Tensor types

The fixed-size, three-dimensional vector and tensor classes MUESLI uses throughout. All of them support 1-based indexing, slicing, iteration and convert to ordinary Julia arrays; see Tensors and vectors for worked examples.

Types

MuesliMaterials.Ivector — Type
Ivector(v::AbstractVector)
Ivector(a, b, c)

A three-component vector, MUESLI's ivector.

Supports 1-based indexing, iteration and convert(Vector{Float64}, v).

Examples

v = Ivector([1.0, 2.0, 3.0])v[2]                              # 2.0convert(Vector{Float64}, v)
source
MuesliMaterials.Itensor — Type
Itensor(m::AbstractMatrix)
Itensor(a::AbstractVector, b::AbstractVector, c::AbstractVector)
Itensor(a00, a01, a02, a10, a11, a12, a20, a21, a22)

A general second-order tensor in three dimensions, MUESLI's itensor.

Details

The matrix constructor takes a 3 × 3 Julia matrix. Julia stores matrices column-major while MUESLI's constructor is row-major, so the components are reordered on the way in — the round trip is faithful and nothing is transposed. Any other shape raises an error.

Behaves like a 3 × 3 array: indexing and slicing (both 1-based), axes, length, ndims, iteration, and convert to Matrix{Float64}. copyto!(A, T) and copyto!(T, A) move a whole tensor in one call and do not allocate for Float64 arrays, which is what to use in a hot loop. With Tensors.jl loaded it also converts to Tensor{2, 3}.

Finite strain material points take the deformation gradient as an Itensor, since it is not symmetric.

Examples

F = Itensor([1.05 0.02 0.0; 0.0 0.98 0.01; 0.0 0.0 1.01])F[1, 2]                           # 0.02F[1, :]                           # first rowconvert(Matrix{Float64}, F)A = zeros(3, 3)copyto!(A, F)                     # no allocationcopyto!(F, 2 .* A)                # overwrite F in place

See Also

source
MuesliMaterials.Istensor — Type
Istensor(m::AbstractMatrix)
Istensor(t00, t11, t22, t12, t20, t01)

A symmetric second-order tensor in three dimensions, MUESLI's istensor.

Details

Only six components are stored. The matrix constructor reads the symmetric part of the 3 × 3 matrix it is given, so a non-symmetric input is symmetrised rather than rejected.

Istensor is a subtype of Itensor on the C++ side, so it supports the same array interface, and converts to SymmetricTensor{2, 3} when Tensors.jl is loaded.

Small strain material points take the strain and return stresses as Istensor; in the finite strain setting the second Piola-Kirchhoff, Kirchhoff and Cauchy stresses are Istensor while the first Piola-Kirchhoff stress is a general Itensor.

Examples

ε = Istensor([0.001 0.0002 0.0; 0.0002 -0.0005 0.0001; 0.0 0.0001 0.0003])σ = Istensor()                    # empty, to be filled by a bang method
source
MuesliMaterials.Itensor4 — Type
Itensor4(a::AbstractArray{<:Real, 4})
Itensor4()

A fourth-order tensor in three dimensions, MUESLI's itensor4.

Details

The constructor takes a 3 × 3 × 3 × 3 array; anything else raises an error. The default constructor gives an uninitialised tensor to be filled by a bang method such as tangentTensor!.

Indexing is 1-based in all four slots.

Examples

ℂ = Itensor4()tangentTensor!(mp, ℂ)ℂ[1, 1, 2, 2]
source

Array interface

Each type implements enough of the AbstractArray interface to be used like one:

  • getindex / setindex!, including UnitRange slices
  • axes, ndims, length, similar
  • iterate, so sum, collect and comprehensions work
  • copyto! to and from a preallocated Vector/Matrix/Array{Float64, 4}, in one call and without allocating (other element types are converted first)
  • convert to Matrix{Float64} / Array{Float64}, and — with Tensors.jl loaded — to Tensor{2,3} and SymmetricTensor{2,3}

Collections of tensors

Some MUESLI routines take a list of tensors, for instance the relaxation strains of a viscoelastic material.

MuesliMaterials.ArrayOfITensors — Type
ArrayOfITensors()

A growable list of Itensor values, for MUESLI routines that take several tensors at once.

Use push! to append a 3 × 3 Julia matrix, clear! to empty it and size for the count.

source
MuesliMaterials.ArrayOfIsTensors — Type
ArrayOfIsTensors()

A growable list of Istensor values, for MUESLI routines that take several symmetric tensors at once — the per-branch relaxation strains of a viscoelastic material, for instance.

Use push! to append a 3 × 3 Julia matrix, clear! to empty it and size for the count.

source