Searcharxiv⌕ Search

arXiv subjects

Xiaoyu Guan

Publications and source records attributed to Xiaoyu Guan.

3 recordsLinked to original sources

Hierarchical interfacial anchoring for durable perovskite solar cells

Although inverted perovskite solar cells now deliver certified efficiencies approaching their single-junction limits, their practical deployment remains constrained by the thermo-mechanical fragility of buried interfaces under realistic temperature fluctuations. Here, we identify thermo-mechanically driven, stress-accumulation-dominated adhesion failure at self-assembled monolayer-based buried interfaces as a critical failure mode in inverted perovskite solar cells. To address this issue, we propose an interfacial robustness design strategy based on a homometallic halide pretreatment, which constructs a chemically compatible and highly anchorable interfacial environment on transparent conductive oxides to guide the ordered assembly of SAMs. This interfacial treatment increases the density of surface hydroxyl species and enables hierarchical multipoint interfacial anchoring dominated by metal-oxygen coordination, with Cl- induced hydrogen-bonding interactions acting cooperatively, thereby enhancing anchoring robustness and improving tolerance to thermo-mechanical stress accumulation. With this interfacial strategy, the devices achieve a power conversion efficiency of 27.3% (26.9% certified) and retain 94% of their initial efficiency after 400 temperature cycles between -40 and 85 degrees C (ISOS-T-3). These results demonstrate that rational regulation of buried interfacial anchoring architectures can simultaneously enhance device efficiency and long-term operational reliability, offering a viable pathway toward high-efficiency and durable perovskite photovoltaics.

physics.optics↗

DPI: Exploiting Parameter Heterogeneity for Interference-Free Fine-Tuning

Supervised fine-tuning (SFT) is a crucial step for adapting large language models (LLMs) to downstream tasks. However, conflicting objectives across heterogeneous SFT tasks often induce the "seesaw effect": optimizing for one task may degrade performance on others, particularly when model parameters are updated indiscriminately. In this paper, we propose a principled approach to disentangle and isolate task-specific parameter regions, motivated by the hypothesis that parameter heterogeneity underlies cross-task interference. Specifically, we first independently fine-tune LLMs on diverse SFT tasks and identify each task's core parameter region as the subset of parameters exhibiting the largest updates. Tasks with highly overlapping core parameter regions are merged for joint training, while disjoint tasks are organized into different stages. During multi-stage SFT, core parameters acquired in prior tasks are frozen, thereby preventing overwriting by subsequent tasks. To verify the effectiveness of our method, we conducted intensive experiments on multiple public datasets. The results showed that our dynamic parameter isolation strategy consistently reduced data conflicts and achieved consistent performance improvements compared to multi-stage and multi-task tuning baselines.

cs.CL↗

Introducing Anisotropic Fields for Enhanced Diversity in Crowd Simulation

Large crowds exhibit intricate behaviors and significant emergent properties, yet existing crowd simulation systems often lack behavioral diversity, resulting in homogeneous simulation outcomes. To address this limitation, we propose incorporating anisotropic fields (AFs) as a fundamental structure for depicting the uncertainty in crowd movement. By leveraging AFs, our method can rapidly generate crowd simulations with intricate behavioral patterns that better reflect the inherent complexity of real crowds. The AFs are generated either through intuitive sketching or extracted from real crowd videos, enabling flexible and efficient crowd simulation systems. We demonstrate the effectiveness of our approach through several representative scenarios, showcasing a significant improvement in behavioral diversity compared to classical methods. Our findings indicate that by incorporating AFs, crowd simulation systems can achieve a much higher similarity to real-world crowd systems. Our code is publicly available at https://github.com/tomblack2014/AF\_Generation.

cs.MA↗