arXiv Science⌕ Search

arXiv subjects

Shifeng Wen

Publications and source records attributed to Shifeng Wen.

3 recordsLinked to original sources

Multichannel Phonon Interconversion Drives 90 Polarization Rotation for Stable Domain Walls in Hafnium Oxide

Polarization rotation is a fundamental mechanism in ferroelectric materials, yet how such rotation arises microscopically in fluorite ferroelectrics and gives rise to distinct domain walls (DW) remains unresolved. Phonon modes offer a microscopic route to such structural transformations, as geometric orientations can modify their frequencies, dispersion relations, and vibrational eigenstates when their characteristic wavelengths become comparable to relevant physical dimensions. Here, by combining condensed phonon mode decomposition with first-principles calculations, we investigate the origin of 90 polarization rotation in the ferroelectric phase of HfO2. Using the cubic parent phase as reference, we show that polarization rotation originates from multiple symmetry equivalent phonon modes of distinct eigenstates at different wavevectors in the Brillouin zone, whose coupling and cooperative reorientation continuously evolve to generate distinct 90 Pca21 variants and switching pathways. Among the 90 variants, the lowest-energy switching pathway through the Pmn21 exhibits an energy barrier comparable to, but slightly lower than, that of the 180 polarization reversal pathway through the P42/nmc phase, indicating that 90 polarization rotation is an energetically accessible alternative switching route. We further analyze 90 DW and find that frustrated phonon modes emerge at the wall center, where the coexistence and competition of condensed modes promote local reconfiguration of the lattice distortions, contributing to the elevated DW energies. Importantly, experimental observations with STEM imaging provide direct evidence of atomic mode distortions in stable 90 DW, consistent with the theoretically identified low-energy DW. These findings provide the fundamental structural mechanism underlying 90 switching dynamics, competing domain-wall physics, and macroscopic ferroelectric behavior.

cond-mat.mtrl-sci↗

SIREN: Unified Multi-Granularity Semantic Interaction for Multi-Modal Lifelong User Interest Modeling

Industrial recommender systems increasingly leverage lifelong user behavior histories and rich multi-modal content to capture evolving user preferences. However, effectively integrating multi-modal features into lifelong interest modeling remains challenging due to the inherent misalignment between multi-modal and collaborative spaces. Existing paradigms typically rely on separate modeling of multi-modal sequence and behavior sequence, and late fusion to alleviate the modality gap, which results in coarse-grained multi-modal representation and limited integration. In this paper, we propose SIREN, a unified multi-granularity semantic interaction framework for multi-modal lifelong user interest modeling. In the General Search Unit stage, we introduce two alternative retrieval strategies: multi-modal similarity-based soft retrieval for retrieval effectiveness, and Semantic ID (SemID)-based hard retrieval for efficient industrial serving. For the Exact Search Unit stage, we explicitly incorporate target-aware relevance via coarse similarity buckets and fine-grained prefix-encoded SemIDs, enabling unified interaction with collaborative ID features within the target-conditioned transformer architecture. Extensive experiments on the offline dataset demonstrate that SIREN achieves a state-of-the-art GAUC. Online A/B tests further demonstrate consistent GMV gains across multiple production scenarios, including +2.28% in Weixin Moments, +3.87% in Weixin Official Accounts, and +1.61% in Weixin Channels. From July 2025, SIREN has been fully launched for full-traffic serving in Tencent's advertising platform.

cs.IR↗

Crocodile: Cross Experts Covariance for Disentangled Learning in Multi-Domain Recommendation

Multi-domain learning (MDL) has become a prominent topic in enhancing the quality of personalized services. It's critical to learn commonalities between domains and preserve the distinct characteristics of each domain. However, this leads to a challenging dilemma in MDL. On the one hand, a model needs to leverage domain-aware modules such as experts or embeddings to preserve each domain's distinctiveness. On the other hand, real-world datasets often exhibit long-tailed distributions across domains, where some domains may lack sufficient samples to effectively train their specific modules. Unfortunately, nearly all existing work falls short of resolving this dilemma. To this end, we propose a novel Cross-experts Covariance Loss for Disentangled Learning model (Crocodile), which employs multiple embedding tables to make the model domain-aware at the embeddings which consist most parameters in the model, and a covariance loss upon these embeddings to disentangle them, enabling the model to capture diverse user interests among domains. Empirical analysis demonstrates that our method successfully addresses both challenges and outperforms all state-of-the-art methods on public datasets. During online A/B testing in Tencent's advertising platform, Crocodile achieves 0.72% CTR lift and 0.73% GMV lift on a primary advertising scenario.

cs.IR↗