arXiv subjects

Fei Meng

Publications and source records attributed to Fei Meng.

At least 19 recordsLinked to original sources

arXiv2026

Weak Permanent Anti-Concentration for Random Gaussian Matrices in Boson Sampling

Recent demonstrations of quantum computational advantage have been driven largely by sampling problems. A prominent model, boson sampling, involves sampling from the output distribution of a linear optical network. However, its classical hardness hinges on two plausible yet less-studied conjectures: the average-case hardness of approximating Gaussian permanents, and the permanent anti-concentration conjecture (PACC). The PACC is a purely mathematical assertion regarding the distributional properties of random Gaussian matrices. While the typical magnitude of the permanent has been established for discrete random matrices, the complex Gaussian case, which governs transition amplitudes in linear optical networks, has remained open. Here, we establish a weak anti-concentration bound by upper-bounding the probability that a random Gaussian permanent is superexponentially smaller than its standard deviation. Tightening this bound to an inverse-polynomial fraction would prove the original PACC. As a corollary, we establish the typical magnitude of Gaussian permanents, on par with Tao and Vu's seminal result for Bernoulli matrices. Combined with the Aaronson-Arkhipov framework, our result implies that classically simulating boson sampling to within a superexponentially small total variation distance would collapse the polynomial hierarchy, assuming the remaining conjectures hold.

quant-ph↗

arXiv2026

Text region detection in historical astronomical diagrams

Text detection is a crucial task in the analysis of historical documents. While datasets and benchmarks exist for text detection in manuscripts and maps, the study of text in mathematical diagrams has received little attention. To address this, we introduce a large-scale, diverse, open-access dataset of 948 historical astronomical diagrams containing 10,940 oriented polygonal text regions. Our dataset spans ten centuries (8th to 18th) and seven main linguistic traditions: Arabic and Persian (115), Chinese (332), Byzantine (233), Latin (185), Hebrew (48), and Sanskrit (35). It captures a wide range of diagram styles and textual content, from symbols to multi-line paragraphs. Each text instance is annotated with ordered polygons that precisely delineate text regions and encode the reading direction. In addition, we annotated the 2,293 regions in Latin diagrams with 20 class labels. We evaluated several strong baselines on our dataset, including TESTR, DeepSolo++, and Poly-DETR, a simple extension of DINO-DETR that we design to predict ordered polygon vertices. Poly-DETR achieves state-of-the-art performance on the MTHv2 and cBAD2019 benchmarks and provides a solid, simple baseline on our dataset. Code and dataset available online.

cs.CV↗

arXiv2026

Robust Fall Recovery for Armless Bipedal-Wheeled Robots Via Force-Guided Learning

Fall recovery is critical for autonomous legged locomotion. Existing methods have demonstrated that some legged robots, such as humanoids and quadrupeds, are capable of fall recovery from diverse postures by utilizing arms or coordinating multi-legs to generate support forces. Without arms or other legs to provide supportive assistance, a bipedal-wheeled robot must rely solely on the actuation of its legs, making recovery particularly difficult. To address this, we introduce FTSR (Force-guided Teacher-student framework with Stage-wise Rewards). The force-guided method constructs an external auxiliary force during simulation training that correlates directly with the robot's real-time height, explicitly formulating this force as an optimizable constraint. Through constrained reinforcement learning, the policy is guided toward reducing force dependency gradually and increasing the body height, developing internal recovery strategies despite having no arms for support. Height-progressive stage-Wise rewards progressively structure posture stabilization during recovery and transition to sustained locomotion, integrated with teacher-student architecture distilling privileged knowledge of force effects and recovery dynamics. After simulation training, the policy is deployed on a physical armless bipedal-wheeled robot and extensively evaluated. Experiments confirm robust and reliable fall recovery under diverse challenging conditions, demonstrating strong environmental adaptability and motion robustness, while maintaining full post-recovery motion capability. The framework also generalizes effectively to a high-DOF humanoid, confirming its practical generalizability. The project page is available at https://2350575870.github.io/force-guided.github.io/

cs.RO↗

arXiv2026

Forward-Assisted Purification: A Spatiotemporal Framework Beyond Conventional Limits

Noise remains the primary obstacle to realizing quantum advantage, continuously degrading the resources that enable quantum technologies. Purification aims to reverse this degradation by extracting high-fidelity resources from noisy ensembles, yet its conventional formulation is intrinsically static, acting only after noise has taken effect. Here we instead recast purification as a dynamical task, introducing a spatiotemporal framework that distributes interventions across the noise process. This formulation reveals operational capabilities inaccessible to existing approaches and gives rise to forward-assisted purifications that extend achievable performance. In certain regimes, a single-copy protocol already exceeds what can be achieved with up to 50 copies under conventional purification, demonstrating a significant overhead in required resources. Beyond these gains, our framework circumvents no-purification theorems within conventional protocols, including for Bell-state ensembles, thereby enabling purification previously considered impossible and pointing toward an efficient route to mitigating noise in quantum systems.

quant-ph↗

arXiv2026

TacSE3: Equivariant SE(3) Motion Estimation from Low-Texture Visuotactile Images for In-Gripper Tracking and Compensation

Robotic in-hand manipulation requires reliable object-motion tracking under frequent visual occlusion, yet low-texture visuotactile images provide few stable correspondences for conventional image- or geometry-matching methods. This paper presents TacSE3, a tactile motion-estimation pipeline that converts low-texture visuotactile observations into a decoupled three-dimensional force field and estimates incremental rigid-body motion on SE(3). The method derives planar translation from contact-centroid motion and estimates rotation primarily from shear-related tactile responses, yielding a physically interpretable signal for in-gripper tracking and compensation. Experiments with paired DM-Tac fingertip sensors show that dual-sensor sensing reduces translation-rotation ambiguity, supports rotation tracking across axes and object geometries, and provides a lightweight compensation signal that improves disturbance tolerance in downstream manipulation tasks without retraining the base policy.

cs.RO↗

arXiv2026

PoM: A Linear-Time Replacement for Attention with the Polynomial Mixer

This paper introduces the Polynomial Mixer (PoM), a novel token mixing mechanism with linear complexity that serves as a drop-in replacement for self-attention. PoM aggregates input tokens into a compact representation through a learned polynomial function, from which each token retrieves contextual information. We prove that PoM satisfies the contextual mapping property, ensuring that transformers equipped with PoM remain universal sequence-to-sequence approximators. We replace standard self-attention with PoM across five diverse domains: text generation, handwritten text recognition, image generation, 3D modeling, and Earth observation. PoM matches the performance of attention-based models while drastically reducing computational cost when working with long sequences. The code is available at https://github.com/davidpicard/pom.

cs.CV↗

arXiv2026

Provably Safe Trajectory Generation for Manipulators Under Motion and Environmental Uncertainties

Robot manipulators operating in uncertain and non-convex environments present significant challenges for safe and optimal motion planning. Existing methods often struggle to provide efficient and formally certified collision risk guarantees, particularly when dealing with complex geometries and non-Gaussian uncertainties. This article proposes a novel risk-bounded motion planning framework to address this unmet need. Our approach integrates a rigid manipulator deep stochastic Koopman operator (RM-DeSKO) model to robustly predict the robot's state distribution under motion uncertainty. We then introduce an efficient, hierarchical verification method that combines parallelizable physics simulations with sum-of-squares (SOS) programming as a filter for fine-grained, formal certification of collision risk. This method is embedded within a Model Predictive Path Integral (MPPI) controller that uniquely utilizes binary collision information from SOS decomposition to improve its policy. The effectiveness of the proposed framework is validated on two typical robot manipulators through extensive simulations and real-world experiments, including a challenging human-robot collaboration scenario, demonstrating sim-to-real transfer of the learned model and its ability to generate safe and efficient trajectories in complex, uncertain settings.

cs.RO↗

arXiv2026

Beyond Retention: Orchestrating Structural Safety and Plasticity in Continual Learning for LLMs

Continual learning in Large Language Models (LLMs) faces the critical challenge of balancing stability (retaining old knowledge) and plasticity (learning new tasks). While Experience Replay (ER) is a standard countermeasure against catastrophic forgetting, its impact across diverse capabilities remains underexplored. In this work, we uncover a critical dichotomy in ER's behavior: while it induces positive backward transfer on robust, unstructured tasks (e.g., boosting performance on previous NLP classification tasks through repeated rehearsal), it causes severe negative transfer on fragile, structured domains like code generation (e.g., a significant relative drop in coding accuracy). This reveals that ER trades structural integrity for broad consolidation. To address this dilemma, we propose \textbf{Orthogonal Subspace Wake-up (OSW)}. OSW identifies essential parameter subspaces of previous tasks via a brief "wake-up" phase and enforces orthogonal updates for new tasks, providing a mathematically grounded "safety guarantee" for established knowledge structures. Empirical results across a diverse four-task sequence demonstrate that OSW uniquely succeeds in preserving fragile coding abilities where Replay fails, while simultaneously maintaining high plasticity for novel tasks. Our findings emphasize the necessity of evaluating structural safety alongside average retention in LLM continual learning.

cs.LG↗

arXiv2025

Non-unique decompositions of mixed states and deterministic energy transfers

We investigate the impact of non-unique decompositions of mixed states on energy transfer. Mixed states generally have non-unique decompositions into pure states in quantum theory and, by definition, in other non-classical probabilistic theories. We consider energy transfers constituting deterministic energy harvesting, wherein the source transfers energy to the harvester but not entropy. We use the possibility of non-unique decompositions to derive that if source states in a set jointly lead to deterministic energy harvesting for the given harvesting system and interaction, then that set can be expanded to include both mixtures and superpositions of the original states in the set. As a paradigmatic example, we model the source as an EM mode transferring energy to a 2-level system harvester via the Jaynes-Cummings model. We show that the set of coherent EM mode states with fixed $|\alpha|$ that jointly achieve deterministic energy transfer can be expanded to include all mixtures and superpositions of those states. More generally, the results link the defining feature of a non-classical probability theory with the ability to achieve energy transfer without entropy transfer.

quant-ph↗

arXiv2025

One Polyp Identifies All: One-Shot Polyp Segmentation with SAM via Cascaded Priors and Iterative Prompt Evolution

Polyp segmentation is vital for early colorectal cancer detection, yet traditional fully supervised methods struggle with morphological variability and domain shifts, requiring frequent retraining. Additionally, reliance on large-scale annotations is a major bottleneck due to the time-consuming and error-prone nature of polyp boundary labeling. Recently, vision foundation models like Segment Anything Model (SAM) have demonstrated strong generalizability and fine-grained boundary detection with sparse prompts, effectively addressing key polyp segmentation challenges. However, SAM's prompt-dependent nature limits automation in medical applications, since manually inputting prompts for each image is labor-intensive and time-consuming. We propose OP-SAM, a One-shot Polyp segmentation framework based on SAM that automatically generates prompts from a single annotated image, ensuring accurate and generalizable segmentation without additional annotation burdens. Our method introduces Correlation-based Prior Generation (CPG) for semantic label transfer and Scale-cascaded Prior Fusion (SPF) to adapt to polyp size variations as well as filter out noisy transfers. Instead of dumping all prompts at once, we devise Euclidean Prompt Evolution (EPE) for iterative prompt refinement, progressively enhancing segmentation quality. Extensive evaluations across five datasets validate OP-SAM's effectiveness. Notably, on Kvasir, it achieves 76.93% IoU, surpassing the state-of-the-art by 11.44%.

cs.CV↗

arXiv2025

High-precision laser spectrum analyzer via digital decoherence

With the continuous advancement of laser technology, accurately evaluating the noise spectrum of high-performance lasers has become increasingly challenging. In this work, we demonstrate a high-precision laser spectrum analyzer based on the proposed digital decoherence method, which can precisely measure the frequency noise spectrum of sub-Hz linewidth lasers. In addition, it has broad wavelength compatibility, which enables convenient switching between lasers with different center wavelengths. Its performance is validated through measurements of ultra-stable lasers. Based on the measured frequency noise power spectral density, a beta-line linewidth is determined to be 570 mHz at 10-second observation time, and the minimum observable linewidth is calculated to be 133 mHz. The system's noise floor is evaluated to be 210 mHz beta-line linewidth at 25-second observation time, and a minimum observable linewidth of 39 mHz.

physics.optics↗

arXiv2025

Practical 1-um GHz fiber comb on silica-based platform

We present a fully stabilized 1-GHz Yb-fiber laser frequency comb built on silica substrates, utilizing "optical cubes" to house all optical components, ensuring long-term stability and practical operation. Both the femtosecond laser and f-to-2f interferometer are constructed to silica bricks, with a compact footprint of 290 mm * 250 mm, and a total weight of 1.8 kg. This system provides a stable repetition rate, offset frequency, and a supercontinuum spanning 460-1560 nm without requiring amplification. The carrier-envelop offset frequency exhibits exceptional in-loop stability, with a fractional frequency instability of 3.07* 10^(-18) at a 1 second averaging time, improving to 2.12*10^(-20) at a 10,000 second, maintaining uninterrupted operation for over 60 hours. This work demonstrates a high-performance GHz fiber-based frequency comb, paving the way for applications beyond laboratory environments, including dual-comb spectroscopy, astronomical spectrograph calibration, and portable optical clocks.

physics.optics↗

arXiv2025

Generalized Cross-Entropy Benchmarking for Random Circuits with Ergodicity

Cross-entropy benchmarking is a central technique used to certify a quantum chip in recent experiments. To better understand its mathematical foundation and develop new benchmarking schemes, we introduce the concept of ergodicity to random circuit sampling and find that the Haar random quantum circuit satisfies an ergodicity condition -- the average of certain types of post-processing function over the output bit strings is close to the average over the unitary ensemble. For noiseless random circuits, we prove that the ergodicity holds for polynomials of degree $t$ with positive coefficients and when the random circuits form a unitary $2t$-design. For strong enough noise, the ergodicity condition is violated. This suggests that ergodicity is a property that can be exploited to certify a quantum chip. We formulate the deviation of ergodicity as a measure for quantum chip benchmarking and show that it can be used to estimate the circuit fidelity for global depolarizing noise and weakly correlated noise. For a quadratic post-processing function, our framework recovers Google's result on estimating the circuit fidelity via linear cross-entropy benchmarking (XEB), and we give a sufficient condition on the noise model characterizing when such estimation is valid. Our results establish an interesting connection between ergodicity and noise in random circuits and provide new insights into designing quantum benchmarking schemes.

quant-ph↗

arXiv2025

A 3-Step Optimization Framework with Hybrid Models for a Humanoid Robot's Jump Motion

High dynamic jump motions are challenging tasks for humanoid robots to achieve environment adaptation and obstacle crossing. The trajectory optimization is a practical method to achieve high-dynamic and explosive jumping. This paper proposes a 3-step trajectory optimization framework for generating a jump motion for a humanoid robot. To improve iteration speed and achieve ideal performance, the framework comprises three sub-optimizations. The first optimization incorporates momentum, inertia, and center of pressure (CoP), treating the robot as a static reaction momentum pendulum (SRMP) model to generate corresponding trajectories. The second optimization maps these trajectories to joint space using effective Quadratic Programming (QP) solvers. Finally, the third optimization generates whole-body joint trajectories utilizing trajectories generated by previous parts. With the combined consideration of momentum and inertia, the robot achieves agile forward jump motions. A simulation and experiments (Fig. \ref{Fig First page fig}) of forward jump with a distance of 1.0 m and 0.5 m height are presented in this paper, validating the applicability of the proposed framework.

cs.RO↗

arXiv2024

Highly coherent two-color laser with stability below 3E-17 at 1 second

Two-color lasers with high coherence are paramount in precision measurement, accurate light-matter interaction, and low-noise photonic microwave generation. However, conventional two-color lasers often suffer from low coherence, particularly when these two colors face large frequency spacings. Here, harnessing the Pound-Drever-Hall technique, we synchronize two lasers to a shared ultra-stable optical reference cavity to break through the thermal noise constraint, achieving a highly coherent two-color laser. With conquering these non-common mode noises, we demonstrate an exceptional fractional frequency instability of 2.7E-17 at 1 second when normalized to the optical frequency. Characterizing coherence across large frequency spacings poses a significant challenge. To tackle this, we employ electro-optical frequency division to transfer the relative stability of a 0.5 THz spacing two-color laser to a 25 GHz microwave signal. As its performance surpasses the sensitivity of the current apparatus, we establish two independent systems for comparative analyses. The resulting 25 GHz signals exhibit exceptional phase noise of -74 dBc/Hz at 1 Hz and -120 dBc/Hz at 100 Hz, demonstrating the two-color laser's performance approaching the quantum noise limit of its synchronization system. It also sets a new record for the two-point frequency division method in photonic microwave generation. Our achievement in highly coherent two-color lasers and low-noise microwave signals will usher in a new era for precision measurements and refine the accuracy of light-matter and microwave-matter interactions to their next decimal place.

physics.optics↗

arXiv2024

Online Time-Informed Kinodynamic Motion Planning of Nonlinear Systems

Sampling-based kinodynamic motion planners (SKMPs) are powerful in finding collision-free trajectories for high-dimensional systems under differential constraints. Time-informed set (TIS) can provide the heuristic search domain to accelerate their convergence to the time-optimal solution. However, existing TIS approximation methods suffer from the curse of dimensionality, computational burden, and limited system applicable scope, e.g., linear and polynomial nonlinear systems. To overcome these problems, we propose a method by leveraging deep learning technology, Koopman operator theory, and random set theory. Specifically, we propose a Deep Invertible Koopman operator with control U model named DIKU to predict states forward and backward over a long horizon by modifying the auxiliary network with an invertible neural network. A sampling-based approach, ASKU, performing reachability analysis for the DIKU is developed to approximate the TIS of nonlinear control systems online. Furthermore, we design an online time-informed SKMP using a direct sampling technique to draw uniform random samples in the TIS. Simulation experiment results demonstrate that our method outperforms other existing works, approximating TIS in near real-time and achieving superior planning performance in several time-optimal kinodynamic motion planning problems.

cs.RO↗

arXiv2024

Quantum harvester enables energy transfer without randomness transfer or dissipation

We consider a foundational question in energy harvesting: given a partly random energy source, is it possible to extract the energy without also transferring randomness or accepting another thermodynamical cost? We answer this in the positive, describing scenarios and protocols where in principle energy is extracted from a field with randomness but without any randomness being transferred, and without energy dissipation. Such protocols fundamentally outperform existing methods of rectification which dissipate power, or feedback demon-like protocols which transfer randomness to the feedback system. The protocols exploit the possibility of the harvesting system taking several trajectories that lead to the same final state at a given time. We explain why these protocols do not violate basic physical principles. A key example involves the experimentally well-established phenomenon of Rabi oscillations between energy levels, exploiting the multitude of rotation axes in the state space that take the lower energy state to the excited state. The quantum system is deterministically excited to the highest energy level after interacting with the source for a fixed amount of time, irrespective of the random initial phase of the external potential.

quant-ph↗

arXiv2024

Symmetry-guided gradient descent for quantum neural networks

Many supervised learning tasks have intrinsic symmetries, such as translational and rotational symmetry in image classifications. These symmetries can be exploited to enhance performance. We formulate the symmetry constraints into a concise mathematical form. We design two ways to adopt the constraints into the cost function, thereby shaping the cost landscape in favour of parameter choices which respect the given symmetry. Unlike methods that alter the neural network circuit ansatz to impose symmetry, our method only changes the classical post-processing of gradient descent, which is simpler to implement. We call the method symmetry-guided gradient descent (SGGD). We illustrate SGGD in entanglement classification of Werner states and in a binary classification task in a 2-D feature space. In both cases, the results show that SGGD can accelerate the training, improve the generalization ability, and remove vanishing gradients, especially when the training data is biased.

quant-ph↗