Блог пользователя risingStark

Автор risingStark, история, 7 недель назад, По-английски

I was asked this question in on online test. So, this question is not possible to submit on some judge. But I found a similar question somewhere else. The question goes like this...

Problem

There are multiple delivery centers and delivery warehouses all over the world! The world is represented by a number line from -10⁹ to 10⁹. There are n delivery centers, with the i-th one at location center[i]. A location x is called a suitable location for a warehouse if it is possible to bring ALL the products to that point AND then return them to their original locations by traveling a distance of no more than d.

At any one time, products can be brought from one delivery center and placed at point x. Given the positions of n delivery centers, calculate the number of suitable locations in the world. That is, calculate the number of points x on the number line (-10⁹ ≤ x ≤ 10⁹) where the travel distance required to bring all the products to that point AND then return them to their original locations is less than or equal to d.

Note: The distance between point x and center[i] is |x — center[i]|, their absolute difference.

Sample test cases

Case 1:
n=3, arr = [0, 1, -2], d = 8

output = 3

Case 2:
n = 4, arr = [2, -4, 0, 3], d = 22

output = 5

Constraints

1 <= n <= 1e5
-1e9 <= arr[i] <= 1e9
0 <= d <= 1e15

Approach

This is the approach I tried. Here’s the idea in plain English:

We want to find a range of positions on the number line where the total travel distance from all the delivery centers to a warehouse is less than or equal to a given distance, called d.

Binary search for the smallest point (xmin): — We start by trying to find the smallest position, xmin, where the sum of the distances from all the delivery centers to xmin is less than or equal to D. — We calculate the total distance by summing the absolute differences between xmin and each delivery center. — If we find such a point xmin, it marks the left end of our range.

Binary search for the largest point (xmax): — Similarly, we look for the largest position, xmax, where the sum of the distances from all the delivery centers to xmax is less than or equal to D. — If we find such a point xmax, it marks the right end of our range.

Range of valid positions: — The positions between xmin and xmax (including both) are all valid locations for the warehouse where the total travel distance from the delivery centers is within the limit D.

By finding xmin and xmax, we define a range [xmin, xmax] on the number line, and every position within this range is a possible location for the warehouse. This range contains all the valid points where the travel distance requirement is satisfied.

Code

bool cal(vi arr, ll d, ll x){
	ll sum = 0;
	for(ll i:arr){
		sum+=abs(x-i);
	}
	return (2*sum)<=d;
}

void solve(){
        int n;
        cin>>n;
        vi arr(n);
        cin>>arr;

	ll d;
	cin>>d;

	// sort(all(arr));

	ll l1 = -1e9, h1 = 1e9;
	ll minx = h1;
	while(l1<=h1){
	  ll m = (l1+h1)/2;
	  if(cal(arr, d, m)){
	    minx = min(minx, m);
			h1=m-1;
	  }else{
	    l1 = m+1;
	  }
	}
	l1=-1e9, h1=1e9;
	ll maxx = l1;
	while(l1<=h1){
	  ll m = (l1+h1)/2;
	  if(cal(arr, d, m)){
	    maxx = max(m, maxx);
			l1=m+1;
	  }else{
	    h1 = m-1;
	  }
	}

	if(maxx<minx)cout<<0;
	else cout<<(maxx-minx+1);
}

Issue (asking help)

But then while submitting, I found out that approach fails on some test cases. I figured that it makes a "V", where before that minx, everything is false, after that maxx, everything is false. between them its true.

FFFFFFFFF......FFFTTTTT....TTTTFFFFFFF......FFFFFF

I have done binary search where there's only 1 inflexion point. How do I tackle this problem with two inflexion points? Is there no way rather than to have multiple if else of checks to see if we are in the right inflexion point or the left inflexion point? What if there were 3 inflexion points? How would we approach that?

  • Проголосовать: нравится
  • +4
  • Проголосовать: не нравится

»
7 недель назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Auto comment: topic has been updated by risingStark (previous revision, new revision, compare).

»
7 недель назад, # |
Rev. 3   Проголосовать: нравится 0 Проголосовать: не нравится

In case of these kind of problems where there are 2 inflexion points the logic of "Rotated sorted array search" comes into play.

For Example lets take a array like : arr = [5 ,6 , 7 , 1 , 2 , 3 , 4 ]

and we have to find if 7 is present or not.

Now when we generally apply binary search we just check for one half right like if the array is [1 , 2 , 3 , 4 ,5] then if our answer is greater than 3 then it will always lie on the right half but in case of these sorted arrays you need to check both the halfs like for the example array:

the first mid is = (0 + 6) / 2 = 3; which is 1 now check the right half as 1 <= 4 we can clearly say the right side is sorted so we can check as 4 is less than 7 that means that 7 is not there in the second half so the answer is in right half so our l = 0 and h = mid — 1 = 2;

Now the new mid = (0 + 2) / 2 = 1; the val = 6 now again repeat the same procedure and you will eventually get the answer for your array lets say we modify our array like for val >= 6 the answer will be true and for less than 5 the answer is false so the array redueces to : Arr = [F , T , T , F , F , F , F]

Now do the very same thing I explained in the first half and you will get your answer(Like the postion of the first true value or the true value exists or not.).

Here is a complete code for this :

int search(vector<int>& arr, int n, int k)
{
    int lo = 0 , hi = n - 1;
    while(lo <= hi) {
        int mid = (lo + hi) >> 1;
        if(arr[mid] == k) return mid;
        if(arr[lo] <= arr[mid]) {
            if(arr[lo] <= k && k <= arr[mid]) {
                hi = mid - 1;
            } else {
                lo = mid + 1;
            }
        } else {
            if(arr[mid] <= k && k <= arr[hi]) {
                lo = mid + 1;
            } else {
                hi = mid - 1;
            }
        }
    }
    return -1;
}

Although I have explained the logic a bit I would highly suggest to read about this from a blog or something

Hope this helps !!

»
7 недель назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Maybe try ternary search since you're dealing with a convex function

»
7 недель назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

What does it mean "I figured that it makes a "V", where before that minx, everything is false". SOmeone posted this solution

https://leetcode.com/playground/824sRnhU