arXiv subjects

Nan Yu

Publications and source records attributed to Nan Yu.

At least 19 recordsLinked to original sources

arXiv2026

Inter-Residue Geometry Attention for Antibody-Specific Epitope Prediction

Antibody-specific epitope prediction aims to identify which antigen residues are recognized by a given antibody, a task that depends on the three-dimensional complementarity between antibody CDRs and the antigen surface. Existing methods usually leverage PLM embeddings and inject structure through additional graph, surface, or point-cloud encoders, where the positional mechanism inside attention remains largely tied to one-dimensional sequence order. For proteins, the analogue of a token offset is not only sequence separation, but also the three-dimensional displacement between residues after folding. This raises a question, can folded residue geometry serve as the positional mechanism of attention itself? We propose Local-Frame 3D Rotary Position Encoding (LF3DRoPE), which expresses inter-residue displacements in backbone-defined local frames and injects them directly into rotary attention. This design preserves continuous directional geometry while ensuring invariance to global $\mathrm{SE}(3)$ transformations. On the AsEP benchmark, LF3DRoPE achieves state-of-the-art $\mathrm{MCC}$ on both ratio and epitope-group splits. Ablations and rigid transformation tests show that local three-dimensional geometry provides information beyond sequence-order attention while preserving invariance to arbitrary global coordinate systems. Mutation ranking results further indicate that LF3DRoPE captures antigen-specific structural compatibility.

cs.AI↗

arXiv2026

Organizational Control Layer: Governance Infrastructure at the Execution Boundary of LLM Agent Systems

LLM-based agents are increasingly deployed in workflows where generated outputs may trigger state-changing actions, such as price offers, refunds, payments, or tool calls. This creates an execution-boundary problem: a platform must decide whether an agent's proposed action is authorized before the action is executed. We introduce the Organizational Control Layer (OCL), a model-agnostic governance layer that separates proposal generation from environment-facing execution. OCL intercepts generated actions, checks them against role, policy, and economic constraints, and either approves, revises, blocks, or escalates them without modifying the underlying LLM generator. We evaluate OCL on adversarial buyer--seller negotiation environments adapted from AgenticPay. Across multiple frontier LLM backends, OCL reduces observed unsafe executions from 88% to 0% while increasing valid success from 12% to 96%. Ablations show that this gain comes from combining pre-execution enforcement with structured recovery, rather than from prompting or blocking alone. These results suggest that deployment-grade LLM agent systems require explicit governance at the boundary between language generation and executable action.

cs.MA↗

arXiv2026

DINORANKCLIP: DINOv3 Distillation and Injection for Vision-Language Pretraining with High-Order Ranking Consistency

Contrastive language-image pretraining (CLIP) suffers from two structural weaknesses: the symmetric InfoNCE loss discards the relative ordering among unmatched in-batch pairs, and global pooling collapses the visual representation into a semantic bottleneck that is poorly sensitive to fine-grained local structure. RANKCLIP partially addresses the first issue with a list-wise Plackett-Luce ranking-consistency loss, but its model is strictly first-order and inherits the second weakness untouched. We propose DINORANKCLIP, a pretraining framework that addresses both jointly. Our principal contribution is injecting a frozen DINOv3 teacher into the contrastive trunk through a dual-branch lightweight student and a multi-scale fusion module with channel-spatial attention, a self-attention refiner, and a conflict-aware gate that preserves the cross-modal alignment up to first order. Complementarily, we introduce a high-order Plackett-Luce ranking model in which the per-position utility is augmented with attention-parameterised pairwise and tuple-wise transition terms; the family contains CLIP and RANKCLIP as nested zero-order and first-order special cases, and the optimal order on every benchmark is $R^*=3$. The full empirical study -- order sweep, Fine-grained Probe on five datasets, four-node Modality-Gap analysis, six-variant Fusion ablation -- fits in 72 hours on a single eight-GPU H100 node and trains entirely on Conceptual Captions 3M. DINORANKCLIP consistently outperforms CLIP, CyCLIP, ALIP, and RANKCLIP under matched compute, with the largest relative gains on the fine-grained and out-of-distribution evaluations that most directly stress local structural reasoning.

cs.CV↗

arXiv2025

HiEAG: Evidence-Augmented Generation for Out-of-Context Misinformation Detection

Recent advancements in multimodal out-of-context (OOC) misinformation detection have made remarkable progress in checking the consistencies between different modalities for supporting or refuting image-text pairs. However, existing OOC misinformation detection methods tend to emphasize the role of internal consistency, ignoring the significant of external consistency between image-text pairs and external evidence. In this paper, we propose HiEAG, a novel Hierarchical Evidence-Augmented Generation framework to refine external consistency checking through leveraging the extensive knowledge of multimodal large language models (MLLMs). Our approach decomposes external consistency checking into a comprehensive engine pipeline, which integrates reranking and rewriting, apart from retrieval. Evidence reranking module utilizes Automatic Evidence Selection Prompting (AESP) that acquires the relevant evidence item from the products of evidence retrieval. Subsequently, evidence rewriting module leverages Automatic Evidence Generation Prompting (AEGP) to improve task adaptation on MLLM-based OOC misinformation detectors. Furthermore, our approach enables explanation for judgment, and achieves impressive performance with instruction tuning. Experimental results on different benchmark datasets demonstrate that our proposed HiEAG surpasses previous state-of-the-art (SOTA) methods in the accuracy over all samples.

cs.CL↗

arXiv2025

SIADAFIX: issue description response for adaptive program repair

We propose utilizing fast and slow thinking to enhance the capabilities of large language model-based agents on complex tasks such as program repair. In particular, we design an adaptive program repair method based on issue description response, called SIADAFIX. The proposed method utilizes slow thinking bug fix agent to complete complex program repair tasks, and employs fast thinking workflow decision components to optimize and classify issue descriptions, using issue description response results to guide the orchestration of bug fix agent workflows. SIADAFIX adaptively selects three repair modes, i.e., easy, middle and hard mode, based on problem complexity. It employs fast generalization for simple problems and test-time scaling techniques for complex problems. Experimental results on the SWE-bench Lite show that the proposed method achieves 60.67% pass@1 performance using the Claude-4 Sonnet model, reaching state-of-the-art levels among all open-source methods. SIADAFIX effectively balances repair efficiency and accuracy, providing new insights for automated program repair. Our code is available at https://github.com/liauto-siada/siada-cli.

cs.SE↗

arXiv2025

ProtTeX-CC: Activating In-Context Learning in Protein LLM via Two-Stage Instruction Compression

Recent advances in protein large language models, such as ProtTeX, represent both side-chain amino acids and backbone structure as discrete token sequences of residue length. While this design enables unified modeling of multimodal protein information, it suffers from two major limitations: (1) The concatenation of sequence and structure tokens approximately doubles the protein length and breaks the intrinsic residue-level alignment between modalities. (2) Constrained by the training corpus and limited context window, ProtTeX is typically trained on single-protein inputs, rendering it incompatible with in-context learning (ICL) and thus limiting its generalization capability. To address these issues, we propose ProtTeX-CC, a lightweight two-stage compression framework designed to enhance ProtTeX under few-shot settings. We first design a joint embedding compression mechanism that fuses sequence and structure representations at the residue level, effectively reducing the protein input length by half without sacrificing performance. Then we propose a self-compression module that aggregates each full demonstration into the latent space of the last few linguistic tokens, reducing the average demonstration length from 751 tokens to less than 16 tokens. Compared to the original ProtTeX, our self-compression approach achieves a compression ratio of approximately 93.68% in the total prompt length under the 16-shot setting. Without modifying the backbone model, ProtTeX-CC introduces only a small number of additional parameters through PEFT-based tuning in the joint embedding compression stage and a single trainable projection layer in the self-compression stage. Extensive experiments on protein function prediction show that ProtTeX-CC improves performance on the in-domain benchmark by 2%, and generalizes well to the out-of-domain dataset with a performance gain of 11%.

cs.LG↗

arXiv2025

Dark Energy Search by Atom Interferometry in the Einstein-Elevator

The DESIRE project aims to test chameleon field theories as potential candidates for dark energy. The chameleon field is a light scalar field that is subject to screening mechanisms in dense environments making them hardly detectable. The project is designed to overcome this challenge. To this end, a specially designed source mass generates periodic gravitational and chameleon potentials. The design of the source mass allows for adjustment of the amplitude and periodicity of the gravitational potential while keeping the chameleon potential unchanged. The periodicity of the potentials makes them distinguishable from the environment and allows for resonant detection using multiloop atom interferometry under microgravity conditions.

physics.atom-ph↗

arXiv2025

ChatMol: A Versatile Molecule Designer Based on the Numerically Enhanced Large Language Model

Goal-oriented de novo molecule design, namely generating molecules with specific property or substructure constraints, is a crucial yet challenging task in drug discovery. Existing methods, such as Bayesian optimization and reinforcement learning, often require training multiple property predictors and struggle to incorporate substructure constraints. Inspired by the success of Large Language Models (LLMs) in text generation, we propose ChatMol, a novel approach that leverages LLMs for molecule design across diverse constraint settings. Initially, we crafted a molecule representation compatible with LLMs and validated its efficacy across multiple online LLMs. Afterwards, we developed specific prompts geared towards diverse constrained molecule generation tasks to further fine-tune current LLMs while integrating feedback learning derived from property prediction. Finally, to address the limitations of LLMs in numerical recognition, we referred to the position encoding method and incorporated additional encoding for numerical values within the prompt. Experimental results across single-property, substructure-property, and multi-property constrained tasks demonstrate that ChatMol consistently outperforms state-of-the-art baselines, including VAE and RL-based methods. Notably, in multi-objective binding affinity maximization task, ChatMol achieves a significantly lower KD value of 0.25 for the protein target ESR1, while maintaining the highest overall performance, surpassing previous methods by 4.76%. Meanwhile, with numerical enhancement, the Pearson correlation coefficient between the instructed property values and those of the generated molecules increased by up to 0.49. These findings highlight the potential of LLMs as a versatile framework for molecule generation, offering a promising alternative to traditional latent space and RL-based approaches.

cs.CE↗

arXiv2025

E2LVLM:Evidence-Enhanced Large Vision-Language Model for Multimodal Out-of-Context Misinformation Detection

Recent studies in Large Vision-Language Models (LVLMs) have demonstrated impressive advancements in multimodal Out-of-Context (OOC) misinformation detection, discerning whether an authentic image is wrongly used in a claim. Despite their success, the textual evidence of authentic images retrieved from the inverse search is directly transmitted to LVLMs, leading to inaccurate or false information in the decision-making phase. To this end, we present E2LVLM, a novel evidence-enhanced large vision-language model by adapting textual evidence in two levels. First, motivated by the fact that textual evidence provided by external tools struggles to align with LVLMs inputs, we devise a reranking and rewriting strategy for generating coherent and contextually attuned content, thereby driving the aligned and effective behavior of LVLMs pertinent to authentic images. Second, to address the scarcity of news domain datasets with both judgment and explanation, we generate a novel OOC multimodal instruction-following dataset by prompting LVLMs with informative content to acquire plausible explanations. Further, we develop a multimodal instruction-tuning strategy with convincing explanations for beyond detection. This scheme contributes to E2LVLM for multimodal OOC misinformation detection and explanation. A multitude of experiments demonstrate that E2LVLM achieves superior performance than state-of-the-art methods, and also provides compelling rationales for judgments.

cs.LG↗

arXiv2024

Additive manufacturing in ceramics: targeting lightweight mirror applications in the visible, ultraviolet and X-ray

Additive manufacturing (AM; 3D printing) has clear benefits in the production of lightweight mirrors for astronomy: it can create optimised lightweight structures and combine multiple components into one. New capabilities in AM ceramics, silicon carbide infiltrated with silicon and fused silica, offer the possibility to combine the design benefits of AM with a material suitable for visible, ultraviolet and X-ray applications. This paper will introduce the printing methods and post-processing steps to convert AM ceramic samples into reflective mirrors. Surface roughness measurements after abrasive polishing of the AM ceramics will be presented.

astro-ph.IM↗

arXiv2024

Searching for new physics in the solar system with tetrahedral spacecraft formations

Tetrahedral configurations of spacecraft on unperturbed heliocentric orbits allow for highly precise observations of small spatial changes in the gravitational field, especially those affecting the gravity gradient tensor (GGT). The resulting high sensitivity may be used to search for new physics that could manifest itself via deviations from general relativistic behavior yielding a non-vanishing trace[GGT]. We study the feasibility of recovering the trace[GGT] with the sensitivity of O(1e-24 s^(-2)) -- the level where some of the recently proposed cosmological models may have observable effects in the solar system. We consider how local measurements provided by precision laser ranging and atom-wave interferometry can be used for that purpose. We report on a preliminary study of such an experiment and precision that may be reached in measuring the trace[GGT], with the assumption of drag-compensated spacecraft by atom interferometer measurements. For that, we study the dynamical behavior of a tetrahedral formation established by four spacecraft on heliocentric nearby elliptical orbits. We formulate the observational equations to measure the trace[GGT] relying only on the observables available within the formation: laser ranging and the Sagnac interferometry. We demonstrate that Sagnac observable is a mission enabling and allows to measure the angular frequency of the tetrahedral rotation with respect to an inertial reference frame with an accuracy much higher than that available from any other modern navigational techniques. We show that the quality of the science measurements is affected by the changes in tetrahedron's orientation and shape as spacecraft follow their orbits. We present the preliminary mission and instrument requirements needed to measure the trace[GGT] to the required accuracy and demonstrate the feasibility of satisfying the science objectives.

gr-qc↗

arXiv2023

NASA's Cold Atom Laboratory: Five Years of Quantum Science Operations in Space

NASA's Cold Atom Laboratory (CAL) is a multi-user science facility for studying quantum gases in the microgravity environment of the International Space Station. The persistent microgravity environment of the ISS enables research with ultracold atoms in a temperature regime and force-free environment inaccessible to terrestrial laboratories, unlocking the potential to observe novel quantum phenomena. CAL launched to the ISS in May 2018, and has operated continuously since then as the worlds first multi-user quantum science facility in space. CAL is the first experimental facility to produce the fifth state of matter known as a Bose-Einstein condensate with ultracold rubidium atoms on orbit and, more recently, with mixtures of rubidium and potassium. We present an overview of CAL's design and operation, review the scientific contributions to date, and discuss recent on-orbit upgrades to extend its useful mission lifetime and provide enhanced science. We also consider opportunities for follow-on missions informed by lessons learned from over five years of operation on orbit.

quant-ph↗

arXiv2022

STE-QUEST -- Space Time Explorer and QUantum Equivalence principle Space Test: The 2022 medium-class mission concept

Space-borne quantum technologies, particularly those based on atom interferometry, are heralding a new era of strategic and robust space exploration. The unique conditions of space, characterized by low noise and low gravity environments, open up diverse possibilities for applications ranging from precise time and frequency transfer to Earth Observation and the search of new Physics. In this paper, we summarise the M-class mission proposal in response to the 2022 call in ESA's science program: Space-Time Explorer and Quantum Equivalence Principle Space Test (STE-QUEST). It consists in a satellite mission featuring a dual-species atom interferometer operating over extended durations. This mission aims to tackle three of the most fundamental questions in Physics: (i) testing the universality of free fall with an accuracy better than one part in $10^{-17}$, (ii) exploring various forms of Ultra-Light Dark Matter, and (iii) scrutinizing the foundations of Quantum Mechanics.

physics.space-ph↗

arXiv2022

Discourse-Aware Emotion Cause Extraction in Conversations

Emotion Cause Extraction in Conversations (ECEC) aims to extract the utterances which contain the emotional cause in conversations. Most prior research focuses on modelling conversational contexts with sequential encoding, ignoring the informative interactions between utterances and conversational-specific features for ECEC. In this paper, we investigate the importance of discourse structures in handling utterance interactions and conversationspecific features for ECEC. To this end, we propose a discourse-aware model (DAM) for this task. Concretely, we jointly model ECEC with discourse parsing using a multi-task learning (MTL) framework and explicitly encode discourse structures via gated graph neural network (gated GNN), integrating rich utterance interaction information to our model. In addition, we use gated GNN to further enhance our ECEC model with conversation-specific features. Results on the benchmark corpus show that DAM outperform the state-of-theart (SOTA) systems in the literature. This suggests that the discourse structure may contain a potential link between emotional utterances and their corresponding cause expressions. It also verifies the effectiveness of conversationalspecific features. The codes of this paper will be available on GitHub.

cs.CL↗

arXiv2021

Fundamental Physics with a State-of-the-Art Optical Clock in Space

Recent advances in optical atomic clocks and optical time transfer have enabled new possibilities in precision metrology for both tests of fundamental physics and timing applications. Here we describe a space mission concept that would place a state-of-the-art optical atomic clock in an eccentric orbit around Earth. A high stability laser link would connect the relative time, range, and velocity of the orbiting spacecraft to earthbound stations. The primary goal for this mission would be to test the gravitational redshift, a classical test of general relativity, with a sensitivity 30,000 times beyond current limits. Additional science objectives include other tests of relativity, enhanced searches for dark matter and drifts in fundamental constants, and establishing a high accuracy international time/geodesic reference.

gr-qc↗

arXiv2021

The Deep Space Quantum Link: Prospective Fundamental Physics Experiments using Long-Baseline Quantum Optics

The National Aeronautics and Space Administration's Deep Space Quantum Link mission concept enables a unique set of science experiments by establishing robust quantum optical links across extremely long baselines. Potential mission configurations include establishing a quantum link between the Lunar Gateway moon-orbiting space station and nodes on or near the Earth. In this publication, we summarize the principal experimental goals of the Deep Space Quantum Link mission. These include long-range teleportation, tests of gravitational coupling to quantum states, and advanced tests of quantum nonlocality.

quant-ph↗

arXiv2021

Discontinuous Named Entity Recognition as Maximal Clique Discovery

Named entity recognition (NER) remains challenging when entity mentions can be discontinuous. Existing methods break the recognition process into several sequential steps. In training, they predict conditioned on the golden intermediate results, while at inference relying on the model output of the previous steps, which introduces exposure bias. To solve this problem, we first construct a segment graph for each sentence, in which each node denotes a segment (a continuous entity on its own, or a part of discontinuous entities), and an edge links two nodes that belong to the same entity. The nodes and edges can be generated respectively in one stage with a grid tagging scheme and learned jointly using a novel architecture named Mac. Then discontinuous NER can be reformulated as a non-parametric process of discovering maximal cliques in the graph and concatenating the spans in each clique. Experiments on three benchmarks show that our method outperforms the state-of-the-art (SOTA) results, with up to 3.5 percentage points improvement on F1, and achieves 5x speedup over the SOTA model.

cs.CL↗

arXiv2020

Generating 500 mW for laser cooling of strontium atoms by injection locking a high power laser diode

We report on the generation of 500 mW of spectrally pure laser light at the 460.86 nm transition used for laser cooling of strontium atoms. To this end we inject a high power single mode laser diode with light from a stabilized extended cavity diode laser. To optimize and monitor the injection status and the spectral purity of the slave diode we developed a novel technique that uses a single passive optical element. A narrow band interference filter generates a suitable monitoring signal without any additional electronics for post processing of the data. Our method greatly simplifies the daily operation of injection locked laser diodes and can be easily adapted to other wavelengths of interest.

physics.atom-ph↗