Quantum computing

Quantum computing: teamwork is key

The biggest hopes are pinned on quantum computers. Because they use a fundamentally different approach to computing, they could solve problems that are impossible for traditional computers. Present-day versions are still too limited and error-prone to truly add value at this point. But there have been great advances recently. The sensitive qubits – superconductors, atoms, photons or ion traps – are subject to thermal, electromagnetic and other influences that lead to computing errors and noise. Correcting for these errors requires adding other qubits to the chip to execute the necessary code. The issue is that the quantity of qubits currently available falls far short of what would be required. On top of that, every additional error-prone computing unit further increases the error rate.

Now, Microsoft is promising a true quantum leap to higher qubit spheres with its innovative Majorana 1 quantum chip, un­veiled in February, which is based on “topological” qubits. “In prin­ciple, this new technology could make it possible to place millions of qubits on a chip, and not only that, but qubits that are much more robust than before,” ex­plains Dr. Jeanette Lorenz, a quantum researcher at the Fraun­hofer Institute for Cognitive Systems IKS, expressing cautious optimism. So far, however, no computing oper­ations or algorithms have yet been shown on the chips that have been produced. If researchers are able to do so, it could cut the time needed to develop practical quantum computers from decades to just a few years. “Up until then, benchmarking will be an important task – not just for industry, to really be able to gauge potential, but also for the quantum community itself when it comes to knowing what direction to go in for aspects like further developing software.”

Error-tolerant and application-friendly software

Quantum software is the research area of Jeanette Lorenz and her team. While work is ongoing to make the qubits themselves less error-prone, they are striving to adjust the algorithms to cope with hard­ware errors. Together with five partners from industry and the research sector, they are work­ing on a project called QUAST, which aims to make quantum computing easily accessible to companies. The researchers are focusing on industrial opti­mization issues that do not yet have a perfect solution today.

“To be able to solve a certain issue, the first thing we need to understand is which algorithm fits which hardware, meaning how a specific industrial opti­mization problem can be import­ed into a quantum computer in the first place,” Lorenz says, outlining the challenge. One crucial factor in this is the software stack, meaning the layered structure of all components that are necessary to the development, op­eration and use of quantum computers. The relevant use case is visualized at the uppermost level of the stack. The lower levels are where the connection to the relevant hardware is made. To date, the assumption has been that different hardware, with its specific pros and cons, will be suitable for different use cases. This is also why the researchers are working with different hardware provid­ers. “We know, for example, that ion traps are slower than superconducting qubits, which is why they tend to be a better fit for questions where we can afford for the machine to compute more slowly in exchange for greater accuracy.” Lorenz points to simulation of molecules as one example.

One of the many applications her research team has explored as part of the QUAST project is optimization of power grids. The goal here is to strike a balance among different distributed energy sources with high volatility. The question that traditional computers become less and less able to answer as complexity increases is when pow­er sources should be brought online or taken offline to stabilize the grid. Quantum computers may be able to help with these kinds of combinatorial optimization problems – but not on their own. “People are increasing­ly realizing that quantum processors will be more of a new processing unit rather than an independent system in their own right,” Lorenz says. “In the future, they will probably be operated in tandem with a traditional high-performance system. That’s why the connection is so important, so people can switch back and forth de­pending on the issue.”

One of the results of QUAST is a detailed and specific decision tree with solution paths for these kinds of optimization problems. It lets companies put together their own quantum-supported solution from various compo­nents to solve their application problem automatically. “They don’t need any quantum experts of their own. Instead, they are guided through the entire pro­cess, so they know which algo­rithm fits which issue,” Lorenz explains. This is of interest both to industrial firms, which want to use the new methods to op­timize their own processes in fields such as the automotive industry, and software providers, which are looking to incorporate these advances into their software tools.