UW Physics alum Kyle Cranmer chosen to lead American Family Insurance Data Science Institute

profile photo of Kyle Cranmer

Cramer, who received his PhD in Physics in 2005 from UW–Madison, will be a faculty member in the Department of Physics in addition to Director of the American Family Data Science Institute

 

This story was originally published by University Communications

Kyle Cranmer, a University of Wisconsin–Madison alumnus who played a significant role in the discovery of the Higgs boson, will become the next director of the American Family Insurance Data Science Institute.

“We are excited to welcome Kyle back to UW–Madison, where he earned his PhD in physics in 2005,” says Amy Wendt, associate vice chancellor for research in the physical sciences. “Kyle brings a background to the position of director that will facilitate research synergies throughout campus, connecting data scientists and domain experts working to address present-day challenges ranging from health care to education, the sciences and beyond.”

Founded in 2019, the institute is working to advance discoveries that benefit society through data science research, the translation of fundamental research into practical applications, and collaboration across disciplines. The institute is a campus focal point for integrating data science into research, and one of its top priorities is to build a thriving data science community at UW–Madison.

Cranmer is currently a physics professor at New York University and will assume leadership of the institute on July 1, 2022, joining the faculty in the UW–Madison Department of Physics, with an affiliate appointment in Statistics. Brian Yandell, the David R. Anderson Founding Director of the data science institute, has served since 2019.

Cranmer arrived at data science through his contributions to the search for the Higgs boson, a fundamental particle that in the 1960s had been theorized to exist and is responsible for giving objects in the universe their mass.

Finding evidence for the particle required navigating enormous amounts of data generated by trillions of high-energy particle collisions. Cranmer developed a method for collaborative statistical modeling that allowed thousands of scientists to work together to seek, and eventually find, strong evidence for the Higgs boson in 2012.

Kyle Cranmer stands next to a statue of Einstein sitting
Drawing on his experiences reaching across traditional academic boundaries, Cranmer aims to build partnerships between people working on data science methodology and those working in the humanities and the natural, physical and social sciences. | CONTRIBUTED PHOTO

“Shortly after the discovery, I pivoted to thinking more broadly about data science and machine learning for the physical sciences, identifying synergies and opportunities, and shaping that discussion internationally,” says Cranmer. His research has expanded beyond particle physics and is influencing astrophysics, cosmology, computational neuroscience, evolutionary biology and other fields.

At NYU, Cranmer is executive director of the Moore-Sloan Data Science Environment, associated faculty at the Center for Data Science, and is affiliated with the core machine learning group. His awards and honors include the Presidential Early Career Award for Science and Engineering in 2007 and the National Science Foundation’s Career Award in 2009. He was elected a 2021 Fellow of the American Physical Society.

Understanding and addressing the impact data science has on society, and the disproportionate effects it can have on marginalized people, is central to Cranmer’s vision.

One of Cranmer’s goals for the American Family Insurance Data Science Institute is to broaden engagement in data science across campus. Drawing on his own experiences reaching across traditional academic boundaries, he aims to build partnerships between people working on data science methodology and those working in the humanities and the natural, physical and social sciences. Understanding and addressing the impact data science has on society, and the disproportionate effects it can have on marginalized people, is central to his vision for this work.

“Issues around equity, inclusion and bias, and how that impacts society, those are very real problems I think everyone can appreciate,” says Cranmer. “But the way that they manifest themselves technically is much more subtle. Raising awareness of just how subtle and challenging those problems are, I think, is going to be useful for broadening the discussion across campus.”

Cranmer grew up in Arkansas and was in the first graduating class of a public, residential high school for math, science and the arts. He describes the school as a “melting pot” where students interested in computer science, physics, math and engineering collaborated on projects that today might be considered data science. Frustrated by a lack of extracurricular activities at his brand-new school, Cranmer got involved in school politics and student government.

“That was one of my first calls for leadership,” he says. “I came into the school and there was nothing set up at all — no student clubs, no activities. That was a very influential moment for me — realizing that you can be part of the solution and shape the environment around you to make it better.”

Cranmer looks forward to connecting and sharing ideas with people and research centers at UW–Madison. He stresses the importance of building trust, both within and outside the university, by demonstrating the potential for data science to positively affect people’s lives and the world.

“With experience as a national leader in data science, Kyle is well prepared to guide the institute in partnerships with enterprise thought leaders,” says Wendt. “His own research focus on data science methods that broaden participation to advance discovery in particle physics is truly rooted in the Wisconsin Idea.”

For Cranmer, contributing to the Wisconsin Idea is an exciting aspect of his new role. He sees opportunities at UW–Madison to engage with the community in research, such as working with the Division of Extension and farmers on problems like agricultural sustainability, carbon capture and climate change.

“This kind of capability is very, very unique, and there are several different entry points for the Data Science Institute to be involved in such research,” says Cranmer. “The role of data science would be really compelling.”

Following his years of experience shaping the first wave of data science at NYU, Cranmer looks forward to leading an institute that is well positioned to have real-world impact.

“I think that’s a pretty exciting thing to be a part of.”

Does the behavior of the Higgs boson match the expectations?

yellow and orange cones radiate from a common center, with green dots around them

Note: This story has been modified slightly from the original, which was published by the CMS Collaboration. Their version has some nice interactive graphics to check out, too!

The standard model of particle physics is our current best theory to describe the most basic building blocks of the universe, the elementary particles, and the interactions among them. At the heart of the standard model is a hypothesis describing how all the elementary particles acquire mass. Importantly, this scheme also envisages the existence of a new type of particle, called the Higgs boson.  It took nearly 50 years, since its postulation, to observe the Higgs boson at the LHC experiments at CERN. It is strongly believed that the Higgs boson, the only scalar particle known to date, is a key to answer some of the questions that standard model cannot answer. Thus a detailed study of the properties of the Higgs boson is the order of the day. Often, specially at the LHC, one of the essential observables concerns the probability that a certain unstable particle is produced momentarily, albeit obeying the laws of nature. In experiments this production cross section is estimated using a specific decay final state of this transient particle in terms of the number of events over a given amount of time. The standard model predicts the cross section for the Higgs boson production as well as the decay rates very precisely. The frequency distribution of a given type of event, as a function of some of the measured variables in the experiment, helps us understand better various aspects of the interactions involved; they are typically lost in the summed or total cross section. Hence measurement of this differential cross section is a powerful tool to vindicate the standard model; also any deviation from the standard model predictions in data would indicate presence of a New Physics.

The Higgs boson is roughly about 125 times more massive than a proton and decays to lighter particles including cascade processes in some cases. Physicists typically use the signatures of stable particles in the detector to trace back suitable decay chains of the Higgs boson. The tau lepton is the heaviest lepton known so far, and as such it is the lepton with strongest ties to the Higgs boson. The probability of a Higgs boson decaying to a pair of tau leptons is reasonably high (about 6%), when compared, for example, to a pair of muons (about 0.02%). But the tau lepton is also an unstable particle and decays quickly to lighter particles always accompanied by its partner, the tau neutrino. Often the decay products from the tau lepton are hadrons producing a shower of particles or jet in the calorimeter system. The tau neutrino goes undetected affecting the accuracy of measurement of the tau lepton energy. It is interesting to study the detailed characteristics of the Higgs boson events using the decay to tau leptons which possess a rest mass of only about 1.4% that of the parent.

profile photo of Andrew Loeliger
Andrew Loeliger

A recent study from the CMS Collaboration, focuses on the events where the Higgs boson decays into a pair of tau leptons using data collected by the experiment between 2016 and 2018. The analysis measures the Higgs boson production cross section as a function of three key variables: the Higgs boson momentum in the direction transverse to the beam, the number of jets produced along with the Higgs boson, and the transverse momentum of the leading jet. New Physics could manifest in excess of events in the frequency distribution of these variables when compared with the standard model predictions.

Says Andrew Loeliger, a UW–Madison physics grad student and one of the lead authors on the study:

The Higgs Boson is the most recent addition to the standard model of particle physics, discovered jointly between the CMS and ATLAS collaborations in 2012, so a big goal of the High Energy Physics field is to make very detailed measurements of its properties, to understand if our predictions are all confirmed, or if there is some kind of new physics or strange properties that might foreshadow or necessitate further discoveries. This work provides, what amounts to, a very fine grained consistency check (alternatively, a search for deviations in the amount) that the Higgs Boson is produced with the amounts/strengths we would expect when categorizing alongside some second interesting property (the transverse momentum of the Higgs Boson is a big one). This type of analysis had not been performed before using the particles we used, so it may open the door for far more precise measurements in places we may not have been able to do before, and a better overall confirmation of the Higgs Boson’s properties.

Other UW–Madison researchers involved in the study include former postdoc Cecile Caillol and Profs. Tulika Bose and Sridhara Dasu.

The analysis employs deep neural networks to exploit simultaneously a variety of tau lepton properties for identifying them with high efficiency. Eventually, to ensure that the selected tau lepton pair is produced from the decay of the Higgs boson and discard those from other processes, such as Z boson decay, the mass of the selected tau pair (m𝝉𝝉 ) is scrutinized. Reconstruction of m𝝉𝝉 , after taking into account the neutrinos involved in the decay as mentioned earlier, required a dedicated algorithm which computes, for each event, a  likelihood function P(m𝝉𝝉) to quantify the level of compatibility of a Higgs boson process.

yellow and orange cones radiate from a common center, with green dots around them
Higgs boson produced in vector boson fusion and decay to tau pair | credit: CMS Collaboration

The Higgs boson typically has more transverse momentum or boost when produced in conjunction with jet(s), compared to the case when it is produced singly. One such event, collected by the CMS detector in 2018 and shown in Figure 1, could correspond to such a boosted Higgs boson decaying to two tau leptons which, in turn, decay hadronically. However, several other less interesting processes could also be the cause of such an event and pose as backgrounds. Such contributions have been measured mostly from the data itself by carefully studying the properties of the jets. Figure 2 shows the good agreement in the m𝝉𝝉 distribution between the prediction and data collected by the CMS experiment for the events with the transverse momentum of the Higgs boson below 45 GeV. The contribution from the Higgs boson process is hardly noticeable due to the overwhelming background.  On the other hand, Figure 3 presents m𝝉𝝉 distribution for the events with highly boosted Higgs boson, when its transverse momentum is above 450 GeV.  Selecting only events with high boost reduces a lot the total number of available events, but  the fraction of the signal events in the collected sample is significantly improved. The data agrees with the sum of predicted contributions from the Higgs boson and all the standard model background processes.

This CMS result presents the first-ever measurement of the differential cross sections for the Higgs boson production decaying to a pair of tau leptons. Run 2 data is allowing us to scrutinize the Higgs boson in the tau lepton decay channel which was only observed a few years back. Future comparison and combination of all Higgs boson decay modes will offer better insights on the interactions of the Higgs boson to different standard model particles. But the story does not end here! The Run 3 of the LHC machine is just around the corner and looking into the future, the high luminosity operation (the HL-LHC) will offer a huge increase in data volume. That could perhaps provide hints of the question if the discovered Higgs boson is the one as predicted by the standard model or if there is any new interaction depicting another fundamental particle contributing to such measurements. That will indeed point to New Physics!

Willy Haeberli remembered as physicist, teacher, and museum supporter

photo of Willy Haeberli
photo of Willy Haeberli
Willy Haeberli in 2013 | Credit: Pupa Gilbert

University of Wisconsin–Madison Professor Emeritus Willy Haeberli passed away October 4, 2021. He was 96.

Born in Zurich, Switzerland on June 17, 1925, Haeberli received his PhD from the University of Basel (Switzerland) in 1952. He joined the faculty of UW–Madison in 1956, retiring in 2005.

Haeberli was a world-class experimental nuclear physicist. His research focused on studying spin effects in nuclear processes and in fundamental interactions. He and his collaborators developed spin-polarized gas targets of atomic hydrogen and deuterium. These “Haeberli cells” were used in many experiments worldwide including the Indiana University Cyclotron Facility, Brookhaven National Laboratory, and DESY Laboratory in Germany, and they were crucial for the success of those experiments.

Haeberli was the Raymond G. Herb Professor of Physics and a Hilldale Professor. He was elected to the American Academy of Arts and Sciences and the National Academy of Sciences, and he won the American Physical Society’s Bonner Prize in nuclear physics in 1979.

In addition to his scientific achievements, Haeberli was an accomplished teacher. He taught physics courses at UW–Madison for 49 years and developed the popular course Physics 109: Physics in the Arts, with Prof. Ugo Camerini. Physics in the Arts has been offered successfully and continuously since 1969, and has been emulated by tens of universities across the country. In the last five years before retiring, he co-taught the course with Prof. Pupa Gilbert. After he retired, Gilbert convinced him to co-write a textbook for Physics in the Arts, published by Academic Press-Elsevier in 2008, and 2011, translated into Chinese and published by Tsinghua University Press in 2011.

“Willy is a giant in my life. He was career changing, life changing, teaching changing, everything. Just the most amazing person I could have ever met,” Gilbert says. “He was, until the last day, my best friend ever, and the closest thing to a father figure I have ever had.”

Gilbert says that Haeberli’s interest in Physics in the Arts may have stemmed from his musician days — he played the flute in a quartet in college — and his wife’s passion for the figurative arts. She continues:

He always loved a lot more the physics of sound compared to the physics of light and color. He and I had feisty disagreements about the physics of light, and I enjoyed every one of them. Very often before classes I would come up with questions, and he could always, always answer them and pacify me. The last one was last spring, when I was teaching sound, and started wondering: Okay, we know that the speed of sound changes dramatically with temperature, but does the frequency change too? In other words, does a tuning fork sound different indoors or outdoors in Madison’s winters? I looked into this seemingly trivial question and could not find any answer I could trust to be right. Until I asked Willy, who (of course!) knew the answer right away, and charmingly explained that the wavelength and the speed of sound vary with temperature for a guitar string or a tuning fork, but the frequency does not. I will miss these elegant answers tremendously!

Haeberli recently made a significant donation to the Ingersoll Physics Museum, which allows for new exhibits to be developed, allows for current exhibits to be improved, and helps fund the docents program which provides tours for visiting school groups. He and his late wife, Dr. Gabriele Haberland, also supported the Madison Museum of Contemporary Art, UW­–Madison’s Chazen Museum, and Tandem Press with generous gifts.

Several current and emeritus department members shared their memories of Willy. Please visit the Willy Haeberli tribute page to read those stories.

Many thanks to Profs. Pupa Gilbert and Baha Balantekin for helping with this obituary

Alex Wang selected for DOE’s Office of Science Graduate Student Research program

profile photo of Alex Wang
profile photo of Alex Wang
Alex Wang

Congratulations to physics graduate student Alex Wang, who is one of 65 graduate students selected for the Department of Energy’s Office of Science Graduate Student Research (SCGSR) program. He is a student in Prof. Sau Lan Wu’s group.

SCGSR awardees work on research projects of significant importance to the Office of Science mission and that address societal challenges at national and international scale. Wang’s research is part of the priority research area of High Energy Physics. Awardees carry out part of their research at a DOE national lab, and his host lab is the SLAC National Accelerator Laboratory.

“The DOE Office of Science provides the scientific foundation for solutions to some of our nation’s most complex challenges, and now more than ever we need to invest in a diverse, talented pipeline of scientists, engineers, and entrepreneurs who can help us build a brighter future,” said Dr. Harriet Kung, Deputy Director for Science Programs in the Office of Science. “These outstanding students will help us tackle mission-critical research at our labs as this experience helps them begin a successful and rewarding career.”

Awardees were selected from a diverse pool of graduate applicants from institutions around the country. Selection was based on merit peer review by external scientific experts.