org.hipparchus.ode.events

Class EventFilter

java.lang.Object

org.hipparchus.ode.events.EventFilter

All Implemented Interfaces:

ODEEventHandler

public class EventFilter
extends Object
implements ODEEventHandler

Wrapper used to detect only increasing or decreasing events.

General events are defined implicitly by a g function crossing zero. This function needs to be continuous in the event neighborhood, and its sign must remain consistent between events. This implies that during an ODE integration, events triggered are alternately events for which the function increases from negative to positive values, and events for which the function decreases from positive to negative values.

Sometimes, users are only interested in one type of event (say increasing events for example) and not in the other type. In these cases, looking precisely for all events location and triggering events that will later be ignored is a waste of computing time.

Users can wrap a regular event handler in an instance of this class and provide this wrapping instance to the ODE solver in order to avoid wasting time looking for uninteresting events. The wrapper will intercept the calls to the g function and to the eventOccurred method in order to ignore uninteresting events. The wrapped regular event handler will the see only the interesting events, i.e. either only increasing events or decreasing events. the number of calls to the g function will also be reduced.

Constructor Summary

Constructors

Constructor and Description
EventFilter(ODEEventHandler rawHandler, FilterType filter) Wrap an event handler.

Method Summary

Modifier and Type	Method and Description
Action	eventOccurred(ODEStateAndDerivative state, boolean increasing) Handle an event and choose what to do next.
double	g(ODEStateAndDerivative state) Compute the value of the switching function.
void	init(ODEStateAndDerivative initialState, double finalTime) Initialize event handler at the start of an ODE integration.
ODEState	resetState(ODEStateAndDerivative state) Reset the state prior to continue the integration.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

EventFilter

public EventFilter(ODEEventHandler rawHandler,
                   FilterType filter)

Wrap an event handler.

Parameters:

rawHandler - event handler to wrap

filter - filter to use

Method Detail

init

public void init(ODEStateAndDerivative initialState,
                 double finalTime)

Initialize event handler at the start of an ODE integration.

This method is called once at the start of the integration. It may be used by the event handler to initialize some internal data if needed.

The default implementation does nothing

Specified by:

init in interface ODEEventHandler

Parameters:

initialState - initial time, state vector and derivative

finalTime - target time for the integration

g

public double g(ODEStateAndDerivative state)

Compute the value of the switching function.

The discrete events are generated when the sign of this switching function changes. The integrator will take care to change the stepsize in such a way these events occur exactly at step boundaries. The switching function must be continuous in its roots neighborhood (but not necessarily smooth), as the integrator will need to find its roots to locate precisely the events.

Also note that for the integrator to detect an event the sign of the switching function must have opposite signs just before and after the event. If this consistency is not preserved the integrator may not detect any events.

This need for consistency is sometimes tricky to achieve. A typical example is using an event to model a ball bouncing on the floor. The first idea to represent this would be to have g(state) = h(state) where h is the height above the floor at time state.getTime(). When g(state) reaches 0, the ball is on the floor, so it should bounce and the typical way to do this is to reverse its vertical velocity. However, this would mean that before the event g(state) was decreasing from positive values to 0, and after the event g(state) would be increasing from 0 to positive values again. Consistency is broken here! The solution here is to have g(state) = sign * h(state), where sign is a variable with initial value set to +1. Each time eventOccurred is called, sign is reset to -sign. This allows the g(state) function to remain continuous (and even smooth) even across events, despite h(state) is not. Basically, the event is used to fold h(state) at bounce points, and sign is used to unfold it back, so the solvers sees a g(state) function which behaves smoothly even across events.

Calling this multiple times with the same state will result in the same value. The definition of the g function may change when an event occurs, as in the above example.

Specified by:

g in interface ODEEventHandler

Parameters:

state - current value of the independent time variable, state vector and derivative

Returns:

value of the g switching function

See Also:

org.hipparchus.ode.events

eventOccurred

public Action eventOccurred(ODEStateAndDerivative state,
                            boolean increasing)

Handle an event and choose what to do next.

This method is called when the integrator has accepted a step ending exactly on a sign change of the function, just before the step handler itself is called (see below for scheduling). It allows the user to update his internal data to acknowledge the fact the event has been handled (for example setting a flag in the org.hipparchus.migration.ode.FirstOrderDifferentialEquations differential equations to switch the derivatives computation in case of discontinuity), or to direct the integrator to either stop or continue integration, possibly with a reset state or derivatives.

• if Action.STOP is returned, the step handler will be called with the isLast flag of the handleStep method set to true and the integration will be stopped,
• if Action.RESET_STATE is returned, the resetState method will be called once the step handler has finished its task, and the integrator will also recompute the derivatives,
• if Action.RESET_DERIVATIVES is returned, the integrator will recompute the derivatives,
• if Action.CONTINUE is returned, no specific action will be taken (apart from having called this method) and integration will continue.

The scheduling between this method and the ODEStepHandler method handleStep(interpolator, isLast) is to call this method first and handleStep afterwards. This scheduling allows the integrator to pass true as the isLast parameter to the step handler to make it aware the step will be the last one if this method returns Action.STOP. As the interpolator may be used to navigate back throughout the last step (as StepNormalizer does for example), user code called by this method and user code called by step handlers may experience apparently out of order values of the independent time variable. As an example, if the same user object implements both this EventHandler interface and the ODEFixedStepHandler interface, a forward integration may call its eventOccurred method with t = 10 first and call its handleStep method with t = 9 afterwards. Such out of order calls are limited to the size of the integration step for variable step handlers and to the size of the fixed step for fixed step handlers.

Specified by:

eventOccurred in interface ODEEventHandler

Parameters:

state - current value of the independent time variable, state vector and derivative

increasing - if true, the value of the switching function increases when times increases around event (note that increase is measured with respect to physical time, not with respect to integration which may go backward in time)

Returns:

indication of what the integrator should do next, this value must be one of Action.STOP, Action.RESET_STATE, Action.RESET_DERIVATIVES or Action.CONTINUE

resetState

public ODEState resetState(ODEStateAndDerivative state)

Reset the state prior to continue the integration.

This method is called after the step handler has returned and before the next step is started, but only when ODEEventHandler.eventOccurred(org.hipparchus.ode.ODEStateAndDerivative, boolean) has itself returned the Action.RESET_STATE indicator. It allows the user to reset the state vector for the next step, without perturbing the step handler of the finishing step.

The default implementation returns its argument.

Specified by:

resetState in interface ODEEventHandler

Parameters:

state - current value of the independent time variable, state vector and derivative

Returns:

reset state (note that it does not include the derivatives, they will be added automatically by the integrator afterwards)