Very shamefully, this is my first post on the network blog: I hope to make up for this by spamming writing a reasonable amount of posts in the near future.
For now, a brief presentation of myself 🙂
My name is Pietro Vischia. I come from Padova, in Italy, where I was born and where I got my bachelor and master degree in Physics. I then moved to Lisbon (you might have read in this blog some posts from Giles and from my former supervisor João Varela), where I got my Ph.D. last July, and now I am in Oviedo in Spain, for my postdoc.
I am an experimental physicist working in the CMS experiment at the LHC, but I originally come from the CDF experiment at Tevatron, where I did my bachelor degree and spent an awesome time as a summer student at FermiLab (the mega laboratory near Chicago which hosts Tevatron, which is to many respects the predecessor of the LHC).
How on Earth did I end up doing these things? During primary school I already knew that I wanted to become a cop: something must have gone wrong!
Well, as far as I can remember, the idea of becoming a cop already changed, right before high school, into the idea of becoming a lawyer (not a judge. Lawyer is cooler: Perry Mason was a lawyer, not a judge): after all, they earn more money and they can have a larger impact on the society.
I went on to humanities high school (in Italian it is called Liceo Classico), where the study of Greek and Latin and their ancient language and culture, as well as philosophy, made me think I actually wanted to become a philosopher, heavily biased towards symbolic logic and such.
In parallel, I was preparing for a backup career as a composer and orchestra director, by studying and passing exams at the Conservatory. I was quite prolific as a classical music composer, by the way (and my Maestro said I was writing reasonably good stuff!).
However, the curriculum of the Liceo included also some math (starting from the first year) and science (starting from the third year with chemistry), and I soon became fascinated by the idea of studying how nature actually works instead of just memorizing the rules of men or elaborating mind constructions without confronting them with reality.
However, chemistry, the way it was taught, seemed not rigorous and “logic” enough for me (“if it’s not logic, it is bullshit”). This opened up the way for falling in love with physics during my fourth year of high school (in Italy, high school lasts five years): things seemed to be built in a more logic way.
In retrospective, I guess I could have arbitrarily settled at any of these steps, depending on the events of life, so somehow I guess I am a physicist by luck: I got fascinated by stuff in the right order at the right time.
Of course, what is the coolest physicist you might think about, if you haven’t really been exposed to science and hence to the different possible flavours of physicists? The nuclear physicist!
Soon after the beginning of the university courses, I started to think that most likely the most interesting (hence cooler) field was astrophysics.
However, an awesome thing happened: the physics department used to give the opportunity to students to work with various physicists on different topics, for two or three weeks during each summer (some sort of summer internship program). Right after my first year, I looked at the list of available internships and decided that since I did not have a real preference I would have chosen on the basis of the most amiable professor.
Back at the time, I though it was vaguely ridiculous. Nowadays, I would just say that it is an estimator as any other and that it simply reflects my prior ranking of all the possible discriminant variables: this way, I would then skip mentioning the fact that this does not make it in any way less or more ridiculous.
In any case, I chose the internship led by Giovanni Busetto, who is very-very-awesome and had taught me a lab course (measuring stuff in electric circuits, measuring stuff in optics – things like that). He happened to be working within the CDF experiment (CDF is a detector that was located in one of the interaction points of the Tevatron accelerator, near Chicago) together with some bloke named Tommaso Dorigo: I am sure you have heard about him, here… Anyways, as a newcomer, I started working at simple tasks like learning how to plot variables.
An additional internship experience with him, the following year, finally convinced me that the funniest and coolest thing is particle physics, so here I am 😀
Now that I established that I am a particle physicists just by sheer luck, let’s venture into a longish description of my working career.
My bachelor degree aimed at improving the accuracy with which CDF measured the energy of jets of particles coming from the hadronization of b quarks. To put it briefly, quarks are usually detected in the form of “jets” of particles (pictorially like a swarm of bullets from a shotgun): jets coming from quarks of type “b” have some peculiar characteristics that can be used to identify them and to improve the measurement of their energy. The results were somehow inconclusive, though, leaving just the hope of being able to exploit one particular variable in a future multivariate study.
During my time in the Fermilab Summer Student program, I had my first real encounter with multivariate analysis methods, in particular machine learning: in university courses, at least in physics, this stuff was not covered at all. Right now the situation seems to be slowly improving, but it still depends on the effort of individual dedicated people that push for creating optional classes or mini-workshops.
My task was to replicate some preliminary studies that had been performed by Ohio University on improving the di-jet mass resolution and the discrimination between events featuring a Z boson decaying into two b quarks from events of pure QCD (basically, events with many jets). Di-jet mass means “take two jets and consider them a single object: which is its mass?”, whereas resolution is a quantity that tells us how well we measure that quantity: higher values are bad, low values are good!
At that time, the software used for the analysis of events, ROOT, had some basic experimental support for Neural Networks, which was however mostly undocumented. Actually, I spent most of the time gathering useful documentation on the basics of Neural Networks! The study yielded some encouraging results, as far as I could judge by digging up old emails: by removing QCD events via the Neural Network the resolution on the di-jet mass went down from 18.0% to 10.7%, according to my notes!!!
For the master thesis, I switched to the CMS experiment at the LHC, where I am still working nowadays. There, I worked on something we call “VBF H→ZZ→μμbb” Higgs boson search: OK, what the hell does this mean?
Let’s start with “H”: that is the Higgs boson, that at the time had not been discovered yet. H→ZZ means that the Higgs boson decays into two Z bosons, ZZ→μμbb means that one of the Z bosons decays into two muons and the other into two b quarks. That was a typical search channel for a Higgs boson, but it is affected by large backgrounds.
A background, in general, is some other process that produces the same type of particles after the collision. In this case, something that produces two muons and two b quarks. A typical example is the top quark pair production.
In order to improve the situation, one can limit the search by imposing constraints on the way the Higgs boson is produced: “VBF”, standing for “vector boson fusion”, is a particular production mechanism whose peculiarity is that in the final state you expect to find two jets with extremely large momentum. The presence of these two jets is not typical for the backgrounds to the Higgs boson production and decay, hence their characteristics can be exploited to classify events into “signal-like” events and “background-like” events.
For this classification problem, I used a machine learning technique called Boosted Decision Trees, which is somehow an extension of the concept of performing cuts on the value of some variable aiming at increasing the fraction of signal-like events on one side of the cut and of background-like events on the other side. (Then you do this many times, for many variables, according to some criteria, and finally you redo the whole process a few hundred or thousand times).
So, what was the result? Well, it turned out that the separation between signal and background did not increase significantly if compared to single-variable cuts. However, using the output of the BDT permitted to enhance significantly (in some cases even by 50%) the ability of choosing the correct pair of jets for the jets coming from the vector boson fusion. This, in turn, allows to assign more often the correct pair of jets to the jets coming from the decay Z→bb, which enhances the capacity of reconstructing correctly the event and thus opens the way to a more detailed study of the Higgs decay topology.
Actually, digging up the past presentations and theses in order to make sure I report things correctly has been a very interesting exercise: I had indeed forgotten many details of these works of mine!
Well, I have probably started to bore you, so I will summarize the rest quite briefly: also, I must say that I have just written an entire thesis about my Ph.D., so right now to describe it again is kind of boooooooooring for me. Programmatically, I try to mostly do fun stuff, because if it is fun it can still be tiring but it is not really “work”.
During my Ph.D. in Lisbon, I started by joining a measurement of the top quark mass in top quark pair events in the dilepton final state. The mass measurement method used was called kinematic method and had been originally developed within the CDF experiment. It consists in computing the top quark mass as a function of the full kinematic quantities of the event.
Since the full kinematics of the event is not known (there are neutrinos that we don’t detect and when we collide two proton beams we don’t know exactly the longitudinal momentum of the individual protons that collided), it is necessary for each event to solve 50000 times the system of equations that determines the event. As a progress with respect to the original CDF method, we exploited the information on the identification of jets coming from b quarks, which has a much better performance for the CMS detector.
I then moved to the measurement of the cross section of top quark pairs in final states containing τ leptons. The τ lepton is the heavier brother of the electron and of the muon and its signature characteristic is that it decays most of the times into a cluster of particles (it decays hadronically), mimicking the hadronization of a quark: this means that the challenge of the work is to distinguish jets from hadronically decaying taus.
I exploited this final state also in the search for something called charged Higgs boson: the idea is that, in case new physics exists, the Higgs boson we discovered could have some heavier brothers, sometimes with an electric charge (hence labeled charged Higgs boson). When the mass of such a boson is lower than the top quark mass (i.e. lower than 173 GeV), the charged Higgs boson decays mostly into a τ and a neutrino: that’s why we chose to investigate a final state with taus despite the challenge of their identification.
The really cool part, though, was when I passed to the search for a heavy charged Higgs boson (i.e. a charged Higgs boson with mass higher than the aforementioned 173 GeV): in that situation, the charged Higgs decays mostly into a top quark and a bottom quark (modulo the electrical charges of the stuff, so either top and anti-bottom or anti-top and bottom). The result I obtained is the first result for a direct search of a heavy charged Higgs boson decaying into a top and a bottom quark!
Across the years of my Ph.D., I realized more and more that statistics is very important for the production of a solid scientific result. It is actually a pity that a non-negligible part of what an experimental physicist should know is not taught in standard university courses. At some point, I had the opportunity of entering the CMS Statistics Committee, a group of people in the CMS experiment that are tasked with “checking the stat stuff” for the experiment, formulating and giving suggestions when possible to colleagues that might have some delicate statistics issue: having direct access to tricky and delicate conversations on statistics boosted and is boosting my understanding and experience in this topic!
To finish this work related jibber-jabber, I must only say that I have just begun my postdoc at the University of Oviedo, in Spain. Here, my analysis work will be oriented to searches of new physics, mostly in the context of a theory called Super Symmetry: SUSY, for friends, while keeping my commitment in the stat stuff for my experiment and expanding it with some personal project.
I joined the AMVA4NewPhysics network during my Ph.D. in the context of my Portuguese institution: so far, I mostly developed together with Tommaso a nice new multivariate algorithm, called inverse bagging, that I presented recently at a conference and that will be the object of a blog post in the near future. Uh, and I am currently the Outreach Officer of the network, which kind of adds shame to this being my first blog post. Ehem.
I am keeping my involvement in AMVA4NP, and am the project PI of the University of Oviedo group that is joining the network: more on the group in another post 😉
As finishing remarks, here are a few non-work-related bits about me!
I have always loved reading (thanks to my parents for inculcating in my mind the love for reading!), and up to a couple of years into university I was also writing music and playing the piano. I should definitely re-take some music into my hands.
Anyways, back to reading: I am an avid reader, ranging from narrative to essays to the shampoo composition and instructions (not kidding). In the most recent years I have enjoyed a lot digging into crime novels, but paradoxically right now I am reading none of them. I just started reading a Spanish series of books called “Las aventuras de capitán Alatriste” (The Adventures of Captain Alatriste) that reportedly is the Spanish equivalent of Dumas’ musketeers series. From the first pages, it already promises to be AWESOME.
When I was a child, I used to draw a lot (seriously, A LOT), but then I stopped, except for my awesome notes taken during courses that always featured pirate ships: I have read too many novels from Salgari about pirates and corsairs. And yes, pirates and corsairs are not the same thing! Corsairs were privateers, i.e. people that, because of some ongoing war, had been granted a commission of war: basically, legalized pirates (if you listened to the country granting the commission. They were still hanged by the enemy country)! Also, the commission was usually valid against a specific nation, so if a British corsair with a commission against Spain attacked a French ship, the British themselves would normally consider him a pirate. Complicated times.
In any case, a couple of years ago I restarted drawing, encouraged by a dear friend, Lara (check out some of her work!!!), and am currently in the process of (very slowly) improving my skills with watercolour. I have been convinced/compelled/tricked into even uploading my (still bad) drawings in a blog, but I won’t tell you where. Haha.
I have a passion for languages. Before high school I had studied English and French, but then I perfected only English in high school and beyond. Nowadays, I practice French mainly during my trips to CERN. Thanks to having lived in Portugal I learned Portuguese. The same friend that pushed me into drawing also taught me Spanish (perk: I arrived in Spain already speaking some kind of version of Spanish with many Asturian expressions – Asturias is the Spanish region Oviedo is the capital of). A couple of years ago I started studying German with an awesome teacher (Nora, that recently created a non-profit association of language teachers working also via skype), but then had to stop for the finishing-the-PhD haze: I plan to restart around December!
I like to play computer games: mainly RPGs-stuff like Gothic or Skyrim or Assassin’s Creed and first-person-shooters-stuff like Call of Duty or Vietcong or Battlefield.
I love roleplaying games (Dungeons and Dragons and Vampire, mostly, plus some indie systems like Dread), and play quite often since something like 15 years ago. In the last couple of years I started to play live roleplaying games. The difference is that you are not seated around a table describing what your characters do, but you are roaming around a room, or a garden, or a town, effectively acting as if you were your character. This implies that you can forget stuff typical of a pen&paper game such as “I break the wall with my fists” or the most common development for the first D&D session of a campaign (“We burn the tavern to the ground”), but it is more immersive and focused on interpersonal skills.
Actually, I must say that this practice helped me with my general dialectic skills. If you drop by Lisbon, check out the folks from Olisippo Obscura! My next project is to create a campaign in Oviedo, by the way!
Well, I think I have to end this post here, lest I incur in warnings by the awesome editor of the blog (Sabine, the Press Office Coordinator of the network), so I thank you for the attention (provided you had the patience of arriving up to here) and encourage you to drop by in the comments for any question.
For now, talk to you in the next post! You can also find me on twitter, with the handle @pietrovischia. I will just finish this one with a photo of my two awesome cats: Maria João (right) and Sborra (left)