Jaebin Lee | Searcharxiv

arXiv2026

Strain-programmable exciton diffusion in moir\'e heterostructures

Moir\'e superlattices in van der Waals heterostructures have recently gained significant attention as an intriguing platform for studying correlated electronic systems and exotic excitonic properties. Previous reports, however, focused on creating and modulating moir\'e heterostructures through interlayer twisting or lattice constant mismatches, limiting controls on symmetry of heterostructures. In this work, we show that strain significantly alters the geometry of moir\'e superlattices by breaking the C3 rotational symmetry. We realize strain-induced moir\'e superlattices by intentionally regulating interlayer strain in WSe2-MoSe2 heterostructures, which is manifested by linearly polarized interlayer exciton emission coupled to the strain direction. Furthermore, interlayer exciton diffusion was preferentially guided along the stretched moir\'e superlattice orientations over a wide spatial range, reflecting the strain-modified moir\'e potentials. Our work highlights strain tuning as a versatile tool for designing moir\'e superlattices and programming excitonic transport, which opens pathways for van der Waals logic and information processing devices.

cond-mat.mes-hall↗

arXiv2025

TimePerceiver: An Encoder-Decoder Framework for Generalized Time-Series Forecasting

In machine learning, effective modeling requires a holistic consideration of how to encode inputs, make predictions (i.e., decoding), and train the model. However, in time-series forecasting, prior work has predominantly focused on encoder design, often treating prediction and training as separate or secondary concerns. In this paper, we propose TimePerceiver, a unified encoder-decoder forecasting framework that is tightly aligned with an effective training strategy. To be specific, we first generalize the forecasting task to include diverse temporal prediction objectives such as extrapolation, interpolation, and imputation. Since this generalization requires handling input and target segments that are arbitrarily positioned along the temporal axis, we design a novel encoder-decoder architecture that can flexibly perceive and adapt to these varying positions. For encoding, we introduce a set of latent bottleneck representations that can interact with all input segments to jointly capture temporal and cross-channel dependencies. For decoding, we leverage learnable queries corresponding to target timestamps to effectively retrieve relevant information. Extensive experiments demonstrate that our framework consistently and significantly outperforms prior state-of-the-art baselines across a wide range of benchmark datasets. The code is available at https://github.com/efficient-learning-lab/TimePerceiver.

cs.LG↗