580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can null models and randomization techniques characterize the structural patterns of real bipartite networks?
This theme investigates the development and application of statistically principled null models for bipartite networks to identify significant structural patterns beyond trivial connectivity. Given the complex nature of bipartite networks, traditional approaches projecting them into monopartite graphs can lose critical information. Creating null models that preserve specific characteristics (such as degree sequences) while randomizing others enables precise pattern detection relevant across domains like world trade and ecology.
Key finding: Introduces a rigorous null model for binary, undirected bipartite networks by extending entropy maximization methods previously applied to monopartite networks. The method enables systematic statistical pattern detection in... Read more
Key finding: Analytically reveals that transitivity and degree assortativity in one-mode projections of social networks arise naturally from the joint degree distributions and cycle structures of their underlying bipartite networks. Using... Read more
Key finding: Employs generating functions and synthetic configuration models to prove that degree distributions of both node types in bipartite networks jointly dictate the degree distribution of projected one-mode networks. The study... Read more
2. What algorithmic and computational advances enable efficient exact extraction of maximum biclique structures in large bipartite graphs?
Identifying maximum balanced bicliques in bipartite graphs is a fundamental NP-hard problem with extensive applications in bioinformatics, VLSI design, and network analysis. This theme covers exact algorithms exploiting graph structural properties (e.g., sparsity, density, degeneracy parameters) to achieve practical computation efficiencies for large-scale graphs. It also includes optimizations and complexity analyses that facilitate near polynomial-time performance in specific instances.
Key finding: Proposes two specialized exact algorithms for solving the maximum balanced biclique problem: denseMBB for dense bipartite graphs (with O*(1.3803^n) time complexity) and sparseMBB for large sparse bipartite graphs based on... Read more
3. How do bipartite graph decompositions and labeling schemes contribute to understanding and applying complex network structures?
This theme explores structural decompositions of complete bipartite and bipartite circulant graphs via graph products and orthogonal double covers, alongside the study of labeling strategies like compact mean labeling, which assign numerical values to nodes and edges under specific algebraic constraints. These approaches yield deeper insights into spectral properties, enable topology design, and facilitate deadlock-free routing in communication networks and parallel systems modeling.
Key finding: Develops new orthogonal double covers (ODCs) of complete bipartite graphs K_x,x using caterpillar graph families, including stars, disjoint cycles, and complete bipartite caterpillars. The work constructs symmetric starter... Read more
Key finding: Introduces decomposition approaches for bipartite circulant graphs via Cartesian and tensor products, revealing diverse categories of decomposed graphs. The study discusses applications to supercomputer topologies and... Read more
Key finding: Demonstrates that complete bipartite graphs K_{n,n}, as well as K_{2,2}, K_{2,3}, and K_{2,4}, admit compact mean labeling schemes where vertex labels and induced edge labels are distinct integers satisfying mean-based... Read more