ell::WalkGradient Class Reference
#include <Walk.hh>
Inheritance diagram for ell::WalkGradient:
Detailed Description
Implements a gradient (steepest decent) walk in the landscape, i.e. every state along the walk is the smallest of all neighbors of its predecessor. Therefore, a gradient walk ends in a local minimum. For each step all neighbors of the current walk end have to be checked if they are smaller according to the 'Stateoperator<(s2)'.WalkGradient defines:
- a static function "walkGradient" which takes all neccesary params for an gradient walk
- a virtual member function "walk" which does what "walkGradient" does
Definition at line 159 of file Walk.hh.
Public Member Functions | |
| virtual StateCollector * | walk (const State *const start, StateCollector *const scWalk, StateCollector *const scRejected) const |
| WalkGradient () | |
| virtual | ~WalkGradient () |
Static Public Member Functions | |
| static StateCollector * | walk (const State *const start, StateCollector *const scWalk, const NeighborGenerator *const ng, const StateAcceptor *const sa, const WalkAbortionCriterion *const wac, StateCollector *scRejected=NULL) |
| static StateCollector * | walkGradient (const State *const start, StateCollector *const scWalk) |
Constructor & Destructor Documentation
| ell::WalkGradient::WalkGradient | ( | ) |
| virtual ell::WalkGradient::~WalkGradient | ( | ) | [inline, virtual] |
Member Function Documentation
| StateCollector * ell::Walk::walk | ( | const State *const | start, | |
| StateCollector *const | scWalk, | |||
| const NeighborGenerator *const | ng, | |||
| const StateAcceptor *const | sa, | |||
| const WalkAbortionCriterion *const | wac, | |||
| StateCollector * | scRejected = NULL | |||
| ) | [static, inherited] |
This function implements a generic walk starting from State start. The properties of this search are defined by the implementations of NeighborhoodGenerator, StateAcceptor WalkAbortionCriterion and StateCollector.
- Parameters:
-
start the begin of the walk scWalk the StateCollector each step of the walk is added to ng the NeighborGenerator used to get the neighbor list of the last walk State to generate the next possible steps sa the StateAcceptor that decides if a step is accepted or rejected wac the walk abortion criterion that decides if the end of the walk has been reached or a further step is to do scRejected the StateCollector each rejected possible step is added to
- Returns:
- the modified scWalk StateCollector
| StateCollector * ell::WalkGradient::walk | ( | const State *const | start, | |
| StateCollector *const | scWalk, | |||
| StateCollector *const | scRejected | |||
| ) | const [virtual] |
pure virtual function, s is start state
- Parameters:
-
start the begin of the walk scWalk the StateCollector each step of the walk is added to scRejected the StateCollector each rejected possible step is added to
- Returns:
- the modified scWalk StateCollector
Implements ell::Walk.
| StateCollector * ell::WalkGradient::walkGradient | ( | const State *const | start, | |
| StateCollector *const | scWalk | |||
| ) | [static] |
Static function to perform a gradient (steepest decent) walk.
- Parameters:
-
start the begin of the walk scWalk the StateCollector each step of the walk is added to
- Returns:
- the modified scWalk StateCollector
The documentation for this class was generated from the following files: