Make LinearMaps callable (#128)

Author: dkarrasch

Project.toml

@@ -1,6 +1,6 @@
 name = "LinearMaps"
 uuid = "7a12625a-238d-50fd-b39a-03d52299707e"
-version = "3.0.1"
+version = "3.1.0"
 
 [deps]
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"

README.md

@@ -7,8 +7,7 @@ The only requirement for a `LinearMap` is that it can act on a vector (by multip
 
 | **Documentation**                                                               | **Build Status**                                                                                |
 |:-------------------------------------------------------------------------------:|:-----------------------------------------------------------------------------------------------:|
-| [![][docs-stable-img]][docs-stable-url] [![][docs-dev-img]][docs-dev-url] | [![][build-img]][build-url] [![][codecov-img]][codecov-url] [![][license-img]][license-url] |
-
+| [![stable docs][docs-stable-img]][docs-stable-url] [![dev docs][docs-dev-img]][docs-dev-url] | [![build status][build-img]][build-url] [![coverage][codecov-img]][codecov-url] [![license][license-img]][license-url] |
 
 ## Installation
 

docs/Project.toml

@@ -7,5 +7,5 @@ Literate = "98b081ad-f1c9-55d3-8b20-4c87d4299306"
 
 [compat]
 BenchmarkTools = "0.4, 0.5"
-Documenter = "0.25"
+Documenter = "0.25, 0.26"
 Literate = "2"

docs/src/history.md

@@ -1,5 +1,10 @@
 # Version history
 
+## What's new in v3.1
+
+* In Julia v1.3 and above, `LinearMap`-typed objects are callable on `AbstractVector`s:
+  For `L::LinearMap` and `x::AbstractVector`, `L(x) = L*x`.
+
 ## What's new in v3.0
 
 * BREAKING change: Internally, any dependence on former `A*_mul_B!` methods is abandonned.

docs/src/index.md

@@ -87,6 +87,14 @@ Arpack.eigs(Δ; nev=3, which=:LR)
 KrylovKit.eigsolve(x -> Δ*x, size(Δ, 1), 3, :LR)
 ```
 
+In Julia v1.3 and above, the last line can be simplified to
+
+```julia
+KrylovKit.eigsolve(Δ, size(Δ, 1), 3, :LR)
+````
+
+leveraging the fact that objects of type `L <: LinearMap` are callable.
+
 ## Philosophy
 
 Several iterative linear algebra methods such as linear solvers or eigensolvers

src/LinearMaps.jl

@@ -90,11 +90,20 @@ convert_to_lmaps(A) = (convert_to_lmaps_(A),)
 
 Compute the action of the linear map `A` on the vector `x`.
 
+!!! compat "Julia 1.3"
+    In Julia versions v1.3 and above, objects `L` of any subtype of `LinearMap`
+    are callable in the sense that `L(x) = L*x` for `x::AbstractVector`.
+
 ## Examples
 ```jldoctest; setup=(using LinearAlgebra, LinearMaps)
 julia> A=LinearMap([1.0 2.0; 3.0 4.0]); x=[1.0, 1.0];
 
 julia> A*x
+2-element Array{Float64,1}:
+ 3.0
+ 7.0
+
+julia> A(x)
 2-element Array{Float64,1}:
  3.0
  7.0
@@ -104,6 +113,9 @@ function Base.:(*)(A::LinearMap, x::AbstractVector)
     check_dim_mul(A, x)
     return mul!(similar(x, promote_type(eltype(A), eltype(x)), size(A, 1)), A, x)
 end
+if VERSION ≥ v"1.3"
+    (L::LinearMap)(x::AbstractVector) = L*x
+end
 
 """
     mul!(Y::AbstractVecOrMat, A::LinearMap, B::AbstractVector) -> Y

src/kronecker.jl

@@ -30,7 +30,7 @@ holds. Upon vector multiplication, this rule is checked for applicability.
 # Examples
 ```jldoctest; setup=(using LinearAlgebra, SparseArrays, LinearMaps)
 julia> J = LinearMap(I, 2) # 2×2 identity map
-LinearMaps.UniformScalingMap{Bool}(true, 2)
+2×2 LinearMaps.UniformScalingMap{Bool} with scaling factor: true
 
 julia> E = spdiagm(-1 => trues(1)); D = E + E' - 2I;
 
@@ -228,7 +228,7 @@ For convenience, one can also use `A ⊕ B` or `⊕(A, B, Cs...)` (typed as
 # Examples
 ```jldoctest; setup=(using LinearAlgebra, SparseArrays, LinearMaps)
 julia> J = LinearMap(I, 2) # 2×2 identity map
-LinearMaps.UniformScalingMap{Bool}(true, 2)
+2×2 LinearMaps.UniformScalingMap{Bool} with scaling factor: true
 
 julia> E = spdiagm(-1 => trues(1)); D = LinearMap(E + E' - 2I);
 

test/linearmaps.jl

@@ -61,6 +61,9 @@ LinearAlgebra.mul!(y::AbstractVector, A::Union{SimpleFunctionMap,SimpleComplexFu
     α = rand(ComplexF64); β = rand(ComplexF64)
     v = rand(ComplexF64, 10); V = rand(ComplexF64, 10, 3)
     w = rand(ComplexF64, 10); W = rand(ComplexF64, 10, 3)
+    if VERSION ≥ v"1.3"
+        F(v) == F*v
+    end    
     @test mul!(w, F, v) === w == F * v
     @test_throws ErrorException F' * v
     @test_throws ErrorException transpose(F) * v