An open race
Computers that calculate according to the rules of quantum physics could possibly solve special tasks faster than today’s supercomputers. However, universal machines of this kind still only exist as laboratory experiments.
The race has long begun: who will build the first quantum computer, far superior to today’s supercomputers? Companies such as Intel, IBM, Google and Microsoft are taking part in the race. The USA, Canada, China and Japan are investing considerable sums in the development of these exotic calculating machines. The US government, for example, supports research into quantum technologies with around € 170 million per year. In the science metropolis of Hefei, Beijing has a quantum laboratory built for almost € 9 billion. The EU has just launched a € 1 billion funding programme for quantum technologies; an important part of this flagship programme is dedicated to quantum computers.
“The developments in this area are being driven forward with great enthusiasm,” says physics professor David DiVincenzo of the Peter Grünberg Institute (PGI-2), a pioneer in the field of quantum information science. The exotic computers raise high expectations. Many experts are certain that such a machine will solve certain mathematical problems much faster than all computers of today combined. For example, quantum computers could, in a matter of seconds, search through huge databases, deal with complex logistics problems and calculate the properties of molecules for chemistry and materials research. They would also be able to cancel out the standard procedures currently used to encrypt data on the Internet. Today’s computers are too slow to cope with the many calculation steps to decrypt the data in a reasonable amount of time. With the computing power of a quantum computer, this would be easily possible.
Yet: what has been achieved so far falls short of expectations. This is no surprise to David DiVincenzo: “The participants in this race don’t run very fast. It’s as if they were all dragging around a 100-pound backpack. These are virtually the enormous technical challenges that still need to be mastered. So no race like at the Olympic Games,” explains the Jülich researcher, smiling. More than 20 years ago, the physicist had already formulated five basic criteria that a universal quantum computer must theoretically fulfil. Every newly developed model must be measured against these in the laboratory.
From bit to qubit
Um die Herausforderungen und Tücken besser zu verstehen, muss man sich klar machen, dass Quantencomputer nicht wie konventionelle Computer arbeiten. Deren Welt besteht aus den Bits, aus Nullen und Einsen. Ein Bit ist die kleinstmögliche Informationseinheit. Um ein Bauteil als Träger einer Information nutzen zu können, muss es vor allem eine Eigenschaft besitzen: Es muss zwei unterschiedliche Zustände einnehmen können, die der Null und der Eins entsprechen. Herkömmliche Computerchips rechnen mit Milliarden von mikroskopisch kleinen Halbleitertransistoren, die wie winzige Kippschalter für den elektrischen Strom funktionieren. Eine Steuerspannung kann sie zwischen „an“ und „aus“, zwischen null und eins hin- und herwechseln.
„ Quantencomputer sollten bei gewissen Aufgabenstellungen konventionelle Rechner, die die Befehle nacheinander abarbeiten, schlagen können.“
David Divincenzo
Quantum computers, on the other hand, calculate with quantum bits, so-called qubits. They can assume not only the states of zero or one, but at the same time also any value in between. This is due to the information carriers that are used. These are the smallest objects we know, such as atoms, ions, electrons or photons. Completely different rules apply for them than for classical microelectronics – they are subject to the often bizarre rules of the quantum world. One of these rules is the possible superposition of states. The experts call it superposition. It allows quantum objects to assume several states simultaneously. Furthermore, the individual particles can be entangled in an arrangement of qubits. They are then connected to each other as if by magic. Because of these two phenomena, quantum computers are capable of performing many operations simultaneously with each switching operation. David DiVincenzo sums it up: “With this parallel processing, they should be able to beat conventional computers, which execute the commands one after the other, in certain tasks.”
At least that’s the theory. In practice, however, a phenomenon called decoherence thwarts the experts’ quantum computation plans. Behind this lies the fact that quantum bits react sensitively to external influences such as heat or radiation. These interference factors cause the entangled states to decay again after a few microseconds. Therefore, the lifetime of a piece of quantum information also depends on how well the researchers manage to shield the computer from the environment.
Käfige, Ketten und Fehlstellen
The research groups involved in the global race for the quantum computer have chosen various ways to achieve this goal. The ideas for the heart of such computers, the physical storage cells, are manifold: some experts rely on defects, which they specifically incorporate into thin diamond layers, or on exotic materials, which are actually insulators but conduct electricity on their surface. Other researchers hold a chain of ions in suspension in a vacuum between two electrodes. David DiVincenzo is concerned with quantum dots. These are semiconductor cages into which individual electrons are locked. Their angular momentum – the direction of their own rotation, so to speak – can store the value of the qubit. “The advantage is that there is already a lot of know-how about the manufacturing of these semiconductor structures: conventional chip production is also based on semiconductors.”
Companies like IBM and Google favour superconducting circuits. In their conductor loops, the current can circle in different directions, thus reflecting the value of the qubit: clockwise, anti-clockwise or any superposition of the two directions of rotation. The outward appearance of these machines is not quite reminiscent of computers as we know them today: only the huge, barrel-shaped cooling container is visible. Inside, it hides the actual measuring apparatus: a maze of wires and metal parts resembling an avant-garde chandelier.
At regular intervals, the companies outdo each other with information on the number of entangled qubits they were able to generate with their systems. Google announced the record with 72 qubits, IBM comes in second with 50 qubits. While the number of entangled quantum bits is increasing, the lifetime of these systems is still in the range of microseconds. This is because the more qubits that are connected, the more fragile their common quantum state becomes. All the more efforts must be made by researchers to shield the systems from the environment. Otherwise, calculation errors occur that have to be laboriously corrected. The physicist Kristel Michielsen from the Jülich Supercomputing Centre (JSC) says: “This is mainly a technical problem. The performance of today’s quantum computers is below what would be expected from theory.” On the supercomputers of the JSC, the researcher tests the algorithms that could run on quantum computers by simulating a quantum computer that performs its calculations without external disturbances. Her record is a system with 48 entangled qubits: “These simulations work perfectly, a textbook case. This allows us to compare the error rate of real quantum systems with that of an ideal quantum calculating machine.” 48 qubits, however, will probably also be the limit of the simulations, as the required memory doubles for every qubit added. That exceeds the capacity of even the fastest supercomputers today.
“You rely on the qubits finding the solution to the calculation problem themselves by jumping back and forth until the system reaches its minimum energy level.”
Kristel Michielsen
Kristel Michielsen explains that at the moment it is difficult to predict which model will eventually make the race, and when that will be the case. The race will probably not be over any time soon. From her point of view, those quantum systems stand good chances which are tailored to specific problems where the aim is to find the best solution out of a large number of possible solutions. These quantum annealers, for example, are not addressed by switching signals in the form of laser beams, microwaves or voltage pulses. “Instead, you rely on the qubits finding the solution to the calculation problem themselves by jumping back and forth until the system reaches its minimum energy level,” says the expert. “You start by setting the parameters and then let the system evolve until you get an answer.” However, this is at the expense of flexibility: Only certain tasks can be solved in this way: optimisation problems, for example for the calculation of traffic flows, but also deep learning problems and quantum simulation problems. The annealers can find the fastest way from A to B for hundreds of motorists in the urban canyons of big cities, but they will never be able to break encryptions
he physicist hopes to be able to establish a focus on quantum computing at the Jülich Supercomputing Centre in the course of the EU flagship programme, which would not only be open to experts from Forschungszentrum Jülich: “I imagine this as a platform on which different types of quantum computers will be available. We would then have a user infrastructure that offers user support and access to different experimental device types with different levels of technological maturity, complemented by our simulations. We might have remote access to quantum computers from commercial providers. And we would operate a European quantum computer and a quantum annealer. This would allow users to choose from a wider range the system that is best suited to their problem.” The signs in the race for the quantum computer would then be pointing to cooperation and not to competition – because together, things may just work out best.
Arndt Reuning