opencmp.solvers package

Subpackages

• opencmp.solvers.adaptive_transient_solvers package
  • Submodules
  • opencmp.solvers.adaptive_transient_solvers.adaptive_IMEX module
  • opencmp.solvers.adaptive_transient_solvers.adaptive_three_step module
  • opencmp.solvers.adaptive_transient_solvers.adaptive_two_step module
  • opencmp.solvers.adaptive_transient_solvers.base_adaptive_transient_RK module
  • opencmp.solvers.adaptive_transient_solvers.base_adaptive_transient_multistep module
  • Module contents

Submodules

opencmp.solvers.base_solver module

class opencmp.solvers.base_solver.Solver(model_class, config)

Bases: abc.ABC

Base class for the different stationary/transient solvers.

abstract _apply_boundary_conditions()

Apply the boundary conditions to all of the GridFunctions used by the model.

Return type

None

abstract _assemble()

Assemble the linear and bilinear forms of the model.

Return type

None

abstract _create_linear_and_bilinear_forms()

Create the linear and bilinear forms of the model and add any required time integration terms.

Return type

None

abstract _create_preconditioners()

Create the preconditioner(s) used by this time integration scheme.

Return type

None

_dt_for_next_time_to_hit()

Function that calculate the time step till the next time that the model MUST be solved at due to a variety of constraints.

This time is used to set the time step in order to ensure that the model is solved at this time.

Return type

float

Returns

The next time that the model MUST be solved at.

abstract _load_and_apply_initial_conditions()

Function to load the initial conditions.

Return type

None

_log_timestep(accepted, error_abs, error_rel, component)

Function to print out information about the current timestep’s iteration

Parameters

• accepted (bool) – Boolean indicating if the current timestep was accepted, or if it being rerun with a smaller dt.

• error_abs (float) – The absolute local error for this timestep. (not absolute value, but raw error value)

• error_rel (float) – The relative local error for this timestep.

Return type

None

abstract _re_assemble()

Assemble the linear and bilinear forms of the model and update the preconditioner.

Return type

None

abstract _single_solve()

Function to solve the model for the current time step.

Return type

None

_solve()

Function to perform a single iteration of the solve.

Return type

None

abstract _startup()

Higher order methods need to be started up with a first order implicit solve.

Return type

None

_update_bcs(bc_dict_patch)

Function to update the model’s BCs to arbitrary values, and then recreate the linear/bilinear forms and the preconditioner.

This function is used by controllers in order to act on the manipulated variables.

Parameters

bc_dict_patch (Dict[str, Dict[str, Dict[str, List[Optional[CoefficientFunction, None]]]]]) – Dictionary containing new values for the BCs being updated.

Return type

None

abstract _update_preconditioners(precond_lst=None)

Update the preconditioner(s) used by this time integration scheme. This is needed because the preconditioner(s) can’t be updated if it is None.

Return type

None

abstract _update_time_step()

Function to calculate the new timestep and update the time (if a step is being taken).

If the model is stationary, this does nothing.

Return type

Tuple[bool, float, float, str]

Returns

Tuple containing a bool, a float, and a string. The bool indicates whether or not the result of the current iteration was accepted based on all of the criteria established by the individual solver (a stationary solver MUST return True). The float indicates the current local error. The string contains the variable name for the variable with the highest local error.

reset_model()

Function to reset certain model variables back to an initial state.

Needed for running convergence tests.

Return type

None

solve()

This function solves the model, either a single stationary solve or the entire transient solve.

Return type

GridFunction

Returns

GridFunction containing the solution. For a transient solve this will be the result of the final time step.

opencmp.solvers.misc module

opencmp.solvers.misc.get_solver_class(config)

Function to get the solver to use.

Parameters

config (ConfigParser) – The config file from which to get information for which solver to use.

Return type

Type[Solver]

Returns

The solver to use.

opencmp.solvers.stationary module

class opencmp.solvers.stationary.StationarySolver(model_class, config)

Bases: opencmp.solvers.base_solver.Solver

Stationary solver.

opencmp.solvers.time_integration_schemes module

opencmp.solvers.time_integration_schemes.CNLF(model, gfu_0, dt)

Crank Nicolson Leap Frog IMEX time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.RK_222(model, gfu_0, dt, step)

Two-step second-order Runge Kutta IMEX time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of intermediate steps in reverse step order (ie: [step n+1 sol, step 2 sol, step 1 sol, step n sol]).

• dt (List[Parameter]) – List of current time step size.

• step (int) – Which step of the Runge Kutta scheme should be assembled.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.RK_232(model, gfu_0, dt, step)

Three-step second-order Runge Kutta IMEX time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of intermediate steps in reverse step order (ie: [step n+1 sol, step 2 sol, step 1 sol, step n sol]).

• dt (List[Parameter]) – List of current time step size.

• step (int) – Which step of the Runge Kutta scheme should be assembled.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.SBDF(model, gfu_0, dt)

Third order semi-implicit backwards differencing time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes._add_dt_terms(a, L, gfu_lst, model, time_discretization_scheme)

Function to handle the adding of the time discretization terms to the linear and bilinear forms.

Parameters

• a (List[BilinearForm]) – List of all of the bilinear forms for the model

• L (List[BilinearForm]) – List of all of the linear forms for the model

• gfu_lst (List[List[GridFunction]]) – List of the solutions of previous time steps in reverse chronological order.

• model (Model) –

• time_discretization_scheme (str) – The name of the time integration discretization being used.

Return type

None

opencmp.solvers.time_integration_schemes._split_gfu(gfu)

Function to separate out the various components of each gridfunction solution to a previous timestep.

The first entry in gfu_lst is None so that its indexing matches the indexing for dt and the step argument in the functions for constructing the weak form (ie: gfu_lst[1], dt[1] and step=1 all refer to values at time n). The final result should be gfu_lst = [None, [component 0, component 1…] at t^n, [component 0, component 1…] at t^n-1, …].

Parameters

gfu (List[GridFunction]) – List of gridfunctions to split up

Return type

List[List[GridFunction]]

Returns

[[component 0, component 1…] at t^n, [component 0, component 1…] at t^n-1, …]

opencmp.solvers.time_integration_schemes.adaptive_IMEX_pred(model, gfu_0, dt)

Predictor for the adaptive time-stepping IMEX time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.crank_nicolson(model, gfu_0, dt)

Crank Nicolson (trapezoidal rule) time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.euler_IMEX(model, gfu_0, dt)

First order IMEX time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.explicit_euler(model, gfu_0, dt)

Explicit Euler time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.time_integration_schemes.implicit_euler(model, gfu_0, dt, step=0)

Implicit Euler time integration scheme.

This function constructs the final bilinear and linear forms for the time integration scheme by adding the necessary time-dependent terms to the model’s stationary terms. The returned bilinear and linear forms have NOT been assembled.

Parameters

• model (Model) – The model to solve.

• gfu_0 (List[GridFunction]) – List of the solutions of previous time steps ordered from most recent to oldest.

• dt (List[Parameter]) – List of timestep sizes ordered from most recent to oldest.

• step (int) – Used for adaptive_three_step to ensure the correct boundary condition and model function values are used when the half steps are taken.

Returns

• a: A list of the final bilinear forms (as a BilinearForm but not assembled).

• L: A list of the final linear forms (as a LinearForm but not assembled).

Return type

Tuple[BilinearForm, LinearForm]

opencmp.solvers.transient_RK module

class opencmp.solvers.transient_RK.TransientRKSolver(model_class, config)

Bases: opencmp.solvers.base_solver.Solver

Transient Runge Kutta solver with a fixed time step.

reset_model()

Function to reset certain model variables back to an initial state.

Needed for running convergence tests.

Return type

None

opencmp.solvers.transient_multistep module

class opencmp.solvers.transient_multistep.TransientMultiStepSolver(model_class, config)

Bases: opencmp.solvers.base_solver.Solver

Transient multistep solver with a fixed time step.

reset_model()

Function to reset certain model variables back to an initial state.

Needed for running convergence tests.

Return type

None

Module contents