com.graphhopper.routing.lm

Class LandmarkStorage

java.lang.Object

com.graphhopper.routing.lm.LandmarkStorage

public class LandmarkStorage
extends Object

This class stores the landmark nodes and the weights from and to all other nodes in every subnetwork. This data is created to apply a speed-up for path calculation but at the same times stays flexible to per-request changes. The class is safe for usage from multiple reading threads across algorithms.

Author:

Peter Karich

Constructor Summary

Constructors

Constructor and Description
LandmarkStorage(BaseGraph graph, EncodedValueLookup encodedValueLookup, Directory dir, LMConfig lmConfig, int landmarks)

Method Summary

Modifier and Type	Method and Description
void	close()
void	createLandmarks() This method calculates the landmarks and initial weightings to & from them.
protected com.carrotsearch.hppc.IntHashSet	findBorderEdgeIds(AreaIndex<SplitArea> areaIndex) This method makes edges crossing the specified border inaccessible to split a bigger area into smaller subnetworks.
void	flush()
long	getCapacity()
int	getLandmarkCount()
int[]	getLandmarks(int subnetwork)
LMConfig	getLMConfig()
Weighting	getLmSelectionWeighting()
int	getMinimumNodes()
int	getSubnetworksWithLandmarks()
Weighting	getWeighting() This method returns the weighting for which the landmarks are originally created
boolean	isClosed()
boolean	isEmpty()
boolean	loadExisting()
void	setAreaIndex(AreaIndex<SplitArea> areaIndex) This method specifies the polygons which should be used to split the world wide area to improve performance and quality in this scenario.
LandmarkStorage	setLandmarkSuggestions(List<LandmarkSuggestion> landmarkSuggestions) This method forces the landmark preparation to skip the landmark search and uses the specified landmark list instead.
void	setLMSelectionWeighting(Weighting lmSelectionWeighting) This weighting is used for the selection heuristic and is per default not the weighting specified in the constructor.
void	setLogDetails(boolean logDetails) By default do not log many details.
LandmarkStorage	setMaximumWeight(double maxWeight) Specify the maximum possible value for your used area.
void	setMinimumNodes(int minimumNodes) This method sets the required number of nodes of a subnetwork for which landmarks should be calculated.
String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

LandmarkStorage

public LandmarkStorage(BaseGraph graph,
                       EncodedValueLookup encodedValueLookup,
                       Directory dir,
                       LMConfig lmConfig,
                       int landmarks)

Method Detail

setMaximumWeight

public LandmarkStorage setMaximumWeight(double maxWeight)

Specify the maximum possible value for your used area. With this maximum weight value you can influence the storage precision for your weights that help A* finding its way to the goal. The same value is used for all subnetworks. Note, if you pick this value too big then too similar weights are stored (some bits of the storage capability will be left unused). If too low then far away values will have the same maximum value associated ("maxed out"). Both will lead to bad performance.

Parameters:

maxWeight - use a negative value to automatically determine this value.

setLogDetails

public void setLogDetails(boolean logDetails)

By default do not log many details.

setLandmarkSuggestions

public LandmarkStorage setLandmarkSuggestions(List<LandmarkSuggestion> landmarkSuggestions)

This method forces the landmark preparation to skip the landmark search and uses the specified landmark list instead. Useful for manual tuning of larger areas to safe import time or improve quality.

setMinimumNodes

public void setMinimumNodes(int minimumNodes)

This method sets the required number of nodes of a subnetwork for which landmarks should be calculated. Every subnetwork below this count will be ignored.

getMinimumNodes

public int getMinimumNodes()

See Also:

setMinimumNodes(int)

setLMSelectionWeighting

public void setLMSelectionWeighting(Weighting lmSelectionWeighting)

This weighting is used for the selection heuristic and is per default not the weighting specified in the constructor. The special weighting leads to a much better distribution of the landmarks and results in better response times.

getLmSelectionWeighting

public Weighting getLmSelectionWeighting()

getWeighting

public Weighting getWeighting()

This method returns the weighting for which the landmarks are originally created

getLMConfig

public LMConfig getLMConfig()

createLandmarks

public void createLandmarks()

This method calculates the landmarks and initial weightings to & from them.

setAreaIndex

public void setAreaIndex(AreaIndex<SplitArea> areaIndex)

This method specifies the polygons which should be used to split the world wide area to improve performance and quality in this scenario.

findBorderEdgeIds

protected com.carrotsearch.hppc.IntHashSet findBorderEdgeIds(AreaIndex<SplitArea> areaIndex)

This method makes edges crossing the specified border inaccessible to split a bigger area into smaller subnetworks. This is important for the world wide use case to limit the maximum distance and also to detect unreasonable routes faster.

getLandmarkCount

public int getLandmarkCount()

getLandmarks

public int[] getLandmarks(int subnetwork)

getSubnetworksWithLandmarks

public int getSubnetworksWithLandmarks()

Returns:

the number of subnetworks that have landmarks

isEmpty

public boolean isEmpty()

toString

public String toString()

Overrides:

toString in class Object

loadExisting

public boolean loadExisting()

flush

public void flush()

close

public void close()

isClosed

public boolean isClosed()

getCapacity

public long getCapacity()