On Friday, September 8 th I attended a Sino-Italian workshop on astrostatistics organized at the Department of Statistical Sciences in Padova. It touched current topics at the interface between Astronomy, Physics and Statistics. At a first glance, I was surprised by the similarity of the research topics that are faced across different fields of science. Often the main difference lays only in the data and the assumptions of the underlying data generating process.
Two talks of the workshop were given by members of our network. The first one by Giovanna Menardi was titled “Nonparametric Semi-Supervised Classification with Application to Signal Detection in High Energy Physics” and introduced an interesting general clustering framework to detect deviations from known processes. In contrast to other talks, which focused on some particular problem (i.e. Neutrino Mass Hierarchy) and described a particular framework construction to solve it, the talk introduced a problem in general terms, introduced the statistical approach and only in the end showed results for a particular case study. In this way, the presentation was understandable for non-experts, in a way I would wish to find also in other scientific presentations, but I have also experienced myself how difficult this is .
The other talk was held by Tommaso Dorigo and touched extraordinary claims in Physics and a controversial “5σ rule”. In an interesting way the presentation described the origin of this ad-hoc arbitrary method for claiming discoveries, with its disadvantages and failures in proclaiming discoveries that later turned out to be incorrect. Of all the talks, this one provoced the most questions and intriguing debate.
After the main part of the talks, a Junior Speed Session was organized during which young researchers, including myself, gave a 3-minute presentation of their study as a teaser for a later poster session. Personally, I find such a speed session to be tiring, because with the speed of a machine gun the audience is given a huge amount of information from which almost none is remembered. On the other hand, it is somehow broadcasted in advance which posters are connected to which part of science, problems and algorithms.
Apart from the mentioned activities, one of the presentations caught my attention in particular. Jessi Cisewski-Kehe showed her investigation of the cosmic web using Topological Data Analysis. I touched the subject of Topological Data Analysis during my undergraduate studies and later when Pablo de Castro deepened the subject during his secondment in SDG, but apart from the clusterization and a neat graphical representation of the data I was not aware of any more useful examples of the technique. In the presented research the large-scale structure of the Universe is analyzed via persistent homology and hypotheses tests using the topological summaries presented. I was glad to learn about it.
In conclusion, I found the workshop very stimulating for my work, as I met researchers who try to solve similar problems in their fields of study using completely different methods. I’ve also experienced how important proper presentation is in order to convey a message to scientists from different fields of science.