all programming  

  Sep 02, 2020

Day 2

When fame was acquired briefly.

The Question

Given an array of integers, find out whether there are two distinct indices i and j in the array such that the absolute difference between nums[i] and nums[j] is at most t and the absolute difference between i and j is at most k.


Example 1:

Input: nums = [1,2,3,1], k = 3, t = 0
Output: true

Example 2:

Input: nums = [1,0,1,1], k = 1, t = 2
Output: true

Example 3:

Input: nums = [1,5,9,1,5,9], k = 2, t = 3
Output: false



  • 0 <= nums.length <= 2 * 104
  • -231 <= nums[i] <= 231 - 1
  • 0 <= k <= 104
  • 0 <= t <= 231 - 1

The Process

Note a few ideas that will help you understand this problem and make the optimal time complexity approach easier for you:

  • Consider the edge cases. What are the edge cases? What happens when the array is empty, singleton etc. And what about when k is 0? Consider edge cases that you know are unlikely to be included in a general solution, and take care of those cases right away.
  • Then, you understand that for any element x in nums, you’re actually looking for its partner. This partner element is at most k hops away from x, and doesn’t differ from x by more than t.
  • So, if you happen to have k elements in your hand, and look at a (k + 1)th element, you ask “does this (k + 1)th guy find a partner in the k I’m already holding on to?” If not, well guess what? The oldest element in the k ones you have can no more find its partner so you can get rid of it and let (k + 1)th guy get in.
  • For an example, let’s consider the first one given i.e. nums = [1,2,3,1] | k = 3 | t = 0.
  • Now, you’re holding on to the first k elements i.e. [1,2,3]. Your supposed loop considers the next element i.e. [..., 1]. Now if this 1 has a partner in [1,2,3], great, we’re done. But if not, then there’s no problem in throwing the first [1...], because well, all the elements you see now are going to be more than k hops away from him. So you end up with [2,3,1].
  • That was the basic processing idea.
  • Finally, you understand that upon processing each element x, if the largest number smaller than x or the smallest number greater than x, in the k elements you already have, produce something bounded by t, you have an answer.
  • Now it’s your turn to research what data structure can hold onto elements such that it can yeild a ceiling/floor relative to some number efficiently (clearly it will have to keep its elements in a sorted manner…)
  • (P.S. think a bit about why ceiling/floor and not just max/min)

The Code

    public boolean containsNearbyAlmostDuplicate(int[] nums, int k, int t) {
        int len = nums.length;

        // return false when given parameters makes two distinct indices impossible
        if (len <= 1 || k <= 0) {
            return false;

        * Important, question definition:
        *   Absolute diff between nums[i] and nums[j] <= t
        *   Absolute diff between indices i and j <= k

        TreeSet<Long> set = new TreeSet();
        for (int i = 0; i < nums.length; i++) {
            Long number = new Long(nums[i]);
            Long ceil = set.ceiling(number);
            Long floor = set.floor(number);

            if (ceil != null && ceil - number <= t) {
                return true;

            if (floor != null && number - floor <= t) {
                return true;

            if (set.size() > k) {
                Long oldest = new Long(nums[i - k]);

        return false;

The Fame

For some reason, my explanation for this question reached a lot of people (~1.7K), and got me my first earned Reputation stars. Check out my proof of brief fame on Leetcode!