@@ -861,9 +861,11 @@ patterns with “more structure”.
861
861
!!! note
862
862
863
863
By default, the SparseArrays.jl are implemented for efficiency by caching the
864
- symbolic factorization. I.e., if `set_A` is used, it is expected that the new
865
- `A` has the same sparsity pattern as the previous `A`. If this algorithm is to
866
- be used in a context where that assumption does not hold, set `reuse_symbolic=false`.
864
+ symbolic factorization. If the sparsity pattern of `A` may change between solves, set `reuse_symbolic=false`.
865
+ If the pattern is assumed or known to be constant, set `reuse_symbolic=true` to avoid
866
+ unnecessary recomputation. To further reduce computational overhead, you can disable
867
+ pattern checks entirely by setting `check_pattern = false`. Note that this may error
868
+ if the sparsity pattern does change unexpectedly.
867
869
"""
868
870
Base. @kwdef struct UMFPACKFactorization <: AbstractSparseFactorization
869
871
reuse_symbolic:: Bool = true
@@ -885,9 +887,11 @@ A fast sparse LU-factorization which specializes on sparsity patterns with “le
885
887
!!! note
886
888
887
889
By default, the SparseArrays.jl are implemented for efficiency by caching the
888
- symbolic factorization. I.e., if `set_A` is used, it is expected that the new
889
- `A` has the same sparsity pattern as the previous `A`. If this algorithm is to
890
- be used in a context where that assumption does not hold, set `reuse_symbolic=false`.
890
+ symbolic factorization. If the sparsity pattern of `A` may change between solves, set `reuse_symbolic=false`.
891
+ If the pattern is assumed or known to be constant, set `reuse_symbolic=true` to avoid
892
+ unnecessary recomputation. To further reduce computational overhead, you can disable
893
+ pattern checks entirely by setting `check_pattern = false`. Note that this may error
894
+ if the sparsity pattern does change unexpectedly.
891
895
"""
892
896
Base. @kwdef struct KLUFactorization <: AbstractSparseFactorization
893
897
reuse_symbolic:: Bool = true
0 commit comments