Welcome to Yoshi’s Home Page

One of the great joys of being a research academic is being able to spend a lot of one's time working with fantastic students. My PhD students who have now graduated include:

Dr Chien Lin (subsequently returned to his home country to carry out his national duties)
Chien joined us in October 2020, having previously studied as an undergraduate at Oxford. He was my first new student after COVID happened, with the country being in the midst of lockdowns and partial liftings of lockdowns when he started. I vividly remember the moment we first met, sitting at opposite ends of a park bench in Imperial's beautiful Princes Gardens, where we discussed his bright future as a PhD student in a world where people were suffering from the virus and there was yet no vaccine, and we did not know when we would be allowed into our offices and to travel abroad.
In the end, it would be two years before Chien was able to move to the J-PARC laboratory in Japan to work on-site for the T2K experiment, but he was able to spend that time fruitfully, not only working for the software contributions to T2K ND280 Upgrade in preparation for his hands-on work with it, but also for the GAMBIT global-fitting framework for Beyond-the-Standard Model physics.
Once in Japan, he joined the team that built and commissioned the SuperFGD and took responsibility for helping build up the calibration scheme for the new detector. There is nothing like turning up every day and spending many long hours putting together the components that make up a detector (two million of them, in the case of the SuperFGD) to get to know not only the detector itself, but also the people with whom you will be working as it is brought to full physics-ready order. Towards the end of his PhD, he also worked to apply machine-learning methods to low-energy reconstruction in the SuperFGD, which made use of his knowledge of the calibration scheme.
At the same time, his role in the GAMBIT project kept growing, to the point where he became the convenor for the neutrino module, helping put together a paper on oscillation fits. Since the work of fitting together the results from the many experiments that have studied neutrino oscillations meant that he needed to understand the details of the analyses for experiments other than his own, he needed to meet physicists from other experiments to sit down and ask them all of the questions he had. Earlier on in his PhD, I joined him on such a visit, for one of the first in-person meetings that we were allowed to have post/mid-pandemic, and I am happy to say that on that memorable occasion, not only did he learn about how other experiments carry out their data analyses, but I was able to educate Chien on something that his undergraduate institution never seems to have managed: the joys of the traditional English pub....
Dr Roden Derveni (subsequently a postdoc at Imperial, and then an employee for a prominent online company)
Roden arrived in 2019 after studying Theoretical Astrophysics at the University of Nottingham. He has been a great example of the fact that you definitely don’t need to have done particle physics as an undergraduate to join our group and do well. He settled into the COMET project quickly and ended up working across a surprisingly wide range of topics, from detailed simulation comparisons early on to analysing the first commissioning data from the newly-built proton beam line once the experiment started producing it.
His PhD years were heavily shaped by COVID too, which closed borders just as he was preparing for his first research trips to Japan. Like many of us in that period, he had to adapt to doing most of his work online, but with COMET being in Japan where internally the rules were relatively relaxed, many collaborators were still able to work together on-site, which added an extra layer of complexity. Roden handled it all calmly and managed to stay closely involved in the international work despite the odd hours and moving targets.
One memorable episode came when he spent several months in Clermont-Ferrand working on a hardware project there, and at the same time helping their students get up to speed on the COMET software. This was shortly after the new post-Brexit rules came into force, and only later did we discover that his extended stay meant he couldn’t return to the EU for a holiday later that year. Not the sort of impact that I was hoping the overseas experience would have on his life!
Within the group and the wider collaboration, Roden became known for being particularly generous with his time, helping younger students, answering questions, and generally making life easier for anyone trying to navigate the COMET software for the first time. After finishing his PhD, he stayed on with us as a postdoc before eventually moving into industry.
Dr Matthias Dubouchet (completed February 2023—subsequently a researcher for a well-known technology company in France)
Matthias joined our group in 2018 after completing his undergraduate degree at the University of Sussex. He chose to work on the COMET experiment, which was then moving from design towards construction and commissioning, and quickly found himself involved in the simulation and analysis side of the project. His thesis focused on understanding the expected sensitivity of COMET Phase-I to the muon-to-electron conversion process, and in particular on estimating the backgrounds from cosmic-ray muons that can imitate a conversion signal. So, he worked with others to apply a backwards Monte Carlo approach that allowed computing time to be spent efficiently on the rare, signal-like events most relevant for the experiment. Using this, he was able to provide one of the most detailed estimates to date of how well COMET will be able to distinguish true conversion electrons from background muons once data taking begins, and to identify strategies that could further suppress those backgrounds. His work was the first time in many years, since the initial conceptual design stage, that the full simulation and analysis flow had been run to determine the experiments sensitivity using the most modern software.
Another aspect of Matthias' work was to use Generative Adversarial Networks to create events that looked like simulated background events in the detector, but were actually “made up” by AI to look like such events. All of this was very new and we will be building on his work to help make sure we can simulate enough events for the experiment, which is not trivial when pushing the very limits of what rare-decay searches can do.
Matthias is another student of mine whose graphical skills stood out, so one time at a group meeting, I asked him whether he could come up with something that helps visualise what actually happens in our detector when we see all the particle whizz through it, and he did, very well indeed!
Like so many students of that period, Matthias’s PhD was disrupted by the pandemic, which forced many things online. He managed this very well though, and I’m glad that he was able to experience working with actual particle beams and detectors during his degree work, as well as visiting our collaborating institutions in Japan and elsewhere. As it happened, I even came down with COVID myself in the week of his viva and couldn’t attend in person—such a disappointment!
Dr Wilf Shorrock (completed September 2021—subsequently a researcher at a well-known technology company in London, and then a postdoc at the University of Sussex on the NOvA and Liquid-O projects)
Wilf joined our group in 2017 from the University of Durham. One of his first projects was to help bring the supporting framework for the T2K ND280 software into the modern era. It had been largely unchanged since the early 2000s, and with the ND280 Upgrade coming along, it was clear that something more up to date was needed. He worked with colleagues from COMET and other experiments to understand how they had organised their software, and then set about applying those ideas to ND280. It was a big task—which not everyone was so confident about—but Wilf quietly got on with it and, over the next three years, co-led the redesign of the framework. The version now in use owes a great deal to the decisions he made during that time, and will likely still be going strong when Hyper-Kamiokande is running.
Early on, I also suggested that he join the R&D work for the SuperFGD detector, which sits at the centre of the ND280 Upgrade. Wilf went on to play important roles in the test beam campaigns at both CERN and Los Alamos, taking part in everything from the on-site data acquisition to the analysis that led to a publication, where he was one of only two UK-based authors. It was particularly nice to see him take a genuine leadership role there, without much prompting from me beyond sending him off and wishing him luck.
Wilf also worked on something completely different: bringing neutrino data into the GAMBIT framework, which had originally been developed for Beyond-the-Standard-Model fits. This was a new direction both for GAMBIT and for us, and his work led to the first incorporation of T2K data into the framework. He proposed ways in which T2K results could be published to make such comparisons easier in future, something the collaboration is still discussing, and was elected by the GAMBIT collaboration as their first neutrino convenor.
Many would say that each of these projects could have been a full PhD in its own right, but Wilf somehow managed to fit all three into his. After finishing his degree he spent some time doing research in industry before returning to academic research.
It also turns out that Wilf is an accomplished football player, and in addition to the annual Bubble Chamber Tournament between the UK particle physics groups, we also played at the kickabouts we have at the T2K Collaboration Meetings. Shouting “Come On Wilf!” from the sidelines was something that came quite naturally to me in those days....
Dr Artur Sztuc (completed March 2021—subsequently a postdoc at UCL on the NOvA and CHIPS experiments)
Artur joined our group having completed his undergraduate degree at the University of Salford. The majority of universities in the UK are not active in the field of particle physics, and Salford counts as one of those, but Artur had worked in X-ray diffraction experimentation as an undergraduate, which has many parallels with particle physics work, and so he came on board the T2K-Imperial group in 2016. I’m sure it is not just me who would say that Artur is quite the character, though it clearly is best if I generally don’t go into too much personal detail about my students here! Suffice to say that he is very much suited to doing independent research, and having plenty of fun while doing so. He even compiles his everyday laptop Linux from scratch, rather than relying on such distractions as “distros”—admittedly not the juiciest tidbit. He did a very wide range of things during his time at Imperial, including spending a lot of time at the Super-Kamiokande experiment helping out with the different UK responsibilities we have out there, but his focus for the later parts of his thesis research was using modern statistical methods to extract the most from T2K and Super-K data. This includes a major part of the results that appeared in our Nature paper of 2020. Most people who spend time in Kamioka live in the city of Toyama, about an hour away, but Artur chose to live in nearby Kamoika itself, and became famous as the only gaijin in town. I stayed for a week early on at the very traditional house he had rented there, introducing him to all the townsfolk who had all been wanting to talk to the new arrival, but whose language skills didn't allow them to. Being a mining town, every other house in Kamoika seemed to be a bar, making it a very memorable student-supervisor bonding experience! One of my fundamental principles is that each student's post-viva party has to be better than the previous one, but Artur was the one student of mine who had a fully-online PhD viva because of COVID, resetting the scale somewhat. As always, I popped open a bottle of Moët Impérial for Artur after he passed his viva, but doing that in my living room in front of a webcam was not quite the same!
Dr Ewen Gillies (completed November 2018—subsequently a machine-learning engineer at a London-based startup)
Ewen came down to Imperial after reading Theoretical Physics at the University of Edinburgh, having grown up in the States and made the jump across the pond to the UK. His nationality meant that he was not eligible for a research council-funded position, so I was delighted that he was able to secure a prestigious Schrödinger Scholarship, which allowed him to join Imperial to work on COMET. Of the many different detectors and physics problems that a student can work on, from a very early point, Ewen homed in on the fact that the needle-in-a-haystack problem (a 105 MeV/c electron amongst trillions of lower-momentum electrons) that the main CyDet detector will face was well-suited to machine-learning methods. To help crack the problem, Ewen was able to secure fellowships at the Yandex Data Factory research laboratory in Moscow and at KEK in Japan to spend several months at a time working on algorithmic, software, firmware and hardware-based solutions, leading to our collaboration choosing his algorithms as the baseline options both for on-detector and offline variants. At the same time he also carried on the tradition of our group playing a leading role in the software for the experiment, working with colleagues across the world to bring it to the state that it is today.
Now, whilst most PhD students tend to spend their time as students secure in the knowledge that their funding is assured for the duration of their degree, one thing about Ewen was that he was able to come up with excellent research proposals almost as a matter of habit; in fact from various scholarships and fellowships to significant research proposals, I believe that he was successful in every funding proposal he submitted with me, and there were more than just few! Ewen is the first of my PhD students to have chosen to go into industry after becoming a doctor, instead of looking for a postdoctoral research position. So his Midas touch will not be applied to bringing in research council funding grants in the future—but I am sure that our loss is the real world's gain as he makes the most of his substantial talents, as well as proving that being my student doesn't have to mean you can't get a proper job!
Dr Luke Pickering (completed June 2017—subsequently a postdoc at Michigan State University on the DUNE and T2K experiments)
Luke was an undergraduate student at Imperial, but I do not think our paths truly crossed until he started as a PhD student in October 2013. From the beginning, he was very adept at coding, so it did not take much time at all for him to come up to speed with his work on T2K and start making real contributions. T2K is now quite a mature experiment, so there are lots of topics a strong student can dig their teeth into, and Luke certainly did that! Not only did he make very high-quality improvements to the long-standing T2K ND280 ECAL neutrino interactions work, but he also made real contributions on several other fronts, which led to his publishing papers with a small number of authors that were separate from the big collaboration papers everyone works together on. He was part of a small and lively group of young physicists from across the UK that worked together very effectively, whilst having a lot of fun! One of those was a postdoc whom I originally encouraged to come to the UK and join T2K (although since we did not have a position open at Imperial, I had to recommend that they apply to a different university), which was very nice to see. The contributions that he has made to the neutrino physics community includes a framework that facilitates the comparisons of different mathematical models for neutrino interactions, called NUISANCE....
The one little tinge of sadness that I have with Luke is his choice of programming editor. It would have been one thing if he had gone for emacs, but he actually uses a certain proprietary, non-free software editor, and is the first student of mine who does not use VIM.... But on the other hand, I'm really happy in particular that Luke has gone on to work as a postdoc for Prof Kendall Mahn at MSU, a very highly-respected neutrino physicist that I know of from when she was an undergraduate student at MIT and I was a postgraduate student in the same lab—to think that I would have a future PhD student who would then be employed by her as a researcher was not something that I particularly envisioned happening at the time!
Dr Ben Krikler (completed November 2016—subsequently a postdoc at Bristol on the CMS and LUX-ZEPLIN experiments)
Ben was also an undergraduate student at Imperial, and worked with me during his gap year after gaining his Masters degree, while also travelling round the world, so by the time he joined our group as a PhD student, he was already used to working several time zones away from Imperial and participating in online meetings involving participants from many countries. Our students are given the opportunity to choose which experiment to work for after they arrive, and following much deliberation, Ben decided to do his PhD with me on the COMET experiment, and never looked back. As is the case for many COMET students, his research spanned hardware, software and physics analyses, including work on the AlCap experiment. His contributions to the COMET software framework, which was being completely revamped under the leadership of Dr Ajit Kurup when he arrived, have been nothing short of spectacular. We now have a reliable software framework and a highly active and very international group of collaborators who work together to continue improving it; much of this is due to Ben's open and collaborative nature as well as his outstanding computing skills. His thesis research represents the most thorough study of the physics of the full Phase-II COMET measurement and was quite the tour de force, digging up obscure references to provide the data to constrain the more subtle sources of background which we need to understand. Even though he has moved on from the group and is working on other experiments, I am happy to say that he still keeps in touch, letting me know whenever he has a clever new idea that might help our experiment!
Dr Dominic Brailsford (completed December 2015—subsequently a postdoc at Lancaster on the T2K and DUNE experiments)
Dom was a student at Durham before he arrived here in 2012. He spent a lot of time improving the event reconstruction algorithms, especially in the ECAL, starting from the bottom up (at the single scintillator bar level) and verifying and improving the individual steps of the reconstruction through to the final outputs. He also studied the effects of the magnetic field inside the iron yokes of the ND280 detector, which had not been treated particularly rigorously until then, as it was thought that the impact on physics was small. Dom came up with an alternative implementation of the magnetic field in the simulation, which improved our understanding of background events coming from the magnet, and which is now a standard part of the T2K software. All of this work tended to involve a lot of meticulous comparisons between real data and simulated events, by choosing specific event samples that could be compared between the two, and investigating how any differences arise, and correcting for them if necessary. He also continued our group’s tradition of leading the design and maintenance of the offline analysis format that everyone uses. His vastly-improved ECAL reconstruction algorithms were then applied to real events, where hundreds of thousands of neutrino interactions on lead were found. An analysis of this sort, with very large event samples, requires the systematic errors to be understood well, and much of Dom’s work involved estimating the effects of many sources of uncertainty that were not needed before.
Dr Ben Smith (completed March 2014—subsequently a postdoc with TRIUMF, on the DEAP dark matter experiment)
Ben also came to us from Durham University in 2010, having spent some time working outside of academia after receiving his undergraduate degree. This was now about a year after T2K had started running, so many of the most important aspects of the experiment, both in hardware and analysis, had been done—at least once. As with any experiment, the challenge of getting everything working for the first time was quickly supplanted by the challenge of doing everything better—equally importantly—demonstrating that is indeed the case. Only then can we improve our results significantly by taking more data. Ben was very quick to jump in and help improve event reconstruction algorithms in the near detector. As a member of the ‘reconstruction task force’, he made significant contributions to this effort. Later on he shifted his focus to analysing the output of the reconstruction, to do his own analyses on electron neutrino interactions in the detector, while also helping set up a framework that everybody else could use to do their analyses.
Ben was a recipient of an STFC STEP award, so he stayed with us as a postdoc. This gave him the time to allow him to take his thesis analysis and turn in into a fully-fledged paper, which has now been published in the journal Physical Review Letters.
Dr Mark Scott (completed May 2013—subsequently a postdoc also with TRIUMF in Vancouver on different areas within the T2K experiment)
Mark arrived here in 2009 from Cambridge, at about the time the T2K experiment turned on. Most of our students choose the experiment that they will be working on a few months after they arrive—having spent that time studying, and being courted by, these experiments and the Imperial physicists who work on them. So the time Mark actually joined T2K was about the time that the experiment was turning on in December 2009. But there was still much work to be done using data that we had taken at CERN earlier that year, using the ECAL in a ‘test beam’ of known particles, and working through this data, so that we could use it to help set up the ECAL correctly and interpret the new data properly. This was the initial focus of his work.
Once this work was done, and used to analysed the data coming from the ECALs, Mark went on to contribute to a wide range of topics in ND280, eventually focusing on using new algorithms to reconstruct neutrino interactions occurring in the ECALs themselves rather than in the other detectors, something that no one had done before. Not only does this tell us about neutrino interactions themselves, but it also helped us pin down the number of “background” events caused by the ECALs in the central parts of ND280, which was valuable for the oscillation analyses.
Dr Sarah Ives (completed November 2012—subsequently a postdoc with Imperial on the CMS experiment)
Sarah joined our group in 2008, after completing her undergraduate degree at Oxford. While our group’s main focus on T2K has been the near detector, it is only in combination with the neutrino beam (of course) and the far detector that we can figure out how neutrinos oscillate. Sarah worked on the far detector, Super-Kamiokande, initially, where she performed some early studies on the possibility of expanding the fiducial volume of the detector. Once the ND280 near detector turned on in late 2009, there was a huge amount of work to be done for us to be able to understand it. Sarah led the time-calibration of the detectors that we had built in the UK, Electromagnetic Calorimeters (ECALs), which required the detectors to be run using cosmic rays, and the timing offsets between different electronics channels to be extracted from the data. We learnt a great deal about the detectors and their electronics through this work that we now take into account when we use the data from the ECAL. She then went on to do a physics analysis of the shape of the energy spectrum of the neutrino events seen in the near detector at energies of over 1 GeV, which is higher than the typical events that we being studied, which yielded information that helped her make a quantitative estimate of how well we understood the neutrino flux, which has always been a notoriously difficult problem for neutrino beam experiments.
Dr Jim Dobson (completed February 2012—subsequently a postdoc with UCL and then the University of Edinburgh on the LUX and LUX-ZEPLIN experiments.)
Jim arrived in our group from Cambridge in 2007, a couple of years before T2K started running. We were still building the near detector, so Jim worked in the lab to help us understand the components that we were using such as the novel Silicon Photomultiplier detectors, and made sure that they were simulated well in our software. He worked with Dr Costas Andreopoulos at RAL, his co-supervisor, on the GENIE neutrino interaction generator package and also a tool that everyone still uses to help ‘reweight’ the simulations so that they fit the data well, and to extract systematic uncertainties while doing so. Jim’s thesis was one of the first to look at more complicated neutrino interactions in near detector, which made use of the ECALs that we built to understand the pions that are produced.
Towards the end of his time here as a student, Jim won a STEP award, that allowed him to stay with us as a short-term postdoc, to develop his analysis further and to allow the results to be presented publicly.

Soon to be added to the above list are C. Lin and J. McKean.

Welcome to Yoshi’s Home Page

Yoshi Uchida

My Inaugural Lecture

Selected Recent Activities

Former PhD Students