Saskatchewan team reports record efficiency hyperbolic codes for modular chips

[Pete14]

Researchers at the University of Saskatchewan say they have found families of hyperbolic surface codes that reach a theoretical efficiency limit, Quantum Computing Report reports. Ahmed Adel Mahmoud and Dr Steven Rayan from the Centre for Quantum Topology and Its Applications describe the work in a preprint, arXiv:2610.10948. Their codes are said to reach the limit set by what is known as Delfosse's bound. It has not yet been peer reviewed

The method places quantum LDPC interactions on compactified hyperbolic surfaces and tunes the periodic boundary conditions. According to the report, this doubled the code distance and quadrupled efficiency without adding qubits or raising check weights. The headline code encodes 17,112 logical qubits into 51,330 physical qubits with a distance of 10. That gives an efficiency of about 33.34, roughly 33 times that of standard two dimensional toric codes

The practical side is just as interesting. A compiler splits the codes into planar modules of 80 or fewer qubits per chip. That matters because most hardware roadmaps now involve linking many smaller chips rather than building one giant processor. Codes that fit naturally into modules could make those designs far more efficient

In circuit level simulations using a standard noise model, the threshold came out at about 0.22%. Even when errors between modules were tripled, the threshold stayed at around 0.17%. That robustness to noisy links is encouraging, since connections between chips are usually the weakest point. Still, these are simulations, and real hardware brings its own surprises

This is the kind of theory work that quietly changes the numbers in everyone's roadmaps. Better codes mean fewer physical qubits for the same number of logical ones. We discussed in the RSA thread how much of the qubit count comes from error correction overhead. Results like this chip away at that

How excited should we get about preprint results like this? And do you think hyperbolic codes will find their way into real hardware?

University of Saskatchewan Researchers Achieve Record Hyperbolic Surface Code Efficiency for Modular Fault-Tolerant Architectures - Quantum Computing Reportquantumcomputingreport.com