Package org.apache.wayang.core.optimizer

Class OptimizationUtils

java.lang.Object

org.apache.wayang.core.optimizer.OptimizationUtils

public class OptimizationUtils
extends java.lang.Object

Utility methods for the optimization process.

Constructor Summary

Constructors

Constructor	Description
OptimizationUtils()

Method Summary

Modifier and Type	Method	Description
static java.util.Set<Slot<?>>	collectConnectedSlots(InputSlot<?> input)	Collects all Slots that are related to the given InputSlot either by a SlotMapping or by OutputSlot/InputSlot occupation.
static java.util.Set<Slot<?>>	collectConnectedSlots(OutputSlot<?> output)	Collects all Slots that are related to the given OutputSlot either by a SlotMapping or by OutputSlot/InputSlot occupation.
static java.util.Set<Slot<?>>	collectConnectedSlots(Slot<?> slot)	Collects all Slots that are related to the given Slot either by a SlotMapping or by OutputSlot/InputSlot occupation.
static <T extends PlanEnumerationPruningStrategy> T	createPruningStrategy(java.lang.Class<T> strategyClass, Configuration configuration)	Creates a new PlanEnumerationPruningStrategy and configures it.
static OutputSlot<?>	findWayangPlanOutputSlotFor(Channel openChannel)	Determine the producing OutputSlot of this Channel that lies within a WayangPlan.
static Channel	getPredecessorChannel(Channel channel)	Finds the single input Channel of the given channel's producing ExecutionTask.
static double	logisticGrowth(double g0, double epsilon, double x0, double x)	Uses the right part of the logistic regression curve to model a strictly growing function between g0 and 1.

Methods inherited from class java.lang.Object

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

Constructor Detail

OptimizationUtils

public OptimizationUtils()

Method Detail

findWayangPlanOutputSlotFor

public static OutputSlot<?> findWayangPlanOutputSlotFor(Channel openChannel)

Determine the producing OutputSlot of this Channel that lies within a WayangPlan. We follow non-WayangPlan ExecutionOperators because they should merely forward data.

getPredecessorChannel

public static Channel getPredecessorChannel(Channel channel)

Finds the single input Channel of the given channel's producing ExecutionTask.

Parameters:

channel - whose predecessor is requested

Returns:

the preceeding Channel

createPruningStrategy

public static <T extends PlanEnumerationPruningStrategy> T createPruningStrategy(java.lang.Class<T> strategyClass,
                                                                                 Configuration configuration)

Creates a new PlanEnumerationPruningStrategy and configures it.

Parameters:

strategyClass - the Class of the PlanEnumerationPruningStrategy; must have a default constructor

configuration - provides any potential configuration values

Returns:

the configured PlanEnumerationPruningStrategy instance

collectConnectedSlots

public static java.util.Set<Slot<?>> collectConnectedSlots(Slot<?> slot)

Collects all Slots that are related to the given Slot either by a SlotMapping or by OutputSlot/InputSlot occupation.

Parameters:

slot - whose related Slots are requested

Returns:

the related Slots including the given Slot

collectConnectedSlots

public static java.util.Set<Slot<?>> collectConnectedSlots(InputSlot<?> input)

Collects all Slots that are related to the given InputSlot either by a SlotMapping or by OutputSlot/InputSlot occupation.

Parameters:

input - whose related Slots are requested

Returns:

the related Slots including the given InputSlot

collectConnectedSlots

public static java.util.Set<Slot<?>> collectConnectedSlots(OutputSlot<?> output)

Collects all Slots that are related to the given OutputSlot either by a SlotMapping or by OutputSlot/InputSlot occupation.

Parameters:

output - whose related Slots are requested

Returns:

the related Slots including the given OutputSlot

logisticGrowth

public static double logisticGrowth(double g0,
                                    double epsilon,
                                    double x0,
                                    double x)

Uses the right part of the logistic regression curve to model a strictly growing function between g0 and 1.

Parameters:

g0 - the starting value at x = 0

epsilon - the function should have a value >= 1 - epsilon for x >= x0

x0 - the convergence x value

x - the actual input value to the function

Returns:

the function value