13 April 2023
Sitting at their fireplace, over which a stew is simmering, they are already working on a cookbook for their electric cooker – which still needs to be designed and built. The situation is similar in quantum computing at the moment. While the hardware is still being developed, researchers are already working on the operations that a quantum com-puter is to run, step by step, when solving certain tasks: the quantum algorithms. Prof. Frank Wil-helm-Mauch, director of the Peter Grünberg Insti-tute for Quantum Computing Analytics (PGI-12), believes that the simultaneous development of hardware and software is sensible and important: “If you only started working on algorithms when mature quantum computers already exist, you would have lost a lot of valuable time.”
The industry also sees it that way. “There are no competitive advantages in waiting until quantum technologies are a completely estab-lished technology. Essential new markets would then already be occupied,” warns the German information and telecommunications industry association, Bitkom, in its guide on quantum technologies in business. As a consequence, large corporations have long since begun to build up competencies. The challenge today is to find out for which problems a quantum computer would actually be of practical use and what suitable algorithms might look like.
Science and industry are working closely together to achieve this. “This way, we learn to understand the problems that are significant for the compa-nies. This helps us basic researchers to develop quantum algorithms – not only out of scientific curiosity, but very specifically to advance society this way,” says group leader Dr. Tobias Stollenwerk from PGI-12. In the Q(AI)2 project, together with BMW, Mercedes-Benz, Volkswagen and Bosch, Stollenwerk, his team and researchers at the Jülich Supercomputing Centre (JSC) are working on solutions for the automotive industry.
The focus is on tasks for which the companies use artificial intelligence (AI) – so far, just relying on conventional computers. For example, an AI is supposed to recognize workers and their body parts in pictures. This is often a prerequisite so that robots and humans may work together directly and safely during the production of a car part. The computer can then, for example, control the robot arm so that it does not collide with a worker. Dr. Dmytro Nabok from the JSC is exploring quantum algorithms that are to improve this AI application in such a way that it always identifies body parts correctly and within milliseconds.
Nabok uses, among other things, the quantum annealer from the company D-Wave Systems at Forschungszentrum Jülich, called JUPSI, to test these algorithms. Unlike other quantum comput-ers, a quantum annealer is not universally pro-grammable, so it is only suitable for special tasks. JUPSI is part of the “Jülich UNified Infrastructure for Quantum computing” (JUNIQ). Through JUNIQ, research teams from industry and academia have access to experimental systems, prototypes and commercial quantum computers, as well as the necessary support, for example, in developing algorithms. Prof. Kristel Michielsen from the JSC and head of JUNIQ illustrates how important it is to test and run quantum algo-rithms on the already existing quantum systems at an early stage. “This generates expertise and contributes to the co-design feedback loop,” she says. “In this process, which is known from supercomputing, users, software and hardware developers work together to improve and design future computing environments.”
Volkswagen AG is dealing with another, exempla-ry problem in the car industry. The starting situ-ation: new cars are to be painted in two separate coating layers. When a car arrives at the coating line, however, the robots there do not apply the two layers directly one after the other. If they did, the robots would have to constantly change the paints, which is time-consuming. This is why a conveyor belt transports each car out of the line after the first painting process and then back in again for the second painting process. This process is to be optimized, that is, the time of a paint change for a certain number of cars must be determined in such a way that the paint needs to be changed as rarely as possible. Tobias Stollen-werk’s team is investigating whether quantum computers can solve such problems faster than conventional computers.
“The computational burden of this problem increases enormously fast relative to the number of paints and the number of cars. In addition, we have to take other influencing factors into account in practice,” explains Stollenwerk. “Sometimes, optimization problems like these cannot at all be solved accurately in realistic time spans, not even with supercomputers.” Unlike Nabok, Stollenwerk’s team tests the algorithms on conventional computers that mimic the behav-iour of quantum computers. “This allows us to specifically co-simulate different types of errors that quantum computers of different maturity levels make. This includes checking the effects of the quantum hardware, for example. This helps us to better understand the influence of these errors on the result,” says Stollenwerk.
JUNIQ project manager Dr. Nils Küchler tells about a pleasant side effect: “Conventional com-puters are not only useful for testing quantum software. They can also benefit from it, because with them, the quantum-inspired software sometimes delivers new solutions.” Stollenwerk has observed another phenomenon during his collaboration with industry: “Companies are shedding new light on problems by considering how quantum computers could calculate these. In the process, they sometimes discover far better classical algorithms than those they had previ-ously known.”
Dr. Daniel Zeuch from the Peter Grünberg Institute (PGI-12) has spoken to over 100 representatives of small and medium-sized enterprises (SMEs) about quantum computing over the last two years. Why, Mr Zeuch?
We wanted to find out whether SMEs can be increasingly involved in the development of components or software for quan-tum computers in the future and, if necessary, help them to invest in this field of development in the near term. We also wanted to know to what extent smaller companies had already familiarized themselves with the topic.
What were the results?
There are marked differences. Some suppliers who produce special power sources or microwave analyzers, for example, are already in the business, others could enter quickly. Users, on the other hand, are still holding back. They lack robust forecasts of computational benefits from quantum computing. Nevertheless, some of them expressed great interest in dealing with the topic, for example in conversations with us.
What do you offer these companies?
Information, advice and support. They are welcome to contact me at any time. Forschungszentrum Jülich is also pursuing various approaches to strengthen contact with industry, including SMEs, for example as a partner in the state-wide network “EIN Quantum NRW”, or with the Aachen Fraunhofer Institute for Laser Technology ILT in the planned Center for Quantum Systems and Engineering, or CQSE for short, which aims to promote quantum technologies in the Rhineland region.
Text+Interview: Frank Frick | Images: Forschungszentrum Jülich/Sascha Kreklau; Sofia Wagner, DLR; Radhika Vaidyanathan | Video: Forschungszentrum Jülich
