ACM-数论-广义欧拉降幂
Posted 31415926535x
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了ACM-数论-广义欧拉降幂相关的知识,希望对你有一定的参考价值。
https://www.cnblogs.com/31415926535x/p/11448002.html
曾今一时的懒,造就今日的泪
记得半年前去武大参加的省赛,当时的A题就是一个广义欧拉降幂的板子题,后来回来补了一下,因为没有交的地方,于是就测了数据就把代码扔了,,,然后,,昨天的南京网络赛就炸了,,,一样的广义欧拉降幂的板子题,,然后因为忘记了当初自己想出来的那中写法,,一直想着回想起之前的写法,,然后到结束都没弄出来,,,emmmm,,
赛后看了一下别人的解法,,别人的处理方法很巧妙,,当然另一个种返回两个值的(pair)的解法就是武大的标程,,,,(到最后之前想出的写法还是每能推出来,,都开始怀疑自己当时有没有真的推出来,,,,,
思路
广义欧拉降幂没啥好说的,,就是那个公式:
对于求 \\(a^b(mod \\ p)\\) 可以转换为:
\\[ a^b = \\begincases a^b \\% \\phi (p) &gcd(a, p)=1 \\ a^b &gcd(a, p) \\neq 1, b < \\phi (p) \\ a^b \\% \\phi (p) + \\phi (p) &gcd(a, p) \\neq 1, b \\ge \\phi (p)\\ \\endcases \\]
公式很简单,,但是如果是求 \\(a_1^a_2^a_3^... (mod \\ p)\\) 类似这样的值的话,显然要递归从上往下求(刚开始弄成了从下往上求,,口胡了一段时间,,,,),,但是再递归求的时候要考虑每一次 \\(b\\) 和 \\(\\phi (p)\\) 的关系,,然后选择哪一个等式,,,这样就麻烦了,,可以用一个 pair
什么的来保存一个标志变量来决定递归的上一层要不要 \\(+ \\phi (p)\\) ,,另一种巧妙地方式是修改一下 取模 的过程,,这样就不用考虑了,,,具体的推导过程在这里
所有的取模的步骤改成这样:
inline ll modulo(ll x, ll mod)return x < mod ? x : x % mod + mod;
这样保证 \\(b \\ge \\phi (p)\\),,然后就少了判断的情况
题目
南京网络赛B supper_log
这道题按题目的意思推几项样例就能看出是要求一个 \\(a^a^a^a^... mod \\ m (一共有b个a)\\) 的值,,直接降幂求就可以了,, 记得特判 b=0 的情况
代码
群里很多大佬用的方法,重置取模的流程
#include <bits/stdc++.h>
#define aaa cout<<233<<endl;
#define endl '\\n'
#define pb push_back
using namespace std;
typedef long long ll;
typedef unsigned long long ull;
typedef long double ld;
// mt19937 rnd(time(0));
const int inf = 0x3f3f3f3f;//1061109567 > 1e9
const ll linf = 0x3f3f3f3f3f3f3f3f;
const double eps = 1e-7;
const double pi = 3.14159265358979;
const int maxn = 1e3 + 5;
const int maxm = 1e3 + 5;
const int mod = 1e9 + 7;
ll f(ll x, ll a)
if(x < 1)return -1;
return 1 + f((ll)(log(x) / log(a)), a);
inline ll modulo(ll x, ll mod)return x < mod ? x : x % mod + mod;
inline ll pow_(ll a, ll b, ll p)
ll ret = 1;
while(b)
if(b & 1)ret = modulo(ret * a, p);
a = modulo(a * a, p);
b >>= 1;
return ret;
inline ll phi(ll x)
ll ans = x;
for(ll i = 2; i * i <= x; ++i)
if(x % i == 0)
ans = ans / i * (i - 1);
while(x % i == 0)x /= i;
if(x > 1)ans = ans / x * (x - 1);
return ans;
ll gcd(ll a, ll b)
if(b == 0)return a;
return gcd(b, a % b);
ll f(ll a, ll b, ll k, ll p)
if(p == 1)return 1;
if(k == 0)return 1;
return pow_(a, f(a, a, k - 1, phi(p)), p);
int main()
// double pp = clock();
// freopen("233.in", "r", stdin);
// freopen("233.out", "w", stdout);
// ios_base::sync_with_stdio(0);
// cin.tie(0);cout.tie(0);
int t; cin >> t;
while(t--)
ll a, b, m;
cin >> a >> b >> m;
// cout << a << b << m << endl;
if(b == 0)
cout << 1 % m << endl;
continue;
ll ans = f(a, a, b, m) % m;
// if(a == 1)ans = 1 % m;
// cout << ans << " " << ans % m << endl;
cout << ans << endl;
// cout << endl << (clock() - pp) / CLOCKS_PER_SEC << endl;
return 0;
pair记录上一层
武大那场的标程,,直接改了下输入,,
#include <bits/stdc++.h>
using namespace std;
#define ll long long
const int N = 1000010;
int prime[N + 1], isprime[N + 1];
int tot, phi[N + 1];
struct P
ll ans;
bool v;
P(ll _ans, bool _v)
ans = _ans;
v = _v;
;
ll gcd(ll a, ll b)
return b ? gcd(b, a % b) : a;
P qpow(ll A, ll B, ll C)
ll re = 1;
bool flag = 1;
while (B)
if (B & 1)
if ((re *= A) >= C)
flag = 0;
re = re % C;
B = B >> 1;
if (B)
if (A >= C)
flag = 0;
A %= C;
if ((A *= A) >= C)
flag = 0;
A %= C;
return P(re, flag);
void getphi()
phi[1] = 1;
isprime[1] = 1;
for (int i = 2; i <= N; i++)
if (!isprime[i])
prime[++tot] = i;
phi[i] = i - 1;
for (int j = 1; j <= tot && i * prime[j] <= N; j++)
isprime[i * prime[j]] = 1;
if (i % prime[j] == 0)
phi[i * prime[j]] = phi[i] * prime[j];
break;
else
phi[i * prime[j]] = phi[i] * phi[prime[j]];
inline ll Euler(ll x)
return phi[x];
//题目可以再复杂一点模数可以到longlong
// ll ans = x;
// for (int i = 1; i <= tot && prime[i] * prime[i] <= x; i++)
//
// if (x % prime[i] == 0)
//
// ans = ans / prime[i] * (prime[i] - 1);
// while (x % prime[i] == 0)
// x /= prime[i];
//
//
// if (x > 1)
// ans = ans / x * (x - 1);
// return ans;
P f(ll a, ll b, ll k, ll p)
if (p == 1)
return P(0, 0);
if (k == 0)
return P(a % p, a < p);
ll ep = Euler(p);
P tmp = f(b, b, k - 1, ep);
if (gcd(a, p) == 1)
return qpow(a, tmp.ans, p);
if (tmp.v == false)
tmp.ans += ep;
return qpow(a, tmp.ans, p);
int main()
//double pp = clock();
// freopen("233.in", "r", stdin);
// freopen("233.out", "w", stdout);
ll a, b, k, p;
getphi();
int t;
while (~scanf("%d", &t))
while (t--)
scanf("%lld %lld %lld", &a, &k, &p);
b = a;
if(k == 0)
printf("%lld\\n", 1 % p);
continue;
printf("%lld\\n", f(a, b, k - 1, p).ans);
//cout<<(clock()-pp)/CLOCKS_PER_SEC;
return 0;
cf-906 D. Power Tower
突然很多人交这道两年前的题啊,,hhhhh
这题也是降幂,他是求的一个指数序列的一个区间的幂的值,,,套路一样,,就是这个模数很大,,不能每次都算他的 phi ,,不然会超时,,所以要记忆化一下 unordered_map
一下,,或者 预处理一下模数的所有phi 因为对一个数一直求 phi 下去,,其实个数不多,,,
#include <bits/stdc++.h>
#define aaa cout<<233<<endl;
#define endl '\\n'
#define pb push_back
using namespace std;
typedef long long ll;
typedef unsigned long long ull;
typedef long double ld;
// mt19937 rnd(time(0));
const int inf = 0x3f3f3f3f;//1061109567 > 1e9
const ll linf = 0x3f3f3f3f3f3f3f3f;
const double eps = 1e-7;
const double pi = 3.14159265358979;
const int maxn = 1e5 + 5;
const int maxm = 1e3 + 5;
const int mod = 1e9 + 7;
ll a[maxn];
inline ll modulo(ll x, ll mod)return x < mod ? x : x % mod + mod;
inline ll pow_(ll a, ll b, ll p)
ll ret = 1;
while(b)
if(b & 1)ret = modulo(ret * a, p);
a = modulo(a * a, p);
b >>= 1;
return ret;
unordered_map<ll, ll> phi_;
inline ll phi(ll x)
if(phi_[x])return phi_[x];
ll ans = x;
ll t = x;
for(ll i = 2; i * i <= x; ++i)
if(x % i == 0)
ans = ans / i * (i - 1);
while(x % i == 0)x /= i;
if(x > 1)ans = ans / x * (x - 1);
phi_[t] = ans;
return ans;
//这里根据题意来更改,k表示共有k个指数
ll f(ll a, ll b, ll k, ll p)
if(p == 1)return 1;
if(k == 0)return 1;
return pow_(a, f(a, a, k - 1, phi(p)), p);
ll f(ll l, ll r, ll p)
if(p == 1)return 1;
if(l == r + 1)return 1;
return pow_(a[l], f(l + 1, r, phi(p)), p);
int main()
// double pp = clock();
// freopen("233.in", "r", stdin);
// freopen("233.out", "w", stdout);
ios_base::sync_with_stdio(0);
cin.tie(0);cout.tie(0);
ll n, m;
cin >> n >> m;
for(int i = 1; i <= n; ++i)cin >> a[i];
int q; cin >> q;
while(q--)
ll l, r; cin >> l >> r;
cout << f(l, r, m) % m << endl;
// cout << endl << (clock() - pp) / CLOCKS_PER_SEC << endl;
return 0;
cf-gym-101550 E Exponial
这题是求一个 \\(n^n-1^n-2^n-3^...^1 mod \\ p\\) ,,,用上面的板子改一改就可以了,,,
#include <bits/stdc++.h>
#define aaa cout<<233<<endl;
#define endl '\\n'
#define pb push_back
using namespace std;
typedef long long ll;
typedef unsigned long long ull;
typedef long double ld;
// mt19937 rnd(time(0));
const int inf = 0x3f3f3f3f;//1061109567 > 1e9
const ll linf = 0x3f3f3f3f3f3f3f3f;
const double eps = 1e-7;
const double pi = 3.14159265358979;
const int maxn = 1e5 + 5;
const int maxm = 1e3 + 5;
const int mod = 1e9 + 7;
inline ll modulo(ll x, ll mod)return x < mod ? x : x % mod + mod;
inline ll pow_(ll a, ll b, ll p)
ll ret = 1;
while(b)
if(b & 1)ret = modulo(ret * a, p);
a = modulo(a * a, p);
b >>= 1;
return ret;
unordered_map<ll, ll> phi_;
inline ll phi(ll x)
if(phi_[x])return phi_[x];
ll ans = x;
ll t = x;
for(ll i = 2; i * i <= x; ++i)
if(x % i == 0)
ans = ans / i * (i - 1);
while(x % i == 0)x /= i;
if(x > 1)ans = ans / x * (x - 1);
phi_[t] = ans;
return ans;
// ll f(ll l, ll r, ll p)
//
// if(p == 1)return 1;
// if(l == r + 1)return 1;
// return pow_(a[l], f(l + 1, r, phi(p)), p);
//
ll f(ll a, ll p)
if(p == 1)return 1;
if(a == 1)return 1;
return pow_(a, f(a - 1, phi(p)), p);
int main()
// double pp = clock();
// freopen("233.in", "r", stdin);
// freopen("233.out", "w", stdout);
ios_base::sync_with_stdio(0);
cin.tie(0);cout.tie(0);
ll n, m;
while(cin >> n >> m)cout << f(n, m) % m << endl;
// cout << endl << (clock() - pp) / CLOCKS_PER_SEC << endl;
return 0;
貌似够了,,,数论是最不想碰的东西,,emmmm,,,但又时不得不稍稍掌握的东西,,,,
(end....)
以上是关于ACM-数论-广义欧拉降幂的主要内容,如果未能解决你的问题,请参考以下文章