arXiv subjects

Peng Yan

Publications and source records attributed to Peng Yan.

At least 37 records · Page 2Linked to original sources

arXiv2025

Curvature-Induced Magnon Frequency Combs

Generating magnon frequency combs (MFCs) with tunable spacing via a single-frequency driving is crucial for practical applications but it typically relies on complex spin textures like skyrmions or vortices. Here, we theoretically and numerically demonstrate MFC generation in geometrically curved ferromagnetic thin films using single-frequency microwave excitation, without topological spin textures. We first show that the curvature transforms the planar ferromagnetic resonance into a localized, redshifted magnon bound state, which, under non-resonant driving, activates sequential three-magnon scattering processes assisted by the curvature-driven effective anisotropy and Dzyaloshinskii-Moriya interaction. It finally produces equally spaced, robust frequency combs with spacing exactly set by the bound mode frequency. Moreover, we find that the curvature gradient at the hybrid interface mimics an analog event horizon, with the bound state's redshift resembling gravitational effects in black hole physics. Micromagnetic simulations confirm these curvature-driven nonlinear phenomenon, unveiling a novel geometric strategy for controlling magnon interactions and advancing compact magnonic devices.

cond-mat.other↗

arXiv2025

Multi-Aspect Cross-modal Quantization for Generative Recommendation

Generative Recommendation (GR) has emerged as a new paradigm in recommender systems. This approach relies on quantized representations to discretize item features, modeling users' historical interactions as sequences of discrete tokens. Based on these tokenized sequences, GR predicts the next item by employing next-token prediction methods. The challenges of GR lie in constructing high-quality semantic identifiers (IDs) that are hierarchically organized, minimally conflicting, and conducive to effective generative model training. However, current approaches remain limited in their ability to harness multimodal information and to capture the deep and intricate interactions among diverse modalities, both of which are essential for learning high-quality semantic IDs and for effectively training GR models. To address this, we propose Multi-Aspect Cross-modal quantization for generative Recommendation (MACRec), which introduces multimodal information and incorporates it into both semantic ID learning and generative model training from different aspects. Specifically, we first introduce cross-modal quantization during the ID learning process, which effectively reduces conflict rates and thus improves codebook usability through the complementary integration of multimodal information. In addition, to further enhance the generative ability of our GR model, we incorporate multi-aspect cross-modal alignments, including the implicit and explicit alignments. Finally, we conduct extensive experiments on three well-known recommendation datasets to demonstrate the effectiveness of our proposed method.

cs.IR↗

arXiv2025

Observation of Magnetostatic Surface Spin Wave Solitons in Yttrium Iron Garnet Thin Film

Magnetostatic surface spin wave (MSSW) solitons hold great promise for magnonic information processing, but their existence has long been debated. In this work, we resolve this issue by advanced time-resolved Brillouin light scattering (TR-BLS) spectroscopy. We observe long-period MSSW soliton trains in yttrium iron garnet (YIG) thin films by demonstrating their quasiparticle behavior, mode beating, and periodic modulations. We reveal that the MSSW soliton originates from the dipole gap mechanism with a unique comb-like frequency spectrum caused by the four-magnon process. By varying the microwave power, we find significant changes in soliton periods linked to the spin wave dispersion renormalization. Additionally, we report exotic transverse solitons across the YIG thickness. These findings deepen the understanding of nonlinear physics, and pave the way for spin-wave-soliton-based information technologies.

cond-mat.mes-hall↗

arXiv2025

What Makes Reasoning Invalid: Echo Reflection Mitigation for Large Language Models

Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of reasoning tasks. Recent methods have further improved LLM performance in complex mathematical reasoning. However, when extending these methods beyond the domain of mathematical reasoning to tasks involving complex domain-specific knowledge, we observe a consistent failure of LLMs to generate novel insights during the reflection stage. Instead of conducting genuine cognitive refinement, the model tends to mechanically reiterate earlier reasoning steps without introducing new information or perspectives, a phenomenon referred to as "Echo Reflection". We attribute this behavior to two key defects: (1) Uncontrollable information flow during response generation, which allows premature intermediate thoughts to propagate unchecked and distort final decisions; (2) Insufficient exploration of internal knowledge during reflection, leading to repeating earlier findings rather than generating new cognitive insights. Building on these findings, we proposed a novel reinforcement learning method termed Adaptive Entropy Policy Optimization (AEPO). Specifically, the AEPO framework consists of two major components: (1) Reflection-aware Information Filtration, which quantifies the cognitive information flow and prevents the final answer from being affected by earlier bad cognitive information; (2) Adaptive-Entropy Optimization, which dynamically balances exploration and exploitation across different reasoning stages, promoting both reflective diversity and answer correctness. Extensive experiments demonstrate that AEPO consistently achieves state-of-the-art performance over mainstream reinforcement learning baselines across diverse benchmarks.

cs.AI↗

arXiv2025

LinkedIn Post Embeddings: Industrial Scale Embedding Generation and Usage across LinkedIn

A post embedding (representation of text in embedding space that effectively captures semantic meaning) is a foundational component of LinkedIn that is consumed by product surfaces in retrieval and ranking (e.g., ranking posts in the feed or video tab). This paper presents the post embeddings used at LinkedIn, where a pre-trained transformer-based large language model (LLM) is taken as input and fine-tuned using multi-task learning across a diverse set of semantic labeling tasks. We observe positive transfer, leading to improved performance across all tasks, compared to training them independently. The generated post embeddings outperform baseline models in zero-shot learning, demonstrating its potential for broader applicability. Furthermore, the generated post embeddings' performance surpasses that of OpenAI's ADA-001 and ADA-002 embeddings on LinkedIn specific datasets and tasks. We also describe the offline evaluation methodology and the deployment to our near-line infrastructure, which makes the post embedding available for use within minutes of post creation for any downstream application. We present how the embeddings were applied in the Feed product surface, in both ranking and retrieval stages, and showcase the real world online impact to demonstrate the superior performance of these embeddings. Finally, we also share the results of applying the embeddings to the retrieval system of our video ranking product surface in LinkedIn. These embeddings have been battle-tested in production at LinkedIn for over two years, consistently powering multiple products.

cs.LG↗

arXiv2025

Topological Magnon Frequency Combs

Exploring the synergy between topological physics and nonlinear dynamics unveils profound insights into emergent states of matter. Inspired by recent experimental demonstrations of topological frequency combs in photonics, we theoretically introduce topological magnon frequency combs (MFCs) in a two-dimensional triangular skyrmion lattice. Computing the Chern numbers of magnon bands reveals robust chiral edge states. Strikingly, these topological MFCs originate from nonlinear four-magnon scattering among the chiral edge modes, activated by dual-frequency driving without an amplitude threshold. Comb spacings are readily tunable through excitation frequency detuning. Micromagnetic simulations validate our predictions with good concordance. This work paves the way for defect-immune magnonic devices exploiting MFCs and sparks investigations into topological-nonlinear phenomena in magnetic systems.

cond-mat.mes-hall↗

arXiv2025

Towards Efficient Verification of Computation in Quantum Devices

Designing quantum processors is a complex task that demands advanced verification methods to ensure their correct functionality. However, traditional methods of comprehensively verifying quantum devices, such as quantum process tomography, face significant limitations because of the exponential growth in computational resources. These limitations arise from treating the system as a black box and ignoring its design structure. Consequently, new testing methods must be developed considering the design structure. In this paper, we investigate the structure of computations on the hardware, focusing on the layered interruptible quantum circuit model and designing a scalable algorithm to verify it comprehensively. Specifically, for a given quantum hardware that claims to process an unknown $n$ qubit $d$ layer circuit via a finite set of quantum gates, our method completely reconstructs the circuits within a time complexity of $O(d^2 t \log (n/δ))$, guaranteeing success with a probability of at least $1-δ$. Here, $t$ represents the maximum execution time for each circuit layer. Our approach significantly reduces execution time for completely verifying computations in quantum devices, achieving double logarithmic scaling in the problem size. Furthermore, we validate our algorithm through experiments using IBM's quantum cloud service, demonstrating its potential applicability in the noisy intermediate-scale quantum era.

quant-ph↗

arXiv2025

Solution for Meta KDD Cup'25: A Comprehensive Three-Step Framework for Vision Question Answering

Vision Large Language Models (VLLMs) have improved multi-modal understanding and visual question answering (VQA), but still suffer from hallucinated answers. Multi-modal Retrieval-Augmented Generation (RAG) helps address these issues by incorporating external information, yet challenges remain in visual context comprehension, multi-source retrieval, and multi-turn interactions. To address these challenges, Meta constructed the CRAG-MM benchmark and launched the CRAG-MM Challenge at KDD Cup 2025, which consists of three tasks. This paper describes the solutions of all tasks in Meta KDD Cup'25 from BlackPearl team. We use a single model for each task, with key methods including data augmentation, RAG, reranking, and multi-task fine-tuning. Our solution achieve automatic evaluation rankings of 3rd, 3rd, and 1st on the three tasks, and win second place in Task3 after human evaluation.

cs.IR↗

arXiv2025

Interaction-Driven Altermagnetic Magnon Chiral Splitting

Nonrelativistic magnon chiral splitting in altermagnets has garnered significant recent attention. In this work, we demonstrate that nonlinear three-wave mixing -- where magnons split or coalesce -- extends this phenomenon into unprecedented relativistic regimes. Employing a bilayer antiferromagnet with Dzyaloshinskii-Moriya interactions, we identify three distinct classes of chiral splitting, each dictated by specific symmetries, such as $C_4T$, $σ_v T$, or their combination. This reveals a novel bosonic mechanism for symmetry-protected chiral splitting, capitalizing on the unique ability of magnons to violate particle-number conservation, a feature absent in low-energy fermionic systems. Our findings pave the way for engineering altermagnetic splitting, with potential applications in advanced magnonic devices and deeper insights into magnon dynamics in complex magnetic systems.

cond-mat.mtrl-sci↗

arXiv2025

Frequency comb in twisted magnonic crystals

While twisted magnonic crystals (MCs) have recently gained attention for their intriguing linear phenomena, such as magnon flat bands, their nonlinear dynamics -- particularly the generation of magnonic frequency combs (MFCs) -- have remained largely unexplored. In this work, we demonstrate the creation of MFCs in twisted MCs using two-tone microwave excitation. We find that finite twist angles significantly enhance three-magnon interactions, driven by the non-collinear ground-state magnetic configuration induced by interlayer dipole-dipole interactions. The number of comb teeth exhibits a plateau-like dependence on the twist angle, with the plateau's width and height saturating as the excitation frequency of the propagating magnon mode increases. This behavior reveals an optimal range of twist angles and frequencies for achieving high-quality MFCs with a large number of comb teeth. Our findings deepen the understanding of nonlinear interactions in twisted MCs and highlight their potential for advancing moiré-based materials in information processing and high-precision metrology.

cond-mat.mtrl-sci↗

arXiv2025

Wave-propagation Based Analysis of the Magnetostatic Waves in Ferrite Films Excited by Metallic Transducers

It is conventional wisdom that the spectra of the impedances of magnetostatic waves (MSWs) determine the transmissions of MSW devices. In this work, we show that the characteristics of propagating MSWs have critical impacts on the characteristics of transmissions. A wave-propagation based analysis considering the inhomogeneous distributions of magnetic fields is presented for investigating the propagations of MSWs. Based on the analysis, it is demonstrated that the metallic nature of transducers causes the high insertion losses in high-frequency bands, while the dips and severe in-band ripples in low-frequency bands are resulted from the complicated interference between the multiple width modes. Simulations in HFSS verify the analysis with good agreements. Our work advances the understanding of MSWs propagating in ferrite films with metallic structures and paves the way to designing MSW devices aimed at implantation in microwave systems.

physics.app-ph↗

arXiv2025

Magnon-Driven Phononic Frequency Comb in Linear Elastic Media

Phononic frequency combs (PFCs) typically require nonlinear elastic media, limiting their frequency range and stability. Here, we propose a transformative approach to generate PFCs in purely linear elastic media by harnessing the magnon nonlinearities, offering a new paradigm for frequency comb physics. By tuning the magnon-phonon coupling confined in a magnetic disk of a vortex state into the strong coupling regime, we demonstrate an efficient nonlinearity transfer from magnons to phonons. This mechanism is able to produce GHz-range PFCs with comb spacing set by the vortex core's gyration frequency. Full micromagnetic simulations verify our theoretical predictions, confirming robust comb formation at 3.5 GHz with 0.4 GHz spacing. This approach overcomes the sub-MHz constraints of conventional PFCs, enabling applications in high-precision metrology, nanoscale sensing, and quantum technologies. Our findings also deepen the understanding of the nonlinear dynamics in hybrid magnon-phonon systems and provide a versatile platform for exploring frequency combs in diverse physical systems.

cond-mat.mes-hall↗

arXiv2025

Strong coupling of chiral magnons in altermagnets

Altermagnets recently are identified as a new class of magnets that break the time-reversal symmetry without exhibiting net magnetization. The role of the dipole-dipole interaction (DDI) on their dynamical properties however is yet to be addressed. In this work, we show that the DDI can induce the strong coupling between exchange magnons with opposite chiralities in altermagnets, manifesting as a significant level repulsion in the magnon spectrum. Crucially, the predicted magnon-magnon coupling is highly anisotropic, and observable in practical experiments. These exotic features are absent in conventional antiferromagnets. Our findings open a new pathway for quantum magnonic information processing based on altermagnetism.

cond-mat.mtrl-sci↗

arXiv2025

Learning Actionable World Models for Industrial Process Control

To go from (passive) process monitoring to active process control, an effective AI system must learn about the behavior of the complex system from very limited training data, forming an ad-hoc digital twin with respect to process inputs and outputs that captures the consequences of actions on the process's world. We propose a novel methodology based on learning world models that disentangles process parameters in the learned latent representation, allowing for fine-grained control. Representation learning is driven by the latent factors influencing the processes through contrastive learning within a joint embedding predictive architecture. This makes changes in representations predictable from changes in inputs and vice versa, facilitating interpretability of key factors responsible for process variations, paving the way for effective control actions to keep the process within operational bounds. The effectiveness of our method is validated on the example of plastic injection molding, demonstrating practical relevance in proposing specific control actions for a notoriously unstable process.

cs.LG↗

arXiv2025

Personalized Image Generation with Large Multimodal Models

Personalized content filtering, such as recommender systems, has become a critical infrastructure to alleviate information overload. However, these systems merely filter existing content and are constrained by its limited diversity, making it difficult to meet users' varied content needs. To address this limitation, personalized content generation has emerged as a promising direction with broad applications. Nevertheless, most existing research focuses on personalized text generation, with relatively little attention given to personalized image generation. The limited work in personalized image generation faces challenges in accurately capturing users' visual preferences and needs from noisy user-interacted images and complex multimodal instructions. Worse still, there is a lack of supervised data for training personalized image generation models. To overcome the challenges, we propose a Personalized Image Generation Framework named Pigeon, which adopts exceptional large multimodal models with three dedicated modules to capture users' visual preferences and needs from noisy user history and multimodal instructions. To alleviate the data scarcity, we introduce a two-stage preference alignment scheme, comprising masked preference reconstruction and pairwise preference alignment, to align Pigeon with the personalized image generation task. We apply Pigeon to personalized sticker and movie poster generation, where extensive quantitative results and human evaluation highlight its superiority over various generative baselines.

cs.IR↗

arXiv2024

Quantum entanglement and Einstein-Podolsky-Rosen steering in magnon frequency comb

Significant progress has been made for the emerging concept of magnon frequency comb (MFC) but mainly in the classical region. The quantum property of the comb structure is yet to be explored. Here we theoretically investigate the quantum fluctuations of frequency combs and demonstrate the continuous-variable quantum entanglement and Einstein-Podolsky-Rosen (EPR) steering between different teeth of MFC. Without loss of generality, we address this issue in a hybrid magnon-skyrmion system. We observe a strong two-mode squeezed entanglement and asymmetric steering between the sum- and difference-frequency magnon teeth mediated by the skyrmion that acts as an effective reservoir to cool the Bogoliubov mode delocalized over the first-order magnon pair in MFC. Our findings show the prominent quantum nature of MFC, which has the potential to be utilized in ultrafast quantum metrology and multi-task quantum information processing.

quant-ph↗

arXiv2024

Simulating Thiele's Equation and Collective Skyrmion Dynamics in Circuit Networks

Developed half a century ago, Thiele's equation has played a crucial role in describing the motion of magnetic textures, ranging from simple magnetic domains to exotic magnetic solitons like skyrmions and hopfions. However, it remains a challenge to fully understand the collective dynamics of magnetic texture crystals, due to the complex many-body interactions. On the other hand, electrical circuits have recently been proved to be a powerful platform to realize rich physical phenomena in solids. In this Letter, we first construct a circuit unit to simulate Thiele's equation, which enables us to experimentally observe the "skyrmion Hall effect". By coupling a pair of such circuit units, we find an emerging frequency comb under strong pumpings, a long-sought nonlinear effect in interacting skyrmion systems. By extending our strategy to circuit arrays, we experimentally observe the topological edge state, thus confirming the theoretical prediction of topological solitonic insulators years ago. Our work builds a faithful connection between electric circuits and magnetic solitons, two seemingly unrelated areas, and opens the door for exploring collective magnetization dynamics in circuit networks.

cond-mat.mes-hall↗

arXiv2024

Calibrating the Confidence of Large Language Models by Eliciting Fidelity

Large language models optimized with techniques like RLHF have achieved good alignment in being helpful and harmless. However, post-alignment, these language models often exhibit overconfidence, where the expressed confidence does not accurately calibrate with their correctness rate. In this paper, we decompose the language model confidence into the \textit{Uncertainty} about the question and the \textit{Fidelity} to the answer generated by language models. Then, we propose a plug-and-play method to estimate the confidence of language models. Our method has shown good calibration performance by conducting experiments with 6 RLHF-LMs on four MCQA datasets. Moreover, we propose two novel metrics, IPR and CE, to evaluate the calibration of the model, and we have conducted a detailed discussion on \textit{Truly Well-Calibrated Confidence}. Our method could serve as a strong baseline, and we hope that this work will provide some insights into the model confidence calibration.

cs.CL↗