arXiv subjects

Feng Liu

Publications and source records attributed to Feng Liu.

At least 37 records · Page 2Linked to original sources

arXiv2026

TRACE: A Multi-Agent System for Autonomous Physical Reasoning for Seismology

Modern seismic networks resolve earthquake sequences in unprecedented detail, yet explaining how large earthquakes emerge from evolving fault systems remains difficult. We introduce TRACE, a seismology-guided artificial intelligence agent that plans and executes workflows while preserving auditable evidence chains from observations to physical interpretation. We evaluated TRACE through 104 benchmark tasks and two complementary earthquake sequences. For the well-studied 2019 Ridgecrest sequence, TRACE constructed a high-resolution catalog from continuous waveforms and retrospectively recovered delayed cascading activation between the Mw 6.4 and Mw 7.1 earthquakes without a prescribed target interpretation. In the less-understood 2025-2026 Sanriku sequence off northeastern Japan, TRACE developed a testable interpretation of progressive destabilization within a segmented megathrust. Its synthesis linked coupled seismic-aseismic activation around the MJ 6.9 sequence and subsequent persistent, spatially segmented shallow-interface activity to a megathrust patch that lay between regions of past large coseismic slip and later hosted the MJ 7.7 rupture. These results open a path from seismic observations to testable physical insight.

physics.geo-ph↗

arXiv2026

Language-Conditioned World Modeling for Visual Navigation

Goal-conditioned visual navigation has been a long-standing testbed for embodied AI. We study a natural language-conditioned variant, language-conditioned visual navigation (LCVN), in which an embodied agent must follow a natural language instruction given only an initial egocentric observation. Without access to goal images, the agent must rely on language to shape its perception and continuous control. We introduce the LCVN Dataset, a benchmark of 39,016 trajectories and 117,048 human-verified instructions spanning diverse environments and instruction styles. Building on this benchmark, we study two complementary paradigms: (i) latent-imagination policy learning, in which a diffusion-based world model (LCVN-WM) imagines future observations and an actor-critic agent (LCVN-AC) learns its policy entirely within the imagined latent space; and (ii) unified autoregressive prediction, in which a single multimodal backbone (LCVN-Uni) jointly predicts actions and observations in one forward pass over a shared token sequence. Experiments show that two paradigms offer complementary strengths: latent imagination produces more temporally coherent rollouts, whereas unified prediction generalizes better to unseen environments. Targeted ablations further isolate the contributions of language guidance, conditioning signals, and instruction style, clarifying when language grounding versus dynamics modeling is the performance bottleneck. Together, these findings position LCVN as a testbed for studying how language, imagination, and decision-making interact in embodied agents.

cs.CV↗

arXiv2026

Altermagnetism produces pair emission and absorption from dark excitons and magnons in La$_2$O$_3$Mn$_2$Se$_2$

Altermagnets' (AMs) non-relativistic spin splitting enables novel states, spintronic and magneto-optical devices, though their optical signatures remain elusive. Here, we report exciton-magnon emission and absorption: optical sidebands from a spin-forbidden dark exciton, a direct consequence of altermagnetic symmetry. Combined optical spectroscopy and first-principles calculations reveal that La$_2$O$_3$Mn$_2$Se$_2$ is an altermagnetic insulator, hosting a strongly bound, spin-forbidden dark exciton and a higher-energy bright exciton. Photoluminescence (PL) and absorption reveal mirror-image sidebands, Stokes-shifted in emission and anti-Stokes-shifted in absorption, symmetric about the dark exciton, whose energy shifts and spectral shapes match the magnon energy scale and density of states measured independently by inelastic neutron scattering. The PL intensity tracks the full equal-time spin-spin correlator, combining static and dynamical contributions, and rules out alternative processes. This directly couples PL to magnetism, with potential for magneto-optical devices. These results establish exciton-magnon spectroscopy as a new route for optically identifying and exploiting AMs.

cond-mat.mtrl-sci↗

arXiv2026

Reprogramming Vision-Language Models via Structured Prompt Reparameterization

Visual reprogramming adapts pretrained models to downstream tasks by modifying their input and output interfaces while keeping the backbone fixed. In vision-language models, existing methods mainly rely on intra-class prompt aggregation and do not explicitly model relationships among classes. However, fine-grained categories often exhibit highly overlapping attribute descriptions and strong inter-class correlation in the text embedding space, where discriminative cues lie in subtle low-variance components. We propose Reparameterized Inter-Class Visual Reprogramming (RVP), a structured framework that aggregates multiple text prompts within each class and applies residual correction across classes. We also show that CLIP-based visual reprogramming with input-independent linear output aggregation can be expressed as a linear mapping from frozen image embeddings to downstream logits, and use this view to design a structured reparameterization that models shared semantic components and class-specific differences. RVP uses only a single visual prompt and can be reparameterized at inference into a frozen backbone followed by a linear classifier, incurring nearly zero computational overhead. Across 11 few-shot classification benchmarks and four CLIP backbones, RVP consistently improves over prior visual reprogramming methods with comparable or better inference efficiency.

cs.CV↗

arXiv2026

Does Execution Require Target KV Fidelity? A Mixed-Fidelity KV Runtime for LLM Serving

Large language model (LLM) serving is increasingly constrained by the GPU memory consumed by key-value (KV) caches. Existing compression, eviction, and offloading techniques alleviate this pressure, but serving runtimes typically treat only the configured target KV representation as execution-ready. Under memory pressure, this target-only contract can turn KV shortage into request stalls and preemptions. We present ElasticKV, a mixed-fidelity KV runtime built on the observation that target fidelity need not gate execution. ElasticKV introduces a compact intermediate KV state, making fidelity a runtime-managed execution property. To realize this state in a paged serving runtime, ElasticKV combines (i) a pair-structured layout that turns fidelity reduction into reusable GPU capacity, (ii) a dual-mode attention backend that directly consumes the compact state while preserving the native target-only path, and (iii) pressure-aware fidelity management that adapts KV fidelity to memory pressure. Our extensive evaluation across diverse workloads, model families and scales, and GPU platforms demonstrates the effectiveness and generality of ElasticKV. Under high concurrency, ElasticKV achieves 3.8-4.0$\times$ lower time-to-first-token (TTFT) and 9.1$\times$ lower P90 TTFT than vLLM while preserving generation quality.

cs.LG↗

arXiv2026

Reshaping Converter-Network Interactions in Microgrids: From Virtual Impedance to Virtual Two-Port Control

Converter terminal characteristics are central to both dynamic interactions with the network and steady-state power sharing in inverter-based microgrids. Conventional additive virtual impedance (VI) shapes these characteristics through a single virtual branch. This paper proposes virtual two-port control, which uses four coordinated transfer channels to reconstruct the terminal impedance. The connected network consequently observes the original converter through a virtually inserted twoport interface, with conventional virtual impedance recovered as a degenerate case. The proposed control transforms the original impedance through a matrix linear-fractional map. We also derive a necessary and sufficient condition for the reconstruction to be stable and proper. Among its various potential applications in microgrids, we develop three representative ones in detail: passivation with reduced control effort, uncertainty compression, and seriesshunt power-flow regulation. Simulations of VSG-controlled converters demonstrate the advantages of the proposed method over conventional VI in these applications.

eess.SY↗

arXiv2026

Destination Support Restoration for Finite-Set Multimodal Trajectory Prediction

Robots operating around pedestrians often reason over a finite set of predicted human futures. Repeated online updates can concentrate this limited prediction budget on dominant destinations and leave plausible alternatives underrepresented or absent, removing those alternatives from the finite representation available to downstream decision making. We introduce Destination Support Restoration (DSR), a causal post-selection operator that repairs destination support without retraining the host predictor or increasing the maintained set size. At a repair step, DSR evaluates a temporary destination-stratified candidate bank from the observed prefix, converts candidate evidence into integer target counts, protects representatives of active modes, and reallocates redundant surplus hypotheses to deficient modes. The maintained and returned sets retain exactly $N$ hypotheses, and DSR replaces at most $\lceilρN\rceil$ entries. Protected representatives preserve current categorical support; lineage-aware particle filters also preserve surviving resampling ancestors. Each replacement reduces the allocation mismatch to the evidence-driven target by one. On the complete 3,719-trajectory Edinburgh protocol over three seeds, DSR reduces MIF weighted ADE and FDE by 13.36% and 13.30% at $N=64$. Paired integrations with CLiFF, PPT, causal GDTS, Social Informer, and PECNet improve both metrics in every evaluated pair. These results show that finite-set support allocation is a useful prediction-side control point when a fixed hypothesis set serves as the interface to downstream systems.

cs.RO↗

arXiv2026

Large-Eddy Simulation and Performance Analysis of a Continuous-Turbine-Burner Stage

This paper reports the first large-eddy simulation of chemically reacting flow in a turbine stage to analyze the influence of fuel injection and combustion on its aerodynamic and thermodynamic performance. Two reacting cases---with four and sixteen fuel injectors at the inlet for each stator passage---are computed and compared against two nonreacting cases, one with four fuel injectors and the other without. The analyses indicate viability for the continuous-turbine-burner (CTB) concept, which offers the potential of significant increase of specific power/thrust of a gas-turbine engine without loss of efficiency. Fuel injection and combustion have minimal influence on the total-pressure loss. Compared with the baseline nonreacting case, the stage work per unit mass increases by 8.5% and 11.5% in the two reacting cases, while the residual work rises by 17.3% and 16.0%, respectively. The two reacting cases exhibit a 14.5% increase of overall work and a thermal efficiency of 44% for the fuel injection. Local high temperature on the rotor blade is suppressed by using a more uniform spanwise distribution of fuel injectors. The work extraction process of a CTB is analyzed from both thermodynamic and mechanical views.

physics.flu-dyn↗

arXiv2026

Sample-Conditioned Representation Selection for Audio Few-Shot Learning

Few-shot audio classifiers may rely on foreground-background co-occurrences and fail when those correlations shift. On SpurAudio, the resulting representation shift is concentrated and class dependent: for ResNet12, the top 10 percent of channels explain 82.80 percent of the null-corrected shift contribution. We propose SAMPLESELECT, which predicts a fixed-budget feature mask independently for each input while keeping the encoder and source classifier frozen. Training uses differentiable Gumbel Top-k selection with foreground classification and cross-background contrastive losses; inference uses deterministic Top-k masks and support-only linear adaptation. Across ResNet12 and Conv64 in 5-way 1-shot and 5-shot evaluation, SAMPLESELECT gives the best OOD accuracy among the compared methods and improves the matched full-representation control by 4.90-8.38 percentage points. Ablations and representation analyses further support the learned selection mechanism. Code is available at https://github.com/Cross-Innovation-Lab/SAMPLESELECT/

cs.AI↗

arXiv2026

Extremal problems about the order and size of nonhamiltonian locally linear graphs

The relation between local structure and global cycle properties is a classical topic in graph theory. A graph $G$ is locally linear if $G[N(v)]$ is a path for every $v\in V(G)$. It is locally Hamiltonian or locally traceable if every vertex neighborhood induces a Hamiltonian or traceable graph, respectively. Earlier work by Pareek and Skupień, Skupień, Davies and Thomassen, Asratian and Oksimets, and de Wet and van Aardt studied extremal questions for these related graph classes. We prove that the minimum order of a nonhamiltonian locally linear graph is $12$ and that, for every integer $n\geq 12$, the minimum size of such a graph of order $n$ is $2n$. We also prove that every nontraceable locally linear graph of order $n$ has at least $2n+3$ edges.

math.CO↗

arXiv2026

Rethinking Procedural Audio Pre-training: Source Scaling and Objective Adaptation

Procedural audio has emerged as a viable source for transferable audio representation learning, but its design principles remain unclear.We revisit two questions: how a procedural source should be scaled, and whether training choices developed on natural audio should transfer unchanged to procedural data.Using a controlled source, we separate scale into formula-class coverage C and within-class rendering diversity I.Experiments with FDSL and AudioMAE show that these two forms of scale provide different benefits and depend on the learning formulation and downstream task. A matched AudioMAE study further shows that procedural audio favors low mask ratios (10%--25%), whereas AudioSet-28K favors 50%--75%. Shared-codebook analysis reveals lower patch diversity and stronger temporal predictability in procedural audio. These results motivate source-aware procedural pre-training, where source scaling and learning configuration are considered jointly.Code is available at https://github.com/Cross-Innovation-Lab/Formula-Bank.

cs.SD↗

arXiv2026

BMO-Type characterizations and endpoint estimates for commutators of intrinsic square functions on Orlicz-Hardy spaces

The commutators of intrinsic square functions associated with the intrinsic Littlewood--Paley $g$-function, the intrinsic $g_λ^{*}$-function and the intrinsic Lusin area function can be used to characterize BMO-type function spaces through their boundedness properties. In this paper, we show that, for $b\in {\rm BMO}(\mathbb{R}^{n})$, the boundedness of each of these commutators from the Musielak--Orlicz Hardy space $H^φ(\mathbb{R}^{n})$ to $L^φ(\mathbb{R}^{n})$ is equivalent to $b\in {\rm BMO}_φ(\mathbb{R}^{n})$ (a nontrivial subspace of $\rm{BMO}(\mathbb{R}^{n})$), under suitable assumptions on the growth function $φ$. This extends the weighted characterization of Han and Wu [Proc. Amer. Math. Soc. 152(1) (2024), 281--293] to the Musielak--Orlicz setting. In addition, under suitable assumptions on a growth function $Φ$, we prove that, for $b\in {\rm BMO}_Φ(\mathbb{R}^{n})$, these commutators are bounded from the $b$-adapted Orlicz--Hardy space $H_{b}^Φ(\mathbb{R}^{n})$ to $L^{1}(\mathbb{R}^{n})$ and from Orlicz Hardy space $H^Φ(\mathbb{R}^{n})$ to $L^{1,\infty}(\mathbb{R}^{n})$.

math.CA↗

arXiv2026

Robust biexciton preparation for entangled photon-pair generation by single-shot Floquet driving

Quantum emitters driven by resonant two-photon excitation are a leading source for deterministically generated entangled photon pairs, essential for scalable photonic quantum technologies. However, conventional excitation schemes pose various constraints, limiting their scalability and practicability. Here, we demonstrate how biexciton preparation schemes with significantly improved robustness and reduced experimental demands can be identified using a novel design principle: ultrafast single-shot Floquet driving. This is achieved by employing two strongly and symmetrically detuned dichromatic pulses, whose superposition generates a stroboscopic Hamiltonian that enables direct coupling between ground and biexciton states. Moreover, a pulse delay serves as a tuning knob, introducing an effective magnetic field making biexciton preparation particularly robust. Experimentally, we achieve a biexciton occupation exceeding 96% and preserve photon-pair entanglement with a fidelity of 93.4%. Our work opens avenues for designing a class of quantum control protocols with superior performance across diverse quantum emitters and qubit platforms.

quant-ph↗

arXiv2026

Weighted Endpoint Variational and Jump Inequalites for Rough Singular Integrals

In this present paper, we study the weighted weak type $(1,\,1)$ bounds for rough maximal singular integral operators as well as the the corresponding variation and jump operators. Our first result is the weighted weak type $(1,\,1)$ bound for the maximal singular integral $$T_Ω^{*}f(x)=\sup\limits_{\varepsilon>0}|T_{\varepsilon,Ω}f(x)|=\sup\limits_{\varepsilon>0}\bigg|\int_{|x-y|>\varepsilon} \frac{Ω(x-y)}{|x-y|^{d}}f(y)dy\bigg|,$$ where $Ω\in L^\infty(\mathbb{S}^{d-1})$, homogeneous of degree zero, and satisfies the cancellation condition. We show that $$\|T_Ω^{*}\|_{L^1(w)\rightarrow L^{1,\infty}(w)}\lesssim[w]_{A_1}[w]_{A_\infty}\log([w]_{A_\infty}+1),$$ where $w$ belongs to the Muckenhoupt class $A_1(\mathbb{R}^d)$. This represents an essential improvement of a result (Honz\'ık, Inter.Math. Res. Not. 2020) and a result (Bhojak and Mohanty, J. Funct. Anal.2023). Our second one is the weighted weak type $(1,\,1)$ bounds for variation and jump operators corresponding to $\{T_{\varepsilon,Ω}\}_{\varepsilon\in2^{\mathbb{Z}}}$ and $\{T_{\varepsilon, Ω}^ϕ\}_{\varepsilon\in\mathbb{R}^{+}}$ under the condition $Ω\in L^\infty(\mathbb{S}^{d-1})$, where $T_{ε,Ω}^ϕ$ represents a smooth truncation of rough singular integral operator.These results of this part are the first weighted weak type $(1,\,1)$ variation inequalities and jump inequalities for rough singular integrals.

math.CA↗

arXiv2026

Qubit-Qutrit Quantum Tomography of hadronic $Λϕ$ and $ΛK^{\ast 0}$ systems

Quantum-information observables have emerged in recent years as new tools in nuclear and particle physics, from entanglement in top-quark pairs to spin correlations in $Λ\barΛ$ production. Extending these studies to unequal-spin hadronic final states poses a fundamental challenge: the $6\times6$ density matrix of a qubit-qutrit system contains 35 independent spin parameters, but the decays of $ΛV$ pairs, with $V=ϕ$ or $K^{*0}$, provide access to only 23 due to the hidden vector polarization from the strong decay. In this Letter, we formulate a qubit-qutrit quantum tomography (QQQT) technique for these spin-$\tfrac{1}{2}\otimes1$ systems and establish exact criteria for entanglement certification from the \textit{incomplete} density matrix. Compared with the $Λ\barΛ$ system, QQQT of $Λϕ$ and $ΛK^{*0}$ provides a new probe of nonperturbative QCD hadronization, enabling a direct comparison of the spin evolution of entangled quark pairs produced from the vacuum as they hadronize into a baryon or a vector meson.

hep-ph↗

arXiv2026

What You See Is Not What AI Gets: DPAgent-in-the-Middle Defense Against AI-Groomed Deceptive Patterns

Privacy deceptive patterns in web interfaces manipulate users into disclosing personal data, yet existing defenses are fragmented, static, and increasingly vulnerable to manipulation by large language models. Moreover, data voids, areas of information scarcity on the web, allow adversaries to inject misleading content that can be scraped and learned by AI systems, amplifying both deceptive design and model misbehavior. In this paper, we formalize AI grooming as a new threat in which adversaries seed benign-looking artifacts carrying machine-consumable manipulative signals into AI-mediated workflows. To address this threat, we present DPAgent, an agentic, reasoning-aware framework that orchestrates four specialized agents combining latent-space purification with defensive prompting to explore, detect, and repair privacy deceptive interfaces in live web environments. Extensive evaluations show that DPAgent filters 91\% of naive whole-page generated samples and consistently reduces attack success across five targeted grooming strategies, achieves state-of-the-art detection with a micro F1 of 0.82, explores over 80\% of pattern types while visiting only about 10\% of the pages required by baselines, and successfully repairs 89.7\% of correctly detected PDP instances. Our results demonstrate the promise of agent-in-the-middle defenses for securing the web UI supply chain against deceptive design and emerging AI threats rooted in data void exploitation.

cs.CR↗

arXiv2026

A Systematic Analysis of Automatic Differentiation versus Discretization-based Constraints for Physics-Informed PDE Solvers

Physics-informed neural networks (PINNs) represent a growing frontier in using artificial intelligence to solve partial differential equations (PDEs). Automatic differentiation (AD) plays a central role in this paradigm, which is mesh-free and replaces traditional iterative solvers with gradient-based optimization in continuous space. However, the inherent limitations of AD, particularly in handling higher-order derivatives and discontinuous solutions, pose significant challenges for complex problems. This has motivated a growing number of researchers to explore discretization-based constraints as an alternative path. Yet, the respective applicability of these two paradigms remains largely unexplored. In this work, we conduct systematic experiments across a wide spectrum of problems, from simple linear Poisson to high-Mach hypersonic flows with strong discontinuities. Through a rigorous decomposition of approximation, optimization, and truncation errors, we systematically elucidate the fundamental trade-offs and error-governing mechanisms of both paradigms, as well as two representative network architectures: multi-layer perceptron (MLP) and graph neural network (GNN). Our results reveal a consistent trend: as nonlinearity strengthens, the accuracy advantage of discretization-based constraints becomes increasingly pronounced, with smaller optimization errors compensating for the truncation errors. Moreover, the more complex the nonlinearity and boundary conditions, the greater the advantage of GNN over MLP. These insights offer a robust practical guideline for configuring neural PDE solvers in demanding engineering applications. Our source data and code are available at https://github.com/guoxing0809/neuropde_analysis.

math.NA↗

arXiv2026

Improving Generalizability and Undetectability for Targeted Adversarial Attacks on Multimodal Pre-trained Models

Multimodal pre-trained models (e.g., ImageBind), which align distinct data modalities into a shared embedding space, have shown remarkable success across downstream tasks. However, their increasing adoption raises serious security concerns, especially regarding targeted adversarial attacks. In this paper, we show that existing targeted adversarial attacks on multimodal pre-trained models still have limitations in two aspects: generalizability and undetectability. Specifically, the crafted targeted adversarial examples (AEs) exhibit limited generalization to partially known or semantically similar targets in cross-modal alignment tasks (i.e., limited generalizability) and can be easily detected by simple anomaly detection methods (i.e., limited undetectability). To address these limitations, we propose a novel method called Proxy Targeted Attack (PTA), which leverages multiple source-modal and target-modal proxies to optimize targeted AEs, ensuring they remain evasive to defenses while aligning with multiple potential targets. We also provide theoretical analyses to highlight the relationship between generalizability and undetectability and to ensure optimal generalizability while meeting the specified requirements for undetectability. Furthermore, experimental results demonstrate that our PTA can achieve a high success rate across various related targets and remain undetectable against multiple anomaly detection methods.

cs.CV↗