前言

在英語面試中可能需要 用英語表達算法的時間復雜度，這篇博文是學習后自我訓練的筆記，用學過的英語來描述算法的時間復雜度。

首先介紹如何用英語表達指數冪，然后是各種循環的復雜度分析。

How to express the exponent?

M to the power of N = M^N

x to the power of two thirds = (x)^(2/3)

x squared = x^2

x cubed = x^3

x plus y quantity squared = (x+y)^2

the sum of X one, X two and so on out to Xn, the whole string in parenthesis,raised to the second power = (x1 + x2 + x3 + ... + xn)^(2)

simple for-loop with an increment of size 1

for (int x = 0; x < n; x++) {
    //statement(s) that take constant time
}

The initialization statement assign 0 to x runs once.

Java for loop increments the value x by 1 in every iteration.So the x is first 0,then 1,then 2,...,then n-1.

The increment statement xplusplus runs n times.

The comparison statement x is less than n runs n plus 1 time.

Summing them up, we get a running time complexity of for loop of n + n + 1 + 1 = 2*n + 2

The statements inside the loop runs n times, supposing these statements account for a constant running time of c in each iteration, they account for a running time of c*n throughout the loop's lifetime.Hence the running time complexity is XXX

for-loop with increments of size k

for (int x = 0; x < n; x+=k) {
    //statement(s) that take constant time
}

The initialization statement runs once.

The incrementation part x+k runs floor(n/k) times

The comparision statement x<n runs the same time as the incrementation part and one more time in the last iteration.

The inside the loop runs c * floor(n/k) times.

Summing them up, the running time complexity is 1 + n/k + n/k + 1 + c * n/k = (c+2)/k* n + 2

Dropping the leading constants (c+2)/k ==> n + 2

Dropping the lower order items ==> O(n)

simple nested for-loop

for (int i=0; i<n; i++){
    for (int j=0; j<m; j++){
        //Statement(s) that take(s) constant time
    }
}

The inner loop runs (1+1+m+m)+c*m = (2m+2)+cm

The outer loop runs (1+1+n+n) + n(2m+2+cm) = (2+c)nm+4n+2 ,which is O(nm)

Nested for-loop with dependent variables

for (int i=0; i<n; i++){
    for (int j=0; j<i; j++){
        //Statement(s) that take(s) constant time
    }
}

Let's start with the inner loop.

The initialization statement j = 0 runs n times throughout the whole lifetime.

The comparison statement j < i runs different times when i is set to different value. For example, the first time when i is 0,j<i runs once, the second time when i is 1, the statement runs 2, so when i is n-1, the statement runs n times. So, this statement runs 1+2+...+n times，which is（1+n）*n/2 times.

The increment statement j++ runs 0 + 1+2+...+n-1 times,which is (n-1)*n/2

The statement inside the inner loop runs also c(n-1)n/2 times

As for the outer loop, the statements runs a total of 2n+2 times

Hence, the running time complexity is n+(1+n）n/2 + (n-1)n/2 + c(n-1)n/2 + 2n+2 which is O(n^2)

Nested for-loop with index modification(a little bit hard)

for (int i=0; i<n; i++){
    i*=2;
    for (int j=0; j<i; j++){
        // Statement(s) that take(s) constant time
    }
}

The initialization statement i=0 runs once;

How much times dose the comparison statement i <n and i++ and the inside of the outer loop run ?

To calculate these, we should ignore the i++ after the first iteration. For example i is first 0 , i is then 1, then i is 2, then i is 4... until i is great than n.

we assume n is greater than 2^k and less than 2^(k+1)

so, the i *=2 statement runs k times, and the i < n statement runs k+1 times, which k is estimately log2(n).

The statement j++ of the inner loop inside the outer loop runs

1+2+4+...+2^log2(n) = 2^(log(n)+1) - 1 = 2*n -1 times

So the total times is (2n-1)2 + 2 + (2n-1)c

Hence we can get the sum time of the code is 1 + 1 + 2(2n-1) + 2(2n-1) + 2+ (2n-1)*c

The term linear in n dominates（支配了） the others and the time complexity is O(n).

Note that the quiz can have a rough but not a tight bound approximation, that is the outer loop runs at most n times, the inner for-loop iterates at most n times each iteration of the outer loop. So we can conclude that the running time complexity is O(n^2).