summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAge
* Drop unused variable.HEADmasterAnton Khirnov2019-08-16
|
* Sync lapse_prev[01] after writing them.Anton Khirnov2019-07-01
| | | | | | Prevents writing garbage sync zones to the output, which would break when recovering a checkpointed multi-component run into one with different decomposition.
* Do not try to free built-in MPI types.Anton Khirnov2019-06-29
|
* Use extrapolated solution as initial guess for outermost solve level.Anton Khirnov2019-06-27
|
* Make initial guess interpolation work with MPI.Anton Khirnov2019-06-27
|
* Add an option to extrapolate on the finest level as well.Anton Khirnov2019-06-23
|
* Parallelize the loops.Anton Khirnov2019-06-23
|
* Make the log callback atomic.Anton Khirnov2019-06-14
| | | | | Use a single call to write the entire line to stderr. Avoids mixed lines with e.g. multi-component runs
* Handle components that lie entirely in the buffer zone.Anton Khirnov2019-06-13
|
* Do not pass NULL to MPI as a datatype.Anton Khirnov2019-06-11
| | | | Use MPI_DATATYPE_NULL.
* Update for the new mg2d diff coeffs API.Anton Khirnov2019-06-10
| | | | Treat the PDE coeff non-regularities properly.
* Add support for MPI.Anton Khirnov2019-06-02
|
* Drop commented out obsolete code.Anton Khirnov2019-05-23
|
* Only allocate the context once.Anton Khirnov2019-05-23
|
* Print stats on terminateAnton Khirnov2019-05-23
|
* Make sure the initial guess conforms to the boundary condition.Anton Khirnov2019-04-19
|
* Use better initial guesses for solving.Anton Khirnov2019-04-07
|
* Print stats for the eval solversAnton Khirnov2019-04-07
|
* Avoid unnecessary solves on the finest reflevel.Anton Khirnov2019-04-07
|
* Use per-substep solvers on the finest level.Anton Khirnov2019-03-18
|
* Fix filling reflection-ghostzone boundary pointsAnton Khirnov2019-03-17
| | | | Reflection ghostzones are not filled on sync.
* Eliminate offset_right.Anton Khirnov2019-03-17
| | | | It is ill-defined and confusing.
* Add an option to scale residual tolerance with dx**-2Anton Khirnov2019-03-15
|
* Add the exact_size parameter.Anton Khirnov2019-03-15
|
* Make the parameters steerable.Anton Khirnov2019-03-15
|
* Remove unused context variable.Anton Khirnov2019-03-15
|
* Fill the corner ghost points for fine-level boundaries.Anton Khirnov2019-03-15
|
* Cosmetics, get rid of accidentally introduced tabs.Anton Khirnov2019-03-15
|
* Fix and enable 4th order phi derivatives for 4th order solver.Anton Khirnov2019-03-15
|
* Disable dissipation close to the boundary.Anton Khirnov2019-02-20
|
* Fix typo.Anton Khirnov2019-02-16
|
* Drop the use of the pseudospectral solver.Anton Khirnov2019-02-16
| | | | It does not produce good enough results.
* Stop filling the boundary values for coarsest level.Anton Khirnov2019-02-15
|
* Extrapolate the solution to fill coarsest-level outer ghostpoints.Anton Khirnov2019-02-15
|
* Adapt to mg2d boundary specification changes.Anton Khirnov2019-02-15
| | | | | | Rename FIXDIFF to REFLECT. Use the new FALLOFF BC on the physical boundary.
* Make the cfl factor used for relaxation runtime-configurable.Anton Khirnov2019-01-04
|
* Use half the coarsest grid size as the outer bound for the spectral solver.Anton Khirnov2019-01-03
| | | | This is only used for the history initialization.
* Make more solver parameters runtime-configurable.Anton Khirnov2018-12-28
|
* Set the number of solver threads from OMP_NUM_THREADS.Anton Khirnov2018-12-28
|
* Drop unused header.Anton Khirnov2018-12-28
|
* Make logging configurable through the parfile.Anton Khirnov2018-12-28
|
* Implement higher-order finite differences.Anton Khirnov2018-11-29
|
* Add timing benchmarks.Anton Khirnov2018-11-29
|
* Rewrite using a new method.Anton Khirnov2018-09-25
| | | | | Based on 'Adaptive mesh refinement for coupled elliptic-hyperbolic problems' by Pretorius and Choptuik, 2006.
* Specify maximum reflevel for interpolation.Anton Khirnov2018-08-07
|
* Initial commit.Anton Khirnov2018-08-07