Strange Towers of Hanoi

题目：

Background 
Charlie Darkbrown sits in another one of those boring Computer Science lessons: At the moment the teacher just explains the standard Tower of Hanoi problem, which bores Charlie to death! 

The teacher points to the blackboard (Fig. 4) and says: "So here is the problem: 

• There are three towers: A, B and C. 
• There are n disks. The number n is constant while working the puzzle. 
• All disks are different in size. 
• The disks are initially stacked on tower A increasing in size from the top to the bottom. 
• The goal of the puzzle is to transfer all of the disks from tower A to tower C. 
• One disk at a time can be moved from the top of a tower either to an empty tower or to a tower with a larger disk on the top.

So your task is to write a program that calculates the smallest number of disk moves necessary to move all the disks from tower A to C." 
Charlie: "This is incredibly boring—everybody knows that this can be solved using a simple recursion.I deny to code something as simple as this!" 
The teacher sighs: "Well, Charlie, let's think about something for you to do: For you there is a fourth tower D. Calculate the smallest number of disk moves to move all the disks from tower A to tower D using all four towers." 
Charlie looks irritated: "Urgh. . . Well, I don't know an optimal algorithm for four towers. . . " 
Problem 
So the real problem is that problem solving does not belong to the things Charlie is good at. Actually, the only thing Charlie is really good at is "sitting next to someone who can do the job". And now guess what — exactly! It is you who is sitting next to Charlie, and he is already glaring at you. 
Luckily, you know that the following algorithm works for n <= 12: At first k >= 1 disks on tower A are fixed and the remaining n-k disks are moved from tower A to tower B using the algorithm for four towers.Then the remaining k disks from tower A are moved to tower D using the algorithm for three towers. At last the n - k disks from tower B are moved to tower D again using the algorithm for four towers (and thereby not moving any of the k disks already on tower D). Do this for all k 2 ∈{1, .... , n} and find the k with the minimal number of moves. 
So for n = 3 and k = 2 you would first move 1 (3-2) disk from tower A to tower B using the algorithm for four towers (one move). Then you would move the remaining two disks from tower A to tower D using the algorithm for three towers (three moves). And the last step would be to move the disk from tower B to tower D using again the algorithm for four towers (another move). Thus the solution for n = 3 and k = 2 is 5 moves. To be sure that this really is the best solution for n = 3 you need to check the other possible values 1 and 3 for k. (But, by the way, 5 is optimal. . . )

题意：四根柱子的汉诺塔问题，依次输出圆盘个数为1~12的最少移动次数。

汉诺塔问题比较经典，先介绍一下三根柱子的汉诺塔问题，假如有A,B,C三根柱子，初始条件下A柱子上有n个圆盘，我们需要求将A柱子上的圆盘全部移动到C柱子上的最少移动次数。不难想到的是我们要移动A柱子上的最后一个圆盘，必定要把前n-1个圆盘全部移动到B柱子上，才能把第n个圆盘移动到C柱子上，将n-1个圆盘全部移动到B柱子上所需的最少移动次数为f[n-1]，将第n个圆盘移动到C柱子上的次数为1，再将B柱子上的n-1个圆盘借助A柱子全部移动到C柱子上的最少移动次数依旧是f[n-1]，所以就有递推式f[n]=f[n-1]*2+1,利用这个递推式我们能比较容易地得出f[n]=2^n-1。（在代码中我就不再计算了，直接利用这个结论了）

三根柱子的汉诺塔问题解决了，下面我们来看下四根柱子的汉诺塔问题，假设四根柱子依次为A，B,C,D,在处理四根柱子的汉诺塔问题时，我们必经的一种状态就是我们需要先将A柱子上的i个圆盘借助C,D两根柱子移动到B柱子上，所需最少次数为dp[i]，然后我们就是将A柱子上剩余的n-i个圆盘借助柱子C移动到D柱子上，所需最少次数为f[n-i]，最后再将B柱子上的i个圆盘借助A，C两根柱子移动到D柱子上，所需最少移动次数为dp[i]，总的移动次数就是f[n-i]+2*d[i],我们需要遍历1~n找一个使得f[n-i]+2*d[i]值最小的i作为最优结果，这样就可以完成四根柱子的汉诺塔问题的求解了。同理n根柱子的汉诺塔问题也是先转化为n-1根柱子的汉诺塔问题进行求解的。

下面是代码：           f[n]=(1<<n)-1

#include<cstdio>
#include<iostream>
#include<cstring>
#include<algorithm>
#include<map>
#include<cmath>
using namespace std;
int f[15];
int main()
{
	memset(f,0x3f,sizeof f);//不要忘记初始化 
	f[0]=0;
	for(int i=1;i<=12;i++)
	{
		for(int j=0;j<i;j++)
			f[i]=min(f[i],2*f[j]+(1<<(i-j))-1);
		printf("%d\n",f[i]);
	}
	return 0;
}