NEWS

dyadicMarkov 0.1.3.9000

Refined argument documentation for rOpenSci review, including explicit state-space constraints, separate documentation of the four bivariate sequence inputs, and clarification that bivariate empirical count matrices may use integer or double storage modes.

Added a sensitivity-analysis vignette reproducing the published sequence-length analysis for 1,000 simulated dyads at five sequence lengths, with shorter worked examples for 30- and 90-point sequences.
Added the fixed simulation datasets used in the sensitivity analysis and documented their structure, binary state space, historical labels, and separate simulation at each sequence length.
Clarified the terminology used in the sensitivity analysis and noted the small numerical differences from the published sensitivity and specificity results.
Documented how exact AIC ties are handled in partial and complete bivariate pattern selection.
Improved validation of the digits argument used by print, summary, and LaTeX methods.
Added a check for state-space sizes that are too large to construct the required transition-count matrices safely.
Updated the statistical-software-review annotations to reflect the legacy implementation parity tests.
Added and updated tests for the changes above.
Added base-R state-strip plot() methods for univariate pattern results and binary bivariate case and pattern results, with accessible defaults and retained sequence metadata.
Added inferential summary tables and LaTeX representations for empirical-count and MLE matrices.

Clarified that the univariate pattern-identification procedure is an LRT procedure evaluated using Pearson Chi-squared, while the global bivariate nested-model/LRT framework implements two chi-squared tests for A1 and B1, also evaluated using Pearson Chi-squared.
Clarified that local bivariate pattern selection computes the G-squared deviance before applying AIC = G^2 + 2k.
Corrected the univariate pattern-identification and global bivariate case boundaries so that p-values equal to alpha are treated as rejection (p <= alpha).
Documented that the univariate workflow supports multiple categorical states, while the bivariate workflow is defined for two dichotomous variables.
Added focused tests that distinguish Pearson's chi-squared statistic from G-squared and verify both partial and complete bivariate AIC paths.
Updated the maintainer email address and package version for this release.
Made the manual simulated-parity script stop when a comparison fails.
Declared srr as a development/documentation dependency.

Updated package wording and metadata for the CRAN submission.
Added S3 classes and print/summary support for pattern and case identification results.
Added S3 classes for empirical count matrices and MLE transition probability matrices while preserving ordinary matrix behavior.
Added two synthetic 90-point example datasets for package workflow examples.
Rewrote the workflow vignette around the built-in univariate and bivariate example datasets.
Improved internal input validation for count, estimation, and pattern-identification functions.
Updated tests and documentation for the new S3 return objects.
Improved validation for extreme state-space inputs, non-finite chain values, and malformed empirical matrices.
Refactored selected internal validation and AIC helper code to reduce function complexity while preserving exported behavior.
Improved bivariate count validation coverage for unsupported and malformed inputs.