Stack problems in Java

Question

I'm having some problems with the stack in java... I'm implementing quicksort using an ArrayList- I'll attach my full code at the end but here are the relevant bits (Keep in mind I've been debugging the hell out of this for a few hours with absolutely no clue what was going wrong, so where you see things done in odd/etc ways it's more than likely because I was trying to check for bugs there):

The call that is apparently being executed too many times:

            QuickSort(numArray, m+1, right);

It fails here:

//Swap pivot and left
        numArray.set(pivot, leftVal);

I'm getting output:

Starting sort number: [1] : RandomPivot [ON] : Sorting [RANDOM]
Sorted: 2496 elements.
Took: 53 milliseconds.

Starting sort number: [2] : RandomPivot [ON] : Sorting [RANDOM]
Sorted: 4988 elements.
Took: 25 milliseconds.

Starting sort number: [3] : RandomPivot [ON] : Sorting [RANDOM]
Sorted: 7478 elements.
Took: 49 milliseconds.

Starting sort number: [4] : RandomPivot [ON] : Sorting [RANDOM]
Sorted: 9953 elements.
Took: 85 milliseconds.

Starting sort number: [5] : RandomPivot [ON] : Sorting [RANDOM]
Sorted: 12416 elements.
Took: 131 milliseconds.

Starting sort number: [1] : RandomPivot [ON] : Sorting [SORTED]
Sorted: 2497 elements.
Took: 1 milliseconds.

Starting sort number: [2] : RandomPivot [ON] : Sorting [SORTED]
Sorted: 4984 elements.
Took: 1 milliseconds.

Starting sort number: [3] : RandomPivot [ON] : Sorting [SORTED]
Sorted: 7482 elements.
Took: 2 milliseconds.

Starting sort number: [4] : RandomPivot [ON] : Sorting [SORTED]
Sorted: 9950 elements.
Took: 2 milliseconds.

Starting sort number: [5] : RandomPivot [ON] : Sorting [SORTED]
Sorted: 12424 elements.
Took: 2 milliseconds.

Starting sort number: [1] : RandomPivot [ON] : Sorting [REVERSE SORTED]
Sorted: 2494 elements.
Took: 2 milliseconds.

Starting sort number: [2] : RandomPivot [ON] : Sorting [REVERSE SORTED]
Sorted: 4988 elements.
Took: 10 milliseconds.

Starting sort number: [3] : RandomPivot [ON] : Sorting [REVERSE SORTED]
Sorted: 7470 elements.
Took: 35 milliseconds.

Starting sort number: [4] : RandomPivot [ON] : Sorting [REVERSE SORTED]
Sorted: 9962 elements.
Took: 50 milliseconds.

Starting sort number: [5] : RandomPivot [ON] : Sorting [REVERSE SORTED]
Sorted: 12419 elements.
Took: 65 milliseconds.

Starting sort number: [1] : RandomPivot [OFF] : Sorting [RANDOM]
Sorted: 2497 elements.
Took: 5 milliseconds.

Starting sort number: [2] : RandomPivot [OFF] : Sorting [RANDOM]
Sorted: 4984 elements.
Took: 54 milliseconds.

Starting sort number: [3] : RandomPivot [OFF] : Sorting [RANDOM]
Sorted: 7473 elements.
Took: 47 milliseconds.

Starting sort number: [4] : RandomPivot [OFF] : Sorting [RANDOM]
Sorted: 9958 elements.
Took: 80 milliseconds.

Starting sort number: [5] : RandomPivot [OFF] : Sorting [RANDOM]
Sorted: 12419 elements.
Took: 130 milliseconds.

Starting sort number: [1] : RandomPivot [OFF] : Sorting [SORTED]
Sorted: 2498 elements.
Took: 11 milliseconds.

Starting sort number: [2] : RandomPivot [OFF] : Sorting [SORTED]
Sorted: 4991 elements.
Took: 44 milliseconds.

Starting sort number: [3] : RandomPivot [OFF] : Sorting [SORTED]
Sorted: 7474 elements.
Took: 97 milliseconds.

Starting sort number: [4] : RandomPivot [OFF] : Sorting [SORTED]
Exception in thread "main" java.lang.StackOverflowError
at java.util.ArrayList.set(Unknown Source)
at Sorter.QuickSort(Sorter.java:64)
at Sorter.QuickSort(Sorter.java:107)
at Sorter.QuickSort(Sorter.java:107)
at Sorter.QuickSort(Sorter.java:107)
at Sorter.QuickSort(Sorter.java:107)
    (on and on and on and on)

Testing and it always fails once I get past ~7500 as the size of my ArrayList. Always fails at "ArrayList.set()" and I haven't the slightest clue as to why. As you can see- all other sorted types work fine with numbers over this amount, but with the sorted list I will have to call "m+1, right" n times, where n is the size of the list.

I've tried this:

    if(m-1>left && m-1<right)
        QuickSort(numArray, left, m-1);
    if(m+1<right && m+1>left)
        QuickSort(numArray, m+1, right);

but I get the same error either way, and I figured it's easier to understand if it's left out.

If I can increase the stack size, somehow, I might be able to postpone the error, which would allow me to at least analyze the different methods.

I'm running this code via eclipse, if that makes any difference. Thanks! (Full code now)

import java.util.ArrayList;
import java.util.HashSet;
import java.util.Random;
import java.util.Set;
import java.util.TreeSet;


public class Sorter {

//STATE:
public boolean test=false;
public boolean randomPivot=false;
Random r = new Random();
public int sortMethod=1;

public static int RANDOM=1;
public static int SORTED=2;
public static int REVERSE_SORTED=3;


public Sorter(){ }


public ArrayList<Integer> SlowSort(ArrayList<Integer> numArray){

    //Add "infinity" to the end
    numArray.add(Integer.MAX_VALUE);


    //TIME AND RUN QUICKSORT
    long startTime = System.currentTimeMillis();
    numArray=QuickSort(numArray, 0, numArray.size()-2);
    long stopTime = System.currentTimeMillis();
    long runTime = stopTime - startTime;

    //Remove "infinity" from the end
    numArray.remove(numArray.size()-1);

    //TODO: Printing effs it up? idk
    //      printArrayWithMessage("Sort Finished.", numArray);

    //PRINT COMPLETION MESSAGE AND RETURN
    System.out.println("Sorted: "+numArray.size()+" elements.\nTook: " + runTime+" milliseconds.\n");
    return numArray;
}


public ArrayList<Integer> QuickSort(ArrayList<Integer> numArray, int left, int right){
    if(left>=right){
        return numArray;
    }

    //Correctly Initialize Pivot
    int pivot=0;
    if(randomPivot){
        pivot = r.nextInt(right-left);
    }
    pivot+=left;

    //Swap pivot and left
    Integer temp = numArray.get(pivot);
//      System.out.println(numArray.size()+" "+pivot);
    int leftVal=numArray.get(left);
    numArray.set(pivot, leftVal);
    pivot=temp;
    numArray.set(left, pivot);

    Integer m=0;

    //REPEAT:
    while(true){
        int i=left+1;
        int j=right+1;

        while(numArray.get(i).intValue()<pivot){
            i++;
        }
        while(numArray.get(j).intValue()>pivot){
            j--;
        }

        if(i<j){
            //Swap i and j
            if(test) printArrayWithMessage("[SWAP] (i="+i+") and (j="+j+")", numArray);

            Integer a=numArray.get(i);
            numArray.set(i, numArray.get(j));
            numArray.set(j, a);
        }

        if(i>j){
            //Swap pivot and j
            if(test) printArrayWithMessage("[SWAP] (j="+j+") and (pivot="+left+")", numArray);

            numArray.set(left, numArray.get(j));
            numArray.set(j, pivot);
            m=j;

            break;
        }
    }

    if(test) printArrayWithMessage("Iteration Finished... Left: "+left+"  Right: "+right, numArray);


        QuickSort(numArray, left, m-1);
        QuickSort(numArray, m+1, right);

    return numArray;
}

public void benchmark(){

    for(int i=1;i<6;i++){
        //CREATE BLANK DATA STRUCTURES
        ArrayList<Integer> numList = new ArrayList<Integer>();
        Set<Integer> doublesFilter;

        if(sortMethod==RANDOM){
            doublesFilter = new HashSet<Integer>();
        }else{
            doublesFilter = new TreeSet<Integer>();
        }
        //FILL ARRAYLIST WITH UNIQUE NUMBERS
        for(int j=0;j<2500*i;j++){
            int num=r.nextInt(1000000);
            if(doublesFilter.add(num)){
                if(sortMethod==RANDOM){
                    numList.add(num);
                }
            }
        }
        if(sortMethod==SORTED){
            numList.add(0);
            numList.ensureCapacity(doublesFilter.size());
            numList.addAll(doublesFilter);
            numList.remove(0);
        }
        else if(sortMethod==REVERSE_SORTED){
            numList.add(0);
            numList.ensureCapacity(doublesFilter.size());
            numList.addAll(((TreeSet<Integer>) doublesFilter).descendingSet());
            numList.remove(0);
        }


        System.out.println("Starting sort number: ["+i+"] : "+getMode());
        numList=SlowSort(numList);
    }
}


public void printArrayWithMessage(String s, ArrayList<Integer> list){
    System.out.println(s);
    System.out.println(list);
    System.out.println();
}

public String getMode(){

    String rPivot="OFF";
    if(randomPivot) rPivot="ON";


    String sortMode="UNDEFINED";
    if(sortMethod==1)sortMode="RANDOM";
    if(sortMethod==2)sortMode="SORTED";
    if(sortMethod==3)sortMode="REVERSE SORTED";

    return "RandomPivot ["+rPivot+"] : "+"Sorting ["+sortMode+"]";
}

}

I think I'll wait for the DVD, but in the meantime, try debugging it using a debugger and step through your code. It's likely that you terminating condition is being retried, eg you're passing an empty list in and that being retried — Bohemian, Jun 05 '11 at 01:53

btreat · Accepted Answer · 2011-06-05T02:04:12.790

3

As you noted, Quicksort has worst case performance when provided a sorted array and ends up making O(n) recursive calls because the partitioning only removes one element during each subdivision step. In the other cases where the array is not sorted, the partitioning is more effective, so you end up with O(lgN) recursive calls. The O(n) recursive calls exceed the max number of stack frames where O(lgN) calls won't.

Edit (additional note): One of benefits and intents of using a random pivot in quick sort is that it ensures that the size of the partitions / sub-problems are O(n) rather than the O(1) behavior in the worst case The latter case (no random pivot + sorted input) is where you're seeing the stack overflow error.

edited Jun 05 '11 at 02:04

answered Jun 05 '11 at 01:56

btreat

1,554
8
10

In other words, your implementation is probably correct, but you've hit the test case where it fails for the smallest input size (no random pivot + sorted input). – btreat Jun 05 '11 at 02:14
There might be some means of increasing the number of stack frames available to you, but I think you'll have better luck decreasing the size of your inputs for the purposes of completing your analysis. – btreat Jun 05 '11 at 02:21

score 2 · Answer 2 · edited May 23 '17 at 11:50

2

A recursive implementation will always result in a stackoverflow under the "right" circumstances (i.e. large dataset). Also, have a look at some answers to similar questions:

Got stackoverflowerror when using quickSort, can I increase the stack and the heap?

Some issues in implementing QuickSort in java

edited May 23 '17 at 11:50

Community

1
1

answered Jun 05 '11 at 02:02

Stefan

736
4
18

Stack problems in Java

2 Answers2