Graph-structured data, operating as an abstraction of data containing nodes and interactions between nodes, is pervasive in the real world. There are numerous ways dedicated to extract graph structure information explicitly or implicitly, but whether it has been adequately exploited remains an unanswered question. This work goes deeper by heuristically incorporating a geometric descriptor, the discrete Ricci curvature (DRC), in order to uncover more graph structure information. We present a curvature-based topology-aware graph transformer, termed Curvphormer. This work expands the expressiveness by using a more illuminating geometric descriptor to quantify the connections within graphs in modern models and to extract the desired structure information, such as the inherent community structure in graphs with homogeneous information. We conduct extensive experiments on a variety of scaled datasets, including PCQM4M-LSC, ZINC, and MolHIV, and obtain a remarkable performance gain on various graph-level tasks and fine-tuned tasks.
1 October 2010 in Japanese Journal of Applied Physics
This work presents a technique forin situmeasurement of light-induced refractive index gratings in epoxy resin using digital holographic microscopy (DHM). The reconstructed phase image derived from a digital hologram can exhibit the grating structure and refractive index profile of an epoxy resin hologram. Reconstruction properties of finite aperture effect in the DHM system are considered and analyzed theoretically. Due to the high spatial frequency components collected by the objective lens, the DHM system can measure fine grating structures and numerically determine the reconstructed images in increased detail. Grating formation and dynamic behavior during the light-induced holographic process can be demonstrated experimentally and characterized using the proposed scheme. Compared with an optical holographic readout, the proposed technique facilitates direct observation and substantial understanding of the holographic recording mechanism in a microscopic view.
Digraphs are generalization of graphs in which each edge is given one or two directions. For each digraph, there exists a transitive digraph containing it. Moreover, all the formal linear combinations of allowed elementary paths form a basis of the path complex for a transitive digraph. Hence, the study of discrete Morse theory on transitive digraphs is very important for the further study of discrete Morse theory on general digraphs. As we know, the definition of discrete Morse function on a digraph is different from that on a simplical complex or a cell complex: each discrete Morse function on a digraph is a discrete flat Witten-Morse function. In this paper, we deform the usual boundary operator, replacing it with a boundary operator with parameters and consider the induced Laplace operators. In addition, we consider the eigenvectors of the eigenvalues of the Laplace operator that approach to zero when the parameters approach infinity, define the generation space of these eigenvectors, and further give the Witten complex of digraphs. Finally, we prove that for a transitive digraph, Witten complex approaches to its Morse complex. However, for general digraphs, the structure of Morse complex is not as simple as that of transitive digraphs and the critical path is not directly related to the eigenvector with zero eigenvalue of Laplace operator. This is explained in the last part of the paper.
This work develops a rigorous Ricci flow on weighted directed graphs whose formulation decouples edge weights from the distance metric, so that the metric remains fixed while the edge weights evolve. The design preserves the triangle inequality automatically and avoids the need for edge surgery required by Bai et al. (Amer. J. Math., 146(4) 2024).To capture the essence of asymmetry in directed networks and to enhance the capability of modeling more flexible structures, we incorporate a node-wise balancing factor β(x) that regulates between outflow and inflow. Since the undirected graph case can be viewed as a bidirectional graph, this model can naturally apply to the undirected case. Building on the continuous Ricci flow evolution framework, we propose a discrete Ricci flow algorithm that is applicable to numerical computing. Numerical studies on various directed graph examples demonstrate the capacity of the proposed flow to reveal structural asymmetry and dynamic evolutions.