Recursive algorithm stops working after certain depth

Question

I was exploring Fork/Join framework and its possible speed benefits through factorial counting, when discovered that my sequential recursive algorithm breaks at a certain point. To be precise, when I try to count 46342! the result from RecursiveCounter is wrong, but before that value it is always right and is the same that result from ParallelCounter and LoopCounter. Does anyone have an idea why that may happen?

Here are the classes:

RecursiveCounter:

public class RecursiveCounter implements FactorialCounter, RangeFactorialCounter {
    @Override
    public BigInteger count(int number) {
        return count(1, number);
    }

    @Override
    public BigInteger count(int from, int to) {
        int middle = (from + to) >> 1;
        BigInteger left;
        BigInteger right;
        if (middle - from > 1)
            left = count(from, middle);
        else
            left = new BigInteger(String.valueOf(from * middle));
        if (to - (middle + 1) > 1)
            right = count(middle + 1, to);
        else
            right = to == middle + 1 ? new BigInteger(String.valueOf(to)) : new BigInteger(String.valueOf((middle + 1) * to));
        return left.multiply(right);
    }
}

LoopCounter:

public class LoopCounter implements FactorialCounter, RangeFactorialCounter {
    @Override
    public BigInteger count(final int number) {
        return count(1, number);
    }

    @Override
    public BigInteger count(final int from, final int to) {
        BigInteger result = new BigInteger("1");
        for (int i = from; i < to + 1; i++) {
            result = result.multiply(new BigInteger(String.valueOf(i)));
        }
        return result;
    }
}

A RecursiveTask for ParallelCounter:

public class FactorialTask extends RecursiveTask<BigInteger> {
    private static final int THRESHOLD = 1000;
    private RangeFactorialCounter iterativeCounter = new LoopCounter();

    private Integer firstVal;
    private Integer lastVal;

    public FactorialTask(Integer from, Integer to) {
        super();
        this.firstVal = from;
        this.lastVal = to;
    }

    @Override
    protected BigInteger compute() {
        return count(firstVal, lastVal);
    }

    private BigInteger count(int from, int to) {
        int middle = (from + to) >> 1;
        if (to - from > THRESHOLD) {
            List<FactorialTask> tasks = Arrays.asList(new FactorialTask(from, middle), new FactorialTask(middle + 1, to));
            tasks.forEach(RecursiveTask::fork);
            return tasks.stream()
                    .map(RecursiveTask::join)
                    .map(BigInteger.class::cast)
                    .reduce(new BigInteger("1"), BigInteger::multiply);
        } else {
            return (from != to) ? countSequential(from, to) : new BigInteger(String.valueOf(from));
        }
    }

    private BigInteger countSequential(int from, int to) {
        return iterativeCounter.count(from, to);
    }
}

Answer 1

In RecursiveCounter, from * middle and (middle + 1) * to might overflow, you need use BigInteger to manipulate them:

...
left = BigInteger.valueOf(from).multiply(BigInteger.valueOf(middle));
...
right = to == middle + 1 ? BigInteger.valueOf(to) : BigInteger.valueOf(to).multiply(BigInteger.valueOf(middle + 1));

Then you can get the same result in RecursiveCounter and LoopCounter:

LoopCounter loopCounter = new LoopCounter();
RecursiveCounter recursiveCounter = new RecursiveCounter();
BigInteger loopResult = loopCounter.count(46342);
BigInteger recursiveResult = recursiveCounter.count(46342);
System.out.println(loopResult.equals(recursiveResult)); // true

Answer 2

This happens because of numeric overflow of an int, not because of recursive depth, which is nicely controlled by your algorithm, which needs O(log₂n) stack frames for recursion.

The overflow happens here:

new BigInteger(String.valueOf((middle + 1) * to))

When to is high, this value can overflow int. Specifically, when middle approaches to in the second "leg" of recursive invocations, you multiply 46341 by 46342, which yields -2147432674 due to an overflow (demo).

You can fix this by using only BigInteger for "payload" multiplication, i.e.

BigInteger.valueOf(middle+1).multiply(BigInteger.valueOf(to))