How to find minimum number of jumps to reach the end of the array in O(n) time
The time complexity of the solution proposed on the site is linear because you only iterate over the array once. The algorithm avoids the inner iteration of my proposed solution by using some clever tricks.
The variable maxReach stores at all time the maximal reachable position in the array. jump stores the amount of jumps necessary to reach that position. step stores the amount of steps we can still take (and is initialized with the amount of steps at the first array position)
During the iteration, the above values are updated as follows:
First we test whether we have reached the end of the array, in which case we just need to return the jump variable.
Next we update the maximal reachable position. This is equal to the maximum of maxReach and i+A[i] (the number of steps we can take from the current position).
We used up a step to get to the current index, so steps has to be decreased.
If no more steps are remaining (i.e. steps=0, then we must have used a jump. Therefore increase jump. Since we know that it is possible somehow to reach maxReach, we initialize the steps to the amount of steps to reach maxReach from position i.
public class Solution {
public int jump(int[] A) {
if (A.length <= 1)
return 0;
int maxReach = A[0];
int step = A[0];
int jump = 1;
for (int i = 1; i < A.length; i++) {
if (i == A.length - 1)
return jump;
if (i + A[i] > maxReach)
maxReach = i + A[i];
step--;
if (step == 0) {
jump++;
step = maxReach - i;
}
}
return jump;
}
}
Example:
int A[] = {1, 3, 5, 8, 9, 2, 6, 7, 6, 8, 9}
int maxReach = A[0]; // A[0]=1, so the maximum index we can reach at the moment is 1.
int step = A[0]; // A[0] = 1, the amount of steps we can still take is also 1.
int jump = 1; // we will always need to take at least one jump.
/*************************************
* First iteration (i=1)
************************************/
if (i + A[i] > maxReach) // 1+3 > 1, we can reach further now!
maxReach = i + A[i] // maxReach = 4, we now know that index 4 is the largest index we can reach.
step-- // we used a step to get to this index position, so we decrease it
if (step == 0) {
++jump; // we ran out of steps, this means that we have made a jump
// this is indeed the case, we ran out of the 1 step we started from. jump is now equal to 2.
// but we can continue with the 3 steps received at array position 2.
steps = maxReach-i // we know that by some combination of 2 jumps, we can reach position 4.
// therefore in the current situation, we can minimaly take 3
// more steps to reach position 4 => step = 3
}
/*************************************
* Second iteration (i=2)
************************************/
if (i + A[i] > maxReach) // 2+5 > 4, we can reach further now!
maxReach = i + A[i] // maxReach = 7, we now know that index 7 is the largest index we can reach.
step-- // we used a step so now step = 2
if (step==0){
// step
}
/*************************************
* Second iteration (i=3)
************************************/
if (i + A[i] > maxReach) // 3+8 > 7, we can reach further now!
maxReach = i + A[i] // maxReach = 11, we now know that index 11 is the largest index we can reach.
step-- // we used a step so now step = 1
if (step==0){
// step
}
/*************************************
* Third iteration (i=4)
************************************/
if (i + A[i] > maxReach) // 4+9 > 11, we can reach further now!
maxReach = i + A[i] // maxReach = 13, we now know that index 13 is the largest index we can reach.
step-- // we used a step so now step = 0
if (step == 0) {
++jump; // we ran out of steps, this means that we have made a jump.
// jump is now equal to 3.
steps = maxReach-i // there exists a combination of jumps to reach index 13, so
// we still have a budget of 9 steps
}
/************************************
* remaining iterations
***********************************
// nothing much changes now until we reach the end of the array.
My suboptimal algorithm which works in O(nk) time with n the number of elements in the array and k the largest element in the array and uses an internal loop over array[i]. This loop is avoided by the below algorithm.
Code
public static int minimum_steps(int[] array) {
int[] min_to_end = new int[array.length];
for (int i = array.length - 2; i >= 0; --i) {
if (array[i] <= 0)
min_to_end[i] = Integer.MAX_VALUE;
else {
int minimum = Integer.MAX_VALUE;
for (int k = 1; k <= array[i]; ++k) {
if (i + k < array.length)
minimum = Math.min(min_to_end[i+k], minimum);
else
break;
}
min_to_end[i] = minimum + 1;
}
}
return min_to_end[0];
}
Here is the basic intuition regarding the above problem's greedy approach and rest are the code requirements.
Given array is Input: a[] = {1, 3, 5, 8, 9, 2, 6, 7, 6, 8, 9}.
Now we start from the 1st element i.e i=0 and a[i] = 1. So seeing this we can take at most a jump of size 1, so since we don't have any other choice so we make this step happen.
Currently we are at i=1 and a[i]=3. So we currently can make a jump of size 3, but instead we consider all possible jumps we can make from the current location and attain the maximum distance which is within bounds(of the array). So what are our choices? we can make a jump of 1 step, or 2 steps or 3 steps. So we investigate from current location for each size jumps and choose the one which can take us maximum further into the array.
Once we have decided, which one we stick to, we take that jump size and update our number of jumps made so far and also where we can reach at most and how many steps we have now to decide our next move. And that's it. This is how finally we select the best option linearly traversing the array. So this is the basic idea of the algo you might be looking for, next is to code it for the algorithm to work. Cheers!
Hope somebody time travels and finds the intuition helpful !! :) :P "Years late to the party" @Vasilescu Andrei - well said. Sometimes it feels to me that we are time travelers.
Years late to the party , but here is another O(n) solution that made sense for me.
/// <summary>
///
/// The actual problem is if it's worth not to jump to the rightmost in order to land on a value that pushes us further than if we jumped on the rightmost.
///
/// However , if we approach the problem from the end, we go end to start,always jumping to the leftmost
///
/// with this approach , these is no point in not jumping to the leftmost from end to start , because leftmost will always be the index that has the leftmost leftmost :) , so always choosing leftmost is the fastest way to reach start
///
/// </summary>
/// <param name="arr"></param>
static void Jumps (int[] arr)
{
var LeftMostReacher = new int[arr.Length];
//let's see , for each element , how far back can it be reached from
LeftMostReacher[0] = -1; //the leftmost reacher of 0 is -1
var unReachableIndex = 1; //this is the first index that hasn't been reached by anyone yet
//we use this unReachableIndex var so each index's leftmost reacher is the first that was able to jump to it . Once flagged by the first reacher , new reachers can't be the leftmost anymore so they check starting from unReachableIndex
// this design insures that the inner loop never flags the same index twice , so the runtime of these two loops together is O(n)
for (int i = 0; i < arr.Length; i++)
{
int maxReach = i + arr[i];
for (; unReachableIndex <= maxReach && unReachableIndex < arr.Length; unReachableIndex++)
{
LeftMostReacher[unReachableIndex] = i;
}
}
// we just go back from the end and then reverse the path
int index = LeftMostReacher.Length - 1;
var st = new Stack<int>();
while (index != -1)
{
st.Push(index);
index = LeftMostReacher[index];
}
while (st.Count != 0)
{
Console.Write(arr[st.Pop()] + " ");
}
Console.WriteLine();
}
static void Main ()
{
var nrs = new[] { 1, 3, 5, 8, 9, 2, 6, 7, 6, 8, 9 };
Jumps(nrs);
}