NEWS.md

In rbbnp: A Bias Bound Approach to Non-Parametric Inference

rbbnp 0.3.0

Major Improvements

• Added automatic bandwidth selection using cross-validation or Silverman's rule of thumb
• Improved performance of get_est_Ar() function by approximately 50%, significantly accelerating main function execution time
• Improved code structure with centralized configuration management
• Added unit tests

New Features

• Implemented automatic bandwidth selection with two methods:
• Cross-validation (h_method = "cv") for optimal accuracy
• Silverman's rule of thumb (h_method = "silverman") for faster computation
• Added create_biasBound_config() function for centralized configuration management
• Added create_kernel_functions() factory for flexible kernel function creation
• Changed default bandwidth parameter h to NULL to enable automatic selection

Performance Optimization

• Optimized get_est_Ar() function by:
• Precomputing avg_phi_log values for all frequencies
• Enhancing the optimization algorithm for determining smoothness parameters
• Increased default resolution for kernel function approximation

Bug Fixes and Edge Case Handling

• Modified get_sigma() and get_sigma_yx() to handle negative density estimates by returning 0 instead of NaN
• Enhanced get_conditional_var() to ensure non-negative variance estimates
• Fixed sinc() function to properly handle x = 0 cases
• Improved handling of NA values in Fourier transform calculations

Documentation

• Updated examples to demonstrate automatic bandwidth selection
• Improved parameter descriptions throughout the package

rbbnp documentation built on June 8, 2025, 10:46 a.m.