To the moon
Time Limit: 4000/2000 MS (Java/Others) Memory Limit: 65536/65536 K (Java/Others)
Total Submission(s): 7293 Accepted Submission(s): 1700
Total Submission(s): 7293 Accepted Submission(s): 1700
Problem Description
Background
To The Moon is a independent game released in November 2011, it is a role-playing adventure game powered by RPG Maker.
The premise of To The Moon is based around a technology that allows us to permanently reconstruct the memory on dying man. In this problem, we'll give you a chance, to implement the logic behind the scene.
You‘ve been given N integers A [1], A [2],..., A [N]. On these integers, you need to implement the following operations:
1. C l r d: Adding a constant d for every {A i | l <= i <= r}, and increase the time stamp by 1, this is the only operation that will cause the time stamp increase.
2. Q l r: Querying the current sum of {A i | l <= i <= r}.
3. H l r t: Querying a history sum of {A i | l <= i <= r} in time t.
4. B t: Back to time t. And once you decide return to a past, you can never be access to a forward edition anymore.
.. N, M ≤ 10 5, |A [i]| ≤ 10 9, 1 ≤ l ≤ r ≤ N, |d| ≤ 10 4 .. the system start from time 0, and the first modification is in time 1, t ≥ 0, and won't introduce you to a future state.
To The Moon is a independent game released in November 2011, it is a role-playing adventure game powered by RPG Maker.
The premise of To The Moon is based around a technology that allows us to permanently reconstruct the memory on dying man. In this problem, we'll give you a chance, to implement the logic behind the scene.
You‘ve been given N integers A [1], A [2],..., A [N]. On these integers, you need to implement the following operations:
1. C l r d: Adding a constant d for every {A i | l <= i <= r}, and increase the time stamp by 1, this is the only operation that will cause the time stamp increase.
2. Q l r: Querying the current sum of {A i | l <= i <= r}.
3. H l r t: Querying a history sum of {A i | l <= i <= r} in time t.
4. B t: Back to time t. And once you decide return to a past, you can never be access to a forward edition anymore.
.. N, M ≤ 10 5, |A [i]| ≤ 10 9, 1 ≤ l ≤ r ≤ N, |d| ≤ 10 4 .. the system start from time 0, and the first modification is in time 1, t ≥ 0, and won't introduce you to a future state.
Input
n m
A 1 A 2 ... A n
... (here following the m operations. )
A 1 A 2 ... A n
... (here following the m operations. )
Output
... (for each query, simply print the result. )
Sample Input
10 5 1 2 3 4 5 6 7 8 9 10 Q 4 4 Q 1 10 Q 2 4 C 3 6 3 Q 2 4 2 4 0 0 C 1 1 1 C 2 2 -1 Q 1 2 H 1 2 1
Sample Output
4 55 9 15 0 1
大致题意:支持Q、C、H、B四种操作,分别对应查询当前区间[l,r]的和,修改区间[l,r]并使得时间加一,查询第x个时刻区间[l,r]的区间和,回到第x个时刻。
纯数据结构题目,支持回到、查询历史版本,显然是一个可持久化数据结构,而支持查询区间和以及区间修改,首选的就是可持久化线段树了。这里有一个时间的定义,所以我们对每一个时间动态建立一个线段树,每个线段树继承之前时间的信息。对于两个查询操作,直接在对应时刻版本的线段树里面查询即可,我们用rt[i]表示第i个时刻对应线段树的根。关键就在于这个C更新操作了。
对于区间更新操作,在普通线段树中,我们用的是找到对应区间,然后打lazy标记,在下次访问该点的时候再下传lazy标记的方式来进行的。但是,如果考虑在主席树种这么操作,我每一次的下传lazy标记又相当于是一次更新,理论上说需要再新开logN个节点来存储变化。或者说如果不需要新开节点,在处理种也会比较麻烦,所以我们考虑不下传lazy标记。其实这种操作在普通线段树中也有,只是我们有用过而已。对于每个修改,我累加lazy的数值,然后在询问的时候,如果询问区间小于当前区间,那么最后的答案就要累加上当前区间的lazy数值。这样从上到下就会累积每一个区间修改的变化,得到正确的结果。
对于需要维护区间极值的问题,我们同样也是可以用这种不下传lazy标记的形式进行。具体见代码:
#include<bits/stdc++.h>
#define LL long long
#define N 150010
using namespace std;
int a[N],b[N],rt[N<<4];
struct Persistent_SegTree
{
struct node{int l,r;LL sum,lazy;} T[N<<4];
int cnt; void init(){cnt=0;}
void build(int &i,int l,int r)
{
i=++cnt;
T[i]=node{0,0,0,0};
if (l==r)
{
T[i].sum=a[l];
return;
}
int mid=(l+r)>>1;
build(T[i].l,l,mid);
build(T[i].r,mid+1,r);
T[i].sum=T[T[i].l].sum+T[T[i].r].sum;
}
void update(int &i,int old,int l,int r,int L,int R,LL x)
{
i=++cnt; T[i]=T[old]; //继承old的信息
T[i].sum+=1LL*(r-l+1)*x;
if (l==L&&r==R)
{
T[i].lazy+=x; return;
}
int mid=(L+R)>>1;
if (r<=mid) update(T[i].l,T[old].l,l,r,L,mid,x);
else if (l>mid) update(T[i].r,T[old].r,l,r,mid+1,R,x);
else
{
update(T[i].l,T[old].l,l,mid,L,mid,x);
update(T[i].r,T[old].r,mid+1,r,mid+1,R,x);
}
}
LL getsum(int i,int l,int r,int L,int R)
{
if (l==L&&R==r) return T[i].sum;
int mid=(L+R)>>1;
LL ans=T[i].lazy*1LL*(r-l+1); //每次要累加所有的区间变化
if (r<=mid) ans+=getsum(T[i].l,l,r,L,mid);
else if (l>mid) ans+=getsum(T[i].r,l,r,mid+1,R);
else ans+=getsum(T[i].l,l,mid,L,mid)+getsum(T[i].r,mid+1,r,mid+1,R);
return ans;
}
} Persist;
int main()
{
int n,q;
while(~scanf("%d%d",&n,&q))
{
int t=0;
Persist.init();
for(int i=1;i<=n;i++)
scanf("%d",&a[i]);
Persist.build(rt[0],1,n);
while(q--)
{
char op[10];
int l,r; LL x;
scanf("%s%d",op,&l);
if (op[0]=='Q')
{
scanf("%d",&r);
printf("%lld\n",Persist.getsum(rt[t],l,r,1,n));
}
if (op[0]=='C')
{
++t; scanf("%d%lld",&r,&x);
Persist.update(rt[t],rt[t-1],l,r,1,n,x);
}
if (op[0]=='H')
{
scanf("%d%lld",&r,&x);
printf("%lld\n",Persist.getsum(rt[x],l,r,1,n));
}
if (op[0]=='B') t=l;
}
}
return 0;
}