题目 | RLE - 颢天笔记

Monoxer Programming Contest 2022（AtCoder Beginner Contest 249）

E - RLE

https://atcoder.jp/contests/abc249/tasks/abc249_e

Time Limit: 3 sec / Memory Limit: 1024 MB

Score : $500$ points

Problem Statement

Consider the following procedure of, given a string $X$ consisting of lowercase English alphabets, obtaining a new string:

Split the string $X$ off at the positions where two different characters are adjacent to each other.
For each string $Y$ that has been split off, replace $Y$ with a string consisting of the character which $Y$ consists of, followed by the length of $Y$.
Concatenate the replaced strings without changing the order.

For example, aaabbcccc is divided into aaa,bb,cccc, which are replaced by a3,b2,c4, respectively, which in turn are concatenated without changing the order, resulting in a3b2c4.If the given string is aaaaaaaaaa , the new string is a10 .

Find the number, modulo $P$, of strings $S$ of lengths $N$ consisting of lowercase English alphabets, such that the length of $T$ is smaller than that of $S$, where $T$ is the string obtained by the procedure above against the string $S$.

Constraints

$1 \le N \le 3000$
$10^8 \le P \le 10^9$
$N$ and $P$ are integers.
$P$ is a prime.

Input

Input is given from Standard Input in the following format:

$N$ $P$

Output

Print the answer.

Sample Input 1

3 998244353

Sample Output 1

26

Those strings of which the $1$-st, $2$-nd, and $3$-rd characters are all the same satisfy the condition.

For example, aaa becomes a3, which satisfies the condition, while abc becomes a1b1c1, which does not.

Sample Input 2

2 998244353

Sample Output 2

0

Note that if a string is transformed into another string of the same length, such as aa that becomes a2, it does not satisfy the condition.

Sample Input 3

5 998244353

Sample Output 3

2626

Strings like aaabb and aaaaa satisfy the condition.

Sample Input 4

3000 924844033

Sample Output 4

607425699

Find the number of strings satisfying the condition modulo $P$.

我的笔记

核心思想：动态规划

定义

$dp[i][j]:=$ 原长为 $i$，转化后长为 $j$ 的字符串种数。

初始化

若一个字符串只含一种字符，如长度为 $6$ 的字符串 aaaaaa，转化后 a6 长度为 $2$，一共有 $26$ 种字母，则可初始化 $dp[i][f(i)] = 26$，$1\leq i \leq N$.

$$ \begin{equation} f(x)=\left\{ \begin{aligned} &2 \quad 1 \leq x<10\\ &3 \quad 10 \leq x<100\\ &4 \quad 100 \leq x<1000\\ &5 \quad 1000 \leq x<10000 \end{aligned} \right . \end{equation} $$

转移

在字符串 aaaaaa 后加上 bb，转化后的字符串从 a6 变为 a6b2。

设加上的字符串长度为 $k$，则：$dp[i+k][j+f(k)]$ += $25\times dp[i][j]$（加的字符串字母不能和之前的一样，所以只有 $26-1$ 种字母）

for (int i = 1; i <= N; i++)
    for (int j = 1; j <= N; j++)
        for (int k = 1; k <= N - i; k++)
            if (j + length(k) <= N)
                {dp[i + k][j + length(k)] += 25 * dp[i][j] % P; dp[i + k][j + length(k)] %= P;}

答案

题目需求解原长为 $N$，转化后长度 $<N$ 的种数，即 $ans=\sum_{i=1}^{N-1} dp[N][i]$.

时间复杂度

$O(N^3)$，超时无疑

优化思想：前缀和

上面的方法第三层循环为：将 $dp[i][j]$ 加到很多个其他数里面去，很费时。

我们可以反向思维：对于一个 $dp[i][j]$，哪些数加到了这里面。

然后我们可以发现，对于一个 $dp[i][j]$，$dp[(i-9)\sim(i-1)][j-2]$、$dp[(i-99)\sim(i-10)][j-3]$、$dp[(i-999)\sim(i-100)][j-4]$、$dp[(i-9999)\sim(i-1000)][j-4]$，这四种数加到了里面。

并且这四种数字，每种数字都是同一列的，那么我们可以将每一列做一个前缀和，这样就可以在 $O(1)$ 时间内获得这四种数字的值，加到 $dp[i][j]$ 里就行了。

每次算完 $dp[i][j]$ 后，更新新一位的前缀和 $presum[i][j]=presum[i-1][j]+dp[i][j]$.

代码

#include <bits/stdc++.h>

using namespace std;

const int MAXN = 3010;
long long dp[MAXN][MAXN * 2], presum[MAXN][MAXN * 2];

inline int length(int x)
{
    if (x >= 1000)
        return 5;
    else if (x >= 100)
        return 4;
    else if (x >= 10)
        return 3;
    else
        return 2;
}

int main()
{
    int N, P;
    cin >> N >> P;
    for (int i = 1; i <= N; i++)
        dp[i][length(i)] = 26;
    for (int i = 1; i <= N; i++)
    {
        for (int j = 1; j <= N; j++)
        {
            for (int k = 1, len = 2; k < i && len < j; k *= 10, len++)
            {
                int l = max(0, i - k * 10), r = i - k;
                dp[i][j] += 25 * (presum[r][j - len] - presum[l][j - len] + P) % P;
            }
            presum[i][j] = (presum[i - 1][j] + dp[i][j]) % P;
        }
    }
    long long ans = 0;
    for (int i = 1; i < N; i++)
    {
        ans += dp[N][i];
        ans %= P;
    }
    cout << ans << endl;
    return 0;
}

题目 | RLE