arXiv subjects

Xinyang Liu

Publications and source records attributed to Xinyang Liu.

At least 19 recordsLinked to original sources

arXiv2026

Hidden Triakontadipole Order in the Triangular-Lattice Magnet CeMgAl$_{11}$O$_{19}$

The absence of magnetic Bragg peaks and a broad excitation continuum are commonly recognized experimental features of a quantum spin liquid. The Ce-based triangular-lattice hexaaluminate CeMgAl$_{11}$O$_{19}$ exhibits both signatures, yet its effective model and ground state remain under debate. Experiments have variously suggested a quantum spin liquid, proximate umbrella point, and disorder-suppressed magnetic order. Here we trace these contradictions to an overlooked multipolar degree of freedom: the transverse pseudospin of the $|\pm5/2\rangle$ Kramers doublet is a rank-5 triakontadipole rather than an ordinary dipole. Analyzing the magnetization, neutron spectra, and thermodynamics on equal footing, we establish a dipole-triakontadipole XXZ model with $J_{\perp} \simeq 0.65$ K and $J_c \simeq -0.13$ K. The resulting planar triakontadipolar superfluid order is hidden from dipolar probes, explaining the absence of a magnetic Bragg peak in neutron scattering. The zero-field excitations separate into a low-energy triakontadipole wave and a broad higher-energy continuum. We further propose nuclear spin-lattice relaxation and spin-current experiments to probe the gapless excitations of this hidden triakontadipole superfluid. Our findings thus open a new avenue for high-rank multipolar physics in rare-earth triangular-lattice magnets.

cond-mat.str-el↗

arXiv2026

Contrasting Crystal-Field and Dipolar Anisotropies in the Triangular Antiferromagnet NaBaEr(BO$_3$)$_2$

The magnetic ground states of rare-earth frustrated magnets can be governed by competing energy scales arising from crystal-field anisotropy, exchange interactions, and long-range dipolar interactions. Here we investigate this interplay in the triangular-lattice antiferromagnet NaBaEr(BO$_3$)$_2$ by combining magnetization, specific heat, and neutron scattering measurements. Neutron powder diffraction resolves rotations of the ErO$_6$ octahedra and supports the lower-symmetry $R\bar{3}$ structure over the previously proposed $R\bar{3}m$ model. Magnetization measurements and inelastic neutron scattering establish an effective spin-1/2 description of its low-temperature magnetism with pronounced easy-axis anisotropy, arising from a well-isolated Kramers ground-state doublet separated from the first excited crystal-electric-field level by approximately 4 meV. Below $T_{\rm N}=0.1$ K, single-crystal neutron diffraction reveals A-type antiferromagnetic order with the ordered moment predominantly oriented within the $ab$ plane, despite the pronounced paramagnetic easy-axis anisotropy. The estimated dipolar interaction scale is comparable to $T_{\rm N}$, and classical dipolar-energy calculations for the observed stacking pattern favor an in-plane moment orientation. These results reveal contrasting crystal-field and dipolar anisotropies in determining the magnetic ground state of NaBaEr(BO$_3$)$_2$.

cond-mat.str-el↗

arXiv2026

MGE-PO: Entropy-Aware Credit Assignment for Mathematical Reasoning

GRPO assigns the same group-relative advantage to equally rewarded responses within a question, leaving their generation uncertainty unused. We introduce MGE-PO, centered on four-quadrant (4Q) advantage shaping that combines verified correctness with rollout generation entropy. Correctness determines the sign of an additional correction, while entropy controls its magnitude: lower-entropy successes receive larger bonuses and higher-entropy failures receive larger penalties. An online entropy-based question curriculum supports this credit rule using cached rollout measurements. At the final checkpoint after eight training epochs on Qwen3-8B-Base, adding 4Q across three policy optimization settings yields absolute gains of 3.23-3.69% in six-benchmark macro avg@32 and 4.24-11.06% in macro pass@32. On Qwen2.5-7B-Base, the complete configuration raises four-benchmark macro avg@32 from 41.5% to 45.4%.

cs.CL↗

arXiv2026

System1-Arena: Which Decisions Can Be Left to System 1?

Large language models (LLMs) often make discrete decisions, such as flagging phishing or escalating to a human. System-1 models, such as Jev and the OpenAI Decisions API, return probabilities over options in a single forward pass, while System-2 approaches generate reasoning steps. Which decisions can System-1 models handle, and which require generative reasoning? We introduce System1-Arena, a benchmark that gives all models the same decision requests and scores their probabilities. It contains 92k questions across 105 tasks from 268 public sources, in four groups: text classification and knowledge, applied decisions, evaluation and control, and agent and action. It also includes 138 closed-loop environments, from web shopping and SWE-bench coding control to robotics, games, and driving; in real-time mode, the world evolves while a model decides. We evaluate 30 models: 7 closed LLMs, 13 open LLMs read at the answer token, and 10 System-1 models. The largest System-1 models match or exceed fast closed LLMs in non-agentic accuracy, with better-calibrated probabilities at one-quarter of the latency. A reasoning LLM is the most accurate, at about 70x the cost of a fast LLM. In closed loops, large open LLMs succeed most often (up to 32.9%); small System-1 models rarely succeed.

cs.CL↗

arXiv2026

GDEPO: Group Dual-dynamic and Equal-right Advantage Policy Optimization for Sample-Efficient Reinforcement Learning

Automated Theorem Proving (ATP) is a fundamental AI challenge, requiring machine-verifiable proofs in formal languages like Lean to evaluate reasoning. Reinforcement Learning (RL), especially Group Relative Policy Optimization (GRPO), is a mainstream approach. However, GRPO faces two critical issues in ATP: its relative advantage estimation may conflict with binary verifier feedback under composite rewards, and its static sampling strategy discards batches without valid proofs, causing data waste. Repeated training on mastered problems also wastes computation. To address this, we propose Group Dual-dynamic and Equal-right Advantage Policy Optimization (GDEPO), incorporating three core mechanisms: 1) dynamic additional sampling, resampling ineffective batches until a valid proof is found; 2) equal-right advantage, decoupling the advantage function's sign (correctness) from its magnitude (auxiliary rewards) to ensure stable policy updates while increasing effective computation; and 3) dynamic additional iterations, applying extra gradient steps to initially failed but eventually successful samples to accelerate learning. Experiments on MinF2F-test, MathOlympiadBench, and PutnamBench confirm GDEPO's effectiveness, with ablation studies validating its synergistic components. The method enhances data utilization and optimization efficiency, offering a novel training paradigm for ATP.

cs.AI↗

arXiv2026

Disclosure-Gated User Simulation for Companion-Agent Evaluation

Playing the user with a large language model is standard in scalable evaluation, and has a repeatedly diagnosed failure: the simulated user is excessively cooperative, so a system under test can score by the number of questions it asks rather than by making the user willing to speak. We answer with a disclosure gate conditioning information release on the companion agent's behaviour: its state is a ladder of five ordered gates, merged onto three observable depth layers. We specify, ablate, and audit it. The gate is a load-bearing component of the environment: on a published companion-agent benchmark's English corpus, withdrawing per-example gate statements from training displaces the ranking of the 12 systems beyond the noise band of a seed re-run, while per-system scores show no detectable change. We state two acceptance criteria: a ranking must be order-preserving, and absolute scores must be scale-stable. Only one candidate passes both -- the simulator we release -- correlating at 0.993 with the published leaderboard. Prompting a frontier model instead barely moves the ranking while shifting every score up, invisible to ranking-only checks; we supply the specification, ablations, and downstream sensitivity analysis the benchmark's environment lacked.

cs.CL↗

arXiv2026

LocalTrain: Integer-Only Local Learning for Energy-Efficient CNN Training on Edge FPGAs

Local learning reduces cross-layer training dependencies, but an efficient accelerator must still retain operands, accumulate gradients, and select quantization scales. We present an FPGA system that trains a CNN and its regional auxiliary classifiers using a shared integer matrix engine. Compact errors are expanded on chip, gradients are folded tile by tile, and historical scales permit direct output conversion. At 200 MHz, the KV260 prototype achieves 76.57% accuracy on CIFAR-10 and 91.85% on SVHN at approximately 4.20 W mean module power. Effective matrix throughput is 31.22 and 31.57 GMAC/s, respectively. Against the same integer method on Jetson Orin Nano, measured training energy is 40.4% and 42.1% lower, with $1.86\times$ and $1.82\times$ longer runtime; measurements use the respective platform power domains. Selected state checks and matching complete-run prediction counts support functional agreement with the software reference.

cs.AR↗

arXiv2026

FFR: Forward-Forward Learning for Regression

The Forward-Forward (FF) algorithm offers a computationally efficient and biologically plausible alternative to backpropagation (BP) by training neural networks through purely local, layer-wise optimization. However, FF is inherently designed for classification via contrastive positive-negative sample pairs, and extending it to regression poses fundamental challenges: continuous target space lacks natural "opposites" for contrastive learning, and the standard goodness function carries no information about target magnitude or ordering. We propose FFR (Forward-Forward for Regression), to our knowledge, the first framework to extend FF to real-world regression and demonstrate competitive performance across diverse realworld datasets. FFR introduces three key innovations: (1) an ordinal competitive goodness function that replaces contrastive pairs with competitive learning between partitioned neuron groups under distance-aware ordinal supervision; (2) a stratified ladder architecture where shallow layers learn coarse ordinal discrimination and deeper layers refine into fine-grained regression, with multi-scale feature aggregation for inter-layer collaboration; and (3) hierarchical prediction with uncertainty estimation, where multi-scale predictors jointly provide robust predictions and a single-pass uncertainty score. Extensive experimental results show FFR recovers on average 98.5% of BP's accuracy across six real-world regression benchmarks while reducing peak training memory to only 27% of BP's at depth 8 and 8% at depth 32, with per-iteration time around 72% of BP's, and substantially outperforms all BP-free competitors.

cs.LG↗

arXiv2026

Do Uncertainty Signals Help? A Systematic Study of Uncertainty-Aware Decoding with Rollback Mechanisms

Prediction uncertainty is a widely adopted metric for quantifying model confidence, with downstream applications spanning model explanation, data selection, and prediction rollback. Despite its demonstrated utility, the potential of uncertainty quantification to enhance code generation in large language models (LLMs) remains largely underexplored, raising a critical question: to what extent can uncertainty serve as an effective signal for improving LLM-based code generation? To answer this question, we study uncertainty-aware rollback decoding, an inference-time strategy that uses uncertainty signals to identify unreliable generation regions and roll back to earlier valid prefixes without retraining the model. We evaluate this framework on seven code LLMs, five code generation benchmarks, and eight token-level uncertainty signals under a unified decoding setup. Our results show that the complete rollback framework improves over equal-budget restart across the evaluated benchmarks and model settings, with gains of up to 0.26 in pass@1 and 0.35 in AvgTestPassRate on functional code generation benchmarks, and an absolute improvement of up to 6.4\% in Patch-Aligned Safe Rate on Dsec-Python. Among the evaluated signals, information-theoretic measures such as token entropy and negative log-likelihood show the most favorable overall trend, frequently achieving the best or near-best results on standard benchmarks. A component-controlled ablation further shows that feedback-guided rollback provides the main improvement, while uncertainty localization provides an additional gain when checking, budget, rollback, and branch decay are held fixed.

cs.LG↗

arXiv2026

Overlooked weak structural connections support human cognition under nonlinear connectome scaling

Human cognition depends on large scale communication constrained by white matter architecture. Although weak connections are abundant in mammalian connectomes, they have long been treated as noise and downweighted because of tractography uncertainty in the human brain, and their relevance to human cognition and large scale functional organization remains unresolved. Across multiple datasets and tractography pipelines, we show that, when tractography derived connectivity weights are interpreted through a nonlinear weighting framework, weak connections make measurable contributions to cognitive prediction, functional connectivity simulation, and structure-function coupling. These effects are selective: nonlinear weighting improves the prediction of general cognitive ability and memory more than that of crystallized intelligence or processing speed, consistent with the notion that weak connections preferentially expand the modal repertoire of brain networks to enhance both large scale integration and fine grained segregation, thereby supporting the functional balance essential for diverse cognitive abilities. Importantly, these effects are replicated in a reliability aware connectome generated by integrating two post tractography filtering methods, in which preserving weak links consistently outperforms conventional thresholding strategies. Finally, we show that weak connections contain functionally informative subsets organized along systems level and transcriptomic gradients. In particular, a specific class of weak connections, predominantly linking visual and motor systems with limbic regions and characterized by negative gene coexpression, exerts a disproportionately large influence on brain function.

q-bio.NC↗

arXiv2026

Ising Supercriticality and Universal Magnetocalorics in Spiral Antiferromagnet Nd$_3$BWO$_9$

The celebrated analogy between the pressure-temperature phase diagram of a liquid-gas system and the field-temperature phase diagram of a ferromagnet has long been a cornerstone for understanding universality of phase transitions and critical phenomena. Here we extend this analogy to a highly frustrated antiferromagnet, the spiral Ising compound Nd$_3$BWO$_9$ with kagome layers. In its phase diagram, we identify a metamagnetic transition line with a critical endpoint (CEP) located at $μ_0H_{\mathrm{c}} \simeq 1.04$ T and $T_{\mathrm{c}} \simeq 0.3$ K. Above the CEP, an Ising supercritical regime emerges with crossover lines that follow a universal scaling law, as evidenced by the specific heat, magnetic susceptibility, and magnetocaloric measurements. Remarkably, we observe highly sensitive magnetic cooling near the emergent CEP, characterized by a divergent magnetic Grüneisen ratio $Γ_H \propto 1/t^{β+γ-1}$, with $β+ γ\simeq 1.563$ the sum of critical exponents of the 3D Ising universality class and $t \equiv (T-T_{\rm c})/T_{\rm c}$ the reduced temperature. Adiabatic demagnetization from 2 K and 4 T reaches a minimum temperature of 195 mK, via a self-cascading process that combines supercritical and topological cooling. Our findings open a new avenue for studying supercritical phenomena and magnetic refrigeration with the frustrated rare-earth compounds RE$_3$BWO$_9$ and, more broadly, in Ising-anisotropic antiferromagnets such as spin ices.

cond-mat.str-el↗

arXiv2026

AutoSOTA: An End-to-End Automated Research System for State-of-the-Art AI Model Discovery

Artificial intelligence research increasingly depends on prolonged cycles of reproduction, debugging, and iterative refinement to achieve State-Of-The-Art (SOTA) performance, creating a growing need for systems that can accelerate the full pipeline of empirical model optimization. In this work, we introduce AutoSOTA, an end-to-end automated research system that advances the latest SOTA models published in top-tier AI papers to reproducible and empirically improved new SOTA models. We formulate this problem through three tightly coupled stages: resource preparation and goal setting; experiment evaluation; and reflection and ideation. To tackle this problem, AutoSOTA adopts a multi-agent architecture with eight specialized agents that collaboratively ground papers to code and dependencies, initialize and repair execution environments, track long-horizon experiments, generate and schedule optimization ideas, and supervise validity to avoid spurious gains. We evaluate AutoSOTA on recent research papers collected from eight top-tier AI conferences under filters for code availability and execution cost. Across these papers, AutoSOTA achieves strong end-to-end performance in both automated replication and subsequent optimization. Specifically, it successfully discovers 105 new SOTA models that surpass the original reported methods, averaging approximately five hours per paper. Case studies spanning LLM, NLP, computer vision, time series, and optimization further show that the system can move beyond routine hyperparameter tuning to identify architectural innovation, algorithmic redesigns, and workflow-level improvements. These results suggest that end-to-end research automation can serve not only as a performance optimizer, but also as a new form of research infrastructure that reduces repetitive experimental burden and helps redirect human attention toward higher-level scientific creativity.

cs.CL↗

arXiv2026

Noise-like pulse laser source with ultrabroadband tunability and coherence-limited sub-structure

High brightness and low coherence laser sources with wideband tunability are essential for many full-field imaging applications aiming for high contrast and speckle free performance. However, this combination of parameters is challenging to achieve. The current solutions focus on decreasing spatial coherence or generation of time-varying speckle patterns, while suppression of temporal coherence typically compromises brightness. Here we demonstrate a wideband pulsed laser source with low temporal coherence and the absence of phase correlation between pulses as an alternative approach with simultaneous time and frequency diversity. The full gain spectrum of a Tm doped fiber laser (1650 nm 2000 nm) is operated in a tunable noise like pulse regime, which by nature is composed of countless structured elementary events with uncorrelated phases randomly varying from bunch to bunch. The measured spectral widths range from 13.8 nm to 18.8 nm, while the average output power varies between 63.3 mW and 213 mW. Numerical simulations reveal that temporal coherence decreases significantly with increasing optical gain, dropping from near unity at low gain to approximately 0.2 at high gain. The startup dynamics of the noise like pulse laser are experimentally studied using the dispersive Fourier transformation (DFT) method. Based on single shot spectra and frequency resolved optical gating traces, the coherence properties of the laser are further analyzed by calculating the mutual coherence function and cross-spectral density. The noise like pulse laser exhibits a coherence time of approximately 100 fs and an average pulse burst duration of about 40 ps in the high-gain regime.

physics.optics↗

arXiv2026

Disentangled Generative Graph Representation Learning

Recently, generative graph models have shown promising results in learning graph representations through self-supervised methods. However, most existing generative graph representation learning (GRL) approaches rely on random masking across the entire graph, which overlooks the entanglement of learned representations. This oversight results in non-robustness and a lack of explainability. Furthermore, disentangling the learned representations remains a significant challenge and has not been sufficiently explored in GRL research. Based on these insights, this paper introduces DiGGR (Disentangled Generative Graph Representation Learning), a self-supervised learning framework. DiGGR aims to learn latent disentangled factors and utilizes them to guide graph mask modeling, thereby enhancing the disentanglement of learned representations and enabling end-to-end joint learning. Extensive experiments on 11 public datasets for two different graph learning tasks demonstrate that DiGGR consistently outperforms many previous self-supervised methods, verifying the effectiveness of the proposed approach.

cs.LG↗

arXiv2026

Damping dynamics of the centroid oscillation of a relativistic laser pulse in a plasma channel

The centroid oscillation of an offset laser pulse propagating in a preformed plasma channel is investigated through theoretical analysis and three-dimensional particle-in-cell simulations. For non-relativistic laser pulses, the mode leakage of a finite channel and the temporal walk-off between the fundamental and high order modes of a finite-duration laser induce a decay in the laser centroid oscillation. An analytical model characterizing these decay mechanisms is derived and validated by simulations. For relativistic laser pulses, the slice-based centroid oscillation frequency develops an axial chirp due to relativistic channel modification and photon deceleration. This chirp leads to phase mixing across different axial slices of the pulse, resulting in a rapid damping of the overall centroid oscillation. Understanding this oscillation damping is crucial for mitigating electron beam pointing jitter and maintaining beam quality in high-energy, channel-guided laser wakefield accelerators.

physics.acc-ph↗

arXiv2026

Route Experts by Sequence, not by Token

Mixture-of-Experts (MoE) architectures scale large language models (LLMs) by activating only a subset of experts per token, but the standard TopK routing assigns the same fixed number of experts to all tokens, ignoring their varying complexity. Prior adaptive routing methods introduce additional modules and hyperparameters, often requiring costly retraining from scratch. We propose Sequence-level TopK (SeqTopK), a minimal modification that shifts the expert budget from the token level to the sequence level. By selecting the top $T \cdot K$ experts across all $T$ tokens, SeqTopK enables end-to-end learned dynamic allocation -- assigning more experts to difficult tokens and fewer to easy ones -- while preserving the same overall budget. SeqTopK requires only a few lines of code, adds less than 1% overhead, and remains fully compatible with pretrained MoE models. Experiments across math, coding, law, and writing show consistent improvements over TopK and prior parameter-free adaptive methods, with gains that become substantially larger under higher sparsity (up to 16.9%). These results highlight SeqTopK as a simple, efficient, and scalable routing strategy, particularly well-suited for the extreme sparsity regimes of next-generation LLMs. Code is available at https://github.com/Y-Research-SBU/SeqTopK.

cs.LG↗

arXiv2026

Ultra-Fast Language Generation via Discrete Diffusion Divergence Instruct

Fast and high-quality language generation is the holy grail that people pursue in the age of AI. In this work, we introduce Discrete Diffusion Divergence Instruct (DiDi-Instruct), a training-based method that initializes from a pre-trained diffusion large language model (dLLM) and distills a few-step student for fast generation. The model distilled with DiDi-Instruct matches or surpasses its dLLM teacher and the GPT-2 baseline while providing up to 64$\times$ acceleration. The theoretical foundation of DiDi-Instruct is a novel framework based on integral KL-divergence minimization, which leads to a practical training algorithm. We further introduce grouped reward normalization, intermediate-state matching, and the reward-guided ancestral sampler to improve training stability, model coverage, and inference quality. On the OpenWebText benchmark, DiDi-Instruct achieves perplexity ranging from 62.2 (8 NFEs) to 18.4 (128 NFEs), outperforming prior accelerated dLLMs and the GPT-2 baseline. These gains incur a negligible entropy loss (around $1$%) and reduce additional training wall-clock time by more than $20\times$ compared to competing dLLM distillation methods. We further validate the robustness and effectiveness of DiDi-Instruct through extensive ablation studies, model scaling, downstream task evaluations, and unconditional protein sequence generation. In conclusion, DiDi-Instruct enables efficient and effective distillation for language generation in the blink of an eye.

cs.CL↗

arXiv2026

Giant Magnetocaloric Effect in a High-Spin Shastry-Sutherland Dipolar Magnet

The Shastry-Sutherland lattice is a prototypical frustrated quantum magnet. It is notable for its exactly solvable dimer-singlet ground state and hosts a wealth of magnetic phenomena under external fields. Here, this work investigates the high-spin (S = 7/2) Eu-based magnet Eu2MgSi2O7 (EMSO) using low-temperature magnetothermal measurements and Monte Carlo simulations, revealing a giant magnetocaloric effect (MCE) in this Shastry-Sutherland compound. The entropy change peak value is found to be 55.0 J kg-1 K-1 under a field change of B = 0-4 T, approximately 1.5 times larger than the commercial Gd3Ga5O12 (GGG). Adiabatic demagnetization refrigeration achieves a lowest temperature of 151 mK, deeply into the sub-Kelvin regime. Furthermore, a distinctive cooling effect persists below about 1 T, a characteristic absent for conventional magnetic coolants. A dipolar Shastry-Sutherland model is introduced as a minimal model to describe this system; in particular, the experimentally revealed 1/3 magnetization pseudo-plateau can be ascribed to the presence of dipolar couplings between Eu2+ ions, further stabilized by the thermal fluctuations, explaining the persistent cooling effect. This work establishes EMSO as a novel platform for exploring the dipolar Shastry-Sutherland system and for sub-Kelvin adiabatic demagnetization refrigeration.

cond-mat.mtrl-sci↗