View more

View more

View more

Image of the Day Submit

IOTD | Top Screenshots

The latest, straight to your Inbox.

Subscribe to GameDev.net Direct to receive the latest updates and exclusive content.

Array Find Largest Recursively

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

8 replies to this topic

#1Cuajin  Members

Posted 18 April 2012 - 04:28 PM

Hi. I understand recursion is not good for this sort of operation, but I am trying to learn.

I want to find the position of the largest int in an array recursively.

myList = {5, 2, 3 , 1, 22, 4, 9} ;

public int findIndex(int[] myList, int size)
{
if (size == 1)
return size-1;

return myList[size-1] > findIndex(myList, size-1) ? size-1 : size ;
}


My code produces this output: position 6. Which is wrong.

How do I do this?

#2SiCrane  Moderators

Posted 18 April 2012 - 04:33 PM

I think you're confusing the array's value at a given index with the index itself.

#3Cuajin  Members

Posted 18 April 2012 - 04:54 PM

I believe I need to check the values at a different position. Then return the larger of the two but as a position instead of a number. I am confused though.

#4SiCrane  Moderators

Posted 18 April 2012 - 05:00 PM

Let me be more explicit. Here:
myList[size-1] > findIndex(myList, size-1)

You're comparing a value in the list against the return value of findIndex(), which is an index.

#5ApochPiQ  Moderators

Posted 18 April 2012 - 05:03 PM

This is the algorithm you probably want:

Given a start index and end index (end = one past the last element in a range)
- If range is one element long (i.e. end = start + 1), return start index
- LargestIndex = Recursively call the function with start+1, end
- Compare the value at LargestIndex with the value at start
- If start is larger, return start; else return LargestIndex

That should work, I think.
Wielder of the Sacred Wands

#6Nypyren  Members

Posted 18 April 2012 - 05:25 PM

General layout of the algorithm looks fine to me, it's identical to a typical functional-language linked list recursion (with the "head" at the last slot, meaning you only need one variable to keep track of position, since 0 will be the end)

I think you just need to change:

findIndex(myList, size-1)

to

myList[findIndex(myList, size-1)]

#7rip-off  Moderators

Posted 19 April 2012 - 04:16 AM

During each call, you return either size or size-1. Thus, it is not possible for your implementation to return an early index in the array - regardless of the recursion.

When designing such a generic algorithm, you should think about corner cases. For example, what should the function do if passed an empty list? An "easy" answer is just to reject such a case using an exception.

A stylistic note is that the name of the function is ambiguous, and the parameter name "size" is ambiguous (as Java arrays know their size, it appears redundant). Most programmers are used to the convention that searches start from the beginning. Thus, I would propose a function signature(s) like the following:
public class Example
{
public static int indexOfLargest(int [] array, int startIndex)
{
// ...
}

public static int indexOfLargest(int [] array)
{
return indexOfLargest(array, 0);
}
}

The most general function is indexOfLargest(int [] array, int begin, int end), if you really want maximum flexibility.

#8Álvaro  Members

Posted 19 April 2012 - 05:41 AM

The version that takes begin and end also allows for a different algorithm, where you divide the incoming data into two parts and recursively compute the maximum of each one. This only uses O(log(N)) stack space as opposed O(N).

#9samoth  Members

Posted 20 April 2012 - 03:17 AM

This won't teach you about recursion, but why not use a priority queue, which is the right tool for the task? Learning to use the right tool is important to learn as well.
Or, sort an array of indices (recursively if you want), which lets you find the largest and next largest element in the original array by iterating over it linearly.

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.