Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

v1.0.0 - 2026-03-01

Added

• Initial release of NDArray PHP - High-performance N-dimensional arrays for PHP powered by Rust via FFI

Array Creation

• NDArray::array() - Create from PHP nested arrays
• NDArray::zeros(), NDArray::ones(), NDArray::full() - Fill value factories
• NDArray::empty() - Create empty arrays for preallocation
• NDArray::fromBuffer() - Import from external FFI buffers (e.g., audio/image libraries)
• NDArray::zerosLike(), NDArray::onesLike(), NDArray::fullLike() - Shape-based factories
• NDArray::eye() - Identity matrix creation
• NDArray::arange(), NDArray::linspace(), NDArray::logspace(), NDArray::geomspace() - Sequence generation
• NDArray::random(), NDArray::randomInt(), NDArray::randn() - Random number generation
• NDArray::normal(), NDArray::uniform() - Statistical distributions

Mathematical Operations

• Element-wise arithmetic: add(), subtract(), multiply(), divide(), rem(), mod()
• Unary operations: abs(), negative(), sqrt(), exp(), log(), ln()
• Trigonometric: sin(), cos(), tan(), sinh(), cosh(), tanh(), asin(), acos(), atan()
• Advanced math: cbrt(), ceil(), exp2(), floor(), log2(), log10(), pow2(), round(), signum(), recip(), ln1p(), toDegrees(), toRadians()
• Power operations: powi(), powf(), hypot()
• Clamping: clamp(), clip()
• Neural network helpers: sigmoid(), softmax()

Array Manipulation

• Shape operations: reshape(), transpose(), swapaxes(), permute(), merge()
• Dimension operations: flatten(), ravel(), squeeze(), expandDims(), insert()
• Axis operations: flip(), pad()
• Repetition: tile(), repeat()

Slicing and Indexing

• slice() - Advanced slicing with step support
• ArrayAccess: $arr[0], $arr['0:5'], $arr['0,1']
• Indexing methods: get(), set(), getAt(), setAt()
• Advanced indexing: take(), takeAlongAxis(), put(), putAlongAxis(), scatterAdd()

Reductions and Statistics

• Basic: sum(), mean(), product(), min(), max()
• Variance/Std: var(), std() with ddof support
• Argmin/Argmax: argmin(), argmax()
• Cumulative: cumsum(), cumprod()
• Sorting: sort(), argsort()
• Search: topk(), bincount()

Linear Algebra

• dot() - Dot product
• matmul() - Matrix multiplication
• diagonal() - Extract diagonal
• trace() - Matrix trace
• norm() - Vector/matrix norms (L1, L2, etc.)

Stacking and Splitting

• concatenate(), stack(), vstack(), hstack() - Joining arrays
• split(), vsplit(), hsplit() - Splitting arrays

Comparison and Logical Operations

• Comparison: eq(), ne(), gt(), gte(), lt(), lte()
• Logical: and(), or(), not(), xor()

Bitwise Operations

• bitand(), bitor(), bitxor(), leftShift(), rightShift()

Type Conversion and Export

• toArray() - Convert to nested PHP array
• toScalar() - Extract 0D array value
• toBytes() - Raw binary export
• intoBuffer() - Export to FFI C buffer with offset/length support
• copy() - Deep copy
• astype() - Type casting

Views and Memory Management

• Zero-copy views for slicing, indexing, transpose, reshape (when contiguous)
• Automatic memory management via Rust allocation
• View chain tracking for proper lifecycle management
• isView(), isContiguous() introspection

Utilities

• flat() - Flat iterator for memory-efficient iteration
• Printing: configurable display options, __toString()
• where() - Conditional array selection

Technical Highlights

• Zero-copy views - Slicing, indexing, and many operations return views sharing memory with parent arrays
• Rust backend - Core computations performed by optimized Rust code with SIMD support
• FFI architecture - Minimal PHP/Rust boundary crossing, batch operations when possible
• Memory safety - Rust owns all data, PHP holds opaque pointers, automatic cleanup
• Type safety - Full type hints, 11 DType variants (Int8-64, UInt8-64, Float32/64, Bool)
• Comprehensive testing - 927 tests, 62,353 assertions covering all functionality
• Extensive documentation - API reference, guides, and examples

Full Changelog: https://github.com/phpmlkit/ndarray/commits/1.0.0