Core C++ Sampling Engine for 'glmbayes'
Overview
Core C++ engine for 'glmbayes': envelope-based iid linear and generalized linear model samplers, prior-family routing, and optional 'OpenCL' acceleration. Sampling for supported non-conjugate models uses accept-reject methods based on likelihood subgradients as in Nygren and Nygren (2006) doi:10.1198/016214506000000357. Intended as a developer backend for the 'glmbayes' formula interface; end users should use 'glmbayes' for modelling with interfaces analogous to 'lm' and 'glm'. Mixed-model engines are planned for a future release.
Bug fixes
- **
residuals.rglmb()/residuals.rlmb()/residuals.summary.rglmb():** Fixed theysimargument to substitute for the **observed response**y(fitted value held fixed at each draw), not the reverse. The previous behavior substitutedysimfor the fitted value while holdingyfixed, which does not correspond to a standard posterior-predictive residual check and contradicted this function's ownysim("simulated responses") documentation. The corrected convention matches **glmbayes**'sresiduals.glmb()(simulate new response data, recompute residuals against the same fitted values, and compare to the actual residuals for outlier diagnostics) — the two packages'ysimsemantics were inconsistent since this function was added post-split; they are now aligned. Behavior for the default case (ysim = NULL) is unchanged.
Show all changes in 0.5.4Hide the rest of the changes in 0.5.4
- **Configure (Linux/macOS):**
-DUSE_OPENCLis set only when a **non-PoCL** OpenCL platform exposes at least one **GPU** device (same policy as **glmbayes** 0.9.75). PoCL-only ICD stacks (typical on CRAN debian-gcc incoming) no longer enable OpenCL at compile time, so kernel builds do not touch~/.cache/pocl(avoids incoming NOTE ontempfile_*there). Real GPU installs (NVIDIA/AMD/Intel) are unchanged.configure.winis unaffected (Windows never ran a runtime OpenCL probe). - **Configure policy:** Removed
tools/rcpp_include.R/tools/patch_rcpp_function_h.Rand theglmbayes_getRegisteredNamespacecompatibility shim (src/glmbayes_getRegisteredNamespace.{cpp,h}) fromconfigureandconfigure.win. Builds now rely on standard **LinkingTo: Rcpp** and the CRAN **Rcpp (>= 1.1.1)** requirement instead of probing/patchingRcpp/Function.hor recommending a GitHub install of Rcpp (same policy fix as **glmbayes** 0.9.73/0.9.75). - **CI:** Replaced
.github/workflows/rhub.yamlwith the current **glmbayes** version, removing dead matrix/report logic that referenced the now-removedtools/patch_rcpp_function_h.Rscript.
From NEWS.md in the release
Install
Health
- OK13 OKNow
- OK8 OK2026-08-06
Documentation
- Help pages
- 38includes a package overview page
- Examples that run
- 100%30 run in a plain check
- Exported functions with examples
- 85%
- Documented parameters
- 98%
- Return-value docs
- 100%
- Help pages citing references
- 61% with a bibliography file
Also includes 3 demo scripts (run with demo())
Downloads
Repository
Stars over time
Issues over time
Commits over time
4 development-tooling and community-health practices detected across 3 families in the upstream repository
Checks run against github.com/knygren/glmbayescore on 2026-10-18.
The R CMD check workflow runs on Linux
This repository shows signs of AI coding tools.
- Cursorfirst seen 2026-04-18Seen in the repository: .cursorNamed in .Rbuildignore: .cursorCommits: at least 3 crediting it in the message
- Claude Codefirst seen 2026-04-24Named in .Rbuildignore: .claude
Dependencies
Installing glmbayesCore adds 8 packages (7 with compiled code, 9.8 MB of source tarballs). Heaviest: nmathopencl 2.2 MB, Rcpp 1.8 MB, opencltools 1.6 MB.
Code & Tests
Datasets
People & History
Pick two releases to compare their code metrics.
- 0.5.4Latest
- 0.5.3first seen 2026-08-05
- RR 4.6.0 released · 2026-04-24
Package metadata
- First seen
- 2026-08-05
- Total releases
- 2 / 1 yrs
- License
- GPL-2 OSI
- Copyright
- see file COPYRIGHTS
- Minimum R
- ≥ 3.5.0
- Bundled data
- 127 KB / 6 files
- Download size
- 1.6 MB
Cite
Cite this package
Run in R for the authors' preferred citation:
citation("glmbayesCore")The package provides a citation file that builds its entry when R runs it. Run citation() to see it.
Cite the R Observatory
For a number measured here: a download total, a coverage figure, an archival date.
From data release v2026-10-21, which the citation names so these numbers can be found later. More on citing and the projects behind them.