Reinventing computing with computers – researchers all over the world are pursuing this vision. For this purpose, they are constructing machines that conform to the sometimes bizarre rules of the quantum cosmos. Such quantum computers are likely to be significantly superior to conventional computers for certain tasks. Experts in this field are able to rely on Jülich for an excellent research environment.
A mysterious world hides deep inside matter – the quantum cosmos. It is governed by laws that are difficult to understand. The things lurking there can be connected through ghostly encounters from afar, both dead and alive at the same time.
Researchers at Jülich have set out to prise the secrets out of the quantum cosmos. One ambitious goal unites them all: they want to learn to control the smallest particles in order to construct a computing machine like the world has never seen before – a powerful quantum computer. This computer could help pharmacists and materials researchers in their search for new active ingredients and catalysts by simulating complex molecules. It could optimise logistics problems and production processes in factories in next to no time – faster than any high-performance computer.
So, the Jülich experts search for exotic materials, using them to design new types of circuits, which they then assemble into the first prototypes. They write singular algorithms and pit their computers against the fastest supercomputers in the world. Forschungszentrum Jülich offers them an optimal environment for their endeavour: modern laboratories and powerful equipment, plus partners from the industry and a local service infrastructure. However, regarding the success of the research site, it is the minds behind the research that will make the difference after all: a team of highly specialised experts who contribute their ideas, knowledge, experience, pioneering spirit and perseverance to achieve the common goal. We would like to introduce them here – who they are, what they have achieved so far and what visions they are pursuing.
Otherwise known as qubits, these are the counterpart to classic computer bits: they can be, for example, atoms or ions floating in a row like a string of beads in an optical trap; or they can be superconducting loops through which circular currents flow; they can even be individual electrons trapped in cages made of semiconductor materials, so-called quantum dots. What is crucial is that all these systems can exist in two states that correspond to the binary values of a classical computer: zero or one. Plus, qubits all have the properties of waves, meaning that they can overlap themselves. This state is called “being in superposition”. As a result, they are also able to assume all states between zero and one. The challenge, however, is to protect the fragile quantum state from external interference factors. The superposition must persist long enough to use the qubits to make calculations.
In the beginning, there is the idea: information can be carried by the smallest particles such as atoms, electrons or photons. If several of these quantum bits, or qubits for short (see box), are coupled together, countless calculations can be carried out in parallel. On quantum computers, certain algorithms require only a tiny fraction of the time that they need on conventional computers. One of the first to draw up this concept was David DiVincenzo.
- David DiVincenzo
“In the early 1990s, I became interested in computing with quanta. The scene was still rather small at that time,” recalls Prof. David DiVincenzo. A breakthrough, however, came as early as 1994. Mathematicians were able to show for the first time that fast quantum algorithms can actually solve relevant problems. “So they approached us physicists and asked if the quantum processors needed to do this could actually be built.”
Two years later, DiVincenzo defined the basic concept behind all computers of this kind, capturing it in the five criteria named after him. The first criterion describes the basic requirement of any computer of this type: it calculates with qubits instead of bits. “Nowadays, these principles sound like trying to explain to a kindergartener how a computer works. But at the time, we had no precise idea of what a quantum computer could look like,” explains the physicist, who is considered one of the major pioneers in quantum computing. He spent a large part of his career in a research laboratory of IT giant IBM. In 2010, he became JARA Professor at Jülich’s Peter Grünberg Institute (PGI-2 and PGI-11) and RWTH Aachen University. “Here, I have the resources and freedom to pursue my own research interests,” says the physicist. “I also appreciate the environment: the diversity of topics and thought approaches.”
In Jülich, he devotes himself to error correction: The unstable nature of the quantum state frequently results in errors slipping into the calculations. The physicist aims to remedy this: “We are currently pursuing the idea of dividing a qubit into three electrons sitting in a semiconductor cage,” DiVincenzo explains. If an error sneaked into one particle of the trio, it could then easily be checked with the other two.
From idea to implementation: quantum systems must first of all be found that can easily be controlled from the outside. They form the foundation – the smallest unit of a quantum computer. Stefan Tautz is one person who is looking for such materials.
- Stefan Tautz
Prof. Stefan Tautz can move individual molecules with his bare hands. He directs the tiny entities by swiping through the air. He detaches them from their cluster, shifts them, turns them sideways or arranges them on a surface as desired. Granted: the literal manoeuvring is done by a scanning probe microscope, whose fine probe tip manipulates the molecules. Infrared cameras record the hand movement in three-dimensional space and transmit it to the tip.
This technique of pushing molecules is not new in itself. Researchers at the Peter Grünberg Institute (PGI-3) are now harnessing it to search for new materials for future quantum computers, since individual molecules with the right electronic properties can be used as qubits. “With our microscopes, we can build targeted, customised structures very quickly, a kind of molecular 3D printing,” Tautz explains. “We then test them for their suitability as qubits.”
“Our excellent infrastructure, which we are expanding even further, such as in the Helmholtz Quantum Center (see box), helps us with all of this,” says Tautz. Much more important, however, are the people who come together here to work towards a common goal. “No one builds a quantum computer all by themselves,” says the researcher. “It takes countless experts with a variety of skills and academic backgrounds. Jülich offers the best conditions for this.”
Helmholtz Quantum Center
Whether atom traps, superconductors, semiconductors, Majorana particles or completely different types of qubits: it is not yet decided which technology will perform best. Accordingly, a central technology laboratory, which is currently being set up at Jülich, is designed to be just as versatile: the Helmholtz Quantum Center (HQC). “It covers the whole development arc from the materials to the quantum computer and the algorithms that will run on it,” says Stefan Tautz, who represents the scientists at the HQC. “It will provide a rich ecosystem where all aspects of quantum computing will be explored.” Therefore, the HQC building will accommodate countless specialised labs while offering a strong infrastructure. Because of the fragile nature of qubits, the experiments need to be conducted only just above absolute zero. The premises must be insulated against interferences from magnetic fields, high-frequency fields and vibrations.
Once materials have been found to create the qubits, several qubits must be connected to form small, switchable units. In the classic computer, transistors made of semiconductor materials take over this task. New concepts are required for quantum computers, which is what Hendrik Bluhm is working on developing.
- Hendrik Bluhm
“It is this pioneering work that excites me, this moon-landing character of the research,” says Prof. Hendrik Bluhm. “We are developing a technology with truly tangible benefits on a completely new basis, which knows no comparison in classical data processing.”
The physicist designs elementary components for a future quantum computer and investigates their properties. In order for quantum processors to be able to even exploit their advantage over classical high-performance computers, they must have a sufficient capacity of qubits. “For many of the planned applications, you need millions of qubits. There is still a long way to go before we get there,” explains Bluhm. It is therefore important to develop architectures that can be easily implemented on a large scale. To do this, he relies on so-called semiconductor spin qubits, which store information in the intrinsic angular momentum of individual electrons.
“For many of the planned applications, you need millions of qubits. There is still a long way to go before we get there.”
In Hendrik Bluhm’s view, these have important advantages: compared to superconducting systems, semiconductor qubits could be used to create much more powerful processors in the long term. To produce these, the standard processes from established semiconductor manufacturing can be deployed. “In addition, compared to superconducting qubits, they are less susceptible to external disturbances such as thermal radiation or the cosmic background radiation,” explains the JARA professor, who conducts research at Jülich’s Peter Grünberg Institute (PGI-11) and at RWTH Aachen University.
Several electrons are coupled together in order to calculate with semiconductor qubits, that is, their electron spins are entangled with each other. The tiny particles have to be brought close together for this. However, the more electrons that are to be connected, the more difficult this becomes, because then the distance also inevitably increases. Therefore, Hendrik Bluhm’s team is working on a way to move the entangled electrons back and forth across the chip as information carriers. With such an electron shuttle, it could be possible to connect a large number simultaneously (see QUASAR infobox).
Such alternative concepts would offer Europe the opportunity to catch up more easily with the market leaders. Jülich would play a key role in this: “Forschungszentrum Jülich has continuously expanded its competencies in quantum technology in recent years. The group of experts has become consistently larger and more diverse.”
QUASAR and QLSI
The principle of the electron shuttle is intended to be applied to the development of a German semiconductor quantum processor in the QUASAR project. The circuits that have already been successfully tested in the laboratory will be scaled up for this purpose with the ultimate outcome being a demonstrator with 25 coupled qubits. Individual electrons on a chip can be transported over longer distances in a controlled manner via an element referred to as a quantum bus. Other partners from academia and industry are also on board QUASAR. Semiconductor manufacturer Infineon, for example, will investigate how quantum chips can be realised with conventional silicon technology and manufactured on an industrial scale. Jülich is also involved in the European Quantum Flagship’s QLSI project, in which similar silicon-based quantum chips are being developed.
The processor of a computer has billions of tiny electronic switches, the transistors, to perform the numerous calculations. Scaling up from just a few qubits to integrated circuits is still a challenge in quantum computers. Frank Wilhelm-Mauch has chosen to take on this challenge.
- Frank Wilhelm-Mauch
Craftsmanship is not one of his virtues, admits Prof. Frank Wilhelm-Mauch: “If there’s a nail to be driven into the wall, it’s always my wife who does it.” So it was only natural that he turned to theoretical physics during his studies. This allowed him to deal with perplexing constructs. “On the other hand, I also found myself in the dichotomy of not simply wanting to explain the world, but wanting to build something as well. And quantum computing is the ideal field for that.”
At Jülich, the researcher from the Peter Grünberg Institute (PGI-12) is working on combining individual quantum systems into circuits like those found on conventional computer chips. There, they consist of individual transistors that are interconnected in such a way that they can perform fundamental computing operations. For quantum computers, Frank Wilhelm-Mauch is focusing on qubits in superconducting contacts, which are similar to the technology used by Google or IBM. “They provide the starting point of our quantum chips,” he explains. “We are developing new devices on this basis and are looking at strategies to make existing hardware simpler, sturdier and smaller.”
Frank Wilhelm-Mauch fondly remembers his first computer, a Sinclair ZX-81. The British calculating machine was one of the first home computers to come onto the market 40 years ago – even before the famous Commodore 64. “In terms of development, today’s quantum computers are not quite as far along as the ZX-81 was back then, but the feeling you get while programming is still comparable. It’s about getting the best out of limited hardware.”
Speaking of hardware, the theorist has in fact also been given the opportunity to build a unique prototype: the OpenSuperQ project aims to create a freely programmable quantum computer that is accessible to all researchers in Europe. Wilhelm-Mauch is coordinating the project (see infobox).
The OpenSuperQ project aims to produce the first freely programmable European quantum computer that is superior to conventional high-performance computers. For this, at least 50 qubits would have to be entangled with each other. The construction of the computer, which is based on superconducting quantum circuits, is coordinated by Jülich researcher Frank Wilhelm-Mauch: ten partners from science and industry are working together in the OpenSuperQ project. The open-architecture computer will then be remotely available to experts around the world as part of the Jülich quantum infrastructure JUNIQ at the Jülich Supercomputing Centre. OpenSuperQ is part of the European Quantum Flagship. Jülich researchers are also involved in other flagship projects.
Only for certain algorithms can quantum computers exploit their parallel computing advantage. This can occur, for example, when large amounts of data need to be sifted through or the properties of molecules and materials need to be calculated. In order for the corresponding programmes to run smoothly on the computers that are still error-prone, the machines need mature firmware. This programme level, which mediates between the hardware and the applications running on it, is the domain of Tommaso Calarco.
- Tommaso Calarco
Tommaso Calarco is a man of many talents. He holds a bachelor’s degree in classical guitar, among other things. The fact that he gave preference to quantum physics should be seen as a stroke of luck for the entire European quantum research community: the physics professor is considered the initiator of the Quantum Manifesto that led to the EU’s billion-dollar flagship programme. In it, he holds the office of Chairman of the Quantum Community Network (QCN), which brings science and industry together (see infobox).
In his research, he looks for ways to improve the basic processes and thus the calculation accuracy in quantum computing – as well as for all types of qubits. “In the quantum world, we use the smallest particles such as atoms and electrons. However, we manipulate them with macroscopic tools. It would be like if I played guitar with boxing gloves,” explains Calarco. The necessary intuition is provided by a mathematical method called control theory. It plays an important role in aerospace, for example, but also in optimising production processes. “We have applied this theory to the field of quantum technologies in order to selectively influence quantum processes and get the best performance out of an existing system,” says the scientist from the Peter Grünberg Institute (PGI-8).
In one process step in a quantum computer, entangled qubits are transferred from one defined state to another. The physicist and his team are developing methods to ensure that potential obstacles and errors are avoided along the way. This happens on a level between the actual quantum computers and the quantum algorithms running on them. “It’s the firmware for quantum computers, exactly between hardware and software,” explains Tommaso Calarco. Firmware performs basic tasks in a computer or other electronic devices. For example, it adapts the quantum computers in such a way that they are able to process new algorithms.
Quantum Community Network
In the European Quantum Flagship’s Quantum Community Network (QCN), each member state has two representatives: one from research and one from industry. “The QCN is an important instrument for international networking – facilitating not only the coming together of theorists and experimental groups, but also of science and industry,” explains the QCN chair, Jülich researcher Tommaso Calarco. He also promotes regular exchange with industry through the Quantum Industry Consortium (QuIC), which was established on the initiative of the QCN and brings European companies together.
Applications for quantum computers, the algorithms, can already be developed and tested now, even without ready-made quantum computers being available: these processes can be simulated on conventional high-performance computers. Kristel Michielsen knows how this works.
- Kristel Michielsen
In the pioneering days of quantum computing, her computing skills were in demand: Prof. Kristel Michielsen and her team from the Jülich Supercomputing Centre contributed to Google’s proof of the so-called quantum supremacy in 2019 with simulations on the Jülich supercomputer JUWELS. Quantum supremacy means the point in time at which a quantum computer is, for the first time, superior to a conventional computer in a particular task. Proving this was considered a great challenge.
Kristel Michielsen uses the Jülich high-performance computer JUWELS to test quantum algorithms, that is, programmes that will run on quantum computers in the future. JUWELS can perform 85 petaflops, which means 85 quadrillion computing operations per second. This computing power exceeds that of 300,000 modern PCs. “Strictly speaking, the quantum algorithms can also run on conventional computing machines. Only above a threshold value of about 50 qubits do bit-based computers reach their limits,” explains the researcher.
According to Michielsen, a major advantage of the simulations is that the simulated quantum bits all work a hundred per cent reliably – in contrast to the real circuits on real quantum processors. Comparing simulation and reality also allows for an assessment of a quantum computer’s computational quality.
However, supercomputers reach their limits when it comes to simulating a quantum computer like the ones that are already available today at companies such as IBM or Google. “In 2018, we simulated a process with 48 entangled qubits,” says the physicist. “That was the world record, which might be difficult to tie because with every qubit added on top of that, the memory required by the computer running the simulation doubles.”
Simulations of quantum computers on high-performance computers are an important component of the Jülich User Infrastructure for Quantum Computing, or JUNIQ for short. It is a kind of machine park for quantum computers that brings devices with different levels of technological maturity together in one place. A 5,000+ Qubit Advantage System, which is a quantum annealer from the Canadian company D-Wave Systems, will be operated at JUNIQ starting mid-year in 2021. By the end of 2022, a quantum simulator from Pasqal will be added, which will be deeply integrated into the JSC’s modular supercomputer architecture. In addition, starting in autumn of 2021, JUNIQ will also be offering remote access to the European quantum computer of the EU flagship project OpenSuperQ, which is operated at the HQC.
By 2024, remote access to a digital-analogue quantum computer will be added, which is being developed in the DAQC project (see the Quantum Ticker) In addition, access to a variety of other systems is planned, including hybrid systems. There are also support, training and collaboration opportunities, and software tools, modelling concepts and algorithms as well as prototype applications are being developed. “Every user of this infrastructure will then have a whole range of systems at their disposal, from which they can choose what is best for them,” says Jülich physicist Kristel Michielsen, one of the initiators of JUNIQ.
Picture, illustrations: Christian Marchionna