#include <petsc/private/tsimpl.h> /*I "petscts.h"  I*/

const char *const TSConvergedReasons_Shifted[] = {"ADJOINT_DIVERGED_LINEAR_SOLVE", "FORWARD_DIVERGED_LINEAR_SOLVE", "DIVERGED_STEP_REJECTED", "DIVERGED_NONLINEAR_SOLVE", "CONVERGED_ITERATING", "CONVERGED_TIME", "CONVERGED_ITS", "CONVERGED_USER", "CONVERGED_EVENT", "CONVERGED_PSEUDO_FATOL", "CONVERGED_PSEUDO_FATOL", "TSConvergedReason", "TS_", NULL};
const char *const *TSConvergedReasons = TSConvergedReasons_Shifted + 4;

/*@
  TSCreate - This function creates an empty timestepper. The problem type can then be set with `TSSetProblemType()` and the
  type of solver can then be set with `TSSetType()`.

  Collective

  Input Parameter:
. comm - The communicator

  Output Parameter:
. ts - The `TS`

  Level: beginner

  Developer Notes:
  `TS` essentially always creates a `SNES` object even though explicit methods do not use it. This is
  unfortunate and should be fixed at some point. The flag snes->usessnes indicates if the
  particular method does use SNES and regulates if the information about the `SNES` is printed
  in `TSView()`. `TSSetFromOptions(`) does call `SNESSetFromOptions()` which can lead to users being confused
  by help messages about meaningless `SNES` options.

.seealso: [](ch_ts), `TS`, `SNES`, `TSSetType()`, `TSSetUp()`, `TSDestroy()`, `TSSetProblemType()`, `TSSetTimeStep()`
@*/
PetscErrorCode TSCreate(MPI_Comm comm, TS *ts)
{
  TS t;

  PetscFunctionBegin;
  PetscAssertPointer(ts, 2);
  PetscCall(TSInitializePackage());

  PetscCall(PetscHeaderCreate(t, TS_CLASSID, "TS", "Time stepping", "TS", comm, TSDestroy, TSView));

  /* General TS description */
  t->problem_type  = TS_NONLINEAR;
  t->equation_type = TS_EQ_UNSPECIFIED;

  t->ptime             = 0.0;
  t->time_step         = 0.1;
  t->initial_time_step = t->time_step;
  t->max_time          = PETSC_MAX_REAL;
  t->exact_final_time  = TS_EXACTFINALTIME_UNSPECIFIED;
  t->steps             = 0;
  PetscObjectParameterSetDefault(t, max_steps, PETSC_INT_MAX);
  PetscObjectParameterSetDefault(t, run_steps, PETSC_INT_MAX);
  t->steprestart = PETSC_TRUE;

  t->max_snes_failures = 1;
  t->max_reject        = 10;
  t->errorifstepfailed = PETSC_TRUE;

  t->rhsjacobian.time  = PETSC_MIN_REAL;
  t->rhsjacobian.scale = 1.0;
  t->ijacobian.shift   = 1.0;

  /* All methods that do adaptivity should specify
   * its preferred adapt type in their constructor */
  t->default_adapt_type = TSADAPTNONE;
  t->atol               = 1e-4;
  t->rtol               = 1e-4;
  t->cfltime            = PETSC_MAX_REAL;
  t->cfltime_local      = PETSC_MAX_REAL;

  t->num_rhs_splits = 0;

  t->axpy_pattern = UNKNOWN_NONZERO_PATTERN;
  *ts             = t;
  PetscFunctionReturn(PETSC_SUCCESS);
}