ISSUE 3763: Fix default solver options in documentation #4254
Conversation
clean.rst is the solving-interface.rst
| @@ -257,7 +257,7 @@ definite, and therefore can choose the conjugate gradient method, | |||
| rather than GMRES. | |||
|
|
|||
| .. code-block:: python3 | |||
|
|
|||
There was a problem hiding this comment.
no need for extra whitespace
docs/source/solving-interface.rst
Outdated
| solver_parameters: { | ||
| * ``mat_type``: "aij" | ||
| * ``ksp_type``: "preonly" | ||
| * ``ksp_rtol``: 1e-7 | ||
| * ``pc_type``: "lu" | ||
| * ``pc_factor``: _DEFAULT_DIRECT_SOLVER_PARAMETERS | ||
| } | ||
|
|
||
| `LinearVariationalSolver defaults <https://github.com/firedrakeproject/firedrake/blob/master/firedrake/variational_solver.py#L134-L143>`_ |
There was a problem hiding this comment.
This is just supposed to be a bulleted list. Please run make -C docs html and check the rendered HTML to make sure things look right.
tcalderon-space
changed the title
Co-authored-by: Connor Ward <[email protected]>
…e/firedrake into tcalderon-space/issue3763 updating solving-interface.rst after feedback
| * ``ksp_atol``: ``1e-50`` | ||
| * ``ksp_divtol``: ``1e4`` | ||
| * ``ksp_max_it``: ``10000`` |
There was a problem hiding this comment.
Also, please make sure the rest of the file is up to date. The options in 753-763 should have:
{'mat_type': 'aij',
'ksp_type': 'preonly',
'ksp_rtol': 1e-05,
'pc_type': 'lu',
'pc_factor_mat_solver_type': 'mumps',
'pc_factor_mat_mumps_icntl_14': 200,
'snes_type': 'newtonls',
'snes_linesearch_type': 'basic'}
clean.rst
Outdated
There was a problem hiding this comment.
Please remove this file from the PR.
docs/source/solving-interface.rst
Outdated
| * 'pc_factor_mat_solver_type': 'mumps' | ||
| * 'pc_factor_mat_mumps_icntl_14': 200 |
There was a problem hiding this comment.
| * 'pc_factor_mat_solver_type': 'mumps' | |
| * 'pc_factor_mat_mumps_icntl_14': 200 | |
| * ``pc_factor_mat_solver_type``: ``mumps`` | |
| * ``pc_factor_mat_mumps_icntl_14``: 200 |
…e/firedrake into tcalderon-space/issue3763
| * ``pc_factor_mat_solver_type`` : ``mumps`` | ||
| * ``pc_factor_mat_mumps_icntl_14``: 200 |
There was a problem hiding this comment.
These parameters are only relevant if MUMPS is present. Do we want to keep them in the documentation? If so, they should also be included above in the linear solver parameters.
There was a problem hiding this comment.
Should I wait for approval for whether I should update mumps or not? @pbrubeck @connorjward
There was a problem hiding this comment.
You have to try quite hard to not have MUMPS installed. I think these options should be documented.
Description
Fixes Issue #3763 Transfered the file from clean.rst to solving-interfaces.rst