General framework for anticoncentration and linear cross-entropy benchmarking in photonic quantum advantage experiments
Zoltán Kolarovszki, Ágoston Kaposi, Zoltán Zimborás, Michał Oszmaniec
Abstract
Photonic architectures are one of the leading platforms for demonstrating quantum computational advantage, with Boson Sampling and Gaussian Boson Sampling as the primary schemes. Yet, we lack for these photonic primitives a systematic theoretical understanding of linear cross-entropy benchmarking (LXEB), which is a central tool for testing quantum advantage proposals. In this work, we develop a representation-theoretic framework for the classical computation of average LXEB scores and second moments of output probability distributions, covering a range of quantum advantage experiments based on scattering -photon states through -mode Haar-random interferometers. Our methods apply in any regime, including the saturated regime, where the (expected) number of photons is comparable to the number of optical modes. The same second-moment techniques also allow us to prove anticoncentration for traditional Fock-state Boson Sampling in the saturated regime. Interestingly, for Gaussian Boson Sampling second moments are not sufficient to establish a meaningful anticoncentration statement. The technical core of our approach rests on decomposing two copies of the -particle bosonic space into irreducible representations of . This reduces two-copy Haar averages to computing purities of initial states after partial traces over particles, highlighting the role that particle entanglement plays for LXEB and anticoncentration.
AI Impact Assessments
(3 models)Scientific Impact Assessment
Core Contribution
This paper develops a representation-theoretic framework for computing average LXEB scores and second moments of output probability distributions in photonic quantum advantage experiments. The two key results are: (1) an efficient classical algorithm for computing LXEB reference values for product input states evolved through Haar-random interferometers, applicable across Boson Sampling, Gaussian Boson Sampling, and several variants; and (2) the first proof of anticoncentration for Boson Sampling in the experimentally relevant saturated regime where m = Θ(n).
The technical core rests on decomposing two copies of the n-particle bosonic symmetric space Sym^n(C^m) into irreducible representations of U(m) via the Pieri rule, then expressing the resulting projectors P_k as linear combinations of "bosonic swap operators" S_q. This reduces Haar-averaged second moments to computing purities of particle-reduced density matrices, elegantly connecting LXEB scores to particle entanglement structure.
Methodological Rigor
The mathematical framework is carefully constructed. The decomposition into irreducible representations (Equation 20-21), the explicit construction of projectors via Young symmetrizers (Lemma 1), and the efficient algorithm for computing bosonic swap expectation values for product states (Proposition 2) form a coherent pipeline. The proof of anticoncentration (Theorem 1) uses the Paley-Zygmund inequality combined with a careful upper bound on anticoncentration scores, yielding AC(m,n) ≤ 19.1(m/n + 1). The asymptotic analysis showing lim AC(m,n) = 1 + m/n (Lemma 5) provides further confidence.
The numerical experiments using the Piquasso framework validate the analytical predictions, showing convergence of LXEB fidelity estimators for both Boson Sampling and Gaussian Boson Sampling. The paper also honestly acknowledges limitations—concentration of the LXEB score around its reference value is supported numerically but not proven, and the bound of 19.1 is likely not tight (numerics suggest ~1.285m/n would suffice).
One notable aspect is the transparent acknowledgment of AI assistance in the manuscript preparation, including a formula (Proposition 9) suggested by AI that substantially simplified the anticoncentration proof—an interesting methodological data point for the field.
Potential Impact
Immediate impact on photonic quantum advantage: The framework fills a critical gap by providing LXEB reference values in the saturated regime, which is where virtually all current experiments operate. Previous results were restricted to the dilute regime m = Ω(n²), severely limiting their practical applicability.
Complexity-theoretic implications: The anticoncentration result (Result 2) strengthens the hardness argument for saturated Boson Sampling by enabling the Stockmeyer reduction to produce multiplicative (rather than merely additive) approximations. This places photonic quantum advantage on comparable footing with RCS, IQP, and Fermion Sampling regarding the structure of hardness arguments.
Broader theoretical contributions: The proof of the t=2 case of the Hunter-Jones conjecture about moments of permanents, the volume-law entanglement result for typical Fock states in the saturated regime, and the exponential lower bounds on certification sample complexity are valuable byproducts.
Practical benchmarking: The poly(m,n)-time algorithm for LXEB reference values provides an immediately usable tool for experimentalists running photonic quantum advantage experiments, especially given its compatibility with losses, non-uniform squeezing, and displacements.
Timeliness & Relevance
This work is exceptionally timely. Recent experiments (Jiuzhang, Borealis, and the 3050-photon GBS experiment) all operate in regimes where prior LXEB theory was inadequate. The recent complexity-theoretic work by Bouland et al. extending hardness to the saturated regime created a need for matching anticoncentration results, which this paper directly addresses. The gap between experimental practice and theoretical understanding of benchmarking in photonic systems was a recognized bottleneck.
Strengths
1. Generality and unification: A single framework covers Boson Sampling, GBS, Scattershot BS, Displaced GBS, and Superposition BS, with and without losses.
2. First-quantization insight: The particle-entanglement interpretation of LXEB scores is conceptually illuminating and suggests natural extensions.
3. Practical computability: The algorithms are polynomial-time with explicit complexity bounds, not just existence results.
4. Sharp distinction between schemes: The framework reveals that anticoncentration behavior differs qualitatively across schemes—strong for Fock-state BS, weak for GBS/Scattershot BS, absent for single-source GBS—providing structural insight into why different photonic architectures have different complexity-theoretic properties.
5. Completeness: The paper includes closed-form expressions for several important cases, numerical validation, and a clear discussion of open problems.
Limitations
1. Haar-random assumption: Real experiments use structured (not Haar-random) interferometers; extending to finite-depth or structured circuits remains open.
2. No concentration proof: The paper does not prove that individual LXEB scores concentrate around the reference value, only supporting this numerically.
3. Threshold detectors excluded: The framework requires photon-number-resolving detection; the experimentally common threshold detection case is not covered.
4. Partial distinguishability not addressed: A major source of experimental imperfection is left for future work.
5. Anticoncentration bound looseness: The constant 19.1 appears significantly loose compared to numerical evidence, though this doesn't affect the qualitative conclusion.
Overall Assessment
This is a technically strong paper that resolves a significant open problem (anticoncentration for photonic sampling in the saturated regime) while simultaneously providing practical tools for benchmarking photonic quantum advantage experiments. The representation-theoretic approach is elegant and yields results that were inaccessible via prior methods. The work substantially advances the theoretical foundations of photonic quantum computing.
Generated Apr 17, 2026
Comparison History (40)
Paper 2 demonstrates large-scale quantum simulation of the Fermi-Hubbard model (120 qubits, 90 Trotter steps) on a real quantum processor, directly observing spin-charge separation and achieving quantitative agreement with classical methods while showing 3000× wall-clock speedup in challenging regimes. This has broader impact: it validates digital quantum processors for studying physically relevant fermionic many-body dynamics, bridges quantum computing and condensed matter physics, and establishes practical quantum utility. Paper 1, while theoretically rigorous and important for benchmarking photonic quantum advantage, addresses a more specialized audience focused on certification methodology rather than demonstrating new physical capabilities.
Paper 2 offers a broad, interdisciplinary perspective connecting quantum information science with nuclear and high-energy physics. While Paper 1 provides a highly rigorous and technical framework for a specific quantum advantage architecture, Paper 2 has a much wider scope, potentially shaping future research directions and algorithms across multiple fundamental physics domains, leading to a broader overall scientific impact.
Paper 1 likely has higher impact due to its broad, platform-agnostic framework for programmable long-range interactions and a concrete hybrid classical-quantum workflow that targets near-term analog quantum simulators. It claims large performance gains (fidelity/energy estimation) at experimentally relevant scales (100–1000 particles) and enables new controlled studies of thermalization, spanning multiple model classes and hardware types. Paper 2 is mathematically novel and timely for validation of photonic quantum advantage, but its impact is narrower (benchmarking theory for specific photonic primitives) and more indirect in near-term applications.
Paper 1 provides a foundational theoretical framework for benchmarking and proving anticoncentration in photonic quantum advantage experiments, addressing a critical and immediate challenge in current quantum hardware validation. In contrast, Paper 2 focuses on a specific circuit implementation for quantum Metropolis-Hastings that relies on future fault-tolerant regimes. Paper 1's rigorous mathematical approach and immediate applicability to state-of-the-art quantum experiments give it a higher potential for near-term scientific impact.
Paper 2 addresses a fundamental theoretical gap in photonic quantum advantage experiments—developing a systematic framework for linear cross-entropy benchmarking (LXEB) and anticoncentration in Boson Sampling variants. This has broad impact across quantum computing theory, quantum advantage verification, and representation theory. It provides foundational tools applicable to multiple experimental platforms and resolves open questions about the saturated regime. Paper 1, while practically useful for NISQ cryptography, is more incremental—optimizing circuit depth for quantum walks and applying them to key generation—with narrower scope and less theoretical depth.
Paper 1 likely has higher impact: it provides a general, representation-theoretic framework for analyzing LXEB and anticoncentration in photonic quantum advantage experiments across regimes (including saturated), directly addressing verification/validation—an urgent bottleneck for near-term quantum computing claims. The methods appear broadly applicable to Boson Sampling variants and connect to entanglement structure, potentially influencing theory, complexity, and experimental benchmarking. Paper 2 is a useful control proposal for optomechanical entanglement/steering with plausible applications, but is more domain-specific and may face higher implementation sensitivity, making its cross-field and near-term impact less certain.
Paper 2 likely has higher impact due to stronger novelty-to-application coupling: it introduces trainable recurrent quantum sequence models for coherent stochastic-process generation with linear-in-horizon circuit growth and a recurrent parameter-shift training rule. This directly targets high-value application areas (risk analysis, importance sampling, bio/sequence data) and intersects quantum ML, generative modeling, and quantum simulation, broadening cross-field influence and timeliness. Paper 1 is rigorous and important for validating photonic quantum advantage (LXEB/anticoncentration), but its impact is more specialized to benchmarking foundations within photonic sampling experiments.
Paper 1 provides a highly practical, computationally efficient solution to a critical bottleneck in quantum error correction (decoding LDPC codes). Its direct applicability to near-term hardware (IBM's targeted codes) and the simplicity of its modulo-w prediction give it significant real-world impact and timeliness, whereas Paper 2 focuses on more theoretical aspects of photonic quantum advantage.
Paper 2 addresses a fundamental gap in the theoretical understanding of photonic quantum advantage experiments—a highly active and competitive research area. It develops a systematic framework for linear cross-entropy benchmarking across multiple photonic schemes, proves anticoncentration in new regimes, and reveals the role of particle entanglement. Its breadth of impact (quantum computing, representation theory, experimental validation) and timeliness given ongoing quantum advantage claims give it higher impact than Paper 1, which, while mathematically elegant, provides incremental improvements to trace inequalities with more narrowly scoped applications in quantum information theory.
Paper 2 addresses a critical and highly relevant challenge in quantum computing: verifying quantum computational advantage. By providing a broad theoretical framework for benchmarking photonic architectures, it has foundational implications for a rapidly growing field. Paper 1, while demonstrating a useful experimental scheme for single-photon generation, is much narrower in scope and application compared to the field-wide implications of Paper 2.
Paper 1 develops a systematic representation-theoretic framework for linear cross-entropy benchmarking across photonic quantum advantage experiments, addressing a fundamental gap in the theoretical understanding of Boson Sampling and Gaussian Boson Sampling. It proves anticoncentration results in the saturated regime and connects particle entanglement to LXEB scores—results with broad implications for quantum computational advantage claims. Paper 2 presents a useful but more incremental variational circuit approach for quantum metrology in noisy systems, limited to small system sizes and a narrower application domain.
Paper 2 addresses fault-tolerant quantum computing (FTQC) via surface codes, which is universally recognized as critical for scalable quantum computing. Its proposed framework provides substantial, quantifiable reductions in both space and time overheads, accelerating the timeline to practical quantum algorithms. While Paper 1 offers a rigorous theoretical framework for evaluating photonic quantum advantage, Paper 2's impact is broader and more applied, directly solving pressing engineering bottlenecks in quantum error correction that apply across multiple hardware modalities.
Paper 2 addresses a critical bottleneck in near-term quantum computing: verifying quantum advantage in photonic architectures. By providing a rigorous framework for linear cross-entropy benchmarking and proving anticoncentration in the saturated regime, it has immediate, high-profile applications for validating leading experimental platforms. While Paper 1 offers valuable algorithmic advancements for simulating open quantum systems, Paper 2's direct relevance to foundational claims of quantum supremacy gives it broader and more timely scientific impact across both theoretical and experimental physics.
Paper 2 likely has higher scientific impact due to broader and more immediate real-world applicability: an end-to-end LLVM-based quantum-classical co-compilation stack affects many users, platforms, and workloads, and can be adopted by industry and HPC communities. Its benchmarking claims (depth/gate reductions on MQTBench) indicate practical performance gains and methodological evaluation. Paper 1 is theoretically novel and rigorous for photonic quantum advantage verification, but its impact is narrower (specialized to boson-sampling/LXEB theory) and less directly enabling across the wider quantum software ecosystem.
Paper 2 likely has higher impact: it demonstrates a concrete, experimentally realized waveguide-QED platform with electrically tunable, long-range (13 μm) radiative coupling, switchable directional emission, and direction-dependent photon statistics—capabilities directly relevant to scalable on-chip quantum networks and quantum photonics. The results are timely for integrated quantum technologies and broadly applicable across nanophotonics, quantum optics, and device engineering. Paper 1 is theoretically novel and rigorous for photonic quantum advantage verification, but its immediate real-world applicability and cross-field reach are narrower than an experimental platform enabling new functionalities.
Paper 2 addresses a fundamental theoretical gap in quantum computational advantage—developing a systematic framework for linear cross-entropy benchmarking in photonic systems. It provides broadly applicable representation-theoretic tools, proves anticoncentration results in new regimes, and reveals structural insights (particle entanglement's role) relevant to the entire quantum advantage community. Paper 1, while achieving impressive speedups for tensor network simulations, is more incremental and narrower in scope, building on existing PTSBE methods with engineering-level optimizations rather than establishing new theoretical foundations.
Paper 1 addresses a fundamental and broadly applicable problem in quantum information—the robustness of quantum state determination from local marginals. It provides universal theoretical guarantees with practical certification via semidefinite programming, classifies multipartite states by robustness exponents, and constructs scalable entanglement witnesses. This has broad impact across quantum tomography, entanglement detection, and experimental quantum information. Paper 2, while technically strong, addresses a more specialized topic (LXEB for photonic quantum advantage) with a narrower audience. Paper 1's results are more likely to influence multiple subfields and experimental practice.
Paper 1 likely has higher impact: it advances a systematic, representation-theoretic framework for LXEB and anticoncentration in photonic quantum advantage—central, timely problems for validating near-term quantum supremacy claims. It applies across regimes (including saturated), connects benchmarking to particle entanglement, and clarifies limits for Gaussian Boson Sampling, offering broadly useful theory for experiments and classical simulability. Paper 2 is novel and rigorous but more specialized (metastable macrospin dynamics) with narrower immediate cross-field influence and fewer direct ties to a high-profile, rapidly moving benchmarking ecosystem.
Paper 2 provides a general framework for benchmarking and proving anticoncentration in photonic quantum advantage experiments. Its focus on linear cross-entropy benchmarking and Boson Sampling addresses critical, highly active challenges in verifying quantum supremacy, offering immediate theoretical and experimental applications. Paper 1, while rigorously analyzing coherence in the HHL algorithm, is more narrowly focused on specific mathematical properties of a single algorithm, making its broader impact more limited.
Paper 1 likely has higher impact: it addresses a timely, high-visibility area (verification of photonic quantum advantage) and provides a broadly applicable, representation-theoretic framework for LXEB and anticoncentration across regimes, including the experimentally relevant saturated regime. The results directly affect benchmarking methodology and interpretation of near-term experiments, with potential cross-field influence (quantum information, photonics, random matrix/rep theory). Paper 2 offers a classification of static SU(2) Yang–Mills solutions with a specialized ansatz; it is more niche, with less clear immediate experimental or broad methodological uptake.