sim.field
Class SparseField

java.lang.Object
  extended bysim.field.SparseField
All Implemented Interfaces:
java.io.Serializable
Direct Known Subclasses:
Continuous2D, Continuous3D, SparseGrid2D, SparseGrid3D

public abstract class SparseField
extends java.lang.Object
implements java.io.Serializable

While it has no abstract fields, SparseField is explicitly an abstract superclass of various sparse field objects. It specifies a many-to-one relationship between objects and locations. It contains two hash tables and a bag, resulting in fairly speedy, O(1) searches for all objects at a location, all objects in the entire field, changes in location for an object, and addition or removal of an object.

There are two flags you can set to trade memory for speed. These flags are here because often in a Sparse Field, you'd load a lot of things into one location, then delete all of them, but the large Bag which held them is still hanging there. Thus: removeEmptyBags will garbage-collect any bags in the field which have been completely emptied (except for the allObjects bag). And replaceLargeBags will replace large bags (over 32 in size) with smaller ones when the bag's contents drop below 1/4 the bag size.

To create a SparseField, you need to specify exactly what type a "location" object is. You do this by overriding the removeObjectsAtlocation() and setObjectLocation() methods, with versions set to the proper location class. Further, you create a method called getObjectLocation(), which returns objects of that class (by calling getRawObjectLocation() and casting the result into the proper type). See the Example Usage below.

Warning About Hashing. Java's hashing method is broken in an important way. One can override the hashCode() and equals() methods of an object so that they hash by the value of an object rather than just the pointer to it. But if this is done, then if you use this object as a key in a hash table, then change those values in the object, it will break the hash table -- the key and the object hashed by it will both be lost in the hashtable, unable to be accessed or removed from it. The moral of the story is: do not override hashCode() and equals() to hash by value unless your object is immutable -- its values cannot be changed. This is the case, for example, with Strings, which hash by value but cannot be modified. It's also the case with Int2D, Int3D, Double2D, and Double3D, as well as Double, Integer, etc. Some of Sun's own objects are broken in this respect: Point, Point2D, etc. are both mutable and hashed by value.

This affects you in the following way. While SparseField can specify any Object as a "location", generally speaking, subclasses should only permit immutable objects as locations (Int2D, Integer, String, etc.), or warn the user to never modify the location object he provides as a key.

Computational Complexity. Adding a new object to a location is O(1). Changing an object's location, or removing the object, is O(M), where M is the number of objects currently located at the object's old location. Scanning through all objects is O(N) and fast, where N is the number of objects total -- just scan through the allObjects Bag. Scanning through all stored locations is potentially slow, probably O(H), where H is the number of hash table buckets in the objectHash hash table -- you'd have to get a hash table iterator and iterate through it. Removing all objects at a given location is O(O), where O is the number of objects at that location. Clearing the hash table is O(1) discounting GC.

Example Usage. Here is an example of a simple subclass which allows locations to be positive, non-zero integers:

    
// make final to inline (one can wish!)
public final class PositiveIntegerSparseField extends SparseField
    {

    // note we return an Integer, not an Object.  If the user wants
    // an abstract supermethod for all SparseFields, he can use
    // getRawObjectLocation instead.  Java's a pain sometimes

    public Integer getObjectLocation(Object obj)
        {
        return (Integer) super.getRawObjectLocation(obj);
        }


    // note we explicitly state that the location has to be an integer

    public Bag removeObjectsAtLocation(Integer location)
        {
        return super.removeObjectsAtLocation(location);
        }


    // note we explicitly state that the location has to be an integer

    public boolean setObjectLocation(Object obj, final Integer location)
        {
        if (location.intValue() >= 0)  // it's a valid location
            return super.setObjectLocation(obj, location);
        else return false;
        }
    }
    
    

See Also:
Serialized Form

Nested Class Summary
static class SparseField.LocationAndIndex
          Objects stored in SparseField's locationAndIndexHash table.
 
Field Summary
 Bag allObjects
          All the objects in the sparse field.
static int LARGE_BAG_RATIO
          A bag must be larger than this amount times the capacity before it is eliminated if replaceLargeBags is true
 java.util.HashMap locationAndIndexHash
          LocationAndIndex objects (locations and indexes into the allObjects array) hashed by Object
static int MIN_BAG_SIZE
          No bags smaller than this size will be replaced regardless of the setting of replaceLargeBags
 java.util.HashMap objectHash
          Bags of objects hashed by location.
 boolean removeEmptyBags
          Should we remove bags in the field if they have been emptied, and let them GC, or should we keep them around? This doesn't include the allObjects bag.
 boolean replaceLargeBags
          When a bag drops to one quarter capacity, should we replace it with a new bag? This doesn't include the allObjects bag.
 
Constructor Summary
SparseField()
           
 
Method Summary
 Bag clear()
          Deletes everything, returning all the objects as a Bag (which you can freely use and modify).
 Bag getAllObjects()
          Returns all the objects in the Sparse Field.
 Bag getObjectsAtLocation(java.lang.Object location)
          Returns a bag containing all the objects at a given location.
 Bag getObjectsAtLocations(Bag locations, Bag result)
          For each location, puts all object at that location into the result bag.
 java.lang.Object getRawObjectLocation(java.lang.Object obj)
          Get the location of the provided object, or null if the object does not exist.
 java.util.Iterator iterator()
          Iterates over all objects.
 java.util.Iterator locationBagIterator()
          Iterates [somewhat inefficiently] over all bags of objects grouped by location.
 java.lang.Object remove(java.lang.Object obj)
          Removes an object if it exists.
protected  Bag removeObjectsAtLocation(java.lang.Object location)
          Removes objects at the given location, and returns a bag of them, or null.
protected  boolean setObjectLocation(java.lang.Object obj, java.lang.Object location)
          Changes the location of an object, or adds if it doesn't exist yet.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

removeEmptyBags

public boolean removeEmptyBags
Should we remove bags in the field if they have been emptied, and let them GC, or should we keep them around? This doesn't include the allObjects bag.


replaceLargeBags

public boolean replaceLargeBags
When a bag drops to one quarter capacity, should we replace it with a new bag? This doesn't include the allObjects bag.


MIN_BAG_SIZE

public static final int MIN_BAG_SIZE
No bags smaller than this size will be replaced regardless of the setting of replaceLargeBags

See Also:
Constant Field Values

LARGE_BAG_RATIO

public static final int LARGE_BAG_RATIO
A bag must be larger than this amount times the capacity before it is eliminated if replaceLargeBags is true

See Also:
Constant Field Values

locationAndIndexHash

public java.util.HashMap locationAndIndexHash
LocationAndIndex objects (locations and indexes into the allObjects array) hashed by Object


objectHash

public java.util.HashMap objectHash
Bags of objects hashed by location. Do not rely on these bags always being the same objects.


allObjects

public Bag allObjects
All the objects in the sparse field. For fast scans. Do not rely on this bag always being the same object.

Constructor Detail

SparseField

public SparseField()
Method Detail

getRawObjectLocation

public final java.lang.Object getRawObjectLocation(java.lang.Object obj)
Get the location of the provided object, or null if the object does not exist. Override this and create a getObjectLocation(Object obj) method which returns a location type appropriate for your kind of Sparse Field.


getObjectsAtLocation

public final Bag getObjectsAtLocation(java.lang.Object location)
Returns a bag containing all the objects at a given location. You should NOT MODIFY THIS BAG. This is the actual container bag, and modifying it will almost certainly break the Sparse Field object. If you want to modify the bag, make a copy and modify the copy instead, using something along the lines of new Bag(foo.getObjectsAtLocation(location)) . Furthermore, changing values in the Sparse Field may result in a different bag being used -- so you should not rely on this bag staying valid.


removeObjectsAtLocation

protected Bag removeObjectsAtLocation(java.lang.Object location)
Removes objects at the given location, and returns a bag of them, or null. You can freely modify this bag.


clear

public Bag clear()
Deletes everything, returning all the objects as a Bag (which you can freely use and modify). If you need the Bag, then this is a useful method -- otherwise it might in fact be faster to just make a brand new Sparse Field and let the garbage collector do its magic.


remove

public java.lang.Object remove(java.lang.Object obj)
Removes an object if it exists. Returns its location, or null if the object didn't exist.


setObjectLocation

protected boolean setObjectLocation(java.lang.Object obj,
                                    java.lang.Object location)
Changes the location of an object, or adds if it doesn't exist yet. Returns false if the object could not be set to that location. For example: you cannot put null objects or null locations into a Sparse Field. Subclasses may have further restrictions on locations.


getAllObjects

public final Bag getAllObjects()
Returns all the objects in the Sparse Field. Do NOT modify the bag that you receive from this method -- it is used internally. If you wish to modify the Bag you receive, make a copy of the Bag first, using something like new Bag(foo.getAllObjects()).


getObjectsAtLocations

public Bag getObjectsAtLocations(Bag locations,
                                 Bag result)
For each location, puts all object at that location into the result bag. Returns the result bag. If the provided result bag is null, one will be created and returned.


iterator

public java.util.Iterator iterator()
Iterates over all objects. NOT fail-fast, and remove() not supported. Use this method only if you're woozy about accessing allObject.numObjs and allObject.objs directly. For the fastest scan, you can do:

for(int x=0;x

... but do NOT modify the allObjects.objs array.


locationBagIterator

public java.util.Iterator locationBagIterator()
Iterates [somewhat inefficiently] over all bags of objects grouped by location. Only used by SparseFieldPortrayal -- generally this should not be interesting to you.