Can the human brain compute quantum algorithms

FRAUNHOFER INSTITUTE FOR OPEN COMMUNICATION SYSTEMS

The euphoria surrounding quantum computers is great: at least since Google researchers reported on possible »quantum supremacy« in the journal »Nature« in 2019, a new age seems to have dawned in which science fiction is becoming a reality. Since 2012, $ 674 million in venture capital has gone into quantum computing companies. The Federal Ministry of Research will invest an additional 650 million euros in the technology until 2022. In the future, quantum computers should help to enable complex calculations in a significantly shorter time than "classic" computers can. As a result, they are theoretically able to crack encryption, comprehensively simulate the human brain, help optimize traffic forecasts, or accelerate the development of batteries or new materials in the chemical and pharmaceutical industries. There is already talk of the "quantum Internet". But what actually is a quantum computer and how does it work?

Annealer and universal quantum computer

So far, the bit has been the measure of all things. "Classical" computers can represent any type of data by stringing together bits that can assume the status 0 or 1. The basis for this are the transistors, which act as switches and thus generate the states 0 or 1. If current flows through a transistor, state 1 is present. If no current is flowing, the status is 0. The performance of today's computers therefore essentially depends on the number of transistors.

Quantum computers, on the other hand, calculate with so-called »qubits« (quantum bits). There are several ways to create them. One is to "lock" charged atoms, the ions, in magnetic and electric fields. In such "ion traps" the ions can be brought into different states with the help of lasers. A laser is then also used to read the result. Every single ion in this trap is a qubit. Interesting: a qubit can assume all possible states in a quantum computer - and at the same time. This means that a quantum computer can carry out many "calculations" at the same time. A quantum computer is therefore much more suitable for solving certain questions.

The problem with this is that as soon as particles have a certain energy, which we know as heat, they start to move. If you want to manipulate and capture (calculate) them, you have to calm them down. This is done by cooling the chips, on which the "ion traps" described above are applied, almost to absolute zero. At minus 273 degrees the quanta are shielded from all environmental influences. A quantum computer works under vacuum conditions and is protected against electromagnetic influences. External forces, such as vibrations, could also cause errors when reading out the results.

Not all quantum computers are the same. Universal quantum computers make it possible in principle to carry out any kind of arithmetic operation. So far, only a few algorithms, i.e. calculation instructions, exist for such computers. In contrast to "classic" computers that work with yes / no queries, they have to be formulated completely differently. In the end, quantum algorithms do not deliver a clear result, but rather probabilities for different results, from which the most likely answer can only be determined after several runs of the calculation. In addition to such universal quantum computers, there are so-called quantum annealers, which are structured differently and are more suitable for very specific tasks.

Quantum computing at Fraunhofer

The Fraunhofer-Gesellschaft is pooling its quantum computing competencies in the newly emerging “Fraunhofer Center for Quantum Computing”. Together with IBM, Fraunhofer is building the first universal quantum computer in Europe in a data center near Stuttgart. The computer should be available for application-oriented quantum computing research from 2021. "One of our central research questions is which specific application scenarios in industry are suitable for calculation with a quantum computer and how the necessary algorithms can be developed and easily translated into applications," says Prof. Manfred Hauswirth, managing director of Fraunhofer FOKUS and Co- Spokesman for the Quantum Computing division of the Fraunhofer Society. With the European quantum computer, digital sovereignty is also preserved. The data are processed in accordance with European law and are subject to European data protection standards. "It will certainly take ten years, maybe twenty, for quantum computers to come into widespread use in laboratories," says Manfred Hauswirth. “But it is more important when the time comes to have the necessary software and expertise available. We already know where we're headed, because we're helping to shape it. And Fraunhofer is of central importance in bringing industry and applied research into position and building it up «.

A platform for quantum-assisted artificial intelligence

The “Platform and Ecosystem for Quantum-Assisted Artificial Intelligence” (PlanQK) project started in January 2020. Over the next three years, the project will work on a platform for quantum-assisted artificial intelligence. Fraunhofer FOKUS, together with 14 project partners, is researching the implementation of various use cases from three typical areas: "Forecasting and Classification", "Maintenance and Detection" and "Planning and Optimization". Among other things, AI-based solutions are to be developed - e.g. B. for the forecast and classification of anomaly and fraud detection in the financial sector, for maintenance in industry and in local public transport, the maintenance of street furniture (e.g. park benches, street lamps) or the condition of the streets, the optimization of production lines in industry or the networking and control of industrial components. In addition, Fraunhofer FOKUS is expanding the PlanQK platform with semantic management of knowledge so that it is also available in machine-readable and interpretable form for AI-based functions. A semantic search can be used to find acronyms such as B. dissolve »ANN« as »Artificial Neural Networks«.

However, such AI applications consume more and more computing time. Here quantum computers can help to provide the necessary resources. The problem with this: In order to develop AI applications that can benefit from quantum computers, AI knowledge is required in addition to domain knowledge. These skills are often difficult for companies to develop. The PlanQK project therefore creates a community of different experts who can work together through technically meaningful interfaces. Users can access a quantum AppStore and put together or commission the solutions for their company. Developers can easily use Quantum platforms to expand and improve their AI algorithms. And specialists provide concepts that make quantum computing easily accessible even without special expertise.

The projects show that quantum computing is no longer just science fiction. Even if quantum computers will probably never be as commonplace as the PC on the home desk, they open up completely new possibilities for us in many areas of application. "It is likely that cloud models will prevail, quantum computers as a service, so to speak, as well as mixed forms of traditional high-performance computing and quantum computing," says Manfred Hauswirth, assessing the development.


Related Links: