:py:mod:`tlm_adjoint.fenics.caches`
===================================

.. py:module:: tlm_adjoint.fenics.caches

.. autoapi-nested-parse::

   This module implements finite element assembly and linear solver data
   caching.

   ..
       !! processed by numpydoc !!


Module Contents
---------------

.. py:class:: AssemblyCache




   
   A :class:`.Cache` for finite element assembly data.
















   ..
       !! processed by numpydoc !!
   .. py:method:: assemble(form, *, bcs=None, form_compiler_parameters=None, linear_solver_parameters=None, replace_map=None)

      
      Perform finite element assembly and cache the result, or return a
      previously cached result.

      :arg form: The :class:`ufl.Form` to assemble.
      :arg bcs: Dirichlet boundary conditions.
      :arg form_compiler_parameters: Form compiler parameters.
      :arg linear_solver_parameters: Linear solver parameters. Required for
          assembly parameters which appear in the linear solver parameters.
      :arg replace_map: A :class:`Mapping` defining a map from symbolic
          variables to values.
      :returns: A :class:`tuple` `(value_ref, value)`, where `value` is the
          result of the finite element assembly, and `value_ref` is a
          :class:`.CacheRef` storing a reference to `value`.

              - For an arity zero or arity one form `value_ref` stores the
                assembled value.
              - For an arity two form `value_ref` is a tuple `(A, b_bc)`. `A`
                is the assembled matrix, and `b_bc` is a boundary condition
                right-hand-side term which should be added after assembling a
                right-hand-side with homogeneous boundary conditions applied.
                `b_bc` may be `None` to indicate that this term is zero.















      ..
          !! processed by numpydoc !!


.. py:class:: LinearSolverCache




   
   A :class:`.Cache` for linear solver data.
















   ..
       !! processed by numpydoc !!
   .. py:method:: linear_solver(form, *, bcs=None, form_compiler_parameters=None, linear_solver_parameters=None, replace_map=None, assembly_cache=None)

      
      Construct a linear solver and cache the result, or return a
      previously cached result.

      :arg form: An arity two :class:`ufl.Form`, defining the matrix.
      :arg bcs: Dirichlet boundary conditions.
      :arg form_compiler_parameters: Form compiler parameters.
      :arg linear_solver_parameters: Linear solver parameters.
      :arg replace_map: A :class:`Mapping` defining a map from symbolic
          variables to values.
      :arg assembly_cache: :class:`.AssemblyCache` to use for finite element
          assembly. Defaults to `assembly_cache()`.
      :returns: A :class:`tuple` `(value_ref, value)`. `value` is a tuple
          `(solver, A, b_bc)`, where `solver` is the linear solver, `A` is
          the assembled matrix, and `b_bc` is a boundary condition
          right-hand-side term which should be added after assembling a
          right-hand-side with homogeneous boundary conditions applied.
          `b_bc` may be `None` to indicate that this term is zero.
          `value_ref` is a :class:`.CacheRef` storing a reference to `value`.















      ..
          !! processed by numpydoc !!


.. py:class:: LocalSolverCache




   
   A :class:`.Cache` for element-wise local block diagonal linear solvers.
















   ..
       !! processed by numpydoc !!
   .. py:method:: local_solver(form, solver_type=None, *, replace_map=None)

      
      Construct an element-wise local block diagonal linear solver and
      cache the result, or return a previously cached result.

      :arg form: An arity two :class:`ufl.Form`, defining the element-wise
          local block diagonal matrix.
      :arg solver_type: `dolfin.LocalSolver.SolverType`. Defaults to
          `dolfin.LocalSolver.SolverType.LU`.
      :arg replace_map: A :class:`Mapping` defining a map from symbolic
          variables to values.
      :returns: A :class:`tuple` `(value_ref, value)`. `value` is a
          `tlm_adjoint.fenics.backend_interface.LocalSolver` and `value_ref`
          is a :class:`.CacheRef` storing a reference to `value`.















      ..
          !! processed by numpydoc !!


.. py:function:: assembly_cache()

   
   :returns: The default :class:`.AssemblyCache`.
















   ..
       !! processed by numpydoc !!

.. py:function:: set_assembly_cache(assembly_cache)

   
   Set the default :class:`.AssemblyCache`.

   :arg assembly_cache: The new default :class:`.AssemblyCache`.















   ..
       !! processed by numpydoc !!

.. py:function:: linear_solver_cache()

   
   :returns: The default :class:`.LinearSolverCache`.
















   ..
       !! processed by numpydoc !!

.. py:function:: set_linear_solver_cache(linear_solver_cache)

   
   Set the default :class:`.LinearSolverCache`.

   :arg linear_solver_cache: The new default :class:`.LinearSolverCache`.















   ..
       !! processed by numpydoc !!

.. py:function:: local_solver_cache()

   
   :returns: The default :class:`.LocalSolverCache`.
















   ..
       !! processed by numpydoc !!

.. py:function:: set_local_solver_cache(local_solver_cache)

   
   Set the default :class:`.LocalSolverCache`.

   :arg local_solver_cache: The new default :class:`.LocalSolverCache`.















   ..
       !! processed by numpydoc !!