v0.6.2 - Windows Support & Cross-Platform Stability

This release introduces full Windows compatibility with Microsoft Visual C++ (MSVC) while maintaining complete backward compatibility with macOS and Linux builds.

🔧 Windows Support Enhancements
• Full MSVC Compatibility: Native support for Visual Studio 2022 and MSVC compiler
• Windows-Specific SIMD Detection: Automatic AVX2/AVX/SSE2 optimization on Windows x64
• Platform-Agnostic Build System: Conditional compilation ensures zero impact on Unix/Linux/macOS builds
• Comprehensive Documentation: New docs/Windows_Support.md with setup and build instructions

⚡ Performance & Optimization
• Enhanced SIMD Detection: Improved runtime CPU feature detection across all platforms
• Cross-Platform Memory Management: Platform-specific aligned allocation (_aligned_malloc on Windows, std::aligned_alloc on Unix)
• Compiler Warning Cleanup: Resolved MSVC warnings while maintaining compatibility

🏗️ Build System Improvements
• CMake Windows Integration: Native Windows library naming and symbol export
• Flexible Dependency Detection: Automatic TBB and GTest detection on Windows via find_package
• Cross-Compilation Support: Environment variable overrides for SIMD settings

🔄 Maintained Compatibility
• Zero Breaking Changes: All existing macOS/Linux code unchanged
• Active Development Branch: windows-support branch remains available for ongoing Windows-specific development
• Version Synchronization: Updated version numbers across CMakeLists.txt and libstats.h

📦 What's New Since v0.6.1
• Version number consistency updates
• Final Windows compatibility stabilization
• Build system refinements

This release enables libstats to be used across Windows, macOS, and Linux development environments.

Full Changelog: v0.6.0...v0.6.2

v0.6.0 - Major Refactoring of Plaform-Independent Code

🏗️ Structure Cleanup (COMPLETED)

• Split monolithic distribution_base.h (1400+ lines → 326 lines)
• Reorganized constants.h (600+ lines → 39 focused files)
• Extracted template implementations to reduce compilation time
• Created focused module structure with clear separation of concerns

🚀 Performance Optimizations

• Fixed redundant cache management in uniform distribution parallel methods
• Eliminated ~13x overhead: parallel methods now have ~0% overhead vs SIMD
• Optimized thread-safe parameter caching across all distributions
• Enhanced SIMD dispatch with runtime CPU detection

Code Quality Improvements

• Eliminated all compilation warnings with [[maybe_unused]] attributes
• Standardized error handling patterns across distribution classes
• Improved memory management with focused allocation utilities
• Enhanced thread safety with simplified synchronization patterns

📁 New Modular Structure

Core constants now organized into focused files:

• mathematical_constants.h - π, e, mathematical values
• precision_constants.h - Tolerances, epsilons, thresholds
• statistical_constants.h - Critical values, significance levels
• probability_constants.h - Probability bounds and safety limits
• benchmark_constants.h - Performance testing parameters
• robust_constants.h - Robust estimation parameters
• statistical_methods_constants.h - Bayesian, bootstrap, cross-validation
• goodness_of_fit_constants.h - Critical values for goodness-of-fit tests

Core infrastructure split into focused modules:

• distribution_interface.h - Pure virtual interface definitions
• distribution_cache.h - Caching infrastructure (platform/)
• distribution_memory.h - Memory management utilities
• distribution_validation.h - Validation structures and utilities

📈 Performance Metrics

• 20-30% faster compilation times through header modularization
• Zero performance overhead for parallel uniform distribution methods
• Eliminated double cache validation/locking across all parallel operations
• Improved SIMD dispatch efficiency with centralized CPU detection

✅ Quality Assurance

• All 34 tests pass successfully
• Zero compilation warnings across all platforms
• Full backward compatibility maintained
• Comprehensive performance verification completed

📚 Documentation

• Added comprehensive refactoring plan (core_platform-independent_improvements.md)
• Enhanced code comments and architectural documentation
• Improved namespace organization for better usability
• Updated API examples with modern C++20 patterns

v0.5.0 - Core Functionality Complete

Major optimization phase milestone featuring:

• Comprehensive header reorganization and modularization
• Advanced SIMD optimization with runtime dispatch
• Parallel execution framework with adaptive thresholds
• Work-stealing thread pools and adaptive caching
• Consolidated distribution implementations with full test coverage
• Enhanced documentation and API simplification roadmap
• Robust build system with diagnostic tools

This release completes the core optimization infrastructure and sets
the foundation for API simplification leading to v1.0.0 stable release.

Key Performance Improvements:

• SIMD-optimized batch operations
• Parallel execution with work distribution
• Memory-efficient caching with predictive eviction
• Enhanced numerical precision and stability

Platform Support: macOS, Linux, Windows with C++20 compliance
Zero external dependencies maintained with thread-safe operations.

v0.4.0 - Improve distribution implementations with performance optimizations

Key Features:

• Complete Uniform, Discrete, Gaussian, and Exponential distributions
• SIMD-optimized batch operations with runtime dispatch
• Work-stealing and cache-aware parallel processing
• Comprehensive performance benchmarks with correctness verification
• Production-ready thread-safe implementations

Performance Highlights:

• Fixed critical uniform distribution performance regression (17,365% improvement)
• Consistent parallelization thresholds across all distributions
• Optimized cache management with intelligent grain sizing
• Full coverage of PDF, LogPDF, and CDF operations

Testing & Quality:

• Comprehensive test suites with statistical validation
• Performance benchmarks with automated correctness checks
• Thread safety verification and stress testing
• Complete API consistency across all distributions

v0.3.2 - Fix compiler warnings for unused functions on Apple platforms

Wrap detect_cache_info and detect_topology_info functions in #if !defined(APPLE)
guards since they are unused on Apple platforms due to using sysctl instead

• Resolves compiler warnings without affecting functionality on any platform

v0.3.1 - Fix Intel Mac compilation and linking issues

This release fixes critical compilation and linking issues that affected Intel Macs with Homebrew LLVM when building v0.3.0:

• Fixed missing forward declaration for detect_macos_topology() function
• Added proper Homebrew LLVM libc++ linking for shared library
• Ensures compatibility across all macOS platforms (Intel & Apple Silicon)

v0.3.0 - Modern macOS Thread Optimization

🚀 Key Features:

• Modern QoS-based thread optimization for macOS
• 1.67x-5.24x performance improvements on large datasets
• Enhanced cross-platform compatibility
• Fixed CMake duplicate library warnings
• 100% test pass rate (27/27 tests)

🎯 Performance:

• Work-stealing pool: 67%+ steal success rate
• Parallel speedups: Up to 5.24x on computational workloads
• Clean QoS integration with macOS scheduler
• Zero performance regressions

🔧 Technical Highlights:

• Replace deprecated thread affinity with pthread_set_qos_class_self_np()
• Architecture-aware testing (Apple Silicon L3 cache handling)
• Enhanced SIMD detection and runtime dispatch
• Improved build system with better GTest integration

✅ Compatibility:

• macOS Intel/Apple Silicon optimized
• Linux x86_64/ARM64 with traditional pthread affinity
• Windows x64 with Windows affinity APIs
• Graceful fallbacks for other platforms

v0.2.2 - Fix build issues and resolve AVX2 SIMD race conditions in Exponential distribution

• Fix AVX2 code compilation errors and SIMD feature conflicts
• Resolve race conditions in Exponential distribution SIMD implementations
• Add proper thread safety for SIMD dispatch in exponential calculations
• Mark unused test variables with [[maybe_unused]] to eliminate warnings
• Improve SIMD runtime detection and fallback mechanisms
• Enhanced numerical stability in vectorized exponential computations
• Update build system to handle SIMD feature detection more reliably
• All tests passing with clean warning-free build

v0.2.1 - Fix SIMD detection missing from repository

This patch release fixes critical CMake configuration issues when cloning the repository by:

• Adding missing cmake/SIMDDetection.cmake file with detect_simd_features() function
• Fixing overly broad .gitignore rule that excluded important cmake modules
• Ensuring SIMD detection system works properly on fresh clones

This resolves build failures for users cloning the repository for the first time.

v0.2.0 - Exponential Distribution and Code Consolidation in Tests/Examples

This release represents increased maturation of the libstats library with:

🎯 Production-Ready Examples:

• Clear, meaningful file names (statistical_validation_demo, parallel_execution_demo)
• Comprehensive SIMD and parallel performance benchmarking
• All examples build and run correctly
• Professional documentation alignment

⚡ Enhanced Performance:

• Complete exponential distribution implementation with SIMD support
• Advanced parallel execution demonstrations
• Comprehensive mathematical validation tools
• Production-grade benchmarking capabilities

🛠️ Code Quality:

• Removed obsolete documentation and test files
• Consolidated test architecture with clear naming
• Enhanced error handling and parameter validation
• Version consistency across all components

📚 Documentation Excellence:

• README accurately reflects actual functionality
• Examples properly categorized (Core Usage vs Performance)
• Clear build instructions and usage patterns
• Professional presentation throughout