HDNRA 2.1.0

New features

• Added WZ2026.GLHTBF.2cNRT(), an F-type normal-reference test for heteroscedastic high-dimensional general linear hypothesis testing (GLHT) problems.
• WZ2026.GLHTBF.2cNRT() supports the grouped-data GLHTBF interface WZ2026.GLHTBF.2cNRT(Y, G, n, p), where Y is a list of groupwise data matrices, G is the raw full-row-rank contrast matrix, n is the vector of group sample sizes, and p is the common data dimension.
• The new GLHTBF F-type routine implements a trace-studentized quadratic contrast statistic with Welch–Satterthwaite two-cumulant F-type normal-reference calibration.
• The returned NRtest object reports the test statistic, p-value, fitted numerator and denominator degrees of freedom, and the corresponding approximation method.
• The new GLHTBF F-type routine can be used for both omnibus one-way MANOVA-type hypotheses and targeted rank-one contrast hypotheses through the same grouped-data interface.
• Added CCXH2024.GLHTBF.2cNRT() for the rank-one scale-invariant GLHTBF normal-reference procedure of Cao et al. (2024).
• Added LHNB2025.GLHTBF.NABT() for the rank-one random-integration GLHTBF normal-approximation procedure of Li et al. (2025).

Documentation and examples

• Updated the GLHTBF documentation to clarify that the software argument G is the raw contrast matrix supplied by the user; the normalized contrast matrix and the induced contrast operator are constructed internally.
• Added examples showing how to call WZ2026.GLHTBF.2cNRT() for omnibus and rank-one GLHTBF analyses using grouped data.
• Updated the method inventory so that the newly added GLHTBF routines are listed under the correct problem class and calibration family.
• Updated references to the proposed statistic TNEW so that its callable routine is consistently recorded as WZ2026.GLHTBF.2cNRT().
• Expanded package references and documentation entries for the newly added GLHTBF procedures.

Improvements

• Standardized input validation for the new GLHTBF routines, including checks on the groupwise data list Y, contrast matrix G, sample-size vector n, and common dimension p.
• Improved consistency of returned approximation fields across GLHTBF normal-reference routines.
• Kept the new F-type implementation inversion-free and compatible with HDLSS settings where the sample covariance matrices may be singular.
• Updated exported functions, manual pages, examples, and package metadata for the new GLHTBF additions.
• Improved consistency between the package description, method inventory, documentation, and simulation workflow.

Bug fixes

• Fixed minor documentation inconsistencies in GLHTBF examples and method descriptions.
• Fixed possible mismatches between newly exported GLHTBF function names, examples, and test scripts.
• Corrected minor typographical and formatting issues in package documentation.

HDNRA 2.0.1

• Renamed the function BS1996.TS.NART to BS1996.TS.NABT for consistency.
• Fixed several typos in the documentation and function names.
• Corrected several typographical errors in the documentation and function names to improve clarity and usability.
• Replaced deprecated save-always with actions/cache@v3 in GitHub Actions workflow.
• Enhanced GitHub Actions performance by implementing caching for R package dependencies, leading to faster build times and improved CI efficiency.

HDNRA 2.0.0

• The function name has been changed. The title, example, output format of the function have been changed, and some typos have also been corrected.

• Added a helper function to ‘HDNRA.cpp’ file in order to improve computation speed; updated the code of all functions to facilitate faster computation.

• Added an ‘NRtest.object’ to output an S3 class ‘NRtest’ for our package, and also constructed a corresponding print function to output the appropriate format.

• Added a ‘zzz.R’ file to manage package startup messages and initialization.

HDNRA 1.0.0

• Initial CRAN submission.