This question is based off of this resource http://algs4.cs.princeton.edu/14analysis.
Can someone break down why Exercise 6 letter b is linear? The outer loop seems to be increasing i by a factor of 2 each time, so I would assume it was logarithmic...
From the link:
int sum = 0;
for (int n = N; n > 0; n /= 2)
for (int i = 0; i < n; i++)
sum++;
This is a geometric series.
The inner loops runs i iterations per iteration of the outer loop, and the outer loop decreases by half each time.
So, summing it up gives you:
n + n/2 + n/4 + ... + 1
This is geometric series, with r=1/2 and a=n - that converges to a/(1-r)=n/(1/2)=2n, so:
T(n) <= 2n
And since 2n is in O(n) - the algorithm runs in linear time.
This is a perfect example to see that complexity is NOT achieved by multiplying the complexity of each nested loop (that would have got you O(nlogn)), but by actually analyzing how many iterations are needed.
Yes its simple
See the value of n is decreasing by half each time and I runs n times.
So for the first time i goes from 1 to n
next time 0 to n/2
and hence 0 to n/k on kth term.
Now total time inner loop would run = Log(n)
So its a GP the number of times i is running.
with terms
n,n/2,n/4,n/8....0
so we can find the sum of the GP
2^(long(n) +1)-1 / (2-1)
2^(long(n)+1) = n
hence n-1/(1) = >O(n)
Related
What is the big-O for the following code :
y=1;
x=3;
for(int i =1 ; i < =n ; i*=2)
for(int j =1; j<= i * i; j++)
if (i % j == 0)
for(int k = 1; k<=j; k++)
y=y*x;
My Thoughts :
Looking at another similar questions I think the inner most loop is O(n) and the first loop is O(log (n))..as for the middle its O(n^2)
so the overall result would be O(log(n)*n^3)
Is my answer and way of thinking right ? I'm new to this so i hope i can get some help explaning how this loops work.
the most inner loop will run j time if i % j == 0. As the middle loop will run i^2 times, only when j < i it will be possible to satisfy the specified condition. Hence, among i^2 iteration of the middle loop, at least i^2 - i times, the condition will not be satisfied.
Suppose we denote the number of divisors of i with tau(i), among j < i only tau(i) times the condition will satisfy that means the total complexity of the most inner loop is equal to the sum of divisions of i which is at most 77/16 i (see this post for the proof).
Hence, the total complexity of the middle loop with the inner loop is at most (i^2 - i) + (i - tau(i)) + 77/16 i = i^2 + 77/16 i - tau(i).
We also know that the tau(i) is in O(i^(1/loglog(i))) (see the proof here). Now, to find the complexity of the whole loop, we need to sum the last expression for i = 1, 2, 4, ..., n. As we desire to find the asymptotic complexity and we have a sum here, we can ignore the lower powers of i. Therefore, the time complexity of the whole loop is 1 + 2^2 + (2^2)^2 + ... + (2^2)^log(n) = ((2^2)^(log(n)+1)-1)/(2^2-1) = Theta(n^2) (a geometric sum with factor of 2^2 and log(n) items).
In sum, the higher time complexity analysis for the specified code is Theta(n^2) which is also in O(n^2) as well.
I found this example problem on the internet that I just cannot understand how the author came to their conclusion.
sum1 = 0;
for(k=1; k<=n; k*=2) // Do log n times
for (j=1; j<=n; j++) // Do n times
sum1++;`
sum2 = 0;
for (k=1; k<=n; k*=2) // Do log n times
for (j=1; j<=k; j++) // Do k times
sum2++;
I understand that the running time for the first loop is O(n) = nlog(n), but the author claims that for the second loop, the running time is O(n) = n.
Why is that?
The closest I can get to an answer is:
O(n) = k * log(n)
k = 2^i
O(n) = 2^i * log(n) ----> this is where I get stuck
I'm guessing some property of logarithms is used, but I can't figure out which one. Can someone point me in the right direction?
Thanks.
In the second example, the complexity is sum_j 2^j, i.e. the total number of operations in the inner loop.
As 2^j <= n, there are logn terms.
This sum is equal to 2^{jmax+1} - 1, with 2^jmax roughly (<=) equal to n.
Then, effectively, a complexity O(2n) = O(n).
sum2++ is executed 1+2+4+8+...+K times, where K is the largest power of 2 less than or equal to n. That sum is equal to 2K-1.
Since n/2 < K <= n (because K is the largest power of 2 less than or equal to n), the number of iterations is between n-1 and 2n-1. That's Theta(n) if you want to express it in asymptotic notation.
I have this code :
int fun(int n)
{
int count = 0;
for (int i = n; i > 0; i /= 2)
for (int j = 0; j < i; j++)
count += 1;
return count;
}
The time complexity of this code can be thought of as O(n) because O(n+n/2+n/4+...) = O(n)
By that logic, the time complexity of this snippet can also be argued to be O(n) :
for(i = 1; i < n; i *= 2)
//O(1) statements
Since O(1+2+4+..+n/4+n/2) = O(n). But since the loop runs log(n) times, it can be log(n) too.
Why is the former one not : log(n) times the outer loop * log(n) times the inner loop so, log(n)log(n)
What am I doing wrong ?
The first snippet has the outer loop that executes O(log n) times, and each iteration the inner loop executes O(i) times. If you sum any number of terms of the form n / 2^k, you'll get O(n).
The second piece of code has O(log n) iterations of O(1) operations, and sum of logarithmic amount of constants is still logarithmic.
In the first example, you don't have an O(1) statement inside your loop, as you have for (int j = 0; j < i; j++) count += 1. If in your second example you put the same inner loop of the first example, you are back to the same complexity. The first loop is not O(n*log(n)); this is easy to demonstrate because you can find an upper bound in O(2n) which is equivalent to O(n).
The time complexity of the 2nd one should not be calculated as a series O(1+2+4+..+n/4+n/2) = O(n), because it is not that series.
Notice the first one. It is being calculated as a series because one counts the number of times the inner for loop executes and then add all of them (series) to get the final time complexity.
When i=n inner for loop executes n times
When i=(n/2) inner for loop executes n/2 times
When i=(n/4) inner for loop executes n/4 times
and so on..
But in the second one, there is no series to add. It just comes to a formula (2^x) = n, which evaluates to x = logn.
(2^x) = n this formula can be obtained by noticing that i starts with 1, and when it becomes 2 it is multiplied by 2 until it reaches n.
So one needs to find out how many times 2 needs to be multiplied by 2 to reach n.
Thus the formula (2^x) = n, and then solve for x.
I need some help finding the complexity or Big-O of this code. If someone could explain what the Big-O of every loop would be that would be great. I think the outter loop would just be O(n) but the inner loop I'm not sure, how does the *=2 effect it?
k = 1;
do
{
j = 1;
do
{
...
j *= 2;
} while (j < n);
k++;
} while (k < n);
The outer loop runs O(n) times, since k starts at 1 and needs to be incremented n-1 times to become equal to 1.
The inner loop runs O(lg(n)) times. This is because on the m-th execcution of the loop, j = 0.5 * 2^(m).
The loop breaks when n = j = 0.5 * 2^m. Rearranging that, we get m = lg(2n) = O(lg(n)).
Putting the two loops together, the total code complexity is O(nlg(n)).
Logarithms can be tricky, but generally, whenever you see something being repeatedly being multiplied or divided by a constant factor, you can guess that the complexity of your algorithm involves a term that is either logarithmic or exponential.
That's why binary search, which repeatedly divides the size of the list it searches in half, is also O(lg(n)).
The inner loop always runs from j=1 to j=n.
For simplicity, let's assume that n is a power of 2 and that the inner loop runs k times.
The values of j for each of the k iterations are,
j = 1
j = 2
j = 4
j = 8
....
j = n
// breaks from the loop
which means that 2^k = n or k = lg(n)
So, each time, it runs for O(lg(n)) times.
Now, the outer loop is executed O(n) times, starting from k=1 to k=n.
Therefore, every time k is incremented, the inner loop runs O(lg(n)) times.
k=1 Innerloop runs for : lg(n)
k=2 Innerloop runs for : lg(n)
k=3 Innerloop runs for : lg(n)
...
k=n Innerloop runs for : lg(n)
// breaks from the loop
Therefore, total time taken is n*lg(n)
Thus, the time complexity of this is O(nlg(n))
As I understand, the complexity of an algorithm is a maximum number of operations performed while sorting. So, the complexity of Bubble sort should be a sum of arithmmetic progression (from 1 to n-1), not n^2.
The following implementation counts number of comparisons:
public int[] sort(int[] a) {
int operationsCount = 0;
for (int i = 0; i < a.length; i++) {
for(int j = i + 1; j < a.length; j++) {
operationsCount++;
if (a[i] > a[j]) {
int temp = a[i];
a[i] = a[j];
a[j] = temp;
}
}
}
System.out.println(operationsCount);
return a;
}
The ouput for array with 10 elements is 45, so it's a sum of arithmetic progression from 1 to 9.
So why Bubble sort's complexity is n^2, not S(n-1) ?
This is because big-O notation describes the nature of the algorithm. The major term in the expansion (n-1) * (n-2) / 2 is n^2. And so as n increases all other terms become insignificant.
You are welcome to describe it more precisely, but for all intents and purposes the algorithm exhibits behaviour that is of the order n^2. That means if you graph the time complexity against n, you will see a parabolic growth curve.
Let's do a worst case analysis.
In the worst case, the if (a[i] > a[j]) test will always be true, so the next 3 lines of code will be executed in each loop step. The inner loop goes from j=i+1 to n-1, so it will execute Sum_{j=i+1}^{n-1}{k} elementary operations (where k is a constant number of operations that involve the creation of the temp variable, array indexing, and value copying). If you solve the summation, it gives a number of elementary operations that is equal to k(n-i-1). The external loop will repeat this k(n-i-1) elementary operations from i=0 to i=n-1 (ie. Sum_{i=0}^{n-1}{k(n-i-1)}). So, again, if you solve the summation you see that the final number of elementary operations is proportional to n^2. The algorithm is quadratic in the worst case.
As you are incrementing the variable operationsCount before running any code in the inner loop, we can say that k (the number of elementary operations executed inside the inner loop) in our previous analysis is 1. So, solving Sum_{i=0}^{n-1}{n-i-1} gives n^2/2 - n/2, and substituting n with 10 gives a final result of 45, just the same result that you got by running the code.
Worst case scenario:
indicates the longest running time performed by an algorithm given any input of size n
so we will consider the completely backward list for this worst-case scenario
int[] arr= new int[]{9,6,5,3,2};
Number of iteration or for loops required to completely sort it = n-1 //n - number of elements in the list
1st iteration requires (n-1) swapping + 2nd iteration requires (n-2) swapping + ……….. + (n-1)th iteration requires (n-(n-1)) swapping
i.e. (n-1) + (n-2) + ……….. +1 = n/2(a+l) //sum of AP
=n/2((n-1)+1)=n^2/2
so big O notation = O(n^2)