Matrix functionality

AbstractAlgebra.jl provides a module, implemented in src/Matrix.jl for matrices over any ring belonging to the AbstractAlgebra abstract type hierarchy. This functionality will work for any matrix type which follows the Matrix interface.

Similarly, AbstractAlgebra.jl provides a module in src/MatRing.jl for matrix algebras over a ring.

Generic matrix types

AbstractAlgebra.jl allows the creation of dense matrices over any computable ring $R$. Generic matrices over a ring are implemented in src/generic/Matrix.jl.

Generic matrix rings of $m\times m$ matrices are implemented in src/generic/MatRing.jl.

Generic matrices in AbstractAlgebra.jl have type Generic.MatSpaceElem{T} for matrices in a matrix space, or Generic.MatRingElem{T} for matrices in a matrix algebra, where T is the type of elements of the matrix. Internally, generic matrices are implemented using an object wrapping a Julia two dimensional array, though they are not themselves Julia arrays. See the file src/generic/GenericTypes.jl for details.

For the most part, one doesn't want to work directly with the MatSpaceElem type though, but with an abstract type called Generic.Mat which includes MatSpaceElem and views thereof.

Parents of generic matrices (matrix spaces) have type MatSpace{T}. Parents of matrices in a matrix algebra have type Generic.MatRing{T}.

The dimensions and base ring $R$ of a generic matrix are stored in its parent object, however to allow creation of matrices without first creating the matrix space parent, generic matrices in Julia do not contain a reference to their parent. They contain the row and column numbers (or degree, in the case of matrix algebras) and the base ring on a per matrix basis. The parent object can then be reconstructed from this data on demand.

Abstract types

The generic matrix types (matrix spaces) belong to the abstract type MatElem{T} and the all matrix space parents are of the concrete type MatSpace{T}. On the other hand, the generic matrix algebra matrix types belong to the abstract type MatRingElem{T} and the parent types belong to the abstract MatRing{T} Note that both the concrete type of a matrix space parent object and the abstract class it belongs to have the name MatElem, therefore disambiguation is required to specify which is intended. The same is true for the abstract types for matrix spaces and their elements.

Matrix space constructors

A matrix space in AbstractAlgebra.jl represents a collection of all matrices with given dimensions and base ring.

In order to construct matrices in AbstractAlgebra.jl, one can first construct the matrix space itself. This is accomplished with the following constructor. We discuss creation of matrix algebras separately in a dedicated section elsewhere in the documentation.

matrix_space(R::Ring, rows::Int, cols::Int)

Construct the space of matrices with the given number of rows and columns over the given base ring.

Here are some examples of creating matrix spaces and making use of the resulting parent objects to coerce various elements into the matrix space.

Examples

julia> R, t = polynomial_ring(QQ, :t)
(Univariate polynomial ring in t over rationals, t)

julia> S = matrix_space(R, 3, 3)
Matrix space of 3 rows and 3 columns
  over univariate polynomial ring in t over rationals

julia> A = S()
[0   0   0]
[0   0   0]
[0   0   0]

julia> B = S(12)
[12    0    0]
[ 0   12    0]
[ 0    0   12]

julia> C = S(R(11))
[11    0    0]
[ 0   11    0]
[ 0    0   11]

Matrix element constructors

There are a few ways to construct matrices other than by coercing elements as shown above. The first method is from an array of elements.

This can be done with either two or one dimensional arrays.

(S::MatSpace{T})(A::Matrix{S}) where {S <: RingElement, T <: RingElement}
(S::MatRing{T})(A::Matrix{S}) where {S <: RingElement, T <: RingElement}

Create the matrix in the given space/algebra whose $(i, j)$ entry is given by A[i, j], where S is the type of elements that can be coerced into the base ring of the matrix.

(S::MyMatSpace{T})(A::Vector{S}) where {S <: RingElem, T <: RingElem}
(S::MyMatAlgebra{T})(A::Vector{S}) where {S <: RingElem, T <: RingElem}

Create the matrix in the given space/algebra of matrices (with dimensions $m\times n$ say), whose $(i, j)$ entry is given by A[i*(n - 1) + j] and where S is the type of elements that can be coerced into the base ring of the matrix.

We also provide the following syntax for constructing literal matrices (similar to how Julia arrays can be be constructed).

R[a b c...;...]

Create the matrix over the base ring $R$ consisting of the given rows (separated by semicolons). Each entry is coerced into $R$ automatically. Note that parentheses may be placed around individual entries if the lists would otherwise be ambiguous, e.g. R[1 2; 2 (- 3)].

Also see the Matrix interface for a list of other ways to create matrices.

Examples

julia> S = matrix_space(QQ, 2, 3)
Matrix space of 2 rows and 3 columns
  over rationals

julia> T = matrix_ring(QQ, 2)
Matrix ring of degree 2
  over rationals

julia> M1 = S(Rational{BigInt}[2 3 1; 1 0 4])
[2//1   3//1   1//1]
[1//1   0//1   4//1]

julia> M2 = S(BigInt[2 3 1; 1 0 4])
[2//1   3//1   1//1]
[1//1   0//1   4//1]

julia> M3 = S(BigInt[2, 3, 1, 1, 0, 4])
[2//1   3//1   1//1]
[1//1   0//1   4//1]

julia> N1 = T(Rational{BigInt}[2 3; 1 0])
[2//1   3//1]
[1//1   0//1]

julia> N2 = T(BigInt[2 3; 1 0])
[2//1   3//1]
[1//1   0//1]

julia> N3 = T(BigInt[2, 3, 1, 1])
[2//1   3//1]
[1//1   1//1]

julia> R, t = polynomial_ring(QQ, :t)
(Univariate polynomial ring in t over rationals, t)

julia> S = matrix_space(R, 3, 3)
Matrix space of 3 rows and 3 columns
  over univariate polynomial ring in t over rationals

julia> M = R[t + 1 1; t^2 0]
[t + 1   1]
[  t^2   0]

julia> N = R[t + 1 2 t] # create a row vector
[t + 1   2   t]

julia> P = R[1; 2; t] # create a column vector
[1]
[2]
[t]

It is also possible to create matrices (in a matrix space only) directly, without first creating the corresponding matrix space (the inner constructor being called directly).

matrix(R::Ring, arr::Matrix{T}) where T <: RingElement

Given an $m\times n$ Julia matrix of entries, construct the corresponding AbstractAlgebra.jl matrix over the given ring R, assuming all the entries can be coerced into R.

matrix(R::Ring, r::Int, c::Int, A::Vector{T}) where T <: RingElement

Construct the given $r\times c$ AbstractAlgebra.jl matrix over the ring R whose $(i, j)$ entry is given by A[c*(i - 1) + j], assuming that all the entries can be coerced into R.

zero_matrix(R::Ring, r::Int, c::Int)

Construct the $r\times c$ AbstractAlgebra.jl zero matrix over the ring R.

Examples

julia> M = matrix(ZZ, BigInt[3 1 2; 2 0 1])
[3   1   2]
[2   0   1]

julia> N = matrix(ZZ, 3, 2, BigInt[3, 1, 2, 2, 0, 1])
[3   1]
[2   2]
[0   1]

julia> P = zero_matrix(ZZ, 3, 2)
[0   0]
[0   0]
[0   0]

julia> R = matrix_ring(ZZ, 2)
Matrix ring of degree 2
  over integers

julia> M = R()
[0   0]
[0   0]

Block diagonal matrix constructors

It is also possible to create block diagonal matrices from a vector of existing matrices. It is also possible to construct them from Julia matrices if one supplies the base ring.

Note that if the input matrices are not square, the output matrix may not be square.

block_diagonal_matrix — Method

block_diagonal_matrix(V::Vector{<:MatElem{T}}) where T <: NCRingElement

Create the block diagonal matrix whose blocks are given by the matrices in V. There must be at least one matrix in V.