高精度算法 (C/C++)
做ACM题的时候,经常遇到大数的加减乘除,乘幂,阶乘的计算,这时给定的数据类型往往不够表示最后结果,这时就需要用到高精度算法。高精度算法的本质是把大数拆成若干固定长度的块,然后对每一块进行相应的运算。这里以考虑4位数字为一块为例,且输入的大数均为正整数(也可以考虑其他位,但要注意在每一块进行相应运算时不能超出数据类型的数值范围;有负整数的话读入时判断一下正负号在决定运算)。
1. 高精度加法
以3479957928375817 + 897259321544245为例:
3479 | 9579 | 2837 | 5817 |
---|
+897 | +2593 | +2154 | +4245 |
= | = | = | = |
4376 | 12172 | 4991 | 10062 |
进位0 | 进位1 | 进位0 | 进位1 |
4377 | 2172 | 4992 | 0062 |
C语言实现代码如下:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define N 200
//整数乘幂运算函数
int Pow(int a, int b)
{
int i = 0, result = 1;
for(i = 0; i < b; ++i)
{
result *= a;
}
return result;
}
//High Precision Of Addition
int main()
{
char stra[N], strb[N]; //字符串数组,以字符形式储存两个大数;
int i = 0, step = 4, carry = 0; //step表示块长,carry为进位位;
int lengtha, lengthb, maxlength, resultsize; //maxlength表示stra和strb二者长度较大的那个;
int numa[N], numb[N],numc[N]; //依次储存被加数,加数,和;
memset(numa, 0, sizeof(numa));
memset(numb, 0, sizeof(numb));
memset(numc, 0, sizeof(numc)); //初始化为零;
scanf("%s%s", stra, strb);
lengtha = strlen(stra);
lengthb = strlen(strb); //计算两个大数的长度
//字符数字转为四位一块的整数数字
for(i = lengtha-1; i >= 0; --i)
{
numa[(lengtha-1-i)/step] += (stra[i]-'0')*Pow(10,(lengtha-1-i)%step);
}
for(i = lengthb-1; i >= 0; --i)
{
numb[(lengthb-1-i)/step] += (strb[i]-'0')*Pow(10,(lengthb-1-i)%step);
}
maxlength = lengtha > lengthb ? lengtha : lengthb;
//逐块相加,并进位
for(i = 0; i <= maxlength/step; ++i)
{
numc[i] = (numa[i] + numb[i])%Pow(10, step) + carry; //计算和
carry = (numa[i] + numb[i])/Pow(10, step); //计算进位
}
//计算最后和的块的总数
resultsize = numc[maxlength/step] > 0 ? maxlength/step : maxlength/step - 1;
printf("%d", numc[resultsize]);
for(i = resultsize-1; i >= 0; --i)
{
printf("%04d", numc[i]); //右对齐,补零输出;
}
printf("\n");
return 0;
}
2. 高精度减法
与加法类似,不同的是要注意正负号和显示位数的变化。以99999037289799 - 100004642015000为例:
先判断被减数和减数哪个大,显然这里减数大,故输出结果为负数。在用大数减去小数,(若某一块相减为负数,则要向高位块借位)如下表
100 | 0046 | 4201 | 5000 |
---|
-99 | -9990 | -3728 | -9799 |
1 | 56 | 473 | 5201 |
借位0 | 借位1 | 借位0 | 借位1 |
0 | 56 | 472 | 5201 |
C语言实现代码如下:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define N 200
//整数乘幂运算函数
int Pow(int a, int b)
{
int i = 0, result = 1;
for(i = 0; i < b; ++i)
{
result *= a;
}
return result;
}
//High Precision Of Subtraction
int main()
{
char stra[N], strb[N]; //字符串数组,以字符形式储存两个大数;
int i = 0, step = 4, borrow = 0, mark = 0; //step表示块长,borrow为借位位, mark为结果符号位;
int lengtha, lengthb, maxlength, resultsize; //maxlength表示stra和strb二者长度较大的那个;
int numa[N], numb[N],numc[N], *maxnum, *minnum; //依次储存被减数,减数,和;
memset(stra, 0, sizeof(stra));
memset(strb, 0, sizeof(strb));
memset(numa, 0, sizeof(numa));
memset(numb, 0, sizeof(numb));
memset(numc, 0, sizeof(numc)); //初始化为零;
scanf("%s%s", stra, strb);
lengtha = strlen(stra);
lengthb = strlen(strb); //计算两个大数的长度
maxlength = lengtha >= lengthb ? lengtha : lengthb;
//字符数字转为四位一块的整数数字
for(i = lengtha-1; i >= 0; --i)
{
numa[(lengtha-1-i)/step] += (stra[i]-'0')*Pow(10,(lengtha-1-i)%step);
}
for(i = lengthb-1; i >= 0; --i)
{
numb[(lengthb-1-i)/step] += (strb[i]-'0')*Pow(10,(lengthb-1-i)%step);
}
//找出较大的数
maxnum = numa;
minnum = numb;
mark = 1;
for(i = (maxlength-1)/step; i >= 0; --i)
{
if(numa[i] > numb[i])
{
maxnum = numa;
minnum = numb;
mark = 1;
break;
}
else if(numa[i] < numb[i])
{
maxnum = numb;
minnum = numa;
mark = -1;
break;
}
}
//逐块相减,并借位
for(i = 0; i <= maxlength/step; ++i)
{
numc[i] = (maxnum[i] - minnum[i] + Pow(10, step) + borrow)%Pow(10,step); //计算差
borrow = (maxnum[i] - minnum[i] + Pow(10, step) + borrow)/Pow(10, step) - 1; //计算借位
}
//计算最后和的块的总数
resultsize = maxlength/step;
while(!numc[resultsize]) --resultsize;
printf("%d", mark*numc[resultsize]);
for(i = resultsize-1; i >= 0; --i)
{
printf("%04d", numc[i]); //右对齐,补零输出;
}
printf("\n");
return 0;
}
3. 高精度乘法
乘法可以看作是乘数每一位与被乘数相乘后再相加,以4296556241 x 56241为例:
被乘数x乘数 | 42 | 9655 | 6241 |
---|
1 | 42 | 9655 | 6241 |
4 | 168*10 | 38620*10 | 24964*10 |
2 | 84*100 | 19310*100 | 12482*100 |
6 | 252*1000 | 57930*1000 | 37446*1000 |
5 | 210*10000 | 48275*10000 | 31205*10000 |
累加和 | 2362122 | 543006855 | 351000081 |
进位(从低位向高位) | 241 | 54304 | 35100 |
C语言实现代码如下:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define N 200
//整数乘幂运算函数
int Pow(int a, int b)
{
int i = 0, result = 1;
for(i = 0; i < b; ++i)
{
result *= a;
}
return result;
}
//High Precision Of Multiplication
int main()
{
char stra[N], strb[N]; //字符串数组,以字符形式储存两个大数;
int i = 0, j = 0, k = 0, step = 4, carry = 0; //step表示块长,carry为进位位;
int lengtha, lengthb, resultsize, tmpsize, eachnum; //resultsize储存块的总数,eachnum用来储存乘数的每一位
int numa[N], numb[N], numc[N], tmp[N]; //依次储存被乘数数&积,乘数;
memset(numa, 0, sizeof(numa));
memset(numb, 0, sizeof(numb));
memset(numc, 0, sizeof(numc)); //初始化为零;
scanf("%s%s", stra, strb);
lengtha = strlen(stra);
lengthb = strlen(strb); //计算两个大数的长度
//将被乘数字符数字转为四位一块的整数数字
for(i = lengtha-1; i >= 0; --i)
{
numa[(lengtha-1-i)/step] += (stra[i]-'0')*Pow(10,(lengtha-1-i)%step);
}
//将乘数数字字符数字转为一位一块的整数数字
for(i = lengthb-1; i >= 0; --i)
{
numb[lengthb-1-i] = strb[i]-'0';
}
resultsize = tmpsize = (lengtha-1)/step;
//取乘数的每一位与被乘数的逐块相乘,并进位;
for(i = 0; i < lengthb; ++i)
{
memcpy(tmp, numa, sizeof(numa)); //将numa数组赋值给tmp数组;
k = i/step; //k储存每一块需要向高位块移动的次数;
if(k)
{
for(j = tmpsize; j >= 0; --j)
{
tmp[j+k] = tmp[j];
tmp[j] = 0;
}
tmpsize += k;
}
//乘以乘数每一位扩展成的块;
eachnum = numb[i]*Pow(10, i%step);
for(j = 0; j <= tmpsize; ++j)
{
tmp[j] *= eachnum;
}
//大数相加
carry = 0; //进位置零;
for(j = 0; j <= resultsize; ++j)
{
numc[j] += tmp[j] + carry;
carry = numc[j]/Pow(10,step);
numc[j] %= Pow(10, step);
}
if(carry)
{
++resultsize;
numc[j] += carry;
}
}
//输出
printf("%d", numc[resultsize]);
for(i = resultsize-1; i >= 0; --i)
{
printf("%04d", numc[i]); //右对齐,补零输出;
}
printf("\n");
return 0;
}
4. 高精度除法
高精度除法有两种,一种是高精度除以低精度,另一种是高精度除以高精度。前者只需将每一块除以低精度除数即可;后者则考虑用高精度减法来实现,即每次减去高精度除数,直到减到小于除数,则减的次数即为商,剩余的即为余数。
- 高精度除以低精度
以9876342876 / 343为例:
C语言代码实现如下:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define N 200
//整数乘幂运算函数
int Pow(int a, int b)
{
int i = 0, result = 1;
for(i = 0; i < b; ++i)
{
result *= a;
}
return result;
}
//High Precision Of division
//(1)高精度除以低精度
int main()
{
char stra[N]; //字符串数组,以字符形式储存高精度被除数;
int i = 0, step = 4, carry = 0; //step表示块长,carry为高位向低位进位位;
int lengtha, resultsize;
int numa[N], numb, numc[N], numd; //依次储存被除数,除数,商, 余数;
memset(numa, 0, sizeof(numa));
memset(numc, 0, sizeof(numc)); //初始化为零;
scanf("%s%d", stra, &numb);
lengtha = strlen(stra); //计算被除数的长度
//字符数字转为四位一块的整数数字
for(i = lengtha-1; i >= 0; --i)
{
numa[(lengtha-1-i)/step] += (stra[i]-'0')*Pow(10,(lengtha-1-i)%step);
}
carry = 0; //高位向低位进位位置零
resultsize = (lengtha-1)/step;
//逐块相除,高位向低位进位
for(i = resultsize; i >= 0; --i)
{
numc[i] = (numa[i] + carry*Pow(10,step))/numb; //计算商
carry = (numa[i] + carry*Pow(10,step))%numb; //计算进位
}
numd = carry; //最低位块的余数即为整个除法的余数
//计算最后和的块的总数
while(!numc[resultsize]) --resultsize;
//输出商
printf("%d", numc[resultsize]);
for(i = resultsize-1; i >= 0; --i)
{
printf("%04d", numc[i]); //右对齐,补零输出;
}
//输出余数
printf("\n%d\n", numd);
return 0;
}
- 高精度除以高精度
以176342876 / 3453452为例:
被除数 | 176342876 |
---|
- (51 x 除数) | 51 x 3453452 |
余数 | 216824 |
商 | 51 |
C语言代码实现如下:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define N 200
//整数乘幂运算函数
int Pow(int a, int b)
{
int i = 0, result = 1;
for(i = 0; i < b; ++i)
{
result *= a;
}
return result;
}
//High Precision Of division
//(2)高精度除以高精度
int main()
{
char stra[N], strb[N]; //字符串数组,以字符形式储存两个大数;
int i = 0, step = 4, borrow = 0; //step表示块长,borrow为进位位;
int lengtha, lengthb, tmpnum, numbsize, numcsize, numdsize, maxsize, mark; //maxlength表示stra和strb二者长度较大的那个;
int numa[N], numb[N], numc[N], numd[N]; //依次储存被除数数,除数数,商,余数;
memset(stra, 0, sizeof(stra));
memset(strb, 0, sizeof(strb));
memset(numa, 0, sizeof(numa));
memset(numb, 0, sizeof(numb));
memset(numc, 0, sizeof(numc));
memset(numd, 0, sizeof(numd)); //初始化为零;
scanf("%s%s", stra, strb);
lengtha = strlen(stra);
lengthb = strlen(strb); //计算两个大数的长度
//字符数字转为四位一块的整数数字
for(i = lengtha-1; i >= 0; --i)
{
numa[(lengtha-1-i)/step] += (stra[i]-'0')*Pow(10,(lengtha-1-i)%step);
}
for(i = lengthb-1; i >= 0; --i)
{
numb[(lengthb-1-i)/step] += (strb[i]-'0')*Pow(10,(lengthb-1-i)%step);
}
memcpy(numd, numa, sizeof(numa));
numbsize = (lengthb-1)/step;
numcsize = 0;
numdsize = (lengtha-1)/step;
do
{
maxsize = numdsize > numbsize ? numdsize : numbsize;
//计算剩余数是否小于除数
mark = 1;
for(i = maxsize; i >= 0; --i)
{
if(numd[i] > numb[i])
{
mark = 1;
break;
}
else if(numd[i] < numb[i])
{
mark = -1;
break;
}
}
//判断是否余数已经小于除数
if(!(mark+1)) break;
borrow = 0; //借位置零;
//逐块相减,并借位
for(i = 0; i <= maxsize; ++i)
{
tmpnum = (numd[i] - numb[i] + Pow(10, step) + borrow)%Pow(10,step); //计算差
borrow = (numd[i] - numb[i] + Pow(10, step) + borrow)/Pow(10,step) - 1; //计算借位
numd[i] = tmpnum;
}
while(!numd[numdsize]) --numdsize;
//每减一个除数,商加一;
borrow = 1;
for(i = 0; i <= numcsize; ++i)
{
numc[i] += borrow;
borrow = numc[i]/Pow(10,step);
numc[i] %= Pow(10,step);
}
if(borrow)
{
++numcsize;
numc[i] += borrow;
}
}while(1);
printf("%d", numc[numcsize]);
for(i = numcsize-1; i >= 0; --i)
{
printf("%04d", numc[i]); //右对齐,补零输出;
}
printf("\n");
printf("%d", numd[numdsize]);
for(i = numdsize-1; i >= 0; --i)
{
printf("%04d", numd[i]); //右对齐,补零输出;
}
printf("\n");
return 0;
}