Giving it her all in both athletics and science
Bettina Heim has succeeded in publishing the results of her semester project in one of the most prestigious scientific journals. Heim was able to show why in certain tests current quantum computing devices were no faster than conventional computers, contrary to previous assumptions. ETH News met the former top athlete and now successful scientist.
When Bettina Heim does something, she gives it 100 percent. Until just a few years ago, she was a figure skater. She was Swiss champion, participated in world championships and completed her training to become a coach. Three years ago, she hung up her skates and has been studying physics at ETH Zurich ever since – with the same level of absolute commitment. For her semester paper during her bachelor’s studies – usually a limited project that takes about three weeks – the work became so involved that she was able to publish the results in the renowned scientific journal Science.
During the interview, Bettina Heim of course praised the team that helped her with the project, headed by Matthias Troyer, Professor of Computational Physics. She reports receiving excellent support from her colleagues and that they made a major contribution to the project. Nonetheless, it is extraordinary for any scientist to have the very first scientific publication of their career in such a prominent journal – and at the age of 26 during their undergraduate studies. “I had to publish more than 30 papers before such a highly renowned journal would accept one from me. I was 34 at the time,” says ETH professor Troyer, who supervised the talented student.
Quantum computing machine no faster
In her research, Heim was able to explain why D-Wave, a highly discussed novel computing device that uses quantum physical effects, is unable to solve certain computational problems any faster than a conventional computer.
Based on simulations, many experts expected that certain mathematical optimisation problems would be solved faster with quantum computing devices, such as D-Wave. Scientists refer to this acceleration, which results from quantum effects, as ‘quantum speedup’. Mathematical problems that may benefit from this effect include flight route planning and optimisation of securities portfolios. And these promising predictions were of course the main reason why the Canadian company D-Wave built one of these so-called quantum optimisers.
Simulation have their limits
However, in tests the device did not prove faster. In the case of certain tasks, it was even slower than a conventional computer, as Troyer demonstrated last year. While this may have won the ETH professor a gallon of Canadian maple syrup in a bet with science bloggers, it also revealed a supposed contradiction between the simulation and the experiment (the device). Heim took a closer look at the simulations and was able to resolve the apparent contradiction: the simulations used to reproduce the quantum effects on conventional computers have their limits, she concluded.
“You can think of the simulation as a system of particles that lie in a landscape of mountains and valleys,” she explains. “The optimisation problem that has to be solved is to find the lowest point of the landscape; the particles can either climb over the mountain or – if you allow for quantum effects – tunnel beneath them.” In the simulation, the path travelled by the particles is simplified to a zigzag path defined by a limited number of milestones.
‘Quantum-inspired’ computing
It was shown that quantum effects accelerated solving the optimisation problem in the simulation, but only if the simulation was very rough; that is, when Heim performed them with a small number of milestones. When the simulation included a large number of closely positioned milestones – a very realistic scenario – the quantum speedup was no longer observed. “Rough simulations like the ones performed by scientists in recent years thus do not reflect the reality in true quantum optimisation devices such as D-Wave,” says Troyer.
Still, these rough quantum simulations can be used to solve certain optimisation problems very quickly on conventional computers without the need to rely on quantum devices. These quantum simulations are therefore significant. Using them to solve optimization problems, however, would not be described as quantum computing but rather as ‘quantum-inspired’ computing, explains Troyer. “At present, it remains to be seen whether optimisation problems can be solved more efficiently using quantum optimisers or quantum-inspired computing on conventional computers. I predict that there will be an interesting rivalry between these two approaches in the years ahead.”
Quantum acceleration as a theoretical possibility
Even if the scientists have not detected any quantum speedup so far on the current generation of quantum computing devices, this certainly does not mean that such a speedup is theoretically impossible. It is possible that a quantum physical acceleration might emerge with a future quantum optimiser of a different design; for example, a device in which quantum particles not only interact with particles in their immediate proximity but also with particles located further away. “It will be interesting to test the potential of new quantum device architectures in a simulation before actually building them,” says Troyer. Heim’s simulation could do just that.
The ETH student is proud to have made a contribution. “I’ve been interested in math and physics for a long time,” says Heim, who completed her final school exam at the top of her year with an average grade of 5.7 out of 6. Even during her time as a top athlete, she was certain that she wanted to study one of these two subjects in the future. “My current interests are definitely in the field of theoretical physics.” The work done in Troyer’s group is described as traversing the point of intersection between theoretical physics and computer science. Heim appreciates that she is able to work on theoretical questions that have a strong connection to reality and which can be tested directly in simulations.
Basic attitude helps when studying
Is there a connection between professional sport and academia? The two worlds are quite different, says Heim. Nevertheless, she reports that her studies benefited from skills that were also important in professional sport; for example, she is good at focusing on the moment. While working on her Bachelor’s degree, Heim was still working part-time as a figure skating coach. She says that at the time it was important to focus on sport while doing sport and on her studies while studying.
She continues to benefit from this same basic attitude she had as a professional athlete. “I’m the one who is responsible for what I do. And if I want to achieve something, no one can stand in the way other than myself,” she says.
While she completes her Master’s degree, Heim continues working in the group led by Professor Troyer at ETH. At some point, she hopes to earn her PhD in theoretical physics or computer science. Thus, it is quite possible that she will publish other high-profile research articles in the years to come.
Literature reference
Heim B, R?nnow TF, Isakow SV, Troyer M: Quantum versus classical annealing of Ising spin glasses. Science, 12 March 2015, doi: external page 10.1126/science.aaa4170