Next: , Previous: , Up: Reference manual   [Contents][Index]


7.25 Package FL.STRATEGY

This package provides methods for solving problems by adaptive FEM.

Variable: *ETA-OBSERVE*

Observe an estimate of the global error.

Variable: *FE-APPROXIMATION-OBSERVE*

Standard observe quantities for fe approximation.

Variable: *MENTRIES-OBSERVE*

Observe entries for the size of the matrix.

Variable: *MODEL-TIME*

Contains the model time during the iteration for use in coefficient functions.

Variable: *ROTHE-OBSERVE*

Standard observe quantities for Rothe.

Variable: *STATIONARY-FE-STRATEGY-OBSERVE*

Standard observe quantities for stationary finite element strategies.

Class: <CELL-RULE-INDICATOR>

Calls cell-rule on a cell for determinining its refinement rule.

Superclasses: <REFINEMENT-INDICATOR>

Direct slots:

Class: <DUALITY-ERROR-ESTIMATOR>

Estimates the error by testing the difference z-IPz against the residual. Here z is the solution of a dual problem in an enriched finite element space.

Superclasses: <SETUP-ENRICHED-ANSATZ-SPACE> <SOLVE-DUAL-PROBLEM> <LOCAL-TEST-WITH-DUAL> <STANDARD-ERROR-ESTIMATOR>

Class: <FE-APPROXIMATION>

This class describes iterative finite element appoximation strategies.

Superclasses: <STRATEGY>

Direct slots:

Class: <FE-INTERPOLATION>

This class implements adaptive finite element interpolation of the given coefficient function as a variant of finite element approximation.

Superclasses: <FE-APPROXIMATION>

Direct slots:

Class: <LARGEST-ETA-INDICATOR>

Puts the fraction of the cells with the largest error contributions in the refinement table. Note that a fraction of 1.0 yields uniform refinement.

Superclasses: <REFINEMENT-INDICATOR>

Direct slots:

Class: <PROJECTION-ERROR-ESTIMATOR>

Estimates the error by measuring the difference between the solution and a projected solution in a hierarchical mesh by a certain norm given by local-p and global-p.

Superclasses: <DIFFERENCE-WITH-PROJECTION> <GLOBAL-AND-LOCAL-NORM> <STANDARD-ERROR-ESTIMATOR>

Class: <REFINEMENT-INDICATOR>

An error indicator can appear as first argument in the indicate which works on a blackboard. Often, it will use quantities computed by an error estimator before.

Direct slots:

Class: <ROTHE>

Rothe strategy for time-dependent problems. The idea of the Rothe method for solving U_t +A U =f is to do an ODE time-stepping scheme in an infinite-dimensional function space. Therefore, in every time-step, the solution has to be approximated sufficiently well in the space variable.

Superclasses: <ITERATION>

Direct slots:

Class: <STATIONARY-FE-STRATEGY>

This class describes some iterative finite element solution strategies for continuous, stationary PDE problems.

Superclasses: <FE-APPROXIMATION>

Direct slots:

Class: <STRATEGY>

A strategy is an iteration for solving a problem defined on a blackboard.

Superclasses: <ITERATIVE-SOLVER>

Class: <UNIFORM-REFINEMENT-INDICATOR>

Marks all cells for refinement.

Superclasses: <REFINEMENT-INDICATOR>

Function: ESTIMATE ERROR-ESTIMATOR BLACKBOARD

Yields both local and global estimate.

Function: INDICATE INDICATOR BLACKBOARD

Puts a list of cells to be refined on the blackboard.

Function: POINT-OBSERVE POSITION &KEY (COMPONENT 0) (INDEX 0)

Generates an observer triple which observes the value of component component of solution number index at the position position.


Next: , Previous: , Up: Reference manual   [Contents][Index]