weighted graph

weighted graph
Name	Weighted graph
Type	Mathematical object
Field	Graph theory
Introduced	20th century
Notable	Dijkstra algorithm; Prim algorithm; Kruskal algorithm

weighted graph

A weighted graph is a graph in which a numerical value, called a weight, is assigned to each edge (and sometimes to vertices) to encode costs, capacities, distances, probabilities, or strengths. Originating in work on networks and optimization, weighted graphs provide a richer model than unweighted graphs for problems in routing, scheduling, and modeling physical systems. They connect to classical results and tools developed by figures and institutions in mathematics and computer science, and they underpin algorithms used in operations research, electrical engineering, and geographic information systems.

Definition

A weighted graph is formally a pair (V, E) together with a weight function w mapping elements of E (or V) to a numerical set such as the real numbers or nonnegative integers. The notion appears in foundational texts and courses across departments at universities such as Massachusetts Institute of Technology, Stanford University, University of Cambridge, and in work by researchers affiliated with Bell Labs, AT&T, and the IBM Research labs. In directed weighted graphs, each ordered pair in E carries a weight, while in undirected weighted graphs the weight is associated to unordered pairs; applications often require nonnegative weights, but negative weights arise in models influenced by developments at institutions like Bell Labs and in formulations related to the Bellman–Ford algorithm lineage.

Types of Weights and Edge Models

Weights may represent lengths, costs, capacities, probabilities, or similarities depending on context and provenance. In transportation modeling influenced by planners at the United States Department of Transportation and consultancies such as McKinsey & Company, weights often denote travel time or distance; in telecommunications work at AT&T and Nokia, weights can denote bandwidth or latency. Stochastic weights appear in probabilistic graph models used in research from Carnegie Mellon University and Microsoft Research, where edges carry distributions rather than scalars. In flow networks (studied at Courant Institute and in textbooks from Princeton University Press), capacities impose upper bounds on weighted edges, while in electrical network analogies dating from work by Leonhard Euler and extended in operations research, weights map to resistances or conductances.

Properties and Metrics

Weighted graphs admit specialized metrics and invariants used across disciplines. Shortest-path distance metrics are central to results developed in courses at University of California, Berkeley and algorithms like those published by researchers at Stanford University; minimum spanning trees and cut metrics trace to work formalized in papers appearing in journals associated with the American Mathematical Society and conferences such as STOC and FOCS. Spectral properties of weighted adjacency and Laplacian matrices are studied in contexts linked to research groups at ETH Zurich and the Institute for Advanced Study; eigenvalues encode connectivity and clustering tendencies exploited in machine learning work from Google Research and Facebook AI Research. Centrality measures (betweenness, closeness) adapted to weighted settings are used in network science studies from the Santa Fe Institute and in epidemiology models developed at institutions like the Centers for Disease Control and Prevention.

Algorithms and Applications

Algorithms for weighted graphs include Dijkstra’s algorithm, Bellman–Ford, Floyd–Warshall, Prim’s and Kruskal’s algorithms; variants and analyses appear in textbooks from MIT Press and in conference proceedings of ACM SIGGRAPH and IEEE. Applications span route planning systems by companies such as Garmin and TomTom, logistics optimization in firms like DHL and Amazon, and mapping services by Google Maps and HERE Technologies. In computational biology labs at Broad Institute and Wellcome Trust Sanger Institute, weighted graph algorithms assist in sequence assembly and protein interaction analysis. In finance, trading and risk models from institutions such as Goldman Sachs and J.P. Morgan leverage weighted network representations for contagion and exposure analysis.

Representations and Data Structures

Practical implementations use adjacency matrices, adjacency lists with weight annotations, incidence matrices, and compact sparse formats favored in high-performance libraries and toolkits from NVIDIA and Intel. Libraries and frameworks such as those developed by teams at Apache Software Foundation and NumPy/SciPy communities provide optimized sparse matrix operations for weighted Laplacians and solvers used in computational pipelines at research centers like Los Alamos National Laboratory. Data serialization formats employed by companies such as Amazon Web Services and standards bodies like World Wide Web Consortium influence how weighted graph data are stored and exchanged.

Examples and Use Cases

Concrete examples include road networks encoding travel time for navigation systems by Uber and Lyft, electrical grid models studied by researchers at the National Grid and Edison Electric Institute, and social networks with interaction strengths analyzed in studies affiliated with Harvard University and Stanford University. Supply-chain optimization examples derive from case studies at multinational firms like Walmart and Procter & Gamble; urban planners at municipal agencies such as Transport for London use weighted graphs for network resilience and capacity planning. Academic case studies appear in journals and conferences supported by organizations like the SIAM and the IEEE Communications Society.

Category:Graph theory