PROBLEM LINK:

Practice
Contest

Author: admin
Tester: Kevin Atienza
Editorialist: Ajay K. Verma
Russian Translator: Sergey Kulik
Mandarian Translator: Gedi Zheng

DIFFICULTY:

Medium

PREREQUISITES:

Combinatorics, Modular Arithmetic

PROBLEM:

Given two integers $N$ and $K$, find how many triplets of strings over $K$ alphabets exist, such that length of all strings is bounded by $N$, and no string in a triplet is a prefix of another string of the triplet.

EXPLANATION:

We compute the number of triplets of strings of bounded length, and then subtract the ones in which one string is a prefix of another. For the sake of brevity, we define a new notation $\prec$ to represent the "prefix" relationship, i.e., "$X$ is a prefix of $Y$" can be represented by $X \prec Y$.

Total Number of Triplets:

There are exactly $K^i$ strings of length $i$. Hence, the number of strings with length not exceeding $N$ will be:
$A = K + K^2 + K^3 + \cdots + K^N$
$A = K (K^N - 1 / (K - 1))$

If the modular inverse of $(K - 1)$ with respect to the given prime $p = 10^9 + 7$ is $x$, then the value of $A \bmod p$ would be $(K (K^N - 1) x) \bmod p$.

Both the modular inverse, and power of a number can be compute in $O (\log N)$ time. Hence, we can compute the value of $A$ in $O (\log N)$ time.

There are $A$ strings of length not exceeding $N$. Therefore, the number of triplets would be $A^3$. Next, we count how many of these triplets are bad triplets, i.e., the ones in which one string is a prefix of another.

Triplets of the Form $X \prec Y \prec Z$:

In this section we count the number of triplets which have strings forming a chain under the relationship $\prec$, i.e., if the triplet is $(P, Q, R)$, then one of the following is true:
$P \prec Q \prec R$
$P \prec R \prec Q$ 
$Q \prec P \prec R$
$Q \prec R \prec P$ 
$R \prec P \prec Q$
$R \prec Q \prec P$

Note that for some triplets more than one of these conditions might be true, e.g., $(X, X, X)$ will satisfy all $6$ conditions. Hence, we consider four categories, so that each triplet is counted exactly once.

1) $X = Y = Z$:
There will be $A$ triplets of this form, as we can choose the string $X$ in $A$ ways, and the other two will be equal to it. These triplets will satisfy all $6$ conditions, i.e., each triplet will be repeated $6$ times.

2) $X \prec Y = Z$ and $X \neq Y$:
For a given string $Y$ of length $r$, there are exactly $(r - 1)$ possible values of string $X$, the ones which are proper substring of $Y$. Hence, the number of such triplets would be:
$K^2 + 2 K^3 + 3 K^4 + \cdots + (N - 1) K^N$.

This is the sum of an arithmetic-geometric series, and can be represented explicitly as shown here. The computation of this sum also requires the computation of power and modular inverse, and hence can be done in $O (\log N)$ time.

These triplets will satisfy two of the above conditions ($X \prec Y \prec Z$, and $X \prec Z \prec Y$).

3) $X = Y \prec Z$, and $Y \neq Z$:
Number of these triplets will be the same as the one computed in (2). This is because once we pick a string $Z$ of length $r$, then $Y$ has exactly $(r - 1)$ choices.

These triplets also satisfy two of the above conditions ($X \prec Y \prec Z$, and $Y \prec X \prec Z$).

4) $X \prec Y \prec Z$, $X \neq Y$, and $Y \neq Z$:
If we pick a string $Z$ of length $r$, then we only need to choose the lengths of $X$ and $Y$. The lengths of $X$ and $Y$ must be distinct and lie between $1$ and $(r - 1)$. Hence, the number of $(X, Y)$ pairs would be the number of ways of choosing two distinct numbers between $1$ and $(r - 1)$, which is given by $(r - 1) \choose 2$. Therefore the number of such triplets would be:
$K^3 + 3 K^4 + 6 K^5 + \cdots + {{N - 1} \choose 2} K^N$

Again this is the sum of arithmetic-geometric sequence of second order, and can be computed using the similar methods. The computation will take $O (\log N)$ time.

These triplets will satisfy exactly one of the above conditions.

Hence, the number of chain triplets would be $6$ ((number of category 1 triplets)/$6$ + (number of category 2 triplets)/$2$ + (number of category 3 triplets)/$2$ + (number of category 4 triplets)/$1$).

Triplets of the Form "$X \prec Y$, $Z$ does not share prefix relationship with $Y$":

In this section we count the number of triplets in which exactly two strings satisfy the "prefix" relationship, i.e., if the triplet is $(P, Q, R)$, then one of the following is true:
$P \prec Q$, $R$ has no prefix relation with $Q$,
$Q \prec P$, $R$ has no prefix relation with $P$,
$P \prec R$, $Q$ has no prefix relation with $R$,
$R \prec P$, $Q$ has no prefix relation with $P$,
$Q \prec R$, $P$ has no prefix relation with $R$,
$R \prec Q$, $P$ has no prefix relation with $Q$,

Once again it is possible that one triplet satisfy more than one condition, however, we should count it only once.

1) $X = Y$, $Z$ has no prefix relation with $Y$: 
If we have a string $Y$ of length $r$, then there are exactly $r$ prefix strings of $Y$. If a string $W$ has $Y$ as a proper prefix, then $W$ can be written as $W = Y + U$, where $U$ is a non-empty string of length at most $(N - r)$. Hence, the number of such string $W$ will be $K + K^2 + \cdots + K^{N - r}$. Since $Z$ is neither a prefix of $Y$, nor has $Y$ as its prefix, the number of possible values of $Z$ would be:
$(K + K^2 + \cdots + K^N) - (K + K^2 + \cdots + K^{N - r}) - r$
$= (K^{N - r + 1} + K^{N - r + 2} + \cdots + K^N - r)$.

Also number of ways of choosing a $r$ length string $Y$ is $K^r$. Hence, the number of such triplets $(X, Y, Z)$, with $Y$ being a $r$-length string will be:
$K^r (K^{N - r + 1} + K^{N - r + 2} + \cdots + K^N - r)$
$= K^{N + 1} * (1 + K + K^2 + \cdots + K^{r - 1}) - r * K^r$

If we take the sum of this expression over all possible values of $r$, we get the number of triplets $(X, Y, Z)$ satisfying the above criteria, and it will be:
$K^{N + 1} (N + (N - 1) K + (N - 2) K^2 + \cdots + K^{N - 1}) - (K + 2 K^2 + 3 K^3 + \cdots + N K^N)$

This is also a sum of arithmetic-geometric sequence, and can be computed in logarithmic time. These triplets satisfy two of the above conditions ($X \prec Y$, with $Z$ having no relationship, and $Y \prec X$, with $Z$ having no relationship).

2) $X \prec Y$, $X \neq Y$, and $Z$ has no prefix relation with $Y$:
Once again if we pick a string $Y$ of length $r$, then $X$ will have $(r - 1)$ choices, and Z will have $(K^{N - r + 1} + K^{N - r + 2} + \cdots + K^N - r)$ choices. Hence, the number of such triplets with $Y$ being of length $r$ would be:

$r K^r (K^{N - r + 1} + K^{N - r + 2} + \cdots + K^N - r)$
$= r K^{N + 1} (1 + K + K^2 + \cdots + K^{r - 1}) - r^2 K^r $

If we sum it over all possible values of $r$, we would get the number of triplets $(X, Y, Z)$ satisfying the above condition, which would be (after some simplifications):

${(N + 1) \choose 2} K^{N + 1} (1 + K + K^2 + \cdots + K^N) - K^{N + 1} ({2 \choose 2} K + {3 \choose 2} K^2 + \cdots + {N \choose 2} K^{N - 1})$
$ - (1^2 K + 2^2 K^2 + 3^2 K^3 + \cdots + N^2 K^N)$.

The number of triplets where exactly two strings satisfy the "prefix" relationship will be $6$ ( (number of triplets of first category)/$2$ + (number of triplets of second category)/$1$).

Now that we have computed the number of bad triplets, we can subtract them from the total number of triplets to get the number of good ones.

Time Complexity:

$O (\log N)$

AUTHOR'S AND TESTER'S SOLUTIONS:

Author's solution will be put up soon.
Tester's solution will be put up soon.

editorial-render A的更多相关文章

  1. React.render和reactDom.render的区别

    刚开始学习react.js.发现网上的资料,有些是写着react.render,有些写着reactDom.render.觉得很奇怪就查阅了一下资料.解释如下: 这个是react最新版api,也就是0. ...

  2. XF custom render 各平台实现类

    目前的XF还是非常简陋的,所以存在大量的自定义工作.一般情况下我们只是要需要派生原生的XF控件,然后在各平台下修改其呈现方法. 所以了解每个XF控件在不同平台上呈现使用的控件类是有所必须要的.以下别人 ...

  3. 塞翁失马,焉知非福:由 Styles.Render 所引发 runAllManagedModulesForAllRequests="true" 的思考

    最近在使用 MVC 开发的时候,遇到一个对我来说"奇怪的问题",就是使用 BundleTable 进行 CSS.JS 文件绑定,然后使用 Styles.Render.Scripts ...

  4. ReactJS分析之入口函数render

    前言 在使用React进行构建应用时,我们总会有一个步骤将组建或者虚拟DOM元素渲染到真实的DOM上,将任务交给浏览器,进而进行layout和paint等步骤,这个函数就是React.render() ...

  5. Cesium原理篇:6 Render模块(3: Shader)

    在介绍Renderer的第一篇,我就提到WebGL1.0对应的是OpenGL ES2.0,也就是可编程渲染管线.之所以单独强调这一点,算是为本篇埋下一个伏笔.通过前两篇,我们介绍了VBO和Textur ...

  6. Cesium原理篇:6 Render模块(4: FBO)

    Cesium不仅仅提供了FBO,也就是Framebuffer类,而且整个渲染过程都是在FBO中进行的.FBO,中文就是帧缓冲区,通常都属于高级用法,但其实,如果你了解了它的基本原理后,用起来还是很简单 ...

  7. Cesium原理篇:6 Render模块(5: VAO&RenderState&Command)

    VAO VAO(Vertext Array Object),中文是顶点数组对象.之前在<Buffer>一文中,我们介绍了Cesium如何创建VBO的过程,而VAO可以简单的认为是基于VBO ...

  8. render :template 和 render :parital

    1 .这两个都可以在controller和view中使用,而且好像可以替换,只是用:template,rails不会自动加下划线,用:partial,rails会自动添加下划线.而且规范的做法,:te ...

  9. AngularJs中,如何在render完成之后,执行Js脚本

    AngularJs是Google开源的前端JS框架.使用AngularJs, 我们能够容易地.健壮的开发出类似于Gmail一样的单页Web应用.AngularJs这个新兴的MVC前端框架,具有以下特点 ...

  10. 利用Render Texture实现游戏的小雷达效果(摄影机分屏)

    最近游戏蛮牛在举办一个活动,就是要做出这样的效果: 题目:实现游戏分屏效果 要求:1.        分屏,且分割线不规则(即非水平或垂直):2.        各屏可单独操作(移动.缩放),操作指该 ...

随机推荐

  1. “耐撕”团队 2016.04.07 站立会议

    1. 时间: 20:00--20:15  共计20分钟. 2. 成员: Z 郑蕊 * 组长 (博客:http://www.cnblogs.com/zhengrui0452/), P 濮成林(博客:ht ...

  2. Servlet Study 1

    this content below are come from the  JSR154 by sun Just for record purpose. if this relate to some ...

  3. 【BZOJ 3732】 Network Kruskal重构树+倍增LCA

    Kruskal重构树裸题, Sunshine互测的A题就是Kruskal重构树,我通过互测了解到了这个神奇的东西... 理解起来应该没什么难度吧,但是我的Peaks连WA,,, 省选估计要滚粗了TwT ...

  4. JS高级设计第七章——复习知识点

    <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/ ...

  5. 点击Cell中的按钮时,如何取所在的Cell

    4.点击Cell中的按钮时,如何取所在的Cell:-(void)OnTouchBtnInCell:(UIButton *)Btu{CGPoint point = btn.center;point = ...

  6. visual studio各个版本的差异

  7. SpringMVC数据库链接池,以及其他相关配置

    1.applicationContext.xml <?xml version="1.0" encoding="UTF-8"?> <beans ...

  8. [NOIP2011] 提高组 洛谷P1003 铺地毯

    题目描述 为了准备一个独特的颁奖典礼,组织者在会场的一片矩形区域(可看做是平面直角坐标系的第一象限)铺上一些矩形地毯.一共有 n 张地毯,编号从 1 到n .现在将这些地毯按照编号从小到大的顺序平行于 ...

  9. UVa 437 The Tower of Babylon

    Description   Perhaps you have heard of the legend of the Tower of Babylon. Nowadays many details of ...

  10. linux java cpu 100%

    1.用top找到最耗资源的进程id [ bin]# toptop - 16:56:14 up 119 days, 6:17, 7 users, load average: 2.04, 2.07, 2. ...