Educational Codeforces Round 85 (Rated for Div. 2)
\(Educational\ Codeforces\ Round\ 85\ (Rated\ for\ Div.2)\)
\(A. Level Statistics\)
每天都可能会有人玩游戏,同时一部分人会过关,玩游戏的人数和过关的人数会每天更新,问记录的数据是否没有矛盾
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
void solve(){
int n, x = 0, y = 0;
cin >> n;
bool ok = true;
for(int i = 1; i <= n; i++){
int a, b;
cin >> a >> b;
if(a<x or b<y or b>a) ok = false;
if(b-y>a-x) ok = false;
x = a, y = b;
}
if(ok) cout << "YES" << endl;
else cout << "NO" << endl;
}
int main(){
____();
int T;
for(cin >> T; T; T--) solve();
return 0;
}
\(B - Middle Class\)
每次可以选几个人把钱均摊,问经过若干次操作之后最多有多少人的钱可以大于\(x\)
从大到小排序,然后判断前\(i\)个人均摊之后能否都大于\(x\),找最大的\(i\)即可
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
const int MAXN = 1e5+7;
using LL = int_fast64_t;
int n,x,A[MAXN];
void solve(){
cin >> n >> x;
for(int i = 1; i <= n; i++) cin >> A[i];
sort(A+1,A+1+n,greater<int>());
LL tot = 0, pr = 0;
for(int i = 1; i <= n; i++){
pr += x;
tot += A[i];
if(tot<pr){
cout << i-1 << endl;
return;
}
}
cout << n << endl;
}
int main(){
____();
int T;
for(cin >> T; T; T--) solve();
return 0;
}
\(C. Circle of Monsters\)
怪物站成一圈,每个怪物有血量\(A_i\),和爆炸伤害\(B_i\),每次打一颗子弹,可以扣除怪物\(1\)滴血,怪物血量\(\le0\)时死亡,死亡之后会对\(i+1\)个怪物造成相应的爆炸伤害,问最少多少子弹可以消灭所有怪物。
\(1~n\)复制一遍到\(n+1~2n\),然后算上一个怪物爆炸的情况下这个怪物还需要多少子弹,求前缀和,然后枚举第一个打的怪物即可
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
const int MAXN = 6e5+7;
using LL = int_fast64_t;
int n;
LL A[MAXN],B[MAXN],cost[MAXN],sum[MAXN];
void solve(){
cin >> n;
for(int i = 1; i <= n; i++){
cin >> A[i] >> B[i];
A[i+n] = A[i]; B[i+n] = B[i];
}
for(int i = 2; i <= n * 2; i++){
cost[i] = max(0ll,A[i]-B[i-1]);
sum[i] = sum[i-1] + cost[i];
}
LL ret = 0x3f3f3f3f3f3f3f3f;
for(int i = 1; i <= n; i++) ret = min(ret,A[i]+sum[i+n-1]-sum[i]);
cout << ret << endl;
}
int main(){
____();
int T;
for(cin >> T; T; T--) solve();
return 0;
}
\(D. Minimum Euler Cycle\)
给一张完全图,任意两个点\(u,v\)之间有\(u \rightarrow v, v \rightarrow u\)两条边,要求走完所有边并且走的点的标号的字典序最小,要求输出一段区间内的顶点标号。
假设为顶点数为\(n\)的完全图,则方案为:
\((1, 2, 1, 3, 1, 4 \cdots 1, n)(\ 2, 3, 2, 4 \cdots 2, n)(\ 3, 4, 3, 5 \cdots 3, n) \cdots (n-3, n-2, n-3, n-1, n-3, n)(\ n-2, n-1, n-2, n)(\ n-1, n)\ 1\)
用括号括出来的可以找出来规律
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
using LL = int_fast64_t;
LL n, l, r;
void solve(){
scanf("%I64d %I64d %I64d",&n,&l,&r);
if(l==n*(n-1)+1){
puts("1");
return;
}
int stpos = 1;
LL tot = 0;
LL L = 0, R = n - 1;
while(L<=R){
LL mid = (L+R)>>1;
if((n-1+n-mid)*mid<l) L = mid + 1;
else R = mid - 1;
}
stpos = L;
tot = (n-1+n-R)*R;
int cur = stpos + (l-tot) / 2;
bool tg = (((l-tot)&1)?true:false);
for(LL i = l; i <= r; i++){
if(stpos==n){
printf("1 ");
break;
}
if(tg){
printf("%d ",stpos);
cur++;
}
else{
printf("%d ",cur);
if(cur==n){
stpos++;
cur = stpos;
}
}
tg = !tg;
}
puts("");
}
int main(){
int T;
for(scanf("%d",&T); T; T--) solve();
return 0;
}
\(E. Divisor Paths\)
给出数\(D\),以这个数建图,建图规则为:所有\(D\)的因子为顶点,\(u,v(u>v)\)两点之间连边的条件为:\(u%v==0 and u/v is prime\),且边权为\(f(u)-f(v)\),其中\(f(x)\)为\(x\)的因子数量,现在有\(Q\)次询问,每次询问\(u,v\)间的最短路条数。
首先我们假设要从\(u\)到\(v\)且\(u\%v==0\),那么走的边必然是\(u\)比\(v\)分解之后多的质因子,略证:
首先假设命题成立
现在从\(a \rightarrow b\),我们走多余质因子边:\(path(a,b)=a \rightarrow c \rightarrow b\),则路径长度为:\(f(a)-f(c)+f(c)-f(b)=f(a)-f(b)\),现在假设不走多余质因子边则:\(path(a,b)=a\rightarrow c \rightarrow d \rightarrow \cdots \rightarrow b\),路径长度为\(f(a)-f(c)+f(d)-f(c)+dist(d,b)=f(a)-f(c)+f(d)-f(c)+f(d)-f(x)+dist(x,b)=dist(a,b)+2f(d)-2f(c)>dist(a,b)\)
如果\(u\%v\ne 0\),则必然是先从\(u\)到\(gcd(u,v)\),再从\(gcd(u,v)到v\)是最优的
如果两个数不互质的话可以先取\(gcd\),然后必然能经过点\(1\),然后只要找出\(1\)到各个点最短路的方案数即可。
假设\(x\)各个素因子的数量分别为\(P_1,P_2,\cdots,P_m\),则最短路数量就是\(\frac{(\sum_{i=1}^{m}P_i)!}{\prod_{i=1}^{m}(P_i!)}\)
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
using LL = int_fast64_t;
const LL MOD = 998244353;
const int MAXN = 1e5+7;
LL D, q, fact[MAXN], invf[MAXN];
map<LL,LL> ret;
LL qpow(LL a, LL b){
LL res = 1;
while(b){
if(b&1) res = res * a % MOD;
b >>= 1;
a = a * a % MOD;
}
return res;
}
void ikun(LL x){
LL v = x;
vector<int> vec;
int tot = 0;
for(LL i = 2; i * i <= x; i++){
if(x%i!=0) continue;
int cnt = 0;
while(x%i==0) cnt++, x /= i;
tot += cnt;
vec.emplace_back(cnt);
if(x==1) break;
}
if(x!=1) tot++;
LL res = fact[tot];
for(int p : vec) res = res * invf[p] % MOD;
ret.insert(make_pair(v,res));
}
void solve(){
fact[0] = 1;
for(int i = 1; i < MAXN; i++) fact[i] = fact[i-1] * i % MOD;
invf[MAXN-1] = qpow(fact[MAXN-1],MOD-2);
for(int i = MAXN - 2; i >= 0; i--) invf[i] = invf[i+1] * (i+1) % MOD;
cin >> D >> q;
for(LL i = 1; i * i <= D; i++){
if(D%i!=0) continue;
ikun(i);
if(i*i!=D) ikun(D/i);
}
while(q--){
LL u, v;
cin >> u >> v;
LL g = __gcd(u,v);
u /= g; v /= g;
cout << ret.at(u) * ret.at(v) % MOD << endl;
}
}
int main(){
____();
solve();
return 0;
}
\(F. Strange Function\)
给出一个序列,\(f(A)\)操作就是取序列\(A\)的前缀递增序列,序列中每个数有一个价值,现在要求你删除序列\(a\)中的一些元素,使得\(f(A)=B\)
考虑朴素的\(DP\),\(dp[i][j]\)表示现在\(A\)串的\(i\)位置和\(B\)串的\(j\)位置匹配,最小的花费,转移如下:
①如果不删除
\(\begin{cases} dp[i+1][j] = dp[i][j] & if\ A[i+1]\le B[j] \\ dp[i+1][j+1] = dp[i][j] & if\ A[i+1]=B[j+1]
\\ 不合法 & if\ A[i+1]>B[j]\ and\ A[i+1]\ne B[j+1] \end{cases}\)
②如果删除
\(dp[i+1][j] = dp[i][j] + cost[i+1]\)
现在整理一下:
\(\begin{cases}dp[i+1][j] = dp[i][j]+min(cost[i+1],0) & if\ A[i+1]\le B[j] \\ dp[i+1][j+1] = min(dp[i][j],dp[i][j+1]) & if\ A[i+1]=B[j+1] \\ dp[i+1][j] = dp[i][j]+cost[i+1] & if\ A[i+1]>B[j] \end{cases}\)
可以发现\(i+1\)的状态都是从\(i\)的状态转移过来的,所以可以把第一维给去掉,然后用线段树或者树状数组维护一下
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
const int MAXN = 5e5+7;
using LL = int_fast64_t;
const LL INF = 0x3f3f3f3f3f3f3f3f;
int n,m,A[MAXN],B[MAXN],p[MAXN];
template<typename T>
struct SegmentTree{
T lazy[MAXN<<2];
int l[MAXN<<2],r[MAXN<<2];
#define ls(rt) rt << 1
#define rs(rt) rt << 1 | 1
void build(int L, int R, int rt){
l[rt] = L; r[rt] = R;
lazy[rt] = 0;
if(L+1==R) return;
int mid = (L + R) >> 1;
build(L,mid,ls(rt)); build(mid,R,rs(rt));
}
void pushdown(int rt){
if(!lazy[rt]) return;
lazy[ls(rt)] += lazy[rt];
lazy[rs(rt)] += lazy[rt];
lazy[rt] = 0;
}
void update(int L, int R, int rt, T x){
if(L>=r[rt] || l[rt]>=R) return;
if(L<=l[rt] && r[rt]<=R){
lazy[rt] += x;
return;
}
pushdown(rt);
update(L,R,ls(rt),x); update(L,R,rs(rt),x);
}
T query(int L, int R, int rt){
if(L>=r[rt] || l[rt]>=R) return 0;
if(L<=l[rt] && r[rt]<=R) return lazy[rt];
pushdown(rt);
return query(L,R,ls(rt))+query(L,R,rs(rt));
}
};
SegmentTree<LL> ST;
int main(){
____();
cin >> n;
for(int i = 1; i <= n; i++) cin >> A[i];
for(int i = 1; i <= n; i++) cin >> p[i];
cin >> m;
for(int i = 1; i <= m; i++) cin >> B[i];
ST.build(0,m+1,1);
ST.update(1,m+1,1,INF);
for(int i = 1; i <= n; i++){
int pos = lower_bound(B+1,B+1+m,A[i]) - B;
if(A[i]==B[pos]){
ST.update(pos,m+1,1,min(0,p[i]));
ST.update(pos,pos+1,1,min(0ll,ST.query(pos-1,pos,1)-ST.query(pos,pos+1,1)));
ST.update(0,pos,1,p[i]);
}
else{
ST.update(pos,m+1,1,min(0,p[i]));
ST.update(0,pos,1,p[i]);
}
}
LL ret = ST.query(m,m+1,1);
if(ret>INF/2) cout << "NO" << endl;
else cout << "YES" << endl << ret << endl;
return 0;
}
\(G. Substring Search\)
给出两个串\(s,t\)要求找到在\(t\)串中所有能和\(s\)串匹配的位置,其中字符匹配的条件是\(s_j==t_{i+j} or P_{s_j}=t_{i+j}\),其中\(P\)为\(0-25\)的一个排列。
如果用朴素的想法,\(26\)个字符每个都匹配一遍,最后找每个位置的匹配数量的话,一共要进行\(78\)遍\(FFT\),但是题目只给了\(1.25s\),显然是行不通的
如果转化成方程的形式,那么如果两个串匹配,则每个相应字符都能匹配上,可以写成:
\]
把\(s\)串翻转后得到:
\]
转换成卷积形式:
\]
每个位置的值即为:
\]
展开:
\]
对于第一项和最后一项可以用前缀和来求,这样可以只做\(7\)次\(FFT\)
hack数据有点恶心,把NTT的998244353模数卡了,用1004535809能过
好像bitset可以水过,而且耗时还很优秀
//#pragma GCC optimize("O3")
//#pragma comment(linker, "/STACK:1024000000,1024000000")
#include<bits/stdc++.h>
using namespace std;
function<void(void)> ____ = [](){ios_base::sync_with_stdio(false); cin.tie(0); cout.tie(0);};
const int MAXN = 1e6+7;
using LL = int_fast64_t;
//const LL MOD = 998244353;
const LL MOD = 1004535809;
const LL g = 3;
int n,P[30],l,limit,r[MAXN],lens,lent;
LL A[MAXN],B[MAXN],C[MAXN];
char s[MAXN],t[MAXN];
LL qpow(LL a, LL b){
if(!a) return 0;
LL ret = 1;
while(b){
if(b&1) ret = ret * a % MOD;
b >>= 1;
a = a * a % MOD;
}
return ret;
}
void NTT(LL arr[], int rev){
for(int i = 0; i < limit; i++) if(i<r[i]) swap(arr[i],arr[r[i]]);
for(int len = 1; len < limit; len <<= 1){
LL wn = qpow(g,(MOD-1)/(len<<1));
if(rev==-1) wn = qpow(wn,MOD-2);
for(int R = 0; R < limit; R += (len<<1)){
LL w = 1;
for(int i = R; i < R + len; i++){
LL x = arr[i], y = w * arr[i+len] % MOD;
arr[i] = (x+y)%MOD;
arr[i+len] = (x-y+MOD)%MOD;
w = w * wn % MOD;
}
}
}
}
int main(){
for(int i = 0; i < 26; i++){
scanf("%d",&P[i]);
P[i]--;
}
scanf("%s %s",s,t);
lens = strlen(s); lent = strlen(t);
reverse(s,s+lens);
limit = 1; l = 0;
while(limit<=(lens+lent-2)) limit <<= 1, l++;
for(int i = 0; i < limit; i++) r[i] = (r[i>>1]>>1) | ((i&1)<<(l-1));
for(int i = lent; i < limit; i++) B[i] = 0;
for(int i = lens; i < limit; i++) C[i] = 0;
for(int i = 0; i < lent; i++){
LL x = t[i] - 'a';
B[i] = x * x * x;
}
for(int i = 0; i < lens; i++) C[i] = (-2*(s[i]-'a'+P[s[i]-'a'])+MOD)%MOD;
NTT(B,1); NTT(C,1);
for(int i = 0; i < limit; i++) A[i] = (A[i] + B[i] * C[i]) % MOD;
for(int i = lent; i < limit; i++) B[i] = 0;
for(int i = lens; i < limit; i++) C[i] = 0;
for(int i = 0; i < lent; i++){
LL x = t[i] - 'a';
B[i] = x * x;
}
for(int i = 0; i < lens; i++) C[i] = qpow(s[i]-'a',2)+qpow(P[s[i]-'a'],2)+4*(s[i]-'a')*P[s[i]-'a'];
NTT(B,1); NTT(C,1);
for(int i = 0; i < limit; i++) A[i] = (A[i] + B[i] * C[i]) % MOD;
for(int i = lent; i < limit; i++) B[i] = 0;
for(int i = lens; i < limit; i++) C[i] = 0;
for(int i = 0; i < lent; i++) B[i] = t[i] - 'a';
for(int i = 0; i < lens; i++) C[i] = (-2*(s[i]-'a')*P[s[i]-'a']*(s[i]-'a'+P[s[i]-'a'])+MOD)%MOD;
NTT(B,1); NTT(C,1);
for(int i = 0; i < limit; i++) A[i] = (A[i] + B[i] * C[i]) % MOD;
NTT(A,-1);
LL inv = qpow(limit,MOD-2);
LL sum = 0;
vector<LL> pre(lent);
for(int i = 0; i < lens; i++) sum += (s[i]-'a')*(s[i]-'a')*P[s[i]-'a']*P[s[i]-'a'];
pre[0] = qpow(t[0]-'a',4);
for(int i = 1; i < lent; i++) pre[i] = pre[i-1] + qpow(t[i]-'a',4);
for(int i = lens - 1; i < lent; i++) A[i] = (A[i] * inv + sum + pre[i] - ((i-lens>=0)?pre[i-lens]:0)) % MOD;
for(int i = lens - 1; i < lent; i++) putchar(A[i]?'0':'1'); puts("");
return 0;
}
Educational Codeforces Round 85 (Rated for Div. 2)的更多相关文章
- Educational Codeforces Round 60 (Rated for Div. 2) - C. Magic Ship
Problem Educational Codeforces Round 60 (Rated for Div. 2) - C. Magic Ship Time Limit: 2000 mSec P ...
- Educational Codeforces Round 60 (Rated for Div. 2) - D. Magic Gems(动态规划+矩阵快速幂)
Problem Educational Codeforces Round 60 (Rated for Div. 2) - D. Magic Gems Time Limit: 3000 mSec P ...
- Educational Codeforces Round 43 (Rated for Div. 2)
Educational Codeforces Round 43 (Rated for Div. 2) https://codeforces.com/contest/976 A #include< ...
- Educational Codeforces Round 35 (Rated for Div. 2)
Educational Codeforces Round 35 (Rated for Div. 2) https://codeforces.com/contest/911 A 模拟 #include& ...
- Codeforces Educational Codeforces Round 44 (Rated for Div. 2) F. Isomorphic Strings
Codeforces Educational Codeforces Round 44 (Rated for Div. 2) F. Isomorphic Strings 题目连接: http://cod ...
- Codeforces Educational Codeforces Round 44 (Rated for Div. 2) E. Pencils and Boxes
Codeforces Educational Codeforces Round 44 (Rated for Div. 2) E. Pencils and Boxes 题目连接: http://code ...
- Educational Codeforces Round 63 (Rated for Div. 2) 题解
Educational Codeforces Round 63 (Rated for Div. 2)题解 题目链接 A. Reverse a Substring 给出一个字符串,现在可以对这个字符串进 ...
- Educational Codeforces Round 39 (Rated for Div. 2) G
Educational Codeforces Round 39 (Rated for Div. 2) G 题意: 给一个序列\(a_i(1 <= a_i <= 10^{9}),2 < ...
- Educational Codeforces Round 48 (Rated for Div. 2) CD题解
Educational Codeforces Round 48 (Rated for Div. 2) C. Vasya And The Mushrooms 题目链接:https://codeforce ...
随机推荐
- Python_列表(list)
list()类中的提供的操作 1.索引取值 li = [11,22,33,44,55] v1 = li[3] print(li[2]) #索引取出33 print(v1) #索引取出44 print( ...
- mybatis入门教程之搭建一个简单的mybatis项目并启动它
一.准备条件: 1.依赖jar包:mybatis核心包(必须).lombok插件包(非必须)以及MySQL数据库连接驱动包(必须) <dependency> <groupId> ...
- Java 栈的使用
讲栈之前,要先讲一下Deque双端队列 既可以添加到队尾,也可以添加到队首 既可以从队首获取又可以从队尾获取 public interface Deque<E> extends Queue ...
- mongodb简单运用
mongodb NoSQL(Not Only SQL),意思是"不仅仅是 SQL",指的是非关系型数据库,是对不同于传统的关系型数据库的数据库管理系统的统称. NoSQL 用于超大 ...
- python-列表包字典-根据字典的某一个键的值来进行排序
python-列表包字典-根据字典的某一个键的值来进行排序 列表包字典的数据结构 要实现按照字典中的某一个键所对应的值进行排序 有两种办法 方法一,使用列表的sort方法 由小到大排 列表.sort( ...
- 处理K8S PVC删除后pod报错
报错如下 Jun 19 17:15:18 node1 kubelet[1722]: E0619 17:15:18.381558 1722 desired_state_of_world_populato ...
- 提示框,对话框,路由跳转页面,跑马灯,幻灯片及list组件的应用
目录: 主页面的js业务逻辑层 主页面视图层 主页面css属性设置 跳转页面一的js业务逻辑层 跳转页面一的视图层 跳转页面二的视图层 跳转页面三的js业务逻辑层 跳转页面三的视图层 跳转页面三的cs ...
- 使用Bat自动打包并通过FTP发送到备份服务器——实战测试
这个bat文件要求本地安装有winrar解压软件,位置是:C:\Program Files\WinRAR\WinRAR.exe 如果执行报错,请检查你复制我的代码是否有问题,有些复制粘贴进去后因为一些 ...
- Python+Selenium+Unittest实现PO模式web自动化框架(2)
1.Common目录下的具体模块讲解. 2.basepage.py basepage.py模块里面是封装的对元素的操作.例如:查找元素.点击元素.文本输入等等. # --^_^-- coding:ut ...
- 计算机网络第7版 PDF+ 计算机网络释疑与习题解答第7版 PDF 计算机网络 课后答案
网上全都是要钱的,要么就是第六版的,属实被恶心到了. 链接:https://pan.baidu.com/s/15jCOH6LXnQfB1RwGpNgBFg提取码:byMB