setfill(A::AbstractGBArray{T, F, N}, x::F2)

Create a new AbstractGBArray with the same underlying data but a new fill x. The fill type of A and the type of x may be different.

setfill!(A::AbstractGBArray{T, F, N}, x::F)

Modify the fill value of A. The fill type of A and the type of x must be the same.

sparsitystatus(A::AbstractGBArray)::AbstractSparsity

Return the current sparsity of A, which is one of Dense, Bitmap, Sparse, or Hypersparse.

format(A::AbstractGBArray) -> (s::AbstractSparsity, o::StorageOrders.StorageOrder)

Return the sparsity status and storage order of A as a tuple.

setstorageorder!(A::AbstractGBArray, o::StorageOrders.StorageOrder)

Set the storage order of A, either StorageOrders.RowMajor() or StorageOrders.ColMajor().

Users must call wait(A) before this will be reflected in A, however operations will perform this wait automatically on input.

gbset(A::GBArray, option, value)
gbset(option, value)

Set an option either for a specific GBArray, or globally. The commonly used options are: - :format = [RowMajor() | ColMajor()]: The global default or array specific column major or row major ordering. - :nthreads = [Integer]: The global number of OpenMP threads to use. - :burble = [Bool]: Print diagnostic output. - :sparsity_control = [:full | :bitmap | :sparse | :hypersparse]: Set the sparsity of a single GBArray.

Descriptor

Context object which may be optionally passed to many SuiteSparse:GraphBLAS functions.

See the SuiteSparse:GraphBLAS User Guide or the SuiteSparseGraphBLAS.jl docs for more information.

Options

• nthreads::Int = Sys.CPU_THREADS ÷ 2: Specify the maximum number of threads to be used by

a function, defaults to avoid hyperthreading, which is typically most performant.

• replace_output: Clear the output array before assignment.
• structural_mask::Bool: Utilize the structure of the mask argument, rather than its values.
• complement_mask::Bool: Values which are true in the complement of the mask will be kept.

set_lib!(path; export_prefs::Bool = false)

Set the shared library path for SuiteSparse:GraphBLAS. Set to "default" to use the provided artifact.

empty!(A::AbstractGBArray)

Clear all the entries from the GBArray. Does not modify the type or dimensions.

Complement{T}

The complement of a GraphBLAS mask. This wrapper will set the mask argument of a GraphBLAS operation to be the negation of the original mask.

It may be nested an arbitrary number of times.

Structural{T}

This wrapper will set a GraphBLAS mask to use the presence of values in the mask rather than their values to determine the mask.

xtype(op::GrBOp)::DataType

Determine type of the first argument to a typed operator.

ytype(op::GrBOp)::DataType

Determine type of the second argument to a typed operator.

ztype(op::GrBOp)::DataType

Determine type of the output of a typed operator.

gbrand(typeorrange, nrows, ncols, density; kwargs...)::GBMatrix
gbrand(rng::AbstractRNG, typeorrange, nrows, ncols, density; kwargs...)::GBMatrix

Construct a random GBMatrix, analogous to sprand from SparseArrays

Arguments

• rng::AbstractRNG: Random number generator for both values and indices.
• typeorrange: Either a type such as Float64, or a range such as 1:10.

Any input which supports eltype(typeorrange).

• nrows::Integer, ncols::Integer: Dimensions of the result.
• density::AbstractFloat: The approximate density of result.

Keywords

• symmetric::Bool: The result matrix is symmetric, Aᵀ = A.
• pattern::Bool: The result matrix consists solely of one(eltype(typeorrange)).
• skewsymmetric::Bool: The result matrix is skew-symmetric, Aᵀ = -A.
• hermitian::Bool: The result matrix is hermitian, aᵢⱼ = āⱼᵢ.
• nodiagonal::Bool: The result matrix has no values on the diagonal.

Returns

• GBMatrix