The Dean of Science Award honours Master’s students in the Faculty of Science who demonstrate outstanding performance. We sat down with the latest winner, IQC researcher Patrick Daley from the Department of Physics and Astronomy, to learn more about his award-winning research.
Congratulations on your Dean of Science Award, Patrick. Let’s start off by talking about your Master’s thesis. What did you research?
I studied quantum foundations under Kevin Resch, a professor in the physics and astronomy department, in the Quantum Optics and Quantum Information Lab and in collaboration with Rob Spekkens at the Perimeter Institute. The overarching goal of our research was to understand causality in quantum mechanics. We’re interested in causality for two reasons: conceptually, it allows us to understand what happens in the universe at the quantum level, but it is also of practical importance.
The classical causality we’re familiar with is useful because it allows us to predict what will happen even in circumstances we haven’t observed before. This is helpful if running an experiment is expensive or difficult. With an understanding of cause and effect, we don’t have to waste resources trying different options. We can predict what will be most worthwhile.
We don’t have that level of understanding in quantum mechanics. We only have statistics that will let us predict things we’ve already seen. Causality will tell you more. It will tell you what will happen if you do something without having to actually do it.
What approach did you take to study causality in quantum mechanics?
We took the simplest example where we know that pre-existing classical causality fails: the Bell scenario where you just have one pair of entangled qubits. The most recent Bell experiments rule out the logical and obvious causal description, which undermines some very fundamental principles like locality, the principle that only objects close enough together can influence each other; their effects on each other can’t travel faster than light.
If we find the correct causal description of this basic quantum mechanical system, then we can better understand how causality works in quantum mechanics, and we can predict the outcomes of this system without having to exhaustively test them all. Very little experimental work has addressed this problem, so that’s what we worked on.
We developed a framework that allows you to assess the relative merits of different causal models using actual data from a lab. We can assess models that allow exotic new influences that can travel faster than the speed of light, superdeterministic models where there’s no freedom of choice, or models where we change the rules of what causal influence actually looks like. We set up an experiment in the lab with polarization entangled photons and used our framework to analyze the results.
What did you find?
Our data showed that we should favour a quantum mechanical model where the nature of causal influences themselves are altered, instead of adding additional influences. If we were to allow faster-than-light influences in our model, we predict things that we don’t see in nature. We can see that the model is unnecessary because it has extra flexibility. It fits noise in the datasets, and actually performs worse. We can’t prove the best model right now, but it offers evidence of which direction we should be investigating.
What makes your approach different from others?
The main focus until now has been finding causal descriptions that allow for the predictions of quantum mechanics to exist. But what if two or more models allow them? Then how do you decide between them?
I think the takeaway lesson is that you can actually measure overfitting, or too much complexity or flexibility in your causal model. You don’t see this kind of measurement as often in physics, but you see it often in the statistical learning and machine learning communities. There’s value in borrowing methods from other fields. You can actually go in the lab and find the downsides when you have overcomplexity. The things we do are not all that special outside of the physics community. It was just finding them and bringing them to these quantum foundations problems.
What challenges did you face during your masters?
Our project was not well-defined initially. We just wanted to explore what we could learn about causality. At first, we came at it from a different angle: a generalized probabilistic framework. It took some time before we realized we should borrow these statistical learning methods and apply them to our research
I was the only student working on this project, so a lot of the problems and questions that came up landed on me, and it’s hard to decide what options are worth your time as a master’s student. Other members of my group who weren’t familiar with the particulars of my project still listened to my problems and helped me talk it through, and I owe them a lot of gratitude for that. Kevin and Rob have also provided so much support, even after I graduated, so I can’t thank them enough for that.
What does the Dean of Science Award mean to you?
It is definitely exciting and encouraging. It helps to build my confidence in the academic world. You question why your ideas are even worth listening to because there is so much research out there, and I’m just a student. It’s exciting that my mentors, my peers and the general physics community at Waterloo found the work that I’m interested in and that I’ve done interesting and noteworthy.
What’s next for you?
Most of my time right now is spent working on the paper resulting from my thesis. Moving forward, I want to continue working on causality issues, though not necessarily with a focus on quantum mechanics. My plans will most likely involve pursuing doctoral studies, but at the moment I’m taking some time to review some opportunities. I want to make sure that whatever I do next has the most impact on the problems in our field that I think need addressing.