IBM announced a new quantum error correction architecture Tuesday that will characterize the design of its forthcoming large-scale quantum processor, IBM Starling.
The starring element of IBM’s Quantum Starling, slated to arrive by 2029, is the low-density parity-check code — in which many logical qubits may be encoded on fewer physical qubits while reducing noise interference — that facilitates the error correction necessary for a quantum-powered computer to smoothly process information and execute commands. Abbreviated as LDPC, the error-correcting coding method is the star element of IBM’s novel fault-tolerant architecture, dubbed “the bicycle architecture.”
“We now have a complete architecture where, if you compare it to the surface code architectures that exist, that is also an order of magnitude or more more efficient,” Jay Gambetta, IBM’s Quantum Initiative vice president, told Nextgov/FCW. “So, this architecture doesn’t require anything we don’t know how to build. It requires less qubits, so it’s easier to do. And for that reason, we feel very confident we can actually execute on it.”
As opposed to surface code architecture — which encodes information on a two-dimensional lattice of qubits, often in short-range — IBM’s bicycle architecture promises better error correction and reliable performance abilities through coupling LDPC codes with modular hardware that supports long-range qubit connectivity, with the goal of enhancing Starling’s computational capacity to fault tolerance.
“The bicycle architecture uses a long-range-connected modular hardware with quantum LDPC codes, in contrast to the conventional surface code architecture based on short-range, monolithic hardware,” IBM’s accompanying paper on its bicycle architecture said.
Eventually, IBM will scale Starling up to thousands of logical qubits to process more complex algorithmic requests, but it is expected to be efficient enough to not require excessive overhead resources.
Starling’s capabilities will be available both in an on-premise format as well as via IBM’s cloud computing software. Initially, Gambetta expects Starling’s capabilities to aid in problems related to chemistry, materials science and optimization, as well as potentially helping with partial differential equations. Starling will serve as the foundation for IBM’s Blue Jay, a 100,000-qubit system set to arrive by 2030.
“We’ve got an end to end architecture now that we completely can describe that allows us to build towards a fault-tolerant quantum computer,” Gambetta said. “These computers that people dreamed about, I feel very, very confident we can build them now.”
The majority of Starling’s research is privately funded by IBM, but Gambetta noted that the company still participates in federal government programs like the Defense Advanced Research Projects Agency’s Quantum Benchmarking Program.
Gambetta said that the Starling announcement related to its bicycle architecture signals a larger shift in the quantum computing industry’s quest to bring a cryptographically-relevant quantum computer to life. While multiple theoretical and scientific avenues exist for practical quantum computing, the engineering and infrastructure challenges remain a larger barrier to scalability.
“The high level view is: the science has been de-risked now by completely solving how we’re going to do every part of [quantum computing],” Gambetta said. “Now it comes down to engineering and building it. So it’s moved from science to engineering, and that is why we feel confident that we can build a large-scale, full time quantum computer by the end of this decade.”
