Dugan O’Neil and his co-workers were part of a group of particle physicists running an experiment called DZero. By the Fall of 2006, DZero had been studying top quarks for 11 years, but even after all that time had failed to find the elusive single top quark it had been looking for. Dugan O’Neil couldn’t care less about the seemingly endless drizzle outside his Burnaby, BC office. He and his graduate students were about to try a new computer program that might finally “see” the top quark buried within the data from DZero.
Particle physicists want to understand fundamentally what matter is made of at the smallest scales. They have learned that our world consists of tiny units, elementary particles such as electrons and quarks. The top quark, also called ‘the top’, is one of six known quarks. So far, however, the top had only ever been observed in pairs. The DZero physicists wanted to prove that the top can also exist in isolation, as a single top quark.
The trouble with finding the single top was that it emitted a very weak signal. Theory predicted exactly what that signal would look like in a particle detector but the signal was so weak and the background noise so intense that nobody had ever been able to detect it. Even though theory had been accurate so far, there was no experimental proof that the single top existed. It needed to be found to validate the theory. The scientists felt their best bet for finding the single top was by designing a smart computer program that will be able to identify the top within the overwhelming noise.
But what computer program should they use? No one knew for sure. O’Neil had a hunch, though. He attended a conference a few years back and something there caught his attention. Part of the conference was dedicated to a programming method called a ‘decision tree’. Decision trees specialize in separating signal from noise in a relatively simple and intuitive way. O’Neil felt that he could actually understand decision trees, and found this utterly refreshing in the otherwise quite convoluted world of advanced analysis techniques.
O’Neil thought that decision trees might even be better than a popular competing computer technique called neural networks. Decision trees are easier to understand and optimize, and they run very fast. But nobody had really proven that they could solve such a difficult problem. Working feverishly, the physicists spent many late nights at Simon Fraser University, training their decision trees with simulated data sets. Then, on that rainy night in 2006, the time had come to run the ultimate test. O’Neil was sitting beside his graduate student and post-doc in their office and the three were polishing the last bits of their program. So far, they had refused to look at any preliminary results the program might produce; they wanted to be sure everything was perfect before the final run. Finally, O’Neil’s voice broke the silence. He felt they were now ready to give it a try.
“So shall I just, you know, shall I run it?” he asked. “If I run it, the answer is going to come up. Shall I do it?”
“Do it,” his post-doc replied.
And so he did. He hit the button. The three held their breath until a number popped up on the screen in front of them. The number showed that they had a signal - just what they had been hoping for. After rigorous cross-checks from their DZero colleagues and a long internal review, they finally published a paper describing the first experimental evidence for the existence of single top quarks.
As A Young Scientist...
O’Neil always knew that he wanted to be a physicist. He wasn’t the type of kid who liked to take apart radios, but he was captivated by science fiction and curious about everything that had to do with space. To understand these things, he concluded that he needed to learn the underlying physics. So when it was time to leave high school, the future was crystal clear. He would sign up for a physics degree at the University of New Brunswick. Studying physics had more to offer O’Neil than simply understanding the nature of things. For a boy growing up in a small town in rural New Brunswick, planning to become a physics professor was an exotic goal. “Most people work in much more industrial settings where I come from. So it was a bit weird. Maybe it attracted me for that reason,” he says.
He ended up majoring in particle physics almost by chance. At 19 O’Neil needed a summer job so he applied for research positions all across Canada. Michel Lefebvre, a young physics professor at the University of Victoria offered him a position. The invitation came as a bit of a shock. O’Neil had not expected it. He pulled out a map and was thrilled to discover that Victoria was on an island, a good place to spend the summer. Working with Lefebvre turned out to be a defining period in his career. Lefebvre became O’Neil’s mentor and PhD advisor and O’Neil has been working in particle physics ever since.
That same innocence and modesty still radiate from O’Neil’s boyish smile even today as an Associate Professor at SFU. When asked to describe himself, he says, “I have quite a calm personality. I rarely get angry, and I rarely get upset. But when we’re talking about things that I am very interested in, I get very enthusiastic about them.”
O’Neil has never regretted going into particle physics, although the field has its challenges. Most elementary particles only exist at very high energies and physicists obtain these levels of energy by accelerating matter to enormous speeds in gigantic circular underground tunnels. The Large Hadron Collider, the biggest particle accelerator in the world, has a diameter of 27 km. Because of the sheer size of the experiments, particle physics can only be done in huge collaborations. O’Neil typically works in groups of up to 3,000 scientists. Organizing this many researchers means rules galore and that can eat away at the freedom most professors enjoy. O’Neil however loves working in this environment. After all, collaborating on big particle physics projects means working with some of the smartest people in the world, something he would not want to miss.
Written by Sigrid Auweter. Based on personal communication, http://www.fnal.gov/, and http://www.particleadventure.org/.
What is the world made of?
This question is as old as mankind and scientists have long aimed to find the basic building blocks that all matter is composed of. What they have been looking for is something that they call ‘fundamental’, meaning that it cannot be further divided. Initially, atoms were believed to be fundamental, but physicists soon discovered that atoms are made of electrons and a nucleus. The nucleus itself is made of protons and neutrons. And protons and neutrons are made of quarks.
As far as we know today, quarks and electrons are fundamental. They are also called ‘elementary particles’ because (so far) there’s nothing smaller.
Image 1: Atom model (Image provided by: The Particle Adventure from the Particle Data Group (LBNL))
How many elementary particles exist in total?
The standard model of particle physics states that there are 6 quarks, 6 leptons, and several force-carrying particles (bosons). The electron is one of the leptons. The photon, the particle associated with light, is one of the bosons.
How do physicists study elementary particles?
Most of the elementary particles exist in isolation only at very high energy. To reach these energies, physicists use particle accelerators. These are large, often circular, underground tunnels in which particles, such as electrons and protons, are brought up to enormous speed using high vacuum and high-powered superconducting magnets, and then they are made to collide with each other. In these high-energy collisions, elementary particles are created. They can be observed in detectors that surround the collision site.
The collision takes place in the center causing particles to fly off in all directions spreading throughout the different layers of the detector. The moving particles trigger electronic signals that are recorded and analyzed by powerful computers. Each elementary particle gives rise to different characteristic signals.
As of 2010 O’Neil is working on an experiment called ATLAS. ATLAS is one of the experiments run by the European Organization for Nuclear Research (CERN), which operates the Large Hadron Collider: the largest particle accelerator in the world located hundreds of feet underneath the border between France and Switzerland. Since March 2010, beams of subatomic particles are accelerated in this tunnel and collided (potentially) at a record-breaking energy of 7 TeV. O’Neil is one of the physicists analyzing the data coming from the ATLAS detector.
Image 2: Schematic of a particle detector, the ATLAS detector at CERN
Image 3: Photograph of the ATLAS detector
O’Neil’s hunt for the single top quark
By 2006, experimental evidence for most elementary particles had been found with the help of particle colliders. The top quark, however, had been observed only in pairs. Theory predicted that it could also be produced in isolation, as the single top quark. An experiment called DZero, which collected data at a particle accelerator located in Batavia, Illinois, was specially designed to find the single top quark.
Initially, however, physicists could not find any evidence of the single top quark in their detectors. This was because the signal that the single top quark emits was very weak and was hidden among background noise. New computing techniques were necessary to differentiate between signal and noise.
Boosted Decision Trees
O’Neil overcame this challenge by designing a computer program called a ‘boosted decision tree’.
Decision trees are programs that separate things based on distinguishing features. For example, imagine you had a pile of apples and you wanted to separate Galas from Granny Smiths and Red Delicious. The most distinguishing feature among the three -given you bought them at a store- is their label, so you would choose to read the labels to separate them.
Now imagine you (the detector) cannot read labels. This means that you have to find other features you can use for separation. You might come up with distinguishing traits such as color, shape, and taste, which will also allow you to eventually separate the apples. This is exactly what O’Neil’s decision trees had to do. They had to sort out signal from noise without a very strong distinguishing feature, but rather using many weak distinguishing traits.
Figure 4: Decision tree
This was still not good enough to solve the problem, though. A single decision tree gave a first description of what might be noise and what might be signal, but there was a lot of data coming from the detector that the first tree could not classify. So O’Neil and his group took the data that the first tree had most difficulties with and fed it into a second tree, a tree that used different distinguishing features. And the data that still proved problematic after the second tree was fed into a third tree and so on. In the end, the physicists had a little grove of 20 decision trees that succeeded in identifying the single top quark signal. This technique is called a ‘boosted’ decision tree and O’Neil and his co-workers were the first to prove that it can be successfully used to identify a weak signal in a particle physics detector.
To learn more, click on the links below:
CERNland (link to https://project-cernland.web.cern.ch/project-CERNland/)
Masterclasses (link to http://physicsmasterclasses.org/neu/)
The ATLAS trilogy:
A New Hope
The Particles Strike Back
Today O’Neil is using boosted decision trees to count tau leptons, elementary particles that were first discovered in the 1960s. So finding their signal will be straightforward this time. The challenge will be to count their exact numbers. O’Neil hopes to find a higher number of taus than expected relative to other particles, as this could be the signature of a new direction in physics. It could be a sign of the Higgs boson, for example, a particle that would explain the origin of mass and the only elementary particle that has not yet been observed. Physicists do not know if the Higgs boson exists, or if there are several different types of Higgs Bosons, so these are all mysteries waiting to be uncovered. In addition to the origin of mass such discoveries could explain the existence of dark matter and dark energy which make up 95% of the universe based on current theories. That is: 95% of what is out there is a mystery to modern science. We simply don't know what it is.
O’Neil suggests looking for a career that both challenges and interests you, rather than simply going for what is easiest or best paid. Difficulties and curiosity will ensure that you stay motivated and that you become good at what you are doing.
- Family Members
- Spouse: Melanie Russell-O'Neil
- calm and enthusiastic
- Favorite Music
- The Tragically Hip, any kind of rock and roll, garage band music
- Other Interests
- tinkering with computers, reading, ball hockey, hiking
- Associate Professor
- Simon Fraser University
- BSc New Brunswick University
- MSc U. of Alberta
- PhD U. of Victoria
- 2007 British Columbia Innovation Council Young Innovator Award
- 2007 Simon Fraser University Faculty of Science Excellence in Teaching Award
- Michel Lefebvre, professor at the University of Victoria, introduced O'Neil to particle physics. 19-year old O'Neil spent a summer in Lefebvre's laboratory and was spellbound by Lefebvre's sheer enthusiasm for science. After many hours of drawing on the blackboard and talking physics with Lefebvre, particle physics was all O'Neil wanted to do.
- Last Updated
- September 17, 2015
Profile viewed 45216 times
Other scientists who may be of interest: