arXiv subjects

Chen Zhang

Publications and source records attributed to Chen Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Typological Feature Induction as a Benchmark for Inductive Metalinguistic Reasoning

Existing metalinguistic reasoning benchmarks require both inference of rules and their application, making failures hard to diagnose. We introduce typological feature induction (TFI) which targets the former step. Given sentences in an unnamed language paired with translations in a known pivot language, models must predict a typological feature of the unnamed language. From Grambank and multi-parallel corpora, we automatically build evaluation questions over 195 languages. We find that LLMs identify 59--80% of these languages from only 16 sentence pairs; and misaligning the translations barely affects their accuracy, suggesting reliance on prior knowledge of the language. Enciphering text to prevent recognition increases the corpus' utility, but mainly for features where target and pivot languages agree. On languages which models cannot identify, neither open model outperforms a majority-value baseline. Since induction from raw text aligns with field linguists' use-case, this work indicates the models tested are not yet ready for this role. TFI offers a scalable path to track progress in this direction.

cs.CL↗

arXiv2026

EMBERS: Empirically Motivated Robustness Testing of Deep Research Systems via Defect-Guided Fuzzing

Deep Research (DR) systems extend large language models into autonomous agents that decompose complex information needs, iteratively search the web, and synthesize evidence to answer challenging questions. However, their multi-stage execution may introduce robustness risks, as variations in user questions or intermediate search inputs may propagate across research stages and lead to incorrect answers. Existing evaluations focus on task-solving capabilities or robustness against external evidence disturbances, leaving intrinsic weaknesses in DR execution insufficiently explored. Thus, we conduct an empirical study of 1,932 execution trajectories from three representative DR systems and identify nine recurring defect categories across subquestion formulation, online search, and final reasoning. Guided by these findings, we propose EMBERS, an empirically motivated, defect-guided fuzzing framework for robustness testing of DR systems. Starting from correctly answered seed questions, EMBERS generates stage-targeted test inputs using defect-guided mutation, web-search-assisted fidelity assurance, and diversity-aware seed maintenance. Experiments across 16 framework--model--dataset settings show that EMBERS achieves 3.21 times the failure-discovery throughput of the strongest baseline under identical testing budgets. Human and LLM-assisted evaluations demonstrate that most generated inputs preserve the original information need and necessary contextual constraints. Moreover, model fine-tuning experiments show consistent-answer performance improvement, highlighting their value for robustness testing and model improvement.

cs.AI↗

arXiv2026

Matching-LLM: Historically Anchored Synthetic Prediction of Public Opinion and Elections

Large language models (LLMs) offer new opportunities for predicting public opinion and electoral behavior, but their synthetic responses may not fully reflect human attitudes and heterogeneity. This study introduces Matching-LLM, a weighted framework that combines LLM-generated responses with historical human behavioral data. Using the World Values Survey, American National Election Studies, and German election survey data, we evaluate its performance in reconstructing survey responses and predicting election outcomes. Results show that Matching-LLM generally improves accuracy relative to LLM-only, including 11 of 12 comparisons on survey questions excluded from calibration, although it does not consistently outperform History-only. In the 2024 U.S. presidential election, Matching-LLM predicts 310 Republican electoral votes, compared with the actual 312, outperforming both single-source benchmarks in electoral-vote accuracy. A supplementary German application extends the framework to a multiparty setting. Overall, the findings highlight the potential of historical data to improve LLM-based predictions while emphasizing the importance of task-specific validation.

econ.GN↗

arXiv2026

EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective

Recent benchmarks for Large Language Model (LLM) agents mainly evaluate reasoning, planning, and execution. However, memory is also essential for agents, as it enables them to store, update, and retrieve information over time. This ability remains under-evaluated, largely because existing benchmarks do not provide a systematic way to assess memory mechanisms. In this paper, we study agent memory from a self-evolving perspective and introduce EvoMemBench, a unified benchmark organized along two axes: memory scope (in-episode vs. cross-episode) and memory content (knowledge-oriented vs. execution-oriented). We compare 15 representative memory methods with strong long-context baselines under a standardized protocol. Results show that current memory systems are still far from a general solution: long-context baselines remain highly competitive, memory helps most when the current context is insufficient or tasks are difficult, and no single memory form works consistently across all settings. Retrieval-based methods remain strong for knowledge-intensive settings, whereas procedural and long-term memory methods are more effective for execution-oriented tasks when their stored experience matches the task structure. We hope EvoMemBench facilitates future research on more effective memory systems for LLM-based agents. Our dataset and code are available at https://github.com/DSAIL-Memory/EvoMemBench.

cs.CL↗

arXiv2026

NeuroBridge: Bio-Inspired Self-Supervised EEG-to-Image Decoding via Cognitive Priors and Bidirectional Semantic Alignment

Visual neural decoding seeks to reconstruct or infer perceived visual stimuli from brain activity patterns, providing critical insights into human cognition and enabling transformative applications in brain-computer interfaces and artificial intelligence. Current approaches, however, remain constrained by the scarcity of high-quality stimulus-brain response pairs and the inherent semantic mismatch between neural representations and visual content. Inspired by perceptual variability and co-adaptive strategy of the biological systems, we propose a novel self-supervised architecture, named NeuroBridge, which integrates Cognitive Prior Augmentation (CPA) with Shared Semantic Projector (SSP) to promote effective cross-modality alignment. Specifically, CPA simulates perceptual variability by applying asymmetric, modality-specific transformations to both EEG signals and images, enhancing semantic diversity. Unlike previous approaches, SSP establishes a bidirectional alignment process through a co-adaptive strategy, which mutually aligns features from two modalities into a shared semantic space for effective cross-modal learning. NeuroBridge surpasses previous state-of-the-art methods under both intra-subject and inter-subject settings. In the intra-subject scenario, it achieves the improvements of 12.3% in top-1 accuracy and 10.2% in top-5 accuracy, reaching 63.2% and 89.9% respectively on a 200-way zero-shot retrieval task. Extensive experiments demonstrate the effectiveness, robustness, and scalability of the proposed framework for neural visual decoding.

cs.CV↗

arXiv2026

Embodied Learning of Reward for Musculoskeletal Control with Vision Language Models

Discovering effective reward functions remains a fundamental challenge in motor control of high-dimensional musculoskeletal systems. While humans can describe movement goals explicitly such as "walking forward with an upright posture," the underlying control strategies that realize these goals are largely implicit, making it difficult to directly design rewards from high-level goals and natural language descriptions. We introduce Motion from Vision-Language Representation (MoVLR), a framework that leverages vision-language models (VLMs) to bridge the gap between goal specification and movement control. Rather than relying on handcrafted rewards, MoVLR iteratively explores the reward space through iterative interaction between control optimization and VLM feedback, aligning control policies with physically coordinated behaviors. Our approach transforms language and vision-based assessments into structured guidance for embodied learning, enabling the discovery and refinement of reward functions for high-dimensional musculoskeletal locomotion and manipulation. These results suggest that VLMs can effectively ground abstract motion descriptions in the implicit principles governing physiological motor control.

cs.RO↗

arXiv2026

VETime: Vision Enhanced Zero-Shot Time Series Anomaly Detection

Time-series anomaly detection (TSAD) requires identifying both immediate Point Anomalies and long-range Context Anomalies. However, existing foundation models face a fundamental trade-off: 1D temporal models provide fine-grained pointwise localization but lack a global contextual perspective, while 2D vision-based models capture global patterns but suffer from information bottlenecks due to a lack of temporal alignment and coarse-grained pointwise detection. To resolve this dilemma, we propose VETime, the first TSAD framework that unifies temporal and visual modalities through fine-grained visual-temporal alignment and dynamic fusion. VETime introduces a Reversible Image Conversion and a Patch-Level Temporal Alignment module to establish a shared visual-temporal timeline, preserving discriminative details while maintaining temporal sensitivity. Furthermore, we design an Anomaly Window Contrastive Learning mechanism and a Task-Adaptive Multi-Modal Fusion to adaptively integrate the complementary perceptual strengths of both modalities. Extensive experiments demonstrate that VETime significantly outperforms state-of-the-art models in zero-shot scenarios, achieving superior localization precision with lower computational overhead than current vision-based approaches. Code available at: https://github.com/yyyangcoder/VETime .

cs.CV↗

arXiv2026

Sandwich Reasoning: An Answer-Reasoning-Answer Approach for Low-Latency Query Correction

Query correction is a critical entry point in modern search pipelines, demanding high accuracy strictly within real-time latency constraints. Chain-of-Thought (CoT) reasoning improves accuracy but incurs prohibitive latency for real-time query correction. A potential solution is to output an answer before reasoning to reduce latency; however, under autoregressive decoding, the early answer is independent of subsequent reasoning, preventing the model from leveraging its reasoning capability to improve accuracy. To address this issue, we propose Sandwich Reasoning (SandwichR), a novel approach that explicitly aligns a fast initial answer with post-hoc reasoning, enabling low-latency query correction without sacrificing reasoning-aware accuracy. SandwichR follows an Answer-Reasoning-Answer paradigm, producing an initial correction, an explicit reasoning process, and a final refined correction. To align the initial answer with post-reasoning insights, we design a consistency-aware reinforcement learning (RL) strategy: a dedicated consistency reward enforces alignment between the initial and final corrections, while margin-based rejection sampling prioritizes borderline samples where reasoning drives the most impactful corrective gains. Additionally, we construct a high-quality query correction dataset, addressing the lack of specialized benchmarks for complex query correction. Experimental results demonstrate that SandwichR achieves SOTA accuracy comparable to standard CoT while delivering a 40-70% latency reduction, resolving the latency-accuracy trade-off in online search.

cs.AI↗

arXiv2026

SpatialCLI: Learning to Reason With Spatial Tools, Then Without Them

Vision-language models (VLMs) are increasingly used in embodied agents to interpret visual inputs, reason about spatial relationships, and make task-level decisions based on that reasoning. However, a fundamental capability mismatch remains: general VLMs can reason about the overall task but often miss the visual details that determine success, while specialist vision models can capture those details but cannot translate them into task-level decisions. In this work, we propose SpatialCLI, a framework that teaches VLMs to reason with spatial tools and progressively internalize the specialist perceptual capabilities they provide. SpatialCLI proceeds in three stages: (1) Call exposes specialist vision models as spatial tools to augment the VLM's perception; (2) Learn uses Cold-Start SFT and agentic RL to improve tool use; and (3) Internalize verbalizes successful tool-use trajectories to internalize specialist perceptual capabilities. We further introduce SpatialCLI-Bench, a 516-example benchmark for compositional perception across localization, segmentation, depth, and pose. On MindCube, SpatialCLI raises Qwen3-VL-8B-Instruct from 29.3% to 84.6% with tools, surpassing GPT-5.6 Sol with tools (72.1%), while retaining 73.8% without tools after internalization.

cs.AI↗

arXiv2026

RSIV-Harness: A Recursive Self-Improvement Harness Framework for Open-Domain Conversational Text-to-Visualization

Users often request charts without knowing which database or table holds the relevant data. Such requests may be ambiguous, and users may ask follow-up questions to deepen their analysis or iteratively refine existing charts. A useful assistant must therefore discover relevant data sources among heterogeneous databases, ask clarifying questions when needed, and track evolving user intent while preserving constraints established in earlier turns. Yet recent text-to-visualization agents typically rely on static agent harnesses that are predefined and remain unchanged despite recurring failures across tasks. To address this limitation, we introduce RSIV-Harness, which keeps the language model weights fixed and applies Recursive Self-Improvement to revise the surrounding harness from the agent's own failure traces. To balance flexible self-improvement with stability, we introduce a four-layer framework that evaluates proposed harness revisions through anchor and memory checks, refining those that pass and restoring the best validated harness otherwise. We further introduce ODAC-Vis, a benchmark with 640 tasks for Open-Domain, Ambiguous, and Conversational text-to-visualization that jointly evaluates open-domain data discovery, ambiguity resolution, multi-turn visualization refinement, and executable chart generation. Experiments show that a single round of RSIV-Harness improves all six evaluation metrics relative to the seed harness, with particularly large gains in executability (+28.02 points) and chart structure (+34.01 points), demonstrating the ability to learn from task-solving experience and turn recurring failures across tasks into substantial performance improvements.

cs.AI↗

arXiv2026

Kagome goldene with flat bands and Dirac nodal line fermions via line-graph epitaxy

The kagome lattice has emerged as a promising platform for investigating exotic quantum phases. However, achieving a single-atomic-layer kagome lattice in elemental materials remains a significant challenge. Here, we introduce line-graph epitaxy, a novel approach that enables the atomic-scale synthesis of goldene, a monolayer of elemental gold atoms arranged in a kagome lattice. Through scanning tunneling microscopy/spectroscopy (STM/STS), and density functional theory (DFT) calculations, we demonstrate the formation of kagome goldene, featuring a flat band with a van Hove singularity approximately 1.1 eV below the Fermi level, signaling strong electron correlation effects. Notably, the flat band is disrupted at the zigzag edges of goldene nanoflakes, revealing substantial edge effects. Furthermore, our calculations show that weak interlayer interactions between goldene and the underlying Au2Ge substrate generate dual Dirac nodal lines through a proximity effect. These findings offer not only a novel strategy for constructing elemental kagome lattices, but also a generalizable framework for fabricating and controlling line-graph materials. This research advances the exploration of quantum phases driven by strong correlations and the design of materials for next-generation quantum technologies.

cond-mat.mes-hall↗

arXiv2026

When Agents Fail: Model or Harness--Who's to Blame?

LLM agents combine models with execution harnesses, so agentic benchmark performance reflects their joint behavior. We argue that harnesses should therefore be evaluated through their interactions with underlying models. We develop a framework combining five complementary measures: (i) Valid Completion Rate (VCR), how reliably agents finish normally and deliver evaluable results, irrespective of task correctness; (ii) Model-Harness Fit, how well a harness supports task achievement through valid completion for a given model; (iii) Cross-Model Fit, a harness's overall suitability across evaluated models; (iv) Interaction Residuals, pairing-specific execution strengths and weaknesses hidden by component-level summaries; and (v) Token Load, the token demands of agent execution, including failed attempts. We evaluate four general-purpose harnesses (OpenCode, Codex CLI, Pi CLI, and DeepAgents) with four core models, plus two fixed native model-harness systems, on six benchmarks covering context management, tool use, long-context planning, multimodal interaction, and artifact handoff. Including native references and stress conditions, 111 evaluation settings yield 52,168 trajectories. Harness rankings vary across models and benchmarks; task performance and valid completion can favor different pairings, and similar task scores can accompany distinct execution profiles. OpenCode ranks first in cross-model performance, while DeepSeek V4 Flash Preview maintains consistently high Mean VCR across harnesses on the shared four-benchmark panel. These findings support evaluating model-harness interactions, training transferable harness-use capabilities in models, and harness designs adapted to the model and task.

cs.AI↗

arXiv2026

Exact Ehrhart Series of Birkhoff Polytopes via Constant Terms and Finite-Field Evaluation

The Ehrhart series of the $n$th Birkhoff polytope is $\sum_{r\geq0}H_n(r)z^r$, where $H_n(r)$ counts nonnegative integer $n\times n$ matrices whose row and column sums all equal $r$. We present an exact method for computing this series using constant terms and finite fields. A root filter expresses $H_n(r)$ as a weighted sum of the values $h_r(M)^n$, where $h_r$ is the complete homogeneous symmetric polynomial and $M$ ranges over multisets of $m$th roots of unity with $m=r+1$. Constant-term cancellation reduces the evaluation of $h_r(M)$ to a sum over repeated elements $a$ of $M$. For a particular $a$ of multiplicity $μ_a$, the computation uses a generalized Todd coefficient of degree $μ_a-2$. Sums of $h_r(M)^n$ over selected multiplicity classes are handled using symmetric function techniques. Together with the remaining individual evaluations, this gives $O_n(m^{n-5}+m^4)$ field operations for each admissible prime and fixed $n\geq5$. An explicit bound and the Chinese remainder theorem recover the integer counts, and Ehrhart symmetry determines the full series. The same method applies to the World Cup problem, which counts the same matrices with diagonal entries required to be $0$. We prove correctness and compute complete series for both families through order $13$. The Birkhoff series for orders $10$--$13$ and the World Cup series for orders $9$--$13$ are tabulated in the appendices.

math.CO↗

arXiv2026

Probing Minihalos in the Late Dark Ages through Velocity Acoustic Oscillations in the 21-cm Power Spectrum

In the Dark Ages, minihalos above the Jeans mass accrete gas from the intergalactic medium (IGM). The accreted gas would be heated to virial temperature and produces 21 cm signal in emission, while the majority of gas retained in the IGM is cold and has 21 cm signal in absorption. The 21 cm signal of the IGM generally dominates over that of minihalos, making it difficult to extract the contribution of minihalos from the detected 21 cm signal. However, in the late Dark Ages, in regions where the gas has not yet been influenced by the radiation from Pop III stars, the 21 cm signal from the IGM is weak because collisional coupling has become less efficient. More importantly, the relic supersonic dark matter-baryon relative streaming velocities suppress the gas accretion of smaller minihalos and encode their large-scale coherent features into the 21 cm signal, generating the velocity acoustic oscillation (VAO) features on the 21 cm power spectrum. This offers a promising avenue for detecting minihalos. We found that, at $z\sim 15$ ($\sim 20$), the first peak of the net VAO wiggle at $k=0.05$ Mpc$^{-1}$ is $\sim 1.4\times 10^{-3}$ ($\sim 3.1\times 10^{-4}$) mK$^2$. With a total of 24 lunar nights/8064 hr (60 lunar nights/20160 hr) of integration, the signal from $z\sim 15$ ($\sim 20$) is detectable for an array composed of 82944 (278784) dipoles, spread over $\sim 3.5$ ($\sim 6.3$) km. If the small-scale matter power spectrum is enhanced and minihalos are much more numerous than in the standard $Λ$CDM model, the signal is detectable for an even smaller scale array.

astro-ph.CO↗

arXiv2026

Sage-A1.5: From Long-Horizon Execution to Scientific Exploration with Capability Composition

In this report, we introduce Sage-A1.5, a 35B Mixture-of-Experts agentic model that builds on A1 to advance long-horizon problem solving through adaptive capability composition. Compared with A1, Sage-A1.5 learns not only to solve individual tasks, but also to organize work and revise its approach during execution. Sage-A1.5 highlights three improvements: agentic orchestration, verification, and open scientific discovery. First, agentic orchestration extends single-agent execution through task delegation and parallel collaboration among subagents with isolated contexts, enabling complex tasks to be decomposed and addressed concurrently and supporting efficient execution while reducing interference across concurrent subagents. Second, agentic verification strengthens long-horizon reliability through progress checking, evidence validation, and reflective strategy revision, allowing the agent to recognize unproductive search paths and pursue alternative approaches. Third, open scientific discovery extends agentic capabilities beyond answering predefined questions to conducting exploratory research, connecting scientific inquiry with interactive analysis and inspectable research artifacts. We showcase these capabilities through complicated agentic tasks and multidisciplinary scientific applications. Besides, we hope Sage-A1.5 provides the community with an accessible foundation for open scientific discovery, offering a practical path from long-horizon task execution to collaborative, verifiable, and inspectable scientific exploration.

cs.AI↗

arXiv2026

Arithmetic purity of strong approximation for complete intersections of large dimension

In this paper, we prove that for a smooth complete intersection $X\subset \mathbb{P}^n$ of type $(d_1,\dots,d_c)$ over a number field $k$, there exists an effective bound $N_{\mathrm{AP}}(d_1,\dots,d_c)$ such that if $n\ge N_{\mathrm{AP}}(d_1,\dots,d_c)$, then $X$ satisfies arithmetic purity of strong approximation. We establish a connecting lemma that promotes weak approximation to strong approximation under a suitable connectivity hypothesis via rational curves over $k$. We construct the required connecting conics by combining the deformation theory of rational curves with results from the circle method.

math.NT↗

arXiv2026

MemForest: Efficient Agent Memory Management via EventTree Partitioning and Progressive Merging

Agent memory systems have demonstrated significant potential in long-term dialogue, personalized assistants, and video understanding. However, continuously accumulated memory introduces substantial storage and retrieval costs during inference. To address this issue, we propose \textbf{MemForest}, a general memory compression framework adaptable to various agent memory systems. Specifically, MemForest partitions historical memory into event-centric units by leveraging global semantic similarity and local temporal continuity. For each unit, it constructs a maximum spanning tree, termed an EventTree, and progressively merges redundant memory nodes by selecting high-weight edges, reducing storage overhead. Furthermore, we introduce an anchor-guided propagation retrieval mechanism that retrieves relevant memory nodes from the temporal neighborhoods of key nodes, improving retrieval accuracy. Extensive experiments demonstrate the effectiveness of MemForest. Under the unimodal Mem0 framework, MemForest retains \textbf{97.1%} of the original performance while compressing \textbf{50%} of historical memory across three benchmarks (LoCoMo, LongMemEval, and PersonaMem), achieving a \textbf{1.89x} retrieval speedup. Under the multimodal M3-Agent framework, it preserves \textbf{99.7%} of the original performance with a \textbf{50%} compression ratio across two benchmarks (M3-Bench-robot and M3-Bench-web), achieving a \textbf{2.24x} retrieval speedup. \textcolor{RoyalBlue}{\textit{Our code is available at [https://github.com/Celina-love-sweet/MemForest.}}](https://github.com/Celina-love-sweet/MemForest.}})

cs.AI↗

arXiv2026

Counterexamples to the Minimum Period Conjecture for Restricted Partition Functions

For a finite sequence of positive integers $\boldsymbol{a}=(a_1,\dots,a_n)$, the restricted partition function $q_{\boldsymbol{a}}(k)$ denote the number of nonnegative integer solutions to the equation $a_1x_1+a_2x_2+\cdots +a_nx_n=k$. It is proved to be a quasi-polynomial of degree $n-1$. Write $q_{\boldsymbol{a}}(k)=\sum_{j=0}^{n-1}c_j(k)k^j$ with periodic coefficient functions $c_j$, and set $b_m=\#\{i:m\mid a_i\}$. In 2008, Beck, Sam, and Woods conjectured that the minimum period of $c_j(k)$ is $\mathrm{lcm}\{m:b_m>j\}$. In this paper, we derive an exact root-of-unity formula for every coefficient function $c_j(k)$. The formula proves the conjectured divisibility upper bound, but it also reveals a lower bound for the period of $c_j(k)$. Both divisibility bounds are sharp. This leads us to construct a family of counterexamples to this conjecture.

math.CO↗