Array fundamentals

Overview

This page introduces the NDArray model, its core properties, and the basic arithmetic and indexing workflow you will use throughout Multik.

NDArray essentials

An NDArray is Multik's core data type for dense, homogeneous numeric data. It is defined by four key properties:

• Dimension (dim) — number of axes, such as D1, D2, D3, D4, or DN.

• Shape (shape) — sizes of each axis, for example (2, 3).

• Size (size) — total number of elements.

• DType (dtype) — element type, such as Int, Double, or complex types.

These properties help keep computations correct and predictable.

Creating arrays

For quick creation, use mk[] with mk.ndarray and let Multik infer type and dimension:

You can also use factory methods like zeros, ones, and identity. For a deeper overview, see Creating multidimensional arrays.

Inspecting properties

Element-wise arithmetic

Arithmetic is element-wise by default:

When operating on two arrays, their shapes must match. For fixed dimensions (D1-D4), dimension mismatch is caught at compile time; shape mismatch is checked at runtime.

In-place operations

Operations like +=, -=, *=, and /= modify the current array:

Indexing and slicing

Indexing uses zero-based indices and works per axis:

For ranges, slices, and multi-axis selection, see Indexing and slicing.