二维树状数组学习之一：彻底理解

 当要频繁的对数组元素进行修改,同时又要频繁的查询数组内任一区间元素之和的时候,可以考虑使用树状数组.

通常对一维数组最直接的算法可以在O(1)时间内完成一次修改,但是需要O(n)时间来进行一次查询.而树状数组的修改和查询均可在O(log(n))的时间内完成.

一、回顾一维树状数组
假设一维数组为A[i](i=1,2,...n),则与它对应的树状数组C[i](i=1,2,...n)是这样定义的：

C1 = A1
C2 = A1 + A2
C3 = A3
C4 = A1 + A2 + A3 + A4
C5 = A5
C6 = A5 + A6
C7 = A7
C8 = A1 + A2 + A3 + A4 + A5 + A6 + A7 + A8
……
C16 = A1 + A2 + A3 + A4 + A5 + A6 + A7 + A8 + A9 + A10 + A11 + A12 + A13 + A14 + A15 + A16
......



(1)C[t]展开以后有多少项？由下面公式计算：

int lowbit(int t){//计算c[t]展开的项数
return t&(-t);
}

C[t]展开的项数就是lowbit(t),C[t]就是从A[t]开始往左连续求lowbit(t)个数的和.


(2)修改
比如修改了A3,必须修改C3,C4,C8,C16,C32,C64...
当我们修改A[i]的值时，可以从C[i]往根节点一路上溯，调整这条路上的所有C[]即可，对于节点i，父节点下标 p=i+lowbit(i)

//给A[i]加上 x后，更新一系列C[j]

Java代码

1. update(int i,int x){
2. while(i<=n){
3. c[i]=c[i]+x;
4. i=i+lowbit(i);
5. }
6. }

update(int i,int x){    
 while(i<=n){   
    c[i]=c[i]+x;    
    i=i+lowbit(i);    
     }    
}    

(3)求数列A[]的前n项和，只需找到n以前的所有最大子树，把其根节点的C加起来即可。

Java代码

1. 如：Sun(1)=C[1]=A[1];
2. Sun(2)=C[2]=A[1]+A[2];
3. Sun(3)=C[3]+C[2]=A[1]+A[2]+A[3];
4. Sun(4)=C[4]=A[1]+A[2]+A[3]+A[4];
5. Sun(5)=C[5]+C[4];
6. Sun(6)=C[6]+C[4];
7. Sun(7)=C[7]+C[6]+C[4];
8. Sun(8)=C[8];
9. ,,,,,,
11. int Sum(int n) //求前n项的和.
12. {
13. int sum=0;
14. while(n>0)
15. {
16. sum+=C[n];
17. n=n-lowbit(n);
18. }
19. return sum;
20. }
24. lowbit(1)=1 lowbit(2)=2 lowbit(3)=1 lowbit(4)=4
25. lowbit(5)=1 lowbit(6)=2 lowbit(7)=1 lowbit(8)=8
26. lowbit(9)=1 lowbit(10)=2 lowbit(11)=1 lowbit(12)=4
27. lowbit(13)=1 lowbit(14)=2 lowbit(15)=1 lowbit(16)=16
28. lowbit(17)=1 lowbit(18)=2 lowbit(19)=1 lowbit(20)=4
29. lowbit(21)=1 lowbit(22)=2 lowbit(23)=1 lowbit(24)=8
30. lowbit(25)=1 lowbit(26)=2 lowbit(27)=1 lowbit(28)=4
31. lowbit(29)=1 lowbit(30)=2 lowbit(31)=1 lowbit(32)=32
32. lowbit(33)=1 lowbit(34)=2 lowbit(35)=1 lowbit(36)=4
33. lowbit(37)=1 lowbit(38)=2 lowbit(39)=1 lowbit(40)=8
34. lowbit(41)=1 lowbit(42)=2 lowbit(43)=1 lowbit(44)=4
35. lowbit(45)=1 lowbit(46)=2 lowbit(47)=1 lowbit(48)=16
36. lowbit(49)=1 lowbit(50)=2 lowbit(51)=1 lowbit(52)=4
37. lowbit(53)=1 lowbit(54)=2 lowbit(55)=1 lowbit(56)=8
38. lowbit(57)=1 lowbit(58)=2 lowbit(59)=1 lowbit(60)=4
39. lowbit(61)=1 lowbit(62)=2 lowbit(63)=1 lowbit(64)=64

  如：Sun(1)=C[1]=A[1];
      Sun(2)=C[2]=A[1]+A[2];
      Sun(3)=C[3]+C[2]=A[1]+A[2]+A[3];
      Sun(4)=C[4]=A[1]+A[2]+A[3]+A[4];
      Sun(5)=C[5]+C[4];
      Sun(6)=C[6]+C[4];
      Sun(7)=C[7]+C[6]+C[4];
      Sun(8)=C[8];
      ,,,,,,
  
int Sum(int n) //求前n项的和.   
{    
    int sum=0;    
    while(n>0)    
    {    
         sum+=C[n];    
         n=n-lowbit(n);    
    }        
    return sum;    
}  

 

lowbit(1)=1       lowbit(2)=2       lowbit(3)=1      lowbit(4)=4  
 lowbit(5)=1       lowbit(6)=2       lowbit(7)=1      lowbit(8)=8  
 lowbit(9)=1      lowbit(10)=2      lowbit(11)=1      lowbit(12)=4  
lowbit(13)=1      lowbit(14)=2      lowbit(15)=1      lowbit(16)=16  
lowbit(17)=1      lowbit(18)=2      lowbit(19)=1      lowbit(20)=4  
lowbit(21)=1      lowbit(22)=2      lowbit(23)=1      lowbit(24)=8  
lowbit(25)=1      lowbit(26)=2      lowbit(27)=1      lowbit(28)=4  
lowbit(29)=1      lowbit(30)=2      lowbit(31)=1      lowbit(32)=32  
lowbit(33)=1      lowbit(34)=2      lowbit(35)=1      lowbit(36)=4  
lowbit(37)=1      lowbit(38)=2      lowbit(39)=1      lowbit(40)=8  
lowbit(41)=1      lowbit(42)=2      lowbit(43)=1      lowbit(44)=4  
lowbit(45)=1      lowbit(46)=2      lowbit(47)=1      lowbit(48)=16  
lowbit(49)=1      lowbit(50)=2      lowbit(51)=1      lowbit(52)=4  
lowbit(53)=1      lowbit(54)=2      lowbit(55)=1      lowbit(56)=8  
lowbit(57)=1      lowbit(58)=2      lowbit(59)=1      lowbit(60)=4  
lowbit(61)=1      lowbit(62)=2      lowbit(63)=1      lowbit(64)=64