【学习笔记】莫比乌斯反演

已于 2022-02-12 10:00:38 修改
阅读量741 收藏 1
点赞数
分类专栏： 学习笔记 文章标签： 概率论 线性代数 机器学习
于 2022-02-10 14:19:42 首次发布
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/C2022lihan/article/details/122857814
版权
本文深入探讨了引理在数论和算法中的应用，结合容斥原理解决了一系列数学问题。通过举例展示了如何利用引理和容斥定理计算数对的GCD（最大公约数）的幂次、数表的性质以及约数个数和等。文章通过严谨的数学推导，将复杂问题简化为易于理解的形式，展现了数学在编程问题解决中的强大能力。
摘要由CSDN通过智能技术生成

其实引理用得更多，人麻了

引理：

f ( n ) = ∑ d ∣ n μ ( d ) ⇒ f ( n ) = { 1 , n = 1 0 , e l s e \begin{aligned} f (n) &= \sum_{d \mid n} \mu (d) \\ \Rightarrow f (n) &= \begin{cases} 1, n =1 \\ 0, else \end{cases} \end{aligned} f(n)⇒f(n)​=d∣n∑​μ(d)={1,n=10,else​​

容斥定理易证

原式：

f ( n ) = ∑ d ∣ n F ( d ) ⇒ F ( n ) = ∑ d ∣ n μ ( n d ) f ( d ) \begin{aligned} f (n) &= \sum_{d \mid n} F (d) \\ \Rightarrow F (n) &= \sum_{d \mid n} \mu (\frac{n}{d}) f (d) \end{aligned} f(n)⇒F(n)​=d∣n∑​F(d)=d∣n∑​μ(dn​)f(d)​

证明：

F ( n ) = ∑ d ∣ n μ ( n d ) ∑ p ∣ d F ( p ) 令 t = d p = ∑ p ∑ t ∣ n p μ ( n t p ) F ( p ) = ∑ p F ( p ) ∑ t ∣ n p μ ( n p t ) = F ( n ) \begin{aligned} F (n) &= \sum_{d \mid n} \mu (\frac{n}{d}) \sum_{p \mid d} F (p) \\ 令 t = &\frac{d}{p} \\ &= \sum_{p} \sum_{t \mid \frac{n}{p}} \mu (\frac{n}{tp}) F (p) \\ &= \sum_{p} F (p) \sum_{t \mid \frac{n}{p}} \mu (\frac{\frac{n}{p}}{t}) \\ &= F (n) \end{aligned} F(n)令t=​=d∣n∑​μ(dn​)p∣d∑​F(p)pd​=p∑​t∣pn​∑​μ(tpn​)F(p)=p∑​F(p)t∣pn​∑​μ(tpn​​)=F(n)​

推论：

f ( n ) = ∑ n ∣ d F ( d ) ⇒ F ( n ) = ∑ n ∣ d μ ( d n ) f ( d ) \begin{aligned} f (n) &= \sum_{n \mid d} F (d) \\ \Rightarrow F (n) &= \sum_{n \mid d} \mu (\frac{d}{n}) f (d) \end{aligned} f(n)⇒F(n)​=n∣d∑​F(d)=n∣d∑​μ(nd​)f(d)​

证明：

类似于原式。

例题

一、 Y Y YY YY 的 G C D GCD GCD

∑ p r i m e ∑ x ∈ [ 1 , n ] ∑ y ∈ [ 1 , m ] [ ( x , y ) = = p r i m e ] \begin{aligned} & \sum_{prime} \sum_{x \in [1, n]}\sum_{y \in [1, m]} [(x, y) == prime] \end{aligned} ​prime∑​x∈[1,n]∑​y∈[1,m]∑​[(x,y)==prime]​

令 f ( i ) f(i) f(i) 表示 g c d = i gcd = i gcd=i 的数的对数， F ( i ) F(i) F(i) 表示 i ∣ g c d i \mid gcd i∣gcd 的数的对数

F ( i ) = ∑ i ∣ j f ( j ) f ( i ) = ∑ i ∣ j μ ( j i ) F ( j ) a n s w e r = ∑ p r i m e f ( p r i m e ) = ∑ p r i m e ∑ p r i m e ∣ j μ ( j p r i m e ) ⌊ n j ⌋ ⌊ m j ⌋ = ∑ j ⌊ n j ⌋ ⌊ m j ⌋ ∑ p r i m e ∣ j μ ( j p r i m e ) 令 h ( x ) = ∑ p r i m e ∣ x μ ( x p r i m e ) = ∑ j ⌊ n j ⌋ ⌊ m j ⌋ h ( j ) \begin{aligned} F (i) &= \sum_{i \mid j} f (j) \\ f (i) &= \sum_{i \mid j} \mu (\frac{j}{i}) F (j) \\ answer &= \sum_{prime} f (prime) \\ &= \sum_{prime} \sum_{prime \mid j} \mu(\frac{j}{prime}) \lfloor \frac{n}{j} \rfloor \lfloor \frac{m}{j} \rfloor \\ &= \sum_{j} \lfloor \frac{n}{j} \rfloor \lfloor \frac{m}{j} \rfloor \sum_{prime \mid j} \mu (\frac{j}{prime}) \\ 令 h (x) = &\sum_{prime \mid x} \mu (\frac{x}{prime}) \\ &= \sum_{j} \lfloor \frac{n}{j} \rfloor \lfloor \frac{m}{j} \rfloor h (j) \end{aligned} F(i)f(i)answer令h(x)=​=i∣j∑​f(j)=i∣j∑​μ(ij​)F(j)=prime∑​f(prime)=prime∑​prime∣j∑​μ(primej​)⌊jn​⌋⌊jm​⌋=j∑​⌊jn​⌋⌊jm​⌋prime∣j∑​μ(primej​)prime∣x∑​μ(primex​)=j∑​⌊jn​⌋⌊jm​⌋h(j)​

把 h h h 打表出来，整数分块，结束了。

//author : LH ——Who just can eat S??t
//worship WJC ——Who can f??k tourist up and down and loves 周歆恬
//worship YJX ——Who can f??k WJC up and down
#include <cmath>
#include <queue>
#include <vector>
#include <cstdio>
#include <cstdlib> 
#include <cstring>
#include <iostream>
#include <algorithm>
using namespace std;
#define fi first
#define se second
#define db double
#define LL long long
#define ULL unsigned long long
#define PII pair <int, int>
#define MP(x,y) make_pair (x, y)
#define rep(i,j,k) for (int i = (j); i <= (k); ++i)
#define per(i,j,k) for (int i = (j); i >= (k); --i)

template <typename T> T Max (T x, T y) { return x > y ? x : y; }
template <typename T> T Min (T x, T y) { return x < y ? x : y; }
template <typename T> T Abs (T x) { return x > 0 ? x : -x; }
template <typename T>
void read (T &x) {
    x = 0; T f = 1;
    char ch = getchar ();
    while (ch < '0' || ch > '9') {
        if (ch == '-') f = -1;
        ch = getchar ();
    }
    while (ch >= '0' && ch <= '9') {
        x = (x << 3) + (x << 1) + ch - '0';
        ch = getchar ();
    }
    x *= f;
}
template <typename T, typename... Args>
void read (T &x, Args&... args) {
    read (x); read (args...);
}
char For_Print[25];
template <typename T>
void write (T x) {
    if (x == 0) { putchar ('0'); return; }
    if (x < 0) { putchar ('-'); x = -x; }
    int poi = 0;
    while (x) {
        For_Print[++poi] = x % 10 + '0';
        x /= 10;
    }
    while (poi) putchar (For_Print[poi--]);
}
template <typename T>
void print (T x, char ch) {
    write (x); putchar (ch);
}

const LL Mod = 1e9 + 7;

LL square (LL x) { return (x * x) % Mod; }
void DEL (LL &x, LL y) { ((x -= y) < 0) && (x += Mod); }
void ADD (LL &x, LL y) { ((x += y) >= Mod) && (x -= Mod); }

const int Maxn = 1e7;

int t, n, m;

int cnt, primes[Maxn + 5];
int mu[Maxn + 5], px[Maxn + 5], h[Maxn + 5];
LL pre[Maxn + 5];
bool vis[Maxn + 5];
void Euler () {
    mu[1] = 1;
    rep (i, 2, Maxn) {
        if (!vis[i]) {
            primes[++cnt] = i;
            mu[i] = -1;
            px[i] = i;
        }
        rep (j, 1, cnt) {
            if (primes[j] > Maxn / i) break;
            vis[i * primes[j]] = 1;
            if (i % primes[j] == 0) {
                mu[i * primes[j]] = 0;
                px[i * primes[j]] = primes[j];
                break;
            }
            mu[i * primes[j]] = -mu[i];
            px[i * primes[j]] = primes[j];
        }
    }
    rep (i, 1, Maxn) {
        int tmp = i;
        while (tmp != 1) {
            int prm = px[tmp];
            h[i] += mu[i / prm];
            while (tmp % prm == 0)
                tmp /= prm;
        }
    }
    rep (i, 1, Maxn) {
        pre[i] = pre[i - 1] + h[i];
    }
}

int main () {
	// freopen ("D:\\lihan\\1.in", "r", stdin);
	// freopen ("D:\\lihan\\1.out", "w", stdout);

    Euler ();

    read (t);
    while (t--) {
        read (n, m);
        if (n > m) swap (n, m);
        int l = 1, r = n;
        LL res = 0;
        while (l <= n) {
            r = Min (n / (n / l), m / (m / l));
            res += (LL)(n / l) * (m / l) * (pre[r] - pre[l - 1]);
            l = r + 1;
        }
        cout << res << endl;
    }
    return 0;
}

二、于神之怒

∑ i = 1 n ∑ j = 1 m g c d ( i , j ) k = ∑ d ∑ i ∑ j d k ∑ p ∣ g c d ( i , j ) d μ ( p ) = ∑ p μ ( p ) ∑ d d k ⌊ n d p ⌋ ⌊ m d p ⌋ = ∑ d d k ∑ p μ ( p ) ⌊ n d p ⌋ ⌊ m d p ⌋ 令 t = d p = ∑ t ⌊ n t ⌋ ⌊ m t ⌋ ∑ d ∣ t d k μ ( t d ) 令 h ( x ) = ∑ d ∣ x d k μ ( x d ) = ∑ t ⌊ n t ⌋ ⌊ m t ⌋ h ( t ) \begin{aligned} &\sum_{i = 1}^n\sum_{j = 1}^m gcd (i, j)^k \\ = &\sum_d\sum_i\sum_j d^k \sum_{p \mid \frac{gcd (i, j)}{d}} \mu (p) \\ = &\sum_p \mu (p) \sum_d d^k \lfloor \frac{n}{dp} \rfloor \lfloor \frac{m}{dp} \rfloor \\ = &\sum_dd^k\sum_p \mu (p) \lfloor \frac{n}{dp} \rfloor \lfloor \frac{m}{dp} \rfloor \\ 令 &t = dp \\ = &\sum_t \lfloor \frac{n}{t} \rfloor \lfloor \frac{m}{t} \rfloor \sum_{d \mid t} d^k \mu (\frac{t}{d}) \\ 令 &h (x) = \sum_{d \mid x} d^k \mu (\frac{x}{d}) \\ = &\sum_t \lfloor \frac{n}{t} \rfloor \lfloor \frac{m}{t} \rfloor h (t) \end{aligned} ===令=令=​i=1∑n​j=1∑m​gcd(i,j)kd∑​i∑​j∑​dkp∣dgcd(i,j)​∑​μ(p)p∑​μ(p)d∑​dk⌊dpn​⌋⌊dpm​⌋d∑​dkp∑​μ(p)⌊dpn​⌋⌊dpm​⌋t=dpt∑​⌊tn​⌋⌊tm​⌋d∣t∑​dkμ(dt​)h(x)=d∣x∑​dkμ(dx​)t∑​⌊tn​⌋⌊tm​⌋h(t)​

//author : LH ——Who just can eat S??t
//worship WJC ——Who can f??k tourist up and down and loves 周歆恬
//worship YJX ——Who can f??k WJC up and down
#include <cmath>
#include <queue>
#include <vector>
#include <cstdio>
#include <cstdlib> 
#include <cstring>
#include <iostream>
#include <algorithm>
using namespace std;
#define fi first
#define se second
#define db double
#define LL long long
#define ULL unsigned long long
#define PII pair <int, int>
#define MP(x,y) make_pair (x, y)
#define rep(i,j,k) for (int i = (j); i <= (k); ++i)
#define per(i,j,k) for (int i = (j); i >= (k); --i)

template <typename T> T Max (T x, T y) { return x > y ? x : y; }
template <typename T> T Min (T x, T y) { return x < y ? x : y; }
template <typename T> T Abs (T x) { return x > 0 ? x : -x; }
template <typename T>
void read (T &x) {
    x = 0; T f = 1;
    char ch = getchar ();
    while (ch < '0' || ch > '9') {
        if (ch == '-') f = -1;
        ch = getchar ();
    }
    while (ch >= '0' && ch <= '9') {
        x = (x << 3) + (x << 1) + ch - '0';
        ch = getchar ();
    }
    x *= f;
}
template <typename T, typename... Args>
void read (T &x, Args&... args) {
    read (x); read (args...);
}
char For_Print[25];
template <typename T>
void write (T x) {
    if (x == 0) { putchar ('0'); return; }
    if (x < 0) { putchar ('-'); x = -x; }
    int poi = 0;
    while (x) {
        For_Print[++poi] = x % 10 + '0';
        x /= 10;
    }
    while (poi) putchar (For_Print[poi--]);
}
template <typename T>
void print (T x, char ch) {
    write (x); putchar (ch);
}

const LL Mod = 1e9 + 7;

LL square (LL x) { return (x * x) % Mod; }
void DEL (LL &x, LL y) { ((x -= y) < 0) && (x += Mod); }
void ADD (LL &x, LL y) { ((x += y) >= Mod) && (x -= Mod); }

const int Maxn = 5 * 1e6;

int t, k, n, m;

int cnt, primes[Maxn + 5];
LL h[Maxn + 5], val[Maxn + 5], pre[Maxn + 5];
bool vis[Maxn + 5];
LL quick_pow (LL x, LL y) {
    LL res = 1;
    while (y) {
        if (y & 1) res = (res * x) % Mod;
        x = (x * x) % Mod; y >>= 1;
    }
    return res;
}
LL md (LL x, LL M) {
	return (x % M + M) % M;
}
void Euler () {
    pre[1] = h[1] = 1;
    rep (i, 2, Maxn) {
        if (!vis[i]) {
            primes[++cnt] = i;
            val[i] = quick_pow (i, k);
            h[i] = md (val[i] - 1, Mod);
        }
        rep (j, 1, cnt) {
            if (primes[j] > Maxn / i) break;
            vis[primes[j] * i] = 1;
            if (i % primes[j] == 0) {
                h[i * primes[j]] = h[i] * val[primes[j]] % Mod;
                break;
            }
            h[i * primes[j]] = h[i] * (val[primes[j]] - 1) % Mod;
        }
    }
    rep (i, 1, Maxn) pre[i] = (pre[i - 1] + h[i]) % Mod;
}

int main () {
//	freopen ("D:\\lihan\\1.in", "r", stdin);
//	freopen ("D:\\lihan\\1.out", "w", stdout);

    read (t, k); Euler ();
    while (t--) {
        read (n, m);
        if (n > m) swap (n, m);

        int l = 1, r;
        LL res = 0;
        while (l <= n) {
            r = Min (n / (n / l), m / (m / l));
            ADD (res, (LL)(n / l) * (m / l) % Mod * md (pre[r] - pre[l - 1], Mod) % Mod);
            l = r + 1;
        }
        print (res, '\n');
    }
    return 0;
}

三、Crash的数字表格

∑ i = 1 n ∑ j = 1 m i ⋅ j g c d ( i , j ) = ∑ d ∑ i ∑ j i ⋅ j d [ g c d ( i , j ) = = d ] = ∑ d ∑ i ∑ j i ⋅ j d ∑ p ∣ g c d ( i d , j d ) μ ( p ) = ∑ p μ ( p ) ∑ d ∑ d p ∣ i ∑ d p ∣ j i ⋅ j d = ∑ p μ ( p ) ∑ d 1 d ∑ d p ∣ i i ∑ d p ∣ j j 令 l m = ⌊ m d p ⌋ = ∑ p μ ( p ) ∑ d 1 d ( 1 + l m ) ∗ l m 2 ⋅ d p ∑ d p ∣ i i 令 l n = ⌊ n d p ⌋ = ∑ p μ ( p ) ∑ d ( 1 + l m ) ∗ l m 2 d ⋅ d p ⋅ ( 1 + l n ) ∗ l n 2 ⋅ d p = ∑ p μ ( p ) ∑ d ( 1 + l m ) ∗ l m ⋅ ( 1 + l n ) ∗ l n ⋅ p 2 ⋅ d 4 = ∑ p μ ( p ) ∑ d ( 1 + l m ) ∗ l m ⋅ ( 1 + l n ) ∗ l n ⋅ p 2 ⋅ d 4 令 t = d p = ∑ p μ ( p ) ⋅ p 2 ∑ p ∣ t t ⋅ ( 1 + ⌊ m t ⌋ ) ∗ ⌊ m t ⌋ ⋅ ( 1 + ⌊ n t ⌋ ) ∗ ⌊ n t ⌋ 4 p = ∑ p μ ( p ) ⋅ p ∑ p ∣ t t ⋅ ( 1 + ⌊ m t ⌋ ) ∗ ⌊ m t ⌋ ⋅ ( 1 + ⌊ n t ⌋ ) ∗ ⌊ n t ⌋ 4 = ∑ t t ⋅ ( 1 + ⌊ m t ⌋ ) ∗ ⌊ m t ⌋ ⋅ ( 1 + ⌊ n t ⌋ ) ∗ ⌊ n t ⌋ 4 ∑ p ∣ t μ ( p ) ⋅ p 令 h ( t ) = t ⋅ ∑ p ∣ t μ ( p ) ⋅ p = ∑ t ( 1 + ⌊ m t ⌋ ) ∗ ⌊ m t ⌋ ⋅ ( 1 + ⌊ n t ⌋ ) ∗ ⌊ n t ⌋ 4 h ( t ) \begin{aligned} &\sum_{i = 1}^n\sum_{j = 1} ^ {m} \frac{i \cdot j}{gcd (i, j)} \\ = &\sum_{d} \sum_{i} \sum_{j} \frac{i \cdot j}{d} [gcd (i, j) == d] \\ = &\sum_{d} \sum_{i} \sum_{j} \frac{i \cdot j}{d} \sum_{p \mid gcd (\frac{i}{d}, \frac{j}{d})} \mu (p) \\ = &\sum_{p} \mu (p) \sum_{d} \sum_{dp \mid i} \sum_{dp \mid j} \frac{i \cdot j}{d} \\ = &\sum_{p} \mu (p) \sum_{d} \frac{1}{d} \sum_{dp \mid i} i \sum_{dp \mid j} j \\ 令 &lm = \lfloor \frac{m}{dp} \rfloor \\ = &\sum_{p} \mu (p) \sum_{d} \frac{1}{d} \frac {(1 + lm) * lm}{2} \cdot dp \sum_{dp \mid i} i \\ 令 &ln = \lfloor \frac{n}{dp} \rfloor \\ = &\sum_{p} \mu (p) \sum_{d} \frac {(1 + lm) * lm}{2d} \cdot dp \cdot \frac{(1+ln) * ln}{2} \cdot dp \\ = &\sum_{p} \mu (p) \sum_{d} \frac {(1 + lm) * lm \cdot (1+ ln) * ln \cdot p ^ 2 \cdot d}{4} \\ = &\sum_{p} \mu (p) \sum_{d} \frac {(1 + lm) * lm \cdot (1+ ln) * ln \cdot p ^ 2 \cdot d}{4} \\ 令 &t = dp \\ = &\sum_{p} \mu (p) \cdot p ^ 2 \frac{\sum_{p \mid t} t \cdot (1 + \lfloor \frac{m}{t} \rfloor) * \lfloor \frac{m}{t} \rfloor \cdot (1 + \lfloor \frac{n}{t} \rfloor) * \lfloor \frac{n}{t} \rfloor}{4p} \\ = &\sum_{p} \mu (p) \cdot p \frac{\sum_{p \mid t} t \cdot (1 + \lfloor \frac{m}{t} \rfloor) * \lfloor \frac{m}{t} \rfloor \cdot (1 + \lfloor \frac{n}{t} \rfloor) * \lfloor \frac{n}{t} \rfloor}{4} \\ = &\sum_{t}\frac{t \cdot (1 + \lfloor \frac{m}{t} \rfloor) * \lfloor \frac{m}{t} \rfloor \cdot (1 + \lfloor \frac{n}{t} \rfloor) * \lfloor \frac{n}{t} \rfloor}{4} \sum_{p \mid t} \mu (p) \cdot p \\ 令 &h (t) = t \cdot \sum_{p \mid t} \mu (p) \cdot p \\ = &\sum_{t}\frac{(1 + \lfloor \frac{m}{t} \rfloor) * \lfloor \frac{m}{t} \rfloor \cdot (1 + \lfloor \frac{n}{t} \rfloor) * \lfloor \frac{n}{t} \rfloor}{4} h (t) \end{aligned} ====令=令===令===令=​i=1∑n​j=1∑m​gcd(i,j)i⋅j​d∑​i∑​j∑​di⋅j​[gcd(i,j)==d]d∑​i∑​j∑​di⋅j​p∣gcd(di​,dj​)∑​μ(p)p∑​μ(p)d∑​dp∣i∑​dp∣j∑​di⋅j​p∑​μ(p)d∑​d1​dp∣i∑​idp∣j∑​jlm=⌊dpm​⌋p∑​μ(p)d∑​d1​2(1+lm)∗lm​⋅dpdp∣i∑​iln=⌊dpn​⌋p∑​μ(p)d∑​2d(1+lm)∗lm​⋅dp⋅2(1+ln)∗ln​⋅dpp∑​μ(p)d∑​4(1+lm)∗lm⋅(1+ln)∗ln⋅p2⋅d​p∑​μ(p)d∑​4(1+lm)∗lm⋅(1+ln)∗ln⋅p2⋅d​t=dpp∑​μ(p)⋅p24p∑p∣t​t⋅(1+⌊tm​⌋)∗⌊tm​⌋⋅(1+⌊tn​⌋)∗⌊tn​⌋​p∑​μ(p)⋅p4∑p∣t​t⋅(1+⌊tm​⌋)∗⌊tm​⌋⋅(1+⌊tn​⌋)∗⌊tn​⌋​t∑​4t⋅(1+⌊tm​⌋)∗⌊tm​⌋⋅(1+⌊tn​⌋)∗⌊tn​⌋​p∣t∑​μ(p)⋅ph(t)=t⋅p∣t∑​μ(p)⋅pt∑​4(1+⌊tm​⌋)∗⌊tm​⌋⋅(1+⌊tn​⌋)∗⌊tn​⌋​h(t)​

//author : LH ——Who just can eat S??t
//worship WJC ——Who can f??k tourist up and down and loves 周歆恬
//worship YJX ——Who can f??k WJC up and down
#include <cmath>
#include <queue>
#include <vector>
#include <cstdio>
#include <cstdlib> 
#include <cstring>
#include <iostream>
#include <algorithm>
using namespace std;
#define fi first
#define se second
#define db double
#define LL long long
#define ULL unsigned long long
#define PII pair <int, int>
#define MP(x,y) make_pair (x, y)
#define rep(i,j,k) for (int i = (j); i <= (k); ++i)
#define per(i,j,k) for (int i = (j); i >= (k); --i)

template <typename T> T Max (T x, T y) { return x > y ? x : y; }
template <typename T> T Min (T x, T y) { return x < y ? x : y; }
template <typename T> T Abs (T x) { return x > 0 ? x : -x; }
template <typename T>
void read (T &x) {
    x = 0; T f = 1;
    char ch = getchar ();
    while (ch < '0' || ch > '9') {
        if (ch == '-') f = -1;
        ch = getchar ();
    }
    while (ch >= '0' && ch <= '9') {
        x = (x << 3) + (x << 1) + ch - '0';
        ch = getchar ();
    }
    x *= f;
}
template <typename T, typename... Args>
void read (T &x, Args&... args) {
    read (x); read (args...);
}
char For_Print[25];
template <typename T>
void write (T x) {
    if (x == 0) { putchar ('0'); return; }
    if (x < 0) { putchar ('-'); x = -x; }
    int poi = 0;
    while (x) {
        For_Print[++poi] = x % 10 + '0';
        x /= 10;
    }
    while (poi) putchar (For_Print[poi--]);
}
template <typename T>
void print (T x, char ch) {
    write (x); putchar (ch);
}

const LL Mod = 20101009;
const LL inv_2 = 10050505;

LL square (LL x) { return (x * x) % Mod; }
void DEL (LL &x, LL y) { ((x -= y) < 0) && (x += Mod); }
void ADD (LL &x, LL y) { ((x += y) >= Mod) && (x -= Mod); }

const int Maxq = 1e6;
const int Maxn = 1e7;

int cnt, primes[Maxq + 5];
LL h[Maxn + 5], pre[Maxn + 5];
bool vis[Maxn + 5];
LL md (LL x, LL M) {
    return (x % M + M) % M;
}
void Euler () {
    h[1] = 1;
    rep (i, 2, Maxn) {
        if (!vis[i]) {
            primes[++cnt] = i;
            h[i] = 1 - i;
        }
        rep (j, 1, cnt) {
            if (primes[j] > Maxn / i) break;
            vis[primes[j] * i] = 1;
            if (i % primes[j] == 0) {
                h[i * primes[j]] = h[i];
                break;
            }
            h[i * primes[j]] = h[i] * (1 - primes[j]) % Mod;
        }
    }

    rep (i, 1, Maxn) pre[i] = md (pre[i - 1] + h[i] * i % Mod, Mod);
}

LL n, m;

int main () {
	// freopen ("D:\\lihan\\1.in", "r", stdin);
	// freopen ("D:\\lihan\\1.out", "w", stdout);

    Euler ();

    read (n, m);
    // cerr << n << " " << m << endl;
    if (n > m) swap (n, m);
    int l = 1, r;
    LL res = 0;
    while (l <= n) {
        LL ln = n / l, lm = m / l;
        r = Min (n / ln, m / lm);
        ADD (res, (1 + ln) * ln % Mod * (1 + lm) % Mod * lm % Mod * md (pre[r] - pre[l - 1], Mod) % Mod * inv_2 % Mod * inv_2 % Mod);
        l = r + 1;
    }
    cout << res;
    return 0;
}

四、 「 S D O I 2014 」 「SDOI2014」 「SDOI2014」 数表

a n s w e r = ∑ i = 1 n ∑ j = 1 m [ σ ( g c d ( i , j ) ) ≤ a ] σ ( g c d ( i , j ) ) 令 g = g c d ( i , j ) = ∑ g ∑ i ∑ j [ σ ( g ) ≤ a ] σ ( g ) = ∑ g [ σ ( g ) ≤ a ] ∑ i ∑ j ∑ d ∣ g c d ( i g , j g ) μ ( d ) σ ( g ) = ∑ g [ σ ( g ) ≤ a ] ∑ d μ ( d ) ∑ d g ∣ i ∑ d g ∣ j σ ( g ) = ∑ g [ σ ( g ) ≤ a ] ∑ d μ ( d ) ⌊ n d g ⌋ ⌊ m d g ⌋ σ ( g ) = ∑ g ∑ d [ σ ( g ) ≤ a ] ⋅ μ ( d ) ⌊ n d g ⌋ ⌊ m d g ⌋ σ ( g ) 令 t = d g = ∑ g ∣ t ∑ t [ σ ( g ) ≤ a ] ⋅ μ ( t g ) ⌊ n t ⌋ ⌊ m t ⌋ σ ( g ) = ∑ g ∣ t ∑ t [ σ ( g ) ≤ a ] ⋅ μ ( t g ) ⌊ n t ⌋ ⌊ m t ⌋ σ ( g ) = ∑ t ⌊ n t ⌋ ⌊ m t ⌋ ∑ g ∣ t [ σ ( g ) ≤ a ] ⋅ μ ( t g ) σ ( g ) 令 h ( t ) = ∑ g ∣ t [ σ ( g ) ≤ a ] ⋅ μ ( t g ) σ ( g ) = ∑ t ⌊ n t ⌋ ⌊ m t ⌋ h ( t ) \begin{aligned} answer &= \sum_{i = 1}^{n} \sum_{j = 1}^{m} [\sigma (gcd (i, j)) \leq a] \sigma (gcd (i, j)) \\ 令 g = gc&d (i, j) \\ &= \sum_{g} \sum_i\sum_j [\sigma (g) \leq a]\sigma (g) \\ &= \sum_{g} [\sigma (g) \leq a] \sum_i \sum_j \sum_{d \mid gcd (\frac{i}{g}, \frac{j}{g})} \mu (d) \sigma (g) \\ &= \sum_{g} [\sigma (g) \leq a] \sum_{d} \mu (d) \sum_{dg \mid i} \sum_{dg \mid j} \sigma (g) \\ &= \sum_{g} [\sigma (g) \leq a] \sum_{d} \mu (d) \lfloor \frac{n}{dg} \rfloor \lfloor \frac{m}{dg} \rfloor \sigma (g) \\ &= \sum_{g} \sum_{d} [\sigma (g) \leq a] \cdot \mu (d) \lfloor \frac{n}{dg} \rfloor \lfloor \frac{m}{dg} \rfloor \sigma (g) \\ 令 t = dg& \\ &= \sum_{g \mid t} \sum_{t} [\sigma (g) \leq a] \cdot \mu (\frac{t}{g}) \lfloor \frac{n}{t} \rfloor \lfloor \frac{m}{t} \rfloor \sigma (g) \\ &= \sum_{g \mid t}\sum_t [\sigma (g) \leq a] \cdot \mu (\frac{t}{g}) \lfloor \frac{n}{t} \rfloor \lfloor \frac{m}{t} \rfloor \sigma (g) \\ &= \sum_t \lfloor \frac{n}{t} \rfloor \lfloor \frac{m}{t} \rfloor \sum_{g \mid t} [\sigma (g) \leq a] \cdot \mu (\frac{t}{g}) \sigma (g) \\ 令 h(t) = &\sum_{g \mid t}[\sigma (g) \leq a] \cdot \mu (\frac{t}{g}) \sigma (g) \\ &= \sum_t \lfloor \frac{n}{t} \rfloor \lfloor \frac{m}{t} \rfloor h (t) \end{aligned} answer令g=gc令t=dg令h(t)=​=i=1∑n​j=1∑m​[σ(gcd(i,j))≤a]σ(gcd(i,j))d(i,j)=g∑​i∑​j∑​[σ(g)≤a]σ(g)=g∑​[σ(g)≤a]i∑​j∑​d∣gcd(gi​,gj​)∑​μ(d)σ(g)=g∑​[σ(g)≤a]d∑​μ(d)dg∣i∑​dg∣j∑​σ(g)=g∑​[σ(g)≤a]d∑​μ(d)⌊dgn​⌋⌊dgm​⌋σ(g)=g∑​d∑​[σ(g)≤a]⋅μ(d)⌊dgn​⌋⌊dgm​⌋σ(g)=g∣t∑​t∑​[σ(g)≤a]⋅μ(gt​)⌊tn​⌋⌊tm​⌋σ(g)=g∣t∑​t∑​[σ(g)≤a]⋅μ(gt​)⌊tn​⌋⌊tm​⌋σ(g)=t∑​⌊tn​⌋⌊tm​⌋g∣t∑​[σ(g)≤a]⋅μ(gt​)σ(g)g∣t∑​[σ(g)≤a]⋅μ(gt​)σ(g)=t∑​⌊tn​⌋⌊tm​⌋h(t)​

我们离线求解，按照 a a a 排序，求解完 i − 1 i - 1 i−1 后，会加入一些 g g g，满足 a i − 1 ≤ σ ( g ) ≤ a i a_{i - 1} \leq \sigma (g) \leq a_i ai−1​≤σ(g)≤ai​，我们把这些 g g g 用类似埃筛的方法加入 h ( k ⋅ g ) ( k ∈ Z , k ⋅ g ≤ m i n ( n , m ) ) h (k\cdot g) (k \in \mathrm{Z}, k \cdot g \leq min (n, m)) h(k⋅g)(k∈Z,k⋅g≤min(n,m))，然后用树状数组动态维护前缀和就行了。

好妙， D J DJ DJ 是神仙。

//author : LH ——Who just can eat S??t
//worship WJC ——Who can f??k tourist up and down and loves 周歆恬
//worship YJX ——Who can f??k WJC up and down
#include <cmath>
#include <queue>
#include <vector>
#include <cstdio>
#include <cstdlib> 
#include <cstring>
#include <iostream>
#include <algorithm>
using namespace std;
#define fi first
#define se second
#define db double
#define LL long long
#define ULL unsigned long long
#define PII pair <int, int>
#define MP(x,y) make_pair (x, y)
#define rep(i,j,k) for (int i = (j); i <= (k); ++i)
#define per(i,j,k) for (int i = (j); i >= (k); --i)

template <typename T> T Max (T x, T y) { return x > y ? x : y; }
template <typename T> T Min (T x, T y) { return x < y ? x : y; }
template <typename T> T Abs (T x) { return x > 0 ? x : -x; }
template <typename T>
void read (T &x) {
    x = 0; T f = 1;
    char ch = getchar ();
    while (ch < '0' || ch > '9') {
        if (ch == '-') f = -1;
        ch = getchar ();
    }
    while (ch >= '0' && ch <= '9') {
        x = (x << 3) + (x << 1) + ch - '0';
        ch = getchar ();
    }
    x *= f;
}
template <typename T, typename... Args>
void read (T &x, Args&... args) {
    read (x); read (args...);
}
char For_Print[25];
template <typename T>
void write (T x) {
    if (x == 0) { putchar ('0'); return; }
    if (x < 0) { putchar ('-'); x = -x; }
    int poi = 0;
    while (x) {
        For_Print[++poi] = x % 10 + '0';
        x /= 10;
    }
    while (poi) putchar (For_Print[poi--]);
}
template <typename T>
void print (T x, char ch) {
    write (x); putchar (ch);
}

const LL Mod = 1ll << 31;

LL square (LL x) { return (x * x) % Mod; }
void DEL (LL &x, LL y) { ((x -= y) < 0) && (x += Mod); }
void ADD (LL &x, LL y) { ((x += y) >= Mod) && (x -= Mod); }

const int Maxn = 1e5;

int t;
LL ans[Maxn + 5];

int cnt, primes[Maxn + 5];
LL px[Maxn + 5], sigma[Maxn + 5], mu[Maxn + 5];
bool vis[Maxn + 5];
deque <PII> c;
void Euler () {
    mu[1] = 1; sigma[1] = 1;
    rep (i, 2, Maxn) {
        if (!vis[i]) {
            primes[++cnt] = i;
            sigma[i] = i + 1;
            mu[i] = -1;
            px[i] = i;
        }
        rep (j, 1, cnt) {
            if (primes[j] > Maxn / i) break;
            vis[i * primes[j]] = 1;
            if (i % primes[j] == 0) {
                mu[i * primes[j]] = 0;
                px[i * primes[j]] = px[i] * primes[j];
                sigma[i * primes[j]] = sigma[i / px[i]] * (sigma[px[i]] + px[i] * primes[j]);
                break;
            }
            mu[i * primes[j]] = -mu[i];
            px[i * primes[j]] = primes[j];
            sigma[i * primes[j]] = sigma[i] * (primes[j] + 1);
        }
    }
    rep (i, 1, Maxn)
        c.push_back (MP (sigma[i], i));
    sort (c.begin (), c.end ());
}

struct qst {
    int n, m, id; LL a;
}q[Maxn + 5];
bool cmp (qst x, qst y) {
    return x.a < y.a;
}

LL BIT[Maxn + 5];
LL md (LL x, LL M) {
    return (x % M + M) % M;
}
int lowbit (int x) { return x & -x; }
void Update (int Index, LL x) {
    for (int i = Index; i <= Maxn; i += lowbit (i))
        BIT[i] += x;
}
LL Sum (int Index) {
    LL res = 0;
    for (int i = Index; i >= 1; i -= lowbit (i))
        res += BIT[i];
    return res;
}
LL Query (int l, int r) {
    return md (Sum (r) - Sum (l - 1), Mod);
}
void Add (int x) {
    for (int i = x; i <= Maxn; i += x)
        Update (i, mu[i / x] * sigma[x]);
}

int main () {
// 	freopen ("D:\\lihan\\1.in", "r", stdin);
//	freopen ("D:\\lihan\\1.out", "w", stdout);

	Euler ();
    
    read (t);
    rep (i, 1, t)
        { read (q[i].n, q[i].m, q[i].a); q[i].id = i; }
    sort (q + 1, q + 1 + t, cmp);

    rep (i, 1, t) {
        while (c.size () && c.begin () -> fi <= q[i].a) 
			{ Add (c.begin () -> se); c.pop_front (); }

        int n = q[i].n, m = q[i].m;
        if (n > m) swap (n, m);
        int l = 1, r;
        LL res = 0;
        while (l <= n) {
            r = Min (n / (n / l), m / (m / l));
            ADD (res, (n / l) * (m / l) % Mod * Query (l, r) % Mod);
            l = r + 1;
        }
        ans[q[i].id] = res;
    }
    rep (i, 1, t) print (ans[i], '\n');
    return 0;
}

五、约数个数和

∑ i = 1 n ∑ j = 1 m d ( i j ) = ∑ i = 1 n ∑ j = 1 m ∑ p ∣ i j 1 = ∑ p = 1 n m ∑ i = 1 n ∑ j = 1 m [ p ∣ i j ] = ∑ p = 1 n m ∑ i = 1 n ∑ j = 1 m [ p g c d ( i , p ) ∣ j ] = ∑ p = 1 n m ∑ i = 1 n ⌊ m p g c d ( i , p ) ⌋ = ∑ g = 1 n ∑ p = 1 n m ∑ i = 1 n ⌊ m ⋅ g p ⌋ [ g c d ( i , p ) = = g ] = ∑ g = 1 n ∑ p = 1 n m ∑ i = 1 n ⌊ m ⋅ g p ⌋ ∑ d ∣ g c d ( i g , p g ) μ ( d ) = ∑ g = 1 n ∑ p = 1 n m g ∑ i = 1 n g ⌊ m ⋅ g p ⋅ i ⌋ ∑ d ∣ g c d ( i , p ) μ ( d ) = ∑ g = 1 n ∑ p = 1 n m g ∑ i = 1 n g ∑ d ∣ g c d ( i , p ) ⌊ m ⋅ g p ⋅ i ⌋ μ ( d ) = ∑ g = 1 n ∑ d ∑ d ∣ p n m g ∑ d ∣ i n g ⌊ m ⋅ g p ⋅ i ⌋ μ ( d ) = ∑ g = 1 n ∑ d ∑ p = 1 n m g d ∑ i = 1 n g d ⌊ m ⋅ g p ⋅ i ⋅ d 2 ⌋ μ ( d ) = ∑ g = 1 n ∑ d ∑ p = 1 n m g d ∑ i = 1 n g d ⌊ m ⋅ g p ⋅ i ⋅ d 2 ⌋ μ ( d ) \begin{aligned} &\sum_{i = 1}^{n} \sum_{j = 1}^{m} d (ij) \\ = &\sum_{i = 1}^{n} \sum_{j = 1}^{m} \sum_{p \mid ij} 1 \\ = &\sum_{p = 1}^{nm} \sum_{i = 1}^{n} \sum_{j = 1}^{m} [p \mid ij] \\ = &\sum_{p = 1}^{nm} \sum_{i = 1}^{n} \sum_{j = 1}^{m} [\frac{p}{gcd (i, p)} \mid j] \\ = &\sum_{p = 1}^{nm} \sum_{i = 1}^{n} \lfloor \frac{m}{\frac{p}{gcd (i, p)}} \rfloor \\ = &\sum_{g = 1}^{n} \sum_{p = 1}^{nm} \sum_{i = 1}^{n} \lfloor \frac{m \cdot g}{p} \rfloor [gcd (i, p) == g] \\ = &\sum_{g = 1}^{n}\sum_{p = 1}^{nm} \sum_{i = 1}^{n} \lfloor \frac{m \cdot g}{p} \rfloor \sum_{d \mid gcd (\frac{i}{g}, \frac{p}{g})} \mu(d) \\ = &\sum_{g = 1}^{n}\sum_{p = 1}^{\frac{nm}{g}} \sum_{i = 1}^{\frac{n}{g}} \lfloor \frac{m \cdot g}{p \cdot i} \rfloor \sum_{d \mid gcd (i, p)} \mu(d) \\ = &\sum_{g = 1}^{n}\sum_{p = 1}^{\frac{nm}{g}} \sum_{i = 1}^{\frac{n}{g}} \sum_{d \mid gcd (i, p)} \lfloor \frac{m \cdot g}{p \cdot i} \rfloor \mu(d) \\ = &\sum_{g = 1}^{n}\sum_{d}\sum_{d \mid p}^{\frac{nm}{g}} \sum_{d \mid i}^{\frac{n}{g}} \lfloor \frac{m \cdot g}{p \cdot i} \rfloor \mu(d) \\ = &\sum_{g = 1}^{n}\sum_{d}\sum_{p = 1}^{\frac{nm}{gd}} \sum_{i = 1}^{\frac{n}{gd}} \lfloor \frac{m \cdot g}{p \cdot i \cdot d^2} \rfloor \mu(d) \\ = &\sum_{g = 1}^{n}\sum_{d}\sum_{p = 1}^{\frac{nm}{gd}} \sum_{i = 1}^{\frac{n}{gd}} \lfloor \frac{m \cdot g}{p \cdot i \cdot d^2} \rfloor \mu(d) \end{aligned} ===========​i=1∑n​j=1∑m​d(ij)i=1∑n​j=1∑m​p∣ij∑​1p=1∑nm​i=1∑n​j=1∑m​[p∣ij]p=1∑nm​i=1∑n​j=1∑m​[gcd(i,p)p​∣j]p=1∑nm​i=1∑n​⌊gcd(i,p)p​m​⌋g=1∑n​p=1∑nm​i=1∑n​⌊pm⋅g​⌋[gcd(i,p)==g]g=1∑n​p=1∑nm​i=1∑n​⌊pm⋅g​⌋d∣gcd(gi​,gp​)∑​μ(d)g=1∑n​p=1∑gnm​​i=1∑gn​​⌊p⋅im⋅g​⌋d∣gcd(i,p)∑​μ(d)g=1∑n​p=1∑gnm​​i=1∑gn​​d∣gcd(i,p)∑​⌊p⋅im⋅g​⌋μ(d)g=1∑n​d∑​d∣p∑gnm​​d∣i∑gn​​⌊p⋅im⋅g​⌋μ(d)g=1∑n​d∑​p=1∑gdnm​​i=1∑gdn​​⌊p⋅i⋅d2m⋅g​⌋μ(d)g=1∑n​d∑​p=1∑gdnm​​i=1∑gdn​​⌊p⋅i⋅d2m⋅g​⌋μ(d)​

IDSY QAQ
关注
专栏目录
莫比乌斯反演学习笔记
CCX_CoolMint的博客
05-22 1014
莫比乌斯反演学习笔记
莫比乌斯反演入门 学习笔记
永爱米蒂
08-14 189
文章目录前言引入简介莫比乌斯函数欧拉函数基本公式解题基本步骤后记 前言 久仰其大名，近日学之，知不易。 引入 【BZOJ2820】 YY的GCD 给定n,mn,mn,m,求1≤x≤n1≤x≤n1≤x≤n,1≤y≤m1≤y≤m1≤y≤m且gcd(x,y)gcd(x,y)gcd(x,y)为质数的(x,y)(x,y)(x,y)有多少对。 若暴力枚举之，时间复杂度过θ(n2)\theta(n^2)θ(n2...
「学习笔记」莫比乌斯反演
_Gion
08-19 509
前言 （今天入坑莫比乌斯反演，感觉全都是不可做题..） 莫比乌斯反演属于数论中较难的部分吧，做这种题一般长片的推导，化简，最后用枚举或者整除分块等求出答案（我做题少只见过这些）. 下面要引入一些东西做铺垫. 数论函数与积性函数 首先定义数论函数f(x)f(x)f(x)：定义域和值域都是整数的函数. 定义积性函数：∀a,b,gcd(a,b)=1∀a,b,gcd(a,b)=1\forall...
数学/数论专题-学习笔记：莫比乌斯反演
Plozia 的小窝
02-23 532
莫比乌斯反演的学习笔记。
学习笔记——莫比乌斯反演
weixin_30432007的博客
05-26 84
整除分块 通常，要求 \[\sum_{i=1}^{n}{\lfloor\frac{n}{i}\rfloor}\] 需要的时间为\(O(n)\)。 但是实际上，对于几块连续的\(i\)，\(\lfloor\frac{n}{i}\rfloor\)的值是相同的。 于是，我们可以计算出每块的长度，然后用乘法求和。 对于每块以\(i\)开始的区间，它的右端点都是\(n/(n/i)\)。 for (int l...
学习笔记 ---- 莫比乌斯反演总结
weixin_55851276的博客
09-02 1107
啦啦啦
[学习笔记]莫比乌斯反演常见模型
女装的你，如此好看！
01-31 797
莫比乌斯反演常用的公式就是这两个： 1、∑d|nμ(d)=[n=1]" role="presentation" style="position: relative;">∑d|nμ(d)=[n=1]∑d|nμ(d)=[n=1]\sum_{d|n}\mu(d)=[n=1] 2、∑d|nϕ(d)=n" role="presentation" s
莫比乌斯反演专题学习笔记
weixin_34162695的博客
03-11 121
莫比乌斯反演专题学习笔记 本文记录一些和莫反有关的内容的笔记 可能存在诸多谬误，阅读时请谨慎分析 若发现文中有谬误，如您愿意，恳请您向我指出，不胜感激！ 为什么要学莫比乌斯反演？ 解决一类与狄利克雷卷积、整除、积性函数有关的问题，通过莫比乌斯反演，往往可以将复杂度优化到可接受的范围内 积性函数 对于任意互质整数\(a,b\)有\(f(ab)=f(a)f(b)\)的函数称为积性函数 对于任意整数...
【算法竞赛学习笔记】莫比乌斯反演-数学提升计划
SC_Linno的博客
11-19 987
title : 莫比乌斯反演 date : 2021-8-19 tags : ACM,数学 author : Linno 前置芝士 数论函数 {单位函数ε={1，n=10,otherwise幂函数Idk(n)=nk,当k=1时为恒等函数Id(n)，当k=1时为常函数除数函数σk(n)=∑d∣ndk，当k=1时为因数和函数σ(n)，当k=0时为因数个数函数σo（n）欧拉函数φ(n) \begin{cases} 单位函数\varepsilon=\begin{cases}1，n=1\\0,otherwise.
【算法笔记】莫比乌斯反演（包含定理，两种形式的证明及入门经典模板）
繁凡さん的博客
10-23 1657
一、莫比乌斯反演 学习笔记，我是看这个博客入门的，讲的非常好，传送门，关键是给出了非常多的定理，好多是数论书上的权威概念。 我自己证明的照片在文末，有点乱 首先，莫比乌斯反演是什么？ 第一种形式： F(n)=∑d∣nf(d)=>f(n)=∑d∣nμ(d)F(nd)F(n)=\sum_{d|n}f(d)=>f(n)=\sum_{d|n}\mu(d)F(\frac{n}{d})F(n)=d∣n∑​f(d)=>f(n)=d∣n∑​μ(d)F(dn​) 第二种形式： F(n)=∑n∣df(d)=&
『数学相关』莫比乌斯反演学习笔记
Ronaldo7_ZYB的博客
04-25 273
文章目录莫比乌斯反演作用莫比乌斯反演具体内容莫比乌斯函数μ 莫比乌斯反演作用 我们规定函数f(n)代表如下含义：f(n) = ∑d∣ng(d)f(n)\ =\ \sum_{d|n} g(d)f(n) = ∑d∣n​g(d)。 在某一些情况下，莫比乌斯反演的f函数很容易求解，但其中的g函数则不容易求解；我们需要通过某一种算法，在知道每一个g的情况下求解出对应的...
学习笔记 | 反演 | 二项式反演 | 莫比乌斯反演 | 狄利克雷卷积 |容斥 | 未完待更 | 在更
Averyta的博客
08-11 398
文章目录学习笔记 | 反演 | 二项式反演 | 莫比乌斯反演 | 狄利克雷卷积 |容斥参考资料。前言引入简单的容斥应用-全错排幕后反演的定义莫比乌斯反演！ 学习笔记 | 反演 | 二项式反演 | 莫比乌斯反演 | 狄利克雷卷积 |容斥 参考资料。 炫酷反演魔术-vfleaking 莫比乌斯反演-PoPoQQQ 【数论】积性函数、莫比乌斯反演、狄利克雷卷积-nanoape 如何证明莫比乌斯反演 前...
什么是标准差？详解
发几十块的吧
10-22 1490
标准差计算详解。
熵与信息论
m0_54373077的博客
10-28 776
经典信息论的核心概念是香农熵。假设我们得到了一个变量X的值，X的香农熵量化了我们在获悉 X的值时所能得到的平均信息量；另一种观点是将X的看作在我们获悉的值前对其不确定程度的度量。这两种观点是互补的；我们既可以将看作在我们获悉X的值前，对其不确定程度的度量，也可以看作在我们已经得到的值后，对获得信息量的度量。 直觉上，随机变量的信息内容不应该依赖于随机变量取值的标签。例如，有一个随机变量取值为“头”和“尾”的概率分别为 1/4与3/4，另一个随机变量取值为0和1的概率分别为1/
基础知识-因果分析-dayfour-随机变量的数字特征
qq_27047075的博客
10-31 89
随机变量的概率分布（离散随机变量是概率分布值，连续随机变量是概率密度函数）对随机变量进行了完整描述，但在实际工作中，难以确定随机变量的概率分布，同时，对于一些实际问题，并不需要掌握随机变量的概率分布，只要知道与随机变量概率分布相关的一些特征就够了，因此，我们在损失一些数据信息的情况下，用与随机变量的概率分布有关的一些特征对该变量的取值情况进行描述，这些特征称为随机变量的数字特征。1.期望值随机变量的期望值（也称为均值）为该变量各个可能的取值乘以其相应的概率，再将得到的乘积加和。随机变量对应取值的概率。
扩展卡尔曼滤波（EKF）的限制
weixin_43608857的博客
10-31 110
当f (x)或h (x)接近线性时，EKF在许多实际问题上表现良好。然而，它在高度非线性的区域中失败了EKF的概念是基于模型的线性化而提出的。EKF估计包括线性化误差。线性化误差取决于相对于传播的不确定度的函数的非线性度，如下图所示。图13.13：线性化误差。
【数学二】线性代数-行列式
最新发布
WEL测试
10-31 169
【数学二】线性代数-行列式
莫比乌斯反演详解与应用
“莫比乌斯反演-莫比乌斯反演经典讲稿” 莫比乌斯反演是一种在数论中用于解决计数问题的技巧，它由德国数学家莫比乌斯提出。在数论和组合数学中，它常常用于简化计算，特别是在处理关于数的乘积的问题时。讲稿来自...
评论 1
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
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
实付元
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