Codeforces 718A A. Efim and Strange Grade

在有限的时间内,Efim需要通过巧妙地四舍五入他的小数成绩来获得尽可能高的分数。本篇介绍了如何利用特定的算法策略,通过考虑成绩的小数部分,最大化其最终得分。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

C. Efim and Strange Grade

time limit per test1 second
memory limit per test256 megabytes
input standard input
output standard output

Efim just received his grade for the last test. He studies in a special school and his grade can be equal to any positive decimal fraction. First he got disappointed, as he expected a way more pleasant result. Then, he developed a tricky plan. Each second, he can ask his teacher to round the grade at any place after the decimal point (also, he can ask to round to the nearest integer).

There are t seconds left till the end of the break, so Efim has to act fast. Help him find what is the maximum grade he can get in no more than t seconds. Note, that he can choose to not use all t seconds. Moreover, he can even choose to not round the grade at all.

In this problem, classic rounding rules are used: while rounding number to the n-th digit one has to take a look at the digit n + 1. If it is less than 5 than the n-th digit remain unchanged while all subsequent digits are replaced with 0. Otherwise, if the n + 1 digit is greater or equal to 5, the digit at the position n is increased by 1 (this might also change some other digits, if this one was equal to 9) and all subsequent digits are replaced with 0. At the end, all trailing zeroes are thrown away.

For example, if the number 1.14 is rounded to the first decimal place, the result is 1.1, while if we round 1.5 to the nearest integer, the result is 2. Rounding number 1.299996121 in the fifth decimal place will result in number 1.3.

Input
The first line of the input contains two integers n and t (1 ≤ n ≤ 200 000, 1 ≤ t ≤ 109) — the length of Efim’s grade and the number of seconds till the end of the break respectively.

The second line contains the grade itself. It’s guaranteed that the grade is a positive number, containing at least one digit after the decimal points, and it’s representation doesn’t finish with 0.

Output
Print the maximum grade that Efim can get in t seconds. Do not print trailing zeroes.

Examples
input

6 1
10.245

output

10.25

input

6 2
10.245

output

10.3

input

3 100
9.2

output

9.2

Note
In the first two samples Efim initially has grade 10.245.

During the first second Efim can obtain grade 10.25, and then 10.3 during the next second. Note, that the answer 10.30 will be considered incorrect.

In the third sample the optimal strategy is to not perform any rounding at all.

n, t = [int(x) for x in raw_input().split()]
a = list(raw_input())

found = False
dot = n
tos = 0
for i,c in enumerate(a):
    if c == '.':
        dot = i
        break
p = n
for i in xrange(dot+1, n):
    if int(a[i]) >= 5:
        p = i
        break

if p < n:
    for i in xrange(p, dot, -1):
        if int(a[i]) >= 5 and t > 0:
            a[i] = str(int(a[i]) % 10)
            if a[i-1] == '.':
                tos = 1
                p = dot-1
                break
            else:
                a[i-1] = str(int(a[i-1]) + 1)
                p -= 1
                t -= 1
        else:
            break
    if tos > 0:
        for i in xrange(dot-1, -1, -1):
            num = int(a[i]) + tos
            tos = 0
            a[i] = str(num%10)
            if num >= 10:
                tos = 1
            else:
                break

print ''.join(a[:p+1]) if tos == 0 else '{0}{1}'.format(tos, ''.join(a[:p+1])) 
### Codeforces 1732A Bestie 题目解析 对于给定的整数数组 \(a\) 和查询次数 \(q\),每次查询给出两个索引 \(l, r\),需要计算子数组 \([l,r]\) 的最大公约数(GCD)。如果 GCD 结果为 1,则返回 "YES";否则返回 "NO"[^4]。 #### 解决方案概述 为了高效解决这个问题,可以预先处理数据以便快速响应多个查询。具体方法如下: - **预处理阶段**:构建辅助结构来存储每一对可能区间的 GCD 值。 - **查询阶段**:利用已有的辅助结构,在常量时间内完成每个查询。 然而,考虑到内存限制以及效率问题,直接保存所有区间的结果并不现实。因此采用更优化的方法——稀疏表(Sparse Table),它允许 O(1) 时间内求任意连续子序列的最大值/最小值/GCD等问题,并且支持静态RMQ(Range Minimum Query)/RANGE_GCD等操作。 #### 实现细节 ##### 构建稀疏表 通过动态规划的方式填充二维表格 `st`,其中 `st[i][j]` 表示从位置 i 开始长度为 \(2^j\) 的子串的最大公约数值。初始化时只需考虑单元素情况即 j=0 的情形,之后逐步扩展至更大的范围直到覆盖整个输入序列。 ```cpp const int MAXN = 2e5 + 5; int st[MAXN][20]; // Sparse table for storing precomputed results. vector<int> nums; void build_sparse_table() { memset(st,-1,sizeof(st)); // Initialize the base case where interval length is one element only. for(int i = 0 ;i < nums.size(); ++i){ st[i][0]=nums[i]; } // Fill up sparse table using previously computed values. for (int j = 1;(1 << j)<=nums.size();++j){ for (int i = 0;i+(1<<j)-1<nums.size();++i){ if(i==0 || st[i][j-1]!=-1 && st[i+(1<<(j-1))][j-1]!=-1) st[i][j]=__gcd(st[i][j-1],st[i+(1<<(j-1))][j-1]); } } } ``` ##### 处理查询请求 当接收到具体的 l 和 r 参数后,可以通过查找对应的 log₂(r-l+1) 来定位合适的跳跃步长 k ,进而组合得到最终答案。 ```cpp string query(int L,int R){ int K=(int)(log2(R-L+1)); return __gcd(st[L][K],st[R-(1<<K)+1][K])==1?"YES":"NO"; } ``` 这种方法能在较短时间内完成大量查询任务的同时保持较低的空间开销,非常适合本题设定下的性能需求。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值