A fast algorithm for identifying Friends-of-Friends halos

TL;DR

This paper introduces a fast, memory-efficient algorithm for identifying friends-of-friends halos in cosmological data, significantly reducing computational complexity and merge operations, especially in high-density regions.

Contribution

The authors present a novel, simple algorithm that improves the efficiency of friends-of-friends halo identification by reducing merge operations and computational costs using hierarchical trees and pruning techniques.

Findings

Reduces merge operations by over 50% in typical cosmological datasets.

Achieves linear time complexity for merging trees, improving over previous methods.

Easily integrable with existing pair enumeration codes.

Abstract

We describe a simple and fast algorithm for identifying friends-of-friends features and prove its correctness. The algorithm avoids unnecessary expensive neighbor queries, uses minimal memory overhead, and rejects slowdown in high over-density regions. We define our algorithm formally based on pair enumeration, a problem that has been heavily studied in fast 2-point correlation codes and our reference implementation employs a dual KD-tree correlation function code. We construct features in a hierarchical tree structure, and use a splay operation to reduce the average cost of identifying the root of a feature from $O[\log L]$ to $O[1]$ ($L$ is the size of a feature) without additional memory costs. This reduces the overall time complexity of merging trees from $O[L\log L]$ to $O[L]$, reducing the number of operations per splay by orders of magnitude. We next introduce a pruning operation that skips merge operations between two fully self-connected KD-tree nodes. This improves the robustness of the algorithm, reducing the number of merge operations in high density peaks from $O[\delta^2]$ to $O[\delta]$. We show that for cosmological data set the algorithm eliminates more than half of merge operations for typically used linking lengths $b \sim 0.2$ (relative to mean separation). Furthermore, our algorithm is extremely simple and easy to implement on top of an existing pair enumeration code, reusing the optimization effort that has been invested in fast correlation function codes.