LinearMaps

A Julia package for defining and working with linear maps, also known as linear transformations or linear operators acting on vectors. The only requirement for a LinearMap is that it can act on a vector (by multiplication) efficiently.

What's new in v2.3.0

• Fully Julia v0.7/v1.0/v1.1 compatible.
• Full support of noncommutative number types such as quaternions.

What's new in v2.2.0

• Fully Julia v0.7/v1.0 compatible.
• A convert(SparseMatrixCSC, A::LinearMap) function, that calls the sparse matrix generating function.

What's new in v2.1.0

• Fully Julia v0.7 compatible; dropped compatibility for previous versions of Julia from LinearMaps.jl v2.0.0 on.
• A 5-argument version for mul!(y, A::LinearMap, x, α=1, β=0), which computes y := α * A * x + β * y and implements the usual 3-argument mul!(y, A, x) for the default α and β.
• Synonymous convert(Matrix, A::LinearMap) and convert(Array, A::LinearMap) functions, that call the Matrix constructor and return the matrix representation of A.
• Multiplication with matrices, interpreted as a block row vector of vectors:
  • mul!(Y::AbstractArray, A::LinearMap, X::AbstractArray, α=1, β=0): applies A to each column of X and stores the result in-place in the corresponding column of Y;
  • for the out-of-place multiplication, the approach is to compute convert(Matrix, A * X); this is equivalent to applying A to each column of X. In generic code which handles both A::AbstractMatrix and A::LinearMap, the additional call to convert is a noop when A is a matrix.
• Full compatibility with Arpack.jl's eigs and svds; previously only eigs was working. For more, nicely collaborating packages see the Example section.

Installation

Install with the package manager, i.e., ]add LinearMaps (or Pkg.add("LinearMaps") in Julia versions below 0.7).

Philosophy

Several iterative linear algebra methods such as linear solvers or eigensolvers only require an efficient evaluation of the matrix vector product, where the concept of a matrix can be formalized / generalized to a linear map (or linear operator in the special case of a square matrix).

The LinearMaps package provides the following functionality:

1. A LinearMap type that shares with the AbstractMatrix type that it responds to the functions size, eltype, isreal, issymmetric, ishermitian and isposdef, transpose and adjoint and multiplication with a vector using both * or the in-place version mul!. Linear algebra functions that use duck-typing for their arguments can handle LinearMap objects similar to AbstractMatrix objects, provided that they can be written using the above methods. Unlike AbstractMatrix types, LinearMap objects cannot be indexed, neither using getindex or setindex!.

2. A single method LinearMap function that acts as a general purpose constructor (though it is only an abstract type) and allows to construct linear map objects from functions, or to wrap objects of type AbstractMatrix or LinearMap. The latter functionality is useful to (re)define the properties (isreal, issymmetric, ishermitian, isposdef) of the existing matrix or linear map.

3. A framework for combining objects of type LinearMap and of type AbstractMatrix using linear combinations, transposition and composition, where the linear map resulting from these operations is never explicitly evaluated but only its matrix vector product is defined (i.e. lazy evaluation). The matrix vector product is written to minimize memory allocation by using a minimal number of temporary vectors. There is full support for the in-place version mul!, which should be preferred for higher efficiency in critical algorithms. In addition, it tries to recognize the properties of combinations of linear maps. In particular, compositions such as A'*A for arbitrary A or even A'*B*C*B'*A with arbitrary A and B and positive definite C are recognized as being positive definite and hermitian. In case a certain property of the resulting LinearMap object is not correctly inferred, the LinearMap method can be called to redefine the properties.

Methods

• LinearMap

  General purpose method to construct LinearMap objects of specific types, as described in the Types section below

  LinearMap{T}(A::AbstractMatrix[; issymmetric::Bool, ishermitian::Bool, isposdef::Bool])
  LinearMap{T}(A::LinearMap[; issymmetric::Bool, ishermitian::Bool, isposdef::Bool])
  

  Create a WrappedMap object that will respond to the methods issymmetric, ishermitian, isposdef with the values provided by the keyword arguments, and to eltype with the value T. The default values correspond to the result of calling these methods on the argument A; in particular {T} does not need to be specified and is set as eltype(A). This allows to use an AbstractMatrix within the LinearMap framework and to redefine the properties of an existing LinearMap.

  LinearMap{T}(f, [fc = nothing], M::Int, [N::Int = M]; ismutating::Bool, issymmetric::Bool, ishermitian::Bool, isposdef::Bool])
  

  Create a FunctionMap instance that wraps an object describing the action of the linear map on a vector as a function call. Here, f can be a function or any other object for which either the call f(src::AbstractVector) -> dest::AbstractVector (when ismutating = false) or f(dest::AbstractVector,src::AbstractVector) -> dest (when ismutating = true) is supported. The value of ismutating can be spefified, by default its value is guessed by looking at the number of arguments of the first method in the method list of f.

  A second function or object can optionally be provided that implements the action of the adjoint (transposed) linear map. Here, it is always assumed that this represents the adjoint/conjugate transpose, though this is of course equivalent to the normal transpose for real linear maps. Furthermore, the conjugate transpose also enables the use of mul!(y, transpose(A), x) using some extra conjugation calls on the input and output vector. If no second function is provided, than mul!(y, transpose(A), x) and mul!(y, adjoint(A), x) cannot be used with this linear map, unless it is symmetric or hermitian.

  M is the number of rows (length of the output vectors) and N the number of columns (length of the input vectors). When the latter is not specified, N = M.

  Finally, one can specify the eltype of the resulting linear map using the type parameter T. If not specified, a default value of Float64 is assumed. Use a complex type T if the function represents a complex linear map.

  In summary, the keyword arguments and their default values are:

  • ismutating: false if the function f accepts a single vector argument corresponding to the input, and true if it accepts two vector arguments where the first will be mutated so as to contain the result. In both cases, the resulting A::FunctionMap will support both the mutating and non-mutating matrix vector multiplication. Default value is guessed based on the number of arguments for the first method in the method list of f; it is not possible to use f and fc where only one of the two is mutating and the other is not.
  • issymmetric [=false]: whether the function represents the multiplication with a symmetric matrix. If true, this will automatically enable A' * x and transpose(A) * x.
  • ishermitian [=T<:Real && issymmetric]: whether the function represents the multiplication with a hermitian matrix. If true, this will automatically enable A' * x and transpose(A) * x.
  • isposdef [=false]: whether the function represents the multiplication with a positive definite matrix.

• Array(A::LinearMap), Matrix(A::LinearMap), convert(Matrix, A::LinearMap) and convert(Array, A::LinearMap)

  Create a dense matrix representation of the LinearMap object, by multiplying it with the successive basis vectors. This is mostly for testing purposes or if you want to have the explicit matrix representation of a linear map for which you only have a function definition (e.g. to be able to use its transpose or adjoint). This way, one may conveniently make A act on the columns of a matrix X, instead of interpreting A * X as a composed linear map: Matrix(A * X). For generic code, that is supposed to handle both A::AbstractMatrix and A::LinearMap, it is recommended to use convert(Matrix, A*X).

• SparseArrays.sparse(A::LinearMap) and convert(SparseMatrixCSC, A::LinearMap)

  Create a sparse matrix representation of the LinearMap object, by multiplying it with the successive basis vectors. This is mostly for testing purposes or if you want to have the explicit sparse matrix representation of a linear map for which you only have a function definition (e.g. to be able to use its transpose or adjoint).

• The following matrix multiplication methods.

  • A * x: applies A to x and returns the result;
  • mul!(y::AbstractVector, A::LinearMap, x::AbstractVector): applies A to x and stores the result in y;
  • mul!(Y::AbstractMatrix, A::LinearMap, X::AbstractMatrix): applies A to each column of X and stores the results in the corresponding columns of Y;
  • mul!(y::AbstractVector, A::LinearMap, x::AbstractVector, α::Number=1, β::Number=0): computes α * A * x + β * y and stores the result in y. Analogously for X,Y::AbstractMatrix.

  Applying the adjoint or transpose of A (if defined) to x works exactly as in the usual matrix case: transpose(A) * x and mul!(y, A', x), for instance.

Types

None of the types below need to be constructed directly; they arise from performing operations between LinearMap objects or by calling the LinearMap constructor described above.

• LinearMap

  Abstract supertype

• FunctionMap

  Type for wrapping an arbitrary function that is supposed to implement the matrix vector product as a LinearMap.

• WrappedMap

  Type for wrapping an AbstractMatrix or LinearMap and to possible redefine the properties isreal, issymmetric, ishermitian and isposdef. An AbstractMatrix will automatically be converted to a WrappedMap when it is combined with other AbstractLinearMap objects via linear combination or composition (multiplication). Note that WrappedMap(mat1)*WrappedMap(mat2) will never evaluate mat1*mat2, since this is more costly then evaluating mat1*(mat2*x) and the latter is the only operation that needs to be performed by LinearMap objects anyway. While the cost of matrix addition is comparable to matrix vector multiplication, this too is not performed explicitly since this would require new storage of the same amount as of the original matrices.

• UniformScalingMap

  Type for representing a scalar multiple of the identity map (a.k.a. uniform scaling) of a certain size M=N, obtained simply as UniformScalingMap(λ, M). The type T of the resulting LinearMap object is inferred from the type of λ. A UniformScalingMap of the correct size will be automatically created if LinearMap objects are multiplied by scalars from the left or from the right, respecting the order of multiplication.

• LinearCombination, CompositeMap, TransposeMap and AdjointMap

  Used to add and multiply LinearMap objects, don't need to be constructed explicitly.

Example

The LinearMap object combines well with methods of the following packages:

• Arpack.jl: iterative eigensolver eigs and SVD svds;
• IterativeSolvers.jl: iterative solvers, eigensolvers, and SVD;
• TSVD.jl: truncated SVD tsvd.

using LinearMaps
import Arpack, IterativeSolvers, TSVD

function leftdiff!(y::AbstractVector, x::AbstractVector) # left difference assuming periodic boundary conditions
    N = length(x)
    length(y) == N || throw(DimensionMismatch())
    @inbounds for i=1:N
        y[i] = x[i] - x[mod1(i-1, N)]
    end
    return y
end

function mrightdiff!(y::AbstractVector, x::AbstractVector) # minus right difference
    N = length(x)
    length(y) == N || throw(DimensionMismatch())
    @inbounds for i=1:N
        y[i] = x[i] - x[mod1(i+1, N)]
    end
    return y
end

D = LinearMap(leftdiff!, mrightdiff!, 100; ismutating=true) # by default has eltype(D) = Float64

Arpack.eigs(D'D; nev=3, which=:SR) # note that D'D is recognized as symmetric => real eigenfact
Arpack.svds(D; nsv=3)

Σ, L = IterativeSolvers.svdl(D; nsv=3)

TSVD.tsvd(D, 3)