Build a Small Reference Set | Directed k-NN graphs with knn_graph_bigmatrix() | Mutual k-NN graphs with mutual_knn_graph_bigmatrix() | Shared-nearest-neighbour graphs with snn_graph_bigmatrix() | Radius graphs with radius_graph_bigmatrix() | Symmetrization options and edge semantics | Converting outputs with as_edge_list(), as_triplet(), and as_sparse_matrix() | Using graph outputs downstream
Package: bigKNN 0.3.0
bigKNN: Exact Search and Graph Construction for 'bigmemory' Matrices
Exact nearest-neighbour and radius-search routines that operate directly on 'bigmemory::big.matrix' objects. The package streams row blocks through 'BLAS' kernels, supports self-search and external-query search, exposes prepared references for repeated queries, and can build exact k-nearest-neighbour, radius, mutual k-nearest-neighbour, and shared-nearest-neighbour graphs. Version 0.3.0 adds execution plans, serializable prepared caches, resumable streamed graph jobs, coercion helpers, exact candidate reranking, and recall summaries for evaluating approximate neighbours.
Authors:
bigKNN_0.3.0.tar.gz
bigKNN_0.3.0.zip(r-4.7)bigKNN_0.3.0.zip(r-4.6)bigKNN_0.3.0.zip(r-4.5)
bigKNN_0.3.0.tgz(r-4.6-x86_64)bigKNN_0.3.0.tgz(r-4.6-arm64)bigKNN_0.3.0.tgz(r-4.5-x86_64)bigKNN_0.3.0.tgz(r-4.5-arm64)
bigKNN_0.3.0.tar.gz(r-4.6-arm64)bigKNN_0.3.0.tar.gz(r-4.7-arm64)bigKNN_0.3.0.tar.gz(r-4.7-x86_64)bigKNN_0.3.0.tar.gz(r-4.6-x86_64)
bigKNN_0.3.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
bigKNN/json (API)
| # Install 'bigKNN' in R: |
| install.packages('bigKNN', repos = c('https://fbertran.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/fbertran/bigknn/issues
Pkgdown/docs site:https://fbertran.github.io
Last updated from:fc76b39aa5. Checks:12 OK. Indexed: yes.
| Target | Result | Time | Files | Syslog |
|---|---|---|---|---|
| linux-devel-arm64 | OK | 160 | ||
| linux-devel-x86_64 | OK | 143 | ||
| source / vignettes | OK | 180 | ||
| linux-release-x86_64 | OK | 137 | ||
| macos-release-arm64 | OK | 142 | ||
| macos-release-x86_64 | OK | 295 | ||
| macos-oldrel-arm64 | OK | 160 | ||
| macos-oldrel-x86_64 | OK | 359 | ||
| windows-devel | OK | 145 | ||
| windows-release | OK | 179 | ||
| windows-oldrel | OK | 133 | ||
| wasm-release | OK | 118 |
Exports:as_edge_listas_sparse_matrixas_tripletcount_within_radius_bigmatrixknn_bigmatrixknn_graph_bigmatrixknn_graph_stream_bigmatrixknn_load_preparedknn_plan_bigmatrixknn_prepare_bigmatrixknn_search_preparedknn_search_stream_preparedknn_stream_bigmatrixknn_validate_preparedmutual_knn_graph_bigmatrixradius_bigmatrixradius_graph_bigmatrixradius_stream_bigmatrixradius_stream_job_bigmatrixrecall_against_exactrerank_candidates_bigmatrixresume_knn_jobsnn_graph_bigmatrix
Exact Graph Construction from big.matrix Data
Last update: 2026-03-26
Started: 2026-03-26
Execution Plans and Streaming Workflows
Why plans and streaming matter for large matrices | Build a Repeatable Example | Building a plan with knn_plan_bigmatrix() | How memory budget maps to block size | Running planned in-memory search | Streaming k-NN output with knn_stream_bigmatrix() | Streaming radius output with radius_stream_bigmatrix() | Dense versus sparse query inputs | Practical guidance for choosing output modes
Last update: 2026-03-26
Started: 2026-03-26
Prepared References for Repeated Exact Search
When prepared references help | Build a File-Backed Reference | Building a prepared reference with knn_prepare_bigmatrix() | Reusing it with knn_search_prepared() | Streaming prepared results with knn_search_stream_prepared() | Persisting caches with cache_path | Reloading with knn_load_prepared() | Validating with knn_validate_prepared() | Common failure modes and how to avoid them
Last update: 2026-03-26
Started: 2026-03-26
Quick Start with bigKNN
Create a Small Reference Matrix | First Exact k-Nearest-Neighbour Search | Searching New Query Points | First Radius Search | Choosing a Metric | Where to Go Next
Last update: 2026-03-26
Started: 2026-03-26
Resumable Streaming Jobs
When resumable jobs are worth using | Build a Durable Working Directory | Checkpointed graph streaming with knn_graph_stream_bigmatrix() | Checkpointed radius streaming with radius_stream_job_bigmatrix() | Restarting work with resume_knn_job() | Graph Job Resume | Radius Job Resume | Destination matrix sizing and storage types | What checkpoint files contain | Failure recovery and restart patterns | Operational tips for long-running jobs
Last update: 2026-03-26
Started: 2026-03-26
Using bigKNN as Exact Ground Truth
Why exact ground truth matters | Build a Small Exact Evaluation Set | Producing exact neighbours with bigKNN | Measuring recall with recall_against_exact() | Reranking candidate sets with rerank_candidates_bigmatrix() | Comparing approximate candidates against exact results | Suggested workflow alongside a separate ANN package | Interpreting recall and reranking results | Limits and caveats
Last update: 2026-03-26
Started: 2026-03-26
