Why do we need quantum computers

By Christian J. Meier

In the foyer the astronaut Alexander Gerst smiles at the visitor in a life-size photo, next to it hangs a model of the International Space Station ISS. Around 80 experts gathered at the control center of the European Space Agency Esa in Darmstadt in February. You want to venture into unknown territory. But instead of space travel, it is about the first applications of quantum computers, which some of the experts here expect in a few years. Potential users discuss with developers of commercial quantum chips.

Ever since Google introduced a quantum chip last year that solved a special task much faster than the most powerful supercomputer in the world could, there has been a gold rush atmosphere in the industry. In addition to tech giants such as Google and IBM, start-ups are looking for their fortune in the new field, mostly founded by the research community and endowed with millions in private capital. But traditional corporations like BASF or Merck also want to know whether they will soon be able to use quantum computers. "The new technology could fundamentally change our industry," says Philipp Harbach from the Darmstadt-based pharmaceutical company Merck.

The new type of computer uses the rules of quantum mechanics that apply in the microcosm. An atom or an electron can have a double existence, for example be in two places at the same time. Used as a data memory, it saves the two values ​​"0" and "1" simultaneously. This "qubit" surpasses the smallest memory unit of a classic computer, the "bit", which only accepts either "0" or "1". With each additional qubit, the capacity doubles. Even 300 qubits can store more values ​​than the known universe contains particles. Since the data can potentially also be processed simultaneously, the quantum computer promises a huge speed in information processing.

The new computers have not yet accelerated a single worthwhile application

But that doesn't work yet. Quantum computers have not yet accelerated a single worthwhile application. Google's machine solved a purely academic problem with 53 qubits. A lot more qubits would be needed to outperform classic computers in applications such as pattern recognition or data searches. These are also extremely sensitive to environmental influences. After a relatively few calculation steps, qubits lose their ability to process values ​​simultaneously. Whether and how long this can be delayed is currently being researched. Experts do not expect a breakthrough for a decade at the earliest.

Billions of euros are still flowing into research worldwide. But a "quantum winter" could soon bring the flow of money to a standstill if tangible results continue to fail. To prevent this, some researchers want to build a kind of "quantum computer light" that could soon solve individual tasks and keep interest alive. Some problems, it is hoped, can be solved with 100 or 200 qubits and relatively few calculation steps. Start-ups from the USA, Great Britain or Germany are playing poker high. In order to gear their hardware and software towards profit, they need specific customer requests. The three-person "Quantum Task Force" at Merck, of which Harbach is a member, tries to formulate such orders.

For Merck, quantum mechanics is part of everyday life, says Harbach. Because chemical reactions follow their rules. Simulating "quantum chemistry" is an obvious task for a quantum computer. That might make it possible to achieve something that is still unprofitable today, says Harbach. He cites a display made of organic light-emitting diodes (OLED) as an example. It consists of several layers of organic material. "To save costs, a customer wanted to make one of the layers thinner." But first you have to find a suitable chemical compound, with a high refractive index, for example. Synthesizing and testing thousands of candidates in the laboratory would cost too much time and money. The alternative: simulating on the computer how a new molecule reacts to electromagnetic waves. "The entire process simulation is quantum mechanical," says Harbach. With a light version of the quantum computer, however, you can only process part of it. "It would be worth using if the added value of the calculation result were significantly greater," says Harbach. "We don't know yet whether that will be possible anytime soon," he admits.

Klaus Merz, who calculates the orbits of satellites at Esa, has a completely different problem in mind. "The traffic in orbit is increasing sharply," says the mathematician. Disused spacecraft and debris are already disrupting operations today. A satellite has to fly one to three evasive maneuvers a year. "That could grow to dozens a year in the future," says Merz. Each swing would trigger a cascade of others. A continuous optimization of the flight paths by computer would be necessary. Each additional satellite multiplies the number of possible sequences of maneuvers. Conventional computers need too much time to find the variant with the lowest risk of collision in the abundance of variants. In the future, quantum computers, on the other hand, could test all alternatives simultaneously and master the ongoing optimization. "That is why we are keeping an eye on the development of the quantum computer," says Merz. A specific test is not imminent: "We are not quantum people," he says.

"Today's quantum computers are reminiscent of the computers of the 1950s," says one expert

In Darmstadt, Merz has the opportunity to build bridges into the quantum world: five start-ups present their developments. However, they do not have ready-made solutions to offer. Rather, they form an ecosystem that tries different approaches to hardware and the associated software. The light version of the new computer is far from being mature. "Today's quantum computers are reminiscent of the computers of the 1950s," says Matthew Hutchings from the US start-up SeeQC. They are assembled from many parts, mostly in the laboratory. What is missing is an integrated chip, a component that brings together all functions - the qubits, error correction and cables for data exchange. The start-up now wants to integrate parts of the control system into the cooling device in which the qubits are located. Fewer cables then lead in from the outside, which should reduce the interference.

Other start-ups develop software for quantum computers. "We are adapting existing algorithms for quantum computers," says Jan Rainer from HQS Quantum Simulations. The Karlsruhe-based company develops software for simulating chemical reactions for customers such as BASF and Merck. "We are currently testing where quantum computers can and cannot be used," explains Rainer. He says he will be able to say in a year whether and how well this will work. He expects the first applications in five years.

However, the potential users in Darmstadt are not really on fire for the quantum computer. They still see it as an object of study, not as a tool. The interest on the part of the chemical industry is most specific. Others remain aloof: "That is far from what we are doing," says Thomas Neff from the Society for Heavy Ion Research in Darmstadt, where atomic nuclei are shot at one another. To simulate the nuclear reactions you need a quantum machine with 180,000 qubits - that is not realistic for the time being. Philipp Harbach expects the new technology to grow slowly: "It will take a generation or two of programmers before users know what makes a quantum computer tick. That was also the case with classic computers in the 1960s and 70s."

Preventing the onset of quantum winter with quick applications will therefore be difficult. If he comes, then he comes.