arXiv subjects

Yiting Zheng

Publications and source records attributed to Yiting Zheng.

7 recordsLinked to original sources

arXiv2026

Topological Models for Categories Associated with the Weighted Projective Lines of Type $(2,2,2,2)$

We construct a geometric model for the derived category of the weighted projective line $\mathbb{CP}^1_ω$ of type $(2,2,2,2)$ using a graded sphere with four binaries. More precisely, we give a bijection between the set of indecomposable rigid objects and the set of certain arcs, such that the dimensions of $\operatorname{Hom}$-spaces are computed by oriented intersection numbers. As applications, we provide a geometric realization of the indecomposable rigid objects in the cluster category $\mathcal{C}(\mathbb{CP}^1_ω)$ via tagged arcs on the four-punctured sphere, with $\operatorname{Hom}$-space dimensions given by tagged intersection numbers. When the weighted points of $\mathbb{CP}^1_ω$ are $(0,1,\infty,\frac{1}{2})$, both correspondences are compatible with the natural actions of the automorphism groups and the corresponding mapping class groups.

math.RT↗

arXiv2026

RxnCLF: Contrastive Transformation-Aware Reaction Foundation Model for Improved Reactivity Prediction

Reaction yield prediction remains challenging because labeled data are scarce and reaction space is both combinatorially large and sparsely populated, limiting the generalization of existing reaction representations. String-, fingerprint-, and graph-based reaction encodings only partially capture chemical transformations, making accurate prediction difficult for reactions with complex substrates. We propose reaction contrastive learning foundation (RxnCLF), a self-supervised contrastive framework for reaction representation learning. RxnCLF is built on a condensed reaction graph (CRG) that unifies reactant and product information into a single graph, enabling the model to learn explicit and enriched transformation structure rather than disconnected graphs. Pretrained on 1.7 million Pistachio reactions, RxnCLF learns a compact and continuous latent space that captures both reaction-center features and broader side chain contexts, making it transformation-aware and chemically interpretable. Fine-tuned on multiple yield prediction benchmarks, including Buchwald-Hartwig, Pd-catalyzed BH coupling, and proprietary HTE C-N coupling and amide formation datasets, RxnCLF consistently outperforms graph and sequence-based baselines, improving R2 and achieving the best performance overall. Our results highlight the promise of CRG-based RxnCLF as a scalable reaction foundation model, with the potential to generalize across broader reaction spaces and support diverse downstream reaction informatics tasks, including regioselectivity prediction, enantioselectivity prediction, and reaction condition optimization.

cs.LG↗

arXiv2025

Combinatorial Approaches to Exceptional Sequences for Weighted Projective Lines of Type $(p,q)$

We provide a combinatorial description of morphisms in the coherent sheaf category ${\rm coh}\mbox{-}\mathbb{X}(p,q)$ over weighted projective line of type $(p,q)$ via a marked annulus. This leads to a geometric realization of exceptional sequences in ${\rm coh}\mbox{-}\mathbb{X}(p,q)$. As applications, we present a classification of complete exceptional sequences, an effective method for enlarging exceptional sequences, and a new proof of the transitivity of the braid group action on complete exceptional sequences. Besides, we offer a combinatorial description of tilting bundles via lattice paths and count the number of tilting sheaves in ${\rm coh}\mbox{-}\mathbb{X}(p,q)$, up to the Auslander-Reiten translation.

math.RT↗

arXiv2025

Unidirectional lasing via vacuum induced coherent in defective atomic lattice

We skillfully utilized vacuum induced coherence to amplify the probe light, and then successfully achieved both nonreciprocal reflection and lasing oscillation in a single physical system by leveraging the distributed feedback and spatial symmetry breaking effect of the one-dimensional defective atomic lattice. This innovative scheme for realizing unidirectional reflection lasing (URL) is based on both non-Hermitian degeneracy and spectral singularity (NHDSS, means $λ_{+}^{-1}\simeqλ_{-}^{-1}\rightarrow0$). Therefore, we analyze the modulation of parameters such as the lattice structure and external optical fields in this system to find NHDSS point, and further verified the conditions for its occurrence by solving the transcendental equation of susceptibility satisfying the NHDSS point, as well as analyzed its physical essence. Our mechanism is not only beneficial for the integration of photonic devices in quantum networks, but also greatly improves the efficiency of optical information transmission.

physics.optics↗

arXiv2025

HFedCKD: Toward Robust Heterogeneous Federated Learning via Data-free Knowledge Distillation and Two-way Contrast

Most current federated learning frameworks are modeled as static processes, ignoring the dynamic characteristics of the learning system. Under the limited communication budget of the central server, the flexible model architecture of a large number of clients participating in knowledge transfer requires a lower participation rate, active clients have uneven contributions, and the client scale seriously hinders the performance of FL. We consider a more general and practical federation scenario and propose a system heterogeneous federation method based on data-free knowledge distillation and two-way contrast (HFedCKD). We apply the Inverse Probability Weighted Distillation (IPWD) strategy to the data-free knowledge transfer framework. The generator completes the data features of the nonparticipating clients. IPWD implements a dynamic evaluation of the prediction contribution of each client under different data distributions. Based on the antibiased weighting of its prediction loss, the weight distribution of each client is effectively adjusted to fairly integrate the knowledge of participating clients. At the same time, the local model is split into a feature extractor and a classifier. Through differential contrast learning, the feature extractor is aligned with the global model in the feature space, while the classifier maintains personalized decision-making capabilities. HFedCKD effectively alleviates the knowledge offset caused by a low participation rate under data-free knowledge distillation and improves the performance and stability of the model. We conduct extensive experiments on image and IoT datasets to comprehensively evaluate and verify the generalization and robustness of the proposed HFedCKD framework.

cs.LG↗

arXiv2025

A geometric realization for maximal almost pre-rigid representations over type $\mathbb{D}$ quivers

By using the equivariant theory of group actions, we give a geometric model for the category of finite dimensional representations over a type $\mathbb{D}$ quiver $Q_{D}$ with $n$ vertices and directional symmetry. Furthermore, we introduce the notion of maximal almost pre-rigid representations over $Q_{D}$, which form a family of objects counted by the generalized Catalan number. We present a geometric realization for maximal almost pre-rigid representations and prove that the endomorphism algebras of maximal almost pre-rigid representations are tilted algebras of type $Q_{\overline{D}}$, where $Q_{\overline{D}}$ is a quiver obtained by adding $n-2$ new vertices and $n-2$ arrows to the quiver $Q_{D}$. Additionally, we define a partial order on the set of maximal almost pre-rigid representations, which therefore presents a representation-theoretic interpretation of the type-$\mathbb{D}$ Cambrian lattice determined by $Q_{D}$. Meanwhile, we obtain a representation-theoretic interpretation of the type-$\mathbb{B}$ Cambrian lattices.

math.RT↗

arXiv2020

Efficient Solar-driven Steam Generation Enabled by An Ultra-black Paint

Solar-driven interfacial steam generation for desalination has attracted broad attention. However, a significant challenge still remains for achieving a higher evaporation rate and high water quality, together with a cost-effective and easy-to-manufacture device to provide a feasible solar-driven steam generation system. In this study, a novel ultra-black paint, Black 3.0, serving as a perfect solar absorber is introduced into the hot-pressed melamine foam networks, allowing us to construct an ultra-black (99% absorptance in the solar region) and self-floating evaporation device. The high performing features of effective solar absorptance and salt-rejection capability contribute to a high-to-date evaporation rate of freshwater at 2.48 kg m-2 h-1 under one sun (1 kW m-2). This interfacial solar evaporator has a daily drinkable water yield of 2.8 kg m-2 even in cloudy winter weather and maintains stability in water with a wide range of acidity and alkalinity (pH 1~14). These features will enable the construction of a facilely fabricated, robust, highly-efficient, and cost-effective solar steam generation system for freshwater production.

physics.app-ph↗