In fact searches (especially model-independent ones) are just the application of BDT or neural networks algorithms with some fine mix of polynomial fitting, regression etc. However I was also pleased to see at CERN the probabilistic approach to model the observed phenomenon – something that fits more to me. For example I attended an interesting talk about modeling cosmic ray with Levy-stable laws.

By spending time here, for sure I came to understand physicists more. I found out that they are not well learned probability and statistics. For example I attended four courses about “Statistics for Particle Physics Analyses” performed by professors L. Lyons and L. Moneta. We were taught about statistical methods (or rather how to interpret the results) without giving almost any assumption, proofs or calculations why the particular variables converge to the given distributions. Someone could say that, as physicist are not statisticians, I should not demand them to sit with pen and paper trying to solve some probability exercises. Sure – but on the other hand using statistical software and interpreting the results without for example knowing the assumptions of the methods looks like a display of ignorance. I rather see the trend of applying neural networks to every possible data set. I hope it is not the future of statistics to become limited mostly to the machine learning.

Besides my analysis I was pleased to have tours and see CERN from the inside. For this I am thankful to prof. Tancredi and Pietro Vischia. I was able to see the LHC and ATLAS control rooms, parts of ATLAS detector, the first CERN accelerator (a 600 MeV synchrocyclotron from 1957), the headquarters of AMS-02 (an experiment on the international space station) and much more. Thanks to the tours I appreciated more the work of researchers who struggle to deliver as good data as possible. I have seen the long chain of processes that needs to be performed before I could work on the data.

Below I want to offer some photos I took during the secondment. On the feature image I present my daughter and myself in building 40. That is it for now. I am packing and getting prepared for our network conference in Oxford. I cannot wait!