New Clavina Photonic Processor Brings Programmable Nonlinearity to Quantum Computing
A new quantum photonic processor developed by researchers at Imperial College London and collaborators could help overcome one of the most persistent obstacles in light-based quantum computing: photons are excellent carriers of information, but they rarely interact with one another. This property allows quantum signals to travel quickly and with relatively low losses, yet it also makes it difficult to perform the nonlinear operations needed for a fully capable quantum computer. The new architecture, called Clavina, combines programmable optical circuits with specialised nonlinear modules, creating a flexible platform that can carry out a wider range of quantum operations within a single system.
Photonic quantum computers encode information in particles of light, or photons. These particles can be manipulated through their paths, phases, frequencies and arrival times. In Clavina, information is encoded into precisely controlled time bins, meaning that individual quantum states occupy selected moments in a sequence of optical pulses. Fast electro-optic modulators then switch these time-bin-encoded states between different parts of the processor. By rapidly changing the routing, the system can direct quantum information into functional modules designed for particular operations, without rebuilding the processor’s underlying optical hardware.
Most photonic quantum platforms are engineered around a specific task. They may be highly effective at generating entanglement, performing optical transformations or simulating a particular physical model, but changing their purpose can require a substantial redesign. Clavina takes a different approach, drawing inspiration from modern classical computer processors, which connect general-purpose control systems with specialised units for graphics, artificial intelligence or data processing. Its central control architecture can direct quantum information through a programmable linear optical network and then into nonlinear components, allowing the same device to be reconfigured for different computational goals.
The distinction between linear and nonlinear operations is crucial. Linear optical processes can split, interfere and recombine photons, enabling powerful transformations of quantum states. However, linear optics alone cannot easily provide the effective interactions required for universal quantum computation. Nonlinear operations allow the evolution of a quantum state to depend on its amplitude or particle number, making it possible to implement interactions between quantum modes and generate more complex states. By integrating these two classes of operations, Clavina moves closer to a photonic processor capable of supporting a universal set of quantum gates at the physical level.
The researchers demonstrated the architecture by applying it to the Bose-Hubbard model, a theoretical description of interacting quantum particles moving through a lattice. The model is widely used to study phenomena such as quantum transport, correlated matter and phase transitions. Simulating these many-body interactions is difficult because the number of possible quantum states grows rapidly as more particles and modes are added. Clavina’s combination of programmable linear transformations and nonlinear operations allowed the team to reproduce the types of interactions represented by the model, demonstrating how the processor could be adapted for complex quantum simulations.
The system was also used to generate Gottesman-Kitaev-Preskill, or GKP, states. These are highly structured quantum states encoded in the continuous variables of light, such as the position and momentum-like quadratures of an optical field. GKP states are important because they can provide protection against certain forms of noise and are considered a promising resource for bosonic quantum error correction. Earlier photonic methods often produced these states only probabilistically, meaning that a successful state would appear only in some experimental attempts. A more consistent generation process could make such states significantly more useful in future fault-tolerant machines.
Clavina’s nonlinear capabilities also support the creation of other nonclassical resources, including Schrödinger cat states. These states represent quantum superpositions involving distinguishable configurations of a field and are valuable in quantum sensing, communication and error-correction research. The ability to generate cat states and GKP states within a programmable architecture is particularly important because error correction is expected to require multiple interconnected operations rather than a single isolated function. A processor that can produce and manipulate these resources on demand could reduce the need for separate, task-specific photonic systems.
According to Dr Shang Yu, lead author of the study and a Marie Skłodowska-Curie Fellow at Imperial, the objective was to create a photonic quantum processor with a major increase in functionality compared with earlier designs. The architecture is intended to be scalable, modular and extensible, so additional functional units can be incorporated as new computational requirements emerge. Co-author Ying Dong noted that switching between modules could allow one set of hardware to address graph problems, quantum simulations, large entangled states and error-correction resources without requiring a complete overhaul of the processor.
The approach could prove especially significant as quantum computers move from laboratory demonstrations toward larger and more specialised machines. Photonic systems offer potential advantages including high-speed information transfer, compatibility with optical communications infrastructure and the ability to operate across interconnected components. They also face serious challenges, including photon loss, the difficulty of producing reliable sources and the need to engineer effective interactions. Clavina does not eliminate these problems, but it offers a framework for addressing several of them together. By placing programmable control, linear processing and nonlinear quantum operations in one reconfigurable architecture, the platform could help turn photonic quantum computing from a collection of specialised experiments into a more adaptable computing technology.
Subject of Research: Photonic quantum computing and programmable quantum processors
Article Title: Extensible universal photonic quantum computing with nonlinearity
News Publication Date: 31-Jul-2026
Web References: https://www.nature.com/articles/s41566-026-01962-8
References: Nature Photonics; Imperial College London researchers and collaborators
Image Credits: Dr Raj Patel, Imperial College London
Keywords: Clavina, quantum computing, photonic quantum computing, photons, nonlinear optics, quantum processors, quantum error correction, GKP states, Schrödinger cat states, Bose-Hubbard model, Imperial College London
