cern.colt.map.tfloat.OpenLongFloatHashMap Maven / Gradle / Ivy
Show all versions of parallelcolt Show documentation
/*
Copyright (C) 1999 CERN - European Organization for Nuclear Research.
Permission to use, copy, modify, distribute and sell this software and its documentation for any purpose
is hereby granted without fee, provided that the above copyright notice appear in all copies and
that both that copyright notice and this permission notice appear in supporting documentation.
CERN makes no representations about the suitability of this software for any purpose.
It is provided "as is" without expressed or implied warranty.
*/
package cern.colt.map.tfloat;
import cern.colt.function.tfloat.LongFloatProcedure;
import cern.colt.function.tlong.LongProcedure;
import cern.colt.list.tbyte.ByteArrayList;
import cern.colt.list.tfloat.FloatArrayList;
import cern.colt.list.tlong.LongArrayList;
import cern.colt.map.HashFunctions;
import cern.colt.map.PrimeFinder;
/**
* Hash map holding (key,value) associations of type (int-->float);
* Automatically grows and shrinks as needed; Implemented using open addressing
* with float hashing. First see the package
* summary and javadoc tree view to get the
* broad picture.
*
* Overrides many methods for performance reasons only.
*
* @author [email protected]
* @version 1.0, 09/24/99
* @see java.util.HashMap
*/
public class OpenLongFloatHashMap extends AbstractLongFloatMap {
/**
*
*/
private static final long serialVersionUID = 1L;
// public static int hashCollisions = 0;
/**
* The hash table keys.
*
* @serial
*/
protected long table[];
/**
* The hash table values.
*
* @serial
*/
protected float values[];
/**
* The state of each hash table entry (FREE, FULL, REMOVED).
*
* @serial
*/
protected byte state[];
/**
* The number of table entries in state==FREE.
*
* @serial
*/
protected int freeEntries;
protected static final byte FREE = 0;
protected static final byte FULL = 1;
protected static final byte REMOVED = 2;
/**
* Constructs an empty map with default capacity and default load factors.
*/
public OpenLongFloatHashMap() {
this(defaultCapacity);
}
/**
* Constructs an empty map with the specified initial capacity and default
* load factors.
*
* @param initialCapacity
* the initial capacity of the map.
* @throws IllegalArgumentException
* if the initial capacity is less than zero.
*/
public OpenLongFloatHashMap(int initialCapacity) {
this(initialCapacity, defaultMinLoadFactor, defaultMaxLoadFactor);
}
/**
* Constructs an empty map with the specified initial capacity and the
* specified minimum and maximum load factor.
*
* @param initialCapacity
* the initial capacity.
* @param minLoadFactor
* the minimum load factor.
* @param maxLoadFactor
* the maximum load factor.
* @throws IllegalArgumentException
* if
*
* initialCapacity < 0 || (minLoadFactor < 0.0 || minLoadFactor >= 1.0) || (maxLoadFactor <= 0.0 || maxLoadFactor >= 1.0) || (minLoadFactor >= maxLoadFactor)
* .
*/
public OpenLongFloatHashMap(int initialCapacity, double minLoadFactor, double maxLoadFactor) {
setUp(initialCapacity, minLoadFactor, maxLoadFactor);
}
/**
* Assigns the result of a function to each value;
* v[i] = function(v[i]).
*
* @param function
* a function object taking as argument the current association's
* value.
*/
public void assign(cern.colt.function.tfloat.FloatFunction function) {
// specialization for speed
if (function instanceof cern.jet.math.tfloat.FloatMult) { // x[i] = mult*x[i]
float multiplicator = ((cern.jet.math.tfloat.FloatMult) function).multiplicator;
if (multiplicator == 1)
return;
if (multiplicator == 0) {
clear();
return;
}
for (int i = table.length; i-- > 0;) {
if (state[i] == FULL)
values[i] *= multiplicator;
}
} else { // the general case x[i] = f(x[i])
for (int i = table.length; i-- > 0;) {
if (state[i] == FULL)
values[i] = function.apply(values[i]);
}
}
}
/**
* Clears the receiver, then adds all (key,value) pairs of other
* values to it.
*
* @param other
* the other map to be copied into the receiver.
*/
public void assign(AbstractLongFloatMap other) {
if (!(other instanceof OpenLongFloatHashMap)) {
super.assign(other);
return;
}
OpenLongFloatHashMap source = (OpenLongFloatHashMap) other;
OpenLongFloatHashMap copy = (OpenLongFloatHashMap) source.copy();
this.values = copy.values;
this.table = copy.table;
this.state = copy.state;
this.freeEntries = copy.freeEntries;
this.distinct = copy.distinct;
this.lowWaterMark = copy.lowWaterMark;
this.highWaterMark = copy.highWaterMark;
this.minLoadFactor = copy.minLoadFactor;
this.maxLoadFactor = copy.maxLoadFactor;
}
/**
* Removes all (key,value) associations from the receiver. Implicitly calls
* trimToSize().
*/
public void clear() {
new ByteArrayList(this.state).fillFromToWith(0, this.state.length - 1, FREE);
// new FloatArrayList(values).fillFromToWith(0, state.length-1, 0); //
// delta
/*
* if (debug) { for (int i=table.length; --i >= 0; ) { state[i] = FREE;
* table[i]= Long.MAX_VALUE; values[i]= Float.NaN; } }
*/
this.distinct = 0;
this.freeEntries = table.length; // delta
trimToSize();
}
/**
* Returns a deep copy of the receiver.
*
* @return a deep copy of the receiver.
*/
public Object clone() {
OpenLongFloatHashMap copy = (OpenLongFloatHashMap) super.clone();
copy.table = copy.table.clone();
copy.values = copy.values.clone();
copy.state = copy.state.clone();
return copy;
}
/**
* Returns true if the receiver contains the specified key.
*
* @return true if the receiver contains the specified key.
*/
public boolean containsKey(long key) {
return indexOfKey(key) >= 0;
}
/**
* Returns true if the receiver contains the specified value.
*
* @return true if the receiver contains the specified value.
*/
public boolean containsValue(float value) {
return indexOfValue(value) >= 0;
}
/**
* Ensures that the receiver can hold at least the specified number of
* associations without needing to allocate new internal memory. If
* necessary, allocates new internal memory and increases the capacity of
* the receiver.
*
* This method never need be called; it is for performance tuning only.
* Calling this method before put()ing a large number of
* associations boosts performance, because the receiver will grow only once
* instead of potentially many times and hash collisions get less probable.
*
* @param minCapacity
* the desired minimum capacity.
*/
public void ensureCapacity(int minCapacity) {
if (table.length < minCapacity) {
int newCapacity = nextPrime(minCapacity);
rehash(newCapacity);
}
}
/**
* Applies a procedure to each key of the receiver, if any. Note: Iterates
* over the keys in no particular order. Subclasses can define a particular
* order, for example, "sorted by key". All methods which can be
* expressed in terms of this method (most methods can) must
* guarantee to use the same order defined by this method, even
* if it is no particular order. This is necessary so that, for example,
* methods keys and values will yield association pairs,
* not two uncorrelated lists.
*
* @param procedure
* the procedure to be applied. Stops iteration if the procedure
* returns false, otherwise continues.
* @return false if the procedure stopped before all keys where
* iterated over, true otherwise.
*/
public boolean forEachKey(LongProcedure procedure) {
for (int i = table.length; i-- > 0;) {
if (state[i] == FULL)
if (!procedure.apply(table[i]))
return false;
}
return true;
}
/**
* Applies a procedure to each (key,value) pair of the receiver, if any.
* Iteration order is guaranteed to be identical to the order used by
* method {@link #forEachKey(LongProcedure)}.
*
* @param procedure
* the procedure to be applied. Stops iteration if the procedure
* returns false, otherwise continues.
* @return false if the procedure stopped before all keys where
* iterated over, true otherwise.
*/
public boolean forEachPair(final LongFloatProcedure procedure) {
for (int i = table.length; i-- > 0;) {
if (state[i] == FULL)
if (!procedure.apply(table[i], values[i]))
return false;
}
return true;
}
/**
* Returns the value associated with the specified key. It is often a good
* idea to first check with {@link #containsKey(long)} whether the given key
* has a value associated or not, i.e. whether there exists an association
* for the given key or not.
*
* @param key
* the key to be searched for.
* @return the value associated with the specified key; 0 if no
* such key is present.
*/
public float get(long key) {
int i = indexOfKey(key);
if (i < 0)
return 0; // not contained
return values[i];
}
/**
* @param key
* the key to be added to the receiver.
* @return the index where the key would need to be inserted, if it is not
* already contained. Returns -index-1 if the key is already
* contained at slot index. Therefore, if the returned index < 0,
* then it is already contained at slot -index-1. If the returned
* index >= 0, then it is NOT already contained and should be
* inserted at slot index.
*/
protected int indexOfInsertion(long key) {
final long tab[] = table;
final byte stat[] = state;
final int length = tab.length;
final int hash = HashFunctions.hash(key) & 0x7FFFFFFF;
int i = hash % length;
int decrement = hash % (length - 2); // float hashing, see
// http://www.eece.unm.edu/faculty/heileman/hash/node4.html
// int decrement = (hash / length) % length;
if (decrement == 0)
decrement = 1;
// stop if we find a removed or free slot, or if we find the key itself
// do NOT skip over removed slots (yes, open addressing is like that...)
while (stat[i] == FULL && tab[i] != key) {
i -= decrement;
// hashCollisions++;
if (i < 0)
i += length;
}
if (stat[i] == REMOVED) {
// stop if we find a free slot, or if we find the key itself.
// do skip over removed slots (yes, open addressing is like that...)
// assertion: there is at least one FREE slot.
int j = i;
while (stat[i] != FREE && (stat[i] == REMOVED || tab[i] != key)) {
i -= decrement;
// hashCollisions++;
if (i < 0)
i += length;
}
if (stat[i] == FREE)
i = j;
}
if (stat[i] == FULL) {
// key already contained at slot i.
// return a negative number identifying the slot.
return -i - 1;
}
// not already contained, should be inserted at slot i.
// return a number >= 0 identifying the slot.
return i;
}
/**
* @param key
* the key to be searched in the receiver.
* @return the index where the key is contained in the receiver, else
* returns -1.
*/
protected int indexOfKey(long key) {
final long tab[] = table;
final byte stat[] = state;
final int length = tab.length;
final int hash = HashFunctions.hash(key) & 0x7FFFFFFF;
int i = hash % length;
int decrement = hash % (length - 2); // float hashing, see
// http://www.eece.unm.edu/faculty/heileman/hash/node4.html
// int decrement = (hash / length) % length;
if (decrement == 0)
decrement = 1;
// stop if we find a free slot, or if we find the key itself.
// do skip over removed slots (yes, open addressing is like that...)
// assertion: there is at least one FREE slot.
while (stat[i] != FREE && (stat[i] == REMOVED || tab[i] != key)) {
i -= decrement;
// hashCollisions++;
if (i < 0)
i += length;
}
if (stat[i] == FREE)
return -1; // not found
return i; // found, return index where key is contained
}
/**
* @param value
* the value to be searched in the receiver.
* @return the index where the value is contained in the receiver, returns
* -1 if the value was not found.
*/
protected int indexOfValue(float value) {
final float val[] = values;
final byte stat[] = state;
for (int i = stat.length; --i >= 0;) {
if (stat[i] == FULL && val[i] == value)
return i;
}
return -1; // not found
}
/**
* Returns the first key the given value is associated with. It is often a
* good idea to first check with {@link #containsValue(float)} whether there
* exists an association from a key to this value. Search order is
* guaranteed to be identical to the order used by method
* {@link #forEachKey(LongProcedure)}.
*
* @param value
* the value to search for.
* @return the first key for which holds get(key) == value; returns
* Long.MIN_VALUE if no such key exists.
*/
public long keyOf(float value) {
// returns the first key found; there may be more matching keys,
// however.
int i = indexOfValue(value);
if (i < 0)
return Long.MIN_VALUE;
return table[i];
}
/**
* Fills all keys contained in the receiver into the specified list. Fills
* the list, starting at index 0. After this call returns the specified list
* has a new size that equals this.size(). Iteration order is
* guaranteed to be identical to the order used by method
* {@link #forEachKey(LongProcedure)}.
*
* This method can be used to iterate over the keys of the receiver.
*
* @param list
* the list to be filled, can have any size.
*/
public void keys(LongArrayList list) {
list.setSize(distinct);
long[] elements = list.elements();
long[] tab = table;
byte[] stat = state;
int j = 0;
for (int i = tab.length; i-- > 0;) {
if (stat[i] == FULL)
elements[j++] = tab[i];
}
}
/**
* Fills all pairs satisfying a given condition into the specified lists.
* Fills into the lists, starting at index 0. After this call returns the
* specified lists both have a new size, the number of pairs satisfying the
* condition. Iteration order is guaranteed to be identical to the
* order used by method {@link #forEachKey(LongProcedure)}.
*
* Example:
*
*
* LongFloatProcedure condition = new LongFloatProcedure() { // match even keys only
* public boolean apply(long key, float value) { return key%2==0; }
* }
* keys = (8,7,6), values = (1,2,2) --> keyList = (6,8), valueList = (2,1)
* </tt>
*
*
* @param condition
* the condition to be matched. Takes the current key as first
* and the current value as second argument.
* @param keyList
* the list to be filled with keys, can have any size.
* @param valueList
* the list to be filled with values, can have any size.
*/
public void pairsMatching(final LongFloatProcedure condition, final LongArrayList keyList,
final FloatArrayList valueList) {
keyList.clear();
valueList.clear();
for (int i = table.length; i-- > 0;) {
if (state[i] == FULL && condition.apply(table[i], values[i])) {
keyList.add(table[i]);
valueList.add(values[i]);
}
}
}
/**
* Associates the given key with the given value. Replaces any old
* (key,someOtherValue) association, if existing.
*
* @param key
* the key the value shall be associated with.
* @param value
* the value to be associated.
* @return true if the receiver did not already contain such a key;
* false if the receiver did already contain such a key -
* the new value has now replaced the formerly associated value.
*/
public boolean put(long key, float value) {
int i = indexOfInsertion(key);
if (i < 0) { // already contained
i = -i - 1;
// if (debug) if (this.state[i] != FULL) throw new LongernalError();
// if (debug) if (this.table[i] != key) throw new LongernalError();
this.values[i] = value;
return false;
}
if (this.distinct > this.highWaterMark) {
int newCapacity = chooseGrowCapacity(this.distinct + 1, this.minLoadFactor, this.maxLoadFactor);
/*
* System.out.print("grow rehashing "); System.out.println("at
* distinct="+distinct+", capacity="+table.length+" to
* newCapacity="+newCapacity+" ...");
*/
// System.out.println("Rehashing in put");
rehash(newCapacity);
return put(key, value);
}
this.table[i] = key;
this.values[i] = value;
if (this.state[i] == FREE)
this.freeEntries--;
this.state[i] = FULL;
this.distinct++;
if (this.freeEntries < 1) { // delta
int newCapacity = chooseGrowCapacity(this.distinct + 1, this.minLoadFactor, this.maxLoadFactor);
rehash(newCapacity);
// System.out.println("Rehashing in put");
}
return true;
}
/**
* Rehashes the contents of the receiver into a new table with a smaller or
* larger capacity. This method is called automatically when the number of
* keys in the receiver exceeds the high water mark or falls below the low
* water mark.
*/
protected void rehash(int newCapacity) {
int oldCapacity = table.length;
// if (oldCapacity == newCapacity) return;
if (newCapacity <= this.distinct)
throw new InternalError();
// if (debug) check();
long oldTable[] = table;
float oldValues[] = values;
byte oldState[] = state;
long newTable[] = new long[newCapacity];
float newValues[] = new float[newCapacity];
byte newState[] = new byte[newCapacity];
this.lowWaterMark = chooseLowWaterMark(newCapacity, this.minLoadFactor);
this.highWaterMark = chooseHighWaterMark(newCapacity, this.maxLoadFactor);
this.table = newTable;
this.values = newValues;
this.state = newState;
this.freeEntries = newCapacity - this.distinct; // delta
for (int i = oldCapacity; i-- > 0;) {
if (oldState[i] == FULL) {
long element = oldTable[i];
int index = indexOfInsertion(element);
newTable[index] = element;
newValues[index] = oldValues[i];
newState[index] = FULL;
}
}
// if (debug) check();
}
/**
* Removes the given key with its associated element from the receiver, if
* present.
*
* @param key
* the key to be removed from the receiver.
* @return true if the receiver contained the specified key,
* false otherwise.
*/
public boolean removeKey(long key) {
int i = indexOfKey(key);
if (i < 0)
return false; // key not contained
// if (debug) if (this.state[i] == FREE) throw new LongernalError();
// if (debug) if (this.state[i] == REMOVED) throw new LongernalError();
this.state[i] = REMOVED;
// this.values[i]=0; // delta
// if (debug) this.table[i]=Long.MAX_VALUE; // delta
// if (debug) this.values[i]=Float.NaN; // delta
this.distinct--;
if (this.distinct < this.lowWaterMark) {
int newCapacity = chooseShrinkCapacity(this.distinct, this.minLoadFactor, this.maxLoadFactor);
/*
* if (table.length != newCapacity) { System.out.print("shrink
* rehashing "); System.out.println("at distinct="+distinct+",
* capacity="+table.length+" to newCapacity="+newCapacity+" ..."); }
*/
rehash(newCapacity);
// System.out.println("Rehashing in removeKey");
}
return true;
}
/**
* Initializes the receiver.
*
* @param initialCapacity
* the initial capacity of the receiver.
* @param minLoadFactor
* the minLoadFactor of the receiver.
* @param maxLoadFactor
* the maxLoadFactor of the receiver.
* @throws IllegalArgumentException
* if
*
* initialCapacity < 0 || (minLoadFactor < 0.0 || minLoadFactor >= 1.0) || (maxLoadFactor <= 0.0 || maxLoadFactor >= 1.0) || (minLoadFactor >= maxLoadFactor)
* .
*/
protected void setUp(int initialCapacity, double minLoadFactor, double maxLoadFactor) {
int capacity = initialCapacity;
super.setUp(capacity, minLoadFactor, maxLoadFactor);
capacity = nextPrime(capacity);
if (capacity == 0)
capacity = 1; // open addressing needs at least one FREE slot at
// any time.
this.table = new long[capacity];
this.values = new float[capacity];
this.state = new byte[capacity];
// memory will be exhausted long before this pathological case happens,
// anyway.
this.minLoadFactor = minLoadFactor;
if (capacity == PrimeFinder.largestPrime)
this.maxLoadFactor = 1.0f;
else
this.maxLoadFactor = maxLoadFactor;
this.distinct = 0;
this.freeEntries = capacity; // delta
// lowWaterMark will be established upon first expansion.
// establishing it now (upon instance construction) would immediately
// make the table shrink upon first put(...).
// After all the idea of an "initialCapacity" implies violating
// lowWaterMarks when an object is young.
// See ensureCapacity(...)
this.lowWaterMark = 0;
this.highWaterMark = chooseHighWaterMark(capacity, this.maxLoadFactor);
}
/**
* Trims the capacity of the receiver to be the receiver's current size.
* Releases any superfluous internal memory. An application can use this
* operation to minimize the storage of the receiver.
*/
public void trimToSize() {
// * 1.2 because open addressing's performance exponentially degrades
// beyond that point
// so that even rehashing the table can take very long
int newCapacity = nextPrime((int) (1 + 1.2 * size()));
if (table.length > newCapacity) {
rehash(newCapacity);
}
}
/**
* Fills all values contained in the receiver into the specified list. Fills
* the list, starting at index 0. After this call returns the specified list
* has a new size that equals this.size(). Iteration order is
* guaranteed to be identical to the order used by method
* {@link #forEachKey(LongProcedure)}.
*
* This method can be used to iterate over the values of the receiver.
*
* @param list
* the list to be filled, can have any size.
*/
public void values(FloatArrayList list) {
list.setSize(distinct);
float[] elements = list.elements();
float[] val = values;
byte[] stat = state;
int j = 0;
for (int i = stat.length; i-- > 0;) {
if (stat[i] == FULL)
elements[j++] = val[i];
}
}
}