DimensionalData.jl

Appearance

API Reference ​

Arrays ​

DimensionalData.AbstractBasicDimArray Type
julia
AbstractBasicDimArray <: AbstractArray

The abstract supertype for all arrays with a dims method that returns a Tuple of Dimension

Only keyword rebuild is guaranteed to work with AbstractBasicDimArray.

source

DimensionalData.AbstractDimArray Type
julia
AbstractDimArray <: AbstractBasicArray

Abstract supertype for all "dim" arrays.

These arrays return a Tuple of Dimension from a dims method, and can be rebuilt using rebuild.

parent must return the source array.

They should have metadata, name and refdims methods, although these are optional.

A rebuild method for AbstractDimArray must accept data, dims, refdims, name, metadata arguments.

Indexing AbstractDimArray with non-range AbstractArray has undefined effects on the Dimension index. Use forward-ordered arrays only"

source

DimensionalData.DimArray Type
julia
DimArray <: AbstractDimArray

DimArray(data, dims, refdims, name, metadata)
DimArray(data, dims::Tuple; refdims=(), name=NoName(), metadata=NoMetadata())
DimArray(gen; kw...)

The main concrete subtype of AbstractDimArray.

DimArray maintains and updates its Dimensions through transformations and moves dimensions to reference dimension refdims after reducing operations (like e.g. mean).

Arguments

  • data: An AbstractArray.

  • gen: A generator expression. Where source iterators are Dimensions the dim args or kw is not needed.

  • dims: A Tuple of Dimension

  • name: A string name for the array. Shows in plots and tables.

  • refdims: refence dimensions. Usually set programmatically to track past slices and reductions of dimension for labelling and reconstruction.

  • metadata: Dict or Metadata object, or NoMetadata()

Indexing can be done with all regular indices, or with Dimensions and/or Selectors.

Indexing AbstractDimArray with non-range AbstractArray has undefined effects on the Dimension index. Use forward-ordered arrays only"

Note that the generator expression syntax requires usage of the semi-colon ; to distinguish generator dimensions from keywords.

Example:

julia
julia> using Dates, DimensionalData

julia> ti = Ti(DateTime(2001):Month(1):DateTime(2001,12));

julia> x = X(10:10:100);

julia> A = DimArray(rand(12,10), (ti, x), name="example");

julia> A[X(Near([12, 35])), Ti(At(DateTime(2001,5)))]
┌ 2-element DimArray{Float64, 1} example ┐
├────────────────────────────────────────┴────────────── dims ┐
  ↓ X Sampled{Int64} [10, 40] ForwardOrdered Irregular Points
└─────────────────────────────────────────────────────────────┘
 10  0.253849
 40  0.637077

julia> A[Near(DateTime(2001, 5, 4)), Between(20, 50)]
┌ 4-element DimArray{Float64, 1} example ┐
├────────────────────────────────────────┴──────────── dims ┐
  ↓ X Sampled{Int64} 20:10:50 ForwardOrdered Regular Points
└───────────────────────────────────────────────────────────┘
 20  0.774092
 30  0.823656
 40  0.637077
 50  0.692235

Generator expression:

julia
julia> DimArray((x, y) for x in X(1:3), y in Y(1:2); name = :Value)
┌ 3×2 DimArray{Tuple{Int64, Int64}, 2} Value ┐
├────────────────────────────────────────────┴──── dims ┐
  ↓ X Sampled{Int64} 1:3 ForwardOrdered Regular Points,
  → Y Sampled{Int64} 1:2 ForwardOrdered Regular Points
└───────────────────────────────────────────────────────┘
 ↓ →  1        2
 1     (1, 1)   (1, 2)
 2     (2, 1)   (2, 2)
 3     (3, 1)   (3, 2)

source

Shorthand AbstractDimArray constructors:

Base.fill Function
julia
Base.fill(x, dims::Dimension...; kw...) => DimArray
Base.fill(x, dims::Tuple{Vararg{Dimension}}; kw...) => DimArray

Create a DimArray with a fill value of x.

There are two kinds of Dimension value acepted:

  • A Dimension holding an AbstractVector will set the dimension index to that AbstractVector, and detect the dimension lookup.

  • A Dimension holding an Integer will set the length of the axis, and set the dimension lookup to NoLookup.

Keywords are the same as for DimArray.

Example

julia
julia> using DimensionalData, Random; Random.seed!(123);

julia> fill(true, X(2), Y(4))
┌ 2×4 DimArray{Bool, 2} ┐
├───────────────── dims ┤
  ↓ X, → Y
└───────────────────────┘
 1  1  1  1
 1  1  1  1

source

Base.rand Function
julia
Base.rand(x, dims::Dimension...; kw...) => DimArray
Base.rand(x, dims::Tuple{Vararg{Dimension}}; kw...) => DimArray
Base.rand(r::AbstractRNG, x, dims::Tuple{Vararg{Dimension}}; kw...) => DimArray
Base.rand(r::AbstractRNG, x, dims::Dimension...; kw...) => DimArray

Create a DimArray of random values.

There are two kinds of Dimension value acepted:

  • A Dimension holding an AbstractVector will set the dimension index to that AbstractVector, and detect the dimension lookup.

  • A Dimension holding an Integer will set the length of the axis, and set the dimension lookup to NoLookup.

Keywords are the same as for DimArray.

Example

julia
julia> using DimensionalData

julia> rand(Bool, X(2), Y(4))
┌ 2×4 DimArray{Bool, 2} ┐
├───────────────── dims ┤
  ↓ X, → Y
└───────────────────────┘
 0  0  0  0
 1  0  0  1

julia> rand(X([:a, :b, :c]), Y(100.0:50:200.0))
┌ 3×3 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────── dims ┐
  ↓ X Categorical{Symbol} [:a, :b, :c] ForwardOrdered,
  → Y Sampled{Float64} 100.0:50.0:200.0 ForwardOrdered Regular Points
└─────────────────────────────────────────────────────────────────────┘
 ↓ →  100.0       150.0       200.0
  :a    0.443494    0.253849    0.867547
  :b    0.745673    0.334152    0.0802658
  :c    0.512083    0.427328    0.311448

source

Base.zeros Function
julia
Base.zeros(x, dims::Dimension...; kw...) => DimArray
Base.zeros(x, dims::Tuple{Vararg{Dimension}}; kw...) => DimArray

Create a DimArray of zeros.

There are two kinds of Dimension value acepted:

  • A Dimension holding an AbstractVector will set the dimension index to that AbstractVector, and detect the dimension lookup.

  • A Dimension holding an Integer will set the length of the axis, and set the dimension lookup to NoLookup.

Keywords are the same as for DimArray.

Example

julia
julia> using DimensionalData

julia> zeros(Bool, X(2), Y(4))
┌ 2×4 DimArray{Bool, 2} ┐
├───────────────── dims ┤
  ↓ X, → Y
└───────────────────────┘
 0  0  0  0
 0  0  0  0

julia> zeros(X([:a, :b, :c]), Y(100.0:50:200.0))
┌ 3×3 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────── dims ┐
  ↓ X Categorical{Symbol} [:a, :b, :c] ForwardOrdered,
  → Y Sampled{Float64} 100.0:50.0:200.0 ForwardOrdered Regular Points
└─────────────────────────────────────────────────────────────────────┘
 ↓ →  100.0  150.0  200.0
  :a    0.0    0.0    0.0
  :b    0.0    0.0    0.0
  :c    0.0    0.0    0.0

source

Base.ones Function
julia
Base.ones(x, dims::Dimension...; kw...) => DimArray
Base.ones(x, dims::Tuple{Vararg{Dimension}}; kw...) => DimArray

Create a DimArray of ones.

There are two kinds of Dimension value acepted:

  • A Dimension holding an AbstractVector will set the dimension index to that AbstractVector, and detect the dimension lookup.

  • A Dimension holding an Integer will set the length of the axis, and set the dimension lookup to NoLookup.

Keywords are the same as for DimArray.

Example

julia
julia> using DimensionalData

julia> ones(Bool, X(2), Y(4))
┌ 2×4 DimArray{Bool, 2} ┐
├───────────────── dims ┤
  ↓ X, → Y
└───────────────────────┘
 1  1  1  1
 1  1  1  1

julia> ones(X([:a, :b, :c]), Y(100.0:50:200.0))
┌ 3×3 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────── dims ┐
  ↓ X Categorical{Symbol} [:a, :b, :c] ForwardOrdered,
  → Y Sampled{Float64} 100.0:50.0:200.0 ForwardOrdered Regular Points
└─────────────────────────────────────────────────────────────────────┘
 ↓ →  100.0  150.0  200.0
  :a    1.0    1.0    1.0
  :b    1.0    1.0    1.0
  :c    1.0    1.0    1.0

source

Functions for getting information from objects:

DimensionalData.Dimensions.dims Function
julia
dims(x, [dims::Tuple]) => Tuple{Vararg{Dimension}}
dims(x, dim) => Dimension

Return a tuple of Dimensions for an object, in the order that matches the axes or columns of the underlying data.

dims can be Dimension, Dimension types, or Symbols for Dim{Symbol}.

The default is to return nothing.

source

julia
dims(x, query) => Tuple{Vararg{Dimension}}
dims(x, query...) => Tuple{Vararg{Dimension}}

Get the dimension(s) matching the type(s) of the query dimension.

Lookup can be an Int or an Dimension, or a tuple containing any combination of either.

Arguments

  • x: any object with a dims method, or a Tuple of Dimension.

  • query: Tuple or a single Dimension or Dimension Type.

Example

julia
julia> using DimensionalData

julia> A = DimArray(ones(2, 3, 2), (X, Y, Z))
┌ 2×3×2 DimArray{Float64, 3} ┐
├────────────────────── dims ┤
  ↓ X, → Y, ↗ Z
└────────────────────────────┘
[:, :, 1]
 1.0  1.0  1.0
 1.0  1.0  1.0

julia> dims(A, (X, Y))
(↓ X, → Y)

source

DimensionalData.Dimensions.refdims Function
julia
refdims(x, [dims::Tuple]) => Tuple{Vararg{Dimension}}
refdims(x, dim) => Dimension

Reference dimensions for an array that is a slice or view of another array with more dimensions.

slicedims(a, dims) returns a tuple containing the current new dimensions and the new reference dimensions. Refdims can be stored in a field or discarded, as it is mostly to give context to plots. Ignoring refdims will simply leave some captions empty.

The default is to return an empty Tuple ().

source

DimensionalData.Dimensions.Lookups.metadata Function
julia
metadata(x) => (object metadata)
metadata(x, dims::Tuple)  => Tuple (Dimension metadata)
metadata(xs::Tuple) => Tuple

Returns the metadata for an object or for the specified dimension(s)

Second argument dims can be Dimensions, Dimension types, or Symbols for Dim{Symbol}.

source

DimensionalData.Dimensions.name Function
julia
name(x) => Symbol
name(xs:Tuple) => NTuple{N,Symbol}
name(x, dims::Tuple) => NTuple{N,Symbol}
name(x, dim) => Symbol

Get the name of an array or Dimension, or a tuple of of either as a Symbol.

Second argument dims can be Dimensions, Dimension types, or Symbols for Dim{Symbol}.

source

DimensionalData.Dimensions.otherdims Function
julia
otherdims(x, query) => Tuple{Vararg{Dimension,N}}

Get the dimensions of an object not in query.

Arguments

  • x: any object with a dims method, a Tuple of Dimension.

  • query: Tuple or single Dimension or dimension Type.

  • f: <: by default, but can be >: to match abstract types to concrete types.

A tuple holding the unmatched dimensions is always returned.

Example

julia
julia> using DimensionalData, DimensionalData.Dimensions

julia> A = DimArray(ones(10, 10, 10), (X, Y, Z));

julia> otherdims(A, X)
(↓ Y, → Z)

julia> otherdims(A, (Y, Z))
(↓ X)

source

DimensionalData.Dimensions.dimnum Function
julia
dimnum(x, query::Tuple) => NTuple{Int}
dimnum(x, query) => Int

Get the number(s) of Dimension(s) as ordered in the dimensions of an object.

Arguments

  • x: any object with a dims method, a Tuple of Dimension or a single Dimension.

  • query: Tuple, Array or single Dimension or dimension Type.

The return type will be a Tuple of Int or a single Int, depending on whether query is a Tuple or single Dimension.

Example

julia
julia> using DimensionalData

julia> A = DimArray(ones(10, 10, 10), (X, Y, Z));

julia> dimnum(A, (Z, X, Y))
(3, 1, 2)

julia> dimnum(A, Y)
2

source

DimensionalData.Dimensions.hasdim Function
julia
hasdim([f], x, query::Tuple) => NTuple{Bool}
hasdim([f], x, query...) => NTuple{Bool}
hasdim([f], x, query) => Bool

Check if an object x has dimensions that match or inherit from the query dimensions.

Arguments

  • x: any object with a dims method, a Tuple of Dimension or a single Dimension.

  • query: Tuple or single Dimension or dimension Type.

  • f: <: by default, but can be >: to match abstract types to concrete types.

Check if an object or tuple contains an Dimension, or a tuple of dimensions.

Example

julia
julia> using DimensionalData

julia> A = DimArray(ones(10, 10, 10), (X, Y, Z));

julia> hasdim(A, X)
true

julia> hasdim(A, (Z, X, Y))
(true, true, true)

julia> hasdim(A, Ti)
false

source

Multi-array datasets ​

DimensionalData.AbstractDimStack Type
julia
AbstractDimStack

Abstract supertype for dimensional stacks.

These have multiple layers of data, but share dimensions.

Notably, their behaviour lies somewhere between a DimArray and a NamedTuple:

  • indexing with a Symbol as in dimstack[:symbol] returns a DimArray layer.

  • iteration and map apply over array layers, as indexed with a Symbol.

  • getindex and many base methods are applied as for DimArray - to avoid the need to always use map.

This design gives very succinct code when working with many-layered, mixed-dimension objects. But it may be jarring initially - the most surprising outcome is that dimstack[1] will return a NamedTuple of values for the first index in all layers, while first(dimstack) will return the first value of the iterator - the DimArray for the first layer.

See DimStack for the concrete implementation. Most methods are defined on the abstract type.

To extend AbstractDimStack, implement argument and keyword version of rebuild and also rebuild_from_arrays.

The constructor of an AbstractDimStack must accept a NamedTuple.

source

DimensionalData.DimStack Type
julia
DimStack <: AbstractDimStack

DimStack(data::AbstractDimArray...; kw...)
DimStack(data::Tuple{Vararg{AbstractDimArray}}; kw...)
DimStack(data::NamedTuple{Keys,Vararg{AbstractDimArray}}; kw...)
DimStack(data::NamedTuple, dims::DimTuple; metadata=NoMetadata(); kw...)

DimStack holds multiple objects sharing some dimensions, in a NamedTuple.

Notably, their behaviour lies somewhere between a DimArray and a NamedTuple:

  • indexing with a Symbol as in dimstack[:symbol] returns a DimArray layer.

  • iteration and map apply over array layers, as indexed with a Symbol.

  • getindex or view with Int, Dimensions or Selectors that resolve to Int will return a NamedTuple of values from each layer in the stack. This has very good performance, and avoids the need to always use map.

  • getindex or view with a Vector or Colon will return another DimStack where all data layers have been sliced.

  • setindex! must pass a Tuple or NamedTuple matching the layers.

  • many base and Statistics methods (sum, mean etc) will work as for a DimArray again removing the need to use map.

julia
function DimStack(A::AbstractDimArray;
    layersfrom=nothing, name=nothing, metadata=metadata(A), refdims=refdims(A), kw...
)

For example, here we take the mean over the time dimension for all layers:

julia
mean(mydimstack; dims=Ti)

And this equivalent to:

julia
map(A -> mean(A; dims=Ti), mydimstack)

This design gives succinct code when working with many-layered, mixed-dimension objects.

But it may be jarring initially - the most surprising outcome is that dimstack[1] will return a NamedTuple of values for the first index in all layers, while first(dimstack) will return the first value of the iterator - the DimArray for the first layer.

DimStack can be constructed from multiple AbstractDimArray or a NamedTuple of AbstractArray and a matching dims tuple.

Most Base and Statistics methods that apply to AbstractArray can be used on all layers of the stack simulataneously. The result is a DimStack, or a NamedTuple if methods like mean are used without dims arguments, and return a single non-array value.

Example

julia
julia> using DimensionalData

julia> A = [1.0 2.0 3.0; 4.0 5.0 6.0];

julia> dimz = (X([:a, :b]), Y(10.0:10.0:30.0))
(↓ X [:a, :b],
→ Y 10.0:10.0:30.0)

julia> da1 = DimArray(1A, dimz; name=:one);

julia> da2 = DimArray(2A, dimz; name=:two);

julia> da3 = DimArray(3A, dimz; name=:three);

julia> s = DimStack(da1, da2, da3);

julia> s[At(:b), At(10.0)]
(one = 4.0, two = 8.0, three = 12.0)

julia> s[X(At(:a))] isa DimStack
true

source

Dimension generators ​

DimensionalData.DimIndices Type
julia
DimIndices <: AbstractArray

DimIndices(x)
DimIndices(dims::Tuple)
DimIndices(dims::Dimension)

Like CartesianIndices, but for Dimensions. Behaves as an Array of Tuple of Dimension(i) for all combinations of the axis indices of dims.

This can be used to view/index into arbitrary dimensions over an array, and is especially useful when combined with otherdims, to iterate over the indices of unknown dimension.

DimIndices can be used directly in getindex like CartesianIndices, and freely mixed with individual Dimensions or tuples of Dimension.

Example

Index a DimArray with DimIndices.

Notice that unlike CartesianIndices, it doesn't matter if the dimensions are not in the same order. Or even if they are not all contained in each.

julia
julia> A = rand(Y(0.0:0.3:1.0), X('a':'f'))
┌ 4×6 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────── dims ┐
  ↓ Y Sampled{Float64} 0.0:0.3:0.9 ForwardOrdered Regular Points,
  → X Categorical{Char} 'a':1:'f' ForwardOrdered
└─────────────────────────────────────────────────────────────────┘
 ↓ →   'a'       'b'       'c'        'd'        'e'       'f'
 0.0  0.9063    0.253849  0.0991336  0.0320967  0.774092  0.893537
 0.3  0.443494  0.334152  0.125287   0.350546   0.183555  0.354868
 0.6  0.745673  0.427328  0.692209   0.930332   0.297023  0.131798
 0.9  0.512083  0.867547  0.136551   0.959434   0.150155  0.941133

julia> di = DimIndices((X(1:2:4), Y(1:2:4)))
┌ 2×2 DimIndices{Tuple{X{Int64}, Y{Int64}}, 2} ┐
├──────────────────────────────────────── dims ┤
  ↓ X 1:2:3,
  → Y 1:2:3
└──────────────────────────────────────────────┘
 ↓ →  1                3
 1     (↓ X 1, → Y 1)   (↓ X 1, → Y 3)
 3     (↓ X 3, → Y 1)   (↓ X 3, → Y 3)

julia> A[di] # Index A with these indices
┌ 2×2 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────── dims ┐
  ↓ Y Sampled{Float64} 0.0:0.6:0.6 ForwardOrdered Regular Points,
  → X Categorical{Char} 'a':2:'c' ForwardOrdered
└─────────────────────────────────────────────────────────────────┘
 ↓ →   'a'       'c'
 0.0  0.9063    0.0991336
 0.6  0.745673  0.692209

source

DimensionalData.DimSelectors Type
julia
DimSelectors <: AbstractArray

DimSelectors(x; selectors, atol...)
DimSelectors(dims::Tuple; selectors, atol...)
DimSelectors(dims::Dimension; selectors, atol...)

Like DimIndices, but returns Dimensions holding the chosen Selectors.

Indexing into another AbstractDimArray with DimSelectors is similar to doing an interpolation.

Keywords

  • selectors: Near, At or Contains, or a mixed tuple of these. At is the default, meaning only exact or within atol values are used.

  • atol: used for At selectors only, as the atol value.

Example

Here we can interpolate a DimArray to the lookups of another DimArray using DimSelectors with Near. This is essentially equivalent to nearest neighbour interpolation.

julia
julia> A = rand(X(1.0:3.0:30.0), Y(1.0:5.0:30.0), Ti(1:2));

julia> target = rand(X(1.0:10.0:30.0), Y(1.0:10.0:30.0));

julia> A[DimSelectors(target; selectors=Near), Ti=2]
┌ 3×3 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────────── dims ┐
  ↓ X Sampled{Float64} [1.0, 10.0, 22.0] ForwardOrdered Irregular Points,
  → Y Sampled{Float64} [1.0, 11.0, 21.0] ForwardOrdered Irregular Points
└─────────────────────────────────────────────────────────────────────────┘
  ↓ →  1.0        11.0       21.0
  1.0  0.691162    0.218579   0.539076
 10.0  0.0303789   0.420756   0.485687
 22.0  0.0967863   0.864856   0.870485

Using At would make sure we only use exact interpolation, while Contains with sampling of Intervals would make sure that each values is taken only from an Interval that is present in the lookups.

source

DimensionalData.DimPoints Type
julia
DimPoints <: AbstractArray

DimPoints(x; order)
DimPoints(dims::Tuple; order)
DimPoints(dims::Dimension; order)

Like CartesianIndices, but for the point values of the dimension index. Behaves as an Array of Tuple lookup values (whatever they are) for all combinations of the lookup values of dims.

Either a Dimension, a Tuple of Dimension or an object x that defines a dims method can be passed in.

Keywords

  • order: determines the order of the points, the same as the order of dims by default.

source

Tables.jl/TableTraits.jl interface ​

DimensionalData.AbstractDimTable Type
julia
AbstractDimTable <: Tables.AbstractColumns

Abstract supertype for dim tables

source

DimensionalData.DimTable Type
julia
DimTable <: AbstractDimTable

DimTable(s::AbstractDimStack; mergedims=nothing)
DimTable(x::AbstractDimArray; layersfrom=nothing, mergedims=nothing)
DimTable(xs::Vararg{AbstractDimArray}; layernames=nothing, mergedims=nothing)

Construct a Tables.jl/TableTraits.jl compatible object out of an AbstractDimArray or AbstractDimStack.

This table will have columns for the array data and columns for each Dimension index, as a [DimColumn]. These are lazy, and generated as required.

Column names are converted from the dimension types using DimensionalData.name. This means type Ti becomes the column name :Ti, and Dim{:custom} becomes :custom.

To get dimension columns, you can index with Dimension (X()) or Dimension type (X) as well as the regular Int or Symbol.

Keywords

  • mergedims: Combine two or more dimensions into a new dimension.

  • layersfrom: Treat a dimension of an AbstractDimArray as layers of an AbstractDimStack.

Example

julia
julia> using DimensionalData, Tables

julia> a = DimArray(ones(16, 16, 3), (X, Y, Dim{:band}))
┌ 16×16×3 DimArray{Float64, 3} ┐
├──────────────────────── dims ┤
  ↓ X, → Y, ↗ band
└──────────────────────────────┘
[:, :, 1]
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0  …  1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0  …  1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0  …  1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0     1.0  1.0  1.0  1.0  1.0  1.0  1.0
 1.0  1.0  1.0  1.0  1.0  1.0  1.0  1.0  …  1.0  1.0  1.0  1.0  1.0  1.0  1.0

julia>

source

Group by methods ​

For transforming DimensionalData objects:

DataAPI.groupby Function
julia
groupby(A::Union{AbstractDimArray,AbstractDimStack}, dims::Pair...)
groupby(A::Union{AbstractDimArray,AbstractDimStack}, dims::Dimension{<:Callable}...)

Group A by grouping functions or Bins over multiple dimensions.

Arguments

  • A: any AbstractDimArray or AbstractDimStack.

  • dims: Pairs such as groups = groupby(A, :dimname => groupingfunction) or wrapped Dimensions like groups = groupby(A, DimType(groupingfunction)). Instead of a grouping function Bins can be used to specify group bins.

Return value

A DimGroupByArray is returned, which is basically a regular AbstractDimArray but holding the grouped AbstractDimArray or AbstractDimStack. Its dims hold the sorted values returned by the grouping function/s.

Base julia and package methods work on DimGroupByArray as for any other AbstractArray of AbstractArray.

It is common to broadcast or map a reducing function over groups, such as mean or sum, like mean.(groups) or map(mean, groups). This will return a regular DimArray, or DimGroupByArray if dims keyword is used in the reducing function or it otherwise returns an AbstractDimArray or AbstractDimStack.

Example

Group some data along the time dimension:

julia
julia> using DimensionalData, Dates

julia> A = rand(X(1:0.1:20), Y(1:20), Ti(DateTime(2000):Day(3):DateTime(2003)));

julia> groups = groupby(A, Ti => month) # Group by month
┌ 12-element DimGroupByArray{DimArray{Float64,2},1} ┐
├───────────────────────────────────────────────────┴───────────── dims ┐
  ↓ Ti Sampled{Int64} [1, 2, …, 11, 12] ForwardOrdered Irregular Points
├───────────────────────────────────────────────────────────── metadata ┤
  Dict{Symbol, Any} with 1 entry:
  :groupby => :Ti=>month
├─────────────────────────────────────────────────────────── group dims ┤
  ↓ X, → Y, ↗ Ti
└───────────────────────────────────────────────────────────────────────┘
  1  191×20×32 DimArray
  2  191×20×28 DimArray
  3  191×20×31 DimArray
  ⋮
 11  191×20×30 DimArray
 12  191×20×31 DimArray

And take the mean:

julia
julia> groupmeans = mean.(groups) # Take the monthly mean
┌ 12-element DimArray{Float64, 1} ┐
├─────────────────────────────────┴─────────────────────────────── dims ┐
  ↓ Ti Sampled{Int64} [1, 2, …, 11, 12] ForwardOrdered Irregular Points
├───────────────────────────────────────────────────────────── metadata ┤
  Dict{Symbol, Any} with 1 entry:
  :groupby => :Ti=>month
└───────────────────────────────────────────────────────────────────────┘
  1  0.500064
  2  0.499762
  3  0.500083
  4  0.499985
  ⋮
 10  0.500874
 11  0.498704
 12  0.50047

Calculate daily anomalies from the monthly mean. Notice we map a broadcast .- rather than -. This is because the size of the arrays to not match after application of mean.

julia
julia> map(.-, groupby(A, Ti=>month), mean.(groupby(A, Ti=>month), dims=Ti));

Or do something else with Y:

julia
julia> groupmeans = mean.(groupby(A, Ti=>month, Y=>isodd))
┌ 12×2 DimArray{Float64, 2} ┐
├───────────────────────────┴────────────────────────────────────── dims ┐
  ↓ Ti Sampled{Int64} [1, 2, …, 11, 12] ForwardOrdered Irregular Points,
  → Y  Sampled{Bool} [false, true] ForwardOrdered Irregular Points
├────────────────────────────────────────────────────────────── metadata ┤
  Dict{Symbol, Any} with 1 entry:
  :groupby => (:Ti=>month, :Y=>isodd)
└────────────────────────────────────────────────────────────────────────┘
  ↓ →  false         true
  1        0.499594     0.500533
  2        0.498145     0.501379
  ⋮
 10        0.501105     0.500644
 11        0.498606     0.498801
 12        0.501643     0.499298

source

DimensionalData.DimGroupByArray Type
julia
DimGroupByArray <: AbstractDimArray

DimGroupByArray is essentially a DimArray but holding the results of a groupby operation.

Its dimensions are the sorted results of the grouping functions used in groupby.

This wrapper allows for specialisations on later broadcast or reducing operations, e.g. for chunk reading with DiskArrays.jl, because we know the data originates from a single array.

source

DimensionalData.Bins Type
julia
Bins(f, bins; labels, pad)
Bins(bins; labels, pad)

Specify bins to reduce groups after applying function f.

  • f: a grouping function of the lookup values, by default identity.

  • bins:

    • an Integer will divide the group values into equally spaced sections.

    • an AbstractArray of values will be treated as exact matches for the return value of f. For example, 1:3 will create 3 bins - 1, 2, 3.

    • an AbstractArray of IntervalSets.Interval can be used to explicitly define the intervals. Overlapping intervals have undefined behaviour.

Keywords

  • pad: fraction of the total interval to pad at each end when Bins contains an Integer. This avoids losing the edge values. Note this is a messy solution - it will often be prefereble to manually specify a Vector of chosen Intervals rather than relying on passing an Integer and pad.

  • labels: a list of descriptive labels for the bins. The labels need to have the same length as bins.

When the return value of f is a tuple, binning is applied to the last value of the tuples.

source

DimensionalData.ranges Function
julia
ranges(A::AbstractRange{<:Integer})

Generate a Vector of UnitRange with length step(A)

source

DimensionalData.intervals Function
julia
intervals(A::AbstractRange)

Generate a Vector of UnitRange with length step(A)

source

DimensionalData.CyclicBins Type
julia
CyclicBins(f; cycle, start, step, labels)

Cyclic bins to reduce groups after applying function f. Groups can wrap around the cycle. This is used for grouping in seasons, months and hours but can also be used for custom cycles.

  • f: a grouping function of the lookup values, by default identity.

Keywords

  • cycle: the length of the cycle, in return values of f.

  • start: the start of the cycle: a return value of f.

  • step the number of sequential values to group.

  • labels: either a vector of labels matching the number of groups, or a function that generates labels from Vector{Int} of the selected bins.

When the return value of f is a tuple, binning is applied to the last value of the tuples.

source

DimensionalData.seasons Function
julia
seasons(; [start=Dates.December, labels])

Generates CyclicBins for three month periods.

Keywords

  • start: By default seasons start in December, but any integer 1:12 can be used.

  • labels: either a vector of four labels, or a function that generates labels from Vector{Int} of the selected quarters.

source

DimensionalData.months Function
julia
months(step; [start=Dates.January, labels])

Generates CyclicBins for grouping to arbitrary month periods. These can wrap around the end of a year.

  • step the number of months to group.

Keywords

  • start: By default months start in January, but any integer 1:12 can be used.

  • labels: either a vector of labels matching the number of groups, or a function that generates labels from Vector{Int} of the selected months.

source

DimensionalData.hours Function
julia
hours(step; [start=0, labels])

Generates CyclicBins for grouping to arbitrary hour periods. These can wrap around the end of the day.

  • steps the number of hours to group.

Keywords

  • start: By default seasons start at 0, but any integer 1:24 can be used.

  • labels: either a vector of four labels, or a function that generates labels from Vector{Int} of the selected hours of the day.

source

Utility methods ​

For transforming DimensionalData objects:

DimensionalData.Dimensions.Lookups.set Function
julia
set(x, val)
set(x, args::Pairs...) => x with updated field/s
set(x, args...; kw...) => x with updated field/s
set(x, args::Tuple{Vararg{Dimension}}; kw...) => x with updated field/s

set(dim::Dimension, index::AbstractArray) => Dimension
set(dim::Dimension, lookup::Lookup) => Dimension
set(dim::Dimension, lookupcomponent::LookupTrait) => Dimension
set(dim::Dimension, metadata::AbstractMetadata) => Dimension

Set the properties of an object, its internal data or the traits of its dimensions and lookup index.

As DimensionalData is so strongly typed you do not need to specify what field of a Lookup to set - there is no ambiguity.

To set fields of a Lookup you need to specify the dimension. This can be done using X => val pairs, X = val keyword arguments, or X(val) wrapped arguments.

You can also set the fields of all dimensions by simply passing a single Lookup or lookup trait - it will be set for all dimensions.

When a Dimension or Lookup is passed to set to replace the existing ones, fields that are not set will keep their original values.

Notes:

Changing a lookup index range/vector will also update the step size and order where applicable.

Setting the Order like ForwardOrdered will not reverse the array or dimension to match. Use reverse and reorder to do this.

Examples

julia
julia> using DimensionalData; const DD = DimensionalData;

julia> da = DimArray(zeros(3, 4), (custom=10.0:010.0:30.0, Z=-20:010.0:10.0));

julia> set(da, ones(3, 4))
┌ 3×4 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────────── dims ┐
  ↓ custom Sampled{Float64} 10.0:10.0:30.0 ForwardOrdered Regular Points,
  → Z      Sampled{Float64} -20.0:10.0:10.0 ForwardOrdered Regular Points
└─────────────────────────────────────────────────────────────────────────┘
  ↓ →  -20.0  -10.0  0.0  10.0
 10.0    1.0    1.0  1.0   1.0
 20.0    1.0    1.0  1.0   1.0
 30.0    1.0    1.0  1.0   1.0

Change the Dimension wrapper type:

julia
julia> set(da, :Z => Ti, :custom => Z)
┌ 3×4 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────── dims ┐
  ↓ Z  Sampled{Float64} 10.0:10.0:30.0 ForwardOrdered Regular Points,
  → Ti Sampled{Float64} -20.0:10.0:10.0 ForwardOrdered Regular Points
└─────────────────────────────────────────────────────────────────────┘
  ↓ →  -20.0  -10.0  0.0  10.0
 10.0    0.0    0.0  0.0   0.0
 20.0    0.0    0.0  0.0   0.0
 30.0    0.0    0.0  0.0   0.0

Change the lookup Vector:

julia
julia> set(da, Z => [:a, :b, :c, :d], :custom => [4, 5, 6])
┌ 3×4 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────────── dims ┐
  ↓ custom Sampled{Int64} [4, 5, 6] ForwardOrdered Regular Points,
  → Z      Sampled{Symbol} [:a, :b, :c, :d] ForwardOrdered Regular Points
└─────────────────────────────────────────────────────────────────────────┘
 ↓ →   :a   :b   :c   :d
 4    0.0  0.0  0.0  0.0
 5    0.0  0.0  0.0  0.0
 6    0.0  0.0  0.0  0.0

Change the Lookup type:

julia
julia> set(da, Z=DD.NoLookup(), custom=DD.Sampled())
┌ 3×4 DimArray{Float64, 2} ┐
├──────────────────────────┴──────────────────────────────────────── dims ┐
  ↓ custom Sampled{Float64} 10.0:10.0:30.0 ForwardOrdered Regular Points,
  → Z
└─────────────────────────────────────────────────────────────────────────┘
 10.0  0.0  0.0  0.0  0.0
 20.0  0.0  0.0  0.0  0.0
 30.0  0.0  0.0  0.0  0.0

Change the Sampling trait:

julia
julia> set(da, :custom => DD.Irregular(10, 12), Z => DD.Regular(9.9))
┌ 3×4 DimArray{Float64, 2} ┐
├──────────────────────────┴────────────────────────────────────────── dims ┐
  ↓ custom Sampled{Float64} 10.0:10.0:30.0 ForwardOrdered Irregular Points,
  → Z      Sampled{Float64} -20.0:10.0:10.0 ForwardOrdered Regular Points
└───────────────────────────────────────────────────────────────────────────┘
  ↓ →  -20.0  -10.0  0.0  10.0
 10.0    0.0    0.0  0.0   0.0
 20.0    0.0    0.0  0.0   0.0
 30.0    0.0    0.0  0.0   0.0

source

DimensionalData.Dimensions.Lookups.rebuild Function
julia
rebuild(x; kw...)

Rebuild an object struct with updated field values.

x can be a AbstractDimArray, a Dimension, Lookup or other custom types.

This is an abstraction that allows inbuilt and custom types to be rebuilt to update their fields, as most objects in DimensionalData.jl are immutable.

Rebuild is mostly automated using ConstructionBase.setproperties. It should only be defined if your object has fields with with different names to DimensionalData objects. Try not to do that!

The arguments required are defined for the abstract type that has a rebuild method.

AbstractBasicDimArray:

  • dims: a Tuple of Dimension

AbstractDimArray:

  • data: the parent object - an AbstractArray

  • dims: a Tuple of Dimension

  • refdims: a Tuple of Dimension

  • name: A Symbol, or NoName and Name on GPU.

  • metadata: A Dict-like object

AbstractDimStack:

  • data: the parent object, often a NamedTuple

  • dims, refdims, metadata

Dimension:

  • val: anything.

Lookup:

  • data: the parent object, an AbstractArray

  • Note: argument rebuild is deprecated on AbstractDimArray and

AbstractDimStack in favour of always using the keyword version. In future the argument version will only be used on Dimension, which only have one argument.

source

DimensionalData.modify Function
julia
modify(f, A::AbstractDimArray) => AbstractDimArray
modify(f, s::AbstractDimStack) => AbstractDimStack
modify(f, dim::Dimension) => Dimension
modify(f, x, lookupdim::Dimension) => typeof(x)

Modify the parent data, rebuilding the object wrapper without change. f must return a AbstractArray of the same size as the original.

This method is mostly useful as a way of swapping the parent array type of an object.

Example

If we have a previously-defined DimArray, we can copy it to an Nvidia GPU with:

julia
A = DimArray(rand(100, 100), (X, Y))
modify(CuArray, A)

This also works for all the data layers in a DimStack.

source

DimensionalData.@d Macro
julia
@d broadcast_expression options

Dimensional broadcast macro extending Base Julia broadcasting to work with missing and permuted dimensions.

Will permute and reshape singleton dimensions so that all AbstractDimArray in a broadcast will broadcast over matching dimensions.

It is possible to pass options as the second argument of the macro to control the behaviour, as a single assignment or as a NamedTuple. Options names must be written explicitly, not passed in namedtuple variable.

Options

  • dims: Pass a Tuple of Dimensions, Dimension types or Symbols to fix the dimension order of the output array. Otherwise dimensions will be in order of appearance. If dims with lookups are passed, these will be applied to the returned array with set.

  • strict: true or false. Check that all lookup values match explicitly.

All other keywords are passed to DimensionalData.rebuild. This means name, metadata, etc for the returned array can be set here, or for example missingval in Rasters.jl.

Example

julia
using DimensionalData
da1 = ones(X(3))
da2 = fill(2, Y(4), X(3))

@d da1 .* da2
@d da1 .* da2 .+ 5 dims=(Y, X)
@d da1 .* da2 .+ 5 (dims=(Y, X), strict=false, name=:testname)

Use with @.

@d does not imply @.. You need to specify each broadcast. But @. can be used with @d as the inner macro.

julia
using DimensionalData
da1 = ones(X(3))
da2 = fill(2, Y(4), X(3))

@d @. da1 * da2
# Use parentheses around `@.` if you need to pass options
@d (@. da1 * da2 .+ 5) dims=(Y, X)

source

DimensionalData.broadcast_dims Function
julia
broadcast_dims(f, sources::AbstractDimArray...) => AbstractDimArray

Broadcast function f over the AbstractDimArrays in sources, permuting and reshaping dimensions to match where required. The result will contain all the dimensions in all passed in arrays in the order in which they are found.

Arguments

  • sources: AbstractDimArrays to broadcast over with f.

This is like broadcasting over every slice of A if it is sliced by the dimensions of B.

source

DimensionalData.broadcast_dims! Function
julia
broadcast_dims!(f, dest::AbstractDimArray, sources::AbstractDimArray...) => dest

Broadcast function f over the AbstractDimArrays in sources, writing to dest. sources are permuting and reshaping dimensions to match where required.

The result will contain all the dimensions in all passed in arrays, in the order in which they are found.

Arguments

  • dest: AbstractDimArray to update.

  • sources: AbstractDimArrays to broadcast over with f.

source

DimensionalData.mergedims Function
julia
mergedims(old_dims => new_dim) => Dimension

Return a dimension new_dim whose indices are a MergedLookup of the indices of old_dims.

source

julia
mergedims(dims, old_dims => new_dim, others::Pair...) => dims_new

If dimensions old_dims, new_dim, etc. are found in dims, then return new dims_new where all dims in old_dims have been combined into a single dim new_dim. The returned dimension will keep only the name of new_dim. Its coords will be a MergedLookup of the coords of the dims in old_dims. New dimensions are always placed at the end of dims_new. others contains other dimension pairs to be merged.

Example

julia
julia> using DimensionalData

julia> ds = (X(0:0.1:0.4), Y(10:10:100), Ti([0, 3, 4]))
(↓ X  0.0:0.1:0.4,
→ Y  10:10:100,
↗ Ti [0, 3, 4])

julia> mergedims(ds, (X, Y) => :space)
(↓ Ti    [0, 3, 4],
→ space MergedLookup{Tuple{Float64, Int64}} [(0.0, 10), (0.1, 10), …, (0.3, 100), (0.4, 100)] (↓ X, → Y))

source

julia
mergedims(A::AbstractDimArray, dim_pairs::Pair...) => AbstractDimArray
mergedims(A::AbstractDimStack, dim_pairs::Pair...) => AbstractDimStack

Return a new array or stack whose dimensions are the result of mergedims(dims(A), dim_pairs).

source

DimensionalData.unmergedims Function
julia
unmergedims(merged_dims::Tuple{Vararg{Dimension}}) => Tuple{Vararg{Dimension}}

Return the unmerged dimensions from a tuple of merged dimensions. However, the order of the original dimensions are not necessarily preserved.

source

julia
unmergedims(A::AbstractDimArray, original_dims) => AbstractDimArray
unmergedims(A::AbstractDimStack, original_dims) => AbstractDimStack

Return a new array or stack whose dimensions are restored to their original prior to calling mergedims(A, dim_pairs).

source

DimensionalData.reorder Function
julia
reorder(A::Union{AbstractDimArray,AbstractDimStack}, order::Pair...)
reorder(A::Union{AbstractDimArray,AbstractDimStack}, order)
reorder(A::Dimension, order::Order)

Reorder every dims index/array to order, or reorder index for the given dimension(s) in order.

order can be an Order, Dimension => Order pairs. A Tuple of Dimensions or any object that defines dims can be used in which case the dimensions of this object are used for reordering.

If no axis reversal is required the same objects will be returned, without allocation.

Example

julia
using DimensionalData

# Create a DimArray
da = DimArray([1 2 3; 4 5 6], (X(10:10:20), Y(300:-100:100)))

# Reverse it
rev = reverse(da, dims=Y)

# using `da` in reorder will return it to the original order
reorder(rev, da) == da

# output
true

source

Global lookup strictness settings ​

Control how strict DimensionalData when comparing Lookups before doing broadcasts and matrix multipications.

In some cases (especially DimVector and small DimArray) checking lookup values match may be too costly compared to the operations. You can turn check the current setting and turn them on or off with these methods.

DimensionalData.strict_broadcast Function
julia
strict_broadcast()

Check if strict broadcasting checks are active.

With strict=true we check Lookup Order and values before brodcasting, to ensure that dimensions match closely.

An exception to this rule is when dimension are of length one, as these is ignored in broadcasts.

We always check that dimension names match in broadcasts. If you don't want this either, explicitly use parent(A) before broadcasting to remove the AbstractDimArray wrapper completely.

source

DimensionalData.strict_broadcast! Function
julia
strict_broadcast!(x::Bool)

Set global broadcasting checks to strict, or not for all AbstractDimArray.

With strict=true we check Lookup Order and values before brodcasting, to ensure that dimensions match closely.

An exception to this rule is when dimension are of length one, as these is ignored in broadcasts.

We always check that dimension names match in broadcasts. If you don't want this either, explicitly use parent(A) before broadcasting to remove the AbstractDimArray wrapper completely.

source

DimensionalData.strict_matmul Function
julia
strict_matmul()

Check if strickt broadcasting checks are active.

With strict=true we check Lookup Order and values before attempting matrix multiplication, to ensure that dimensions match closely.

We always check that dimension names match in matrix multiplication. If you don't want this either, explicitly use parent(A) before multiplying to remove the AbstractDimArray wrapper completely.

source

DimensionalData.strict_matmul! Function
julia
strict_matmul!(x::Bool)

Set global matrix multiplication checks to strict, or not for all AbstractDimArray.

With strict=true we check Lookup Order and values before attempting matrix multiplication, to ensure that dimensions match closely.

We always check that dimension names match in matrix multiplication. If you don't want this either, explicitly use parent(A) before multiplying to remove the AbstractDimArray wrapper completely.

source

Base methods

Base.cat Function
julia
Base.cat(stacks::AbstractDimStack...; [keys=keys(stacks[1])], dims)

Concatenate all or a subset of layers for all passed in stacks.

Keywords

  • keys: Tuple of Symbol for the stack keys to concatenate.

  • dims: Dimension of child array to concatenate on.

Example

Concatenate the :sea_surface_temp and :humidity layers in the time dimension:

julia
cat(stacks...; keys=(:sea_surface_temp, :humidity), dims=Ti)

source

Base.copy! Function
julia
Base.copy!(dst::AbstractArray, src::AbstractDimStack, key::Key)

Copy the stack layer key to dst, which can be any AbstractArray.

Example

Copy the :humidity layer from stack to array.

julia
copy!(array, stack, :humidity)

source

julia
Base.copy!(dst::AbstractDimStack, src::AbstractDimStack, [keys=keys(dst)])

Copy all or a subset of layers from one stack to another.

Example

Copy just the :sea_surface_temp and :humidity layers from src to dst.

julia
copy!(dst::AbstractDimStack, src::AbstractDimStack, keys=(:sea_surface_temp, :humidity))

source

Base.eachslice Function
julia
Base.eachslice(A::AbstractDimArray; dims,drop=true)

Create a generator that iterates over dimensions dims of A, returning arrays that select all the data from the other dimensions in A using views.

The generator has size and axes equivalent to those of the provided dims if drop=true. Otherwise it will have the same dimensionality as the underlying array with inner dimensions having size 1.

source

julia
Base.eachslice(stack::AbstractDimStack; dims, drop=true)

Create a generator that iterates over dimensions dims of stack, returning stacks that select all the data from the other dimensions in stack using views.

The generator has size and axes equivalent to those of the provided dims.

Examples

julia
julia> ds = DimStack((
           x=DimArray(randn(2, 3, 4), (X([:x1, :x2]), Y(1:3), Z)),
           y=DimArray(randn(2, 3, 5), (X([:x1, :x2]), Y(1:3), Ti))
       ));

julia> slices = eachslice(ds; dims=(Z, X));

julia> size(slices)
(4, 2)

julia> map(dims, axes(slices))
(↓ Z Base.OneTo(4),
→ X Base.OneTo(2))

julia> first(slices)
┌ 3×5 DimStack ┐
├──────────────┴─────────────────────────────────── dims ┐
  ↓ Y  Sampled{Int64} 1:3 ForwardOrdered Regular Points,
  → Ti
├──────────────────────────────────────────────── layers ┤
  :x eltype: Float64 dims: Y size: 3
  :y eltype: Float64 dims: Y, Ti size: 3×5
└────────────────────────────────────────────────────────┘

source

Most base methods work as expected, using Dimension wherever a dims keyword is used. They are not all specifically documented here.

Name ​

DimensionalData.AbstractName Type
julia
AbstractName

Abstract supertype for name wrappers.

source

DimensionalData.Name Type
julia
Name <: AbstractName

Name(name::Union{Symbol,Name) => Name
Name(name::NoName) => NoName

Name wrapper. This lets arrays keep symbol names when the array wrapper needs to be isbits, like for use on GPUs. It makes the name a property of the type. It's not necessary to use in normal use, a symbol is probably easier.

source

DimensionalData.NoName Type
julia
NoName <: AbstractName

NoName()

NoName specifies an array is not named, and is the default name value for all AbstractDimArrays.

source

Internal interface ​

DimensionalData.DimArrayInterface Type
julia
    DimArrayInterface

An Interfaces.jl Interface with mandatory components (:dims, :refdims_base, :ndims, :size, :rebuild_parent, :rebuild_dims, :rebuild_parent_kw, :rebuild_dims_kw, :rebuild) and optional components (:refdims, :name, :metadata).

This is an early stage of inteface definition, many things are not yet tested.

Pass constructed AbstractDimArrays as test data.

They must not be zero dimensional, and should test at least 1, 2, and 3 dimensions.

Extended help

Mandatory keys:

  • dims:

    • defines a dims method

    • dims are updated on getindex

  • refdims_base: refdims returns a tuple of Dimension or empty

  • ndims: number of dims matches dimensions of array

  • size: length of dims matches dimensions of array

  • rebuild_parent: rebuild parent from args

  • rebuild_dims: rebuild paaarnet and dims from args

  • rebuild_parent_kw: rebuild parent from args

  • rebuild_dims_kw: rebuild dims from args

  • rebuild: all rebuild arguments and keywords are accepted

Optional keys:

  • refdims:

    • refdims are updated in args rebuild

    • refdims are updated in kw rebuild

    • dropped dimensions are added to refdims

  • name:

    • rebuild updates name in arg rebuild

    • rebuild updates name in kw rebuild

  • metadata:

    • rebuild updates metadata in arg rebuild

    • rebuild updates metadata in kw rebuild

source

DimensionalData.DimStackInterface Type
julia
    DimStackInterface

An Interfaces.jl Interface with mandatory components (:dims, :refdims_base, :ndims, :size, :rebuild_parent, :rebuild_dims, :rebuild_layerdims, :rebuild_dims_kw, :rebuild_parent_kw, :rebuild_layerdims_kw, :rebuild) and optional components (:refdims, :metadata).

This is an early stage of inteface definition, many things are not yet tested.

Pass constructed AbstractDimArrays as test data.

They must not be zero dimensional, and should test at least 1, 2, and 3 dimensions.

Extended help

Mandatory keys:

  • dims:

    • defines a dims method

    • dims are updated on getindex

  • refdims_base: refdims returns a tuple of Dimension or empty

  • ndims: number of dims matches ndims of stack

  • size: length of dims matches size of stack

  • rebuild_parent: rebuild parent from args

  • rebuild_dims: rebuild paaarnet and dims from args

  • rebuild_layerdims: rebuild paaarnet and dims from args

  • rebuild_dims_kw: rebuild dims from args

  • rebuild_parent_kw: rebuild parent from args

  • rebuild_layerdims_kw: rebuild parent from args

  • rebuild: all rebuild arguments and keywords are accepted

Optional keys:

  • refdims:

    • refdims are updated in args rebuild

    • refdims are updated in kw rebuild

    • dropped dimensions are added to refdims

  • metadata:

    • rebuild updates metadata in arg rebuild

    • rebuild updates metadata in kw rebuild

source

DimensionalData.rebuild_from_arrays Function
julia
rebuild_from_arrays(s::AbstractDimStack, das::NamedTuple{<:Any,<:Tuple{Vararg{AbstractDimArray}}}; kw...)

Rebuild an AbstractDimStack from a Tuple or NamedTuple of AbstractDimArray and an existing stack.

Keywords

Keywords are simply the fields of the stack object:

  • data

  • dims

  • refdims

  • metadata

  • layerdims

  • layermetadata

source

DimensionalData.show_main Function
julia
show_main(io::IO, mime, A::AbstractDimArray)
show_main(io::IO, mime, A::AbstractDimStack)

Interface methods for adding the main part of show

At the least, you likely want to call:

julia
print_top(io, mime, A)

But read the DimensionalData.jl show.jl code for details.

source

DimensionalData.show_after Function
julia
show_after(io::IO, mime, A::AbstractDimArray)
show_after(io::IO, mime, A::AbstractDimStack)

Interface methods for adding additional show text for AbstractDimArray/AbstractDimStack subtypes.

Always include kw to avoid future breaking changes

Additional keywords may be added at any time.

blockwidth is passed in context

julia
blockwidth = get(io, :blockwidth, 10000)

Note - a ANSI box is left unclosed. This method needs to close it, or add more. blockwidth is the maximum length of the inner text.

Most likely you always want to at least close the show blocks with:

julia
print_block_close(io, blockwidth)

But read the DimensionalData.jl show.jl code for details.

source

DimensionalData.refdims_title Function
julia
refdims_title(A::AbstractDimArray)
refdims_title(refdims::Tuple)
refdims_title(refdim::Dimension)

Generate a title string based on reference dimension values.

source