[英]Time complexity of a function in Big-O
I'm trying to find the time complexity of this function: 我正在尝试查找此函数的时间复杂度:
int bin_search(int a[], int n, int x); // Binary search on an array with size n.
int f(int a[], int n) {
int i = 1, x = 1;
while (i < n) {
if (bin_search(a, i, x) >= 0) {
return x;
}
i *= 2;
x *= 2;
}
return 0;
}
The answer is (log n)^2. 答案是(log n)^ 2。 How come? 怎么会?
Best I could get is log n
. 我能得到的最好是log n
。 First the i
is 1
, so the while will be run log n
times. 首先, i
是1
,因此while将被运行log n
次。
At first interaction, when i=1
, the binary search will have only one interaction because the array's size is 1(i). 在第一次交互时,当i=1
,二进制搜索将只有一个交互,因为数组的大小为1(i)。 Then, when i=2
, two interactions, and so on until it's log n
interactions. 然后,当i=2
,将进行两次交互,依此类推,直到log n
交互。
So the formula I thought would fit is this . 所以我认为合适的公式就是这个 。
The summation is for the while and the inner equation is because for i=1
it's log(1)
, for i=2
it's log(2)
and so on until it's log(n)
at the last. 求和是针对一会儿的,内部方程式是因为对于i=1
它是log(1)
,对于i=2
它是log(2)
,依此类推,直到最后一个log(n)
。
Where am I wrong? 我哪里错了?
Each iteration performs a binary search on the first 2^i
elements of the array. 每次迭代都对数组的前2^i
元素执行二进制搜索。
You can compute the number of operations (comparisons): 您可以计算操作数(比较):
log2(1) + log2(2) + log2(4) + ... + log2(2^m)
log(2^n)
equals n
, so this series simplifies into: log(2^n)
等于n
,因此该系列简化为:
0 + 1 + 2 + ... + m
Where m
is floor(log2(n))
. 其中m
是floor(log2(n))
。
The series evaluates to m * (m + 1) / 2
, replacing m
we get 该系列的计算结果为m * (m + 1) / 2
,替换m
得到
floor(log2(n)) * (floor(log2(n)) + 1) / 2
-> 0.5 * floor(log2(n))^2 + 0.5 * floor(log2(n))
The first element dominates the second, hence the complexity is O(log(n)^2)
第一个元素主导第二个元素,因此复杂度为O(log(n)^2)
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.