Merge branch 'rmills/doc-misc-small-fixes' into 'master'Rmills/doc misc small fixesSee merge request petsc/petsc!2070
Copy edits to SNESNASM manual page.
Capitalized "Jacobian" in SNESNASMSetComputeFinalJacobian() manual page.
Small copy edits to SNESASPIN manual page.
Knepley/is separate section
Documentation and trivial code improvements for SNES GS and its secant computationUse tolerances depending on compiler precisionUse Read versions of VecGetArray when possibleCall VecGetArray at s
Documentation and trivial code improvements for SNES GS and its secant computationUse tolerances depending on compiler precisionUse Read versions of VecGetArray when possibleCall VecGetArray at suitable times so that vector operations are not done on it while the array is activeMinor improvements to documentation for SNES GSCommit-type: bug-fix, documentation
show more ...
Fixed error in example, missing one of the changed arguments; fix docs and add docsCommit-type: testing-fix, documentationReported-by: gitlab-ci
Moved update of vec_sol_update to after Pre/PostCheck()
Refactored NewtonTR to include a PreCheck() and vector naming convention similar to LineSearch (i.e. added Vec W).
Update public API, and internal naming conventions for DM to eliminate use of "default" (previously refered to default sections)Commit-type: style-fix, documentation
Ugly typos.
Fix docs for SNESNCG, add NCG linesearch that was in the code and not connected, call SNESSetFromOptions() on NPCCommit-type: feature, documentation
Merge branch 'jczhang/sf-more-opts' [PR #1567]* jczhang/sf-more-opts:Add more optimizations in SF and use it as the default for VecScatter.
Fix a bunch of PetscValidPointer() -> PetscValidXXXPointer()
Do not scatter to the same vector at the same timeIn vecscatter, the target vector is write-locked
Merge branch 'dalcinl/fas' [PR #1879]* dalcinl/fas:SNESFAS: Minor refactors and fixes and support in TSBDF
Merge branch 'barry/9-7/2019/SNES_CONVERGED_TR_DELTA-Error' [PR #1857]* barry/9-7/2019/SNES_CONVERGED_TR_DELTA-Error:Make SNES_CONVERGED_TR_DELTA negative to indicate the SNES did not converge.
SNESMS: Minor fixes* Add missing call to KSPSetOperators()* Refactor duplicated norm code to use a utility routine* Fix SNESReset_MS() to actually remove composed function
SNESFAS: Minor refactor* Remove workarounds in SNESFASSetLevels() related to SNESReset()* Perform proper error checking in the many SNEFAS-specific routines* Fix code duplication in the setup of
SNESFAS: Minor refactor* Remove workarounds in SNESFASSetLevels() related to SNESReset()* Perform proper error checking in the many SNEFAS-specific routines* Fix code duplication in the setup of smoothers and line searches* Replace sprintf -> PetscStrncpy/PetscSNPrintf
Merged in barry/newtontr-post-check (pull request #1828)Add SNESNewtonTRSetPostCheck() that sets a function that is called immediately before the next function SNES function evaluation. This al
Merged in barry/newtontr-post-check (pull request #1828)Add SNESNewtonTRSetPostCheck() that sets a function that is called immediately before the next function SNES function evaluation. This allows one to potentially change the updated solution or log properties of the update for a later convergence te
Merge branch 'gropp/doc-fixes-1' [PR #1838]
Add SNESNewtonTRSetPostCheck() that sets a function that is called immediately before the next function SNES function evaluation. This allows one to potentially change the updated solution or
Add SNESNewtonTRSetPostCheck() that sets a function that is called immediately before the next function SNES function evaluation. This allows one to potentially change the updated solution or log properties of the update for a later convergence test Commit-type: feature Reported-by: Glenn E. Hammond <gehammo@sandia.gov> Reported-by: Heeho Park <heepark@sandia.gov>
Make SNES_CONVERGED_TR_DELTA negative to indicate the SNES did not converge.I cannot explain why previously this was handled as converged since the trust region going tozero doesn't seem to corres
Make SNES_CONVERGED_TR_DELTA negative to indicate the SNES did not converge.I cannot explain why previously this was handled as converged since the trust region going tozero doesn't seem to correspond with convergence of the nonlinear solverCommit-type: bug-fixReported-by: Heeho Park <heepark@sandia.gov>
DM: Simplify implementation of DMHasCreateInjection()* Add log event for DMCreateInjection()* Get rid of DMHasCreateInjection_Default()* Rename dm->ops->getinjection to dm->ops->createinjection*
DM: Simplify implementation of DMHasCreateInjection()* Add log event for DMCreateInjection()* Get rid of DMHasCreateInjection_Default()* Rename dm->ops->getinjection to dm->ops->createinjection* Improve error checking here and there
Merge from doc-fixes branch
1...<<11121314151617181920>>...73