Implementation of the rotational leapfrog integration scheme. More...

#include <Leapfrog.h>

Inheritance diagram for Leapfrog:

Public Types
enum	{ STATE_UNKNOWN = 0 , STATE_NEW_TIMESTEP = 1 , STATE_PRE_FORCE_CALCULATION = 2 , STATE_POST_FORCE_CALCULATION = 3 }

Public Member Functions
	Leapfrog (double timestepLength)
	The constructor.

	~Leapfrog ()
	The destructor.

virtual void	readXML (XMLfileUnits &xmlconfig)
	Read in XML configuration for leapfrog integrator and all its included objects. More...

virtual void	init ()

void	eventForcesCalculated (ParticleContainer molCont, Domain domain)
	steps between the force calculation and the end of the time step More...

void	eventNewTimestep (ParticleContainer molCont, Domain domain)
	performs all steps that can be done before new forces are needed More...

Public Member Functions inherited from Integrator
	Integrator ()
	The constructor.

	Integrator (double timestepLength)

virtual	~Integrator ()
	The destructor.

void	setTimestepLength (double dt)
	set the time between two time steps

double	getTimestepLength ()
	get the time between two time steps

Additional Inherited Members
Protected Attributes inherited from Integrator
double	_timestepLength
	time between time step n and time step (n+1)

Detailed Description

Implementation of the rotational leapfrog integration scheme.

TODO: unnecessarily complicated and cluttered:

This Leapfrog integrator is implemented as a deterministic finite automaton (DFA):

A state of the DFA corresponds to a coarse step in the integration loop. The following states are used:
- state 3: "starting state". The simulation has either just started, or the last steps of a time step have just been done. In state 3, the leapfrog integrator waits for the start of the next time time step
- state 1: a new time step has just begun, the integrator is ready to start integrating (transition to state 2)
- state 2: the first part is finished, now forces are needed to continue the integration
consequently, the transitions do the following things:
- 3 to 1: Nothing to do
- 1 to 2: calculate "preF"
- 2 to 3: calculate "postF"
For the transition to another state, a condition has to be fulfilled There are two possibilities: The integrator has to wait for external information e.g. the new forces. In this case, the integrator is informed, that new forces have been calculated. The automaton can then do all necessary computations that have to be done do get from the current state to the next state

For details about the algorithm see David Fincham's paper "Leapfrog rotational algorithms". [1]

Member Function Documentation

void Leapfrog::eventForcesCalculated	(	ParticleContainer *	molCont,
		Domain *	domain
	)

virtual

steps between the force calculation and the end of the time step

checks whether the current state of the integrator allows that this method is called

Implements Integrator.

void Leapfrog::eventNewTimestep	(	ParticleContainer *	molCont,
		Domain *	domain
	)

virtual

performs all steps that can be done before new forces are needed

checks whether the current state of the integrator allows that this method is called

Implements Integrator.

void Leapfrog::init ( )

virtual

Initialize Integrator

Implements Integrator.

void Leapfrog::readXML ( XMLfileUnits & xmlconfig )

virtual

Read in XML configuration for leapfrog integrator and all its included objects.

The following xml object structure is handled by this method:

<integrator type="Leapfrog" >
  <timestep>DOUBLE</timestep>
</integrator>

Implements Integrator.

The documentation for this class was generated from the following files: