As mechanical orreries have been used to simulate movements of the heavenly bodies, so quantum systems can be used to simulate models of physical systems such as superconductors. A point emphasised in the new review is that the simulation is of a model: the performance must still be checked against the behaviour of the real system. For example, researchers setting out to simulate superconductivity will really be simulating the 'Fermi-Hubbard model': some expect to find that it's a good description for how superconductors work, others to find that it's not. Image: Wellcome Library, London (ICV No 25150) CC-BY 4
Quantum systems that can solve otherwise intractable problems through simulation are touted as one of the most promising technologies emerging from quantum research programmes. But what is required for a simulation to count as quantum? For a system's operation to count as a simulation? For the simulation to count as successful? For it to be trusted? These are questions that continue to stir debate. Researchers from the Centre for Quantum Technologies in Singapore and the University of Oxford, UK, tackle them in a new review.
"What is a quantum simulator?", a concise 12-page review by Tomi Johnson, Stephen Clark and Dieter Jaksch, is published 23 July in EPJ Quantum Technology.
The article appears as part of a thematic series on quantum simulation, edited by CQT's Dimitris Angelakis.
Learn more from Tomi, the paper's first author, in this Q&A:
Tomi: There are questions – not unimportant ones either, like 'how does a superconductor work?' – for which you'd have to wait an unimaginably long, long time for your normal computer to come up with an answer. We'd all be stardust by the time it happens. But with a quantum simulation, looking at how a few well-controlled atoms move in a lattice of laser light, say, we might get the answer out during our working lives.
We wrote it because we felt we had a perspective on the topic that was important to get across. Quantum simulation is a really hot topic right now. There are lots of theorists proposing ways to simulate, and as many experimentalists building simulators in state-of-the-art labs. Things seem to be at a transition point: we are able or on the verge of being able to learn things using quantum simulations that we simply could not learn in any other way, such as with a normal computer. To steal a popular phrase: with great power comes great responsibility. The very fact that quantum simulators are becoming so powerful means that it's really time we started asking hard questions about what it means for a simulator to be good or useful. Despite physicists spending so much time doing simulations, it seems the answers to such questions are not always obvious, and are sometimes even contentious. When we tried to answer them, we eventually were forced to go back and address even more basic questions, such as what is a simulator, how is it different from a computer, how is one used, and even what it means for a simulator to be quantum.
Stephen, Dieter and I had quite a few discussions, debates and perhaps even arguments in trying to pin down sensible answers to these questions. You could have caught us discussing it during breaks at Spinelli's cafe, dinner at the NUS Staff Club, and over lunch at the Senior Common Room of Keble College back in Oxford. While there's no such thing as a definitive answer to such questions of definitions, we eventually felt we had a pretty clear and sensible set of answers.
Definitely. We went into writing the review with some clear ideas, thinking it was just a matter of committing these thoughts to paper. While many of our initial ideas survived, we were surprised by the number of basic aspects that when addressed were suddenly hard to pin down clearly, drawing us in deeper and forcing us to revise or clarify some of our basic assumptions.
Hopefully our review can shape how people judge and even plan future simulations, as well as providing a nice summary and introduction for someone who wants to know what quantum simulations are all about. We've already had some reactions that were great to receive. For example, one colleague was pleasantly surprised to have their perspective on quantum simulation changed by reading the paper. They had the basic idea of quantum simulation as one quantum system mimicking another. But they hadn't previously appreciated the important intermediary role played by a theoretical model in this process, or the possibility of mimicking a classical system. Such things really do make a difference to how one assesses and evaluates simulations.