This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Friends-of-Friends algorithm | |
|---|---|
| Name | Friends-of-Friends algorithm |
| Type | Clustering algorithm |
| Domain | Astronomy; Computational Science |
| Introduced | 1980s |
| Notable for | Group finding in point distributions |
Friends-of-Friends algorithm
The Friends-of-Friends algorithm is a proximity-based clustering method widely used in Max Planck Society-scale cosmology and Harvard University-level data analysis, with origins tied to early surveys such as the Sloan Digital Sky Survey and collaborations involving institutions like Lawrence Berkeley National Laboratory and California Institute of Technology. It groups points by linking neighbors within a fixed linking length, a strategy applied across projects from Hubble Space Telescope programs to simulations run on Oak Ridge National Laboratory supercomputers, and has been discussed in conferences at Institute of Physics and American Astronomical Society meetings.
The algorithm assigns membership by connecting particles or objects if pairwise separations are below a chosen threshold, a concept operationalized in studies from Princeton University and University of Cambridge and incorporated into pipelines at European Southern Observatory and NASA. It has been compared to other techniques developed at Massachusetts Institute of Technology and Stanford University and evaluated alongside methods from Los Alamos National Laboratory and Fermilab.
Starting from an initial catalog such as those produced by Two Micron All Sky Survey or simulation outputs from Millennium Simulation, the procedure selects an object, finds all neighbors within a linking length informed by works at Max Planck Institute for Astrophysics and University of Chicago, and iteratively adds their neighbors until no new members are found, echoing algorithmic patterns discussed at ACM and IEEE symposia. Implementations reference data structures and libraries developed at National Institute of Standards and Technology and Argonne National Laboratory, and have been incorporated into analysis toolkits used at European Research Council-funded projects.
Analyses of percolation thresholds and cluster statistics tie into theoretical work from Cambridge University Press publications and mathematical frameworks advanced at Princeton Plasma Physics Laboratory and Courant Institute of Mathematical Sciences. Metrics such as multiplicity functions, two-point correlation functions, and halo mass functions are computed and compared to predictions from models cited in journals associated with Royal Society and Nature Publishing Group. Convergence properties and scaling relations are discussed in the context of results from International Astronomical Union workshops and texts used at Yale University.
The algorithm has been deployed to identify galaxy groups in surveys like DEEP2 Redshift Survey and 2dF Galaxy Redshift Survey, to extract dark matter halos from simulations such as those run by European Grid Infrastructure teams, and to support analyses in projects led by Space Telescope Science Institute and National Aeronautics and Space Administration. It has also been used in cross-disciplinary settings involving datasets curated by British Geological Survey and toolchains developed at IBM Research and Google Research for large-scale clustering tasks encountered in collaborations with Microsoft Research.
Extensions include adaptive linking length schemes influenced by studies at Institut d'Astrophysique de Paris and hybrid methods combining density-based approaches from Lawrence Livermore National Laboratory with graph-based community detection techniques researched at University of California, Berkeley and Princeton University. Other variants integrate corrective calibrations from works associated with Max Planck Institute for Extraterrestrial Physics and statistical adjustments discussed at Columbia University and Brown University.
Practical implementations appear in codes developed at Kavli Institute for Cosmology and high-performance libraries optimized on architectures from NVIDIA Corporation and Cray Inc., and run on clusters housed at National Energy Research Scientific Computing Center. Complexity depends on neighbor search strategies using spatial indexing schemes popularized by teams at University of Toronto and ETH Zurich, with typical performance considerations discussed in proceedings from SIGMOD and Supercomputing Conference events.
Critiques raised in literature from Royal Astronomical Society and presented at American Physical Society meetings note sensitivity to the choice of linking length, issues with separating close substructures in analyses performed by Harvard-Smithsonian Center for Astrophysics, and biases compared against methods validated by NASA Ames Research Center teams. Debates about robustness and parameter dependence have been prominent in reviews commissioned by National Science Foundation and at panels convened by European Southern Observatory.
Category:Clustering algorithms