NDVek

NDVek is a high-performance Go library for multi-dimensional array operations, powered by the vek SIMD backend. It supports both float64 and float32 data types, automatic broadcasting, and a some common operations.

Methods

Category	Method	Description
Creation	NewNdArray	Creates a new array from a shape and data ([]float64 or []float32).
	Zeros	Creates an array of zeros with the specified shape.
	Linspace	Generates linearly spaced values.
Arithmetic	Add, Subtract	Element-wise addition and subtraction (supports broadcasting).
	Multiply, Divide	Element-wise multiplication and division (supports broadcasting).
	AddScalar, SubScalar	Scalar addition and subtraction.
	MulScalar, DivScalar	Scalar multiplication and division.
Inplace Arithmetic	Add_Inplace, Subtract_Inplace	Element-wise in-place addition and subtraction (requires equal shapes).
	Multiply_Inplace, Divide_Inplace	Element-wise in-place multiplication and division (requires equal shapes).
	AddScalar_Inplace, SubScalar_Inplace	Scalar in-place addition and subtraction.
	MulScalar_Inplace, DivScalar_Inplace	Scalar in-place multiplication and division.
	Abs_Inplace, Neg_Inplace	Element-wise in-place absolute value and negation.
	Sqrt_Inplace, Round_Inplace	Element-wise in-place square root and rounding.
	Floor_Inplace, Ceil_Inplace	Element-wise in-place floor and ceil.
	CumSum_Inplace, CumProd_Inplace	In-place cumulative sum and product.
Broadcasting	ApplyOp	Applies a custom binary function with broadcasting (promotes to float64).
	ApplyHadamardOp	Applies a custom unary function element-wise (promotes to float64).
Aggregation	Sum	Sum of all elements.
	Mean	Arithmetic mean of all elements.
	Min, Max	Minimum and maximum values.
	Prod	Product of all elements.
Unary Ops	Abs	Element-wise absolute value.
	Neg	Element-wise negation.
	Sqrt	Element-wise square root.
	Round, Floor, Ceil	Element-wise rounding operations.
	CumSum	Cumulative sum.
	CumProd	Cumulative product.
Manipulation	Reshape	Changes the shape of the array.
	InsertAxis	Inserts a new axis at the specified position.
	Get	Retrieves an element at a specific index.
	Shape	Returns the shape of the array.
	DType	Returns the data type (Float32, Float64, or Bool).
Boolean Logic	Eq, Neq	Element-wise equality/inequality comparison (returns Bool array).
	Lt, Lte, Gt, Gte	Element-wise comparison (returns Bool array).
	And, Or, Xor	Element-wise logical operations (requires Bool arrays).
	Not	Element-wise logical NOT (requires Bool array).
	Any, All	Returns true if any/all elements are true (requires Bool array).

Usage Examples

Basic Arithmetic and Broadcasting

package main

import (
    "fmt"
    "github.com/mederrata/ndvek"
)

func main() {
    // Create a 2x3 matrix
    shape := []int{2, 3}
    data := []float64{1.0, 2.0, 3.0, 4.0, 5.0, 6.0}
    arr, _ := ndvek.NewNdArray(shape, data)

    // Create a scalar (broadcastable)
    scalarArr, _ := ndvek.NewNdArray([]int{1}, []float64{10.0})

    // Add scalar to every element
    result, _ := ndvek.Add(arr, scalarArr)
    
    // Result is still 2x3
    // [11.0, 12.0, 13.0]
    // [14.0, 15.0, 16.0]
    fmt.Println(result.Mean()) // 13.5
}

Boolean Logic

// Create two arrays
a, _ := ndvek.NewNdArray([]int{4}, []float64{1, 2, 3, 4})
b, _ := ndvek.NewNdArray([]int{4}, []float64{2, 2, 2, 2})

// Compare them
// Lt: [1<2, 2<2, 3<2, 4<2] -> [T, F, F, F]
lt, _ := ndvek.Lt(a, b) 

// Check if any element is true
fmt.Println(lt.Any()) // true

Note: For more comprehensive examples, please refer to the ndarray_test.go file in the source code.