Skip to main content

​
transpose

Permute the dimensions of an array. When called without axes, reverses the dimension order. Returns a view (no data is copied). Also available as np.linalg.transpose(...). 
function transpose(a: ArrayLike, axes?: number[]): NDArray
NameTypeDefaultDescription
aArrayLike—Input array.
axesnumber[]undefinedPermutation of [0, 1, ..., N-1] where N is the number of dimensions. If omitted, the axes are reversed.
Returns: NDArray — View of a with its axes permuted. 
import * as np from 'numpy-ts';

const a = np.array([[1, 2, 3], [4, 5, 6]]);
console.log(a.shape);  // [2, 3]

// Default: reverse all axes
const b = np.transpose(a);
console.log(b.shape);  // [3, 2]
console.log(b.toArray());
// [[1, 4],
//  [2, 5],
//  [3, 6]]

// Shorthand via property
const c = a.T;
console.log(c.shape);  // [3, 2]

// Custom permutation on a 3-D array
const d = np.zeros([2, 3, 4]);
const e = np.transpose(d, [1, 2, 0]);
console.log(e.shape);  // [3, 4, 2]

​
swapaxes

Interchange two axes of an array. Returns a view. 
function swapaxes(a: ArrayLike, axis1: number, axis2: number): NDArray
NameTypeDefaultDescription
aArrayLike—Input array.
axis1number—First axis.
axis2number—Second axis.
Returns: NDArray — View of a with the two specified axes swapped. 
import * as np from 'numpy-ts';

const a = np.zeros([2, 3, 4]);
console.log(a.shape);  // [2, 3, 4]

const b = np.swapaxes(a, 0, 2);
console.log(b.shape);  // [4, 3, 2]

// Swapping the same axis returns an unchanged view
const c = np.swapaxes(a, 1, 1);
console.log(c.shape);  // [2, 3, 4]

​
moveaxis

Move axes of an array to new positions. Other axes remain in their original order. Returns a view. 
function moveaxis(
  a: ArrayLike,
  source: number | number[],
  destination: number | number[]
): NDArray
NameTypeDefaultDescription
aArrayLike—Input array.
sourcenumber | number[]—Original position(s) of the axes to move.
destinationnumber | number[]—Target position(s) for each axis.
Returns: NDArray — View of a with the specified axes moved. 
import * as np from 'numpy-ts';

const a = np.zeros([3, 4, 5]);

// Move axis 0 to position 2
const b = np.moveaxis(a, 0, 2);
console.log(b.shape);  // [4, 5, 3]

// Move multiple axes
const c = np.moveaxis(a, [0, 2], [2, 0]);
console.log(c.shape);  // [5, 4, 3]

​
rollaxis

Roll the specified axis backwards until it lies at the given position. Returns a view. 
function rollaxis(a: ArrayLike, axis: number, start?: number): NDArray
NameTypeDefaultDescription
aArrayLike—Input array.
axisnumber—The axis to roll.
startnumber0The axis is rolled until it lies before this position.
Returns: NDArray — View of a with the axis rolled to the target position. 
import * as np from 'numpy-ts';

const a = np.zeros([3, 4, 5, 6]);

// Roll axis 3 to position 1
const b = np.rollaxis(a, 3, 1);
console.log(b.shape);  // [3, 6, 4, 5]

// Roll axis 2 to position 0
const c = np.rollaxis(a, 2);
console.log(c.shape);  // [5, 3, 4, 6]

​
permute_dims

Permute the dimensions of an array. This is the Array API-compatible equivalent of transpose. Also available as np.linalg.permute_dims(...). 
function permute_dims(a: ArrayLike, axes?: number[]): NDArray
NameTypeDefaultDescription
aArrayLike—Input array.
axesnumber[]—Permutation of dimensions.
Returns: NDArray — View of a with its axes permuted. 
import * as np from 'numpy-ts';

const a = np.zeros([2, 3, 4]);

const b = np.permute_dims(a, [2, 0, 1]);
console.log(b.shape);  // [4, 2, 3]

​
matrix_transpose

Transpose the last two dimensions of an array. The input must have at least 2 dimensions. Returns a view. Also available as np.linalg.matrix_transpose(...). 
function matrix_transpose(a: ArrayLike): NDArray
NameTypeDefaultDescription
aArrayLike—Input array with ndim >= 2.
Returns: NDArray — View of a with the last two axes swapped. 
import * as np from 'numpy-ts';

// 2-D case: same as transpose
const a = np.array([[1, 2, 3], [4, 5, 6]]);
const b = np.matrix_transpose(a);
console.log(b.shape);  // [3, 2]

// Batch of matrices (3-D): transposes each matrix in the batch
const c = np.zeros([5, 3, 4]);
const d = np.matrix_transpose(c);
console.log(d.shape);  // [5, 4, 3]