A good Sorted List for Java

Question

I\'m looking for a good sorted list for java. Googling around give me some hints about using TreeSet/TreeMap. But these components is lack of one thing: random access to an

Answer 1

Generally you can't have constant time look up and log time deletions/insertions, but if you're happy with log time look ups then you can use a SortedList.

Not sure if you'll trust my coding but I recently wrote a SortedList implementation in Java, which you can download from http://www.scottlogic.co.uk/2010/12/sorted_lists_in_java/. This implementation allows you to look up the i-th element of the list in log time.

Answer 2

Phuong:

Sorting 40,000 random numbers:

0.022 seconds

import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.Random;


public class test
{
    public static void main(String[] args)
    {
        List<Integer> nums = new ArrayList<Integer>();
        Random rand = new Random();
        for( int i = 0; i < 40000; i++ )
        {
            nums.add( rand.nextInt(Integer.MAX_VALUE) );
        }

        long start = System.nanoTime();
        Collections.sort(nums);
        long end = System.nanoTime();

        System.out.println((end-start)/1e9);
    }
}   

Since you rarely need sorting, as per your problem statement, this is probably more efficient than it needs to be.

Answer 3

To test the efficiancy of earlier awnser by Konrad Holl, I did a quick comparison with what I thought would be the slow way of doing it:

package util.collections;

import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.Iterator;
import java.util.List;
import java.util.ListIterator;

/**
 *
 * @author Earl Bosch
 * @param <E> Comparable Element
 *
 */
public class SortedList<E extends Comparable> implements List<E> {

    /**
     * The list of elements
     */
    private final List<E> list = new ArrayList();

    public E first() {
        return list.get(0);
    }

    public E last() {
        return list.get(list.size() - 1);
    }

    public E mid() {
        return list.get(list.size() >>> 1);
    }

    @Override
    public void clear() {
        list.clear();
    }

    @Override
    public boolean add(E e) {
        list.add(e);
        Collections.sort(list);
        return true;
    }

    @Override
    public int size() {
        return list.size();
    }

    @Override
    public boolean isEmpty() {
        return list.isEmpty();
    }

    @Override
    public boolean contains(Object obj) {
        return list.contains((E) obj);
    }

    @Override
    public Iterator<E> iterator() {
        return list.iterator();
    }

    @Override
    public Object[] toArray() {
        return list.toArray();
    }

    @Override
    public <T> T[] toArray(T[] arg0) {
        return list.toArray(arg0);
    }

    @Override
    public boolean remove(Object obj) {
        return list.remove((E) obj);
    }

    @Override
    public boolean containsAll(Collection<?> c) {
        return list.containsAll(c);
    }

    @Override
    public boolean addAll(Collection<? extends E> c) {

        list.addAll(c);
        Collections.sort(list);
        return true;
    }

    @Override
    public boolean addAll(int index, Collection<? extends E> c) {
        throw new UnsupportedOperationException("Not supported.");
    }

    @Override
    public boolean removeAll(Collection<?> c) {
        return list.removeAll(c);
    }

    @Override
    public boolean retainAll(Collection<?> c) {
        return list.retainAll(c);
    }

    @Override
    public E get(int index) {
        return list.get(index);
    }

    @Override
    public E set(int index, E element) {
        throw new UnsupportedOperationException("Not supported.");
    }

    @Override
    public void add(int index, E element) {
        throw new UnsupportedOperationException("Not supported.");
    }

    @Override
    public E remove(int index) {
        return list.remove(index);
    }

    @Override
    public int indexOf(Object obj) {
        return list.indexOf((E) obj);
    }

    @Override
    public int lastIndexOf(Object obj) {
        return list.lastIndexOf((E) obj);
    }

    @Override
    public ListIterator<E> listIterator() {
        return list.listIterator();
    }

    @Override
    public ListIterator<E> listIterator(int index) {
        return list.listIterator(index);
    }

    @Override
    public List<E> subList(int fromIndex, int toIndex) {
        throw new UnsupportedOperationException("Not supported.");
    }

}

Turns out its about twice as fast! I think its because of SortedLinkList slow get - which make's it not a good choice for a list.

Compared times for same random list:

• SortedLinkList : 15731.460
• SortedList : 6895.494
• ca.odell.glazedlists.SortedList : 712.460
• org.apache.commons.collections4.TreeList : 3226.546

Seems glazedlists.SortedList is really fast...

Answer 4

You need no sorted list. You need no sorting at all.

I need to add/remove keys from the list when object is added / removed from database.

But not immediately, the removal can wait. Use an ArrayList containing the ID's all alive objects plus at most some bounded percentage of deleted objects. Use a separate HashSet to keep track of deleted objects.

private List<ID> mostlyAliveIds = new ArrayList<>();
private Set<ID> deletedIds = new HashSet<>();

I want to randomly select few dozens of element from the whole list.

ID selectOne(Random random) {
    checkState(deletedIds.size() < mostlyAliveIds.size());
    while (true) {
        int index = random.nextInt(mostlyAliveIds.size());
        ID id = mostlyAliveIds.get(index);
        if (!deletedIds.contains(ID)) return ID;
    }
}

Set<ID> selectSome(Random random, int count) {
    checkArgument(deletedIds.size() <= mostlyAliveIds.size() - count);
    Set<ID> result = new HashSet<>();
    while (result.size() < count) result.add(selectOne(random));
}

For maintaining the data, do something like

void insert(ID id) {
    if (!deletedIds.remove(id)) mostlyAliveIds.add(ID);
} 

void delete(ID id) {
    if (!deletedIds.add(id)) {
         throw new ImpossibleException("Deleting a deleted element);
    }
    if (deletedIds.size() > 0.1 * mostlyAliveIds.size()) {
        mostlyAliveIds.removeAll(deletedIds);
        deletedIds.clear();
    }
}

The only tricky part is the insert which has to check if an already deleted ID was resurrected.

The delete ensures that no more than 10% of elements in mostlyAliveIds are deleted IDs. When this happens, they get all removed in one sweep (I didn't check the JDK sources, but I hope, they do it right) and the show goes on.

With no more than 10% of dead IDs, the overhead of selectOne is no more than 10% on the average.

I'm pretty sure that it's faster than any sorting as the amortized complexity is O(n).