arXiv subjects

Feng Pan

Publications and source records attributed to Feng Pan.

At least 19 recordsLinked to original sources

arXiv2026

ACE-Brain-0.6: Scaling Spatial Grounding for Physical Foundation Models

General-purpose multimodal models, including GPT-6 Astra and Claude Opus 5.5, demonstrate broad capabilities in embodied manipulation. Our analysis indicates that this generality is accompanied by strong spatial perception, as Astra can determine where to interact with task-relevant entities amid visual distractors. Because spatial perception connects high-level reasoning to the physical scene, we investigate scaling visual grounding to strengthen the perceptual foundation of general embodied operation. We develop a unified data pipeline that combines robot manipulation recordings with complementary visual data from diverse sources. After automatically annotating the robot recordings, the pipeline converts the collected data into verifiable grounding supervision. Fine-tuning Qwen3.5-9B on approximately 1.5 million samples produced by this pipeline yields ACE-Brain-0.6. The model outperforms GPT-6 Astra on four of seven public grounding benchmarks, with an overall average score of 75.41 compared with Astra's 73.68. Under a fixed Qwen3.5-9B fine-tuning setup, aggregate grounding performance improves steadily as successive data releases expand the training collection to cover a broader range of grounding scenarios. These results show that scaling verifiable grounding supervision from diverse sources can strengthen the spatial perception of an open 9B model for general embodied operation.

cs.RO↗

arXiv2026

Localized high-Q resonances enable electrically defined beam steering in lithium niobate metasurfaces

High-speed beam steering for LiDAR and optical communication requires independent, reconfigurable control of optical phase. We demonstrate a metasurface combining lithium niobate's Pockels effect with high-Q (Q > 1700) silicon guided-mode resonance nanoantennas whose localized resonances allow individual electro-optic control at a 1 $μ$m pitch. Identical arrays steer light to 53.5$^\circ$ or 32.4$^\circ$, with the angle set by the bias pattern rather than antenna geometry. We measure 18.0% 0th-order modulation depth, 0.09% of incident power in the steered order, and a 940 MHz detected modulation bandwidth. Mode-overlap analysis shows how inter-element coupling limits usable Q, providing design rules for compact, high-speed beam steering.

physics.optics↗

arXiv2026

Field-free Magnetoacoustic Nonreciprocity via Exchange-bias Engineering

Nonreciprocal wave propagation is essential for directional signal processing and isolation. Magnetoacoustic systems are attractive platforms for nonreciprocal devices because of their small footprint and passive operation. However, pronounced nonreciprocity typically requires an external bias field,limiting device miniaturization and integration. Here, we demonstrate field-free magnetoacoustic nonreciprocity in an IrMn/FeGaB/Cu/FeGaB multilayer integrated with a shear-horizontal-wave delay line. Exchange-bias engineering stabilizes a remanent magnetic state with broken time-reversal and spatial-inversion symmetries while tailoring the spin-wave dispersion to achieve resonant magnetoacoustic coupling at zero field. The normalized isolation reaches 21 dB/mm, with its sign programmable by the field-sweep history. Further numerical optimization predicts a zero-field normalized isolation of 91 dB/mm by increasing the dynamical magnetic asymmetry. Our work establishes exchange-bias engineering as a promising route toward compact RF isolators and circulators and as a general strategy for realizing magnetoacoustic devices without external magnetic fields.

cond-mat.mtrl-sci↗

arXiv2026

Differentiable Maximum Likelihood Noise Estimation for Quantum Error Correction

Accurate noise estimation is essential for fault-tolerant quantum computing, since decoding per formance depends critically on the fidelity of circuit-level noise parameters. We introduce a differ entiable Maximum Likelihood Estimation (dMLE) framework that learns these parameters directly from experimental syndrome data using likelihood gradients, with no decoder in the training loop. Mapping syndrome likelihoods to statistical-mechanical models enables exact and efficient evalua tion via planar solvers for repetition codes and tensor networks (TNs) for surface codes, extending the exact treatment to code distance d = 5 with a small number of rounds, and, via spatial subsam pling combined with time-translation symmetry, to larger codes and larger round counts. Across four experimental superconducting datasets, dMLE reduces logical error rates by up to 30.6% for repetition codes. On the Sycamore d = 3 surface code, a TN decoder using the dMLE prior attains a logical error rate comparable to that reported for state-of-the-art AlphaQubit, using standard syndromes alone without auxiliary leakage data. On the larger d = 5 surface code the dMLE prior lowers the logical error rate of the TN and Tesseract decoders by 7.18% and 5.23% relative to a prior optimized by reinforcement learning; at d = 7 it lowers that of Tesseract by 4.56% and 4.83% in the X and Z bases.

quant-ph↗

arXiv2026

CT-bMPS: Approximate Tensor Network Contraction with Boundary MPS on Contraction Trees

We introduce contraction tree boundary matrix product states (CT-bMPS), a scalable approximate contraction method for general closed tensor networks. CT-bMPS represents the boundary tensor and its complementary environment at each edge of a contraction tree as matrix product states. This representation allows local compression based on a reduced transition matrix to incorporate environment information along arbitrary contraction trees, while keeping the time and memory costs polynomial in the network size and bond dimension. We benchmark the method on a variety of models and applications, including Ising partition functions, random tensor networks, kicked Ising dynamics, and quantum error correction decoding. These benchmarks demonstrate improved accuracy and computational efficiency over existing general-purpose approximate contraction methods, together with the ability to handle larger and more complex networks. Optimized contraction trees and environment updates substantially improve accuracy without increasing the bond dimension. These results establish CT-bMPS as a general and scalable approximation framework with flexible contraction orders and environment-based compression for a wide range of applications.

quant-ph↗

arXiv2026

Lightweight, Rubric-Guided Trajectory Evaluation for Production AI Agents

Trajectory evaluation is essential for improving the reliability of LLM-based agents, but production use makes it expensive to run repeatedly. Modern agents generate long traces containing tool calls, observations, retries, and external outputs, while not all raw tokens are equally useful for diagnosis. We present \textit{LiteTrajEval}, a lightweight architecture for budget-bounded trajectory evaluation. LiteTrajEval derives compact domain-specific rule profiles offline, then preprocesses each trajectory online, marks heuristic failure signals, serializes it under a fixed global budget, and invokes a single rubric-guided LLM judge to produce structured diagnostic reports. Evaluated on public Magentic-One-style and $τ$-bench-style trajectory datasets, LiteTrajEval improves failure-localization alignment with human annotations by roughly 20--35 percentage points on Magentic-One and up to 23 percentage points on $τ$-retail compared with AgentRx, while reducing cost by about 6$\times$ and evaluation time by more than 8$\times$. This solution has also been deployed in our enterprise agentic platform.

cs.AI↗

arXiv2026

Exact Maximum Likelihood Decoding beyond Treewidth via Rank-Decomposition Dynamic Programming

Maximum-likelihood decoding provides an optimal decoding strategy for quantum error correction under stochastic Pauli noise. However, computing logical-class probabilities is challenging, and the leading exact tensor-network contraction requires time exponential in treewidth. In this work, we introduce a new decoding algorithm based on rank-decomposition dynamic programming (Rank DP). We express decoding partition functions with independent local fault factors as quadratic sums of powers and apply Rank DP. The resulting exact evaluator has arithmetic complexity polynomial in the input size and exponential in the Tanner graph's rank-width with respect to the fault partition, including decomposition construction. After Gaussian elimination, it gives polynomial-time decoding for punctured quantum Reed-Muller codes and a Steane-concatenated family with growing distance under independent single-qubit Pauli noise. Standard contraction of the corresponding tensor networks requires superpolynomial time. A nonnegative realization gives relative floating-point error bounds under explicit arithmetic assumptions. Numerical experiments demonstrate runtime advantages over the tested tensor-network implementations on selected code-capacity and circuit-level instances, with full likelihood evaluation for Reed-Muller codes up to $1{,}023$ qubits. We further use these likelihoods to learn circuit noise parameters from syndromes, evaluate rare postselection probabilities, and quantify decoder optimality gaps. Our work opens new avenues for exploiting algebraic structure in quantum decoding and noise characterization.

quant-ph↗

arXiv2026

Bridging Ab Initio Symmetries and Global Nuclear Masses with Interpretable Neural Networks

Ab initio theory establishes Wigner's SU(4) and Elliott's SU(3) as dominant symmetries of the nuclear force in light and intermediate-mass nuclei. Previous work shows the relevance of the former symmetry for nuclear binding, whether the latter organizes binding remains elusive. We probe whether both these symmetries organize nuclear masses, aiming at physical insights through interpretable models and predictive capability. From the SU(3) and SU(4) Casimirs we build 3 neural-network models. Two are conventional, a feature-informed NN (FINN) and a Gaussian variant (GINN) with predictive spread, while Wigner-informed network (WINN) is a new design constraining the mass formula to be linear in the operators, learning their (N,Z)-dependent couplings, so that the model is intrinsically explainable. All are trained on AME2016 subtracted by the liquid drop model at 4 data fractions and validated on nuclei new to AME2020, with extrapolation benchmarked against HFB-26 and r-process. The Casimir features carry binding information far beyond the bulk, and SHAP analysis suggests the quadratic SU(4) Casimir as the leading contributor to the residual binding. The WINN yields the best performance, reaching a 0.412 MeV validation error and competitive with state-of-the-art models, and importantly, when trained on the sparsest dataset it outperforms the other NNs trained on the densest. Off the known chart its masses track HFB-26 as closely as WS3 and reproduce the solar abundance peaks of a neutron-star-merger simulation. The WINN's coupling fields reveal an enhanced even-SU(4) contribution toward the neutron dripline, hinting at restoration of Wigner's symmetry. The SU(4) and SU(3) structures reach beyond individual nuclei to organize binding, and embedding symmetry-preserving operators directly in a domain-informed interpretable architecture yields a physically transparent model less hungry for data.

nucl-th↗

arXiv2026

Sampling hard circuits with verifiably high fidelity

Sampling-based proposals are prominent candidates for demonstrating quantum computations beyond the reach of classical supercomputers. However, it has been difficult to combine their complexity-theoretic hardness with two capabilities needed for scalable quantum computing more generally: suppressing hardware errors, and verifying the quantum computation itself. Here we address both issues by introducing structured circuits, which, in addition to provable hardness guarantees, admit an encoding in a quantum code. This allows us to simultaneously reach high fidelities at high circuit depths, and to certify an experimental fidelity via the circuit structure and measurement of code syndromes. The resulting certificate is device dependent, but requires substantially weaker noise assumptions than existing fidelity proxy benchmarks. We demonstrate our proposal with a $64$-qubit, depth-$73$ Clifford circuit, doped with $314$ $T$ gates. We use a total of $76$ physical qubits to encode this computation in spacetime codes, effectively suppressing gate error rates by $10\times$ after syndrome post-selection, and yielding a state with a fidelity lower bound of $0.349$ with $95\%$ confidence. Our construction is a systematic method for promoting a stabilizer state to a magic state while keeping an error-detected fidelity certificate.

quant-ph↗

arXiv2026

Classical Simulation and Design Frontiers for IBM's Doped Clifford Sampling Experiment

We classically simulate the IBM doped Clifford random circuit sampling experiment, comprising $70$ qubits, $70$ entangling layers, and $468$ inserted $T$ gates. A deterministic temporal-boundary tensor network contraction approach is specifically designed to tackle such open-boundary one-dimensional brickwork circuits with operator-Schmidt-rank-$2$ entangling gates. For an $n$-qubit circuit of depth $d$, the resulting unsliced path evaluates an exact amplitude with contraction width $\lceil d/2\rceil$; Ratcatcher calculations certify that no smaller width is possible for the tested instances. Because one-qubit gates are absorbed without changing the network topology, the width and dense scheduled contraction cost are independent of their values and of the number and placement of $T$ gates. For the IBM instance, its largest intermediate tensor contains $2^{35}$ complex64 entries (256 times smaller than IBM's estimation), corresponding to a tensor payload of $256$ GiB, and is distributed across eight GPUs within a node. Using 32 nodes, with eight NVIDIA H100 GPUs per node, we completed all 2051 amplitude batches corresponding to IBM's published output bitstrings in 37.3 minutes. The resulting probabilities yield a log-XEB estimate of $0.35034$ with a 95\% interval of $[0.29763,0.40305]$. Under the Porter--Thomas and scrambled-noise assumptions, this is numerically compatible with IBM's fidelity lower bound; separately, fidelity-weighted resource accounting projects a 583-contraction workload with a 10.6-minute makespan on the same 32 nodes. More broadly, the approach provides a practical diagnostic for experimental outputs and a quantitative tool for designing future doped Clifford sampling experiments.

quant-ph↗

arXiv2026

Credit the Right Box: Marginal Contribution Assignment for Structured Visual Perception

Multimodal Large Language Models (MLLMs) are increasingly expected to solve structured perception tasks that require visual recognition, language-to-object binding, object cardinality preservation, and precisely localized grounding and segmentation outputs. However, existing group-relative reinforcement learning methods provide only response-level supervision, creating a granularity mismatch for structured multi-object prediction: a single advantage is broadcast to all tokens in a response, without distinguishing individual box contributions. To address this mismatch, we propose MCR-GRPO, a marginal contribution assignment framework that derives box-level credit directly from each sampled response. Specifically, Marginal Contribution Reward (MCR) estimates each predicted box's contribution through a leave-one-out comparison, measuring how the matched set value changes when the box is removed from the response. After within-response normalization, records that improve the set value receive positive credit, while redundant or harmful ones are suppressed. To make marginal attribution stable and informative, we further introduce a Continuous Matched Set Value Evaluator that integrates permutation-invariant matching, count-aware normalization, and graded localization. MCR-GRPO maps normalized box-level marginal advantages to the token spans that generated each box, preserving GRPO's response-level comparison while enabling box-aware optimization of structured multi-object grounding. Experiments across REC, DOD, segmentation, and counting benchmarks show state-of-the-art performance over prior GRPO-based baselines.

cs.CV↗

arXiv2026

ACE-Brain-0.5: A Unified Embodied Foundational Model for Physical Agentic AI

Embodied AI is moving from isolated perception or action modules toward physical agents that understand, plan under goals, act through robot bodies, monitor progress, and improve from experience. Existing systems address this loop only in parts: end-to-end policies generate actions but often lack spatial reasoning, planning, and execution assessment, while robot-agent systems orchestrate tools or specialists but do not learn a shared representation. This fragmentation limits general Physical Agentic AI. We present ACE-Brain-0.5, a unified embodied foundation model that organizes robot intelligence into five coupled functions: spatial perception, decision making, embodied interaction, self-monitoring, and self-improvement. Built on ACE-Brain-0, which established spatial intelligence as a shared scaffold across robot platforms, ACE-Brain-0.5 extends an understanding-centric model into a closed-loop foundation model. A single 8B backbone instantiates the first four functions: grounding objects and affordances, reasoning over 3D and egocentric spatial relations, decomposing instructions into subgoals, generating navigation and manipulation actions, and estimating progress for verification and recovery. To unify these capabilities without cross-task interference, we introduce SSR+, which extends Scaffold-Specialize-Reconcile with a Reactivate stage after task-vector merging. The fifth function, self-improvement, is realized by a companion framework that updates external execution state, including task schemas, spatial memory, and failure-recovery cases, from rollouts. Across fifteen benchmarks, ACE-Brain-0.5 improves over ACE-Brain-0 on 14 of 18 spatial perception and grounding benchmarks, achieves competitive navigation and manipulation performance, and provides strong progress estimation in ID and OOD settings. Together, these results mark an early step toward general Physical Agentic AI.

cs.RO↗

arXiv2026

UAV-MapFusion: RTK-Aligned Uncertainty-Aware Coarse-to-Fine Multi-Session UAV Mapping

Large-scale point cloud maps are essential for robotics and spatial intelligence tasks. UAVs provide an efficient means for large-scale map acquisition; however, due to limited flight endurance and onboard storage, mapping a large-scale scene within a single flight remains difficult. Existing multi-session map merging methods can extend the mapping range, yet in UAV scenarios they still struggle to simultaneously suppress long-range drift and preserve local geometric accuracy. To address this issue, an uncertainty-aware multi-session point cloud map merging and coarse-to-fine optimization system is proposed. The proposed method first performs initial multi-session map merging based on a scene graph, and then incorporates RTK observations through an RTK spatiotemporal alignment module, where temporal offsets are estimated using Dynamic Time Warping (DTW), and continuous RTK constraints are recovered using Multi-Output Gaussian Processes (MOGP) under incomplete sampling and frame dropouts. On this basis, a unified uncertainty-aware factor graph is constructed, and local geometric accuracy is further improved through iterative plane-factor refinement. Experiments on real-world datasets validate the effectiveness and robustness of the proposed method. To facilitate further research and development in the community, our code and dataset will be publicly released.

cs.RO↗

arXiv2026

Efficient foundation decoders for fault-tolerant quantum computing

Foundation decoders, a class of high-capacity neural decoders, are leading candidates for fault-tolerant quantum computing, with accurate and efficient decoding at large code distances. However, their construction often faces a steep scaling barrier, as larger code distances rapidly amplify the cost of syndrome generation and neural optimization. To address this bottleneck, here we devise neural transfer unification (NTU), a unified framework for efficient foundation decoders. A central feature of NTU is its ability to align decoding tasks across code distances via algebraic structures shared by scalable code families, which enables knowledge learned on smaller codes to accelerate large-scale decoder training. We instantiate NTU as NTU-Transformer, a transformer-based neural decoder tailored for planar surface codes and bivariate bicycle codes. For planar surface codes under circuit-level noise, NTU-Transformer outperforms correlation-aware matching on the $[\![361,1,19]\!]$ code and further scales to the $[\![625,1,25]\!]$ code, where it exceeds standard matching through transfer adaptation. For the bivariate bicycle code with $[\![72,12,6]\!]$, it surpasses Relay-BP in the low-physical-error regime. These results establish our proposal as a scalable route to amortized cross-distance training of foundation decoders for fault-tolerant quantum processors.

quant-ph↗

arXiv2026

FAST-LIVGO: A Degeneracy-Robust LiDAR-Inertial-Visual-GNSS Fusion Odometry

Robust state estimation and mapping in long-term, large-scale, and highly dynamic environments remains a key challenge in robotics. Existing LiDAR-Inertial-Visual Odometry (LIVO) systems achieve strong local accuracy but suffer from accumulated drift over long distances and may fail in geometrically degraded or textureless scenes. Meanwhile, GNSS-aided fusion frameworks often rely on LiDAR or visual odometry for state prediction and outlier rejection, making them vulnerable when odometry degenerates. To address these limitations, we propose a tightly coupled LiDAR-Inertial-Visual-GNSS fusion framework based on an Error-State Iterated Kalman Filter. An online spatiotemporal alignment module using Dynamic Time Warping is introduced for highly dynamic conditions. To better exploit GNSS precision, we develop observation models based on Doppler shifts and fixed-anchor Time-Differenced Carrier Phase, providing millimeter-level relative constraints without augmenting historical anchor states. We further design a degeneracy-aware dual-mode outlier rejection strategy that switches between LIVO-prior-guided rejection and GNSS-aided recovery according to the LIVO degeneracy level. Experiments on the public M3DGR dataset and a custom 20~m/s fixed-wing UAV dataset demonstrate that our system reduces accumulated drift and map ghosting, outperforming state-of-the-art methods in accuracy and robustness.

cs.RO↗

arXiv2026

Parallelizing Large-Scale Tensor Network Contraction on Multiple GPUs

Exact tensor network contraction underpins quantum circuit simulation, quantum error correction, combinatorial optimization, and many-body dynamics. The dominant parallelization strategy, slicing, scales exponentially and incurs redundant computation. We present a multi-GPU framework that instead distributes intermediate tensors across devices with explicit communication, converting a fixed contraction path into a communication-efficient schedule via GEMM-oriented mode reordering and communication-aware mode distribution planning. Within a single DGX H100 node (8 GPUs, NVLink), distribution delivers $7$--$173\times$ extra speedup beyond embarrassingly parallel slicing, capturing nearly all of the available compute reduction (87--101%) because NVLink's high bandwidth keeps communication small relative to compute. Scaling the same four workloads to 1024 H100 GPUs over InfiniBand, the extra speedup beyond slicing ranges from $42\times$ to $67{,}869\times$, demonstrating that communication-aware distributed contraction far surpasses slicing-based scaling limits for frontier tensor networks.

cs.DC↗

arXiv2026

Maximum Likelihood Decoding of Quantum Error Correction Codes

Quantum error correction (QEC) is indispensable for realizing fault-tolerant quantum computation, yet its effectiveness hinges critically on the classical decoding algorithm that interprets noisy syndrome measurements. Among all possible decoding strategies, maximum likelihood decoding (MLD) is provably optimal, since it identifies the logical group with largest likelihood by summing over all possible errors within logical class consistent with the observed syndrome. Despite its optimality, MLD is computationally intractable in general (#P-hard), motivating a rich landscape of exact and approximate algorithms. In this topical review, we provide a unified perspective on MLD by surveying recent advances through three complementary lenses: statistical mechanics, tensor networks, and artificial intelligence. From the statistical mechanics viewpoint, the MLD problem maps onto evaluating partition functions of disordered spin models, enabling exact solutions for certain codes and noise models as well as threshold estimation via phase-transition analysis. From the tensor network perspective, approximate contraction of tensor networks on the code's factor graph yields decoders that closely approach MLD accuracy with polynomial computational cost. From the artificial intelligence perspective, neural-network-based decoders, including autoregressive generative models and recurrent transformers, learn to approximate the MLD distribution from data, achieving high accuracy with the parallelism afforded by modern hardware accelerators. We discuss the connections among these three approaches, review their application to both simulated and experimental quantum hardware, and outline open challenges including real-time decoding, scalability to large code distances, and generalization to high-rate quantum low-density parity-check codes.

quant-ph↗

arXiv2026

Spin dissymmetry in optical cavities

We introduce the spin dissymmetry factor, a measure of the spin-selectivity in the optical transition rate of quantum particles. This spin dissymmetry factor is valid locally, including at material interfaces and within optical cavities. We design and numerically demonstrate a metasurface optical cavity with three-fold rotational symmetry that maximizes spin dissymmetry, thereby maximizing the spin-selective radiative coupling of a cavity-coupled emitter. We also show the near-field and far-field response of spin and chiral dipoles to these cavities that preferentially enhance either spin or chirality. Our approach emphasizes the difference between spin and chirality in the near-field and reveals a compact parameter for designing more efficient quantum optical devices.

physics.app-ph↗