D. Persistent Bookcase

Recently in school Alina has learned what are the persistent data structures: they are data structures that always preserves the previous version of itself and access to it when it is modified.

After reaching home Alina decided to invent her own persistent data structure. Inventing didn't take long: there is a bookcase right behind her bed. Alina thinks that the bookcase is a good choice for a persistent data structure. Initially the bookcase is empty, thus there is no book at any position at any shelf.

The bookcase consists of n shelves, and each shelf has exactly m positions for books at it. Alina enumerates shelves by integers from 1to n and positions at shelves — from 1 to m. Initially the bookcase is empty, thus there is no book at any position at any shelf in it.

Alina wrote down q operations, which will be consecutively applied to the bookcase. Each of the operations has one of four types:

  • i j — Place a book at position j at shelf i if there is no book at it.
  • i j — Remove the book from position j at shelf i if there is a book at it.
  • i — Invert book placing at shelf i. This means that from every position at shelf i which has a book at it, the book should be removed, and at every position at shelf i which has not book at it, a book should be placed.
  • k — Return the books in the bookcase in a state they were after applying k-th operation. In particular, k = 0 means that the bookcase should be in initial state, thus every book in the bookcase should be removed from its position.

After applying each of operation Alina is interested in the number of books in the bookcase. Alina got 'A' in the school and had no problem finding this values. Will you do so?

Input

The first line of the input contains three integers nm and q (1 ≤ n, m ≤ 103, 1 ≤ q ≤ 105) — the bookcase dimensions and the number of operations respectively.

The next q lines describes operations in chronological order — i-th of them describes i-th operation in one of the four formats described in the statement.

It is guaranteed that shelf indices and position indices are correct, and in each of fourth-type operation the number k corresponds to some operation before it or equals to 0.

Output

For each operation, print the number of books in the bookcase after applying it in a separate line. The answers should be printed in chronological order.

Examples

input

2 3 3
1 1 1
3 2
4 0

output

1
4
0

input

4 2 6
3 2
2 2 2
3 3
3 2
2 2 2
3 2

output

2
1
3
3
2
4

input

2 2 2
3 2
2 2 1

output

2
1

Note

This image illustrates the second sample case.

Solution

 题目大意:

给出一个矩阵,要支持如下操作

1.(x,y)位置变成1

2.(x,y)位置变成0

3.整行取反,0变成1,1变成0

4.退回到第k次操作后的状态

一共Q次询问,每次询问后输出矩阵中1的个数

首先,把矩阵展成序列,对其建线段树,这样,1,2,3操作就是简单的单点修改,区间修改

操作4的难处在于空间不允许保存历史状态,

考虑离线。

首先假设我们得到$i$之前的所有操作的答案,$i+1$次操作是退回操作,显然$i+1$次操作的答案,可以通过以前的答案得到,但问题涉及状态的变化

很显然,一次退回操作就相当于将这个操作之后的,到下一次退回操作之前的所有操作,从其退回到的状态开始修改

这显然是个树形的结构,于是我们的方法就非常直观了

对于所有的操作,我们假定$i$操作是向$i+1$操作连一条单向边的,那么对于一个退回操作$k$,它所退回到的操作是$x$,就相当于从$x$也向$k+1$连一条单向边,然后我们用$x$把$k$的答案更新,去掉$k$既可

那么从一号操作为根的树上DFS,每次修改,记录答案,修改完后回溯,直到遍历整棵树

而这样,状态不能记录的问题就被解决了,只需要一棵线段树,不过是修改2Q次

一个操作,可能不合法,这时候需要记录一下,回溯的时候特判

Code

code from yveh

#include<cstdio>
#include<iostream>
#include<cstring>
using namespace std;
struct edgetype{
int s,t,next;
}e[];
int head[],cnt=;
void addedge(int s,int t)
{
e[cnt].s=s;e[cnt].t=t;e[cnt].next=head[s];head[s]=cnt++;
}
struct Node{
int data,size;
bool rev;
Node()
{
data=rev=;
}
};
bool flag;
namespace Segtree
{
Node tree[];
void pushup(int node)
{
tree[node].data=tree[node<<].data+tree[node<<|].data;
}
void build(int l,int r,int node)
{
tree[node].size=r-l+;
if (l==r)
return;
int mid=(l+r)>>;
build(l,mid,node<<);
build(mid+,r,node<<|);
}
void pushdown(int node)
{
if (tree[node].rev)
{
tree[node<<].data=tree[node<<].size-tree[node<<].data;
tree[node<<].rev^=;
tree[node<<|].data=tree[node<<|].size-tree[node<<|].data;
tree[node<<|].rev^=;
tree[node].rev=;
}
}
void modify_pos(int pos,int l,int r,int node,int val)
{
if (l==r)
{
flag=tree[node].data==val;
tree[node].data=val;
return;
}
pushdown(node);
int mid=(l+r)>>;
if (pos<=mid)
modify_pos(pos,l,mid,node<<,val);
else
modify_pos(pos,mid+,r,node<<|,val);
pushup(node);
}
void modify_rev(int L,int R,int l,int r,int node)
{
if (L<=l&&r<=R)
{
tree[node].data=tree[node].size-tree[node].data;
tree[node].rev^=;
return;
}
pushdown(node);
int mid=(l+r)>>;
if (L<=mid)
modify_rev(L,R,l,mid,node<<);
if (R>mid)
modify_rev(L,R,mid+,r,node<<|);
pushup(node);
}
int query()
{
return tree[].data;
}
}
int n,m,q,opt,u,v,k;
int a[][],ans[];
void init()
{
scanf("%d%d%d",&n,&m,&q);
Segtree::build(,n*m,);
}
void dfs(int node)
{
if (a[node][]==)
{
Segtree::modify_pos((a[node][]-)*m+a[node][],,n*m,,a[node][]);
if (!flag)
a[node][]=;
else
a[node][]=;
}
if (a[node][]==)
{
Segtree::modify_pos((a[node][]-)*m+a[node][],,n*m,,a[node][]);
if (!flag)
a[node][]=;
else
a[node][]=;
}
if (a[node][]==)
Segtree::modify_rev((a[node][]-)*m+,a[node][]*m,,n*m,);
ans[node]=Segtree::query();
for (int i=head[node];i!=-;i=e[i].next)
dfs(e[i].t);
if (a[node][]==)
Segtree::modify_pos((a[node][]-)*m+a[node][],,n*m,,a[node][]);
if (a[node][]==)
Segtree::modify_pos((a[node][]-)*m+a[node][],,n*m,,a[node][]);
if (a[node][]==)
Segtree::modify_rev((a[node][]-)*m+,a[node][]*m,,n*m,);
}
void work()
{
memset(head,0xff,sizeof(head));
cnt=;
for (int i=;i<=q;i++)
{
scanf("%d",&a[i][]);
if (a[i][]==)
{
scanf("%d%d",&a[i][],&a[i][]);
a[i][]=;
}
if (a[i][]==)
{
scanf("%d%d",&a[i][],&a[i][]);
a[i][]=;
} if (a[i][]==)
scanf("%d",&a[i][]);
if (a[i][]==)
{
scanf("%d",&k);
addedge(k,i);
}
else
addedge(i-,i);
}
dfs();
for (int i=;i<=q;i++)
printf("%d\n",ans[i]);
}
int main()
{
init();
work();
return ;
}

YveH打CF时问我的题...当时蹦出这个想法,但是他没能来得及当场A掉

觉得思路挺有意义的一道题,所以留下了想法...

实际上我还不知道题解是什么.....

【Codeforces-707D】Persistent Bookcase DFS + 线段树的更多相关文章

  1. Codeforces 707D Persistent Bookcase(时间树)

    [题目链接] http://codeforces.com/problemset/problem/707/D [题目大意] 给出一个矩阵,要求满足如下操作,单个位置x|=1或者x&=0,一行的数 ...

  2. 【离线】【深搜】【树】Codeforces 707D Persistent Bookcase

    题目链接: http://codeforces.com/problemset/problem/707/D 题目大意: 一个N*M的书架,支持4种操作 1.把(x,y)变为有书. 2.把(x,y)变为没 ...

  3. CodeForces 707D Persistent Bookcase ——(巧妙的dfs)

    一个n*m的矩阵,有四种操作: 1.(i,j)处变1: 2.(i,j)处变0: 3.第i行的所有位置1,0反转: 4.回到第k次操作以后的状态: 问每次操作以后整个矩阵里面有多少个1. 其实不好处理的 ...

  4. CodeForces 707D Persistent Bookcase

    $dfs$,优化. $return$操作说明该操作完成之后的状态和经过操作$k$之后的状态是一样的.因此我们可以建树,然后从根节点开始$dfs$一次(回溯的时候复原一下状态)就可以算出所有状态的答案. ...

  5. HDU 5877 dfs+ 线段树(或+树状树组)

    1.HDU 5877  Weak Pair 2.总结:有多种做法,这里写了dfs+线段树(或+树状树组),还可用主席树或平衡树,但还不会这两个 3.思路:利用dfs遍历子节点,同时对于每个子节点au, ...

  6. codeforces Good bye 2016 E 线段树维护dp区间合并

    codeforces Good bye 2016 E 线段树维护dp区间合并 题目大意:给你一个字符串,范围为‘0’~'9',定义一个ugly的串,即串中的子串不能有2016,但是一定要有2017,问 ...

  7. Codeforces1110F Nearest Leaf dfs + 线段树 + 询问离线

    Codeforces1110F dfs + 线段树 + 询问离线 F. Nearest Leaf Description: Let's define the Eulerian traversal of ...

  8. dfs+线段树 zhrt的数据结构课

    zhrt的数据结构课 这个题目我觉得是一个有一点点思维的dfs+线段树 虽然说看起来可以用树链剖分写,但是这个题目时间卡了树剖 因为之前用树剖一直在写这个,所以一直想的是区间更新,想dfs+线段树,有 ...

  9. codeforces 707D D. Persistent Bookcase(dfs)

    题目链接: D. Persistent Bookcase time limit per test 2 seconds memory limit per test 512 megabytes input ...

随机推荐

  1. 个人PHP开发环境的选择与搭建

    入职一个多月,重新调整了一下自己电脑的开发环境,现在写出来,算是作为自己的笔记. 如果你是该文章的读者,请忍受文章内的所有小章节都没有具体的步骤. 因为平时还要打游戏(划掉),所以电脑系统一直是Win ...

  2. 投入Html5的怀抱,最近在研究的Egret

    html5没有办法不关注,实在太火热了,几年前还不行,如今确是环境较好,typescript语言很好学习,可能基于之前的基础,不到一个星期就基本上差不多了,虽然还有一些小问题,但那都是经验积累下来可以 ...

  3. font和lineheight冲突。

    font:14px bold arial; line-height:40px; 这样写font的话line-height不会有效,只要把font拆分写就有效,chrome ie ff下都是.

  4. NOI2018准备Day4

    上午9点20至11点50就做出了一道题,一个很基础的二分挡住了,原因是浮点数精度问题的处理,现在还搞不懂,为什么用double存进去两位小数过不了,用double存进去两位小数再*100再/100就能 ...

  5. jboss CLI 命令行接口学习(适用JBOSS EAP 6.2+)

    一.确认CLI所使用的端口 以domain模式为例,查看domain controller(也就是master主机)上的host.xml <management-interfaces> & ...

  6. 基于Spring的简易SSO设计

    通常稍微规模大一些的企业,内部已经有很多的应用系统,多个系统整合首先要解决的便是“统一登录(SSO)”问题,之前写过一篇 利用Membership实现SSO(单点登录) ,java环境下已经有一些开源 ...

  7. C118+Osmocom-bb+Openbts搭建小型基站

    演示图片: 演示视频: 交流论坛:GsMsEc 交流Q群:

  8. CUDA1.1-函数类型限定符与变量类型限定符

    这部分来自于<CUDA_C_Programming_Guide.pdf>,看完<GPU高性能变成CUDA实战>的第四章,觉得这本书还是很好的,是一种循序渐进式的书,值得看,而不 ...

  9. location.href 实现点击下载功能

    如果页面上要实现一个点击下载的功能,传统做法是使用一个 a 标签,然后将该标签的 href 属性地址指向下载文件在服务端的地址(相对地址或者绝对地址),比如这样: 能这样实现是因为,在浏览器地址栏输入 ...

  10. MAC OS上Nginx安装

    admin@admindeMac:local]$ brew install nginx ==> Installing dependencies for nginx: pcre, openssl ...