I have been practicing analyzing algorithms lately. I feel like I have a pretty good understanding of analyzing non-recursive algorithms but I am unsure, and have just begun to embark on a full understanding of recursive algorithm as well. Although, I have not had a formal check on my methods and if what I have been doing is indeed correct
Would it be too much to ask if someone could check a few algorithms that I have implemented and analyzed and see if my understanding is along the right lines or if I am completely off.
here:
1)
sum = 0;
for (i = 0; i < n; i++){
for (j = 0; j < i*i; j++){
if (j % i == 0) {
for (k = 0; k < j; k++){
sum++;
}
}
}
}
My analysis of this one was O(n^5) due to:
Sum(i = 0 to n)[Sum(j = 0 to i^2)[Sum(k = 0 to j) of 1]]
which evaluated to:
(1/2)(n^5/5 + n^4/2 + n^3/3 - n/30) + (1/2)(n^3/3 + n^2/2 + n/6) + (1/2)(n^3/3 + n^2/2 + n/6) + n + 1.
Hence it is O(n^5)
Is this correct as an evaluation of the summations of the loops?
a triple summation. I have assumed that the if statement will always pass for worse case complexity. Is this a correct assumption for worst case?
2)
int tonyblair (int n, int a) {
if (a < 12) {
for (int i = 0; i < n; i++){
System.out.println("*");
}
tonyblair(n-1, a);
} else {
for (int k = 0; k < 3000; k++){
for (int j = 0; j < nk; j++){
System.out.println("#");
}
}
}
}
My analysis of this algorithm is O(infinity) due to the infinite recursion in the if statement if it is assumed to be true, which would be the worst case. Although, for pure analysis, I analyzed if this were not true and the if statement would not run. I then got a complexity of O(nk) due to:
Sum(k = 0 to 3000)[Sum(j = 0 to nk) of 1]
which then evaluated to nk(3001) + 3001. Hence is O(nk), where k is not discarded due to it controlling the number of iterations of the loop.
Number 1
I can't tell how you've derived your formula. Usually adding terms happens when there are multiple steps in an algorithm, such as precomputing data and then looking up values from the data. Instead, nested for loops implies multiplication. Also, the worst case is the best case for this snippet of code, because given a value of n, sum will be the same at the end.
To find the complexity, we want to find the number of times that the inner loop is evaluated. Summations are often easy to solve if they go from 1 to n, so I'm going to drop the 0s from them later on. If i is 0, the middle loop won't run, and if j is 0, the inner loop won't run. We can rewrite the code equivalently as:
sum = 0;
for (i = 1; i < n; i++)
{
for (j = 1; j < i*i; j++)
{
if (j % i == 0)
{
for (k = 0; k < j; k++)
{
sum++;
}
}
}
}
I could make my life harder by forcing the outer loop to start at 2, but I'm not going to. The outer loop now runs from 1 to n-1. The middle loop runs based on the current value of i, so we need to do a summation:
The middle for loop always goes to (i^2 - 1), and j will only be divisible by i for a total of (i - 1) times (i, i*2, i*3, ..., i*(i-2), i*(i-1)). With this, we get:
The middle loop then executes j times. The j in our summation is not the same as the j in the code though. The j in the summation represents each time the middle loop executes. Each time the middle loop executes, the j in the code will be (i * (number of executions so far)) = i * (the j in the summation). Therefore, we have:
We can move the i to in-between the two summations, as it is a constant for the inner summation. Then, the formula for the sum of 1 to n is well known: n*(n+1)/2. Because we are going to n - 1, we must subtract n out. This gives:
The summations for the sum of squares and the sum of cubes are also well known. Keeping in mind that we are only summing to n-1 in both cases, we must remember to subtract n^3 and n^2, respectively, and we get:
This is obviously n^4. If we solve it all the way, we get:
Number 2
For the last one, it is in fact O(infinity) if a < 12 because of the if statement. Well, technically everything is O(infinity), because Big-O only provides an upper bound on runtime. If a < 12, it is also omega(infinity) and theta(infinity). If only the else runs, then we have the summation from 1 to 2999 of i*n:
It's very important to notice that the summation from 1 to 2999 is a constant (it's 4498500). No matter how large a constant is, it's still a constant, and not dependent on n. We will end up throwing it out of the runtime calculations. Sometimes, when a theoretically fast algorithm has a large constant, it is practically slower than other algorithms that are theoretically slow. One example I can think of is Chazelle's linear time triangulation algorithm. No one has ever implemented it. In any case, we have 4498500 * n. This is theta(n):
Related
So many loops, I stuck at counting how many times the last loop runs.
I also don't know how to simplify summations to get big Theta. Please somebody help me out!
int fun(int n) {
int sum = 0
for (int i = n; i > 0; i--) {
for (int j = i; j < n; j *= 2) {
for (int k = 0; k < j; k++) {
sum += 1
}
}
}
return sum
}
Any problem has 2 stages:
You guess the answer
You prove it
In easy problems, step 1 is easy and then you skip step 2 or explain it away as "obvious". This problem is a bit more tricky, so both steps require some more formal thinking. If you guess incorrectly, you will get stuck at your proof.
The outer loop goes from n to 0, so the number of iterations is O(n). The middle loop is uncomfortable to analyze because its bounds depend on current value of i. Like we usually do in guessing O-rates, let's just replace its bounds to be from 1 to n.
for (int i = n; i > 0; i--) {
for (int j = 1; j < n; j *= 2) {
perform j steps
}
}
The run-time of this new middle loop, including the inner loop, is 1+2+4+...+n, or approximately 2*n, which is O(n). Together with outer loop, you get O(n²). This is my guess.
I edited the code, so I may have changed the O-rate when I did. So I must now prove that my guess is right.
To prove this, use the "sandwich" technique - edit the program in 2 different ways, one which makes its run-time smaller and one which makes its run-time greater. If you manage to make both new programs have the same O-rate, you will prove that the original code has the same O-rate.
Here is a "smaller" or "faster" code:
do n/2 iterations; set i=n/2 for each of them {
do just one iteration, where you set j = i {
perform j steps
}
}
This code is faster because each loop does less work. It does something like n²/4 iterations.
Here is a "greater" or "slower" code:
do n iterations; set i=n for each of them {
for (int j = 1; j <= 2 * n; j *= 2) {
perform j steps
}
}
I made the upper bound for the middle loop 2n to make sure its last iteration is for j=n or greater.
This code is slower because each loop does more work. The number of iterations of the middle loop (and everything under it) is 1+2+4+...+n+2n, which is something like 4n. So the number of iterations for the whole program is something like 4n².
We got, in a somewhat formal manner:
n²/4 ≤ runtime ≤ 4n²
So runtime = O(n²).
Here I use O where it should be Θ. O is usually defined as "upper bound", while sometimes it means "upper or lower bound, depending on context". In my answer O means "both upper and lower bound".
What's the big O of this?
for (int i = 1; i < n; i++) {
for (int j = 1; j < (i*i); j++) {
if (j % i == 0) {
for (int k = 0; k < j; k++) {
// Simple computation
}
}
}
}
Can't really figure it out. Inclined to say O(n^4 log(n)) but feel like i'm wrong here.
This is quite a confusing analysis, so let's break it down bit by bit to make sense of the calculations:
The outermost loop runs for n-1 iterations (since 1 ≤ i < n).
The next loop inside it makes (i² - 1) iterations for each index i of the outer loop (since 1 ≤ j < i²).
In total, this means the number of iterations for these two loops is equal to calculating the sum of (i²-1) for each 1 ≤ i < n. This is similar to computing the sum of the first n squares, and is order of magnitude of O(n³).
Note the modulo operator % takes constant time (O(1)) to compute, therefore checking the condition if (j % i == 0) for all iterations of these two loops will not affect the O(n³) runtime.
Now let's talk about the inner loop inside the conditional.
We are interested in seeing how many times (and for which values of j) this if condition evaluates to true, since this would dictate how many iterations the innermost loop will run.
Practically speaking, (j % i) will never equal 0 if j < i, so the second loop could actually be shortened to start from i rather than from 1, however this will not impact the Big-O upper bound of the algorithm.
Notice that for a given number i, (j % i == 0) if and only if i is a divisor of j. Since our range is (1 ≤ j < i²), there will be a total of (i-1) values of j for which this will be true, for any given i. If this is confusing, consider this example:
Let's assume i = 4. Then our index j would iterate through all values 1,..,15=i²,
and (j%i == 0) would be true for j = 4, 8, 12 - exactly (i - 1) values.
The innermost loop would therefore make a total of (12 + 8 + 4 = 24) iterations. Thus for a general index i, we would look for the sum: i + 2i + 3i + ... + (i-1)i to indicate the number of iterations the innermost loop would make.
And this could be generalized by calculating the sum of this arithmetic progression. The first value is i and the last value is (i-1)i, which results in a sum of (i³ - i²)/2 iterations of the k loop for every value of i. In turn, the sum of this for all values of i could be computed by calculating the sum of cubes and the sum of squares - for a total runtime of O(n⁴) iterations of the innermost loop (the k loop) for all values of i.
Thus in total, the runtime of this algorithm would be the total of both runtimes we calculated above. We checked the if statement O(n³) times and the innermost loop ran for O(n⁴), so assuming // Simple computation runs in constant time, our total runtime would come down to:
O(n³) + O(n⁴)*O(1) = O(n⁴)
Let us assume that i = 2.Then j can be [1,2,3].The "k" loop will run for j = 2 only.
Similarly for i=3,j can be[1,2,3,4,5,6,7,8].hence, k can run for j = 3,6. You can see a pattern here that for any value of i, the 'k' loop will run (i-1) times.The length of loops will be [i,2*i,3*i,....i*i].
Hence the time complexity of k loop is
=i+(2*i)+(3*i)+ ..... +(i*i)
=(i^2)(i+1)/2
Hence the final complexity will be
= (n^3)(n+3)/2
After looking at this question, this article, and several other questions, I have still not been able to find a general way to determine the computational complexity of algorithms with looping variables dependent on the parent loop's variable. For example,
for (int i = 0; i < n; i++) {
for (int j = i; j < n; j++) {
for (int k = i; k < j; k++) {
//one statement
}
}
}
I know that the first loop has a complexity of n, but the inner loops are confusing me. The second loop seems to be executed n-i times and the third loop seems to be executed j-i times. However, I'm not sure how to turn this into a regular Big-O statement. I don't think I can say O(n(n-i)(j-i)), so how can I get rid of the i and j variables here?
I know this is something on the order of n^3, but how can I show this? Do I need to use series?
Thanks for your help!
(If you were wondering, this is from a brute force implementation of the maximum sum contiguous subsequence problem.)
First loop hits N items on average.
Second loop hits N / 2 items on avarage
Third loop hits N / 4 items on average
O(N * N / 2 * N / 4) is about O((N^3)/8) is about O(N^3)
This is my first question on stackoverflow. I've been solving some exercises from "Algorithm design" by Goodrich, Tamassia. However, I'm quite clueless about this problem. Unusre where to start from and how to proceed. Any advice would be great. Here's the problem:
Boolean matrices are matrices such that each entry is 0 or 1, and matrix multiplication is performed by using AND for * and OR for +. Suppose we are given two NxN random Boolean matrices A and B, so that the probability that any entry
in either is 1, is 1/k. Show that if k is a constant, then there is an algorithm for multiplying A and B whose expected running time is O(n^2). What if k is n?
Matrix multiplication using the standard iterative approach is O(n3), because you have to iterate over n rows and n columns, and for each element do a vector multiply of one of the rows and one of the columns, which takes n multiplies and n-1 additions.
Psuedo code to multiply matrix a by matrix b and store in matrix c:
for(i = 0; i < n; i++)
{
for(j = 0; j < n; j++)
{
int sum = 0;
for(m = 0; m < n; m++)
{
sum += a[i][m] * b[m][j];
}
c[i][j] = sum;
}
}
For a boolean matrix, as specified in the problem, AND is used in
place of multiplication and OR in place of addition, so it becomes
this:
for(i = 0; i < n; i++)
{
for(j = 0; j < n; j++)
{
boolean value = false;
for(m = 0; m < n; m++)
{
value ||= a[i][m] && b[m][j];
if(value)
break; // early out
}
c[i][j] = value;
}
}
The thing to notice here is that once our boolean "sum" is true, we can stop calculating and early out of the innermost loop, because ORing any subsequent values with true is going to produce true, so we can immediately know that the final result will be true.
For any constant k, the number of operations before we can do this early out (assuming the values are random) is going to depend on k and will not increase with n. At each iteration there will be a (1/k)2 chance that the loop will terminate, because we need two 1s for that to happen and the chance of each entry being a 1 is 1/k. The number of iterations before terminating will follow a Geometric distribution where p is (1/k)2, and the expected number of "trials" (iterations) before "success" (breaking out of the loop) doesn't depend on n (except as an upper bound for the number of trials) so the innermost loop runs in constant time (on average) for a given k, making the overall algorithm O(n2). The Geometric distribution formula should give you some insight about what happens if k = n. Note that in the formula given on Wikipedia k is the number of trials.
I am reading some information on time complexity and I'm quite confused as to how the following time complexities are achieved and if there is a particular set of rules or methods for working this out?
1)
Input: int n
for(int i = 0; i < n; i++){
print("Hello World, ");
}
for(int j = n; j > 0; j--){
print("Hello World");
}
Tight: 6n + 5
Big O: O(n)
2)
Input: l = array of comparable items
Output: l = array of sorted items
Sort:
for(int i = 0; i < l.length; i++){
for(int j = 0; j < l.length; j++){
if(l{i} > l{j}){
} }
Swap(l{i},l{j});
}
return ls;
Worst Case Time Complexity: 4n2 +3n+2 = O(n2)
For a given algorithm, time complexity or Big O is a way to provide some fair enough estimation of "total elementary operations performed by the algorithm" in relationship with the given input size n.
Type-1
Lets say you have an algo like this:
a=n+1;
b=a*n;
there are 2 elementary operations in the above code, no matter how big your n is, for the above code a computer will always perform 2 operations, as the algo does not depend on the size of the input, so the Big-O of the above code is O(1).
Type-2
For this code:
for(int i = 0; i < n; i++){
a=a+i;
}
I hope you understand the Big-O in O(n), as elementary operation count directly depend on the size of n
Type-3
Now what about this code:
//Loop-1
for(int i = 0; i < n; i++){
print("Hello World, ");
}
//Loop-2
for(int i = 0; i < n; i++){
for(int j = 0; j < n; j++) {
x=x+j;
}
}
As you can see loop-1 is O(n) and loop-2 is O(n^2). So it feel like total complexity should be O(n)+O(n^2). But no, the time complexity of the above code is O(n^2). Why? Because we are trying to know the fair enough count of elementary operations performed by the algorithm for a given input size n, which will be comparatively easy to understand by another person. With this logic, O(n)+O(n^2) become O(n^2), or O(n^2)+O(n^3)+O(n^4) become O(n^4)!
Again, you may ask: But how? How all the lower power of Big-O become so insignificant as we add it with a higher power of Big-O, that we can completely omit them (lower powers) when we are describing the complexity of our algorithm to another human?
I will try show the reason for this case: O(n)+O(n^2)=O(n^2).
Lets say n=1000 then the exact count for O(n) is 1000 operations and the exact count for O(n^2) is 1000*1000=1000000, so O(n^2) is 1000 time bigger than O(n), which means your program will spend most of the execution time in O(n^2) and thus it is not worth to mention that your algorithm also has some O(n).
PS. Pardon my English :)
In the first example, the array has n elements, and you go through these elements Twice. The first time you start from index 0 until i, and the second time you start from index n to 0. So, to simplify this, we can say that it took you 2n. When dealing with Big O notation, you should keep in mind that we care about the bounds:
As a result, O(2n)=O(n)
and O(an+b)=O(n)
Input: int n // operation 1
for(int i = 0; i < n; i++){ // operation 2
print("Hello World, "); // Operation 3
}
for(int j = n; j > 0; j--) // Operation 4
{
print("Hello World"); //Operation 5
}
As you can see, we have a total of 5 operations outside the loops.
Inside the first loop, we do three internal operations: checking if i is less than n, printing "Hello World", and incrementing i .
Inside the second loop, we also have three internal operations.
So, the total number of of opetations that we need is: 3n ( for first loop) + 3n ( second loop) + 5 ( operations outside the loop). As a result, the total number of steps required is 6n+5 ( that is your tight bound).
As I mentioned before, O( an +b )= n because once an algorithm is linear, a and b do not have a great impact when n is very large.
So, your time complexity will become : O(6n+5) =O(n).
You can use the same logic for the second example keeping in mind that two nested loops take n² instead of n.
I will slightly modify Johns answer. Defining n is one constant operation, defining integer i and assigning it to 0 is 2 constant operations. defining integer j and assigning with n is another 2 constant operations. checking the conditions for i,j inside for loop,increment,print statement depends on n so the total will be 3n+3n+5 which is equal to 6n+5. Here we cannot skip any of the statements during execution so its average case running time will also be its worst case running time which is O(n)