Twente-based startup Quix had to make one of the most elaborate photonic integrated circuits to date to create the world’s most powerful photonic quantum processor.
Quix Quantum has announced an update for its photonic quantum processor line-up. Featuring a record number of 20 qumodes (the optical equivalent of qubits) and the best operating specifications on the market, the new processor outperforms the current generation of processors by almost a factor of 2, the Enschede-based startup claims.
Like the more familiar ‘conventional’ quantum computer, photonic quantum computing exploits quantum mechanical phenomena to produce potentially vast computing power but goes about it in an entirely different way. Instead of creating and manipulating delicate, usually subatomic quantum systems (qubits), it involves leading photons through an optical circuit consisting of a matrix of programmable intersections. Non-classical processing power emerges from the quantum mechanical interactions between the light particles when they meet along the way.
Recently, Chinese researchers – assisted by Quix’s CTO Jelmer Renema – used the quantum photonic approach to sift through a problem with 1043 possible outcomes at a sampling rate that’s 1024 faster than brute-force simulation on classical supercomputers would be able to achieve. This result is one of less than a handful of demonstrations of a quantum advantage to date. A quantum advantage, also called quantum primacy or quantum supremacy, is reached when a quantum device solves a problem that no classical computer can solve in any feasible amount of time.
The problem solved by the Chinese has no practical value. The approach, however, can be extended to speed up chemical simulations, for example to develop new materials or drugs, or boost machine learning. This will require more than 20 optical modes, but how many more is still an open question. A large-scale ‘universal’ optical quantum computer, which can conduct a wide range of calculations, will probably require millions of qumodes, but devices with significantly fewer optical qubits might still be able to do things that aren’t merely of academic interest.
PhD pitches at the Benelux RF Conference
Learn about the latest trends and developments in high-end RF techniques. On 24 May, the Benelux RF Conference will take place in Nijmegen. New this year are the PhD pitches, in which young professionals present their research results. Make sure to reserve your seat in time and register now.
Penciled in
The Chinese used discrete optical components to construct their ‘processor.’ Such bulky setups will never scale to the size needed to solve practically relevant problems. That’s why Quix implements its processors in integrated photonics, which massively improves scaling potential while containing cost. This involves testing the limits of integrated photonics; the 20-qumode photonic quantum processor is already one of the most complex photonic integrated circuits (PICs) ever made, Renema said at the launch event.
Quix doesn’t manufacture hardware itself; it operates as a system integrator. The PIC is handled by Lionix, like Quix a spinoff from the University of Twente (UT). Lionix developed its own ‘flavor’ of integrated photonics, Triplex, which is based on silicon nitride waveguides. Thanks to low light losses and other characteristics, Triplex allows for scaling to larger sizes (number of qumodes) than would be possible using other integrated-photonics recipes. Lionix’s manufacturing facility is located in South Korea.
The complex job of packaging the PIC is handled by Phix Photonics Assembly, like Quix and Lionix located on the UT campus. The packaging process involves connecting optical fibers to the inputs and outputs of the PIC, creating the electrical connections and taking care of proper heat dissipation.
Quix has sold five processors so far, all to European research outfits, making it “the de-facto standard for photonic quantum computing across Europe,” the startup claims. A 50-qumode photonic quantum processor is scheduled for launch in Q4 2022; a universal version is penciled in on the roadmap for 2026.