The quantum industry is seeing more activity than ever. Scott Aaronson’s recent blog post highlights some of the latest advancements generating buzz, both in algorithms and in hardware. Error correction is a central focus in the field today, especially as excitement grows around advances toward near-term demonstrations of quantum advantage. Progress is accelerating.
We found it encouraging to see recently that error detection has been gaining more traction as an important tool in enhancing algorithm performance. After all, you can’t correct what you can’t detect. Detect first, then correct. Then scale.
Error detection is often overlooked by vendors who are just now getting on the error correction bandwagon. It is sometimes portrayed as a near-term approach that is not scalable and generates extra overhead. It turns out there’s more to it.
If you have access to the right qubit, like the Dual-Rail Cavity Qubit, you can leverage error detection as a steppingstone to much more efficient error correction. And then scale. In the meantime, you can utilize unique error data to explore new applications possible only on the kinds of platforms Quantum Circuits provides.
At Quantum Circuits, we’ve been consistent in our focus on how error detection dramatically enhances the efficiency of error correction and enables users to explore new classes of unique algorithms in the near term. Based on our novel Dual-Rail Cavity Qubit architecture (DRQs), error detection is built in at the single qubit level, an industry first. Not only does this substantially reduce the overhead for quantum error correction (QEC), it enables near-term quantum algorithms that use detected error flags as powerful information to enhance the quality of results.
The simplest way to benefit from quantum error detection (QED) is to discard data with detected errors through a process known as (strict) post-selection. In its most stringent form, any results that do not contain any measured errors are kept, while remaining results that contain one or more errors are discarded. In this simple approach, problematic errors are removed from results.
Larger quantum circuits with more qubits and gates generate fewer shots without detected errors, so it is neither a mystery nor a secret that strict post-selection is not the right fit for large algorithms. To bring post-selected fidelity to large and general circuits, QEC enhanced by error detection is the optimal solution.
Conventional qubits without error detection suffer from a worse version of the same problem. Large quantum circuits generate few or no shots without errors. But these errors are silent (without adding higher-layer redundancy), so post-selection isn’t even an option. Again, the long-term scalable solution for conventional qubits is QEC, albeit a less efficient form without QED. In other words, detect, then correct.
Whether in the near-term or long-term, conventional qubits offer no real advantage over error-detected qubits like DRQs.
In fact, QED enables significantly more flexibility for near-term applications than extremes like post-selection and QEC. Intermediate regimes exist where some or all results with errors are preserved, labeled, and used in algorithms alongside error-free data. With DRQs, error data is not devoid of information but rather can be used as an asset. As we stress in our work on the intersection of quantum and ML, the error-detected results reveal ever-increasing hidden patterns as the yield of error-free data decreases. We see this as a promising line of forward-looking research and look ahead to working with partners via our Strategic Quantum Release (SQR) program.
We are excited to see more focus on error detection in the industry. It’s a critical step to commercial quantum readiness that we have focused on since our inception. Recent work demonstrates the application of error detection to the simulation of certain noisy open quantum systems. In this example, errors are intentionally left uncorrected to simulate noise, an application of error detection that does not use post-selection because deep high-fidelity circuits are not required. Despite what this recent announcement suggests, we never view error detection as a non-scalable “stop-gap”, and this work does not alter our (nor the industry’s) understanding of QED’s role to achieve useful quantum advantage.
Indeed, we view error detection as both a core component of our scalable approach to fault tolerance in the error correction code that leverages the unique DRQ architecture and as a powerful capability to enable and enhance near-term applications. In fact, the same experiments demonstrated above require just a single DRQ in the logical encoding instead of five trapped ions – that’s hardware efficiency in action.
From new application spaces to error correction with low overhead – error detection is powerful and a real enabler for efficient quantum computing. Surprising? We don’t think so. We’ve been talking about it for years, it’s integral to our architecture, we’re working with a host of partners to explore its capabilities in greater detail, and it’s live online with our Seeker system.
We look forward to continued engagement with the quantum industry on how error detection can benefit both near-term application discovery and yield significant gains in error correction longer-term. Working with key stakeholders is essential to build the best systems and software for the community. The Seeker system is available via our SQR program, where users can work with our team of experts and learn more about the unique capabilities of the first DRQ-based quantum computer online today. More to come. Stay tuned.
In the meantime, remember this pragmatic approach. Detect. Correct. Then scale.
