(1).秦九韶算法:
把一个N次的多项式,改写成如下形式:
[egin{array}{l}
f( x ) = a_n{x^n} + a_{n - 1}x^{n - 1} + cdots + {a_1}x + a_0\
= ( {a_n}x^{n - 1} + a_{n - 1}x^{n - 2} + cdots + {a_1} )x + a_0\
= ( ( ( {a_n}x + a_{n - 1} )x + a_{n - 2}) + cdots + a_1)x + a_0
end{array}]令:
[egin{array}{*{20}{l}}
{{b_1} = {a_n}x + {a_{n - 1}}}\
{{b_2} = {b_1}x + {a_{n - { m{2}}}}}\
{ cdots cdots }\
{{b_n} = {b_{n - 1}}x + {a_{ m{1}}}}
end{array}]则:
[f(x) = {a_n}{x^n} + cdots + a_0 = b_n]
这样做的好处就是:通过递归求解数列 {bn} 我们最终获得多项式的值,减少了幂运算和防止INT范围溢出。
(2).快速幂算法
求解:${a^b}mod c = ?$令:
[b_{( 10 )} = overline {a_n}a_{n - 1} cdots a_1{a_0} _{( 2 )} = a_n{2^n} + a_{n - 1}2^{n - 1} + cdots + a_1{2^1} + a_0]故:
[a^b = a^{a_n{2^n}}a^{a_{n - 1}2^{n - 1}} cdots a^{a_1{2^1}} a^{a_0}]
[{a^b}mod c = [ {( {{a^{{a_n}{2^n}}}mod c} )( {{a^{{a_{n - 1}}{2^{n - 1}}}}mod c} ) cdots ( {{a^{{a_1}{2^1}}}mod c} )( {{a^{{a_0}}}mod c} )} mod c,{a_i} = { {0,1|i in [ {1,n} ]} }]实际上,我们还可以用移位方式表示 ai :
[a_i = b > > i & 1,i in [ 1,n ]]
(3).代码模板:
1: //整数的快速幂 m^n % k 的快速幂:2: long long quickpow(long long m , long long n , long long k){3: long long ans = 1;4: while(n){5: if(n&1)//如果n是奇数6: ans = (ans * m) % k;7: n = n >> 1;//位运算“右移1类似除1”8: m = (m * m) % k;9: }10: return ans;11: }
(4).代码解释:
这里的a为底数,需要区别于ai :
[egin{array}{l}
if( a_i == 1),a^{a_i{2^i}} = a^{2^i}\
if (a_i == 0),a^{a_i{2^i}} = 1
end{array}]在这里我们记:(i代表第i位)
[p_k = a^k,k = 2^i]则我们可以获得数列 { pk },递推式为:
[p_k = ( {p_{k - 1}^2} )mod c, p_0 = amod c]
(5).样例实验
hdu 2035 - 人见人爱A^B
题意:求解(A^B)%1000
解法:自然是快速幂。
附上我的代码:
1: #include<cstdio>2: #include<cstdlib>3: using namespace std;4: //计算(a^p)%m5: long long int fast_pow(long long a,long long p,long long m){6: long long ans=1;7: while(p){8: if(p&1) ans=(ans*a)%m;9: p=p>>1;10: a=(a*a)%m;11: }12: return ans;13: }14: int main(){15: int n,m;16: while(scanf("%d %d",&n,&m)!=EOF && (n||m)){17: printf("%d ",fast_pow(n,m,1000));18: }19: }