pub struct ArrayBase<S, D>where
S: RawData,{ /* private fields */ }
Expand description
An n-dimensional array.
The array is a general container of elements. The array supports arithmetic operations by applying them elementwise, if the elements are numeric, but it supports non-numeric elements too.
The arrays rarely grow or shrink, since those operations can be costly. On the other hand there is a rich set of methods and operations for taking views, slices, and making traversals over one or more arrays.
In n-dimensional we include for example 1-dimensional rows or columns, 2-dimensional matrices, and higher dimensional arrays. If the array has n dimensions, then an element is accessed by using that many indices.
The ArrayBase<S, D>
is parameterized by S
for the data container and
D
for the dimensionality.
Type aliases Array
, ArcArray
, CowArray
, ArrayView
, and
ArrayViewMut
refer to ArrayBase
with different types for the data
container: arrays with different kinds of ownership or different kinds of array views.
§Contents
- Array
- ArcArray
- CowArray
- Array Views
- Indexing and Dimension
- Loops, Producers and Iterators
- Slicing
- Subviews
- Arithmetic Operations
- Broadcasting
- Conversions
- Constructor Methods for Owned Arrays
- Methods For All Array Types
- Methods For 1-D Arrays
- Methods For 2-D Arrays
- Methods for Dynamic-Dimensional Arrays
- Numerical Methods for Arrays
§Array
Array
is an owned array that owns the underlying array
elements directly (just like a Vec
) and it is the default way to create and
store n-dimensional data. Array<A, D>
has two type parameters: A
for
the element type, and D
for the dimensionality. A particular
dimensionality’s type alias like Array3<A>
just has the type parameter
A
for element type.
An example:
// Create a three-dimensional f64 array, initialized with zeros
use ndarray::Array3;
let mut temperature = Array3::<f64>::zeros((3, 4, 5));
// Increase the temperature in this location
temperature[[2, 2, 2]] += 0.5;
§ArcArray
ArcArray
is an owned array with reference counted
data (shared ownership).
Sharing requires that it uses copy-on-write for mutable operations.
Calling a method for mutating elements on ArcArray
, for example
view_mut()
or get_mut()
,
will break sharing and require a clone of the data (if it is not uniquely held).
§CowArray
CowArray
is analogous to std::borrow::Cow
.
It can represent either an immutable view or a uniquely owned array. If a
CowArray
instance is the immutable view variant, then calling a method
for mutating elements in the array will cause it to be converted into the
owned variant (by cloning all the elements) before the modification is
performed.
§Array Views
ArrayView
and ArrayViewMut
are read-only and read-write array views
respectively. They use dimensionality, indexing, and almost all other
methods the same way as the other array types.
Methods for ArrayBase
apply to array views too, when the trait bounds
allow.
Please see the documentation for the respective array view for an overview
of methods specific to array views: ArrayView
, ArrayViewMut
.
A view is created from an array using .view()
,
.view_mut()
, using
slicing (.slice()
, .slice_mut()
) or from one of
the many iterators that yield array views.
You can also create an array view from a regular slice of data not
allocated with Array
— see array view methods or their From
impls.
Note that all ArrayBase
variants can change their view (slicing) of the
data freely, even when their data can’t be mutated.
§Indexing and Dimension
The dimensionality of the array determines the number of axes, for example a 2D array has two axes. These are listed in “big endian” order, so that the greatest dimension is listed first, the lowest dimension with the most rapidly varying index is the last.
In a 2D array the index of each element is [row, column]
as seen in this
4 × 3 example:
[[ [0, 0], [0, 1], [0, 2] ], // row 0
[ [1, 0], [1, 1], [1, 2] ], // row 1
[ [2, 0], [2, 1], [2, 2] ], // row 2
[ [3, 0], [3, 1], [3, 2] ]] // row 3
// \ \ \
// column 0 \ column 2
// column 1
The number of axes for an array is fixed by its D
type parameter: Ix1
for a 1D array, Ix2
for a 2D array etc. The dimension type IxDyn
allows
a dynamic number of axes.
A fixed size array ([usize; N]
) of the corresponding dimensionality is
used to index the Array
, making the syntax array[[
i, j, …]]
use ndarray::Array2;
let mut array = Array2::zeros((4, 3));
array[[1, 1]] = 7;
Important traits and types for dimension and indexing:
- A
Dim
value represents a dimensionality or index. - Trait
Dimension
is implemented by all dimensionalities. It defines many operations for dimensions and indices. - Trait
IntoDimension
is used to convert into aDim
value. - Trait
ShapeBuilder
is an extension ofIntoDimension
and is used when constructing an array. A shape describes not just the extent of each axis but also their strides. - Trait
NdIndex
is an extension ofDimension
and is for values that can be used with indexing syntax.
The default memory order of an array is row major order (a.k.a “c” order), where each row is contiguous in memory. A column major (a.k.a. “f” or fortran) memory order array has columns (or, in general, the outermost axis) with contiguous elements.
The logical order of any array’s elements is the row major order
(the rightmost index is varying the fastest).
The iterators .iter(), .iter_mut()
always adhere to this order, for example.
§Loops, Producers and Iterators
Using Zip
is the most general way to apply a procedure
across one or several arrays or producers.
NdProducer
is like an iterable but for
multidimensional data. All producers have dimensions and axes, like an
array view, and they can be split and used with parallelization using Zip
.
For example, ArrayView<A, D>
is a producer, it has the same dimensions
as the array view and for each iteration it produces a reference to
the array element (&A
in this case).
Another example, if we have a 10 × 10 array and use .exact_chunks((2, 2))
we get a producer of chunks which has the dimensions 5 × 5 (because
there are 10 / 2 = 5 chunks in either direction). The 5 × 5 chunks producer
can be paired with any other producers of the same dimension with Zip
, for
example 5 × 5 arrays.
§.iter()
and .iter_mut()
These are the element iterators of arrays and they produce an element sequence in the logical order of the array, that means that the elements will be visited in the sequence that corresponds to increasing the last index first: 0, …, 0, 0; 0, …, 0, 1; 0, …0, 2 and so on.
§.outer_iter()
and .axis_iter()
These iterators produce array views of one smaller dimension.
For example, for a 2D array, .outer_iter()
will produce the 1D rows.
For a 3D array, .outer_iter()
produces 2D subviews.
.axis_iter()
is like outer_iter()
but allows you to pick which
axis to traverse.
The outer_iter
and axis_iter
are one dimensional producers.
§.rows()
, .columns()
and .lanes()
.rows()
is a producer (and iterable) of all rows in an array.
use ndarray::Array;
// 1. Loop over the rows of a 2D array
let mut a = Array::zeros((10, 10));
for mut row in a.rows_mut() {
row.fill(1.);
}
// 2. Use Zip to pair each row in 2D `a` with elements in 1D `b`
use ndarray::Zip;
let mut b = Array::zeros(a.nrows());
Zip::from(a.rows())
.and(&mut b)
.for_each(|a_row, b_elt| {
*b_elt = a_row[a.ncols() - 1] - a_row[0];
});
The lanes of an array are 1D segments along an axis and when pointed along the last axis they are rows, when pointed along the first axis they are columns.
A m × n array has m rows each of length n and conversely n columns each of length m.
To generalize this, we say that an array of dimension a × m × n has a m rows. It’s composed of a times the previous array, so it has a times as many rows.
All methods: .rows()
, .rows_mut()
,
.columns()
, .columns_mut()
,
.lanes(axis)
, .lanes_mut(axis)
.
Yes, for 2D arrays .rows()
and .outer_iter()
have about the same
effect:
rows()
is a producer with n - 1 dimensions of 1 dimensional itemsouter_iter()
is a producer with 1 dimension of n - 1 dimensional items
§Slicing
You can use slicing to create a view of a subset of the data in
the array. Slicing methods include .slice()
, .slice_mut()
,
.slice_move()
, and .slice_collapse()
.
The slicing argument can be passed using the macro s![]
,
which will be used in all examples. (The explicit form is an instance of
SliceInfo
or another type which implements SliceArg
; see their docs
for more information.)
If a range is used, the axis is preserved. If an index is used, that index
is selected and the axis is removed; this selects a subview. See
Subviews for more information about subviews. If a
NewAxis
instance is used, a new axis is inserted. Note that
.slice_collapse()
panics on NewAxis
elements and behaves like
.collapse_axis()
by preserving the number of dimensions.
When slicing arrays with generic dimensionality, creating an instance of
SliceInfo
to pass to the multi-axis slicing methods like .slice()
is awkward. In these cases, it’s usually more convenient to use
.slice_each_axis()
/.slice_each_axis_mut()
/.slice_each_axis_inplace()
or to create a view and then slice individual axes of the view using
methods such as .slice_axis_inplace()
and .collapse_axis()
.
It’s possible to take multiple simultaneous mutable slices with
.multi_slice_mut()
or (for ArrayViewMut
only)
.multi_slice_move()
.
use ndarray::{arr2, arr3, s, ArrayBase, DataMut, Dimension, NewAxis, Slice};
// 2 submatrices of 2 rows with 3 elements per row, means a shape of `[2, 2, 3]`.
let a = arr3(&[[[ 1, 2, 3], // -- 2 rows \_
[ 4, 5, 6]], // -- /
[[ 7, 8, 9], // \_ 2 submatrices
[10, 11, 12]]]); // /
// 3 columns ..../.../.../
assert_eq!(a.shape(), &[2, 2, 3]);
// Let’s create a slice with
//
// - Both of the submatrices of the greatest dimension: `..`
// - Only the first row in each submatrix: `0..1`
// - Every element in each row: `..`
let b = a.slice(s![.., 0..1, ..]);
let c = arr3(&[[[ 1, 2, 3]],
[[ 7, 8, 9]]]);
assert_eq!(b, c);
assert_eq!(b.shape(), &[2, 1, 3]);
// Let’s create a slice with
//
// - Both submatrices of the greatest dimension: `..`
// - The last row in each submatrix: `-1..`
// - Row elements in reverse order: `..;-1`
let d = a.slice(s![.., -1.., ..;-1]);
let e = arr3(&[[[ 6, 5, 4]],
[[12, 11, 10]]]);
assert_eq!(d, e);
assert_eq!(d.shape(), &[2, 1, 3]);
// Let’s create a slice while selecting a subview and inserting a new axis with
//
// - Both submatrices of the greatest dimension: `..`
// - The last row in each submatrix, removing that axis: `-1`
// - Row elements in reverse order: `..;-1`
// - A new axis at the end.
let f = a.slice(s![.., -1, ..;-1, NewAxis]);
let g = arr3(&[[ [6], [5], [4]],
[[12], [11], [10]]]);
assert_eq!(f, g);
assert_eq!(f.shape(), &[2, 3, 1]);
// Let's take two disjoint, mutable slices of a matrix with
//
// - One containing all the even-index columns in the matrix
// - One containing all the odd-index columns in the matrix
let mut h = arr2(&[[0, 1, 2, 3],
[4, 5, 6, 7]]);
let (s0, s1) = h.multi_slice_mut((s![.., ..;2], s![.., 1..;2]));
let i = arr2(&[[0, 2],
[4, 6]]);
let j = arr2(&[[1, 3],
[5, 7]]);
assert_eq!(s0, i);
assert_eq!(s1, j);
// Generic function which assigns the specified value to the elements which
// have indices in the lower half along all axes.
fn fill_lower<S, D>(arr: &mut ArrayBase<S, D>, x: S::Elem)
where
S: DataMut,
S::Elem: Clone,
D: Dimension,
{
arr.slice_each_axis_mut(|ax| Slice::from(0..ax.len / 2)).fill(x);
}
fill_lower(&mut h, 9);
let k = arr2(&[[9, 9, 2, 3],
[4, 5, 6, 7]]);
assert_eq!(h, k);
§Subviews
Subview methods allow you to restrict the array view while removing one
axis from the array. Methods for selecting individual subviews include
.index_axis()
, .index_axis_mut()
, .index_axis_move()
, and
.index_axis_inplace()
. You can also select a subview by using a single
index instead of a range when slicing. Some other methods, such as
.fold_axis()
, .axis_iter()
, .axis_iter_mut()
,
.outer_iter()
, and .outer_iter_mut()
operate on all the subviews
along an axis.
A related method is .collapse_axis()
, which modifies the view in the
same way as .index_axis()
except for removing the collapsed axis, since
it operates in place. The length of the axis becomes 1.
Methods for selecting an individual subview take two arguments: axis
and
index
.
use ndarray::{arr3, aview1, aview2, s, Axis};
// 2 submatrices of 2 rows with 3 elements per row, means a shape of `[2, 2, 3]`.
let a = arr3(&[[[ 1, 2, 3], // \ axis 0, submatrix 0
[ 4, 5, 6]], // /
[[ 7, 8, 9], // \ axis 0, submatrix 1
[10, 11, 12]]]); // /
// \
// axis 2, column 0
assert_eq!(a.shape(), &[2, 2, 3]);
// Let’s take a subview along the greatest dimension (axis 0),
// taking submatrix 0, then submatrix 1
let sub_0 = a.index_axis(Axis(0), 0);
let sub_1 = a.index_axis(Axis(0), 1);
assert_eq!(sub_0, aview2(&[[ 1, 2, 3],
[ 4, 5, 6]]));
assert_eq!(sub_1, aview2(&[[ 7, 8, 9],
[10, 11, 12]]));
assert_eq!(sub_0.shape(), &[2, 3]);
// This is the subview picking only axis 2, column 0
let sub_col = a.index_axis(Axis(2), 0);
assert_eq!(sub_col, aview2(&[[ 1, 4],
[ 7, 10]]));
// You can take multiple subviews at once (and slice at the same time)
let double_sub = a.slice(s![1, .., 0]);
assert_eq!(double_sub, aview1(&[7, 10]));
§Arithmetic Operations
Arrays support all arithmetic operations the same way: they apply elementwise.
Since the trait implementations are hard to overview, here is a summary.
§Binary Operators with Two Arrays
Let A
be an array or view of any kind. Let B
be an array
with owned storage (either Array
or ArcArray
).
Let C
be an array with mutable data (either Array
, ArcArray
or ArrayViewMut
).
The following combinations of operands
are supported for an arbitrary binary operator denoted by @
(it can be
+
, -
, *
, /
and so on).
&A @ &A
which produces a newArray
B @ A
which consumesB
, updates it with the result, and returns itB @ &A
which consumesB
, updates it with the result, and returns itC @= &A
which performs an arithmetic operation in place
Note that the element type needs to implement the operator trait and the
Clone
trait.
use ndarray::{array, ArrayView1};
let owned1 = array![1, 2];
let owned2 = array![3, 4];
let view1 = ArrayView1::from(&[5, 6]);
let view2 = ArrayView1::from(&[7, 8]);
let mut mutable = array![9, 10];
let sum1 = &view1 + &view2; // Allocates a new array. Note the explicit `&`.
// let sum2 = view1 + &view2; // This doesn't work because `view1` is not an owned array.
let sum3 = owned1 + view1; // Consumes `owned1`, updates it, and returns it.
let sum4 = owned2 + &view2; // Consumes `owned2`, updates it, and returns it.
mutable += &view2; // Updates `mutable` in-place.
§Binary Operators with Array and Scalar
The trait ScalarOperand
marks types that can be used in arithmetic
with arrays directly. For a scalar K
the following combinations of operands
are supported (scalar can be on either the left or right side, but
ScalarOperand
docs has the detailed conditions).
&A @ K
orK @ &A
which produces a newArray
B @ K
orK @ B
which consumesB
, updates it with the result and returns itC @= K
which performs an arithmetic operation in place
§Unary Operators
Let A
be an array or view of any kind. Let B
be an array with owned
storage (either Array
or ArcArray
). The following operands are supported
for an arbitrary unary operator denoted by @
(it can be -
or !
).
@&A
which produces a newArray
@B
which consumesB
, updates it with the result, and returns it
§Broadcasting
Arrays support limited broadcasting, where arithmetic operations with
array operands of different sizes can be carried out by repeating the
elements of the smaller dimension array. See
.broadcast()
for a more detailed
description.
use ndarray::arr2;
let a = arr2(&[[1., 1.],
[1., 2.],
[0., 3.],
[0., 4.]]);
let b = arr2(&[[0., 1.]]);
let c = arr2(&[[1., 2.],
[1., 3.],
[0., 4.],
[0., 5.]]);
// We can add because the shapes are compatible even if not equal.
// The `b` array is shape 1 × 2 but acts like a 4 × 2 array.
assert!(
c == a + b
);
§Conversions
§Conversions Between Array Types
This table is a summary of the conversions between arrays of different ownership, dimensionality, and element type. All of the conversions in this table preserve the shape of the array.
Output | Input | ||||
---|---|---|---|---|---|
|
|
|
|
|
|
|
no-op |
||||
|
no-op |
||||
|
no-op |
||||
|
|||||
|
illegal |
||||
equivalent with dim |
|||||
equivalent with dim |
|||||
|
§Conversions Between Arrays and Vec
s/Slices/Scalars
This is a table of the safe conversions between arrays and
Vec
s/slices/scalars. Note that some of the return values are actually
Result
/Option
wrappers around the indicated output types.
Input | Output | Methods |
---|---|---|
Vec<A> | ArrayBase<S: DataOwned, Ix1> | ::from_vec() |
Vec<A> | ArrayBase<S: DataOwned, D> | ::from_shape_vec() |
&[A] | ArrayView1<A> | ::from() |
&[A] | ArrayView<A, D> | ::from_shape() |
&mut [A] | ArrayViewMut1<A> | ::from() |
&mut [A] | ArrayViewMut<A, D> | ::from_shape() |
&ArrayBase<S, Ix1> | Vec<A> | .to_vec() |
Array<A, D> | Vec<A> | .into_raw_vec() 1 |
&ArrayBase<S, D> | &[A] | .as_slice() 2, .as_slice_memory_order() 3 |
&mut ArrayBase<S: DataMut, D> | &mut [A] | .as_slice_mut() 2, .as_slice_memory_order_mut() 3 |
ArrayView<A, D> | &[A] | .to_slice() 2 |
ArrayViewMut<A, D> | &mut [A] | .into_slice() 2 |
Array0<A> | A | .into_scalar() |
1Returns the data in memory order.
2Works only if the array is contiguous and in standard order.
3Works only if the array is contiguous.
The table above does not include all the constructors; it only shows
conversions to/from Vec
s/slices. See
below for more constructors.
§Conversions from Nested Vec
s/Array
s
It’s generally a good idea to avoid nested Vec
/Array
types, such as
Vec<Vec<A>>
or Vec<Array2<A>>
because:
-
they require extra heap allocations compared to a single
Array
, -
they can scatter data all over memory (because of multiple allocations),
-
they cause unnecessary indirection (traversing multiple pointers to reach the data),
-
they don’t enforce consistent shape within the nested
Vec
s/ArrayBase
s, and -
they are generally more difficult to work with.
The most common case where users might consider using nested
Vec
s/Array
s is when creating an array by appending rows/subviews in a
loop, where the rows/subviews are computed within the loop. However, there
are better ways than using nested Vec
s/Array
s.
If you know ahead-of-time the shape of the final array, the cleanest solution is to allocate the final array before the loop, and then assign the data to it within the loop, like this:
use ndarray::{array, Array2, Axis};
let mut arr = Array2::zeros((2, 3));
for (i, mut row) in arr.axis_iter_mut(Axis(0)).enumerate() {
// Perform calculations and assign to `row`; this is a trivial example:
row.fill(i);
}
assert_eq!(arr, array![[0, 0, 0], [1, 1, 1]]);
If you don’t know ahead-of-time the shape of the final array, then the
cleanest solution is generally to append the data to a flat Vec
, and then
convert it to an Array
at the end with
::from_shape_vec()
. You just have to be careful
that the layout of the data (the order of the elements in the flat Vec
)
is correct.
use ndarray::{array, Array2};
let ncols = 3;
let mut data = Vec::new();
let mut nrows = 0;
for i in 0..2 {
// Compute `row` and append it to `data`; this is a trivial example:
let row = vec![i; ncols];
data.extend_from_slice(&row);
nrows += 1;
}
let arr = Array2::from_shape_vec((nrows, ncols), data)?;
assert_eq!(arr, array![[0, 0, 0], [1, 1, 1]]);
If neither of these options works for you, and you really need to convert
nested Vec
/Array
instances to an Array
, the cleanest solution is
generally to use Iterator::flatten()
to get a flat Vec
, and then convert the Vec
to an Array
with
::from_shape_vec()
, like this:
use ndarray::{array, Array2, Array3};
let nested: Vec<Array2<i32>> = vec![
array![[1, 2, 3], [4, 5, 6]],
array![[7, 8, 9], [10, 11, 12]],
];
let inner_shape = nested[0].dim();
let shape = (nested.len(), inner_shape.0, inner_shape.1);
let flat: Vec<i32> = nested.iter().flatten().cloned().collect();
let arr = Array3::from_shape_vec(shape, flat)?;
assert_eq!(arr, array![
[[1, 2, 3], [4, 5, 6]],
[[7, 8, 9], [10, 11, 12]],
]);
Note that this implementation assumes that the nested Vec
s are all the
same shape and that the Vec
is non-empty. Depending on your application,
it may be a good idea to add checks for these assumptions and possibly
choose a different way to handle the empty case.
Implementations§
source§impl<S, A> ArrayBase<S, Ix1>where
S: DataOwned<Elem = A>,
impl<S, A> ArrayBase<S, Ix1>where
S: DataOwned<Elem = A>,
§Constructor Methods for Owned Arrays
Note that the constructor methods apply to Array
and ArcArray
,
the two array types that have owned storage.
§Constructor methods for one-dimensional arrays.
sourcepub fn from_vec(v: Vec<A>) -> Self
pub fn from_vec(v: Vec<A>) -> Self
Create a one-dimensional array from a vector (no copying needed).
Panics if the length is greater than isize::MAX
.
use ndarray::Array;
let array = Array::from_vec(vec![1., 2., 3., 4.]);
sourcepub fn from_iter<I: IntoIterator<Item = A>>(iterable: I) -> Self
pub fn from_iter<I: IntoIterator<Item = A>>(iterable: I) -> Self
Create a one-dimensional array from an iterator or iterable.
Panics if the length is greater than isize::MAX
.
use ndarray::Array;
let array = Array::from_iter(0..10);
sourcepub fn linspace(start: A, end: A, n: usize) -> Selfwhere
A: Float,
pub fn linspace(start: A, end: A, n: usize) -> Selfwhere
A: Float,
Create a one-dimensional array with n
evenly spaced elements from
start
to end
(inclusive). A
must be a floating point type.
Note that if start > end
, the first element will still be start
,
and the following elements will be decreasing. This is different from
the behavior of std::ops::RangeInclusive
, which interprets start > end
to mean that the range is empty.
Panics if n
is greater than isize::MAX
or if converting n - 1
to type A
fails.
use ndarray::{Array, arr1};
let array = Array::linspace(0., 1., 5);
assert!(array == arr1(&[0.0, 0.25, 0.5, 0.75, 1.0]))
sourcepub fn range(start: A, end: A, step: A) -> Selfwhere
A: Float,
pub fn range(start: A, end: A, step: A) -> Selfwhere
A: Float,
Create a one-dimensional array with elements from start
to end
(exclusive), incrementing by step
. A
must be a floating point type.
Panics if the length is greater than isize::MAX
.
use ndarray::{Array, arr1};
let array = Array::range(0., 5., 1.);
assert!(array == arr1(&[0., 1., 2., 3., 4.]))
sourcepub fn logspace(base: A, start: A, end: A, n: usize) -> Selfwhere
A: Float,
pub fn logspace(base: A, start: A, end: A, n: usize) -> Selfwhere
A: Float,
Create a one-dimensional array with n
logarithmically spaced
elements, with the starting value being base.powf(start)
and the
final one being base.powf(end)
. A
must be a floating point type.
If base
is negative, all values will be negative.
Panics if n
is greater than isize::MAX
or if converting n - 1
to type A
fails.
use approx::assert_abs_diff_eq;
use ndarray::{Array, arr1};
let array = Array::logspace(10.0, 0.0, 3.0, 4);
assert_abs_diff_eq!(array, arr1(&[1e0, 1e1, 1e2, 1e3]));
let array = Array::logspace(-10.0, 3.0, 0.0, 4);
assert_abs_diff_eq!(array, arr1(&[-1e3, -1e2, -1e1, -1e0]));
sourcepub fn geomspace(start: A, end: A, n: usize) -> Option<Self>where
A: Float,
pub fn geomspace(start: A, end: A, n: usize) -> Option<Self>where
A: Float,
Create a one-dimensional array with n
geometrically spaced elements
from start
to end
(inclusive). A
must be a floating point type.
Returns None
if start
and end
have different signs or if either
one is zero. Conceptually, this means that in order to obtain a Some
result, end / start
must be positive.
Panics if n
is greater than isize::MAX
or if converting n - 1
to type A
fails.
use approx::assert_abs_diff_eq;
use ndarray::{Array, arr1};
let array = Array::geomspace(1e0, 1e3, 4)?;
assert_abs_diff_eq!(array, arr1(&[1e0, 1e1, 1e2, 1e3]), epsilon = 1e-12);
let array = Array::geomspace(-1e3, -1e0, 4)?;
assert_abs_diff_eq!(array, arr1(&[-1e3, -1e2, -1e1, -1e0]), epsilon = 1e-12);
source§impl<S, A> ArrayBase<S, Ix2>where
S: DataOwned<Elem = A>,
impl<S, A> ArrayBase<S, Ix2>where
S: DataOwned<Elem = A>,
§Constructor methods for two-dimensional arrays.
sourcepub fn eye(n: Ix) -> Self
pub fn eye(n: Ix) -> Self
Create an identity matrix of size n
(square 2D array).
Panics if n * n
would overflow isize
.
sourcepub fn from_diag<S2>(diag: &ArrayBase<S2, Ix1>) -> Self
pub fn from_diag<S2>(diag: &ArrayBase<S2, Ix1>) -> Self
Create a 2D matrix from its diagonal
Panics if diag.len() * diag.len()
would overflow isize
.
use ndarray::{Array2, arr1, arr2};
let diag = arr1(&[1, 2]);
let array = Array2::from_diag(&diag);
assert_eq!(array, arr2(&[[1, 0], [0, 2]]));
sourcepub fn from_diag_elem(n: usize, elem: A) -> Self
pub fn from_diag_elem(n: usize, elem: A) -> Self
Create a square 2D matrix of the specified size, with the specified element along the diagonal and zeros elsewhere.
Panics if n * n
would overflow isize
.
use ndarray::{array, Array2};
let array = Array2::from_diag_elem(2, 5.);
assert_eq!(array, array![[5., 0.], [0., 5.]]);
source§impl<S, A, D> ArrayBase<S, D>
impl<S, A, D> ArrayBase<S, D>
§Constructor methods for n-dimensional arrays.
The shape
argument can be an integer or a tuple of integers to specify
a static size. For example 10
makes a length 10 one-dimensional array
(dimension type Ix1
) and (5, 6)
a 5 × 6 array (dimension type Ix2
).
With the trait ShapeBuilder
in scope, there is the method .f()
to select
column major (“f” order) memory layout instead of the default row major.
For example Array::zeros((5, 6).f())
makes a column major 5 × 6 array.
Use IxDyn
for the shape to create an array with dynamic
number of axes.
Finally, the few constructors that take a completely general
Into<StrideShape>
argument optionally support custom strides, for
example a shape given like (10, 2, 2).strides((1, 10, 20))
is valid.
sourcepub fn from_elem<Sh>(shape: Sh, elem: A) -> Selfwhere
A: Clone,
Sh: ShapeBuilder<Dim = D>,
pub fn from_elem<Sh>(shape: Sh, elem: A) -> Selfwhere
A: Clone,
Sh: ShapeBuilder<Dim = D>,
Create an array with copies of elem
, shape shape
.
Panics if the product of non-zero axis lengths overflows isize
.
use ndarray::{Array, arr3, ShapeBuilder};
let a = Array::from_elem((2, 2, 2), 1.);
assert!(
a == arr3(&[[[1., 1.],
[1., 1.]],
[[1., 1.],
[1., 1.]]])
);
assert!(a.strides() == &[4, 2, 1]);
let b = Array::from_elem((2, 2, 2).f(), 1.);
assert!(b.strides() == &[1, 2, 4]);
sourcepub fn zeros<Sh>(shape: Sh) -> Self
pub fn zeros<Sh>(shape: Sh) -> Self
Create an array with zeros, shape shape
.
Panics if the product of non-zero axis lengths overflows isize
.
sourcepub fn ones<Sh>(shape: Sh) -> Self
pub fn ones<Sh>(shape: Sh) -> Self
Create an array with ones, shape shape
.
Panics if the product of non-zero axis lengths overflows isize
.
sourcepub fn default<Sh>(shape: Sh) -> Selfwhere
A: Default,
Sh: ShapeBuilder<Dim = D>,
pub fn default<Sh>(shape: Sh) -> Selfwhere
A: Default,
Sh: ShapeBuilder<Dim = D>,
Create an array with default values, shape shape
Panics if the product of non-zero axis lengths overflows isize
.
sourcepub fn from_shape_simple_fn<Sh, F>(shape: Sh, f: F) -> Selfwhere
Sh: ShapeBuilder<Dim = D>,
F: FnMut() -> A,
pub fn from_shape_simple_fn<Sh, F>(shape: Sh, f: F) -> Selfwhere
Sh: ShapeBuilder<Dim = D>,
F: FnMut() -> A,
Create an array with values created by the function f
.
f
is called with no argument, and it should return the element to
create. If the precise index of the element to create is needed,
use from_shape_fn
instead.
This constructor can be useful if the element order is not important, for example if they are identical or random.
Panics if the product of non-zero axis lengths overflows isize
.
sourcepub fn from_shape_fn<Sh, F>(shape: Sh, f: F) -> Self
pub fn from_shape_fn<Sh, F>(shape: Sh, f: F) -> Self
Create an array with values created by the function f
.
f
is called with the index of the element to create; the elements are
visited in arbitrary order.
Panics if the product of non-zero axis lengths overflows isize
.
use ndarray::{Array, arr2};
// Create a table of i × j (with i and j from 1 to 3)
let ij_table = Array::from_shape_fn((3, 3), |(i, j)| (1 + i) * (1 + j));
assert_eq!(
ij_table,
arr2(&[[1, 2, 3],
[2, 4, 6],
[3, 6, 9]])
);
sourcepub fn from_shape_vec<Sh>(shape: Sh, v: Vec<A>) -> Result<Self, ShapeError>where
Sh: Into<StrideShape<D>>,
pub fn from_shape_vec<Sh>(shape: Sh, v: Vec<A>) -> Result<Self, ShapeError>where
Sh: Into<StrideShape<D>>,
Create an array with the given shape from a vector. (No cloning of elements needed.)
For a contiguous c- or f-order shape, the following applies:
Errors if shape
does not correspond to the number of elements in
v
or if the shape/strides would result in overflowing isize
.
For custom strides, the following applies:
Errors if strides and dimensions can point out of bounds of v
, if
strides allow multiple indices to point to the same element, or if the
shape/strides would result in overflowing isize
.
use ndarray::Array;
use ndarray::ShapeBuilder; // Needed for .strides() method
use ndarray::arr2;
let a = Array::from_shape_vec((2, 2), vec![1., 2., 3., 4.]);
assert!(a.is_ok());
let b = Array::from_shape_vec((2, 2).strides((1, 2)),
vec![1., 2., 3., 4.]).unwrap();
assert!(
b == arr2(&[[1., 3.],
[2., 4.]])
);
sourcepub unsafe fn from_shape_vec_unchecked<Sh>(shape: Sh, v: Vec<A>) -> Selfwhere
Sh: Into<StrideShape<D>>,
pub unsafe fn from_shape_vec_unchecked<Sh>(shape: Sh, v: Vec<A>) -> Selfwhere
Sh: Into<StrideShape<D>>,
Creates an array from a vector and interpret it according to the provided shape and strides. (No cloning of elements needed.)
§Safety
The caller must ensure that the following conditions are met:
-
The ndim of
dim
andstrides
must be the same. -
The product of non-zero axis lengths must not exceed
isize::MAX
. -
For axes with length > 1, the pointer cannot move outside the slice.
-
If the array will be empty (any axes are zero-length), the difference between the least address and greatest address accessible by moving along all axes must be ≤
v.len()
.If the array will not be empty, the difference between the least address and greatest address accessible by moving along all axes must be <
v.len()
. -
The strides must not allow any element to be referenced by two different indices.
sourcepub fn uninit<Sh>(shape: Sh) -> ArrayBase<S::MaybeUninit, D>where
Sh: ShapeBuilder<Dim = D>,
pub fn uninit<Sh>(shape: Sh) -> ArrayBase<S::MaybeUninit, D>where
Sh: ShapeBuilder<Dim = D>,
Create an array with uninitialized elements, shape shape
.
The uninitialized elements of type A
are represented by the type MaybeUninit<A>
,
an easier way to handle uninit values correctly.
Only when the array is completely initialized with valid elements, can it be
converted to an array of A
elements using .assume_init()
.
Panics if the number of elements in shape
would overflow isize.
§Safety
The whole of the array must be initialized before it is converted
using .assume_init()
or otherwise traversed/read with the element type A
.
§Examples
It is possible to assign individual values through *elt = MaybeUninit::new(value)
and so on.
use ndarray::{s, Array2};
// Example Task: Let's create a column shifted copy of the input
fn shift_by_two(a: &Array2<f32>) -> Array2<f32> {
// create an uninitialized array
let mut b = Array2::uninit(a.dim());
// two first columns in b are two last in a
// rest of columns in b are the initial columns in a
a.slice(s![.., -2..]).assign_to(b.slice_mut(s![.., ..2]));
a.slice(s![.., 2..]).assign_to(b.slice_mut(s![.., ..-2]));
// Now we can promise that `b` is safe to use with all operations
unsafe {
b.assume_init()
}
}
sourcepub fn build_uninit<Sh, F>(
shape: Sh,
builder: F,
) -> ArrayBase<S::MaybeUninit, D>
pub fn build_uninit<Sh, F>( shape: Sh, builder: F, ) -> ArrayBase<S::MaybeUninit, D>
Create an array with uninitialized elements, shape shape
.
The uninitialized elements of type A
are represented by the type MaybeUninit<A>
,
an easier way to handle uninit values correctly.
The builder
closure gets unshared access to the array through a view and can use it to
modify the array before it is returned. This allows initializing the array for any owned
array type (avoiding clone requirements for copy-on-write, because the array is unshared
when initially created).
Only when the array is completely initialized with valid elements, can it be
converted to an array of A
elements using .assume_init()
.
Panics if the number of elements in shape
would overflow isize.
§Safety
The whole of the array must be initialized before it is converted
using .assume_init()
or otherwise traversed/read with the element type A
.
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
§Methods For All Array Types
sourcepub fn len_of(&self, axis: Axis) -> usize
pub fn len_of(&self, axis: Axis) -> usize
Return the length of axis
.
The axis should be in the range Axis(
0 .. n )
where n is the
number of dimensions (axes) of the array.
Panics if the axis is out of bounds.
sourcepub fn dim(&self) -> D::Pattern
pub fn dim(&self) -> D::Pattern
Return the shape of the array in its “pattern” form, an integer in the one-dimensional case, tuple in the n-dimensional cases and so on.
sourcepub fn raw_dim(&self) -> D
pub fn raw_dim(&self) -> D
Return the shape of the array as it’s stored in the array.
This is primarily useful for passing to other ArrayBase
functions, such as when creating another array of the same
shape and dimensionality.
use ndarray::Array;
let a = Array::from_elem((2, 3), 5.);
// Create an array of zeros that's the same shape and dimensionality as `a`.
let b = Array::<f64, _>::zeros(a.raw_dim());
sourcepub fn shape(&self) -> &[usize]
pub fn shape(&self) -> &[usize]
Return the shape of the array as a slice.
Note that you probably don’t want to use this to create an array of the
same shape as another array because creating an array with e.g.
Array::zeros()
using a shape of type &[usize]
results in a dynamic-dimensional array. If you want to create an array
that has the same shape and dimensionality as another array, use
.raw_dim()
instead:
use ndarray::{Array, Array2};
let a = Array2::<i32>::zeros((3, 4));
let shape = a.shape();
assert_eq!(shape, &[3, 4]);
// Since `a.shape()` returned `&[usize]`, we get an `ArrayD` instance:
let b = Array::zeros(shape);
assert_eq!(a.clone().into_dyn(), b);
// To get the same dimension type, use `.raw_dim()` instead:
let c = Array::zeros(a.raw_dim());
assert_eq!(a, c);
sourcepub fn stride_of(&self, axis: Axis) -> isize
pub fn stride_of(&self, axis: Axis) -> isize
Return the stride of axis
.
The axis should be in the range Axis(
0 .. n )
where n is the
number of dimensions (axes) of the array.
Panics if the axis is out of bounds.
sourcepub fn view_mut(&mut self) -> ArrayViewMut<'_, A, D>where
S: DataMut,
pub fn view_mut(&mut self) -> ArrayViewMut<'_, A, D>where
S: DataMut,
Return a read-write view of the array
sourcepub fn cell_view(&mut self) -> ArrayView<'_, MathCell<A>, D>where
S: DataMut,
pub fn cell_view(&mut self) -> ArrayView<'_, MathCell<A>, D>where
S: DataMut,
Return a shared view of the array with elements as if they were embedded in cells.
The cell view requires a mutable borrow of the array. Once borrowed the cell view itself can be copied and accessed without exclusivity.
The view acts “as if” the elements are temporarily in cells, and elements can be changed through shared references using the regular cell methods.
sourcepub fn to_owned(&self) -> Array<A, D>
pub fn to_owned(&self) -> Array<A, D>
Return an uniquely owned copy of the array.
If the input array is contiguous, then the output array will have the same
memory layout. Otherwise, the layout of the output array is unspecified.
If you need a particular layout, you can allocate a new array with the
desired memory layout and .assign()
the data.
Alternatively, you can collectan iterator, like this for a result in
standard layout:
Array::from_shape_vec(arr.raw_dim(), arr.iter().cloned().collect()).unwrap()
or this for a result in column-major (Fortran) layout:
Array::from_shape_vec(arr.raw_dim().f(), arr.t().iter().cloned().collect()).unwrap()
Return a shared ownership (copy on write) array, cloning the array elements if necessary.
sourcepub fn into_owned(self) -> Array<A, D>
pub fn into_owned(self) -> Array<A, D>
Turn the array into a uniquely owned array, cloning the array elements if necessary.
sourcepub fn try_into_owned_nocopy(self) -> Result<Array<A, D>, Self>where
S: Data,
pub fn try_into_owned_nocopy(self) -> Result<Array<A, D>, Self>where
S: Data,
Converts the array into Array<A, D>
if this is possible without
cloning the array elements. Otherwise, returns self
unchanged.
use ndarray::{array, rcarr2, ArcArray2, Array2};
// Reference-counted, clone-on-write `ArcArray`.
let a: ArcArray2<_> = rcarr2(&[[1., 2.], [3., 4.]]);
{
// Another reference to the same data.
let b: ArcArray2<_> = a.clone();
// Since there are two references to the same data, `.into_owned()`
// would require cloning the data, so `.try_into_owned_nocopy()`
// returns `Err`.
assert!(b.try_into_owned_nocopy().is_err());
}
// Here, since the second reference has been dropped, the `ArcArray`
// can be converted into an `Array` without cloning the data.
let unique: Array2<_> = a.try_into_owned_nocopy().unwrap();
assert_eq!(unique, array![[1., 2.], [3., 4.]]);
Turn the array into a shared ownership (copy on write) array, cloning the array elements if necessary.
If you want to generalize over Array
and ArcArray
inputs but avoid
an A: Clone
bound, use Into::<ArcArray<A, D>>::into
instead of this
method.
sourcepub fn first(&self) -> Option<&A>where
S: Data,
pub fn first(&self) -> Option<&A>where
S: Data,
Returns a reference to the first element of the array, or None
if it
is empty.
§Example
use ndarray::Array3;
let mut a = Array3::<f64>::zeros([3, 4, 2]);
a[[0, 0, 0]] = 42.;
assert_eq!(a.first(), Some(&42.));
let b = Array3::<f64>::zeros([3, 0, 5]);
assert_eq!(b.first(), None);
sourcepub fn first_mut(&mut self) -> Option<&mut A>where
S: DataMut,
pub fn first_mut(&mut self) -> Option<&mut A>where
S: DataMut,
Returns a mutable reference to the first element of the array, or
None
if it is empty.
§Example
use ndarray::Array3;
let mut a = Array3::<f64>::zeros([3, 4, 2]);
*a.first_mut().unwrap() = 42.;
assert_eq!(a[[0, 0, 0]], 42.);
let mut b = Array3::<f64>::zeros([3, 0, 5]);
assert_eq!(b.first_mut(), None);
sourcepub fn last(&self) -> Option<&A>where
S: Data,
pub fn last(&self) -> Option<&A>where
S: Data,
Returns a reference to the last element of the array, or None
if it
is empty.
§Example
use ndarray::Array3;
let mut a = Array3::<f64>::zeros([3, 4, 2]);
a[[2, 3, 1]] = 42.;
assert_eq!(a.last(), Some(&42.));
let b = Array3::<f64>::zeros([3, 0, 5]);
assert_eq!(b.last(), None);
sourcepub fn last_mut(&mut self) -> Option<&mut A>where
S: DataMut,
pub fn last_mut(&mut self) -> Option<&mut A>where
S: DataMut,
Returns a mutable reference to the last element of the array, or None
if it is empty.
§Example
use ndarray::Array3;
let mut a = Array3::<f64>::zeros([3, 4, 2]);
*a.last_mut().unwrap() = 42.;
assert_eq!(a[[2, 3, 1]], 42.);
let mut b = Array3::<f64>::zeros([3, 0, 5]);
assert_eq!(b.last_mut(), None);
sourcepub fn iter(&self) -> Iter<'_, A, D> ⓘwhere
S: Data,
pub fn iter(&self) -> Iter<'_, A, D> ⓘwhere
S: Data,
Return an iterator of references to the elements of the array.
Elements are visited in the logical order of the array, which is where the rightmost index is varying the fastest.
Iterator element type is &A
.
sourcepub fn iter_mut(&mut self) -> IterMut<'_, A, D> ⓘwhere
S: DataMut,
pub fn iter_mut(&mut self) -> IterMut<'_, A, D> ⓘwhere
S: DataMut,
Return an iterator of mutable references to the elements of the array.
Elements are visited in the logical order of the array, which is where the rightmost index is varying the fastest.
Iterator element type is &mut A
.
sourcepub fn indexed_iter(&self) -> IndexedIter<'_, A, D> ⓘwhere
S: Data,
pub fn indexed_iter(&self) -> IndexedIter<'_, A, D> ⓘwhere
S: Data,
Return an iterator of indexes and references to the elements of the array.
Elements are visited in the logical order of the array, which is where the rightmost index is varying the fastest.
Iterator element type is (D::Pattern, &A)
.
See also Zip::indexed
sourcepub fn indexed_iter_mut(&mut self) -> IndexedIterMut<'_, A, D> ⓘwhere
S: DataMut,
pub fn indexed_iter_mut(&mut self) -> IndexedIterMut<'_, A, D> ⓘwhere
S: DataMut,
Return an iterator of indexes and mutable references to the elements of the array.
Elements are visited in the logical order of the array, which is where the rightmost index is varying the fastest.
Iterator element type is (D::Pattern, &mut A)
.
sourcepub fn slice_mut<I>(&mut self, info: I) -> ArrayViewMut<'_, A, I::OutDim>
pub fn slice_mut<I>(&mut self, info: I) -> ArrayViewMut<'_, A, I::OutDim>
sourcepub fn multi_slice_mut<'a, M>(&'a mut self, info: M) -> M::Outputwhere
M: MultiSliceArg<'a, A, D>,
S: DataMut,
pub fn multi_slice_mut<'a, M>(&'a mut self, info: M) -> M::Outputwhere
M: MultiSliceArg<'a, A, D>,
S: DataMut,
Return multiple disjoint, sliced, mutable views of the array.
See Slicing for full documentation. See also
MultiSliceArg
, s!
, SliceArg
, and
SliceInfo
.
Panics if any of the following occur:
- if any of the views would intersect (i.e. if any element would appear in multiple slices)
- if an index is out of bounds or step size is zero
- if
D
isIxDyn
andinfo
does not match the number of array axes
§Example
use ndarray::{arr2, s};
let mut a = arr2(&[[1, 2, 3], [4, 5, 6]]);
let (mut edges, mut middle) = a.multi_slice_mut((s![.., ..;2], s![.., 1]));
edges.fill(1);
middle.fill(0);
assert_eq!(a, arr2(&[[1, 0, 1], [1, 0, 1]]));
sourcepub fn slice_move<I>(self, info: I) -> ArrayBase<S, I::OutDim>where
I: SliceArg<D>,
pub fn slice_move<I>(self, info: I) -> ArrayBase<S, I::OutDim>where
I: SliceArg<D>,
sourcepub fn slice_collapse<I>(&mut self, info: I)where
I: SliceArg<D>,
pub fn slice_collapse<I>(&mut self, info: I)where
I: SliceArg<D>,
Slice the array in place without changing the number of dimensions.
In particular, if an axis is sliced with an index, the axis is
collapsed, as in .collapse_axis()
, rather than removed, as in
.slice_move()
or .index_axis_move()
.
See Slicing for full documentation.
See also s!
, SliceArg
, and SliceInfo
.
Panics in the following cases:
- if an index is out of bounds
- if a step size is zero
- if
SliceInfoElem::NewAxis
is ininfo
, e.g. ifNewAxis
was used in thes!
macro - if
D
isIxDyn
andinfo
does not match the number of array axes
sourcepub fn slice_axis(&self, axis: Axis, indices: Slice) -> ArrayView<'_, A, D>where
S: Data,
pub fn slice_axis(&self, axis: Axis, indices: Slice) -> ArrayView<'_, A, D>where
S: Data,
Return a view of the array, sliced along the specified axis.
Panics if an index is out of bounds or step size is zero.
Panics if axis
is out of bounds.
sourcepub fn slice_axis_mut(
&mut self,
axis: Axis,
indices: Slice,
) -> ArrayViewMut<'_, A, D>where
S: DataMut,
pub fn slice_axis_mut(
&mut self,
axis: Axis,
indices: Slice,
) -> ArrayViewMut<'_, A, D>where
S: DataMut,
Return a mutable view of the array, sliced along the specified axis.
Panics if an index is out of bounds or step size is zero.
Panics if axis
is out of bounds.
sourcepub fn slice_axis_inplace(&mut self, axis: Axis, indices: Slice)
pub fn slice_axis_inplace(&mut self, axis: Axis, indices: Slice)
Slice the array in place along the specified axis.
Panics if an index is out of bounds or step size is zero.
Panics if axis
is out of bounds.
sourcepub fn slice_axis_move(self, axis: Axis, indices: Slice) -> Self
pub fn slice_axis_move(self, axis: Axis, indices: Slice) -> Self
Slice the array in place along the specified axis, then return the sliced array.
Panics if an index is out of bounds or step size is zero.
Panics if axis
is out of bounds.
sourcepub fn slice_each_axis<F>(&self, f: F) -> ArrayView<'_, A, D>
pub fn slice_each_axis<F>(&self, f: F) -> ArrayView<'_, A, D>
Return a view of a slice of the array, with a closure specifying the slice for each axis.
This is especially useful for code which is generic over the dimensionality of the array.
Panics if an index is out of bounds or step size is zero.
sourcepub fn slice_each_axis_mut<F>(&mut self, f: F) -> ArrayViewMut<'_, A, D>
pub fn slice_each_axis_mut<F>(&mut self, f: F) -> ArrayViewMut<'_, A, D>
Return a mutable view of a slice of the array, with a closure specifying the slice for each axis.
This is especially useful for code which is generic over the dimensionality of the array.
Panics if an index is out of bounds or step size is zero.
sourcepub fn slice_each_axis_inplace<F>(&mut self, f: F)
pub fn slice_each_axis_inplace<F>(&mut self, f: F)
Slice the array in place, with a closure specifying the slice for each axis.
This is especially useful for code which is generic over the dimensionality of the array.
Panics if an index is out of bounds or step size is zero.
sourcepub fn get<I>(&self, index: I) -> Option<&A>
pub fn get<I>(&self, index: I) -> Option<&A>
Return a reference to the element at index
, or return None
if the index is out of bounds.
Arrays also support indexing syntax: array[index]
.
use ndarray::arr2;
let a = arr2(&[[1., 2.],
[3., 4.]]);
assert!(
a.get((0, 1)) == Some(&2.) &&
a.get((0, 2)) == None &&
a[(0, 1)] == 2. &&
a[[0, 1]] == 2.
);
sourcepub fn get_ptr<I>(&self, index: I) -> Option<*const A>where
I: NdIndex<D>,
pub fn get_ptr<I>(&self, index: I) -> Option<*const A>where
I: NdIndex<D>,
Return a raw pointer to the element at index
, or return None
if the index is out of bounds.
use ndarray::arr2;
let a = arr2(&[[1., 2.], [3., 4.]]);
let v = a.raw_view();
let p = a.get_ptr((0, 1)).unwrap();
assert_eq!(unsafe { *p }, 2.);
sourcepub fn get_mut<I>(&mut self, index: I) -> Option<&mut A>
pub fn get_mut<I>(&mut self, index: I) -> Option<&mut A>
Return a mutable reference to the element at index
, or return None
if the index is out of bounds.
sourcepub fn get_mut_ptr<I>(&mut self, index: I) -> Option<*mut A>where
S: RawDataMut,
I: NdIndex<D>,
pub fn get_mut_ptr<I>(&mut self, index: I) -> Option<*mut A>where
S: RawDataMut,
I: NdIndex<D>,
Return a raw pointer to the element at index
, or return None
if the index is out of bounds.
use ndarray::arr2;
let mut a = arr2(&[[1., 2.], [3., 4.]]);
let v = a.raw_view_mut();
let p = a.get_mut_ptr((0, 1)).unwrap();
unsafe {
*p = 5.;
}
assert_eq!(a.get((0, 1)), Some(&5.));
sourcepub unsafe fn uget<I>(&self, index: I) -> &A
pub unsafe fn uget<I>(&self, index: I) -> &A
Perform unchecked array indexing.
Return a reference to the element at index
.
Note: only unchecked for non-debug builds of ndarray.
§Safety
The caller must ensure that the index is in-bounds.
sourcepub unsafe fn uget_mut<I>(&mut self, index: I) -> &mut A
pub unsafe fn uget_mut<I>(&mut self, index: I) -> &mut A
Perform unchecked array indexing.
Return a mutable reference to the element at index
.
Note: Only unchecked for non-debug builds of ndarray.
§Safety
The caller must ensure that:
-
the index is in-bounds and
-
the data is uniquely held by the array. (This property is guaranteed for
Array
andArrayViewMut
, but not forArcArray
orCowArray
.)
sourcepub fn swap<I>(&mut self, index1: I, index2: I)
pub fn swap<I>(&mut self, index1: I, index2: I)
Swap elements at indices index1
and index2
.
Indices may be equal.
Panics if an index is out of bounds.
sourcepub unsafe fn uswap<I>(&mut self, index1: I, index2: I)
pub unsafe fn uswap<I>(&mut self, index1: I, index2: I)
Swap elements unchecked at indices index1
and index2
.
Indices may be equal.
Note: only unchecked for non-debug builds of ndarray.
§Safety
The caller must ensure that:
-
both
index1
andindex2
are in-bounds and -
the data is uniquely held by the array. (This property is guaranteed for
Array
andArrayViewMut
, but not forArcArray
orCowArray
.)
sourcepub fn index_axis(
&self,
axis: Axis,
index: usize,
) -> ArrayView<'_, A, D::Smaller>where
S: Data,
D: RemoveAxis,
pub fn index_axis(
&self,
axis: Axis,
index: usize,
) -> ArrayView<'_, A, D::Smaller>where
S: Data,
D: RemoveAxis,
Returns a view restricted to index
along the axis, with the axis
removed.
See Subviews for full documentation.
Panics if axis
or index
is out of bounds.
use ndarray::{arr2, ArrayView, Axis};
let a = arr2(&[[1., 2. ], // ... axis 0, row 0
[3., 4. ], // --- axis 0, row 1
[5., 6. ]]); // ... axis 0, row 2
// . \
// . axis 1, column 1
// axis 1, column 0
assert!(
a.index_axis(Axis(0), 1) == ArrayView::from(&[3., 4.]) &&
a.index_axis(Axis(1), 1) == ArrayView::from(&[2., 4., 6.])
);
sourcepub fn index_axis_mut(
&mut self,
axis: Axis,
index: usize,
) -> ArrayViewMut<'_, A, D::Smaller>where
S: DataMut,
D: RemoveAxis,
pub fn index_axis_mut(
&mut self,
axis: Axis,
index: usize,
) -> ArrayViewMut<'_, A, D::Smaller>where
S: DataMut,
D: RemoveAxis,
Returns a mutable view restricted to index
along the axis, with the
axis removed.
Panics if axis
or index
is out of bounds.
use ndarray::{arr2, aview2, Axis};
let mut a = arr2(&[[1., 2. ],
[3., 4. ]]);
// . \
// . axis 1, column 1
// axis 1, column 0
{
let mut column1 = a.index_axis_mut(Axis(1), 1);
column1 += 10.;
}
assert!(
a == aview2(&[[1., 12.],
[3., 14.]])
);
sourcepub fn index_axis_move(
self,
axis: Axis,
index: usize,
) -> ArrayBase<S, D::Smaller>where
D: RemoveAxis,
pub fn index_axis_move(
self,
axis: Axis,
index: usize,
) -> ArrayBase<S, D::Smaller>where
D: RemoveAxis,
Collapses the array to index
along the axis and removes the axis.
See .index_axis()
and Subviews for full documentation.
Panics if axis
or index
is out of bounds.
sourcepub fn collapse_axis(&mut self, axis: Axis, index: usize)
pub fn collapse_axis(&mut self, axis: Axis, index: usize)
Selects index
along the axis, collapsing the axis into length one.
Panics if axis
or index
is out of bounds.
sourcepub fn select(&self, axis: Axis, indices: &[Ix]) -> Array<A, D>
pub fn select(&self, axis: Axis, indices: &[Ix]) -> Array<A, D>
Along axis
, select arbitrary subviews corresponding to indices
and and copy them into a new array.
Panics if axis
or an element of indices
is out of bounds.
use ndarray::{arr2, Axis};
let x = arr2(&[[0., 1.],
[2., 3.],
[4., 5.],
[6., 7.],
[8., 9.]]);
let r = x.select(Axis(0), &[0, 4, 3]);
assert!(
r == arr2(&[[0., 1.],
[8., 9.],
[6., 7.]])
);
sourcepub fn rows(&self) -> Lanes<'_, A, D::Smaller>where
S: Data,
pub fn rows(&self) -> Lanes<'_, A, D::Smaller>where
S: Data,
Return a producer and iterable that traverses over the generalized rows of the array. For a 2D array these are the regular rows.
This is equivalent to .lanes(Axis(n - 1))
where n is self.ndim()
.
For an array of dimensions a × b × c × … × l × m it has a × b × c × … × l rows each of length m.
For example, in a 2 × 2 × 3 array, each row is 3 elements long and there are 2 × 2 = 4 rows in total.
Iterator element is ArrayView1<A>
(1D array view).
use ndarray::arr3;
let a = arr3(&[[[ 0, 1, 2], // -- row 0, 0
[ 3, 4, 5]], // -- row 0, 1
[[ 6, 7, 8], // -- row 1, 0
[ 9, 10, 11]]]); // -- row 1, 1
// `rows` will yield the four generalized rows of the array.
for row in a.rows() {
/* loop body */
}
sourcepub fn rows_mut(&mut self) -> LanesMut<'_, A, D::Smaller>where
S: DataMut,
pub fn rows_mut(&mut self) -> LanesMut<'_, A, D::Smaller>where
S: DataMut,
Return a producer and iterable that traverses over the generalized rows of the array and yields mutable array views.
Iterator element is ArrayView1<A>
(1D read-write array view).
sourcepub fn columns(&self) -> Lanes<'_, A, D::Smaller>where
S: Data,
pub fn columns(&self) -> Lanes<'_, A, D::Smaller>where
S: Data,
Return a producer and iterable that traverses over the generalized columns of the array. For a 2D array these are the regular columns.
This is equivalent to .lanes(Axis(0))
.
For an array of dimensions a × b × c × … × l × m it has b × c × … × l × m columns each of length a.
For example, in a 2 × 2 × 3 array, each column is 2 elements long and there are 2 × 3 = 6 columns in total.
Iterator element is ArrayView1<A>
(1D array view).
use ndarray::arr3;
// The generalized columns of a 3D array:
// are directed along the 0th axis: 0 and 6, 1 and 7 and so on...
let a = arr3(&[[[ 0, 1, 2], [ 3, 4, 5]],
[[ 6, 7, 8], [ 9, 10, 11]]]);
// Here `columns` will yield the six generalized columns of the array.
for column in a.columns() {
/* loop body */
}
sourcepub fn columns_mut(&mut self) -> LanesMut<'_, A, D::Smaller>where
S: DataMut,
pub fn columns_mut(&mut self) -> LanesMut<'_, A, D::Smaller>where
S: DataMut,
Return a producer and iterable that traverses over the generalized columns of the array and yields mutable array views.
Iterator element is ArrayView1<A>
(1D read-write array view).
sourcepub fn lanes(&self, axis: Axis) -> Lanes<'_, A, D::Smaller>where
S: Data,
pub fn lanes(&self, axis: Axis) -> Lanes<'_, A, D::Smaller>where
S: Data,
Return a producer and iterable that traverses over all 1D lanes
pointing in the direction of axis
.
When pointing in the direction of the first axis, they are columns, in the direction of the last axis rows; in general they are all lanes and are one dimensional.
Iterator element is ArrayView1<A>
(1D array view).
use ndarray::{arr3, aview1, Axis};
let a = arr3(&[[[ 0, 1, 2],
[ 3, 4, 5]],
[[ 6, 7, 8],
[ 9, 10, 11]]]);
let inner0 = a.lanes(Axis(0));
let inner1 = a.lanes(Axis(1));
let inner2 = a.lanes(Axis(2));
// The first lane for axis 0 is [0, 6]
assert_eq!(inner0.into_iter().next().unwrap(), aview1(&[0, 6]));
// The first lane for axis 1 is [0, 3]
assert_eq!(inner1.into_iter().next().unwrap(), aview1(&[0, 3]));
// The first lane for axis 2 is [0, 1, 2]
assert_eq!(inner2.into_iter().next().unwrap(), aview1(&[0, 1, 2]));
sourcepub fn lanes_mut(&mut self, axis: Axis) -> LanesMut<'_, A, D::Smaller>where
S: DataMut,
pub fn lanes_mut(&mut self, axis: Axis) -> LanesMut<'_, A, D::Smaller>where
S: DataMut,
Return a producer and iterable that traverses over all 1D lanes
pointing in the direction of axis
.
Iterator element is ArrayViewMut1<A>
(1D read-write array view).
sourcepub fn outer_iter(&self) -> AxisIter<'_, A, D::Smaller> ⓘwhere
S: Data,
D: RemoveAxis,
pub fn outer_iter(&self) -> AxisIter<'_, A, D::Smaller> ⓘwhere
S: Data,
D: RemoveAxis,
Return an iterator that traverses over the outermost dimension and yields each subview.
This is equivalent to .axis_iter(Axis(0))
.
Iterator element is ArrayView<A, D::Smaller>
(read-only array view).
sourcepub fn outer_iter_mut(&mut self) -> AxisIterMut<'_, A, D::Smaller> ⓘwhere
S: DataMut,
D: RemoveAxis,
pub fn outer_iter_mut(&mut self) -> AxisIterMut<'_, A, D::Smaller> ⓘwhere
S: DataMut,
D: RemoveAxis,
Return an iterator that traverses over the outermost dimension and yields each subview.
This is equivalent to .axis_iter_mut(Axis(0))
.
Iterator element is ArrayViewMut<A, D::Smaller>
(read-write array view).
sourcepub fn axis_iter(&self, axis: Axis) -> AxisIter<'_, A, D::Smaller> ⓘwhere
S: Data,
D: RemoveAxis,
pub fn axis_iter(&self, axis: Axis) -> AxisIter<'_, A, D::Smaller> ⓘwhere
S: Data,
D: RemoveAxis,
Return an iterator that traverses over axis
and yields each subview along it.
For example, in a 3 × 4 × 5 array, with axis
equal to Axis(2)
,
the iterator element
is a 3 × 4 subview (and there are 5 in total), as shown
in the picture below.
Iterator element is ArrayView<A, D::Smaller>
(read-only array view).
See Subviews for full documentation.
Panics if axis
is out of bounds.
sourcepub fn axis_iter_mut(&mut self, axis: Axis) -> AxisIterMut<'_, A, D::Smaller> ⓘwhere
S: DataMut,
D: RemoveAxis,
pub fn axis_iter_mut(&mut self, axis: Axis) -> AxisIterMut<'_, A, D::Smaller> ⓘwhere
S: DataMut,
D: RemoveAxis,
Return an iterator that traverses over axis
and yields each mutable subview along it.
Iterator element is ArrayViewMut<A, D::Smaller>
(read-write array view).
Panics if axis
is out of bounds.
sourcepub fn axis_chunks_iter(
&self,
axis: Axis,
size: usize,
) -> AxisChunksIter<'_, A, D> ⓘwhere
S: Data,
pub fn axis_chunks_iter(
&self,
axis: Axis,
size: usize,
) -> AxisChunksIter<'_, A, D> ⓘwhere
S: Data,
Return an iterator that traverses over axis
by chunks of size
,
yielding non-overlapping views along that axis.
Iterator element is ArrayView<A, D>
The last view may have less elements if size
does not divide
the axis’ dimension.
Panics if axis
is out of bounds or if size
is zero.
use ndarray::Array;
use ndarray::{arr3, Axis};
let a = Array::from_iter(0..28).into_shape_with_order((2, 7, 2)).unwrap();
let mut iter = a.axis_chunks_iter(Axis(1), 2);
// first iteration yields a 2 × 2 × 2 view
assert_eq!(iter.next().unwrap(),
arr3(&[[[ 0, 1], [ 2, 3]],
[[14, 15], [16, 17]]]));
// however the last element is a 2 × 1 × 2 view since 7 % 2 == 1
assert_eq!(iter.next_back().unwrap(), arr3(&[[[12, 13]],
[[26, 27]]]));
sourcepub fn axis_chunks_iter_mut(
&mut self,
axis: Axis,
size: usize,
) -> AxisChunksIterMut<'_, A, D> ⓘwhere
S: DataMut,
pub fn axis_chunks_iter_mut(
&mut self,
axis: Axis,
size: usize,
) -> AxisChunksIterMut<'_, A, D> ⓘwhere
S: DataMut,
Return an iterator that traverses over axis
by chunks of size
,
yielding non-overlapping read-write views along that axis.
Iterator element is ArrayViewMut<A, D>
Panics if axis
is out of bounds or if size
is zero.
sourcepub fn exact_chunks<E>(&self, chunk_size: E) -> ExactChunks<'_, A, D>where
E: IntoDimension<Dim = D>,
S: Data,
pub fn exact_chunks<E>(&self, chunk_size: E) -> ExactChunks<'_, A, D>where
E: IntoDimension<Dim = D>,
S: Data,
Return an exact chunks producer (and iterable).
It produces the whole chunks of a given n-dimensional chunk size, skipping the remainder along each dimension that doesn’t fit evenly.
The produced element is a ArrayView<A, D>
with exactly the dimension
chunk_size
.
Panics if any dimension of chunk_size
is zero
(Panics if D
is IxDyn
and chunk_size
does not match the
number of array axes.)
sourcepub fn exact_chunks_mut<E>(&mut self, chunk_size: E) -> ExactChunksMut<'_, A, D>where
E: IntoDimension<Dim = D>,
S: DataMut,
pub fn exact_chunks_mut<E>(&mut self, chunk_size: E) -> ExactChunksMut<'_, A, D>where
E: IntoDimension<Dim = D>,
S: DataMut,
Return an exact chunks producer (and iterable).
It produces the whole chunks of a given n-dimensional chunk size, skipping the remainder along each dimension that doesn’t fit evenly.
The produced element is a ArrayViewMut<A, D>
with exactly
the dimension chunk_size
.
Panics if any dimension of chunk_size
is zero
(Panics if D
is IxDyn
and chunk_size
does not match the
number of array axes.)
use ndarray::Array;
use ndarray::arr2;
let mut a = Array::zeros((6, 7));
// Fill each 2 × 2 chunk with the index of where it appeared in iteration
for (i, mut chunk) in a.exact_chunks_mut((2, 2)).into_iter().enumerate() {
chunk.fill(i);
}
// The resulting array is:
assert_eq!(
a,
arr2(&[[0, 0, 1, 1, 2, 2, 0],
[0, 0, 1, 1, 2, 2, 0],
[3, 3, 4, 4, 5, 5, 0],
[3, 3, 4, 4, 5, 5, 0],
[6, 6, 7, 7, 8, 8, 0],
[6, 6, 7, 7, 8, 8, 0]]));
sourcepub fn windows<E>(&self, window_size: E) -> Windows<'_, A, D>where
E: IntoDimension<Dim = D>,
S: Data,
pub fn windows<E>(&self, window_size: E) -> Windows<'_, A, D>where
E: IntoDimension<Dim = D>,
S: Data,
Return a window producer and iterable.
The windows are all distinct overlapping views of size window_size
that fit into the array’s shape.
This is essentially equivalent to [.windows_with_stride()
] with unit stride.
sourcepub fn windows_with_stride<E>(
&self,
window_size: E,
stride: E,
) -> Windows<'_, A, D>where
E: IntoDimension<Dim = D>,
S: Data,
pub fn windows_with_stride<E>(
&self,
window_size: E,
stride: E,
) -> Windows<'_, A, D>where
E: IntoDimension<Dim = D>,
S: Data,
Return a window producer and iterable.
The windows are all distinct views of size window_size
that fit into the array’s shape.
The stride is ordered by the outermost axis.
Hence, a (x₀, x₁, …, xₙ) stride will be applied to
(A₀, A₁, …, Aₙ) where Aₓ stands for Axis(x)
.
This produces all windows that fit within the array for the given stride, assuming the window size is not larger than the array size.
The produced element is an ArrayView<A, D>
with exactly the dimension
window_size
.
Note that passing a stride of only ones is similar to
calling ArrayBase::windows()
.
Panics if any dimension of window_size
or stride
is zero.
(Panics if D
is IxDyn
and window_size
or stride
does not match the
number of array axes.)
This is the same illustration found in ArrayBase::windows()
,
2×2 windows in a 3×4 array, but now with a (1, 2) stride:
──▶ Axis(1)
│ ┏━━━━━┳━━━━━┱─────┬─────┐ ┌─────┬─────┲━━━━━┳━━━━━┓
▼ ┃ a₀₀ ┃ a₀₁ ┃ │ │ │ │ ┃ a₀₂ ┃ a₀₃ ┃
Axis(0) ┣━━━━━╋━━━━━╉─────┼─────┤ ├─────┼─────╊━━━━━╋━━━━━┫
┃ a₁₀ ┃ a₁₁ ┃ │ │ │ │ ┃ a₁₂ ┃ a₁₃ ┃
┡━━━━━╇━━━━━╃─────┼─────┤ ├─────┼─────╄━━━━━╇━━━━━┩
│ │ │ │ │ │ │ │ │ │
└─────┴─────┴─────┴─────┘ └─────┴─────┴─────┴─────┘
┌─────┬─────┬─────┬─────┐ ┌─────┬─────┬─────┬─────┐
│ │ │ │ │ │ │ │ │ │
┢━━━━━╈━━━━━╅─────┼─────┤ ├─────┼─────╆━━━━━╈━━━━━┪
┃ a₁₀ ┃ a₁₁ ┃ │ │ │ │ ┃ a₁₂ ┃ a₁₃ ┃
┣━━━━━╋━━━━━╉─────┼─────┤ ├─────┼─────╊━━━━━╋━━━━━┫
┃ a₂₀ ┃ a₂₁ ┃ │ │ │ │ ┃ a₂₂ ┃ a₂₃ ┃
┗━━━━━┻━━━━━┹─────┴─────┘ └─────┴─────┺━━━━━┻━━━━━┛
sourcepub fn axis_windows(
&self,
axis: Axis,
window_size: usize,
) -> AxisWindows<'_, A, D>where
S: Data,
pub fn axis_windows(
&self,
axis: Axis,
window_size: usize,
) -> AxisWindows<'_, A, D>where
S: Data,
Returns a producer which traverses over all windows of a given length along an axis.
The windows are all distinct, possibly-overlapping views. The shape of each window
is the shape of self
, with the length of axis
replaced with window_size
.
Panics if axis
is out-of-bounds or if window_size
is zero.
use ndarray::{Array3, Axis, s};
let arr = Array3::from_shape_fn([4, 5, 2], |(i, j, k)| i * 100 + j * 10 + k);
let correct = vec![
arr.slice(s![.., 0..3, ..]),
arr.slice(s![.., 1..4, ..]),
arr.slice(s![.., 2..5, ..]),
];
for (window, correct) in arr.axis_windows(Axis(1), 3).into_iter().zip(&correct) {
assert_eq!(window, correct);
assert_eq!(window.shape(), &[4, 3, 2]);
}
sourcepub fn diag(&self) -> ArrayView1<'_, A>where
S: Data,
pub fn diag(&self) -> ArrayView1<'_, A>where
S: Data,
Return a view of the diagonal elements of the array.
The diagonal is simply the sequence indexed by (0, 0, .., 0), (1, 1, …, 1) etc as long as all axes have elements.
sourcepub fn diag_mut(&mut self) -> ArrayViewMut1<'_, A>where
S: DataMut,
pub fn diag_mut(&mut self) -> ArrayViewMut1<'_, A>where
S: DataMut,
Return a read-write view over the diagonal elements of the array.
sourcepub fn is_standard_layout(&self) -> bool
pub fn is_standard_layout(&self) -> bool
Return true
if the array data is laid out in contiguous “C order” in
memory (where the last index is the most rapidly varying).
Return false
otherwise, i.e. the array is possibly not
contiguous in memory, it has custom strides, etc.
sourcepub fn as_standard_layout(&self) -> CowArray<'_, A, D>
pub fn as_standard_layout(&self) -> CowArray<'_, A, D>
Return a standard-layout array containing the data, cloning if necessary.
If self
is in standard layout, a COW view of the data is returned
without cloning. Otherwise, the data is cloned, and the returned array
owns the cloned data.
use ndarray::Array2;
let standard = Array2::<f64>::zeros((3, 4));
assert!(standard.is_standard_layout());
let cow_view = standard.as_standard_layout();
assert!(cow_view.is_view());
assert!(cow_view.is_standard_layout());
let fortran = standard.reversed_axes();
assert!(!fortran.is_standard_layout());
let cow_owned = fortran.as_standard_layout();
assert!(cow_owned.is_owned());
assert!(cow_owned.is_standard_layout());
sourcepub fn as_ptr(&self) -> *const A
pub fn as_ptr(&self) -> *const A
Return a pointer to the first element in the array.
Raw access to array elements needs to follow the strided indexing scheme: an element at multi-index I in an array with strides S is located at offset
Σ0 ≤ k < d Ik × Sk
where d is self.ndim()
.
sourcepub fn as_mut_ptr(&mut self) -> *mut Awhere
S: RawDataMut,
pub fn as_mut_ptr(&mut self) -> *mut Awhere
S: RawDataMut,
Return a mutable pointer to the first element in the array.
This method attempts to unshare the data. If S: DataMut
, then the
data is guaranteed to be uniquely held on return.
§Warning
When accessing elements through this pointer, make sure to use strides obtained after calling this method, since the process of unsharing the data may change the strides.
sourcepub fn raw_view(&self) -> RawArrayView<A, D>
pub fn raw_view(&self) -> RawArrayView<A, D>
Return a raw view of the array.
sourcepub fn raw_view_mut(&mut self) -> RawArrayViewMut<A, D>where
S: RawDataMut,
pub fn raw_view_mut(&mut self) -> RawArrayViewMut<A, D>where
S: RawDataMut,
Return a raw mutable view of the array.
This method attempts to unshare the data. If S: DataMut
, then the
data is guaranteed to be uniquely held on return.
sourcepub fn as_slice(&self) -> Option<&[A]>where
S: Data,
pub fn as_slice(&self) -> Option<&[A]>where
S: Data,
Return the array’s data as a slice, if it is contiguous and in standard order.
Return None
otherwise.
If this function returns Some(_)
, then the element order in the slice
corresponds to the logical order of the array’s elements.
sourcepub fn as_slice_mut(&mut self) -> Option<&mut [A]>where
S: DataMut,
pub fn as_slice_mut(&mut self) -> Option<&mut [A]>where
S: DataMut,
Return the array’s data as a slice, if it is contiguous and in standard order.
Return None
otherwise.
sourcepub fn as_slice_memory_order(&self) -> Option<&[A]>where
S: Data,
pub fn as_slice_memory_order(&self) -> Option<&[A]>where
S: Data,
Return the array’s data as a slice if it is contiguous,
return None
otherwise.
If this function returns Some(_)
, then the elements in the slice
have whatever order the elements have in memory.
sourcepub fn as_slice_memory_order_mut(&mut self) -> Option<&mut [A]>where
S: DataMut,
pub fn as_slice_memory_order_mut(&mut self) -> Option<&mut [A]>where
S: DataMut,
Return the array’s data as a slice if it is contiguous,
return None
otherwise.
In the contiguous case, in order to return a unique reference, this method unshares the data if necessary, but it preserves the existing strides.
sourcepub fn to_shape<E>(
&self,
new_shape: E,
) -> Result<CowArray<'_, A, E::Dim>, ShapeError>
pub fn to_shape<E>( &self, new_shape: E, ) -> Result<CowArray<'_, A, E::Dim>, ShapeError>
Transform the array into new_shape
; any shape with the same number of elements is
accepted.
order
specifies the logical order in which the array is to be read and reshaped.
The array is returned as a CowArray
; a view if possible, otherwise an owned array.
For example, when starting from the one-dimensional sequence 1 2 3 4 5 6, it would be understood as a 2 x 3 array in row major (“C”) order this way:
1 2 3
4 5 6
and as 2 x 3 in column major (“F”) order this way:
1 3 5
2 4 6
This example should show that any time we “reflow” the elements in the array to a different
number of rows and columns (or more axes if applicable), it is important to pick an index
ordering, and that’s the reason for the function parameter for order
.
The new_shape
parameter should be a dimension and an optional order like these examples:
(3, 4) // Shape 3 x 4 with default order (RowMajor)
((3, 4), Order::RowMajor)) // use specific order
((3, 4), Order::ColumnMajor)) // use specific order
((3, 4), Order::C)) // use shorthand for order - shorthands C and F
Errors if the new shape doesn’t have the same number of elements as the array’s current shape.
§Example
use ndarray::array;
use ndarray::Order;
assert!(
array![1., 2., 3., 4., 5., 6.].to_shape(((2, 3), Order::RowMajor)).unwrap()
== array![[1., 2., 3.],
[4., 5., 6.]]
);
assert!(
array![1., 2., 3., 4., 5., 6.].to_shape(((2, 3), Order::ColumnMajor)).unwrap()
== array![[1., 3., 5.],
[2., 4., 6.]]
);
sourcepub fn into_shape_with_order<E>(
self,
shape: E,
) -> Result<ArrayBase<S, E::Dim>, ShapeError>where
E: ShapeArg,
pub fn into_shape_with_order<E>(
self,
shape: E,
) -> Result<ArrayBase<S, E::Dim>, ShapeError>where
E: ShapeArg,
Transform the array into shape
; any shape with the same number of
elements is accepted, but the source array must be contiguous.
If an index ordering is not specified, the default is RowMajor
.
The operation will only succeed if the array’s memory layout is compatible with
the index ordering, so that the array elements can be rearranged in place.
If required use .to_shape()
or .into_shape_clone
instead for more flexible reshaping of
arrays, which allows copying elements if required.
Errors if the shapes don’t have the same number of elements.
Errors if order RowMajor is given but input is not c-contiguous.
Errors if order ColumnMajor is given but input is not f-contiguous.
If shape is not given: use memory layout of incoming array. Row major arrays are reshaped using row major index ordering, column major arrays with column major index ordering.
The new_shape
parameter should be a dimension and an optional order like these examples:
(3, 4) // Shape 3 x 4 with default order (RowMajor)
((3, 4), Order::RowMajor)) // use specific order
((3, 4), Order::ColumnMajor)) // use specific order
((3, 4), Order::C)) // use shorthand for order - shorthands C and F
§Example
use ndarray::{aview1, aview2};
use ndarray::Order;
assert!(
aview1(&[1., 2., 3., 4.]).into_shape_with_order((2, 2)).unwrap()
== aview2(&[[1., 2.],
[3., 4.]])
);
assert!(
aview1(&[1., 2., 3., 4.]).into_shape_with_order(((2, 2), Order::ColumnMajor)).unwrap()
== aview2(&[[1., 3.],
[2., 4.]])
);
sourcepub fn into_shape<E>(self, shape: E) -> Result<ArrayBase<S, E::Dim>, ShapeError>where
E: IntoDimension,
👎Deprecated since 0.16.0: Use .into_shape_with_order()
or .to_shape()
pub fn into_shape<E>(self, shape: E) -> Result<ArrayBase<S, E::Dim>, ShapeError>where
E: IntoDimension,
.into_shape_with_order()
or .to_shape()
Transform the array into shape
; any shape with the same number of
elements is accepted, but the source array or view must be in standard
or column-major (Fortran) layout.
Note that .into_shape()
“moves” elements differently depending on if the input array
is C-contig or F-contig, it follows the index order that corresponds to the memory order.
Prefer to use .to_shape()
or .into_shape_with_order()
.
Because of this, the method is deprecated. That reshapes depend on memory order is not intuitive.
Errors if the shapes don’t have the same number of elements.
Errors if the input array is not c- or f-contiguous.
use ndarray::{aview1, aview2};
assert!(
aview1(&[1., 2., 3., 4.]).into_shape((2, 2)).unwrap()
== aview2(&[[1., 2.],
[3., 4.]])
);
sourcepub fn into_shape_clone<E>(
self,
shape: E,
) -> Result<ArrayBase<S, E::Dim>, ShapeError>
pub fn into_shape_clone<E>( self, shape: E, ) -> Result<ArrayBase<S, E::Dim>, ShapeError>
Transform the array into shape
; any shape with the same number of
elements is accepted. Array elements are reordered in place if
possible, otherwise they are copied to create a new array.
If an index ordering is not specified, the default is RowMajor
.
§.to_shape
vs .into_shape_clone
to_shape
supports views and outputting viewsto_shape
borrows the original array,into_shape_clone
consumes the originalinto_shape_clone
preserves array type (Array vs ArcArray), but does not support views.
Errors if the shapes don’t have the same number of elements.
sourcepub fn reshape<E>(&self, shape: E) -> ArrayBase<S, E::Dim>
👎Deprecated since 0.16.0: Use .into_shape_with_order()
or .to_shape()
pub fn reshape<E>(&self, shape: E) -> ArrayBase<S, E::Dim>
.into_shape_with_order()
or .to_shape()
Note: Reshape is for ArcArray
only. Use .into_shape_with_order()
for
other arrays and array views.
Transform the array into shape
; any shape with the same number of
elements is accepted.
May clone all elements if needed to arrange elements in standard layout (and break sharing).
Panics if shapes are incompatible.
This method is obsolete, because it is inflexible in how logical order
of the array is handled. See [.to_shape()
].
use ndarray::{rcarr1, rcarr2};
assert!(
rcarr1(&[1., 2., 3., 4.]).reshape((2, 2))
== rcarr2(&[[1., 2.],
[3., 4.]])
);
sourcepub fn flatten(&self) -> CowArray<'_, A, Ix1>
pub fn flatten(&self) -> CowArray<'_, A, Ix1>
Flatten the array to a one-dimensional array.
The array is returned as a CowArray
; a view if possible, otherwise an owned array.
use ndarray::{arr1, arr3};
let array = arr3(&[[[1, 2], [3, 4]], [[5, 6], [7, 8]]]);
let flattened = array.flatten();
assert_eq!(flattened, arr1(&[1, 2, 3, 4, 5, 6, 7, 8]));
sourcepub fn flatten_with_order(&self, order: Order) -> CowArray<'_, A, Ix1>
pub fn flatten_with_order(&self, order: Order) -> CowArray<'_, A, Ix1>
Flatten the array to a one-dimensional array.
order
specifies the logical order in which the array is to be read and reshaped.
The array is returned as a CowArray
; a view if possible, otherwise an owned array.
use ndarray::{arr1, arr2};
use ndarray::Order;
let array = arr2(&[[1, 2], [3, 4], [5, 6], [7, 8]]);
let flattened = array.flatten_with_order(Order::RowMajor);
assert_eq!(flattened, arr1(&[1, 2, 3, 4, 5, 6, 7, 8]));
let flattened = array.flatten_with_order(Order::ColumnMajor);
assert_eq!(flattened, arr1(&[1, 3, 5, 7, 2, 4, 6, 8]));
sourcepub fn into_flat(self) -> ArrayBase<S, Ix1>
pub fn into_flat(self) -> ArrayBase<S, Ix1>
Flatten the array to a one-dimensional array, consuming the array.
If possible, no copy is made, and the new array use the same memory as the original array. Otherwise, a new array is allocated and the elements are copied.
use ndarray::{arr1, arr3};
let array = arr3(&[[[1, 2], [3, 4]], [[5, 6], [7, 8]]]);
let flattened = array.into_flat();
assert_eq!(flattened, arr1(&[1, 2, 3, 4, 5, 6, 7, 8]));
sourcepub fn into_dyn(self) -> ArrayBase<S, IxDyn>
pub fn into_dyn(self) -> ArrayBase<S, IxDyn>
Convert any array or array view to a dynamic dimensional array or array view (respectively).
use ndarray::{arr2, ArrayD};
let array: ArrayD<i32> = arr2(&[[1, 2],
[3, 4]]).into_dyn();
sourcepub fn into_dimensionality<D2>(self) -> Result<ArrayBase<S, D2>, ShapeError>where
D2: Dimension,
pub fn into_dimensionality<D2>(self) -> Result<ArrayBase<S, D2>, ShapeError>where
D2: Dimension,
Convert an array or array view to another with the same type, but different dimensionality type. Errors if the dimensions don’t agree (the number of axes must match).
Note that conversion to a dynamic dimensional array will never fail (and is equivalent to
the into_dyn
method).
use ndarray::{ArrayD, Ix2, IxDyn};
// Create a dynamic dimensionality array and convert it to an Array2
// (Ix2 dimension type).
let array = ArrayD::<f64>::zeros(IxDyn(&[10, 10]));
assert!(array.into_dimensionality::<Ix2>().is_ok());
sourcepub fn broadcast<E>(&self, dim: E) -> Option<ArrayView<'_, A, E::Dim>>where
E: IntoDimension,
S: Data,
pub fn broadcast<E>(&self, dim: E) -> Option<ArrayView<'_, A, E::Dim>>where
E: IntoDimension,
S: Data,
Act like a larger size and/or shape array by broadcasting into a larger shape, if possible.
Return None
if shapes can not be broadcast together.
Background
- Two axes are compatible if they are equal, or one of them is 1.
- In this instance, only the axes of the smaller side (self) can be 1.
Compare axes beginning with the last axis of each shape.
For example (1, 2, 4) can be broadcast into (7, 6, 2, 4) because its axes are either equal or 1 (or missing); while (2, 2) can not be broadcast into (2, 4).
The implementation creates a view with strides set to zero for the axes that are to be repeated.
The broadcasting documentation for Numpy has more information.
use ndarray::{aview1, aview2};
assert!(
aview1(&[1., 0.]).broadcast((10, 2)).unwrap()
== aview2(&[[1., 0.]; 10])
);
sourcepub fn swap_axes(&mut self, ax: usize, bx: usize)
pub fn swap_axes(&mut self, ax: usize, bx: usize)
Swap axes ax
and bx
.
This does not move any data, it just adjusts the array’s dimensions and strides.
Panics if the axes are out of bounds.
use ndarray::arr2;
let mut a = arr2(&[[1., 2., 3.]]);
a.swap_axes(0, 1);
assert!(
a == arr2(&[[1.], [2.], [3.]])
);
sourcepub fn permuted_axes<T>(self, axes: T) -> ArrayBase<S, D>where
T: IntoDimension<Dim = D>,
pub fn permuted_axes<T>(self, axes: T) -> ArrayBase<S, D>where
T: IntoDimension<Dim = D>,
Permute the axes.
This does not move any data, it just adjusts the array’s dimensions and strides.
i in the j-th place in the axes sequence means self
’s i-th axis
becomes self.permuted_axes()
’s j-th axis
Panics if any of the axes are out of bounds, if an axis is missing, or if an axis is repeated more than once.
§Examples
use ndarray::{arr2, Array3};
let a = arr2(&[[0, 1], [2, 3]]);
assert_eq!(a.view().permuted_axes([1, 0]), a.t());
let b = Array3::<u8>::zeros((1, 2, 3));
assert_eq!(b.permuted_axes([1, 0, 2]).shape(), &[2, 1, 3]);
sourcepub fn reversed_axes(self) -> ArrayBase<S, D>
pub fn reversed_axes(self) -> ArrayBase<S, D>
Transpose the array by reversing axes.
Transposition reverses the order of the axes (dimensions and strides) while retaining the same data.
sourcepub fn t(&self) -> ArrayView<'_, A, D>where
S: Data,
pub fn t(&self) -> ArrayView<'_, A, D>where
S: Data,
Return a transposed view of the array.
This is a shorthand for self.view().reversed_axes()
.
See also the more general methods .reversed_axes()
and .swap_axes()
.
sourcepub fn axes(&self) -> Axes<'_, D> ⓘ
pub fn axes(&self) -> Axes<'_, D> ⓘ
Return an iterator over the length and stride of each axis.
sourcepub fn max_stride_axis(&self) -> Axis
pub fn max_stride_axis(&self) -> Axis
Return the axis with the greatest stride (by absolute value), preferring axes with len > 1.
sourcepub fn invert_axis(&mut self, axis: Axis)
pub fn invert_axis(&mut self, axis: Axis)
Reverse the stride of axis
.
Panics if the axis is out of bounds.
sourcepub fn merge_axes(&mut self, take: Axis, into: Axis) -> bool
pub fn merge_axes(&mut self, take: Axis, into: Axis) -> bool
If possible, merge in the axis take
to into
.
Returns true
iff the axes are now merged.
This method merges the axes if movement along the two original axes
(moving fastest along the into
axis) can be equivalently represented
as movement along one (merged) axis. Merging the axes preserves this
order in the merged axis. If take
and into
are the same axis, then
the axis is “merged” if its length is ≤ 1.
If the return value is true
, then the following hold:
-
The new length of the
into
axis is the product of the original lengths of the two axes. -
The new length of the
take
axis is 0 if the product of the original lengths of the two axes is 0, and 1 otherwise.
If the return value is false
, then merging is not possible, and the
original shape and strides have been preserved.
Note that the ordering constraint means that if it’s possible to merge
take
into into
, it’s usually not possible to merge into
into
take
, and vice versa.
use ndarray::Array3;
use ndarray::Axis;
let mut a = Array3::<f64>::zeros((2, 3, 4));
assert!(a.merge_axes(Axis(1), Axis(2)));
assert_eq!(a.shape(), &[2, 1, 12]);
Panics if an axis is out of bounds.
sourcepub fn insert_axis(self, axis: Axis) -> ArrayBase<S, D::Larger>
pub fn insert_axis(self, axis: Axis) -> ArrayBase<S, D::Larger>
Insert new array axis at axis
and return the result.
use ndarray::{Array3, Axis, arr1, arr2};
// Convert a 1-D array into a row vector (2-D).
let a = arr1(&[1, 2, 3]);
let row = a.insert_axis(Axis(0));
assert_eq!(row, arr2(&[[1, 2, 3]]));
// Convert a 1-D array into a column vector (2-D).
let b = arr1(&[1, 2, 3]);
let col = b.insert_axis(Axis(1));
assert_eq!(col, arr2(&[[1], [2], [3]]));
// The new axis always has length 1.
let b = Array3::<f64>::zeros((3, 4, 5));
assert_eq!(b.insert_axis(Axis(2)).shape(), &[3, 4, 1, 5]);
Panics if the axis is out of bounds.
sourcepub fn remove_axis(self, axis: Axis) -> ArrayBase<S, D::Smaller>where
D: RemoveAxis,
pub fn remove_axis(self, axis: Axis) -> ArrayBase<S, D::Smaller>where
D: RemoveAxis,
Remove array axis axis
and return the result.
This is equivalent to .index_axis_move(axis, 0)
and makes most sense to use if the
axis to remove is of length 1.
Panics if the axis is out of bounds or its length is zero.
sourcepub fn assign<E: Dimension, S2>(&mut self, rhs: &ArrayBase<S2, E>)
pub fn assign<E: Dimension, S2>(&mut self, rhs: &ArrayBase<S2, E>)
Perform an elementwise assigment to self
from rhs
.
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
sourcepub fn assign_to<P>(&self, to: P)
pub fn assign_to<P>(&self, to: P)
Perform an elementwise assigment of values cloned from self
into array or producer to
.
The destination to
can be another array or a producer of assignable elements.
AssignElem
determines how elements are assigned.
Panics if shapes disagree.
sourcepub fn zip_mut_with<B, S2, E, F>(&mut self, rhs: &ArrayBase<S2, E>, f: F)
pub fn zip_mut_with<B, S2, E, F>(&mut self, rhs: &ArrayBase<S2, E>, f: F)
Traverse two arrays in unspecified order, in lock step,
calling the closure f
on each element pair.
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
sourcepub fn fold<'a, F, B>(&'a self, init: B, f: F) -> B
pub fn fold<'a, F, B>(&'a self, init: B, f: F) -> B
Traverse the array elements and apply a fold, returning the resulting value.
Elements are visited in arbitrary order.
sourcepub fn map<'a, B, F>(&'a self, f: F) -> Array<B, D>
pub fn map<'a, B, F>(&'a self, f: F) -> Array<B, D>
Call f
by reference on each element and create a new array
with the new values.
Elements are visited in arbitrary order.
Return an array with the same shape as self
.
use ndarray::arr2;
let a = arr2(&[[ 0., 1.],
[-1., 2.]]);
assert!(
a.map(|x| *x >= 1.0)
== arr2(&[[false, true],
[false, true]])
);
sourcepub fn map_mut<'a, B, F>(&'a mut self, f: F) -> Array<B, D>
pub fn map_mut<'a, B, F>(&'a mut self, f: F) -> Array<B, D>
Call f
on a mutable reference of each element and create a new array
with the new values.
Elements are visited in arbitrary order.
Return an array with the same shape as self
.
sourcepub fn mapv<B, F>(&self, f: F) -> Array<B, D>
pub fn mapv<B, F>(&self, f: F) -> Array<B, D>
Call f
by value on each element and create a new array
with the new values.
Elements are visited in arbitrary order.
Return an array with the same shape as self
.
use ndarray::arr2;
let a = arr2(&[[ 0., 1.],
[-1., 2.]]);
assert!(
a.mapv(f32::abs) == arr2(&[[0., 1.],
[1., 2.]])
);
sourcepub fn mapv_into<F>(self, f: F) -> Self
pub fn mapv_into<F>(self, f: F) -> Self
Call f
by value on each element, update the array with the new values
and return it.
Elements are visited in arbitrary order.
sourcepub fn mapv_into_any<B, F>(self, f: F) -> Array<B, D>
pub fn mapv_into_any<B, F>(self, f: F) -> Array<B, D>
Consume the array, call f
by value on each element, and return an
owned array with the new values. Works for any F: FnMut(A)->B
.
If A
and B
are the same type then the map is performed by delegating
to mapv_into
and then converting into an owned array. This avoids
unnecessary memory allocations in mapv
.
If A
and B
are different types then a new array is allocated and the
map is performed as in mapv
.
Elements are visited in arbitrary order.
sourcepub fn map_inplace<'a, F>(&'a mut self, f: F)
pub fn map_inplace<'a, F>(&'a mut self, f: F)
Modify the array in place by calling f
by mutable reference on each element.
Elements are visited in arbitrary order.
sourcepub fn mapv_inplace<F>(&mut self, f: F)
pub fn mapv_inplace<F>(&mut self, f: F)
Modify the array in place by calling f
by value on each element.
The array is updated with the new values.
Elements are visited in arbitrary order.
use approx::assert_abs_diff_eq;
use ndarray::arr2;
let mut a = arr2(&[[ 0., 1.],
[-1., 2.]]);
a.mapv_inplace(f32::exp);
assert_abs_diff_eq!(
a,
arr2(&[[1.00000, 2.71828],
[0.36788, 7.38906]]),
epsilon = 1e-5,
);
sourcepub fn for_each<'a, F>(&'a self, f: F)
pub fn for_each<'a, F>(&'a self, f: F)
Call f
for each element in the array.
Elements are visited in arbitrary order.
sourcepub fn fold_axis<B, F>(
&self,
axis: Axis,
init: B,
fold: F,
) -> Array<B, D::Smaller>
pub fn fold_axis<B, F>( &self, axis: Axis, init: B, fold: F, ) -> Array<B, D::Smaller>
Fold along an axis.
Combine the elements of each subview with the previous using the fold
function and initial value init
.
Return the result as an Array
.
Panics if axis
is out of bounds.
sourcepub fn map_axis<'a, B, F>(
&'a self,
axis: Axis,
mapping: F,
) -> Array<B, D::Smaller>
pub fn map_axis<'a, B, F>( &'a self, axis: Axis, mapping: F, ) -> Array<B, D::Smaller>
Reduce the values along an axis into just one value, producing a new array with one less dimension.
Elements are visited in arbitrary order.
Return the result as an Array
.
Panics if axis
is out of bounds.
sourcepub fn map_axis_mut<'a, B, F>(
&'a mut self,
axis: Axis,
mapping: F,
) -> Array<B, D::Smaller>
pub fn map_axis_mut<'a, B, F>( &'a mut self, axis: Axis, mapping: F, ) -> Array<B, D::Smaller>
Reduce the values along an axis into just one value, producing a new array with one less dimension. 1-dimensional lanes are passed as mutable references to the reducer, allowing for side-effects.
Elements are visited in arbitrary order.
Return the result as an Array
.
Panics if axis
is out of bounds.
sourcepub fn remove_index(&mut self, axis: Axis, index: usize)
pub fn remove_index(&mut self, axis: Axis, index: usize)
Remove the index
th elements along axis
and shift down elements from higher indexes.
Note that this “removes” the elements by swapping them around to the end of the axis and shortening the length of the axis; the elements are not deinitialized or dropped by this, just moved out of view (this only matters for elements with ownership semantics). It’s similar to slicing an owned array in place.
Decreases the length of axis
by one.
Panics if axis
is out of bounds
Panics if not index < self.len_of(axis)
.
sourcepub fn accumulate_axis_inplace<F>(&mut self, axis: Axis, f: F)
pub fn accumulate_axis_inplace<F>(&mut self, axis: Axis, f: F)
Iterates over pairs of consecutive elements along the axis.
The first argument to the closure is an element, and the second argument is the next element along the axis. Iteration is guaranteed to proceed in order along the specified axis, but in all other respects the iteration order is unspecified.
§Example
For example, this can be used to compute the cumulative sum along an axis:
use ndarray::{array, Axis};
let mut arr = array![
[[1, 2], [3, 4], [5, 6]],
[[7, 8], [9, 10], [11, 12]],
];
arr.accumulate_axis_inplace(Axis(1), |&prev, curr| *curr += prev);
assert_eq!(
arr,
array![
[[1, 2], [4, 6], [9, 12]],
[[7, 8], [16, 18], [27, 30]],
],
);
source§impl<A> ArrayBase<OwnedRepr<A>, Dim<[usize; 0]>>
impl<A> ArrayBase<OwnedRepr<A>, Dim<[usize; 0]>>
Methods specific to Array0
.
See also all methods for ArrayBase
sourcepub fn into_scalar(self) -> A
pub fn into_scalar(self) -> A
Returns the single element in the array without cloning it.
use ndarray::{arr0, Array0};
// `Foo` doesn't implement `Clone`.
#[derive(Debug, Eq, PartialEq)]
struct Foo;
let array: Array0<Foo> = arr0(Foo);
let scalar: Foo = array.into_scalar();
assert_eq!(scalar, Foo);
source§impl<A, D> ArrayBase<OwnedRepr<A>, D>where
D: Dimension,
impl<A, D> ArrayBase<OwnedRepr<A>, D>where
D: Dimension,
Methods specific to Array
.
See also all methods for ArrayBase
sourcepub fn into_raw_vec_and_offset(self) -> (Vec<A>, Option<usize>)
pub fn into_raw_vec_and_offset(self) -> (Vec<A>, Option<usize>)
Return a vector of the elements in the array, in the way they are stored internally, and the index in the vector corresponding to the logically first element of the array (or 0 if the array is empty).
If the array is in standard memory layout, the logical element order
of the array (.iter()
order) and of the returned vector will be the same.
use ndarray::{array, Array2, Axis};
let mut arr: Array2<f64> = array![[1., 2.], [3., 4.], [5., 6.]];
arr.slice_axis_inplace(Axis(0), (1..).into());
assert_eq!(arr[[0, 0]], 3.);
let copy = arr.clone();
let shape = arr.shape().to_owned();
let strides = arr.strides().to_owned();
let (v, offset) = arr.into_raw_vec_and_offset();
assert_eq!(v, &[1., 2., 3., 4., 5., 6.]);
assert_eq!(offset, Some(2));
assert_eq!(v[offset.unwrap()], 3.);
for row in 0..shape[0] {
for col in 0..shape[1] {
let index = (
offset.unwrap() as isize
+ row as isize * strides[0]
+ col as isize * strides[1]
) as usize;
assert_eq!(v[index], copy[[row, col]]);
}
}
In the case of zero-sized elements, the offset to the logically first
element is somewhat meaningless. For convenience, an offset will be
returned such that all indices computed using the offset, shape, and
strides will be in-bounds for the Vec<A>
. Note that this offset won’t
necessarily be the same as the offset for an array of nonzero-sized
elements sliced in the same way.
use ndarray::{array, Array2, Axis};
let mut arr: Array2<()> = array![[(), ()], [(), ()], [(), ()]];
arr.slice_axis_inplace(Axis(0), (1..).into());
let shape = arr.shape().to_owned();
let strides = arr.strides().to_owned();
let (v, offset) = arr.into_raw_vec_and_offset();
assert_eq!(v, &[(), (), (), (), (), ()]);
for row in 0..shape[0] {
for col in 0..shape[1] {
let index = (
offset.unwrap() as isize
+ row as isize * strides[0]
+ col as isize * strides[1]
) as usize;
assert_eq!(v[index], ());
}
}
sourcepub fn into_raw_vec(self) -> Vec<A>
👎Deprecated since 0.16.0: Use .into_raw_vec_and_offset() instead
pub fn into_raw_vec(self) -> Vec<A>
Return a vector of the elements in the array, in the way they are stored internally.
Depending on slicing and strides, the logically first element of the
array can be located at an offset. Because of this, prefer to use
.into_raw_vec_and_offset()
instead.
source§impl<A> ArrayBase<OwnedRepr<A>, Dim<[usize; 2]>>
impl<A> ArrayBase<OwnedRepr<A>, Dim<[usize; 2]>>
Methods specific to Array2
.
See also all methods for ArrayBase
sourcepub fn push_row(&mut self, row: ArrayView<'_, A, Ix1>) -> Result<(), ShapeError>where
A: Clone,
pub fn push_row(&mut self, row: ArrayView<'_, A, Ix1>) -> Result<(), ShapeError>where
A: Clone,
Append a row to an array
The elements from row
are cloned and added as a new row in the array.
Errors with a shape error if the length of the row does not match the length of the rows in the array.
The memory layout of the self
array matters for ensuring that the append is efficient.
Appending automatically changes memory layout of the array so that it is appended to
along the “growing axis”. However, if the memory layout needs adjusting, the array must
reallocate and move memory.
The operation leaves the existing data in place and is most efficent if one of these is true:
- The axis being appended to is the longest stride axis, i.e the array is in row major (“C”) layout.
- The array has 0 or 1 rows (It is converted to row major)
Ensure appending is efficient by, for example, appending to an empty array and then always pushing/appending along the same axis. For pushing rows, ndarray’s default layout (C order) is efficient.
When repeatedly appending to a single axis, the amortized average complexity of each append is O(m), where m is the length of the row.
use ndarray::{Array, ArrayView, array};
// create an empty array and append
let mut a = Array::zeros((0, 4));
a.push_row(ArrayView::from(&[ 1., 2., 3., 4.])).unwrap();
a.push_row(ArrayView::from(&[-1., -2., -3., -4.])).unwrap();
assert_eq!(
a,
array![[ 1., 2., 3., 4.],
[-1., -2., -3., -4.]]);
sourcepub fn push_column(
&mut self,
column: ArrayView<'_, A, Ix1>,
) -> Result<(), ShapeError>where
A: Clone,
pub fn push_column(
&mut self,
column: ArrayView<'_, A, Ix1>,
) -> Result<(), ShapeError>where
A: Clone,
Append a column to an array
The elements from column
are cloned and added as a new column in the array.
Errors with a shape error if the length of the column does not match the length of the columns in the array.
The memory layout of the self
array matters for ensuring that the append is efficient.
Appending automatically changes memory layout of the array so that it is appended to
along the “growing axis”. However, if the memory layout needs adjusting, the array must
reallocate and move memory.
The operation leaves the existing data in place and is most efficent if one of these is true:
- The axis being appended to is the longest stride axis, i.e the array is in column major (“F”) layout.
- The array has 0 or 1 columns (It is converted to column major)
Ensure appending is efficient by, for example, appending to an empty array and then always pushing/appending along the same axis. For pushing columns, column major layout (F order) is efficient.
When repeatedly appending to a single axis, the amortized average complexity of each append is O(m), where m is the length of the column.
use ndarray::{Array, ArrayView, array};
// create an empty array and append
let mut a = Array::zeros((2, 0));
a.push_column(ArrayView::from(&[1., 2.])).unwrap();
a.push_column(ArrayView::from(&[-1., -2.])).unwrap();
assert_eq!(
a,
array![[1., -1.],
[2., -2.]]);
sourcepub fn reserve_rows(&mut self, additional: usize) -> Result<(), ShapeError>
pub fn reserve_rows(&mut self, additional: usize) -> Result<(), ShapeError>
Reserve capacity to grow array by at least additional
rows.
Existing elements of array
are untouched and the backing storage is grown by
calling the underlying reserve
method of the OwnedRepr
.
This is useful when pushing or appending repeatedly to an array to avoid multiple allocations.
Errors with a shape error if the resultant capacity is larger than the addressable
bounds; that is, the product of non-zero axis lengths once axis
has been extended by
additional
exceeds isize::MAX
.
use ndarray::Array2;
let mut a = Array2::<i32>::zeros((2,4));
a.reserve_rows(1000).unwrap();
assert!(a.into_raw_vec().capacity() >= 4*1002);
sourcepub fn reserve_columns(&mut self, additional: usize) -> Result<(), ShapeError>
pub fn reserve_columns(&mut self, additional: usize) -> Result<(), ShapeError>
Reserve capacity to grow array by at least additional
columns.
Existing elements of array
are untouched and the backing storage is grown by
calling the underlying reserve
method of the OwnedRepr
.
This is useful when pushing or appending repeatedly to an array to avoid multiple allocations.
Errors with a shape error if the resultant capacity is larger than the addressable
bounds; that is, the product of non-zero axis lengths once axis
has been extended by
additional
exceeds isize::MAX
.
use ndarray::Array2;
let mut a = Array2::<i32>::zeros((2,4));
a.reserve_columns(1000).unwrap();
assert!(a.into_raw_vec().capacity() >= 2*1002);
source§impl<A, D> ArrayBase<OwnedRepr<A>, D>where
D: Dimension,
impl<A, D> ArrayBase<OwnedRepr<A>, D>where
D: Dimension,
sourcepub fn move_into<'a, AM>(self, new_array: AM)where
AM: Into<ArrayViewMut<'a, A, D>>,
A: 'a,
pub fn move_into<'a, AM>(self, new_array: AM)where
AM: Into<ArrayViewMut<'a, A, D>>,
A: 'a,
Move all elements from self into new_array
, which must be of the same shape but
can have a different memory layout. The destination is overwritten completely.
The destination should be a mut reference to an array or an ArrayViewMut
with
A
elements.
Panics if the shapes don’t agree.
§Example
use ndarray::Array;
// Usage example of move_into in safe code
let mut a = Array::default((10, 10));
let b = Array::from_shape_fn((10, 10), |(i, j)| (i + j).to_string());
b.move_into(&mut a);
sourcepub fn move_into_uninit<'a, AM>(self, new_array: AM)
pub fn move_into_uninit<'a, AM>(self, new_array: AM)
Move all elements from self into new_array
, which must be of the same shape but
can have a different memory layout. The destination is overwritten completely.
The destination should be a mut reference to an array or an ArrayViewMut
with
MaybeUninit<A>
elements (which are overwritten without dropping any existing value).
Minor implementation note: Owned arrays like self
may be sliced in place and own elements
that are not part of their active view; these are dropped at the end of this function,
after all elements in the “active view” are moved into new_array
. If there is a panic in
drop of any such element, other elements may be leaked.
Panics if the shapes don’t agree.
§Example
use ndarray::Array;
let a = Array::from_iter(0..100).into_shape_with_order((10, 10)).unwrap();
let mut b = Array::uninit((10, 10));
a.move_into_uninit(&mut b);
unsafe {
// we can now promise we have fully initialized `b`.
let b = b.assume_init();
}
sourcepub fn push(
&mut self,
axis: Axis,
array: ArrayView<'_, A, D::Smaller>,
) -> Result<(), ShapeError>where
A: Clone,
D: RemoveAxis,
pub fn push(
&mut self,
axis: Axis,
array: ArrayView<'_, A, D::Smaller>,
) -> Result<(), ShapeError>where
A: Clone,
D: RemoveAxis,
Append an array to the array along an axis.
The elements of array
are cloned and extend the axis axis
in the present array;
self
will grow in size by 1 along axis
.
Append to the array, where the array being pushed to the array has one dimension less than
the self
array. This method is equivalent to append in this way:
self.append(axis, array.insert_axis(axis))
.
Errors with a shape error if the shape of self does not match the array-to-append;
all axes except the axis along which it being appended matter for this check:
the shape of self
with axis
removed must be the same as the shape of array
.
The memory layout of the self
array matters for ensuring that the append is efficient.
Appending automatically changes memory layout of the array so that it is appended to
along the “growing axis”. However, if the memory layout needs adjusting, the array must
reallocate and move memory.
The operation leaves the existing data in place and is most efficent if axis
is a
“growing axis” for the array, i.e. one of these is true:
- The axis is the longest stride axis, for example the 0th axis in a C-layout or the n-1th axis in an F-layout array.
- The axis has length 0 or 1 (It is converted to the new growing axis)
Ensure appending is efficient by for example starting from an empty array and/or always appending to an array along the same axis.
The amortized average complexity of the append, when appending along its growing axis, is O(m) where m is the number of individual elements to append.
The memory layout of the argument array
does not matter to the same extent.
use ndarray::{Array, ArrayView, array, Axis};
// create an empty array and push rows to it
let mut a = Array::zeros((0, 4));
let ones = ArrayView::from(&[1.; 4]);
let zeros = ArrayView::from(&[0.; 4]);
a.push(Axis(0), ones).unwrap();
a.push(Axis(0), zeros).unwrap();
a.push(Axis(0), ones).unwrap();
assert_eq!(
a,
array![[1., 1., 1., 1.],
[0., 0., 0., 0.],
[1., 1., 1., 1.]]);
sourcepub fn append(
&mut self,
axis: Axis,
array: ArrayView<'_, A, D>,
) -> Result<(), ShapeError>where
A: Clone,
D: RemoveAxis,
pub fn append(
&mut self,
axis: Axis,
array: ArrayView<'_, A, D>,
) -> Result<(), ShapeError>where
A: Clone,
D: RemoveAxis,
Append an array to the array along an axis.
The elements of array
are cloned and extend the axis axis
in the present array;
self
will grow in size by array.len_of(axis)
along axis
.
Errors with a shape error if the shape of self does not match the array-to-append;
all axes except the axis along which it being appended matter for this check:
the shape of self
with axis
removed must be the same as the shape of array
with
axis
removed.
The memory layout of the self
array matters for ensuring that the append is efficient.
Appending automatically changes memory layout of the array so that it is appended to
along the “growing axis”. However, if the memory layout needs adjusting, the array must
reallocate and move memory.
The operation leaves the existing data in place and is most efficent if axis
is a
“growing axis” for the array, i.e. one of these is true:
- The axis is the longest stride axis, for example the 0th axis in a C-layout or the n-1th axis in an F-layout array.
- The axis has length 0 or 1 (It is converted to the new growing axis)
Ensure appending is efficient by for example starting from an empty array and/or always appending to an array along the same axis.
The amortized average complexity of the append, when appending along its growing axis, is O(m) where m is the number of individual elements to append.
The memory layout of the argument array
does not matter to the same extent.
use ndarray::{Array, ArrayView, array, Axis};
// create an empty array and append two rows at a time
let mut a = Array::zeros((0, 4));
let ones = ArrayView::from(&[1.; 8]).into_shape_with_order((2, 4)).unwrap();
let zeros = ArrayView::from(&[0.; 8]).into_shape_with_order((2, 4)).unwrap();
a.append(Axis(0), ones).unwrap();
a.append(Axis(0), zeros).unwrap();
a.append(Axis(0), ones).unwrap();
assert_eq!(
a,
array![[1., 1., 1., 1.],
[1., 1., 1., 1.],
[0., 0., 0., 0.],
[0., 0., 0., 0.],
[1., 1., 1., 1.],
[1., 1., 1., 1.]]);
sourcepub fn reserve(
&mut self,
axis: Axis,
additional: usize,
) -> Result<(), ShapeError>where
D: RemoveAxis,
pub fn reserve(
&mut self,
axis: Axis,
additional: usize,
) -> Result<(), ShapeError>where
D: RemoveAxis,
Reserve capacity to grow array along axis
by at least additional
elements.
The axis should be in the range Axis(
0 .. n )
where n is the
number of dimensions (axes) of the array.
Existing elements of array
are untouched and the backing storage is grown by
calling the underlying reserve
method of the OwnedRepr
.
This is useful when pushing or appending repeatedly to an array to avoid multiple allocations.
Panics if the axis is out of bounds.
Errors with a shape error if the resultant capacity is larger than the addressable
bounds; that is, the product of non-zero axis lengths once axis
has been extended by
additional
exceeds isize::MAX
.
use ndarray::{Array3, Axis};
let mut a = Array3::<i32>::zeros((0,2,4));
a.reserve(Axis(0), 1000).unwrap();
assert!(a.into_raw_vec().capacity() >= 2*4*1000);
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
Methods specific to arrays with MaybeUninit
elements.
See also all methods for ArrayBase
sourcepub unsafe fn assume_init(self) -> ArrayBase<<S as RawDataSubst<A>>::Output, D>
pub unsafe fn assume_init(self) -> ArrayBase<<S as RawDataSubst<A>>::Output, D>
Promise that the array’s elements are all fully initialized, and convert
the array from element type MaybeUninit<A>
to A
.
For example, it can convert an Array<MaybeUninit<f64>, D>
to Array<f64, D>
.
§Safety
Safe to use if all the array’s elements have been initialized.
Note that for owned and shared ownership arrays, the promise must include all of the array’s storage; it is for example possible to slice these in place, but that must only be done after all elements have been initialized.
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
§Parallel methods
These methods require crate feature rayon
.
sourcepub fn par_map_inplace<F>(&mut self, f: F)
pub fn par_map_inplace<F>(&mut self, f: F)
Parallel version of map_inplace
.
Modify the array in place by calling f
by mutable reference on each element.
Elements are visited in arbitrary order.
source§impl<A, S> ArrayBase<S, Ix2>where
S: RawData<Elem = A>,
impl<A, S> ArrayBase<S, Ix2>where
S: RawData<Elem = A>,
§Methods For 2-D Arrays
sourcepub fn row(&self, index: Ix) -> ArrayView1<'_, A>where
S: Data,
pub fn row(&self, index: Ix) -> ArrayView1<'_, A>where
S: Data,
Return an array view of row index
.
Panics if index
is out of bounds.
use ndarray::array;
let array = array![[1., 2.], [3., 4.]];
assert_eq!(array.row(0), array![1., 2.]);
sourcepub fn row_mut(&mut self, index: Ix) -> ArrayViewMut1<'_, A>where
S: DataMut,
pub fn row_mut(&mut self, index: Ix) -> ArrayViewMut1<'_, A>where
S: DataMut,
Return a mutable array view of row index
.
Panics if index
is out of bounds.
use ndarray::array;
let mut array = array![[1., 2.], [3., 4.]];
array.row_mut(0)[1] = 5.;
assert_eq!(array, array![[1., 5.], [3., 4.]]);
sourcepub fn nrows(&self) -> usize
pub fn nrows(&self) -> usize
Return the number of rows (length of Axis(0)
) in the two-dimensional array.
use ndarray::{array, Axis};
let array = array![[1., 2.],
[3., 4.],
[5., 6.]];
assert_eq!(array.nrows(), 3);
// equivalent ways of getting the dimensions
// get nrows, ncols by using dim:
let (m, n) = array.dim();
assert_eq!(m, array.nrows());
// get length of any particular axis with .len_of()
assert_eq!(m, array.len_of(Axis(0)));
sourcepub fn column(&self, index: Ix) -> ArrayView1<'_, A>where
S: Data,
pub fn column(&self, index: Ix) -> ArrayView1<'_, A>where
S: Data,
Return an array view of column index
.
Panics if index
is out of bounds.
use ndarray::array;
let array = array![[1., 2.], [3., 4.]];
assert_eq!(array.column(0), array![1., 3.]);
sourcepub fn column_mut(&mut self, index: Ix) -> ArrayViewMut1<'_, A>where
S: DataMut,
pub fn column_mut(&mut self, index: Ix) -> ArrayViewMut1<'_, A>where
S: DataMut,
Return a mutable array view of column index
.
Panics if index
is out of bounds.
use ndarray::array;
let mut array = array![[1., 2.], [3., 4.]];
array.column_mut(0)[1] = 5.;
assert_eq!(array, array![[1., 2.], [5., 4.]]);
sourcepub fn ncols(&self) -> usize
pub fn ncols(&self) -> usize
Return the number of columns (length of Axis(1)
) in the two-dimensional array.
use ndarray::{array, Axis};
let array = array![[1., 2.],
[3., 4.],
[5., 6.]];
assert_eq!(array.ncols(), 2);
// equivalent ways of getting the dimensions
// get nrows, ncols by using dim:
let (m, n) = array.dim();
assert_eq!(n, array.ncols());
// get length of any particular axis with .len_of()
assert_eq!(n, array.len_of(Axis(1)));
sourcepub fn is_square(&self) -> bool
pub fn is_square(&self) -> bool
Return true if the array is square, false otherwise.
§Examples
Square:
use ndarray::array;
let array = array![[1., 2.], [3., 4.]];
assert!(array.is_square());
Not square:
use ndarray::array;
let array = array![[1., 2., 5.], [3., 4., 6.]];
assert!(!array.is_square());
source§impl<A, S> ArrayBase<S, IxDyn>where
S: Data<Elem = A>,
impl<A, S> ArrayBase<S, IxDyn>where
S: Data<Elem = A>,
§Methods for Dynamic-Dimensional Arrays
sourcepub fn insert_axis_inplace(&mut self, axis: Axis)
pub fn insert_axis_inplace(&mut self, axis: Axis)
Insert new array axis of length 1 at axis
, modifying the shape and
strides in-place.
Panics if the axis is out of bounds.
use ndarray::{Axis, arr2, arr3};
let mut a = arr2(&[[1, 2, 3], [4, 5, 6]]).into_dyn();
assert_eq!(a.shape(), &[2, 3]);
a.insert_axis_inplace(Axis(1));
assert_eq!(a, arr3(&[[[1, 2, 3]], [[4, 5, 6]]]).into_dyn());
assert_eq!(a.shape(), &[2, 1, 3]);
sourcepub fn index_axis_inplace(&mut self, axis: Axis, index: usize)
pub fn index_axis_inplace(&mut self, axis: Axis, index: usize)
Collapses the array to index
along the axis and removes the axis,
modifying the shape and strides in-place.
Panics if axis
or index
is out of bounds.
use ndarray::{Axis, arr1, arr2};
let mut a = arr2(&[[1, 2, 3], [4, 5, 6]]).into_dyn();
assert_eq!(a.shape(), &[2, 3]);
a.index_axis_inplace(Axis(1), 1);
assert_eq!(a, arr1(&[2, 5]).into_dyn());
assert_eq!(a.shape(), &[2]);
sourcepub fn squeeze(self) -> Self
pub fn squeeze(self) -> Self
Remove axes of length 1 and return the modified array.
If the array has more the one dimension, the result array will always have at least one dimension, even if it has a length of 1.
use ndarray::{arr1, arr2, arr3};
let a = arr3(&[[[1, 2, 3]], [[4, 5, 6]]]).into_dyn();
assert_eq!(a.shape(), &[2, 1, 3]);
let b = a.squeeze();
assert_eq!(b, arr2(&[[1, 2, 3], [4, 5, 6]]).into_dyn());
assert_eq!(b.shape(), &[2, 3]);
let c = arr2(&[[1]]).into_dyn();
assert_eq!(c.shape(), &[1, 1]);
let d = c.squeeze();
assert_eq!(d, arr1(&[1]).into_dyn());
assert_eq!(d.shape(), &[1]);
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
§Numerical Methods for Arrays
sourcepub fn sum(&self) -> A
pub fn sum(&self) -> A
Return the sum of all elements in the array.
use ndarray::arr2;
let a = arr2(&[[1., 2.],
[3., 4.]]);
assert_eq!(a.sum(), 10.);
sourcepub fn mean(&self) -> Option<A>
pub fn mean(&self) -> Option<A>
Returns the arithmetic mean x̅ of all elements in the array:
1 n
x̅ = ― ∑ xᵢ
n i=1
If the array is empty, None
is returned.
Panics if A::from_usize()
fails to convert the number of elements in the array.
sourcepub fn product(&self) -> A
pub fn product(&self) -> A
Return the product of all elements in the array.
use ndarray::arr2;
let a = arr2(&[[1., 2.],
[3., 4.]]);
assert_eq!(a.product(), 24.);
sourcepub fn var(&self, ddof: A) -> Awhere
A: Float + FromPrimitive,
pub fn var(&self, ddof: A) -> Awhere
A: Float + FromPrimitive,
Return variance of elements in the array.
The variance is computed using the Welford one-pass algorithm.
The parameter ddof
specifies the “delta degrees of freedom”. For
example, to calculate the population variance, use ddof = 0
, or to
calculate the sample variance, use ddof = 1
.
The variance is defined as:
1 n
variance = ―――――――― ∑ (xᵢ - x̅)²
n - ddof i=1
where
1 n
x̅ = ― ∑ xᵢ
n i=1
and n
is the length of the array.
Panics if ddof
is less than zero or greater than n
§Example
use ndarray::array;
use approx::assert_abs_diff_eq;
let a = array![1., -4.32, 1.14, 0.32];
let var = a.var(1.);
assert_abs_diff_eq!(var, 6.7331, epsilon = 1e-4);
sourcepub fn std(&self, ddof: A) -> Awhere
A: Float + FromPrimitive,
pub fn std(&self, ddof: A) -> Awhere
A: Float + FromPrimitive,
Return standard deviation of elements in the array.
The standard deviation is computed from the variance using the Welford one-pass algorithm.
The parameter ddof
specifies the “delta degrees of freedom”. For
example, to calculate the population standard deviation, use ddof = 0
,
or to calculate the sample standard deviation, use ddof = 1
.
The standard deviation is defined as:
⎛ 1 n ⎞
stddev = sqrt ⎜ ―――――――― ∑ (xᵢ - x̅)²⎟
⎝ n - ddof i=1 ⎠
where
1 n
x̅ = ― ∑ xᵢ
n i=1
and n
is the length of the array.
Panics if ddof
is less than zero or greater than n
§Example
use ndarray::array;
use approx::assert_abs_diff_eq;
let a = array![1., -4.32, 1.14, 0.32];
let stddev = a.std(1.);
assert_abs_diff_eq!(stddev, 2.59483, epsilon = 1e-4);
sourcepub fn sum_axis(&self, axis: Axis) -> Array<A, D::Smaller>
pub fn sum_axis(&self, axis: Axis) -> Array<A, D::Smaller>
Return sum along axis
.
use ndarray::{aview0, aview1, arr2, Axis};
let a = arr2(&[[1., 2., 3.],
[4., 5., 6.]]);
assert!(
a.sum_axis(Axis(0)) == aview1(&[5., 7., 9.]) &&
a.sum_axis(Axis(1)) == aview1(&[6., 15.]) &&
a.sum_axis(Axis(0)).sum_axis(Axis(0)) == aview0(&21.)
);
Panics if axis
is out of bounds.
sourcepub fn product_axis(&self, axis: Axis) -> Array<A, D::Smaller>
pub fn product_axis(&self, axis: Axis) -> Array<A, D::Smaller>
Return product along axis
.
The product of an empty array is 1.
use ndarray::{aview0, aview1, arr2, Axis};
let a = arr2(&[[1., 2., 3.],
[4., 5., 6.]]);
assert!(
a.product_axis(Axis(0)) == aview1(&[4., 10., 18.]) &&
a.product_axis(Axis(1)) == aview1(&[6., 120.]) &&
a.product_axis(Axis(0)).product_axis(Axis(0)) == aview0(&720.)
);
Panics if axis
is out of bounds.
sourcepub fn mean_axis(&self, axis: Axis) -> Option<Array<A, D::Smaller>>
pub fn mean_axis(&self, axis: Axis) -> Option<Array<A, D::Smaller>>
Return mean along axis
.
Return None
if the length of the axis is zero.
Panics if axis
is out of bounds or if A::from_usize()
fails for the axis length.
use ndarray::{aview0, aview1, arr2, Axis};
let a = arr2(&[[1., 2., 3.],
[4., 5., 6.]]);
assert!(
a.mean_axis(Axis(0)).unwrap() == aview1(&[2.5, 3.5, 4.5]) &&
a.mean_axis(Axis(1)).unwrap() == aview1(&[2., 5.]) &&
a.mean_axis(Axis(0)).unwrap().mean_axis(Axis(0)).unwrap() == aview0(&3.5)
);
sourcepub fn var_axis(&self, axis: Axis, ddof: A) -> Array<A, D::Smaller>
pub fn var_axis(&self, axis: Axis, ddof: A) -> Array<A, D::Smaller>
Return variance along axis
.
The variance is computed using the Welford one-pass algorithm.
The parameter ddof
specifies the “delta degrees of freedom”. For
example, to calculate the population variance, use ddof = 0
, or to
calculate the sample variance, use ddof = 1
.
The variance is defined as:
1 n
variance = ―――――――― ∑ (xᵢ - x̅)²
n - ddof i=1
where
1 n
x̅ = ― ∑ xᵢ
n i=1
and n
is the length of the axis.
Panics if ddof
is less than zero or greater than n
, if axis
is out of bounds, or if A::from_usize()
fails for any any of the
numbers in the range 0..=n
.
§Example
use ndarray::{aview1, arr2, Axis};
let a = arr2(&[[1., 2.],
[3., 4.],
[5., 6.]]);
let var = a.var_axis(Axis(0), 1.);
assert_eq!(var, aview1(&[4., 4.]));
sourcepub fn std_axis(&self, axis: Axis, ddof: A) -> Array<A, D::Smaller>
pub fn std_axis(&self, axis: Axis, ddof: A) -> Array<A, D::Smaller>
Return standard deviation along axis
.
The standard deviation is computed from the variance using the Welford one-pass algorithm.
The parameter ddof
specifies the “delta degrees of freedom”. For
example, to calculate the population standard deviation, use ddof = 0
,
or to calculate the sample standard deviation, use ddof = 1
.
The standard deviation is defined as:
⎛ 1 n ⎞
stddev = sqrt ⎜ ―――――――― ∑ (xᵢ - x̅)²⎟
⎝ n - ddof i=1 ⎠
where
1 n
x̅ = ― ∑ xᵢ
n i=1
and n
is the length of the axis.
Panics if ddof
is less than zero or greater than n
, if axis
is out of bounds, or if A::from_usize()
fails for any any of the
numbers in the range 0..=n
.
§Example
use ndarray::{aview1, arr2, Axis};
let a = arr2(&[[1., 2.],
[3., 4.],
[5., 6.]]);
let stddev = a.std_axis(Axis(0), 1.);
assert_eq!(stddev, aview1(&[2., 2.]));
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
§Element-wise methods for float arrays
Element-wise math functions for any array type that contains float number.
sourcepub fn is_nan(&self) -> Array<bool, D>
pub fn is_nan(&self) -> Array<bool, D>
If the number is NaN
(not a number), then true
is returned for each element.
sourcepub fn is_all_nan(&self) -> bool
pub fn is_all_nan(&self) -> bool
Return true
if all elements are NaN
(not a number).
sourcepub fn is_any_nan(&self) -> bool
pub fn is_any_nan(&self) -> bool
Return true
if any element is NaN
(not a number).
sourcepub fn is_infinite(&self) -> Array<bool, D>
pub fn is_infinite(&self) -> Array<bool, D>
If the number is infinity, then true
is returned for each element.
sourcepub fn is_all_infinite(&self) -> bool
pub fn is_all_infinite(&self) -> bool
Return true
if all elements are infinity.
sourcepub fn is_any_infinite(&self) -> bool
pub fn is_any_infinite(&self) -> bool
Return true
if any element is infinity.
sourcepub fn signum(&self) -> Array<A, D>
pub fn signum(&self) -> Array<A, D>
Sign number of each element.
1.0
for all positive numbers.-1.0
for all negative numbers.NaN
for allNaN
(not a number).
sourcepub fn to_degrees(&self) -> Array<A, D>
pub fn to_degrees(&self) -> Array<A, D>
Converts radians to degrees for each element.
sourcepub fn to_radians(&self) -> Array<A, D>
pub fn to_radians(&self) -> Array<A, D>
Converts degrees to radians for each element.
sourcepub fn powi(&self, rhs: i32) -> Array<A, D>
pub fn powi(&self, rhs: i32) -> Array<A, D>
Integer power of each element.
This function is generally faster than using float power.
sourcepub fn log(&self, rhs: A) -> Array<A, D>
pub fn log(&self, rhs: A) -> Array<A, D>
Logarithm of each element with respect to an arbitrary base.
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
sourcepub fn clamp(&self, min: A, max: A) -> Array<A, D>
pub fn clamp(&self, min: A, max: A) -> Array<A, D>
Limit the values for each element, similar to NumPy’s clip
function.
use ndarray::array;
let a = array![0., 1., 2., 3., 4., 5., 6., 7., 8., 9.];
assert_eq!(a.clamp(1., 8.), array![1., 1., 2., 3., 4., 5., 6., 7., 8., 8.]);
assert_eq!(a.clamp(3., 6.), array![3., 3., 3., 3., 4., 5., 6., 6., 6., 6.]);
§Panics
Panics if !(min <= max)
.
source§impl<A, S> ArrayBase<S, Ix1>where
S: Data<Elem = A>,
impl<A, S> ArrayBase<S, Ix1>where
S: Data<Elem = A>,
sourcepub fn dot<Rhs>(&self, rhs: &Rhs) -> <Self as Dot<Rhs>>::Outputwhere
Self: Dot<Rhs>,
pub fn dot<Rhs>(&self, rhs: &Rhs) -> <Self as Dot<Rhs>>::Outputwhere
Self: Dot<Rhs>,
Perform dot product or matrix multiplication of arrays self
and rhs
.
Rhs
may be either a one-dimensional or a two-dimensional array.
If Rhs
is one-dimensional, then the operation is a vector dot
product, which is the sum of the elementwise products (no conjugation
of complex operands, and thus not their inner product). In this case,
self
and rhs
must be the same length.
If Rhs
is two-dimensional, then the operation is matrix
multiplication, where self
is treated as a row vector. In this case,
if self
is shape M, then rhs
is shape M × N and the result is
shape N.
Panics if the array shapes are incompatible.
Note: If enabled, uses blas dot
for elements of f32, f64
when memory
layout allows.
source§impl<A, S> ArrayBase<S, Ix2>where
S: Data<Elem = A>,
impl<A, S> ArrayBase<S, Ix2>where
S: Data<Elem = A>,
sourcepub fn dot<Rhs>(&self, rhs: &Rhs) -> <Self as Dot<Rhs>>::Outputwhere
Self: Dot<Rhs>,
pub fn dot<Rhs>(&self, rhs: &Rhs) -> <Self as Dot<Rhs>>::Outputwhere
Self: Dot<Rhs>,
Perform matrix multiplication of rectangular arrays self
and rhs
.
Rhs
may be either a one-dimensional or a two-dimensional array.
If Rhs is two-dimensional, they array shapes must agree in the way that
if self
is M × N, then rhs
is N × K.
Return a result array with shape M × K.
Panics if shapes are incompatible or the number of elements in the
result would overflow isize
.
Note: If enabled, uses blas gemv/gemm
for elements of f32, f64
when memory layout allows. The default matrixmultiply backend
is otherwise used for f32, f64
for all memory layouts.
use ndarray::arr2;
let a = arr2(&[[1., 2.],
[0., 1.]]);
let b = arr2(&[[1., 2.],
[2., 3.]]);
assert!(
a.dot(&b) == arr2(&[[5., 8.],
[2., 3.]])
);
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
sourcepub fn scaled_add<S2, E>(&mut self, alpha: A, rhs: &ArrayBase<S2, E>)
pub fn scaled_add<S2, E>(&mut self, alpha: A, rhs: &ArrayBase<S2, E>)
Perform the operation self += alpha * rhs
efficiently, where
alpha
is a scalar and rhs
is another array. This operation is
also known as axpy
in BLAS.
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
source§impl<A, S, D> ArrayBase<S, D>
impl<A, S, D> ArrayBase<S, D>
sourcepub fn abs_diff_eq<S2>(
&self,
other: &ArrayBase<S2, D>,
epsilon: A::Epsilon,
) -> bool
pub fn abs_diff_eq<S2>( &self, other: &ArrayBase<S2, D>, epsilon: A::Epsilon, ) -> bool
A test for equality that uses the elementwise absolute difference to compute the approximate equality of two arrays.
Requires crate feature "approx"
source§impl<'a, A, D> ArrayBase<ViewRepr<&'a A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<ViewRepr<&'a A>, D>where
D: Dimension,
Methods for read-only array views.
sourcepub fn from_shape<Sh>(shape: Sh, xs: &'a [A]) -> Result<Self, ShapeError>where
Sh: Into<StrideShape<D>>,
pub fn from_shape<Sh>(shape: Sh, xs: &'a [A]) -> Result<Self, ShapeError>where
Sh: Into<StrideShape<D>>,
Create a read-only array view borrowing its data from a slice.
Checks whether shape
are compatible with the slice’s
length, returning an Err
if not compatible.
use ndarray::ArrayView;
use ndarray::arr3;
use ndarray::ShapeBuilder;
// advanced example where we are even specifying exact strides to use (which is optional).
let s = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12];
let a = ArrayView::from_shape((2, 3, 2).strides((1, 4, 2)),
&s).unwrap();
assert!(
a == arr3(&[[[0, 2],
[4, 6],
[8, 10]],
[[1, 3],
[5, 7],
[9, 11]]])
);
assert!(a.strides() == &[1, 4, 2]);
sourcepub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *const A) -> Selfwhere
Sh: Into<StrideShape<D>>,
pub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *const A) -> Selfwhere
Sh: Into<StrideShape<D>>,
Create an ArrayView<A, D>
from shape information and a raw pointer to
the elements.
§Safety
The caller is responsible for ensuring all of the following:
-
The elements seen by moving
ptr
according to the shape and strides must live at least as long as'a
and must not be not mutably aliased for the duration of'a
. -
ptr
must be non-null and aligned, and it must be safe to.offset()
ptr
by zero. -
It must be safe to
.offset()
the pointer repeatedly along all axes and calculate thecount
s for the.offset()
calls without overflow, even if the array is empty or the elements are zero-sized.In other words,
-
All possible pointers generated by moving along all axes must be in bounds or one byte past the end of a single allocation with element type
A
. The only exceptions are if the array is empty or the element type is zero-sized. In these cases,ptr
may be dangling, but it must still be safe to.offset()
the pointer along the axes. -
The offset in units of bytes between the least address and greatest address by moving along all axes must not exceed
isize::MAX
. This constraint prevents the computed offset, in bytes, from overflowingisize
regardless of the starting point due to past offsets. -
The offset in units of
A
between the least address and greatest address by moving along all axes must not exceedisize::MAX
. This constraint prevents overflow when calculating thecount
parameter to.offset()
regardless of the starting point due to past offsets.
-
-
The product of non-zero axis lengths must not exceed
isize::MAX
. -
Strides must be non-negative.
This function can use debug assertions to check some of these requirements, but it’s not a complete check.
source§impl<'a, A, D> ArrayBase<ViewRepr<&'a mut A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<ViewRepr<&'a mut A>, D>where
D: Dimension,
Methods for read-write array views.
sourcepub fn from_shape<Sh>(shape: Sh, xs: &'a mut [A]) -> Result<Self, ShapeError>where
Sh: Into<StrideShape<D>>,
pub fn from_shape<Sh>(shape: Sh, xs: &'a mut [A]) -> Result<Self, ShapeError>where
Sh: Into<StrideShape<D>>,
Create a read-write array view borrowing its data from a slice.
Checks whether dim
and strides
are compatible with the slice’s
length, returning an Err
if not compatible.
use ndarray::ArrayViewMut;
use ndarray::arr3;
use ndarray::ShapeBuilder;
let mut s = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12];
let mut a = ArrayViewMut::from_shape((2, 3, 2).strides((1, 4, 2)),
&mut s).unwrap();
a[[0, 0, 0]] = 1;
assert!(
a == arr3(&[[[1, 2],
[4, 6],
[8, 10]],
[[1, 3],
[5, 7],
[9, 11]]])
);
assert!(a.strides() == &[1, 4, 2]);
sourcepub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *mut A) -> Selfwhere
Sh: Into<StrideShape<D>>,
pub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *mut A) -> Selfwhere
Sh: Into<StrideShape<D>>,
Create an ArrayViewMut<A, D>
from shape information and a
raw pointer to the elements.
§Safety
The caller is responsible for ensuring all of the following:
-
The elements seen by moving
ptr
according to the shape and strides must live at least as long as'a
and must not be aliased for the duration of'a
. -
ptr
must be non-null and aligned, and it must be safe to.offset()
ptr
by zero. -
It must be safe to
.offset()
the pointer repeatedly along all axes and calculate thecount
s for the.offset()
calls without overflow, even if the array is empty or the elements are zero-sized.In other words,
-
All possible pointers generated by moving along all axes must be in bounds or one byte past the end of a single allocation with element type
A
. The only exceptions are if the array is empty or the element type is zero-sized. In these cases,ptr
may be dangling, but it must still be safe to.offset()
the pointer along the axes. -
The offset in units of bytes between the least address and greatest address by moving along all axes must not exceed
isize::MAX
. This constraint prevents the computed offset, in bytes, from overflowingisize
regardless of the starting point due to past offsets. -
The offset in units of
A
between the least address and greatest address by moving along all axes must not exceedisize::MAX
. This constraint prevents overflow when calculating thecount
parameter to.offset()
regardless of the starting point due to past offsets.
-
-
The product of non-zero axis lengths must not exceed
isize::MAX
. -
Strides must be non-negative.
This function can use debug assertions to check some of these requirements, but it’s not a complete check.
sourcepub fn reborrow<'b>(self) -> ArrayViewMut<'b, A, D>where
'a: 'b,
pub fn reborrow<'b>(self) -> ArrayViewMut<'b, A, D>where
'a: 'b,
Convert the view into an ArrayViewMut<'b, A, D>
where 'b
is a lifetime
outlived by 'a'
.
source§impl<'a, A, D> ArrayBase<ViewRepr<&'a A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<ViewRepr<&'a A>, D>where
D: Dimension,
Methods for read-only array views.
sourcepub fn reborrow<'b>(self) -> ArrayView<'b, A, D>where
'a: 'b,
pub fn reborrow<'b>(self) -> ArrayView<'b, A, D>where
'a: 'b,
Convert the view into an ArrayView<'b, A, D>
where 'b
is a lifetime
outlived by 'a'
.
sourcepub fn to_slice(&self) -> Option<&'a [A]>
pub fn to_slice(&self) -> Option<&'a [A]>
Return the array’s data as a slice, if it is contiguous and in standard order.
Return None
otherwise.
Note that while the method is similar to ArrayBase::as_slice()
, this method transfers
the view’s lifetime to the slice, so it is a bit more powerful.
sourcepub fn to_slice_memory_order(&self) -> Option<&'a [A]>
pub fn to_slice_memory_order(&self) -> Option<&'a [A]>
Return the array’s data as a slice, if it is contiguous.
Return None
otherwise.
Note that while the method is similar to
ArrayBase::as_slice_memory_order()
, this method transfers the view’s
lifetime to the slice, so it is a bit more powerful.
source§impl<'a, A> ArrayBase<ViewRepr<&'a A>, Dim<[usize; 0]>>
impl<'a, A> ArrayBase<ViewRepr<&'a A>, Dim<[usize; 0]>>
sourcepub fn into_scalar(self) -> &'a A
pub fn into_scalar(self) -> &'a A
Consume the view and return a reference to the single element in the array.
The lifetime of the returned reference matches the lifetime of the data the array view was pointing to.
use ndarray::{arr0, Array0};
// `Foo` doesn't implement `Clone`.
#[derive(Debug, Eq, PartialEq)]
struct Foo;
let array: Array0<Foo> = arr0(Foo);
let view = array.view();
let scalar: &Foo = view.into_scalar();
assert_eq!(scalar, &Foo);
source§impl<'a, A> ArrayBase<ViewRepr<&'a mut A>, Dim<[usize; 0]>>
impl<'a, A> ArrayBase<ViewRepr<&'a mut A>, Dim<[usize; 0]>>
Methods specific to ArrayViewMut0
.
See also all methods for ArrayViewMut
and ArrayBase
sourcepub fn into_scalar(self) -> &'a mut A
pub fn into_scalar(self) -> &'a mut A
Consume the mutable view and return a mutable reference to the single element in the array.
The lifetime of the returned reference matches the lifetime of the data the array view was pointing to.
use ndarray::{arr0, Array0};
let mut array: Array0<f64> = arr0(5.);
let view = array.view_mut();
let scalar = view.into_scalar();
*scalar = 7.;
assert_eq!(scalar, &7.);
assert_eq!(array[()], 7.);
source§impl<'a, A, D> ArrayBase<ViewRepr<&'a mut A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<ViewRepr<&'a mut A>, D>where
D: Dimension,
Methods for read-write array views.
sourcepub fn into_slice(self) -> Option<&'a mut [A]>
pub fn into_slice(self) -> Option<&'a mut [A]>
Return the array’s data as a slice, if it is contiguous and in standard order.
Return None
otherwise.
Note that while this is similar to ArrayBase::as_slice_mut()
, this method transfers the
view’s lifetime to the slice.
sourcepub fn into_slice_memory_order(self) -> Option<&'a mut [A]>
pub fn into_slice_memory_order(self) -> Option<&'a mut [A]>
Return the array’s data as a slice, if it is contiguous.
Return None
otherwise.
Note that while this is similar to
ArrayBase::as_slice_memory_order_mut()
, this method transfers the
view’s lifetime to the slice.
sourcepub fn into_cell_view(self) -> ArrayView<'a, MathCell<A>, D>
pub fn into_cell_view(self) -> ArrayView<'a, MathCell<A>, D>
Return a shared view of the array with elements as if they were embedded in cells.
The cell view itself can be copied and accessed without exclusivity.
The view acts “as if” the elements are temporarily in cells, and elements can be changed through shared references using the regular cell methods.
source§impl<'a, A, D> ArrayBase<ViewRepr<&'a A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<ViewRepr<&'a A>, D>where
D: Dimension,
Methods for read-only array views.
sourcepub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
pub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
Split the array view along axis
and return one view strictly before the
split and one view after the split.
Panics if axis
or index
is out of bounds.
Examples:
let a = aview2(&[[0, 1, 2, 3],
[4, 5, 6, 7],
[8, 9, 0, 1]]);
The array view a
has two axes and shape 3 × 4:
──▶ Axis(1)
┌─────┬─────┬─────┬─────┐ 0
│ │ a₀₀ │ a₀₁ │ a₀₂ │ a₀₃ │
▼ ├─────┼─────┼─────┼─────┤ 1
Axis(0)│ a₁₀ │ a₁₁ │ a₁₂ │ a₁₃ │
├─────┼─────┼─────┼─────┤ 2
│ a₂₀ │ a₂₁ │ a₂₂ │ a₂₃ │
└─────┴─────┴─────┴─────┘ 3 ↑
0 1 2 3 4 ← possible split_at indices.
Row indices increase along Axis(0)
, and column indices increase along
Axis(1)
. Note that we split “before” an element index, and that
both 0 and the endpoint are valid split indices.
Example 1: Split a
along the first axis, in this case the rows, at
index 2.
This produces views v1 and v2 of shapes 2 × 4 and 1 × 4:
let (v1, v2) = a.split_at(Axis(0), 2);
┌─────┬─────┬─────┬─────┐ 0 ↓ indices
│ a₀₀ │ a₀₁ │ a₀₂ │ a₀₃ │ along Axis(0)
├─────┼─────┼─────┼─────┤ v1 1
│ a₁₀ │ a₁₁ │ a₁₂ │ a₁₃ │
└─────┴─────┴─────┴─────┘
┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄ 2
┌─────┬─────┬─────┬─────┐
│ a₂₀ │ a₂₁ │ a₂₂ │ a₂₃ │ v2
└─────┴─────┴─────┴─────┘ 3
Example 2: Split a
along the second axis, in this case the
columns, at index 2.
This produces views u1 and u2 of shapes 3 × 2 and 3 × 2:
let (u1, u2) = a.split_at(Axis(1), 2);
u1 u2
┌─────┬─────┐┊┌─────┬─────┐
│ a₀₀ │ a₀₁ │┊│ a₀₂ │ a₀₃ │
├─────┼─────┤┊├─────┼─────┤
│ a₁₀ │ a₁₁ │┊│ a₁₂ │ a₁₃ │
├─────┼─────┤┊├─────┼─────┤
│ a₂₀ │ a₂₁ │┊│ a₂₂ │ a₂₃ │
└─────┴─────┘┊└─────┴─────┘
0 1 2 3 4 indices →
along Axis(1)
source§impl<'a, T, D> ArrayBase<ViewRepr<&'a Complex<T>>, D>where
D: Dimension,
impl<'a, T, D> ArrayBase<ViewRepr<&'a Complex<T>>, D>where
D: Dimension,
sourcepub fn split_complex(self) -> Complex<ArrayView<'a, T, D>>
pub fn split_complex(self) -> Complex<ArrayView<'a, T, D>>
Splits the view into views of the real and imaginary components of the elements.
use ndarray::prelude::*;
use num_complex::{Complex, Complex64};
let arr = array![
[Complex64::new(1., 2.), Complex64::new(3., 4.)],
[Complex64::new(5., 6.), Complex64::new(7., 8.)],
[Complex64::new(9., 10.), Complex64::new(11., 12.)],
];
let Complex { re, im } = arr.view().split_complex();
assert_eq!(re, array![[1., 3.], [5., 7.], [9., 11.]]);
assert_eq!(im, array![[2., 4.], [6., 8.], [10., 12.]]);
source§impl<'a, A, D> ArrayBase<ViewRepr<&'a mut A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<ViewRepr<&'a mut A>, D>where
D: Dimension,
Methods for read-write array views.
sourcepub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
pub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
Split the array view along axis
and return one mutable view strictly
before the split and one mutable view after the split.
Panics if axis
or index
is out of bounds.
sourcepub fn multi_slice_move<M>(self, info: M) -> M::Outputwhere
M: MultiSliceArg<'a, A, D>,
pub fn multi_slice_move<M>(self, info: M) -> M::Outputwhere
M: MultiSliceArg<'a, A, D>,
Split the view into multiple disjoint slices.
This is similar to .multi_slice_mut()
, but .multi_slice_move()
consumes self
and produces views with lifetimes matching that of
self
.
See Slicing for full documentation. See also
MultiSliceArg
, s!
, SliceArg
, and
SliceInfo
.
Panics if any of the following occur:
- if any of the views would intersect (i.e. if any element would appear in multiple slices)
- if an index is out of bounds or step size is zero
- if
D
isIxDyn
andinfo
does not match the number of array axes
source§impl<'a, T, D> ArrayBase<ViewRepr<&'a mut Complex<T>>, D>where
D: Dimension,
impl<'a, T, D> ArrayBase<ViewRepr<&'a mut Complex<T>>, D>where
D: Dimension,
sourcepub fn split_complex(self) -> Complex<ArrayViewMut<'a, T, D>>
pub fn split_complex(self) -> Complex<ArrayViewMut<'a, T, D>>
Splits the view into views of the real and imaginary components of the elements.
use ndarray::prelude::*;
use num_complex::{Complex, Complex64};
let mut arr = array![
[Complex64::new(1., 2.), Complex64::new(3., 4.)],
[Complex64::new(5., 6.), Complex64::new(7., 8.)],
[Complex64::new(9., 10.), Complex64::new(11., 12.)],
];
let Complex { mut re, mut im } = arr.view_mut().split_complex();
assert_eq!(re, array![[1., 3.], [5., 7.], [9., 11.]]);
assert_eq!(im, array![[2., 4.], [6., 8.], [10., 12.]]);
re[[0, 1]] = 13.;
im[[2, 0]] = 14.;
assert_eq!(arr[[0, 1]], Complex64::new(13., 4.));
assert_eq!(arr[[2, 0]], Complex64::new(9., 14.));
source§impl<A, D> ArrayBase<RawViewRepr<*const A>, D>where
D: Dimension,
impl<A, D> ArrayBase<RawViewRepr<*const A>, D>where
D: Dimension,
sourcepub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *const A) -> Selfwhere
Sh: Into<StrideShape<D>>,
pub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *const A) -> Selfwhere
Sh: Into<StrideShape<D>>,
Create an RawArrayView<A, D>
from shape information and a raw pointer
to the elements.
§Safety
The caller is responsible for ensuring all of the following:
-
ptr
must be non-null, and it must be safe to.offset()
ptr
by zero. -
It must be safe to
.offset()
the pointer repeatedly along all axes and calculate thecount
s for the.offset()
calls without overflow, even if the array is empty or the elements are zero-sized.In other words,
-
All possible pointers generated by moving along all axes must be in bounds or one byte past the end of a single allocation with element type
A
. The only exceptions are if the array is empty or the element type is zero-sized. In these cases,ptr
may be dangling, but it must still be safe to.offset()
the pointer along the axes. -
The offset in units of bytes between the least address and greatest address by moving along all axes must not exceed
isize::MAX
. This constraint prevents the computed offset, in bytes, from overflowingisize
regardless of the starting point due to past offsets. -
The offset in units of
A
between the least address and greatest address by moving along all axes must not exceedisize::MAX
. This constraint prevents overflow when calculating thecount
parameter to.offset()
regardless of the starting point due to past offsets.
-
-
The product of non-zero axis lengths must not exceed
isize::MAX
. -
Strides must be non-negative.
This function can use debug assertions to check some of these requirements, but it’s not a complete check.
sourcepub unsafe fn deref_into_view<'a>(self) -> ArrayView<'a, A, D>
pub unsafe fn deref_into_view<'a>(self) -> ArrayView<'a, A, D>
Converts to a read-only view of the array.
§Safety
From a safety standpoint, this is equivalent to dereferencing a raw pointer for every element in the array. You must ensure that all of the data is valid, ensure that the pointer is aligned, and choose the correct lifetime.
sourcepub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
pub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
Split the array view along axis
and return one array pointer strictly
before the split and one array pointer after the split.
Panics if axis
or index
is out of bounds.
sourcepub fn cast<B>(self) -> RawArrayView<B, D>
pub fn cast<B>(self) -> RawArrayView<B, D>
Cast the raw pointer of the raw array view to a different type
Panics if element size is not compatible.
Lack of panic does not imply it is a valid cast. The cast works the same way as regular raw pointer casts.
While this method is safe, for the same reason as regular raw pointer casts are safe, access through the produced raw view is only possible in an unsafe block or function.
source§impl<T, D> ArrayBase<RawViewRepr<*const Complex<T>>, D>where
D: Dimension,
impl<T, D> ArrayBase<RawViewRepr<*const Complex<T>>, D>where
D: Dimension,
sourcepub fn split_complex(self) -> Complex<RawArrayView<T, D>>
pub fn split_complex(self) -> Complex<RawArrayView<T, D>>
Splits the view into views of the real and imaginary components of the elements.
source§impl<A, D> ArrayBase<RawViewRepr<*mut A>, D>where
D: Dimension,
impl<A, D> ArrayBase<RawViewRepr<*mut A>, D>where
D: Dimension,
sourcepub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *mut A) -> Selfwhere
Sh: Into<StrideShape<D>>,
pub unsafe fn from_shape_ptr<Sh>(shape: Sh, ptr: *mut A) -> Selfwhere
Sh: Into<StrideShape<D>>,
Create an RawArrayViewMut<A, D>
from shape information and a raw
pointer to the elements.
§Safety
The caller is responsible for ensuring all of the following:
-
ptr
must be non-null, and it must be safe to.offset()
ptr
by zero. -
It must be safe to
.offset()
the pointer repeatedly along all axes and calculate thecount
s for the.offset()
calls without overflow, even if the array is empty or the elements are zero-sized.In other words,
-
All possible pointers generated by moving along all axes must be in bounds or one byte past the end of a single allocation with element type
A
. The only exceptions are if the array is empty or the element type is zero-sized. In these cases,ptr
may be dangling, but it must still be safe to.offset()
the pointer along the axes. -
The offset in units of bytes between the least address and greatest address by moving along all axes must not exceed
isize::MAX
. This constraint prevents the computed offset, in bytes, from overflowingisize
regardless of the starting point due to past offsets. -
The offset in units of
A
between the least address and greatest address by moving along all axes must not exceedisize::MAX
. This constraint prevents overflow when calculating thecount
parameter to.offset()
regardless of the starting point due to past offsets.
-
-
The product of non-zero axis lengths must not exceed
isize::MAX
. -
Strides must be non-negative.
This function can use debug assertions to check some of these requirements, but it’s not a complete check.
sourcepub unsafe fn deref_into_view<'a>(self) -> ArrayView<'a, A, D>
pub unsafe fn deref_into_view<'a>(self) -> ArrayView<'a, A, D>
Converts to a read-only view of the array.
§Safety
From a safety standpoint, this is equivalent to dereferencing a raw pointer for every element in the array. You must ensure that all of the data is valid, ensure that the pointer is aligned, and choose the correct lifetime.
sourcepub unsafe fn deref_into_view_mut<'a>(self) -> ArrayViewMut<'a, A, D>
pub unsafe fn deref_into_view_mut<'a>(self) -> ArrayViewMut<'a, A, D>
Converts to a mutable view of the array.
§Safety
From a safety standpoint, this is equivalent to dereferencing a raw pointer for every element in the array. You must ensure that all of the data is valid, ensure that the pointer is aligned, and choose the correct lifetime.
sourcepub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
pub fn split_at(self, axis: Axis, index: Ix) -> (Self, Self)
Split the array view along axis
and return one array pointer strictly
before the split and one array pointer after the split.
Panics if axis
or index
is out of bounds.
sourcepub fn cast<B>(self) -> RawArrayViewMut<B, D>
pub fn cast<B>(self) -> RawArrayViewMut<B, D>
Cast the raw pointer of the raw array view to a different type
Panics if element size is not compatible.
Lack of panic does not imply it is a valid cast. The cast works the same way as regular raw pointer casts.
While this method is safe, for the same reason as regular raw pointer casts are safe, access through the produced raw view is only possible in an unsafe block or function.
source§impl<T, D> ArrayBase<RawViewRepr<*mut Complex<T>>, D>where
D: Dimension,
impl<T, D> ArrayBase<RawViewRepr<*mut Complex<T>>, D>where
D: Dimension,
sourcepub fn split_complex(self) -> Complex<RawArrayViewMut<T, D>>
pub fn split_complex(self) -> Complex<RawArrayViewMut<T, D>>
Splits the view into views of the real and imaginary components of the elements.
source§impl<'a, A, D> ArrayBase<CowRepr<'a, A>, D>where
D: Dimension,
impl<'a, A, D> ArrayBase<CowRepr<'a, A>, D>where
D: Dimension,
Methods specific to CowArray
.
See also all methods for ArrayBase
source§impl<A, D> ArrayBase<OwnedArcRepr<A>, D>where
D: Dimension,
impl<A, D> ArrayBase<OwnedArcRepr<A>, D>where
D: Dimension,
Methods specific to ArcArray
.
See also all methods for ArrayBase
source§impl<S, A, D> ArrayBase<S, D>
impl<S, A, D> ArrayBase<S, D>
sourcepub fn triu(&self, k: isize) -> Array<A, D>
pub fn triu(&self, k: isize) -> Array<A, D>
Upper triangular of an array.
Return a copy of the array with elements below the k-th diagonal zeroed.
For arrays with ndim
exceeding 2, triu
will apply to the final two axes.
For 0D and 1D arrays, triu
will return an unchanged clone.
See also ArrayBase::tril
use ndarray::array;
let arr = array![
[1, 2, 3],
[4, 5, 6],
[7, 8, 9]
];
assert_eq!(
arr.triu(0),
array![
[1, 2, 3],
[0, 5, 6],
[0, 0, 9]
]
);
sourcepub fn tril(&self, k: isize) -> Array<A, D>
pub fn tril(&self, k: isize) -> Array<A, D>
Lower triangular of an array.
Return a copy of the array with elements above the k-th diagonal zeroed.
For arrays with ndim
exceeding 2, tril
will apply to the final two axes.
For 0D and 1D arrays, tril
will return an unchanged clone.
See also ArrayBase::triu
use ndarray::array;
let arr = array![
[1, 2, 3],
[4, 5, 6],
[7, 8, 9]
];
assert_eq!(
arr.tril(0),
array![
[1, 0, 0],
[4, 5, 0],
[7, 8, 9]
]
);
Trait Implementations§
source§impl<A, B, S, S2, D> AbsDiffEq<ArrayBase<S2, D>> for ArrayBase<S, D>
impl<A, B, S, S2, D> AbsDiffEq<ArrayBase<S2, D>> for ArrayBase<S, D>
Requires crate feature "approx"
.
source§fn default_epsilon() -> A::Epsilon
fn default_epsilon() -> A::Epsilon
source§fn abs_diff_eq(&self, other: &ArrayBase<S2, D>, epsilon: A::Epsilon) -> bool
fn abs_diff_eq(&self, other: &ArrayBase<S2, D>, epsilon: A::Epsilon) -> bool
source§fn abs_diff_ne(&self, other: &Rhs, epsilon: Self::Epsilon) -> bool
fn abs_diff_ne(&self, other: &Rhs, epsilon: Self::Epsilon) -> bool
AbsDiffEq::abs_diff_eq
.source§impl<'a, A, B, S, S2, D, E> Add<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> Add<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
addition
between references self
and rhs
,
and return the result as a new Array
.
If their shapes disagree, self
and rhs
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> Add<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> Add<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
addition
between self
and reference rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> Add<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> Add<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
addition
between reference self
and rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<A, B, S, S2, D, E> Add<ArrayBase<S2, E>> for ArrayBase<S, D>
impl<A, B, S, S2, D, E> Add<ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
addition
between self
and rhs
,
and return the result.
self
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape.
Panics if broadcasting isn’t possible.
source§impl<'a, A, S, D, B> Add<B> for &'a ArrayBase<S, D>
impl<'a, A, S, D, B> Add<B> for &'a ArrayBase<S, D>
Perform elementwise
addition
between the reference self
and the scalar x
,
and return the result as a new Array
.
source§impl<A, S, D, B> Add<B> for ArrayBase<S, D>
impl<A, S, D, B> Add<B> for ArrayBase<S, D>
Perform elementwise
addition
between self
and the scalar x
,
and return the result (based on self
).
self
must be an Array
or ArcArray
.
source§impl<'a, A, S, S2, D, E> AddAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, S, S2, D, E> AddAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform self += rhs
as elementwise addition (in place).
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
source§fn add_assign(&mut self, rhs: &ArrayBase<S2, E>)
fn add_assign(&mut self, rhs: &ArrayBase<S2, E>)
+=
operation. Read moresource§impl<A, S, D> AddAssign<A> for ArrayBase<S, D>
impl<A, S, D> AddAssign<A> for ArrayBase<S, D>
Perform self += rhs
as elementwise addition (in place).
source§fn add_assign(&mut self, rhs: A)
fn add_assign(&mut self, rhs: A)
+=
operation. Read moresource§impl<A: Binary, S, D: Dimension> Binary for ArrayBase<S, D>where
S: Data<Elem = A>,
impl<A: Binary, S, D: Dimension> Binary for ArrayBase<S, D>where
S: Data<Elem = A>,
Format the array using Binary
and apply the formatting parameters used
to each element.
The array is shown in multiline style.
source§impl<'a, A, B, S, S2, D, E> BitAnd<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitAnd<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
bit and
between references self
and rhs
,
and return the result as a new Array
.
If their shapes disagree, self
and rhs
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> BitAnd<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitAnd<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
bit and
between self
and reference rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> BitAnd<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitAnd<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
bit and
between reference self
and rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<A, B, S, S2, D, E> BitAnd<ArrayBase<S2, E>> for ArrayBase<S, D>
impl<A, B, S, S2, D, E> BitAnd<ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
bit and
between self
and rhs
,
and return the result.
self
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape.
Panics if broadcasting isn’t possible.
source§impl<'a, A, S, D, B> BitAnd<B> for &'a ArrayBase<S, D>
impl<'a, A, S, D, B> BitAnd<B> for &'a ArrayBase<S, D>
Perform elementwise
bit and
between the reference self
and the scalar x
,
and return the result as a new Array
.
source§impl<A, S, D, B> BitAnd<B> for ArrayBase<S, D>
impl<A, S, D, B> BitAnd<B> for ArrayBase<S, D>
Perform elementwise
bit and
between self
and the scalar x
,
and return the result (based on self
).
self
must be an Array
or ArcArray
.
source§impl<'a, A, S, S2, D, E> BitAndAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, S, S2, D, E> BitAndAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform self &= rhs
as elementwise bit and (in place).
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
source§fn bitand_assign(&mut self, rhs: &ArrayBase<S2, E>)
fn bitand_assign(&mut self, rhs: &ArrayBase<S2, E>)
&=
operation. Read moresource§impl<A, S, D> BitAndAssign<A> for ArrayBase<S, D>
impl<A, S, D> BitAndAssign<A> for ArrayBase<S, D>
Perform self &= rhs
as elementwise bit and (in place).
source§fn bitand_assign(&mut self, rhs: A)
fn bitand_assign(&mut self, rhs: A)
&=
operation. Read moresource§impl<'a, A, B, S, S2, D, E> BitOr<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitOr<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
bit or
between references self
and rhs
,
and return the result as a new Array
.
If their shapes disagree, self
and rhs
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> BitOr<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitOr<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
bit or
between self
and reference rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> BitOr<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitOr<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
bit or
between reference self
and rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<A, B, S, S2, D, E> BitOr<ArrayBase<S2, E>> for ArrayBase<S, D>
impl<A, B, S, S2, D, E> BitOr<ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
bit or
between self
and rhs
,
and return the result.
self
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape.
Panics if broadcasting isn’t possible.
source§impl<'a, A, S, D, B> BitOr<B> for &'a ArrayBase<S, D>
impl<'a, A, S, D, B> BitOr<B> for &'a ArrayBase<S, D>
Perform elementwise
bit or
between the reference self
and the scalar x
,
and return the result as a new Array
.
source§impl<A, S, D, B> BitOr<B> for ArrayBase<S, D>
impl<A, S, D, B> BitOr<B> for ArrayBase<S, D>
Perform elementwise
bit or
between self
and the scalar x
,
and return the result (based on self
).
self
must be an Array
or ArcArray
.
source§impl<'a, A, S, S2, D, E> BitOrAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, S, S2, D, E> BitOrAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform self |= rhs
as elementwise bit or (in place).
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
source§fn bitor_assign(&mut self, rhs: &ArrayBase<S2, E>)
fn bitor_assign(&mut self, rhs: &ArrayBase<S2, E>)
|=
operation. Read moresource§impl<A, S, D> BitOrAssign<A> for ArrayBase<S, D>
impl<A, S, D> BitOrAssign<A> for ArrayBase<S, D>
Perform self |= rhs
as elementwise bit or (in place).
source§fn bitor_assign(&mut self, rhs: A)
fn bitor_assign(&mut self, rhs: A)
|=
operation. Read moresource§impl<'a, A, B, S, S2, D, E> BitXor<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitXor<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
bit xor
between references self
and rhs
,
and return the result as a new Array
.
If their shapes disagree, self
and rhs
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> BitXor<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitXor<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
bit xor
between self
and reference rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> BitXor<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> BitXor<ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
bit xor
between reference self
and rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<A, B, S, S2, D, E> BitXor<ArrayBase<S2, E>> for ArrayBase<S, D>
impl<A, B, S, S2, D, E> BitXor<ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
bit xor
between self
and rhs
,
and return the result.
self
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape.
Panics if broadcasting isn’t possible.
source§impl<'a, A, S, D, B> BitXor<B> for &'a ArrayBase<S, D>
impl<'a, A, S, D, B> BitXor<B> for &'a ArrayBase<S, D>
Perform elementwise
bit xor
between the reference self
and the scalar x
,
and return the result as a new Array
.
source§impl<A, S, D, B> BitXor<B> for ArrayBase<S, D>
impl<A, S, D, B> BitXor<B> for ArrayBase<S, D>
Perform elementwise
bit xor
between self
and the scalar x
,
and return the result (based on self
).
self
must be an Array
or ArcArray
.
source§impl<'a, A, S, S2, D, E> BitXorAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, S, S2, D, E> BitXorAssign<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform self ^= rhs
as elementwise bit xor (in place).
If their shapes disagree, rhs
is broadcast to the shape of self
.
Panics if broadcasting isn’t possible.
source§fn bitxor_assign(&mut self, rhs: &ArrayBase<S2, E>)
fn bitxor_assign(&mut self, rhs: &ArrayBase<S2, E>)
^=
operation. Read moresource§impl<A, S, D> BitXorAssign<A> for ArrayBase<S, D>
impl<A, S, D> BitXorAssign<A> for ArrayBase<S, D>
Perform self ^= rhs
as elementwise bit xor (in place).
source§fn bitxor_assign(&mut self, rhs: A)
fn bitxor_assign(&mut self, rhs: A)
^=
operation. Read moresource§impl<A: Debug, S, D: Dimension> Debug for ArrayBase<S, D>where
S: Data<Elem = A>,
impl<A: Debug, S, D: Dimension> Debug for ArrayBase<S, D>where
S: Data<Elem = A>,
Format the array using Debug
and apply the formatting parameters used
to each element.
The array is shown in multiline style.
source§impl<A, S, D> Default for ArrayBase<S, D>
impl<A, S, D> Default for ArrayBase<S, D>
Create an owned array with a default state.
The array is created with dimension D::default()
, which results
in for example dimensions 0
and (0, 0)
with zero elements for the
one-dimensional and two-dimensional cases respectively.
The default dimension for IxDyn
is IxDyn(&[0])
(array has zero
elements). And the default for the dimension ()
is ()
(array has
one element).
Since arrays cannot grow, the intention is to use the default value as placeholder.
source§impl<'de, A, Di, S> Deserialize<'de> for ArrayBase<S, Di>
impl<'de, A, Di, S> Deserialize<'de> for ArrayBase<S, Di>
Requires crate feature "serde"
source§fn deserialize<D>(deserializer: D) -> Result<ArrayBase<S, Di>, D::Error>where
D: Deserializer<'de>,
fn deserialize<D>(deserializer: D) -> Result<ArrayBase<S, Di>, D::Error>where
D: Deserializer<'de>,
source§impl<A: Display, S, D: Dimension> Display for ArrayBase<S, D>where
S: Data<Elem = A>,
impl<A: Display, S, D: Dimension> Display for ArrayBase<S, D>where
S: Data<Elem = A>,
Format the array using Display
and apply the formatting parameters used
to each element.
The array is shown in multiline style.
source§impl<'a, A, B, S, S2, D, E> Div<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> Div<&'a ArrayBase<S2, E>> for &'a ArrayBase<S, D>
Perform elementwise
division
between references self
and rhs
,
and return the result as a new Array
.
If their shapes disagree, self
and rhs
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.
source§impl<'a, A, B, S, S2, D, E> Div<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
impl<'a, A, B, S, S2, D, E> Div<&'a ArrayBase<S2, E>> for ArrayBase<S, D>
Perform elementwise
division
between self
and reference rhs
,
and return the result.
rhs
must be an Array
or ArcArray
.
If their shapes disagree, self
is broadcast to their broadcast shape,
cloning the data if needed.
Panics if broadcasting isn’t possible.