arXiv subjects

Weiyi Kong

Publications and source records attributed to Weiyi Kong.

12 recordsLinked to original sources

arXiv2026

Multi-LLM Systems Exhibit Robust Semantic Collapse

Whether machines can originate novel content has been debated for nearly two centuries, from Lovelace's assertion that no engine can "originate anything" to Turing's question of whether a machine can amplify ideas brought in from outside. Systems of multiple interacting LLMs, increasingly deployed for autonomous generation, reopen this question empirically. Here we show that such systems, operating in inference-only setups, exhibit semantic collapse: systematic convergence in semantic representations despite apparent lexical variation. Across model families, extended simulations of 200 to 1,000 rounds, the pattern remains consistent. Thirteen intervention strategies, spanning decoding parameters, prompt design, agent composition, activation engineering, and reinforcement learning, fail to restore semantic diversity. Mechanistic analyses suggest that semantic collapse is not explained by alignment or conformity biases, but is consistent with intrinsic properties of autoregressive generation. Our results point to persistent constraints on the ability of multi-LLM systems to sustain open-ended exploration in closed-loop settings.

cs.MA↗

arXiv2026

CARE: Context-Aware Ranking Evolution with Executable Scoring Programs for Budgeted Reaction Optimization

High-throughput experimentation can evaluate many reaction conditions, yet combinatorial condition spaces still exceed the available experiment budget. This makes experiment selection a sequential decision problem: each new condition must be chosen from limited observations before its outcome is known. LLMs can express task-specific selection logic. A direct recommendation, however, is neither a persistent executable object that can be validated and revised nor an independently auditable decision rule. We introduce CARE, a reference-conditioned controller that separates program synthesis from experiment selection. An LLM writes an executable scoring program that ranks the remaining conditions, while a non-LLM reference policy supplies a numerical candidate and support summary. CARE forms an optional alternative from the program, applies a reference-conditioned intervention gate to compare it with the reference, and records the decision before the selected outcome is revealed. Each new outcome updates controller state and can trigger retention, revision, or regeneration of the active program. This outcome-guided program evolution changes the scoring logic without updating the LLM parameters. In matched offline replay with 30 seeds on eight reaction-optimization tasks, CARE attains the lowest normalized regret, the highest normalized best-so-far AUC, and the highest Top-1% Success@15 among the evaluated methods. These results support using a scoring program written by an LLM as one component of a reference-conditioned optimizer rather than as a standalone experiment selector.

cs.LG↗

arXiv2026

Parabolic BMO Spaces, Muckenhoupt Weights, and Reverse Hölder Classes with Time Lag: Equivalence and Characterizations

For any given time lag $γ\in(0,1)$, we prove that the one-sided parabolic BMO space $\mathrm{BMO}^+(γ)$ coincides with the parabolic BMO space $\mathrm{PBMO}^-(γ)$ with equivalent norms, the parabolic Muckenhoupt class $A_{\infty}^+(γ)$ defined via the reverse Jensen inequality can be represented as the union of the parabolic Muckenhoupt classes $A_r^+(γ)$ with $r\in[1,\infty)$, and the parabolic reverse Hölder classes $\bigcup_{q\in(1,\infty]}RH_q^+$ coincide with the parabolic Muckenhoupt classes $\bigcup_{r\in[1,\infty)}A_r^+(γ)$, and hence give affirmative answers to Questions 4.5 and 4.6 posed by Kinnunen and Saari [Nonlinear Anal. 131 (2016)]. To show them, we establish the uniform parabolic space-time shifting property for parabolic reverse Hölder weights, and develop the one-sided stopping time argument which yields a new parabolic John--Nirenberg inequality for $\mathrm{BMO}^+(γ)$. As applications, we obtain John--Nirenberg and exponential integrability characterizations of $\mathrm{BMO}^+(γ)$, prove that $\mathrm{BMO}^+(γ)$ is independent of the positive time lag, and identify its null space.

math.CA↗

arXiv2026

CDEP Agent: Connecting Meteorologically Detected Temporal Compound Events to Real-World Documentary Evidence

Compound drought-to-extreme-precipitation (CDEP) events are recognized in climate science as a growing driver of extreme impact, but whether this recognition carries over into real-world early warning and post-event documentation is unknown, so a meteorologically real CDEP event may pass with neither advance warning nor any later record. Here we present CDEP Agent, an auditable LLM-agent framework that tests this mismatch directly by linking CDEP candidates detected from meteorological reanalysis to real-world hazard and impact evidence across sources with different spatial scales, temporal resolutions, and reporting conventions. Using California as a case study, we identify 408 candidate CDEP events from ERA5 observations during 2021-2025 and evaluate each against the U.S. Drought Monitor, NOAA Storm Events, and public webpages along five dimensions: antecedent drought, extreme rainfall, local impact, hazard-impact attribution, and explicit drought-to-rainfall linkage. Only 34.3% of candidates are corroborated on both hazard components, and just 1.5% are ever explicitly linked to their antecedent drought, indicating that most meteorologically detected CDEP events go undocumented and their compound nature almost never enters the record at all. Our framework gives climate scientists a way to test physical event definitions against what actually gets documented, and gives social scientists, economists, and disaster-response agencies a provenance-linked evidence base for compound events that current warning and reporting systems largely fail to capture.

cs.AI↗

arXiv2026

Same Evidence, Different Target: Decoding How Diagnostic Evidence Bears on Causal Questions from Language-Model States

The same diagnostic result can support or challenge one causal claim yet fail to address another when the claims concern different populations, outcomes, estimands, pathways, or identifying assumptions. When the evidence and target vary together, a correct answer may reflect favorable or adverse wording, lexical overlap, or a familiar diagnostic pattern rather than matching the evidence to the causal question. We introduce paired prompts that repeat the same diagnostic evidence verbatim while changing the causal target. Each prompt is labeled Favors, Challenges, Unresolved, or Wrong Target according to how the evidence bears on the causal question. A pair is recovered only when both prompts are classified correctly. Using linear readouts trained on a separate development set, we analyze the final-token hidden state from the penultimate transformer block of Qwen2.5-7B-Instruct, Qwen3-8B, and Llama-3.1-8B-Instruct. On the 49-pair primary benchmark spanning nine diagnostic families, balanced accuracy ranges from 0.654 to 0.659 and 18-21 pairs are recovered. Two independent human reviewers assigned the same label to 95 of the 98 prompts (96.9%). Across checkpoints, balanced accuracy and complete-pair recovery exceed permutation nulls that preserve development scenario groups. In Qwen2.5, full-prompt balanced accuracy exceeds both restricted inputs, with paired-bootstrap intervals for both differences above zero. Readouts trained without development examples from the evaluated diagnostic family recover 21 pairs, including at least one in each of the nine families. The hidden-state readout exceeds a linear classifier on answer-option logits and text baselines in balanced accuracy and recovered pairs. These results show that the hidden state contains linearly decodable information about whether diagnostic evidence favors, challenges, or fails to address the causal target.

cs.CL↗

arXiv2026

Sobolev--Morrey Spaces and Divergence-Form Degenerate Second-Order Elliptic Equations on Domains with Higher Co-Dimensional Boundaries

In this article, we study the weighted homogeneous Sobolev--Morrey spaces on domains in $\mathbb{R}^n$ with higher co-dimensional boundaries. Precisely, we systematically establish a real-variable theory of these spaces, including completeness, embedding theorems, Riesz potential characterizations, continuity, trace and extension theorems, and complex interpolation. Applying the boundedness of the trace and the extension operators, we obtain sharp weighted a priori estimates for solutions to the Dirichlet problem of divergence-form degenerate second-order elliptic equations on such domains in weighted Lebesgue spaces. The absence of a boundary manifold structure of these domains poses some essential difficulties, which are overcome by using some tools, such as the intrinsic properties of distance weights and the geometric structure of domains, different from those available in Lipschitz domains.

math.AP↗

arXiv2026

Large Language Models as Explainable Cyberattack Detectors for Energy Industrial Control Systems

In modern energy systems, industrial control systems (ICS) and power-system SCADA require intrusion detection that is not only accurate but also auditable by operators. The ICS intrusion-detection landscape is currently dominated by established supervised detectors. In this paper, we study whether an off-the-shelf large language model (LLM) can serve as a complementary, human-in-the-loop layer for Modbus traffic. We cast this as a binary network-side normal/critical decision task on two public ICS Modbus datasets, collapsing attack periods and other safety-critical behaviors into a single critical class. Each Modbus communication instance is converted into a compact token string derived from discretized protocol fields, and a prompt-configured LLM produces a normal/critical alert together with a concise, token-grounded incident record for analyst review. Under matched event information and shared evaluation splits, the resulting LLM-based triage pipeline achieves high predictive performance on both benchmarks and is broadly comparable to strong supervised baselines, while requiring no task-specific weight updates. To assess the audit record, we apply intervention-based diagnostics, including sufficiency- and necessity-style tests, which provide evidence that the cited tokens are often decision-relevant to the model's own prediction. These records are intended as audit signals rather than full human-grounded explanations.

cs.CR↗

arXiv2026

A unified SPH framework for shell-related interactions

A unified Smoothed Particle Hydrodynamics (SPH) framework is proposed to simulate interaction dynamics involving thin shells modeled by a reduced-dimensional, single-layer particle discretization, as opposed to full-dimensional SPH solids. The framework encompasses one-sided fluid-shell interactions, with the fluid present on only one side of the shell, as well as solid-shell, shell-shell, and shell-self interactions The study introduces a novel concept of imaginary shell contact particles, generated by projecting real shell particles along the local normal direction within the cut-off radius of the fluid particle, thereby mapping this reduced-dimensional shell model into a full-dimensional representation. With the volume of the imaginary particles defined based on the local shell curvature, the projection preserves kernel completeness for fluid-shell interactions while leaving the fluid-structure interaction (FSI) dynamics unchanged, such that the fluid-shell coupling algorithm is the same as in standard fluid-solid coupling. In addition, a particle-to-particle contact model for solid-solid interactions is developed by analogy to fluid dynamics: a contact density is computed using a fluid-style density initialization, and the resulting contact forces follow a momentum-equation-inspired formulation. Combined with the projection strategy, this contact formulation is directly extended to efficiently handle shell-related contact problems. The proposed method is validated using a series of benchmark tests, demonstrating stable and accurate performance across diverse interaction scenarios.

physics.flu-dyn↗

arXiv2026

One-Sided and Parabolic BLO Spaces with Time Lag and Their Applications to Muckenhoupt $A_1$ Weights and Doubly Nonlinear Parabolic Equations

In this article, we first introduce the one-sided BLO space $\mathrm{BLO}^+(\mathbb{R})$ and characterize it, respectively, in terms of the one-sided Muckenhoupt class $A_1^+(\mathbb{R})$ and the one-sided John--Nirenberg inequality. Using these, we establish the Coifman--Rochberg type decomposition of $\mathrm{BLO}^+(\mathbb{R})$ functions and show that $\mathrm{BLO}^+(\mathbb{R})$ is independent of the distance between the two intervals, which further induces the characterization of this space in terms of the one-sided BMO space $\mathrm{BMO}^+(\mathbb{R})$ (the Bennett type lemma). As applications, we prove that any $\mathrm{BMO}^+(\mathbb{R})$ function can split into the sum of two $\mathrm{BLO}^+(\mathbb{R})$ functions and we provide an explicit description of the distance from $\mathrm{BLO}^+(\mathbb{R})$ functions to $L^\infty(\mathbb{R})$. Finally, as a higher-dimensional analogue we introduce the parabolic BLO space $\mathrm{PBLO}_γ^-(\mathbb{R}^{n+1})$ with time lag, and we extend all the above one-dimensional results to $\mathrm{PBLO}_γ^-(\mathbb{R}^{n+1})$; furthermore, as applications, we not only establish the relationships between $\mathrm{PBLO}_γ^-(\mathbb{R}^{n+1})$ and the solutions of doubly nonlinear parabolic equations, but also provide a necessary condition for the negative logarithm of the parabolic distance function to belong to $\mathrm{PBLO}_γ^-(\mathbb{R}^{n+1})$ in terms of the weak porosity of the set.

math.AP↗

arXiv2025

Toward Efficient FSI Modeling in Patient-Specific Arteries: SPH Simulation of Blood Flow in Thin Deformable Vessels

Accurate simulation of blood flow in deformable vessels is critical in cardiovascular research for understanding disease progression and informing clinical decision-making. However, due to the thin-walled nature of arteries, traditional smoothed particle hydrodynamics (SPH) approaches based on full-dimensional volume modeling often require extremely fine particle spacing to ensure numerical convergence for the solid mechanics. This, in turn, leads to redundant resolution in the fluid domain to maintain sufficient kernel support near the fluid-solid interface in fluid-structure interaction (FSI) simulations. To address this limitation, we propose an efficient reduced-dimensional shell-based SPH method for modeling thin-walled deformable arteries, and conduct FSI for capturing hemodynamics and arterial wall mechanics. Through a series of validation cases, the proposed shell model demonstrates comparable accuracy in fluid dynamics to the volume model, while achieving faster convergence in solid mechanics and reduced computational cost. We further investigate the influence of wall compliance on flow transitions and key hemodynamic indices, highlighting the necessity of FSI modeling over rigid-wall assumptions. Finally, the method is applied to two patient-specific vascular geometries, i.e. the carotid artery and the aorta, which demonstrates its robustness, efficiency and physiological relevance in realistic cardiovascular simulations.

cs.CE↗

arXiv2025

Parabolic Extrapolation and Its Applications to Characterizing Parabolic BMO Spaces via Parabolic Fractional Commutators

In this article, we establish the parabolic version of the celebrated Rubio de Francia extrapolation theorem. As applications, we obtain new characterizations of parabolic BMO-type spaces in terms of various commutators of parabolic fractional operators with time lag. The key tools to achieve these include to establish the appropriate form in the parabolic setting of the parabolic Rubio de Francia iteration algorithm, the Cauchy integral trick, and a modified Fourier series expansion argument adapted to the parabolic geometry. The novelty of these results lies in the fact that, for the first time, we not only introduce a new class of commutators associated with parabolic fractional integral operators with time lag, but also utilize them to provide a characterization of the parabolic BMO-type space in the high-dimensional case.

math.FA↗

arXiv2024

Parabolic Muckenhoupt Weights Characterized by Parabolic Fractional Maximal and Integral Operators with Time Lag

In this article, motivated by the regularity theory of the solutions of doubly nonlinear parabolic partial differential equations the authors introduce the off-diagonal two-weight version of the parabolic Muckenhoupt class with time lag. Then the authors introduce the uncentered parabolic fractional maximal operator with time lag and characterize its two-weighted boundedness (including the endpoint case) via these weights under an extra mild assumption (which is not necessary for one-weight case). The most novelty of this article exists in that the authors further introduce a new parabolic shaped domain and its corresponding parabolic fractional integral with time lag and, moreover, applying the aforementioned two-weighted boundedness of the uncentered parabolic fractional maximal operator with time lag, the authors characterize the (two-)weighted boundedness (including the endpoint case) of these parabolic fractional integrals in terms of the off-diagonal (two-weight) parabolic Muckenhoupt class with time lag; as applications, the authors further establish a parabolic weighted Sobolev embedding and a priori estimate for the solution of the heat equation. The key tools to achieve these include the parabolic Calderón--Zygmund-type decomposition, the chaining argument, and the parabolic Welland inequality which is obtained by making the utmost of the geometrical relation between the parabolic shaped domain and the parabolic rectangle.

math.AP↗