Runtime-efficient zero-noise extrapolation from mixed physical and logical data
D. V. Babukhin, W. V. Pogosov
Abstract
Partial quantum error correction and quantum error mitigation are expected to coexist in the pre-fault-tolerant regime, yet the resource advantage of combining them remains insufficiently quantified. We study zero-noise extrapolation constructed from mixed datasets that contain a small number of error-corrected data points together with data obtained without error correction. The low-noise logical points anchor the extrapolation, while the higher-noise physical points enlarge the noise baseline at a much smaller runtime cost. Under a simple model in which error correction suppresses the effective gate error rate from p to p, we derive the variance of the zero-noise estimator and compare the physical runtime required to reach a target precision. For Richardson extrapolation, the mixed-data strategy reduces variance amplification and can lower the required physical runtime by several orders of magnitude when . As a proof of principle, we apply the method to digital quantum simulation of a six-spin transverse-field Ising model and find that mixed physical/logical datasets yield lower-variance zero-noise estimates and outperform extrapolation based only on error-corrected data in the parameter regime studied here. These results identify hybrid error correction and error mitigation as a practical route to resource-efficient quantum computation before full fault tolerance.
AI Impact Assessments
(3 models)Scientific Impact Assessment
Core Contribution
This paper proposes a "mixed-data" strategy for zero-noise extrapolation (ZNE) in the pre-fault-tolerant quantum computing regime. The key idea is to combine a small number of expensive, error-corrected (logical) data points with cheaper, uncorrected (physical) data points to construct the ZNE extrapolation. The low-noise logical points serve as anchors near zero noise, while physical points extend the noise baseline at lower runtime cost. The authors derive variance expressions for Richardson extrapolation under this mixed strategy and show that the required physical runtime can be reduced by orders of magnitude when error correction suppresses gate errors by a factor γ ≤ 0.1.
Methodological Rigor
The analytical framework is straightforward but internally consistent. The paper uses a simplified noise model where error correction reduces gate error from p to γp (equation 10), which is a first-order approximation of the distance-3 surface code behavior (p_L = cp²). The variance analysis for Richardson extrapolation (equations 19-22) follows standard propagation of uncertainty through polynomial interpolation coefficients.
However, several methodological concerns arise:
1. Oversimplified noise model: The assumption p_L = γp rather than p_L ∝ p² is a significant simplification. Real error correction has nonlinear error suppression, and the effective noise model matters substantially for extrapolation curve fitting. The paper acknowledges this but does not explore consequences.
2. Equal single-shot variance assumption: The assumption that σ_s is identical for physical and logical circuits is questionable. Logical circuits may have different output distributions, and depolarizing noise at different levels changes the effective variance.
3. Bias-variance tradeoff inadequately addressed: The paper focuses almost exclusively on variance reduction but gives limited treatment to potential bias. Extrapolating across very different noise regimes (physical vs. logical) assumes the noise dependence O(λ) is well-captured by a polynomial across the entire range, which becomes increasingly suspect as the noise lever arm grows. The numerical results in Fig. 3 hint at this issue but don't rigorously quantify it.
4. Numerical demonstration is limited: The proof-of-principle uses a 6-spin Ising model with depolarizing noise — a relatively benign scenario. N_shots = 10⁴ with 64 computational basis initial states provides some statistical coverage, but the results are presented somewhat qualitatively (distribution plots) rather than with rigorous statistical tests.
Potential Impact
The paper addresses a genuinely practical question: how to optimally allocate expensive logical circuit executions versus cheap physical ones in the pre-fault-tolerant era. This is a relevant resource optimization problem as quantum hardware transitions toward partial error correction capabilities. The Tables I and II provide useful quick-reference results for practitioners.
However, the impact is somewhat limited by:
Timeliness & Relevance
The paper is well-timed. With recent demonstrations of quantum error correction on superconducting (Google) and trapped-ion platforms, the pre-fault-tolerant regime is becoming a practical reality. The question of how to optimally combine QEC and QEM resources is increasingly relevant. Several concurrent works (references 8-16) address related questions, though the specific runtime-optimization angle is somewhat distinct.
Strengths
1. Clear geometric intuition: The visual explanation in Section III A and Fig. 1 makes the core insight accessible.
2. Closed-form results: The variance prefactors #₁ and #₂ (equations 20-21) and runtime ratio (equation 22) are analytically tractable and easy to evaluate.
3. Practical framing: The focus on physical runtime rather than just shot count is practically motivated, particularly for platforms where circuit execution time is a bottleneck.
4. Honest scope: The paper acknowledges its limitations and identifies natural extensions.
Limitations
1. Extrapolation bias is underexplored: Mixing data from very different noise regimes may introduce systematic bias if the polynomial model is inadequate across the full range — this is the elephant in the room.
2. No shot-budget optimization: The equal-shots-per-point allocation is suboptimal. Optimal shot allocation (which would weight points differently) could significantly change the comparative advantage.
3. Narrow noise model: Only depolarizing noise is considered; coherent errors, crosstalk, and non-Markovian effects could alter conclusions.
4. No comparison with alternative hybrid methods: The paper doesn't benchmark against other QEC-QEM hybrid approaches, making it hard to assess relative advantage.
5. Writing quality: Some sections read as preliminary (e.g., Section II is a textbook-level overview that adds little). The paper could be more concise.
6. Scalability unclear: The 6-qubit example doesn't test whether the polynomial extrapolation assumption holds for larger, more complex circuits where noise dependence may be more complex.
Overall Assessment
This paper presents a sensible and practically motivated observation — that mixing cheap physical data with expensive logical data can improve the cost-effectiveness of ZNE. The analytical results are correct within their assumptions, and the idea is clearly communicated. However, the contribution is incremental: the core insight is relatively straightforward once stated, the analysis relies on strong simplifying assumptions, and the numerical validation is limited in scope. The paper opens a useful direction but leaves significant work (bias analysis, shot optimization, realistic noise models, experimental validation) to future studies.
Generated Apr 17, 2026
Comparison History (41)
Paper 2 represents a fundamental experimental breakthrough by demonstrating coherent cavity-mediated all-to-all interactions in a solid-state ensemble for the first time. The dramatic extension of coherence times (from microseconds to milliseconds) without decoupling pulses, alongside the realization of one-axis twisting for spin squeezing, has profound implications for quantum metrology, sensing, and memory. While Paper 1 offers a highly practical hybrid error mitigation strategy for near-term quantum computing, Paper 2's foundational advancement in solid-state many-body physics and hardware capabilities promises broader and more foundational cross-disciplinary impact.
Paper 1 demonstrates a new solid-state platform achieving coherent cavity-mediated all-to-all interactions, superradiance, one-axis twisting, and gap-protected coherence extension from microseconds to milliseconds—an experimental advance with clear implications for quantum memories, metrology (spin squeezing), and many-body physics. Its novelty (bringing cold-atom-style collective dynamics to solids) and broad applicability across quantum sensing, microwave photonics, and condensed-matter/AMO interfaces suggest high impact. Paper 2 is timely and useful for NISQ-era resource estimation, but is more incremental and model-dependent, with narrower cross-field reach.
Paper 2 addresses a critical and highly timely challenge in quantum computing: bridging the gap between near-term error mitigation and long-term fault tolerance. Its approach of mixing physical and logical data is broadly applicable across various algorithms in the pre-fault-tolerant regime, offering orders-of-magnitude runtime improvements. While Paper 1 provides a significant algorithmic improvement for Hamiltonian simulation, Paper 2 has broader potential real-world impact on near-to-medium-term practical quantum hardware utilization.
Paper 1 is likely to have higher impact due to its direct, timely relevance to near-term quantum computing: it proposes a practical hybrid error-mitigation/error-correction protocol, provides runtime/variance analysis with clear resource claims (orders-of-magnitude improvements under plausible parameters), and demonstrates performance on a concrete simulation task. Its applicability spans NISQ algorithm execution, benchmarking, and experimental workflows. Paper 2 is mathematically novel and rigorous, but its impact is more specialized (code/AME theory in heterogeneous dimensions) and may translate to practice more slowly.
Paper 2 likely has higher impact: it introduces a new simulation framework for bosonic quantum systems with provable guarantees (observable estimation and sampling), identifies clear complexity regimes (quasi-polynomial for log Kerr gates; polynomial in weak-nonlinearity), and benchmarks on Bose–Hubbard models. This breadth spans quantum computing, complexity theory, and many-body physics, with immediate utility for validating bosonic hardware and studying driven bosonic dynamics. Paper 1 is timely and practically useful for NISQ-era error mitigation, but is more incremental and narrower in scope (optimizing ZNE using a mixed physical/logical dataset under a simplified noise model).
Paper 2 is more conceptually novel and broadly impactful: it reframes single-particle interference as measurement-basis–defined rather than purely path-phase–defined, demonstrated with high visibility and a distinctive three-scan equivalence that suggests a new operational principle. Its claimed unification across atomic Λ-systems (CPT/EIT), photonic interference, and diffraction indicates cross-field relevance and potential new tools for quantum photonics and measurement-engineered mode control. Paper 1 is timely and practically useful for near-term quantum computing, but is more incremental within established error-mitigation theory.
Paper 2 proposes a fundamentally new dissipative protocol for preparing correlated quantum states across the many-body spectrum without requiring prior Hamiltonian knowledge. This is a more novel conceptual contribution with broader applicability across quantum platforms and many-body physics. Paper 1 offers a practical but incremental improvement combining existing techniques (QEC + QEM) with useful but narrower impact in the pre-fault-tolerant regime. Paper 2's flexibility, scalability, and relevance to programmable quantum simulators give it higher potential for broad scientific impact across quantum computing, simulation, and many-body physics.
Paper 1 proposes a highly timely and practical hybrid approach bridging the gap between near-term error mitigation and long-term error correction. By combining physical and logical data, it significantly reduces the runtime overhead for zero-noise extrapolation. This transitionary framework has broad applicability for scaling pre-fault-tolerant quantum computers. While Paper 2 offers a valuable optimization for fault-tolerant syndrome extraction, Paper 1's approach addresses a more immediate, pervasive bottleneck in quantum computing with potentially wider adoption in the short-to-medium term.
Paper 2 has higher estimated impact: it introduces a practical, near-term technique for combining partial error correction with error mitigation, with clear quantitative runtime/variance advantages and a proof-of-principle application. This directly addresses a pressing bottleneck for NISQ-era quantum computing and is broadly relevant across algorithms and hardware platforms. Methodological rigor is stronger (explicit model, variance analysis, runtime comparison, simulation demo). Paper 1 is conceptually interesting for quantum networking interfaces, but reads more as a design proposal with less demonstrated validation and narrower immediate applicability.
Paper 2 has higher likely impact due to timeliness and broad applicability in the near-term quantum regime: it targets a practical hybrid of error correction and error mitigation, offers analytic variance/runtime scaling results, and demonstrates gains on a concrete simulation task. Its potential real-world application—reducing physical runtime by orders of magnitude when limited logical data are available—can influence experimental practice across platforms. Paper 1 is novel and rigorous for fault-tolerant compilation under magic-state constraints, but its impact is narrower and more specialized to later-stage fault-tolerant scheduling/architecture.
Paper 1 addresses a fundamental question about quantum advantage claims by introducing a scalable classical algorithm that outperforms current GBS experiments up to 1152 modes. This directly challenges major experimental claims of quantum supremacy, which has broad implications for the entire quantum computing field. Paper 2 presents a useful but more incremental contribution combining error correction with error mitigation for the pre-fault-tolerant regime. While practically relevant, Paper 1's impact is larger because it reshapes understanding of quantum advantage benchmarks and affects how the community evaluates quantum supremacy claims.
Paper 2 likely has higher impact due to a stronger combination of novelty and rigor: it provides a parameter-free closed-form theory (A0, stopping distance, and validated P_peel formula) for decoding performance under circuit-level noise, backed by broad validation across codes/noise/T and cross-platform experimental reproduction. Its real-world applicability is immediate (large latency reductions and streaming decoding relevant to hardware control loops). The results generalize across a code family and connect to practical decoder design, giving broader cross-field relevance within quantum error correction and systems engineering than Paper 1’s more context-dependent mitigation/runtime analysis.
Paper 1 addresses a critical bottleneck in near-term quantum computing by bridging quantum error mitigation and error correction. Its method significantly reduces runtime costs and variance, offering a highly practical and broadly applicable framework for the pre-fault-tolerant regime. While Paper 2 provides valuable hardware-specific insights and a novel particle detection application, Paper 1's algorithmic advancements have a more immediate and widespread potential impact on improving computational capabilities across various quantum platforms.
Paper 1 presents a fundamental algorithmic improvement to quantum phase estimation—a cornerstone quantum algorithm—achieving an exponential reduction in two-qubit gates. This addresses a core bottleneck in quantum computing with broad applicability to any algorithm using phase kickback. Paper 2 offers a practical but incremental improvement to error mitigation strategies in the pre-fault-tolerant regime, combining known techniques (ZNE + partial error correction). While timely, its impact is more niche and transitional. Paper 1's exponential improvement to a foundational primitive has greater potential for lasting, broad scientific impact.
Paper 2 likely has higher impact due to broad, timely relevance to near-term quantum computing. It introduces an analytically grounded, runtime-efficient hybrid error mitigation/error correction strategy, with quantified variance/runtime advantages and a demonstrated application, making it immediately actionable across platforms and algorithms. Its potential real-world applications (reducing cost of useful NISQ computations) and cross-field breadth (quantum algorithms, error correction, metrology/statistics) are strong. Paper 1 is rigorous and valuable for hBN quantum emitters, but its impact is more specialized to solid-state defect physics.
Paper 2 has higher potential impact: it introduces a broadly applicable conceptual tool (Clifford lensing) to refocus delocalized phase information in interacting many-body systems, linking metrology, shadow tomography, and quantum error-correcting code structure. This bridges multiple fields (quantum sensing, information theory, tomography, many-body physics) and targets a scalable bottleneck in metrology with an experimental demonstration up to 15 qubits. Paper 1 is timely and practically useful for NISQ-era resource reduction, but is more specialized to error mitigation/extrapolation and depends on model assumptions about logical noise suppression.
Paper 1 addresses a critical bottleneck in the transition from NISQ to fault-tolerant quantum computing by combining error correction and mitigation. Demonstrating orders of magnitude reduction in runtime costs has immediate, broad, and highly practical implications for scaling quantum computation. Paper 2, while conceptually fascinating, focuses on quantum thermometry, which has a narrower scope and less immediate widespread application compared to advancing practical quantum error mitigation.
Paper 1 has higher likely impact: it targets a timely bottleneck in near-term quantum computing by quantifying resource/runtime advantages of combining error mitigation with partial error correction, with clear practical implications for NISQ-to-pre-FT workflows. The approach is novel in leveraging mixed logical/physical datasets to reduce variance amplification in ZNE, includes analytic variance/runtime scaling under a concrete noise-suppression model, and demonstrates performance on a nontrivial digital simulation. Its relevance spans quantum algorithms, hardware, and error control, whereas Paper 2, while rigorous and useful, is a more incremental theoretical-method extension with narrower application scope.
Paper 1 addresses a critical bottleneck in near-term quantum computing by combining quantum error correction and mitigation. Its potential to significantly reduce runtime costs and variance in the pre-fault-tolerant regime offers broad, immediate applications across quantum algorithms. Paper 2 presents a valuable but narrower advancement in quantum optics and single-photon generation.
Paper 2 demonstrates a significantly larger scale of experimental validation (up to 50 qubits on a 156-qubit processor) compared to Paper 1 (6-spin model). By overcoming the exponential sampling bottleneck of Probabilistic Error Cancellation (PEC) and demonstrating it on state-of-the-art hardware, Paper 2 offers a more immediate and transformative impact for large-scale NISQ applications. Its combination of variational blockwise aggregation and CUDA-Q accelerated validation provides a highly scalable and practical solution.