Question

I am doing some personal research and came across a need to efficiently modify data in a (possibly very large) tree of data. The data consists of simple data, just an integer or possibly a small object. While parsing the tree, I will need to modify all similar objects (but only objects within the subtree). This is a bit confusing to explain, so I have added an image as an example to help.

In the above image, letters represent a value for the data (identical letters represent identical data), and the numbers in parenthesis are a unique value so I can refer to them.

So lets say I am parsing the tree and I am currently at node 2. I have determined that I need to modify the value of ‘a’ and change it to ‘g’, but I only want to modify the current node and all children. (so 2 and 9 will change to ‘g’, while 1 and 7 remain ‘a’). I could recursively search all children and change it manually, but the tree may be very large. If I wanted to change ALL of the values of a, I could simply store data as a double pointer, and change the value that the pointer points to. So my question is, is there any known method to modify the data in the way I want, in possibly constant time?

Answer 1

The way I would do it would be on insertion to the tree.

Make a class for node

class node
{
    node* dataParent;
    CustomType* value;
    // put regular tree pointers here
}

Then when you insert as you traverse, dataParent points to the last parent that had identical data. If there are no parents that are identical it will be null and value will not be null. If it has a parent that is identical, value will be null and something will be in Custom parent. Your retrieval will not be constant time though, you will have to traverse until value is not null.

This is the best way I can think of. It is at least log n i think. Its been awhile.

One thing is insertion will be costly.