Cat Qubits Offer Efficient Path To Error Correction: Alice & Bob Make 5 Criteria For LQ Evals

Alice & Bob are tackling a core challenge in quantum computing: establishing how to reliably measure progress toward fault-tolerant machines. The company’s approach utilizes cat qubits, chosen for their efficiency in error correction by requiring fewer physical qubits per logical one compared to conventional architectures. Recognizing a lack of standardization in recent claims, Alice & Bob commented and laid out three criteria, breakeven, below-threshold operation, and repeatable error-correction cycles, after Google’s Willow chip announcement, a framework that has since expanded to five. At Alice & Bob, we were talking about FTQC as the only path to useful quantum computing before it became industry vocabulary,” the company states, emphasizing the need for a common understanding of what constitutes a truly useful building block for future quantum computers. These criteria, designed to be accessible even without a quantum physics background, aim to provide a modality-agnostic benchmark for evaluating logical qubit experiments.

Cat Qubit Efficiency and Fault-Tolerance Requirements

A key challenge is the lack of standardization across the industry; claims do not demonstrate the same thing, and there is no standardized, common definition for what constitutes a meaningful logical qubit milestone. To address this, Alice & Bob commented and laid out a set of criteria the result satisfied. Beyond simply achieving low error rates, the framework emphasizes the importance of running experiments long enough to capture realistic error patterns and avoiding post-selection. Discarding runs where errors were detected is not an accurate way to establish a logical error rate and is not reproducible while running real algorithms. Ultimately, the goal is to build logical qubits capable of sustaining computations for extended periods, potentially hours or even weeks, to account for rare physical events like cosmic ray impacts; an experiment that runs for seconds or minutes simply has not been exposed to those error mechanisms.

Five Criteria for Evaluating Logical Qubit Performance

The pursuit of practical quantum computing has shifted from theoretical possibility to demonstrable, albeit incremental, progress; however, evaluating these advancements requires standardized metrics, a challenge currently facing the field. This framework has since expanded to five, remaining relevant as the industry matures and claims of milestones proliferate without consistent definitions. These criteria focus on memory experiments utilizing single logical qubits, designed to determine if they represent genuinely useful building blocks for FTQC. The first, “breakeven,” assesses whether the logical qubit outperforms the physical qubits it comprises, demanding a longer logical lifetime than physical ones, as Google Quantum AI demonstrated with its surface code. Beyond simply functioning, a scalable architecture is crucial; the second criterion requires a tunable parameter, such as code distance, to systematically reduce logical error rates.

Sufficient quantum error correction cycles are also essential, ensuring experiments run long enough for error patterns to emerge and be accurately measured. Avoiding artificially low error rates requires complete datasets; post-selection, discarding runs where errors were detected, is not an accurate way to establish a logical error rate and is not reproducible while running real algorithms. Finally, a “bonus” criterion considers utility timescales, acknowledging that algorithms like Shor’s algorithm for RSA, FeMoco simulation, or Hubbard model calculations, run for extended periods, exposing logical qubits to rare error mechanisms like cosmic-ray impacts. Every meaningful quantum computation will run inside a classical HPC environment,” the company notes, emphasizing the need for collaboration between quantum and high-performance computing communities to accelerate progress.

A key consideration is whether a logical qubit can demonstrably outperform its constituent physical qubits; the error-correction layer must actively reduce errors, extending the logical qubit’s lifetime beyond that of the underlying hardware. Beyond mere functionality, scalability is paramount, requiring a tunable parameter, often code distance, to systematically lower logical error rates. Ultimately, the goal is to achieve a state where logical qubits can sustain computations lasting hours, days, or even weeks, though this remains a significant challenge.