Lines Matching refs:methods

26 overview of the theory of such methods.
29 iterative methods; the `PC` module, described in
54 particular, `KSP` *does* support matrix-free methods. The routine
56 provides further information regarding matrix-free methods. Typically,
168 The Krylov subspace methods accept a number of options, many of which
186 GMRES methods:
260 Since the rate of convergence of Krylov projection methods for a
290 some methods by using the options database command
456 convergence testing of all left-preconditioned `KSP` methods. For the
457 conjugate gradient, Richardson, and Chebyshev methods the true residual
579 Since the convergence of Krylov subspace methods depends strongly on the
620 Standard Krylov methods require that the preconditioner be a linear operator, thus, for example, a …
622 Flexible Krylov methods are a subset of methods that allow (with modest additional requirements
624 The flexible `KSP` methods have the label "Flexible" in {any}`tab-kspdefaults`.
628 In addition to supporting `PCKSP`, the flexible methods support `KSPFlexibleSetModifyPC()` to
639 Standard Krylov methods have one or more global reductions resulting from the computations of inner…
642 methods overlap the reduction operations with local computations (generally the application of the …
645 The pipeline `KSP` methods have the label "Pipeline" in {any}`tab-kspdefaults`.
648 performance of pipelined methods. See {any}`doc_faq_pipelined` for details.
671 One should not destroy this vector. For certain `KSP` methods (e.g.,
681 Again, for GMRES and certain other methods this is an expensive
688 As discussed in {any}`sec_ksppc`, Krylov subspace methods
699 preconditioning methods supported in PETSc. See the `PCType` manual
711 PETSc provides several domain decomposition methods/preconditioners including
914 The block Jacobi and overlapping additive Schwarz (domain decomposition) methods in PETSc are
917 implementations of these methods employ ILU(0) factorization on each
1089 …TSc provides a fully supported (smoothed) aggregation AMG, but supports the addition of new methods
1191 AMG methods require knowledge of the number of degrees of freedom per
1281 **Troubleshooting algebraic multigrid methods:** If `PCGAMG`, *ML*, *AMGx* or
1283 methods. Often, the default parameters or just the strengths of different
1286 of AMG solvers and often special purpose methods must be developed to
1304 **I am converging slowly, what do I do?** AMG methods are sensitive to
1305 coarsening rates and methods; for GAMG use `-pc_gamg_threshold <x>`
1377 easily add AMG capabilities, like new AMG methods or an AMG component
1520 Unlike conventional non-overlapping methods that iterates just on the
1585 `PCBDDC` methods is still an active topic of research, its implementation is
1796 `PCMG` uses only one level! This is different from the algebraic multigrid methods
1993 the methods described in this section because there are fewer copies and
1995 components, but it is not possible to use many other methods with