Internal API

reinit_buffer!(cb::CellBuffer, sim::Simulation, coupled, cellnum::Int)

Reinitialize the cb::CellBuffer for cell number cellnum. The global degree of freedom vectors a (current) and aold are used to update the cell degree of freedom vectors in c. If the global vectors are not included in sim, the corresponding local vectors are set to NaN The element stiffness, cb.Ke, and residual, cb.re, are also zeroed.

create_states(sdh::SubDofHandler, material, cellvalues, a, cellset, dofrange)

Create a Dict of states for the cells in cellset, where the user should define the create_cell_state function for their material (and corresponding cellvalues) dofrange::NamedTuple is passed onto create_cell_state and contains the local dof ranges for each field.

_copydofs!(edofs::Vector, gdofs::Vector, inds::Vector{Int})

Internal function for faster copying of global values into the element values. Equivalent to edofs .= gdofs[inds]

_copydofs!(edofs::Vector, gdofs::Nothing, inds::Vector{Int})

Fill edofs with NaN

_create_cell_state(cell, material, cellvalues, a, ae, dofrange, cellnr)

Internal function to reinit and extract the relevant quantities from the cell::CellCache, reinit cellvalues, update ae from a, and pass these into the create_cell_state function that the user should specify.

CellBuffer(
    numdofs::Int, numnodes::Int, ::Val{sdim}, 
    cellvalues, material, state, dofrange, user_data=nothing) -> CellBuffer

Create a cell cache for an element with numdofs degrees of freedom and numnodes nodes with dimension sdim. cellvalues are reinit!ed for each cell, and the state is updated to the old cell state. material will be passed as-is to the element. The given dofrange::NamedTuple, user_data::Any, and cache::Any are available to the element via the buffer input.

AutoDiffCellBuffer(cb::CellBuffer)

FacetBuffer(
    numdofs::Int, numnodes::Int, ::Val{sdim}, 
    facetvalues, material, state, dofrange, user_data=nothing)

Create a facetbuffer for an element with numdofs degrees of freedom and numnodes nodes with dimension sdim. facetvalues are reinit!ed for each cell, and the state is updated to the old cell state. material will be passed as-is to the element. The given dofrange::NamedTuple, user_data::Any, and cache::Any are available to the element via the buffer input.

get_itembuffer(dbs::Dict{String,AbstractDomainBuffer}, domain::String)
get_itembuffer(db::AbstractDomainBuffer)
get_itembuffer(sim::Simulation[, domain::String])

Get the AbstractItemBuffer stored in db or dbs[domain]. This internal function might change, but currently the full TaskLocals is returned for a ThreadedDomainBuffer (used internally).

skip_this_domain(worker, name::String)

Should the domain with key name be skipped during work? Defaults to false. Can be used to e.g. only loop over parts of a domain.

can_thread(worker)::Bool

Does the worker support multithreaded work? Defaults to false. If this returns true, the worker must support the TaskLocals interface.

fast_getindex(collection)

If the output is known from the element type of the collection, this can be returned directly. Currently, this is implemented for AbstractDict with Nothing as type, which takes away overhead when state variables are not used.

work_single_cell!(worker, cellbuffer)

Each worker that supports a cellbuffer should overload this function.

work_single_facet!(worker, facetbuffer)

Each worker that supports for a facetbuffer should overload this function.

autogenerate_cellvalues(cv::AbstractCellValues, args...)

Just return the provided CellValues

autogenerate_cellvalues(order::Int, ip_fun::Interpolation{RefShape}, ip_geo::Interpolation{RefShape})

Using quadrature rule, qr = QuadratureRule{RefShape}(order), return CellValues(qr, ip_fun, ip_geo)

autogenerate_facetvalues(fv::AbstractFacetValues, args...)

Just return the provided FacetValues

autogenerate_facetvalues(order::Int, ip_fun::Interpolation{RefShape}, ip_geo::Interpolation{RefShape})

Using quadrature rule, fqr = FacetQuadratureRule{RefShape}(order), create FacetValues(fqr, ip_fun, ip_geo)

TaskChunks(chunks)

This is strictly an internal implementation detail for now. It is used to provide a variant of Base.Channel, but where data is not removed, the index is just incremented.