Package: sfgpu
Title: Explicit CPU, CUDA, and Metal Distances for Projected Points
Version: 0.1.0
Authors@R: person("Egor", "Kotov", email = "kotov.egor@gmail.com", role = c("aut", "cre", "cph"))
Description: Computes dense Euclidean distance matrices and sparse
        within-radius point matches for Cartesian two-column
        coordinates and projected 'sf' POINT geometries. A portable
        double-precision CPU backend is always available; optional
        NVIDIA CUDA and Apple Metal backends must be requested
        explicitly at build time and at each call.
License: MIT + file LICENSE
Encoding: UTF-8
Suggests: sf, units, testthat (>= 3.0.0)
Config/testthat/edition: 3
SystemRequirements: C++17; optional NVIDIA CUDA toolkit for
        --enable-cuda; Apple Metal SDK for --enable-metal
RoxygenNote: 7.3.3
Config/pak/sysreqs: nvidia-cuda-dev
Repository: https://e-kotov.r-universe.dev
Date/Publication: 2026-09-27 15:09:24 UTC
RemoteUrl: https://github.com/e-kotov/sfgpu
RemoteRef: dev
RemoteSha: c40de03841be33422ac7628415411468f77336ef
NeedsCompilation: yes
Packaged: 2026-09-27 16:08:11 UTC; root
Author: Egor Kotov [aut, cre, cph]
Maintainer: Egor Kotov <kotov.egor@gmail.com>