arXiv subjects

Siyuan Li

Publications and source records attributed to Siyuan Li.

At least 19 recordsLinked to original sources

arXiv2026

ENGINE: Endogenous Variational MultiScale Optimization for Zeroth-Order LLM Fine-Tuning

Zeroth-order (ZO) optimization enables memory-efficient large language model (LLM) fine-tuning, and recent methods incorporate curvature-aware strategies. However, in the high-dimensional parameter space of LLMs, these methods inherently suffer from exploding gradient variance and prohibitive costs. Conversely, confining updates to a low-dimensional subspace reduces variance but discards critical high-dimensional information, degrading performance. This creates a fundamental dilemma: full-space optimization is informative but has high variance, while subspace optimization has low variance but loses information. To break this trade-off, we propose ENGINE (Endogenous Variational MultiScale). Inspired by computational mechanics, we adapt static Variational MultiScale (VMS) theory to dynamic optimization, treating low- and high-dimensional spaces as resolved coarse and unresolved fine scales. Through a multigrid-style V-cycle, ENGINE alternates between spaces, dynamically probing fine-grained information in the high-dimensional space and injecting it to correct the efficient low-dimensional sampling process. Under appropriate assumptions, statistical analysis characterizes variance reduction and correction of curvature-induced bias, and convergence analysis establishes a stationarity bound for the corrected coarse updates. Evaluations on standard LLM benchmarks show that ENGINE achieves higher accuracy and faster convergence than existing ZO methods, achieving up to a $2.61\times$ speed-up in forward passes with zero additional peak memory overhead. Code is available at https://github.com/AI4Science-WestlakeU/ENGINE.

cs.LG↗

arXiv2026

KnowMT-Bench: Benchmarking Knowledge-Intensive Long-Form Question Answering in Multi-Turn Dialogues

Multi-Turn Long-Form Question Answering (MT-LFQA) is a key application paradigm of Large Language Models (LLMs) in knowledge-intensive domains. However, existing benchmarks are limited to single-turn dialogue, while multi-turn dialogue benchmarks typically assess other orthogonal capabilities rather than knowledge-intensive factuality. To bridge this critical gap, we introduce \textbf{KnowMT-Bench}, a new benchmark designed to systematically evaluate MT-LFQA for LLMs across knowledge-intensive fields, including medicine, finance, and law. To faithfully assess the model's real-world performance, KnowMT-Bench employs a dynamic evaluation setting where models generate their own multi-turn dialogue histories given logically progressive question sequences. Our experiments on a diverse suite of LLMs show a clear degradation in both factual capability and information delivery efficiency in multi-turn dialogue. We further probe the underlying causes and investigate mitigation strategies spanning domain-specific finetuning, active context denoising, and RAG. The results show domain-dependent gains from finetuning, while active context denoising and RAG more consistently mitigate these issues, with RAG notably reversing this performance degradation. These findings underscore the importance of our benchmark for evaluating LLMs' conversational factual capabilities in real-world applications. Code and data are publicly available.\footnote{\href{https://github.com/Sven-LI-sankyuu/KnowMT-Bench}{\texttt{https://github.com/Sven-LI-sankyuu/KnowMT-Bench}}}

cs.CL↗

arXiv2026

Bottom-Charm $[bc]$ Diquark Masses from QCD Laplace Sum-Rules

Constituent mass predictions for bottom-charm $[bc]$ color-antitriplet diquarks are generated using QCD Laplace sum rules. We calculate the diquark correlator within the operator product expansion to next-to-leading-order (NLO), including terms proportional to the four- and six-dimensional gluon and six-dimensional quark condensates. Existing analytic expressions for the resulting multi-scale master integrals are insufficient for these calculations, so we utilize numerical $D$-dimensional loop integration techniques. The sum rule analyses stabilize in all four channels yielding constituent masses in GeV of: scalar ($0^+$) $(5.60\pm0.56)$, pseudoscalar ($0^-$) $(6.04\pm0.60)$, axial vector ($1^+$) $(5.60\pm0.56)$ and vector ($1^-$) $(5.99\pm0.60)$. These constituent mass results are discussed in comparison with values used in diquark models of tetraquark hadronic states. We also summarize QCD sum rule predictions for diquark masses for all possible quark-flavor combinations, providing QCD-based criteria for assessing the diquark mass parameters used in constituent diquark models of hadrons.

hep-ph↗

arXiv2026

A Neuro-Symbolic Reasoning Engine for Compilation-Free Patch Presence Testing

Code reuse in modern software ecosystems facilitates the spread of 1-day vulnerabilities, posing significant security risks. Determining if patches are present is crucial, but traditional methods suffer from limited usability and accuracy, often depending on a fragile compilation process. In our prior work, we introduced Lares, the first compilation-free framework for patch presence testing that combines large language models (LLMs) with SMT solvers. However, Lares treats the LLM and the SMT solver as two separate, parallel decision channels: the symbolic channel (Z3) succeeded in only 20% of code cases, and the remaining cases fell back to the neural channel (the LLM) for the final decision. Letting the LLM act as the final arbiter turns verification into an informal, non-deterministic process that is prone to hallucination errors, while the two channels never reinforce each other. To overcome these limitations, we propose SyLares, which reformulates compilation-free patch presence testing as neuro-symbolic reasoning. Instead of letting the LLM emit a final verdict, SyLares uses the LLM as a neural feature extractor that perceives heterogeneous source code and decompiled pseudocode and produces specification-compliant SMT-LIB inputs, and delegates the final, deterministic judgment to the symbolic Z3 solver. This neural-perceives, symbolic-decides coupling is realized by a Code2SMT reasoning engine that combines deterministic syntactic parsing with LLM-driven code normalization and symbol alignment, so that every case is reduced to a formal equivalence query rather than an opaque LLM guess.

cs.SE↗

arXiv2026

Short-Distance-Free Leptonic $g-2$

We note that the hadronic contributions to $a_e$ and $a_μ$ are strongly correlated, which potentially allows one to reduce the uncertainties associated with higher-energy contributions (and address the current discrepancy between lattice and data-driven evaluations) by improved measurements of $a_e$ and the fine-structure constant.

hep-ph↗

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

Witness Overlap: Directional Provenance Inside Open-Weight Model Families

Open-weight models are often released, fine-tuned, aligned, merged, and re-released, making provenance audits ask not only whether checkpoints are related, but also which checkpoint came first. Many existing model-provenance methods are designed for a base-known audit setting: given a victim or source model, they test whether a suspect model is related to it. Although these audits are framed as source-to-suspect tests, their underlying evidence is often symmetric, relying on representation similarity, weight similarity, behavioral fingerprints, or correlation statistics. Symmetric pairwise comparisons can detect relatedness, but they cannot by themselves orient relationship between checkpoints A and B. We therefore introduce a local geometric comparison: instead of comparing two checkpoints directly, we add a third same-family checkpoint as a witness and compare the geometry around each candidate endpoint. Direction is inferred by asking which candidate behaves more like a branching parent. Motivated by this idea, and by the empirically observed asymmetry between parent-anchored and child-anchored witness-overlap distributions, we propose Witness Overlap, a prompt-free, training-free white-box test for directional provenance. On 176 LLM checkpoints from 16 families, our one-witness test orients 95.3\% of parent-child decisions using Frobenius cosine. We further evaluate root identification, sibling discrimination, generalizations to VLM and diffusion families, and chain-structured ordering. The signal is robust to weight noise and sparse pruning, with a proposed SVD weight reduction variant showing greater robustness than Frobenius cosine.

cs.AI↗

arXiv2026

Bounds on hadronic contributions to the anomalous magnetic moment of the $τ$ lepton from QCD Sum Rules

A QCD sum-rule study is performed to place bounds on the leading-order (LO) hadronic vacuum polarization (HVP) contribution to the anomalous magnetic moment of the $τ$ lepton ($a_τ^\mathrm{HVP,\,LO},\, a_τ= (g-2)_τ/2$). The QCD contributions to $a_τ^\mathrm{HVP,\,LO}$ are modeled as a combination of Laplace QCD sum rules (LSRs) and finite-energy QCD sum rules (FESRs) of various weights, which are bounded by Hölder and related fundamental inequalities. The light-quark ($u,\,d,\,s$) contributions are evaluated up to five-loop order in perturbative contributions in the chiral limit, LO in light quark mass corrections, next-to-leading order (NLO) in dimension-four QCD condensates and to LO in dimension-six QCD condensates. Extending previous work on $a_μ^\mathrm{HVP,\,LO}$, we present a robust methodology for using QCD sum rules to constrain $a_τ^\mathrm{HVP,\,LO}$, and more generally any lepton $g-2$. A criterion is developed to extract from the bounds analysis a final result of $a_τ^\mathrm{QCD} = \left( 3.385\pm 0.048 \right) \times 10^{-6}$, which after inclusion of charmonium and bottomonium contributions gives a final result $a_τ^\mathrm{HVP,\,LO} = \left( 3.501\pm0.048\right) \times 10^{-6}$, comparable to data-driven and lattice predictions. Lattice QCD results along with our value for $a_τ^\mathrm{HVP,\,LO}$ are used to obtain a combined QCD determination from lattice and sum-rule methods of $a_μ^\mathrm{HVP,\,LO} = (7.217\pm0.233)\times10^{-8}$. Our analysis provides fundamental QCD sum-rule inputs into the theoretical determinations of $a_τ$, and establishes new methodologies for QCD sum-rule determinations of low-energy observables.

hep-ph↗

arXiv2026

EverMine: Dissecting the Self-Evolution of Research Capabilities in Long-Horizon Alpha Research

Self-evolving agents aim to turn research feedback into reusable skills, tools, and research rules. Whether these accumulated capabilities continue to improve later research requires controlled evaluation. Long-horizon alpha discovery provides a state-dependent setting: once a new factor enters the portfolio, the predictive information already covered changes, so the value of the same candidate or experience may change over time. We introduce EverMine, an empirical framework for studying self-evolving research capabilities in long-horizon alpha discovery. EverMine decomposes the research state into history (Hist), the current factor portfolio (Frontier), and reusable capabilities (Cap). Under matched resource limits, we compare complete runs with fixed or evolving Cap, and replace Cap while holding Hist and Frontier fixed to estimate the conditional value of accumulated capabilities. We also combine full trajectories with historical-state replay to examine how experience-based decisions affect candidate selection and portfolio outcomes. Across 18 long-horizon trajectories, end-to-end comparisons show no consistent gain from Cap evolution. Across 48 continuation branches from shared Hist and Frontier states, accumulated Cap also does not consistently outperform the initial Cap. Parameter tuning of existing factor structures can still improve the portfolio. In an exploratory replay of two screening batches from one Evolving trajectory, some screened-out candidates have positive marginal value at the original state, yet submitting all screened-out candidates sequentially slightly lowers final portfolio IC in both batches. These results show that candidate value depends on the evolving portfolio and submission order, and motivate evaluating self-evolving research capabilities through end-to-end outcomes, conditional capability value, and the consequences of experience-based decisions.

cs.AI↗

arXiv2026

LoRa Fluid Antenna Multiple Access

Concurrent long-range (LoRa) transmissions over the same time-frequency and spreading factor (SF) resources generally result in packet collisions, as the gateway cannot distinguish the overlapping signals from different end devices (EDs). This paper advocates a new fluid antenna multiple access (FAMA) framework for LoRa, referred to as {\it lora}-FAMA, to provide spatial opportunities for LoRa multiuser communications. In {\it lora}-FAMA, a gateway employs a single fluid antenna connected to only one radio-frequency (RF) chain, whose radiating element traverses the antenna aperture by sequentially visiting all candidate positions, i.e., fluid antenna `ports', within each symbol interval. The signal segments collected along the trajectory are compensated using the channel state information for the target ED. As a result, the desired signal is coherently accumulated, whereas the signals from other EDs experience unmatched channel variations and thus cannot be coherently combined. Applying the compensation separately to each active ED enables simultaneous multiuser transmission. We analyze the statistical performance of {\it lora}-FAMA under asynchronous transmissions and spatially correlated fading, as well as the large-aperture limiting case with independent and identically distributed fading. Numerical results show close agreement between the analytical and Monte Carlo results. It is revealed that, with a normalized aperture of $4\times4$ and $\mathrm{SF}=9$, a gateway can simultaneously serve more than $10$ EDs over the same frequency and SF resources while maintaining a symbol error rate below $10^{-4}$. These results demonstrate the potential of fluid antennas to enable LoRa multiple access without multiple RF chains.

eess.SP↗

arXiv2026

Collab-Solver: Collaborative Solving Policy Learning for Mixed-Integer Linear Programming

Mixed-integer linear programming (MILP) has been a fundamental problem in combinatorial optimization. Conventional MILP solving mainly relies on carefully designed heuristics embedded in the branch-and-bound framework. Driven by the strong capabilities of neural networks, recent research is exploring the value of machine learning alongside conventional MILP solving. Although learning-based MILP methods have shown great promise, existing works typically learn policies for individual modules in MILP solvers in isolation, without considering their interdependence, which limits both solving efficiency and solution quality. To address this limitation, we propose Collab-Solver, a novel multi-agent-based policy learning framework for MILP that enables collaborative policy optimization for multiple modules. Specifically, we formulate the collaboration between cut selection and branching in MILP solving as a Stackelberg game. Under this formulation, we develop a two-phase learning paradigm to stabilize collaborative policy learning: the first phase performs data-communicated policy pretraining, and the second phase further orchestrates the policy learning for various modules. Extensive experiments on both synthetic and large-scale real-world MILP datasets demonstrate that the jointly learned policies significantly improve solving performance. Moreover, the policies learned by Collab-Solver have also demonstrated excellent generalization abilities across different instance sets.

cs.AI↗

arXiv2026

Complex Problem Solving in Large Language Models: A Statistical Control Survey and Diagnostic Framework

Complex problem solving (CPS) with large language models (LLMs) is often framed as a matter of stronger reasoning or longer generation. Yet early-step error amplification, prompt brittleness, and failures to revise incorrect commitments are difficult to explain by missing knowledge or expressive capacity alone. This survey interprets CPS as a sequential estimation-and-decision problem over a latent solution state. A controller maintains a belief about an unobserved solution trajectory, updates it as noisy intermediate evidence arrives, and decides whether to commit, verify, branch, roll back, or abstain to minimize expected loss. Reasoning supplies candidate transitions and interpretations, whereas process control shapes and evaluates those proposals and regulates subsequent transitions and observations. Within this framework, we organize existing methods around five components: explicit state representation, transition structuring, validation and constraint enforcement, search and rollback, and uncertainty management. We also interpret evaluation metrics according to the statistical quantities they estimate. The framework further yields a diagnostic hypothesis: interventions should be most effective when they target the error or uncertainty component implicated by an observed failure. We distinguish systematic, stochastic, and irreducible error together with epistemic and aleatoric uncertainty, and call this alignment problem-control fit and its failure control mismatch. For example, additional sampling may reduce sampling variability while leaving a shared systematic error unchanged. This perspective clarifies what current methods estimate and control, what remains uncontrolled, and why reliable validation, targeted recovery, calibrated uncertainty, and matched-budget evaluation are central open problems.

stat.ML↗

arXiv2026

Anchoring What Matters: A Dual-Level Learning Framework for Visually-Grounded Multimodal Reasoning

Reinforcement learning with verifiable rewards (RLVR) has significantly improved the reasoning capabilities of large vision-language models (LVLMs). However, standard on-policy RLVR algorithms face a critical optimization bottleneck in preserving and reinforcing visually grounded reasoning behaviors: valuable visually-grounded reasoning trajectories are discarded after a single update, while uniform token advantage allocation prevents the model from reinforcing critical perception or reasoning steps. To bridge this gap, we propose PIVOT, a dual-level learning framework that anchors policy optimization around informative visual reasoning signals. Specifically, PIVOT introduces a self-calibrated experience replay mechanism, which selectively collects and replays visually-grounded historical experiences as stable reference anchors for policy optimization. Building upon this, we further design a vision-guided advantage allocation mechanism to allocate additional vision-aware advantages to tokens based on their local visual support and impact on downstream reasoning. Extensive experiments across diverse benchmarks demonstrate that PIVOT achieves highly competitive performance in enhancing the multimodal reasoning capabilities of LVLMs.

cs.AI↗

arXiv2026

SoK: Intent-Oriented Systematization of Multi-Turn LLM Jailbreaks

Large Language Models (LLMs) are increasingly deployed in interactive settings, where user intent commonly unfolds through multi-turn dialogue. Multi-turn jailbreaks exploit this pattern by advancing a harmful intent across turns, so that no single message exposes the full objective. However, existing work treats these attacks as a loose collection of prompt patterns and does not analyze how the adversary organizes and advances harmful intent across an interaction. We develop a four-part, intent-oriented taxonomy that organizes multi-turn jailbreaks by adversarial intent structure. Through controlled ablations, we find that effectiveness is driven by how deliberately intent is organized across turns rather than by context length or query count. We further show that the way intent is organized determines the level at which it becomes detectable, pushing the required detection surface outward from the turn level to the session level to the cross-session level. These findings indicate that turn-local safety mechanisms are structurally insufficient and that single-point evaluation overlooks how intent is organized, motivating evaluation protocols aligned to the level at which harmful intent becomes observable. The code is available at: https://github.com/SiyuanLi00/INTACT.

cs.CR↗

arXiv2026

Toward Robust Personalized Alignment for LLMs: Mitigating Persona Drift in Multi-Turn Dialogue

Persona drift remains a central challenge for personalized language models, as user profiles evolve over long interactions rather than remain permanently fixed. Models must therefore revise persistent persona states when preferences genuinely change, while avoiding updates driven by transient, ambiguous, or unresolved observations. We propose CORE, which separates turn-local evidence from persistent persona-state revision and selectively updates grounded user preferences through uncertainty-aware belief revision. We also introduce PERSIST, a held-out post-anchor benchmark for persona-state robustness under sequential interaction stress, covering ambiguity, conflict, and controlled social influence. Across ALOE, PersonaChat, and PERSIST, CORE improves personalized alignment and robustness, with complementary gains in normalized closed-slot state fidelity. Human evaluation and mechanistic controls further support explicit update control beyond stronger generation or persistent memory alone.

cs.AI↗

arXiv2026

New Evidence, Same Choice: Testing Physical Experiment Selection in Vision Language Models

A model first sees an image from one physical measurement experiment, such as how far a block coasted, and must answer a question about a new trial, such as whether the block will pass a target after a fixed push. The initial experiment may provide enough information to answer, or the model may need another measurement, such as the object's mass, friction, restitution, or spring stiffness. We study whether vision language models can decide when to answer immediately and, when more evidence is needed, which experiment to perform. Current physical reasoning benchmarks usually evaluate only the final answer, so they do not directly measure this decision-making ability. We introduce a controlled evaluation where each problem provides one measurement image and four possible physical worlds created by combining two possible masses and two possible values of another relevant property. The model must either stop and answer or select the cheapest additional experiment that can resolve the question. We construct matched problem pairs where changing either the observed measurement or the question changes the optimal action. Since all possible worlds and experiment costs are known, we can explicitly determine the optimal choice. Across six open models and 144 physical parameter sets, direct responses repeat the same action for 95.1% to 100% of image pairs even when the correct action changes. Brief reasoning improves action switching, but the best model makes both decisions correctly for only 5.9% of image pairs. Additional analysis reveals failures in measurement interpretation, physical reasoning, and response formatting. By evaluating evidence selection separately from final answers, our benchmark reveals limitations in physical reasoning that conventional answer accuracy can overlook.

cs.CV↗

arXiv2026

UBone3D: Physics-Rectified Conditional Flow Matching for Anatomical 3D Shape Completion from Ultrasound

Three-dimensional ultrasound (US) is a safe, radiation-free complementary modality to CT and X-rays for longitudinal monitoring, yet its segmentation-derived partial point clouds are extremely artifact-laden. Consequently, it is challenging to recover a clean and complete anatomical structure from such US point clouds. In this paper, we present UBone3D, a novel framework based on physics-rectified conditional flow matching (CFM) that performs point cloud completion directly from partial US observations. UBone3D models deterministic physics artifacts (e.g., surface thickening, streaking, dropouts) via a simulated physics proxy, and introduces test-time physics rectification to steer the shape completion. At inference, the completion is jointly steered by two decoupled forces: (1) anatomical plausibility enforced by a CT-trained generative shape prior, BoneFM, and (2) physics consistency enforced by USimNet in the ultrasound formation space. Extensive experiments on simulated and in-vivo data demonstrate significant improvements in reconstruction accuracy and anatomical fidelity over existing baselines.

cs.CV↗

arXiv2026

Safe Task Planning with Long-Term Graph Memory for Embodied Agents

Large language models (LLMs) and vision-language models (VLMs) have significantly advanced zero-shot task planning for embodied agents. However, most LLM- and VLM-driven methods struggle to generate safe high-level actions due to a lack of physical risk awareness, particularly under partial observability, where hazards lie outside the immediate field of view. To address this challenge, we propose a novel safe task-planning framework, SafeMem, which constructs and maintains a long-term semantic graph memory of the open and dynamic environment. Based on egocentric observations, the proposed framework incrementally accumulates knowledge about surrounding objects and their relationships with a graph. Then, an LLM-based risk predictor evaluates candidate actions using the graph memory, triggering a conservatism-modulated replanning loop with explanations for detected hazards. Extensive experiments on the IS-Bench benchmark and a real-world robot platform demonstrate that the SafeMem framework substantially improves safe success rates compared to state-of-the-art VLM-driven task planners. Video results are available on our webpage: https://sites.google.com/view/safemem.

cs.RO↗