Efficient hierarchical clustering for single-dimensional data using CUDA

This repository contains the implementation of the clustering algorithm presented in the paper:

• Adam Rehn, Aidan Possemiers, and Jason Holdsworth. Efficient hierarchical clustering for single-dimensional data using CUDA. In Proceedings of the Australasian Computer Science Week Multiconference, ACSW '18, 14:1–14:10. New York, NY, USA, 2018. ACM.

Building from source

The following dependencies are required in order to build Cluster1D from source:

• A modern, C++11-compliant compiler (Clang, GCC, Visual Studio 2015 or newer)
• CMake 3.5 or newer
• NVIDIA CUDA Toolkit

To build under macOS or Linux, run the following commands:

mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
cmake --build .

To build under Windows, run the following commands:

mkdir build && cd build
cmake -A x64 ..
cmake --build . --config Release

Performing clustering

(Running the built Cluster1D executable without any arguments will also print these usage instructions.)

Usage: Cluster1D <INFILE> <LINKAGE> [--r-format]

Input file should be a text file containing floats, one per line.

Supported linkage metrics:

• single
• complete

Specifying --r-format will transform the output into a form compatible with the default dendrogram representation used by the R hclust() function.

Output is written to stdout in CSV format.