Codeforces 390 E. Inna and Large Sweet Matrix


主要是树状数组的改段求段操作。。。

E. Inna and Large Sweet Matrix
time limit per test
1 second
memory limit per test
512 megabytes
input
standard input
output
standard output

Inna loves sweets very much. That's why she wants to play the "Sweet Matrix" game with Dima and Sereja. But Sereja is a large person, so the game proved small for him. Sereja suggested playing the "Large Sweet Matrix" game.

The "Large Sweet Matrix" playing field is an n × m matrix. Let's number the rows of the matrix from 1 to n, and the columns — from 1 tom. Let's denote the cell in the i-th row and j-th column as (i, j). Each cell of the matrix can contain multiple candies, initially all cells are empty. The game goes in w moves, during each move one of the two following events occurs:

  1. Sereja chooses five integers x1, y1, x2, y2, v (x1 ≤ x2, y1 ≤ y2) and adds v candies to each matrix cell (i, j) (x1 ≤ i ≤ x2y1 ≤ j ≤ y2).
  2. Sereja chooses four integers x1, y1, x2, y2 (x1 ≤ x2, y1 ≤ y2). Then he asks Dima to calculate the total number of candies in cells(i, j) (x1 ≤ i ≤ x2y1 ≤ j ≤ y2) and he asks Inna to calculate the total number of candies in the cells of matrix (p, q), which meet the following logical criteria: (p < x1 OR p > x2) AND (q < y1 OR q > y2). Finally, Sereja asks to write down the difference between the number Dima has calculated and the number Inna has calculated (D - I).

Unfortunately, Sereja's matrix is really huge. That's why Inna and Dima aren't coping with the calculating. Help them!

Input

The first line of the input contains three integers nm and w (3 ≤ n, m ≤ 4·106; 1 ≤ w ≤ 105).

The next w lines describe the moves that were made in the game.

  • A line that describes an event of the first type contains 6 integers: 0x1y1x2y2 and v (1 ≤ x1 ≤ x2 ≤ n; 1 ≤ y1 ≤ y2 ≤ m; 1 ≤ v ≤ 109).
  • A line that describes an event of the second type contains 5 integers: 1x1y1x2y2 (2 ≤ x1 ≤ x2 ≤ n - 1; 2 ≤ y1 ≤ y2 ≤ m - 1).

It is guaranteed that the second type move occurs at least once. It is guaranteed that a single operation will not add more than 109candies.

Be careful, the constraints are very large, so please use optimal data structures. Max-tests will be in pretests.

Output

For each second type move print a single integer on a single line — the difference between Dima and Inna's numbers.

Sample test(s)
input
4 5 5
0 1 1 2 3 2
0 2 2 3 3 3
0 1 5 4 5 1
1 2 3 3 4
1 3 4 3 4
output
2
-21
Note

Note to the sample. After the first query the matrix looks as:

22200
22200 
00000
00000

After the second one it is:

22200
25500
03300 
00000

After the third one it is:

22201
25501
03301
00001

For the fourth query, Dima's sum equals 5 + 0 + 3 + 0 = 8 and Inna's sum equals 4 + 1 + 0 + 1 = 6. The answer to the query equals 8 - 6 = 2. For the fifth query, Dima's sum equals 0 and Inna's sum equals 18 + 2 + 0 + 1 = 21. The answer to the query is 0 - 21 = -21.





#include <iostream>
#include <cstdio>
#include <cstring>
#include <algorithm>

using namespace std;

typedef long long int LL;

const int maxn=4000400;

int n,m,w,nm;
LL B[2][maxn],C[2][maxn];

int lowbit(int x)
{
	return x&(-x);
}

void add_b(int id,int p,LL v)
{
	for(int i=p;i;i-=lowbit(i)) B[id][i]+=v;
}

void add_c(int id,int p,LL v)
{
	for(int i=p;i<=nm;i+=lowbit(i)) C[id][i]+=v*p;
}

LL sum_b(int id,int p)
{
	LL sum=0;
	for(int i=p;i<=nm;i+=lowbit(i)) sum+=B[id][i];
	return sum;
}

LL sum_c(int id,int p)
{
	LL sum=0;
	for(int i=p;i;i-=lowbit(i)) sum+=C[id][i];
	return sum;
}

void ADD(int id,int l,int r,LL v)
{
	add_b(id,r,v);
	add_c(id,r,v);
	if(l-1)
	{
		add_b(id,l-1,-v);
		add_c(id,l-1,-v);
	}
}

LL sUm(int id,int p)
{
	if(p)	return sum_b(id,p)*p+sum_c(id,p-1);
	else return 0;
}

LL SUM(int id,int l,int r)
{
	return sUm(id,r)-sUm(id,l-1);
}

int main()
{
	scanf("%d%d%d",&n,&m,&w);
	nm=max(n,m)+10;
	while(w--)
	{
		int c,x1,x2,y1,y2,v,dx,dy;
		scanf("%d%d%d%d%d",&c,&x1,&y1,&x2,&y2);
		dx=x2-x1+1;dy=y2-y1+1;
		if(c==0)
		{
			scanf("%d",&v);
			ADD(0,x1,x2,v*dy);
			ADD(1,y1,y2,v*dx);
		}
		else
		{
			printf("%I64d\n",SUM(1,y1,y2)-SUM(0,1,x1-1)-SUM(0,x2+1,nm));
		}
	}
	return 0;
}





深度学习是机器学习的一个子领域,它基于人工神经网络的研究,特别是利用多层次的神经网络来进行学习和模式识别。深度学习模型能够学习数据的高层次特征,这些特征对于图像和语音识别、自然语言处理、医学图像分析等应用至关重要。以下是深度学习的一些关键概念和组成部分: 1. **神经网络(Neural Networks)**:深度学习的基础是人工神经网络,它是由多个层组成的网络结构,包括输入层、隐藏层和输出层。每个层由多个神经元组成,神经元之间通过权重连接。 2. **前馈神经网络(Feedforward Neural Networks)**:这是最常见的神经网络类型,信息从输入层流向隐藏层,最终到达输出层。 3. **卷积神经网络(Convolutional Neural Networks, CNNs)**:这种网络特别适合处理具有网格结构的数据,如图像。它们使用卷积层来提取图像的特征。 4. **循环神经网络(Recurrent Neural Networks, RNNs)**:这种网络能够处理序列数据,如时间序列或自然语言,因为它们具有记忆功能,能够捕捉数据中的时间依赖性。 5. **长短期记忆网络(Long Short-Term Memory, LSTM)**:LSTM 是一种特殊的 RNN,它能够学习长期依赖关系,非常适合复杂的序列预测任务。 6. **生成对抗网络(Generative Adversarial Networks, GANs)**:由两个网络组成,一个生成器和一个判别器,它们相互竞争,生成器生成数据,判别器评估数据的真实性。 7. **深度学习框架**:如 TensorFlow、Keras、PyTorch 等,这些框架提供了构建、训练和部署深度学习模型的工具和库。 8. **激活函数(Activation Functions)**:如 ReLU、Sigmoid、Tanh 等,它们在神经网络中用于添加非线性,使得网络能够学习复杂的函数。 9. **损失函数(Loss Functions)**:用于评估模型的预测与真实值之间的差异,常见的损失函数包括均方误差(MSE)、交叉熵(Cross-Entropy)等。 10. **优化算法(Optimization Algorithms)**:如梯度下降(Gradient Descent)、随机梯度下降(SGD)、Adam 等,用于更新网络权重,以最小化损失函数。 11. **正则化(Regularization)**:技术如 Dropout、L1/L2 正则化等,用于防止模型过拟合。 12. **迁移学习(Transfer Learning)**:利用在一个任务上训练好的模型来提高另一个相关任务的性能。 深度学习在许多领域都取得了显著的成就,但它也面临着一些挑战,如对大量数据的依赖、模型的解释性差、计算资源消耗大等。研究人员正在不断探索新的方法来解决这些问题。
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