Searcharxiv⌕ Search

arXiv subjects

Siwei Song

Publications and source records attributed to Siwei Song.

2 recordsLinked to original sources

Linear Dynamics of Composition Operators on Zygmund Spaces of Rooted Trees

We conjugate composition operators on the Zygmund and little Zygmund spaces of an infinite, locally finite rooted tree $T$ by the second-difference isometries. A path Green kernel gives row-finite matrices that characterize boundedness and yield exact operator norms. A rigidity result holds on the chain $\mathbb{N}_0$. Boundedness on $\mathcal{Z}_0(\mathbb{N}_0)$ is equivalent to $φ(n)=qn+b$ for all sufficiently large $n$, with $q\in\mathbb{N}_0$ and $b\in\mathbb{Z}$. For these symbols, $φ(n)>n$ eventually characterizes Li-Yorke chaos of $C_φ$ on either Zygmund space, allowing noninjectivity and periodic vertices. If $C_φ$ is bounded on $\mathcal{Z}_0(T)$ and $φ^k(T)$ eventually avoids every finite set, then for each $λ\ne0$, $λC_φ$ is Li-Yorke chaotic on either space exactly when the norms of its powers are unbounded. For $φ_q(n)=q(n+1)-1$ with integer $q\ge2$, the corresponding conditions are $|λ|q^2>1$ on both Zygmund spaces and $|λ|q>1$ on the little Lipschitz space. For bounded $C_φ$ on $\mathcal{Z}_0(\mathbb{N}_0)$, topological mixing and Devaney chaos each hold exactly for injective symbols with no periodic vertices and eventual slope $q\ge2$. For such bounded operators with strictly increasing $φ$ and $φ(0)>0$, $q\ge2$ also characterizes hypercyclicity and frequent hypercyclicity.

math.FA↗

From Models to Systems: A Comprehensive Survey of Efficient Multimodal Learning

The rapid expansion of multimodal models has surfaced formidable bottlenecks in computation, memory, and deployment, catalyzing the rise of Efficient Multimodal Learning (EML) as a pivotal research frontier. Despite intensive progress, a cohesive understanding of what, how, and where efficiency is manifested across the learning stack remains fragmented. This survey systematizes the EML landscape by introducing the first structured, model-to-system taxonomy. We distill insights from over 300 seminal works into three hierarchical levels--model, algorithm, and system--addressing architectural parsimony, execution refinement, and hardware-aware orchestration, respectively. Moving beyond a purely categorical review, we offer a methodological synthesis of the vertical synergies between these layers, elucidating how cross-layer co-design contributes to the fundamental "Efficiency-Utility-Privacy" trade-off. Through an integrative case study of Multimodal Large Language Models (MLLMs), we trace the field's evolutionary trajectory from initial structural adjustments to modern full-stack resource orchestration. Furthermore, we provide a holistic discussion and application-specific optimization blueprints for diverse domains and posit a paradigm shift toward self-regulating intelligence, where efficiency is an intrinsic, emergent property of the model's fundamental design rather than a post-hoc constraint. Finally, we present open challenges and future directions that will define the trajectory of EML research. This survey establishes a structured framework for multimodal systems that are not only high-performing and generalizable but natively efficient and ready for ubiquitous deployment. A continuously updated version is available at https://github.com/pwang322/Efficient-Multimodal-Learning-Survey.

cs.MM↗