The question how to realise and to assess the power of (programmable) quantum simulators is very close to my heart. Quantum simulators promise to allow for new insights into strongly correlated quantum matter (and presumably also for industrially relevant use cases).
But before this aim can be fully achieved, we have to find ways to clearly define what the quantum simulator is actually doing precisely. Therefore, ideas of benchmarking, certification and the seemingly innocent question of read-out are at the centre of our activities. Only if we have predictive power, quantum simulators can be seen as genuine quantum technological devices. And this is far from trivial: After all, quantum simulators outperform classical simulation methods, so oftentimes, we cannot simply check whether the quantum simulation has been right.
Some of the questions we tackle within PASQuanS are very hands-on, practically minded, and provide tools relevant for the consortium as a whole. Here, we see ourselves as service providers for the entire network. Other questions are rather visionary aiming to give the project a medium-term scope and perspective.
The project has so far been enormously productive for us. We have gained a much deeper understanding of how we can benchmark and verify quantum simulators, a topic that is very important for PASQuanS. Plus, we have helped develop quantum simulations of non-equilibrium phenomena presumably beyond the classical realm, and developed new classical tensor network methods to identify applications and to be used in future benchmarking.
Maybe most importantly, we have identified quantum simulation architectures with a provable quantum advantage over classical supercomputers, and ones that can at the same time be verified. This has been one of the core tasks.
However, perhaps it is the unexpected applications, those we did not foresee at the beginning, that are the most exciting. These new directions render such a project particularly interesting. For example, that quantum devices have a provable scope for being better in certain learning tasks than classical ones. Or we have seen that there is an intermediate realm between analog and digital quantum simulation with applications in quantum chemistry – unexpected, but spot on to the aims of PASQuanS.
The interactions with others have been tremendously stimulating. This interaction is of key importance, as it measures theoretical ideas against what is possible and important in practice.
This division of labour works very well and we contribute complementing insights. And even in case of overlap, e.g., when developing methods of verification in Berlin and Innsbruck, the endeavours are highly mutually stimulating and beneficial. We focus more on benchmarking, while Innsbruck brings in new ideas related to self-verification. As for tensor networks, we bring in unique expertise in developing methods for higher dimensional quantum systems. Overall, the project has been going extremely well also in this respect.
It is the pragmatic attitude guided by experiments combined with deep thinking about potential applications of quantum devices that makes the project special. And we should not forget that quantum simulation is a field where Europe is at the forefront of things. Hence, in this area we are in the privileged position of being able to define research directions.
These are ambitious aims. To start with, we should identify quantum simulators as quantum technological devices, with clear predictive power and which have been properly benchmarked. Then, there are already lots of interesting applications for analog quantum simulation. In the long term, and with regard to the next steps, other applications will emerge. High energy physics is probably next, as the problems at stake have already been cast into a form amenable to programmable quantum simulators.
Moreover, there comes quantum chemistry: even though there is still too big a gap between basic variational eigensolvers and programmable quantum simulators to tackle interacting problems in quantum chemistry. We have worked hard to bring the effort down. Maybe the last application will be to find better approximations in combinatorical optimization problems. Here, we should be aware of the fact that it is all about approximation levels. Nobody expects quantum devices to efficiently solve NP-hard problems.
I see what might make it appealing to use powerful buzz words such as ”revolution”, and this can in instances provide some stimulating rhetoric. But we should not forget that our future prospects and advances have only little in common with a revolution. It is true, quantum devices offer an enormous potential for tackling exciting problems beyond the reach of supercomputers. But this will be a long and winding journey. It will require more expertise, stamina, ideas, and funding. And PASQuanS sets, to my understanding, precisely the right aims.
It does not promise hyped applications in the financial world, does not speak of quantum advantages in risk portfolio management. Instead, it sets out to bring analog quantum simulators to a new level, to understand them as devices with enormous predictive power, and to add elements of programming to them. For this, we have assembled a world-class team. And then, if these steps have been achieved, we go for the next steps. And if the applications laid out materialize, yes, then we are in some sense about to have a revolution.
Download the PDF version of this interview: PASQuanS_Interview_JensEisert