SPEC.md

PHP NDArray Library Specification

1. Project Overview

1.1 Project Name

NDArray PHP - A high-performance N-dimensional array library for PHP, backed by Rust

1.2 Purpose

To provide PHP developers with a NumPy-like interface for efficient numerical computing, scientific computing, and data manipulation, leveraging Rust's performance and safety through FFI.

1.3 Core Goals

• Performance: Near-native performance through Rust implementation
• Familiarity: NumPy-inspired API for easy adoption
• Safety: Proper memory management across FFI boundary
• Flexibility: Support for various data types, memory layouts, and operations
• Optional BLAS: High-performance linear algebra through optional BLAS integration

2. Core Array Types and Properties

2.1 NDArray Class (REQ-2.1)

Priority: CRITICAL

The primary array type representing N-dimensional arrays.

Requirements:

• 2.1.1: Support for arbitrary dimensions (1D, 2D, 3D, ..., ND)
• 2.1.2: Support for dynamic dimension sizing
• 2.1.3: Efficient memory layout (row-major by default)
• 2.1.4: Type-safe element access
• 2.1.5: Automatic reference counting for memory management

2.2 Supported Data Types (REQ-2.2)

Priority: CRITICAL

Requirements:

• 2.2.1: int8, int16, int32, int64 - Signed integers
• 2.2.2: uint8, uint16, uint32, uint64 - Unsigned integers
• 2.2.3: float32, float64 - Floating-point numbers
• 2.2.4: bool - Boolean values
• 2.2.5: complex64, complex128 - Complex numbers (future)

2.3 Array Properties (REQ-2.3)

Priority: CRITICAL

Requirements:

• 2.3.1: shape: Tuple/array representing dimensions
• 2.3.2: ndim: Number of dimensions
• 2.3.3: size: Total number of elements
• 2.3.4: dtype: Data type of elements
• 2.3.5: strides: Byte-steps in each dimension
• 2.3.6: itemsize: Size of each element in bytes
• 2.3.7: nbytes: Total bytes consumed
• 2.3.8: flags: Memory layout flags (C_CONTIGUOUS, F_CONTIGUOUS, etc.)

3. Array Creation

3.1 Basic Creation Functions (REQ-3.1)

Priority: CRITICAL

Requirements:

• 3.1.1: NDArray::array($data, $dtype = null) - From PHP array
• 3.1.2: NDArray::zeros($shape, $dtype = 'float64') - Array of zeros
• 3.1.3: NDArray::ones($shape, $dtype = 'float64') - Array of ones
• 3.1.4: NDArray::full($shape, $fill_value, $dtype = null) - Filled array
• 3.1.5: NDArray::empty($shape, $dtype = 'float64') - Uninitialized array
• 3.1.6: NDArray::eye($n, $m = null, $k = 0, $dtype = 'float64') - Identity matrix

3.2 Range and Sequence Creation (REQ-3.2)

Priority: HIGH

Requirements:

• 3.2.1: NDArray::arange($start, $stop = null, $step = 1, $dtype = null) - Evenly spaced values
• 3.2.2: NDArray::linspace($start, $stop, $num = 50, $endpoint = true, $dtype = null) - Linear spacing
• 3.2.3: NDArray::logspace($start, $stop, $num = 50, $base = 10.0, $dtype = null) - Logarithmic spacing
• 3.2.4: NDArray::geomspace($start, $stop, $num = 50, $dtype = null) - Geometric spacing

3.3 Random Array Creation (REQ-3.3)

Priority: HIGH

Requirements:

• 3.3.1: NDArray::random($shape, $dtype = 'float64') - Uniform [0, 1)
• 3.3.2: NDArray::randomInt($low, $high, $shape, $dtype = 'int64') - Random integers
• 3.3.3: NDArray::randn($shape, $dtype = 'float64') - Standard normal distribution
• 3.3.4: NDArray::normal($mean, $std, $shape, $dtype = 'float64') - Normal distribution
• 3.3.5: NDArray::uniform($low, $high, $shape, $dtype = 'float64') - Uniform distribution

3.4 Like Functions (REQ-3.4)

Priority: MEDIUM

Requirements:

• 3.4.1: NDArray::zerosLike($array) - Zeros with same shape/dtype
• 3.4.2: NDArray::onesLike($array) - Ones with same shape/dtype
• 3.4.3: NDArray::fullLike($array, $value) - Filled with same shape/dtype
• 3.4.4: NDArray::emptyLike($array) - Empty with same shape/dtype (not planned)

4. Indexing and Slicing

4.1 Basic Indexing (REQ-4.1)

Priority: CRITICAL

Requirements:

• 4.1.1: Integer indexing: $array[0], $array[1, 2]
• 4.1.2: Negative indexing: $array[-1]
• 4.1.3: Multi-dimensional indexing: $array[0, 1, 2]
• 4.1.4: Return scalar for single element, NDArray for sub-array

4.2 Slicing (REQ-4.2)

Priority: CRITICAL

Requirements:

• 4.2.1: Basic slice: $array->slice([':']) or $array['0:5']
• 4.2.2: Step slicing: $array->slice(['::2'])
• 4.2.3: Multi-dimensional slicing: $array->slice([':', '0:5'])
• 4.2.4: Negative indices in slices: $array->slice(['-5:'])
• 4.2.5: Ellipsis support: $array->slice(['...', 0])

4.3 Advanced Indexing (REQ-4.3)

Priority: MEDIUM

Requirements:

• 4.3.1: Boolean indexing: $array->at($bool_array)
• 4.3.2: Integer array indexing: $array->at([0, 2, 4])
• 4.3.3: Fancy indexing with multiple arrays
• 4.3.4: Return views where possible, copies when necessary
• 4.3.5: Flat scalar indexing write: $array->setAt($flatIndex, $value)
• 4.3.6: Gather by flat indices: $array->take($indices, $axis = null)
• 4.3.7: Gather along axis: $array->takeAlongAxis($indices, $axis)
• 4.3.8: Conditional select: NDArray::where($condition, $x, $y)

4.4 Assignment (REQ-4.4)

Priority: CRITICAL

Requirements:

• 4.4.1: Scalar assignment: $array[0] = 5
• 4.4.2: Array assignment: $array['0:5'] = $other_array
• 4.4.3: Broadcast assignment: $array[':', 0] = 5
• 4.4.4: Boolean mask assignment: $array->set($mask, $value)
• 4.4.5: Scatter write by flat indices: $array->put($indices, $values, $mode = 'raise')
• 4.4.6: Scatter write along axis: $array->putAlongAxis($indices, $values, $axis)
• 4.4.7: Accumulating scatter add: $array->scatterAdd($indices, $updates)

5. Array Views and Copies

5.1 View Creation (REQ-5.1)

Priority: HIGH

Requirements:

• 5.1.1: Slicing operations return views by default
• 5.1.2: $array->view() - Create explicit view
• 5.1.3: Views share data with parent array
• 5.1.4: Modifications to views affect parent
• 5.1.5: Proper reference counting for view lifetime

5.2 Copy Operations (REQ-5.2)

Priority: HIGH

Requirements:

• 5.2.1: $array->copy() - Deep copy
• 5.2.2: $array->astype($dtype) - Copy with type conversion
• 5.2.3: Copies are independent of parent
• 5.2.4: Contiguous memory for copies

5.3 Memory Layout Flags (REQ-5.3)

Priority: MEDIUM

Requirements:

• 5.3.1: isContiguous() - Check if C-contiguous
• 5.3.2: isFortranContiguous() - Check if Fortran-contiguous
• 5.3.3: asContiguousArray() - Return C-contiguous copy
• 5.3.4: asFortranArray() - Return Fortran-contiguous copy

6. Shape Manipulation

6.1 Reshaping (REQ-6.1)

Priority: CRITICAL

Requirements:

• 6.1.1: $array->reshape($new_shape) - Change shape (view if possible)
• 6.1.2: Support for -1 in shape (infer dimension)
• 6.1.3: $array->flatten() - 1D view/copy
• 6.1.4: $array->ravel() - 1D view if possible
• 6.1.5: Maintain element count constraint

6.2 Transposition (REQ-6.2)

Priority: HIGH

Requirements:

• 6.2.1: $array->transpose($axes = null) - Permute dimensions
• 6.2.2: $array->T property - 2D transpose shorthand
• 6.2.3: $array->swapaxes($axis1, $axis2) - Swap two axes
• 6.2.4: $array->moveaxis($source, $destination) - Move axis position
• 6.2.5: $array->permuteAxes($axes) - Explicit axis permutation

6.3 Dimension Manipulation (REQ-6.3)

Priority: HIGH

Requirements:

• 6.3.1: $array->squeeze($axis = null) - Remove single-dimensional axes
• 6.3.2: $array->expandDims($axis) - Add dimension
• 6.3.3: NDArray::newaxis constant for adding dimensions
• 6.3.4: $array->broadcast_to($shape) - Broadcast to shape
• 6.3.5: $array->insertAxis($axis) - Insert new axis at position
• 6.3.6: $array->mergeAxes($take, $into) - Merge two axes together
• 6.3.7: $array->invertAxis($axis) - Reverse stride of axis
• 6.3.8: $array->pad($padWidth, $mode = PadMode::Constant, $constantValues = 0) - Pad array
• 6.3.9: Pad modes enum via PadMode::{Constant, Symmetric, Reflect}

6.4 Joining and Splitting (REQ-6.4)

Priority: MEDIUM

Requirements:

• 6.4.1: NDArray::concatenate($arrays, $axis = 0) - Join arrays
• 6.4.2: NDArray::stack($arrays, $axis = 0) - Stack arrays along new axis
• 6.4.3: NDArray::vstack($arrays) - Vertical stack
• 6.4.4: NDArray::hstack($arrays) - Horizontal stack
• 6.4.5: $array->split($indices_or_sections, $axis = 0) - Split array
• 6.4.6: $array->vsplit($indices) - Vertical split
• 6.4.7: $array->hsplit($indices) - Horizontal split

7. Mathematical Operations

7.1 Element-wise Arithmetic (REQ-7.1)

Priority: CRITICAL

Requirements:

• 7.1.1: Addition: $a + $b, $a->add($b)
• 7.1.2: Subtraction: $a - $b, $a->subtract($b)
• 7.1.3: Multiplication: $a * $b, $a->multiply($b)
• 7.1.4: Division: $a / $b, $a->divide($b)
• 7.1.5: Floor division: $a->floorDivide($b)
• 7.1.6: Modulo: $a % $b, $a->mod($b)
• 7.1.7: Power: $a->power($b), $a ** $b
• 7.1.8: Negation: -$a, $a->negative()

7.2 Element-wise Comparison (REQ-7.2)

Priority: HIGH

Requirements:

• 7.2.1: Equal: $a == $b, $a->eq($b)
• 7.2.2: Not equal: $a != $b, $a->ne($b)
• 7.2.3: Greater: $a > $b, $a->gt($b)
• 7.2.4: Greater or equal: $a >= $b, $a->gte($b)
• 7.2.5: Less: $a < $b, $a->lt($b)
• 7.2.6: Less or equal: $a <= $b, $a->lte($b)

7.3 Element-wise Mathematical Functions (REQ-7.3)

Priority: HIGH

Requirements:

• 7.3.1: $array->abs() - Absolute value
• 7.3.2: $array->sqrt() - Square root
• 7.3.3: $array->exp() - Exponential
• 7.3.4: $array->log() - Natural logarithm
• 7.3.5: $array->log10() - Base-10 logarithm
• 7.3.6: $array->log2() - Base-2 logarithm
• 7.3.7: $array->sin(), cos(), tan() - Trigonometric
• 7.3.8: $array->arcsin(), arccos(), arctan() - Inverse trigonometric
• 7.3.9: $array->sinh(), cosh(), tanh() - Hyperbolic
• 7.3.10: $array->floor(), ceil(), round() - Rounding
• 7.3.11: $array->sign() - Sign function
• 7.3.12: $array->clip($min, $max) - Clip values
• 7.3.13: $array->sigmoid() - Sigmoid: 1 / (1 + exp(-x))
• 7.3.14: $array->softmax($axis = -1) - Softmax along axis (numerically stable)

7.4 Broadcasting (REQ-7.4)

Priority: CRITICAL

Requirements:

• 7.4.1: Automatic broadcasting for binary operations
• 7.4.2: Broadcasting rules compatible with NumPy
• 7.4.3: Support for scalar broadcasting
• 7.4.4: Error on incompatible shapes
• 7.4.5: $array->broadcastTo($shape) for explicit broadcasting

8. Reduction Operations

8.1 Statistical Reductions (REQ-8.1)

Priority: HIGH

Requirements:

• 8.1.1: $array->sum($axis = null, $keepdims = false) - Sum
• 8.1.2: $array->mean($axis = null, $keepdims = false) - Mean
• 8.1.3: $array->std($axis = null, $ddof = 0, $keepdims = false) - Standard deviation
• 8.1.4: $array->var($axis = null, $ddof = 0, $keepdims = false) - Variance
• 8.1.5: $array->min($axis = null, $keepdims = false) - Minimum
• 8.1.6: $array->max($axis = null, $keepdims = false) - Maximum
• 8.1.7: $array->median($axis = null, $keepdims = false) - Median
• 8.1.8: $array->quantile($q, $axis = null, $keepdims = false) - Quantile

8.2 Logical Reductions (REQ-8.2)

Priority: MEDIUM

Requirements:

• 8.2.1: $array->all($axis = null, $keepdims = false) - All true
• 8.2.2: $array->any($axis = null, $keepdims = false) - Any true

8.3 Index-based Operations (REQ-8.3)

Priority: MEDIUM

Requirements:

• 8.3.1: $array->argmin($axis = null) - Index of minimum
• 8.3.2: $array->argmax($axis = null) - Index of maximum
• 8.3.3: $array->argsort($axis = -1, $kind = SortKind::QuickSort) - Indices that would sort
• 8.3.4: $array->sort($axis = -1, $kind = SortKind::QuickSort) - Return sorted copy
• 8.3.5: $array->nonzero() - Indices of non-zero elements
• 8.3.6: $array->bincount($minlength = null) - Histogram of non-negative integer values
• 8.3.7: $array->topk($k, $axis = -1, $largest = true, $sorted = true, $kind = SortKind::QuickSort) - Top-k values and indices

Sort kind selection is enum-based via SortKind:

• SortKind::QuickSort
• SortKind::MergeSort
• SortKind::HeapSort
• SortKind::Stable

8.4 Cumulative Operations (REQ-8.4)

Priority: MEDIUM

Requirements:

• 8.4.1: $array->cumsum($axis = null) - Cumulative sum
• 8.4.2: $array->cumprod($axis = null) - Cumulative product

9. Linear Algebra

9.1 Matrix Operations (REQ-9.1)

Priority: HIGH

Requirements:

• 9.1.1: $a->dot($b) - Dot product / matrix multiplication
• 9.1.2: $a->matmul($b) - Matrix multiplication (@ operator support)
• 9.1.3: $array->trace() - Sum of diagonal elements
• 9.1.4: $array->diagonal($offset = 0) - Extract diagonal

9.2 Matrix Decompositions (REQ-9.2)

Priority: MEDIUM (requires BLAS)

Requirements:

• 9.2.1: $array->svd($full_matrices = true) - Singular value decomposition
• 9.2.2: $array->qr() - QR decomposition
• 9.2.3: $array->cholesky() - Cholesky decomposition
• 9.2.4: $array->eig() - Eigenvalue decomposition
• 9.2.5: $array->lu() - LU decomposition

9.3 Matrix Properties (REQ-9.3)

Priority: MEDIUM (requires BLAS)

Requirements:

• 9.3.1: $array->det() - Determinant
• 9.3.2: $array->inv() - Matrix inverse
• 9.3.3: $array->pinv() - Pseudo-inverse
• 9.3.4: $array->norm($ord = null, $axis = null, $keepdims = false) - Matrix/vector norm (supports 1, 2, inf, -inf, fro)
• 9.3.5: $array->cond($p = null) - Condition number
• 9.3.6: $array->rank() - Matrix rank

9.4 Solving Linear Systems (REQ-9.4)

Priority: MEDIUM (requires BLAS)

Requirements:

• 9.4.1: NDArray::solve($a, $b) - Solve linear equations
• 9.4.2: NDArray::lstsq($a, $b) - Least squares solution

10. Iteration and Application

10.1 Iterator Support (REQ-10.1)

Priority: HIGH

Requirements:

• 10.1.1: Implement PHP Iterator interface for 1D iteration
• 10.1.2: $array->nditer() - N-dimensional iterator
• 10.1.3: Iterate over first axis by default
• 10.1.4: Support for foreach loops
• 10.1.5: Efficient iteration without copying

10.2 Functional Operations (REQ-10.2)

Priority: MEDIUM

Requirements:

• 10.2.1: $array->apply($callable) - Apply function element-wise
• 10.2.2: $array->map($callable) - Map function (returns new array)
• 10.2.3: $array->reduce($callable, $initial = null) - Reduce operation

11. Type Conversion and I/O

11.1 Type Conversion (REQ-11.1)

Priority: HIGH

Requirements:

• 11.1.1: $array->astype($dtype) - Convert data type
• 11.1.2: $array->toArray() - Convert to PHP array
• 11.1.3: $array->toScalar() - Convert 0-d array to scalar
• 11.1.4: Automatic type inference from PHP arrays
• 11.1.5: $array->itemsize() - Bytes per element
• 11.1.6: $array->nbytes() - Total bytes in array/view
• 11.1.7: $array->intoBuffer($buffer, $maxElements = null) - Bulk copy to caller C buffer
• 11.1.8: $array->toBytes() - Raw bytes in C-order

11.2 Serialization (REQ-11.2)

Priority: MEDIUM

Requirements:

• 11.2.1: $array->save($filename) - Save to binary file (.npy format)
• 11.2.2: NDArray::load($filename) - Load from binary file
• 11.2.3: $array->toString() - String representation
• 11.2.4: $array->toJson() - JSON serialization
• 11.2.5: Support for serialize() / unserialize()

12. BLAS Integration

12.1 Optional BLAS Backend (REQ-12.1)

Priority: MEDIUM

Requirements:

• 12.1.1: Configurable at runtime (not compile-time requirement)
• 12.1.2: Graceful fallback to pure Rust implementation
• 12.1.3: Support for OpenBLAS, Intel MKL, Apple Accelerate
• 12.1.4: Configuration via: NDArray::enableBLAS($backend = 'openblas')
• 12.1.5: Query BLAS status: NDArray::blasEnabled()

12.2 BLAS-Accelerated Operations (REQ-12.2)

Priority: MEDIUM

Requirements:

• 12.2.1: Matrix multiplication (GEMM)
• 12.2.2: Dot products (DOT)
• 12.2.3: Matrix-vector operations (GEMV)
• 12.2.4: Linear system solving
• 12.2.5: Eigenvalue/SVD decompositions

13. Memory Management

13.1 Memory Allocation (REQ-13.1)

Priority: CRITICAL

Requirements:

• 13.1.1: Rust handles all memory allocation
• 13.1.2: PHP holds opaque pointer to Rust data
• 13.1.3: Reference counting for shared data (views)
• 13.1.4: Automatic cleanup on PHP object destruction
• 13.1.5: Manual cleanup via $array->dispose() if needed

13.2 Memory Safety (REQ-13.2)

Priority: CRITICAL

Requirements:

• 13.2.1: No use-after-free errors
• 13.2.2: No double-free errors
• 13.2.3: No memory leaks
• 13.2.4: Thread-safe reference counting
• 13.2.5: Proper exception handling across FFI boundary

13.3 Memory Optimization (REQ-13.3)

Priority: MEDIUM

Requirements:

• 13.3.1: Copy-on-write where applicable
• 13.3.2: In-place operations where possible
• 13.3.3: View creation instead of copying
• 13.3.4: Memory pooling for small allocations (optional)

14. Error Handling

14.1 Exception Types (REQ-14.1)

Priority: HIGH

Requirements:

• 14.1.1: NDArrayException - Base exception class
• 14.1.2: ShapeException - Shape mismatch errors
• 14.1.3: DTypeException - Data type errors
• 14.1.4: IndexException - Indexing errors
• 14.1.5: BroadcastException - Broadcasting errors
• 14.1.6: LinearAlgebraException - Linear algebra errors

14.2 Error Propagation (REQ-14.2)

Priority: HIGH

Requirements:

• 14.2.1: Rust panics converted to PHP exceptions
• 14.2.2: Clear error messages with context
• 14.2.3: Stack traces preserved where possible
• 14.2.4: No silent failures

15. Performance Requirements

15.1 Benchmarks (REQ-15.1)

Priority: HIGH

Requirements:

• 15.1.1: Matrix multiplication: 10x faster than pure PHP
• 15.1.2: Element-wise operations: 5-10x faster than pure PHP
• 15.1.3: Reductions: 10x faster than pure PHP
• 15.1.4: FFI overhead < 5% for large arrays (>1000 elements)
• 15.1.5: Memory usage comparable to NumPy

15.2 Optimization Strategies (REQ-15.2)

Priority: HIGH

Requirements:

• 15.2.1: SIMD vectorization in Rust
• 15.2.2: Multi-threading for large operations (configurable)
• 15.2.3: Cache-friendly memory access patterns
• 15.2.4: Minimize FFI crossings
• 15.2.5: Batch operations in Rust layer

16. Testing and Quality

16.1 Test Coverage (REQ-16.1)

Priority: HIGH

Requirements:

• 16.1.1: Unit tests for all public methods
• 16.1.2: Integration tests for complex workflows
• 16.1.3: Property-based testing for mathematical correctness
• 16.1.4: Benchmark suite for performance regression
• 16.1.5: Memory leak detection tests

16.2 Compatibility Testing (REQ-16.2)

Priority: MEDIUM

Requirements:

• 16.2.1: NumPy compatibility tests (behavior comparison)
• 16.2.2: Cross-platform testing (Linux, macOS, Windows)
• 16.2.3: PHP version testing (8.1, 8.2, 8.3+)
• 16.2.4: Different BLAS backends

17. Documentation

17.1 API Documentation (REQ-17.1)

Priority: HIGH

Requirements:

• 17.1.1: PHPDoc annotations for all public APIs
• 17.1.2: User guide with examples
• 17.1.3: Migration guide from NumPy
• 17.1.4: Performance tuning guide
• 17.1.5: API reference (auto-generated)

17.2 Examples (REQ-17.2)

Priority: MEDIUM

Requirements:

• 17.2.1: Quick start tutorial
• 17.2.2: Common operations cookbook
• 17.2.3: Performance comparison examples
• 17.2.4: Advanced usage patterns
• 17.2.5: Integration examples (with Laravel, Symfony, etc.)

18. Installation and Distribution

18.1 Package Distribution (REQ-18.1)

Priority: HIGH

Requirements:

• 18.1.1: Composer package
• 18.1.2: Pre-built binaries for common platforms
• 18.1.3: Build instructions for source compilation
• 18.1.4: Docker images for development

18.2 Platform Support (REQ-18.2)

Priority: HIGH

Requirements:

• 18.2.1: Linux (x86_64, ARM64)
• 18.2.2: macOS (x86_64, Apple Silicon)
• 18.2.3: Windows (x86_64)
• 18.2.4: PHP 8.1+ with FFI extension enabled

19. Future Enhancements (Optional)

19.1 Phase 2 Features (REQ-19.1)

Priority: LOW

• 19.1.1: GPU acceleration via CUDA/ROCm
• 19.1.2: Sparse matrix support
• 19.1.3: Complex number support
• 19.1.4: FFT operations
• 19.1.5: Image processing utilities
• 19.1.6: Pandas-like DataFrame functionality

19.2 Integration Features (REQ-19.2)

Priority: LOW

• 19.2.1: Laravel facade
• 19.2.2: Symfony bundle
• 19.2.3: CSV/Excel import utilities
• 19.2.4: Database query result conversion

20. Success Criteria

20.1 Functional Criteria

• All CRITICAL priority requirements implemented
• 90%+ HIGH priority requirements implemented
• Passes all NumPy compatibility tests
• Zero critical bugs in core operations

20.2 Performance Criteria

• Meets or exceeds benchmark targets (REQ-15.1)
• Memory usage within 10% of NumPy
• FFI overhead negligible for large arrays

20.3 Quality Criteria

• 90% code coverage

• No memory leaks detected
• Zero use-after-free or double-free errors
• Clean Valgrind reports

20.4 Usability Criteria

• Complete API documentation
• 10+ example projects
• Migration guide from NumPy
• Active community engagement

Appendix A: NumPy Compatibility Matrix

Feature	NumPy Equivalent	Priority	Status
Array creation	np.array()	CRITICAL	Planned
Slicing	arr[0:5]	CRITICAL	Planned
Broadcasting	Automatic	CRITICAL	Done
Matrix mult	@ operator	HIGH	Planned
Reductions	sum(), mean()	HIGH	Planned
SVD	np.linalg.svd()	MEDIUM	Planned
FFT	np.fft.fft()	LOW	Future

Appendix B: Glossary

• FFI: Foreign Function Interface - allows PHP to call Rust code
• BLAS: Basic Linear Algebra Subprograms - optimized linear algebra routines
• View: Array that shares data with another array
• Broadcasting: Automatic shape alignment for operations
• Contiguous: Memory layout where elements are stored sequentially
• Stride: Number of bytes to step in each dimension