fluids: Add TSPostStep for SGS DD Training

- This will shutdown the simulation if there's a tensor called
`check-run` equal to zeros

fluids: Add data-driven SGS training setup, TSMonitor

- Adds `smartsim_weak.c` to handle case where SmartRedis is not built

fluids: Add SmartSim setup

- Adds `-smartsim_diable_calls` which can run with `-sgs_train_enable`,
but doesn't call SmartSim library calls. This allows for debugging the
training related functio

fluids: Add SmartSim setup

- Adds `-smartsim_diable_calls` which can run with `-sgs_train_enable`,
but doesn't call SmartSim library calls. This allows for debugging the
training related functions without having to run inside a SmartSim
experiment.

show more ...

fluids: Store diff filter field ids

fluids: Agnosticize VeloGradient projection setup

- It can now setup more than just user->grad_velo_proj

fluids: Cleanup unused struct members

Many of these are set, but never actually used.

Plex project coord renamed (#1391)

* fixing ProjectCoordinates

* copy of Ratel fix

* set for future, preserve backward

* break out versions

* PETSc is actually at 3.20 not 3.21 so decrem

Plex project coord renamed (#1391)

* fixing ProjectCoordinates

* copy of Ratel fix

* set for future, preserve backward

* break out versions

* PETSc is actually at 3.20 not 3.21 so decremment the minor version switch

* fixed style and copied same to solids and petsc examples (thought they have lots of other compile warnings)

* blank line

* Breaking again with compounded versions

* Update remaining DMPlexProjectCoordinates

* ex - update DMPlexCreateCoordinateSpace

* cx - update ICC/IFORT job

---------

Co-authored-by: Jed Brown <jed@jedbrown.org>
Co-authored-by: Jeremy L Thompson <jeremy@jeremylt.org>

show more ...

Drop Coordinate Basis from IJacobian (without breaking IDL) (#1380)

* compute IDL coefficient in IFunction that has access to coordinates, append this to jac_data so that IJacobian can avoid pulling

Drop Coordinate Basis from IJacobian (without breaking IDL) (#1380)

* compute IDL coefficient in IFunction that has access to coordinates, append this to jac_data so that IJacobian can avoid pulling coordinates through basis to apply IDL

Co-authored-by: James Wright <james@jameswright.xyz>

show more ...

fluids: Don't set ts_max_steps if not TESTTYPE_NONE

fix(fluids): Don't use IntArrayC2P in diff filter

IntArrayC2P frees the input array, when I need both the `PetscInt` array
and the `CeedInt` array.

Merge pull request #1349 from CEED/jrwrigh/remove_ctx_update

fluids: Use normal `CeedOperatorSetContextDouble` instead of `UpdateContextLabel`

fluids: Use normal CeedOperatorSetContextDouble

That instead of `UpdateContextLabel`, as #1347 built that functionality
into the normal label setting functions themselves

fluids: Fix 64bit Int typing issues

Add CeedOperatorLinearAssemblePointBlockDiagonalSymbolic

interface - rename CEED_BASIS_COLLOCATED to CEED_BASIS_NONE

fixup! fluids: Add DMPlexCeedElemRestriction*Create

fluids: Document and clarify spanstats restriction creation

- The restriction creation for the spanwise statistics is a pretty
special case, so better name the functions and document what exactl

fluids: Document and clarify spanstats restriction creation

- The restriction creation for the spanwise statistics is a pretty
special case, so better name the functions and document what exactly
it is doing

show more ...

fluids: Add DMPlexCeedElemRestriction*Create

Replaces CreateRestrictionFromPlex and GetRestrictionForDomain

fluids: Misc changes to functions

fluids: Rename files with *_NS suffix

fluids: Correct PetscFunction{Return,BeginUser} placement

- Respect the PETSc standard for style (https://petsc.org/release/developers/style/#coding-conventions-and-style)

fluids: Remove misc comments

- Either by inaccuracy of unnecessary

fluids: Make function names in consistent format

- Mainly remove `_` and acronymns with more than 2 letters should
respect CamelCase (ie. `StgFunction` not `STGFunction`)

fluids: Fix preloading vector

- Preloading should use Q_preload, not Q. Otherwise the initial
condition for the "actual" solve is changed by the preload step

fluids: Add DMSetupByOrder_* helper function