有没有O(1/n)种算法?

或者其他小于O(1)的数?


当前回答

我猜小于O(1)是不可能的。算法所花费的任何时间都称为O(1)。但是对于O(1/n)下面的函数呢。(我知道这个解决方案中已经出现了许多变体,但我猜它们都有一些缺陷(不是主要的,它们很好地解释了这个概念)。这里有一个,只是为了方便讨论:

def 1_by_n(n, C = 10):   #n could be float. C could be any positive number
  if n <= 0.0:           #If input is actually 0, infinite loop.
    while True:
      sleep(1)           #or pass
    return               #This line is not needed and is unreachable
  delta = 0.0001
  itr = delta
  while delta < C/n:
    itr += delta

因此,随着n的增加,函数将花费越来越少的时间。此外,如果输入实际为0,则函数将永远返回。

有人可能会说,这将受到机器精度的限制。因此,由于c eit有一个上界,它是O(1)。但我们也可以绕过它,通过在字符串中输入n和C。加法和比较是对字符串进行的。用这个方法,我们可以把n减小到任意小。因此,即使忽略n = 0,函数的上限也是无界的。

我也相信我们不能说运行时间是O(1/n)。我们应该写成O(1 + 1/n)

其他回答

我相信量子算法可以通过叠加“一次”进行多次计算……

我怀疑这是一个有用的答案。

sharptooth是正确的,O(1)是可能的最佳性能。然而,这并不意味着一个快速的解决方案,只是一个固定时间的解决方案。

一个有趣的变种,也许是真正的建议,是随着人口的增长,哪些问题会变得更容易。我能想出一个虽然是做作的半开玩笑的答案:

一组中有两个人生日相同吗?当n超过365时,返回true。虽然小于365,这是O(nln n)。也许不是一个很好的答案,因为问题不会慢慢变得简单,而是变成O(1)对于n > 365。

inline void O0Algorithm() {}

正如已经指出的,除了null函数可能的例外,不可能有O(1/n)个函数,因为所花费的时间必须接近0。

当然,有一些算法,比如康拉德定义的算法,它们至少在某种意义上应该小于O(1)

def get_faster(list):
    how_long = 1/len(list)
    sleep(how_long)

If you want to investigate these algorithms, you should either define your own asymptotic measurement, or your own notion of time. For example, in the above algorithm, I could allow the use of a number of "free" operations a set amount of times. In the above algorithm, if I define t' by excluding the time for everything but the sleep, then t'=1/n, which is O(1/n). There are probably better examples, as the asymptotic behavior is trivial. In fact, I am sure that someone out there can come up with senses that give non-trivial results.

这是一个简单的O(1/n)算法。它甚至做了一些有趣的事情!

function foo(list input) {
  int m;
  double output;

  m = (1/ input.size) * max_value;  
  output = 0;
  for (int i = 0; i < m; i++)
    output+= random(0,1);

  return output;
}

O(1/n) is possible as it describes how the output of a function changes given increasing size of input. If we are using the function 1/n to describe the number of instructions a function executes then there is no requirement that the function take zero instructions for any input size. Rather, it is that for every input size, n above some threshold, the number of instructions required is bounded above by a positive constant multiplied by 1/n. As there is no actual number for which 1/n is 0, and the constant is positive, then there is no reason why the function would constrained to take 0 or fewer instructions.