GLBFP 0.6.1

GLBFP 0.6.1 updates the package documentation and the published benchmark records. The package provides fixed-grid leave-one-out scores via compute_Di() and supports fields indexed in multiple dimensions.

• Vignettes
• Fields indexed in multiple dimensions
• Quebec climate examples

Install this GitHub release with remotes::install_github("AurelienNicosiaULaval/GLBFP", ref = "v0.6.1"). The attached R source package includes the built vignettes.

GLBFP 0.6.0

GLBFP 0.6.0 is available on GitHub. The CRAN release is 0.5.2.

• compute_Di() computes fixed-grid leave-one-out scores. The single-grid LBFP case is described in arXiv:2607.14335.
• glbfp_field() supports fields indexed by any positive number of dimensions. The Quebec climate vignette illustrates spatial and space-time fields.
• Prepared grid counts can be reused for estimation and diagnostics. The package also provides density plots for up to three observation dimensions.
• Pointwise asymptotic LBFP intervals for complete rectangular fields are available under the assumptions documented in the indexed fields vignette.
• Equations are embedded in the HTML vignettes for online and local viewing.

Install this release from GitHub with remotes:

remotes::install_github("AurelienNicosiaULaval/GLBFP", ref = "v0.6.0")

The attached R source package includes the built vignettes.

GLBFP 0.5.2.9001: scientific review

GLBFP 0.5.2.9001 is a development version for scientific review.

It corrects finite-grid boundary extrapolation and two bandwidth-formula transcription errors, adds reusable native direct/prefix computation and integer-indexed fields, and includes real Quebec climate examples for N=2 and N=3 with d=2. The source archive includes built vignettes and source/licence records for the bundled data.

The Quebec climate vignette explains indexed fields for N=2 and N=3. The validation results and figures document the estimator, conventions, measured performance and remaining restrictions. The vignette follows the development documentation, and the validation link identifies the later documentation reorganization. The pre-release source archive and tag remain fixed at the original commit. Local R CMD check is OK with 318 passing assertions. GitHub checks passed on Windows, macOS and Linux (release, devel and oldrel), with successful documentation and coverage jobs. Platform checks are available for the exact review commit.

Download the source package for installation, or the review ZIP for the package, manual, guide and figures together. SHA256 checksums are attached. This pre-release was published on GitHub for scientific review.

install.packages("remotes")
remotes::install_github("AurelienNicosiaULaval/GLBFP", ref = "v0.5.2.9001")
library(GLBFP)
packageVersion("GLBFP")

Dependent-field confidence intervals, general GLBFP asymptotic normality, universal bandwidth optimality and calibration of D_i as a hypothesis test are not claimed. Default uncertainty is disabled; the opt-in IID calculation has explicit fixed-parameter restrictions. Legacy calls remain available, while corrected boundary values and automatic bandwidths intentionally change.

GLBFP 0.5.2

CRAN-facing release of GLBFP 0.5.2.

Changes since 0.5.1:

• Updated package authorship and citation metadata.
• Added CRAN status, CRAN checks, and CRAN download badges to the README.
• Updated README installation instructions now that GLBFP is available from CRAN.

Version scope:

• CRAN 0.5.2 does not include the fixed-grid leave-one-out D_i function compute_Di().
• compute_Di() is available only in the GitHub development version 0.5.2.9000 and accompanies the single-grid LBFP method in Nicosia, Duchesne and Carbon (2026), arXiv:2607.14335.

GLBFP 0.5.1

Release 0.5.1 prepares GLBFP for pre-CRAN and pre-R Journal review.

Highlights:

• strengthened input validation and error messages;
• added testthat coverage for core estimators, edge cases, S3 helpers, and grid behavior;
• added summary() and predict() methods for fit and grid objects;
• refreshed README, roxygen documentation, vignettes, NEWS, and citation files;
• added reproducible benchmarks, data provenance notes, pkgdown configuration, and GitHub Actions for checks, coverage, and pkgdown;
• added pre-R Journal audit and planning material.

Validation before release:

• devtools::test(): 50 passing tests;
• devtools::check(args = '--as-cran', manual = FALSE): 0 errors, 0 warnings, 0 notes;
• rcmdcheck::rcmdcheck(args = '--as-cran'): 0 errors, 0 warnings, 2 expected notes;
• urlchecker::url_check(): all URLs correct;
• lintr::lint_package(): no lints.

Known remaining work:

• verify the complete bibliographic record for the original GLBFP methodological article;
• complete the raw-data provenance script for ashua before CRAN or journal submission.

Consistency and CI/CD

Changes in version 0.5.0

Fixes & Improvements

• S3 Method Consistency: Fixed argument naming in print methods for ASH, GLBFP, and LBFP objects to comply with R standards (renamed object to x).
• Metadata: Updated DESCRIPTION file to include proper Authors@R fields and standardized the package description.
• Documentation: Regenerated package documentation to reflect code updates.

CI/CD

• Verified package passes R CMD check on Windows, macOS, and Linux via GitHub Actions.

"Version 0.4.0: Add output information on the *_estimate functions"

Add standard deviation and confidence interval in the output of each *_estimate function.

"Version 0.3.0: Correction of GLBFP function"

error in GLBFP function has been corrected

"Version 0.2.0: Updated functions"

• fix error in ASH and GLBFP functions

"GLBFP 0.1.0 - First Public Release"

🚀 First official release of GLBFP!

• Initial functions included
• Basic documentation available
• Installation via remotes::install_github("AurelienNicosiaULaval/GLBFP")