莫比乌斯变换
约定
n = p 1 a 1 p 2 a 2 ⋯ p r a r \ n=p_{1}^{a_{1}} p_{2}^{a_{2}} \cdots p_{r}^{a_{r}} n=p1a1p2a2⋯prar
a ∣ b \ a \mid b a∣b: a \ a a整除 b \ b b
a ∤ b \ a \nmid b a∤b: a \ a a不整除 b \ b b
a k ∥ b \ a^{k} \parallel b ak∥b: a k ∣ b \ a^{k} \mid b ak∣b且 a k + 1 ∤ b \ a^{k+1} \nmid b ak+1∤b
( a , b ) \ (a,b) (a,b)最大公约数
[ a , b ] \ [a,b] [a,b]最小公倍数
定义
我们设数论函数 f ( n ) \ f(n) f(n),我们考虑一种运算:
F ( n ) = ∑ d ∣ n f ( d ) ( 1 ) F(n)= \sum_{d \mid n} f(d)\,\,\,\,\,\,\,\,\, (1) F(n)=d∣n∑f(d)(1)
我们看到过很多这样关系:
f ( n ) \ f(n) f(n) | F ( n ) \ F(n) F(n) |
---|---|
1 \ 1 1 | τ ( n ) \ \tau(n) τ(n) |
n \ n n | σ ( n ) \ \sigma(n) σ(n) |
φ ( n ) \ \varphi(n) φ(n) | n \ n n |
μ ( n ) \ \mu(n) μ(n) | [ 1 n ] \ \left[ \frac{1}{n} \right] [n1] |
μ ( n ) n \ \frac{ \mu(n)}{n} nμ(n) | φ ( n ) n \ \frac{ \varphi(n)}{n} nφ(n) |
λ ( n ) \ \lambda(n) λ(n) | ln n \ \ln n lnn |
h ( n ) \ h(n) h(n) | N ( n ) 4 \ \frac{N(n)}{4} 4N(n) |
证明略。
我们称 F ( n ) \ F(n) F(n)为 f ( n ) \ f(n) f(n)的莫比乌斯变换。
定理
定理1
F ( n ) = { 1 , n = 1 ∑ e 1 = 0 a 1 ∑ e 2 = 0 a 2 ⋯ ∑ e r = 0 a r f ( p 1 e 1 p 2 e 2 ⋯ p r e r ) , n > 1 F(n)= \begin{cases} 1,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,n=1 \\ \displaystyle \sum_{e_{1}=0}^{a_{1}} \sum_{e_{2}=0}^{a_{2}} \cdots \sum_{e_{r}=0}^{a_{r}} f(p_{1}^{e_{1}} p_{2}^{e_{2}} \cdots p_{r}^{e_{r}}), \,\,\,n>1 \end{cases} F(n)=⎩⎪⎨⎪⎧1,n=1e1=0∑a1e2=0∑a2⋯er=0∑arf(p1e1p2e2⋯prer),n>1
若 f ( n ) \ f(n) f(n)积性, F ( n ) \ F(n) F(n)积性:
F ( n ) = ∏ j = 1 r ( 1 + f ( p j ) + f ( p j 2 ) + ⋯ f ( p j a r ) ) = ∏ p a ∥ n ( 1 + f ( p ) + f ( p 2 ) + ⋯ f ( p a ) ) ( 2 ) F(n)= \prod_{j=1}^{r}(1+f(p_{j})+f(p_{j}^{2})+ \cdots f(p_{j}^{a_{r}})) \\ \,\,\,\,\,\,\,\,\,\,\,\,= \prod_{p^{a} \parallel n}(1+f(p)+f(p^{2})+ \cdots f(p^{a}))\, \\ (2) F(n)=j=1∏r(1+f(pj)+f(pj2)+⋯f(pjar))=pa∥n∏(1+f(p)+f(p2)+⋯f(pa))(2)
若 f ( n ) \ f(n) f(n)完全积性, F ( n ) \ F(n) F(n)完全积性:
F ( n ) = ∏ j = 1 r ( 1 + f ( p j ) + f 2 ( p j ) + ⋯ f a j ( p j ) ) = ∏ p a ∥ n ( 1 + f ( p ) + f 2 ( p ) + ⋯ f a ( p ) ) ( 4 ) F(n)= \prod_{j=1}^{r}(1+f(p_{j})+f^{2}(p_{j})+ \cdots f^{a_{j}}(p_{j})) \\ \,\,\,\,\,\,\,\,\,=\prod_{p^{a} \parallel n}(1+f(p)+f^{2}(p)+ \cdots f^{a}(p))\,\,\,\, \\ (4) F(n)=j=1∏r(1+f(pj)+f2(pj)+⋯faj(pj))=pa∥n∏(1+f(p)+f2(p)+⋯fa(p))(4)
证明
这
F
(
1
)
=
f
(
1
)
\ F(1)=f(1)
F(1)=f(1)的证明是显然的。
F ( n ) = ∑ e 1 = 0 a 1 ∑ e 2 = 0 a 2 ⋯ ∑ e r = 0 a r f ( p 1 e 1 p 2 e 2 ⋯ p r e r ) = { ∑ e 1 = 0 a 1 f ( p 1 e 1 ) } { ∑ e 2 a 2 f p 2 e 2 } ⋯ { ∑ e r = 0 a r f ( p r e r ) } F(n)=\sum_{e_{1}=0}^{a_{1}} \sum_{e_{2}=0}^{a_{2}} \cdots \sum_{e_{r}=0}^{a_{r}} f(p_{1}^{e_{1}} p_{2}^{e_{2}} \cdots p_{r}^{e_{r}})\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ \\ \,\,\,\,\,\,= \{ \sum_{e_{1}=0}^{a_{1}}f(p_{1}^{e_{1}}) \} \{ \sum_{e_{2}}^{a_{2}}f_{p_{2}^{e_{2}}} \} \cdots \{ \sum_{e_{r}=0}^{a_{r}}f(p_{r}^{e_{r}}) \} F(n)=e1=0∑a1e2=0∑a2⋯er=0∑arf(p1e1p2e2⋯prer) ={e1=0∑a1f(p1e1)}{e2∑a2fp2e2}⋯{er=0∑arf(prer)}
显然由(3)式得:
F ( n ) = F ( p 1 a 1 ) F ( p 2 a 2 ) ⋯ F ( p r a r ) ( 5 ) F(n)=F(p_{1}^{a_{1}}) F(p_{2}^{a_{2}}) \cdots F(p_{r}^{a_{r}})\,\,\,\,\,\,\,\,\,(5) F(n)=F(p1a1)F(p2a2)⋯F(prar)(5)
由此推出其积性。
显然
f
(
n
)
\ f(n)
f(n)完全积性的时候,显然证明
(4)式。
证毕
引理2
设 ( m , n ) = 1 \ (m,n)=1 (m,n)=1, k \ k k是给定的正整数,那么对于每个 d ∣ m n \ d \mid mn d∣mn成立的充要条件是存在唯一的一对正整数 d 1 , d 2 \ d_{1},d_{2} d1,d2满足 d = d 1 d 2 , d 1 k ∣ m , d 2 k ∣ n ( 6 ) \ d=d_{1}d_{2},\,\,\,\,\,\,\,\,\,d_{1}^{k} \mid m,\,\,\,\,\,\,\,\,\, d_{2}^{k} \mid n\,\,\,\,\,\,\,\,\,(6) d=d1d2,d1k∣m,d2k∣n(6)
证明:
( m , n ) = 1 , d k ∣ m n ⇒ d k = ( d k , m n ) = ( d k , m ) ( d k , n ) ( 7 ) (m,n)=1,\,\,\,d^{k} \mid mn \Rightarrow d^{k}=(d^{k},mn)=(d^{k},m)(d^{k},n)\,\,\,\,\,\,\,\,\,(7) (m,n)=1,dk∣mn⇒dk=(dk,mn)=(dk,m)(dk,n)(7)
显见 ( ( d k , m ) , ( d k , n ) ) = 1 \ ((d^{k},m),(d^{k},n))=1 ((dk,m),(dk,n))=1,所以:
( d k , m ) = ( ( d k , m ) , d ) k = ( d , m ) k ( 8 ) ( d k , n ) = ( ( d k , n ) , d ) k = ( d , n ) k ( 9 ) (d^{k},m)=((d^{k},m),d)^{k}=(d,m)^{k}\,\,\,\,\,\,\,\,\,(8) \\ (d^{k},n)=((d^{k},n),d)^{k}=(d,n)^{k}\,\,\,\,\,\,\,\,\,(9) (dk,m)=((dk,m),d)k=(d,m)k(8)(dk,n)=((dk,n),d)k=(d,n)k(9)
取 d 1 = ( d , m ) , d 2 = ( d , n ) \ d_{1}=(d,m),d_{2}=(d,n) d1=(d,m),d2=(d,n),由(6)推出,反过来,若(6)成立, d k ∣ m n \ d^{k} \mid mn dk∣mn是显然的,再由(7)(8)(9)知:
d 1 k d 2 k = ( d k , m ) ( d k , n ) = ( d , m ) k ( d , n ) k d_{1}^{k}d_{2}^{k}=(d^{k},m)(d^{k},n)=(d,m)^{k}(d,n)^{k} d1kd2k=(dk,m)(dk,n)=(d,m)k(d,n)k
我们注意到:
( m , n ) = ( d 1 , n ) = ( d 2 , m ) = 1 (m,n)=(d_{1},n)=(d_{2},m)=1 (m,n)=(d1,n)=(d2,m)=1
所以:
d 1 = ( d , m ) , d 2 = ( d , n ) d_{1}=(d,m),d_{2}=(d,n) d1=(d,m),d2=(d,n)
证毕
\ \\\,\\\,\\\,
定理1的另一种证明
我们发现定理1和引理2是独立无关的。
我们设
(
m
,
n
)
=
1
,
d
1
,
d
2
\ (m,n)=1,d_{1},d_{2}
(m,n)=1,d1,d2。
F
(
m
n
)
=
∑
d
∣
m
n
f
(
d
)
=
∑
d
1
∣
m
∑
d
2
∣
n
f
(
d
1
d
2
)
\ F(mn)= \displaystyle \sum_{d \mid mn}f(d)= \sum_{d_{1} \mid m} \sum_{d_{2} \mid n}f(d_{1}d_{2})\,\,\,\,\,\,\,
F(mn)=d∣mn∑f(d)=d1∣m∑d2∣n∑f(d1d2)
=
∑
d
1
∣
m
f
(
d
1
)
∑
d
2
∣
m
f
(
d
2
)
=
F
(
m
)
F
(
n
)
(
10
)
\ \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,= \displaystyle \sum_{d_{1} \mid m} f(d_{1}) \sum_{d_{2} \mid m} f(d_{2})=F(m)F(n)\,\,\,\,\,\,\,\,\,(10)
=d1∣m∑f(d1)d2∣m∑f(d2)=F(m)F(n)(10)
推论3
设积性函数
f
(
n
)
\ f(n)
f(n),我们有:
∑
d
∣
m
μ
(
d
)
f
(
d
)
=
∏
p
∣
n
(
1
−
f
(
p
)
)
(
11
)
\sum_{d \mid m} \mu(d)f(d) =\prod_{p\mid n}(1-f(p))\,\,\,\,\,\,\,\,\,(11)
d∣m∑μ(d)f(d)=p∣n∏(1−f(p))(11)
和
∑ d ∣ n μ 2 ( d ) f ( n ) = ∏ p ∣ n ( 1 + f ( p ) ) ( 12 ) \sum_{d \mid n} \mu^{2}(d)f(n)= \prod_{p \mid n}(1+f(p))\,\,\,\,\,\,\,\,\,(12) d∣n∑μ2(d)f(n)=p∣n∏(1+f(p))(12)
这是容易得到的,就不展示证明了。
取 f ( n ) = 1 n \ f(n)= \frac{1}{n} f(n)=n1,得
∑ d ∣ n μ ( d ) d = ∏ p ∣ n ( 1 − 1 p ) = φ ( n ) n ( 13 ) \sum_{d \mid n} \frac{\mu (d)}{d}=\prod_{p \mid n}(1- \frac{1}{p})=\frac{\varphi(n)}{n}\,\,\,\,\,\,\,\,\,(13) d∣n∑dμ(d)=p∣n∏(1−p1)=nφ(n)(13)
我们也能想到莫比乌斯反演不是对整数成立的。所以我们可以用此优化很多算法包括优化精度。
特别的,取 f ( n ) = 1 \ f(n)=1 f(n)=1.
∑ d ∣ n μ ( d ) = { 1 , n = 1 0 , n > 1 ( 14 ) \sum_{d \mid n} \mu(d)= \begin{cases} 1,\,\,\,n=1 \\ 0,\,\,\,n>1 \end{cases} \\ (14) d∣n∑μ(d)={1,n=10,n>1(14)
定理4
设函数 f ( n ) , F ( n ) \ f(n),F(n) f(n),F(n),若要使(1)式成立,充要条件3为:
f
(
n
)
=
∑
d
∣
n
μ
(
d
)
F
(
n
d
)
(
15
)
f(n)= \sum_{d \mid n} \mu(d)F(\frac{n}{d})\,\,\,\,\,\,(15)
f(n)=d∣n∑μ(d)F(dn)(15)
充分性证明:
∑ d ∣ n f ( d ) = ∑ d ∣ n { ∑ k ∣ d μ ( k ) F ( d k ) } = ∑ k ∣ d μ ( k ) ∑ k ∣ d , d ∣ n F ( d k ) \sum_{d \mid n}f(d)=\sum_{d \mid n} \{ \sum_{k \mid d} \mu(k)F(\frac{d}{k}) \}= \sum_{k\mid d}\mu(k) \sum_{k \mid d,d \mid n} F(\frac{d}{k}) d∣n∑f(d)=d∣n∑{k∣d∑μ(k)F(kd)}=k∣d∑μ(k)k∣d,d∣n∑F(kd)
令
d
=
k
l
\ d=kl
d=kl,得:
∑
d
∣
n
f
(
d
)
=
∑
k
∣
n
μ
(
k
)
∑
l
∣
n
k
F
(
l
)
=
∑
l
∣
n
F
(
l
)
∑
k
∣
n
l
μ
(
k
)
=
F
(
n
)
\sum_{d \mid n}f(d)=\sum_{k \mid n} \mu(k) \sum_{l \mid \frac{n}{k}}F(l)= \sum_{l \mid n}F(l) \sum_{k \mid \frac{n}{l}} \mu(k)=F(n)
d∣n∑f(d)=k∣n∑μ(k)l∣kn∑F(l)=l∣n∑F(l)k∣ln∑μ(k)=F(n)
证毕
必要性有多种方法:
直接带入(15)于(1)
∑
d
∣
n
μ
(
d
)
F
(
n
d
)
=
∑
d
∣
n
μ
(
d
)
∑
l
∣
n
d
f
(
l
)
=
∑
l
∣
n
f
(
l
)
∑
d
∣
n
l
μ
(
d
)
=
f
(
n
)
\sum_{d \mid n} \mu (d)F( \frac{n}{d})=\sum_{d \mid n} \mu(d) \sum_{l \mid \frac{n}{d}}f(l)= \sum_{l \mid n}f(l)\sum_{d \mid \frac{n}{l}} \mu(d)=f(n)
d∣n∑μ(d)F(dn)=d∣n∑μ(d)l∣dn∑f(l)=l∣n∑f(l)d∣ln∑μ(d)=f(n)
证毕
但我们不能看出为什么有(15)这个式子。所以我们要利用(14)来理解:
f ( n ) = ∑ k ∣ n f ( n k ) ∑ d ∣ k μ ( d ) = ∑ d ∣ n μ ( d ) ∑ d ∣ k , k ∣ n f ( n k ) f(n)= \sum_{k \mid n}f(\frac{n}{k}) \sum_{d \mid k} \mu(d) = \sum_{d \mid n} \mu(d) \sum_{d \mid k,k \mid n}f(\frac{n}{k}) f(n)=k∣n∑f(kn)d∣k∑μ(d)=d∣n∑μ(d)d∣k,k∣n∑f(kn)
令 k = d l \ k=dl k=dl,得:
f ( n ) = ∑ d ∣ n μ ( d ) ∑ l ∣ n d f ( n d l ) = ∑ d ∣ n μ ( d ) F ( n d ) f(n)= \sum_{d \mid n}\mu(d) \sum_{l \mid \frac{n}{d}} f(\frac{n}{dl})=\sum_{d \mid n}\mu(d)F(\frac{n}{d}) f(n)=d∣n∑μ(d)l∣dn∑f(dln)=d∣n∑μ(d)F(dn)
证毕
以上为莫比乌斯翻转公式
定理5
设数论函数
f
(
n
)
,
g
(
n
)
,
h
(
n
)
\ f(n),g(n),h(n)
f(n),g(n),h(n)。
h
(
n
)
=
∑
d
∣
n
f
(
n
)
g
(
n
d
)
(
16
)
h(n)=\sum_{d \mid n}f(n)g(\frac{n}{d})\,\,\,\,\,\,\,\,\,(16)
h(n)=d∣n∑f(n)g(dn)(16)
f ( n ) , g ( n ) \ f(n),g(n) f(n),g(n)积性时, h ( n ) \ h(n) h(n)积性。
证明
f ( 1 ) = g ( 1 ) = 1 ⇒ h ( 1 ) = 1 f(1)=g(1)=1 \Rightarrow h(1)=1 f(1)=g(1)=1⇒h(1)=1
设 ( m , n ) = 1 \ (m,n)=1 (m,n)=1,利用引理2
h ( m n ) = ∑ d ∣ m n f ( d ) g ( m n d ) = ∑ d 1 ∣ m , d 2 ∣ n f ( d 1 d 2 ) g ( m n d 1 d 2 ) = h ( m ) h ( n ) h(mn)=\sum_{d \mid mn}f(d)g(\frac{mn}{d})=\sum_{d_{1} \mid m,d_{2} \mid n}f(d_{1}d_{2})g(\frac{mn}{d_{1}d_{2}})=h(m)h(n) h(mn)=d∣mn∑f(d)g(dmn)=d1∣m,d2∣n∑f(d1d2)g(d1d2mn)=h(m)h(n)
证毕
推论6
f ( n ) \ f(n) f(n)积性的充要条件是 F ( n ) \ F(n) F(n)积性。
证明:
我们显然得知:
f ( p a ) = F ( p a ) − F ( p a − 1 ) f(p^{a})=F(p^{a})-F(p^{a-1}) f(pa)=F(pa)−F(pa−1)
几乎瞬间证毕
应用例子
ex1
F ( n ) = n t \ F(n)=n^{t} F(n)=nt求 f ( n ) \ f(n) f(n)
F ( n ) \ F(n) F(n)积性,所以有:
f ( p a ) = p a t − p ( a − 1 ) t = p a t ( a − p − t ) f(p^{a})=p^{at}-p^{(a-1)t}=p^{at}(a-p^{-t}) f(pa)=pat−p(a−1)t=pat(a−p−t)
所以有
f ( n ) = n t ∏ p ∣ n ( 1 − p − t ) f(n)=n^{t} \prod_{p \mid n}(1-p^{-t}) f(n)=ntp∣n∏(1−p−t)
ex2
F ( n ) = φ ( n ) \ F(n)= \varphi(n) F(n)=φ(n),求 f ( n ) \ f(n) f(n)。
F ( n ) \ F(n) F(n)积性,所以有:
f ( p a ) = φ ( p a ) − φ ( p a − 1 ) = { p ( 1 − 2 p ) , a = 1 p a ( 1 − 1 p ) 2 , a > 1 f(p^{a})=\varphi(p^{a})-\varphi(p^{a-1})= \begin{cases} p(1-\frac{2}{p}), \,\,\,\,\,\,\,\,a=1 \\ p^{a}(1-\frac{1}{p})^{2}, \,\,\,a>1 \end{cases} f(pa)=φ(pa)−φ(pa−1)={p(1−p2),a=1pa(1−p1)2,a>1
ex3
f ( n ) = μ 2 ( n ) φ ( n ) \ f(n)=\frac{\mu^{2}(n)}{\varphi(n)} f(n)=φ(n)μ2(n),求 F ( n ) \ F(n) F(n)。
∑ d ∣ p a μ 2 ( d ) φ ( d ) = 1 + 1 p − 1 = ( 1 − 1 p ) − 1 \sum_{d \mid p^{a}}\frac{\mu^{2}(d)}{\varphi(d)}=1+\frac{1}{p-1}=(1-\frac{1}{p})^{-1} d∣pa∑φ(d)μ2(d)=1+p−11=(1−p1)−1
所以得:
∑ d ∣ n μ 2 ( d ) φ ( d ) = ∏ p ∣ n ( 1 − 1 p ) − 1 = n φ ( n ) \sum_{d \mid n} \frac{\mu^{2}(d)}{\varphi(d)}=\prod_{p \mid n}(1- \frac{1}{p})^{-1}=\frac{n}{\varphi(n)} d∣n∑φ(d)μ2(d)=p∣n∏(1−p1)−1=φ(n)n
ex5
求
f
(
n
,
m
)
=
∑
i
n
∑
j
m
[
(
i
,
j
)
=
1
]
f(n,m)=\sum_{i}^{n} \sum_{j}^{m}[(i,j)=1]
f(n,m)=i∑nj∑m[(i,j)=1]
证明
f ( n , m ) = ∑ i n ∑ j m ∑ d ∣ ( i , j ) μ ( d ) = ∑ d μ ( d ) ∑ d ∣ i ∑ d ∣ j 1 = ∑ d μ ( d ) [ n d ] [ m d ] \begin{array}{crl} & &\,\,\,\,\,\,\,f(n,m) \\ & &=\sum_{i}^{n}\sum_{j}^{m}\sum_{d \mid (i,j)} \mu(d) \\ & &=\sum_{d} \mu(d) \sum_{d \mid i} \sum_{d \mid j} 1 \\ & &=\sum_{d} \mu(d) \left[ \frac{n}{d}\right] \left[ \frac{m}{d}\right] \end{array} f(n,m)=∑in∑jm∑d∣(i,j)μ(d)=∑dμ(d)∑d∣i∑d∣j1=∑dμ(d)[dn][dm]
ex5.5
求
f
(
n
,
m
)
=
∑
i
n
∑
j
m
(
i
,
j
)
f(n,m)=\sum_{i}^{n} \sum_{j}^{m}(i,j)
f(n,m)=i∑nj∑m(i,j)
证明
f ( n , m ) = ∑ i n ∑ j m ∑ d ∣ ( i , j ) φ ( d ) = ∑ d φ ( d ) ∑ d ∣ i ∑ d ∣ j 1 = ∑ d φ ( d ) [ n d ] [ m d ] \begin{array}{crl} & &\,\,\,\,\,\,\,f(n,m) \\ & &=\sum_{i}^{n}\sum_{j}^{m}\sum_{d \mid (i,j)} \varphi(d) \\ & &=\sum_{d} \varphi(d) \sum_{d \mid i} \sum_{d \mid j} 1 \\ & &=\sum_{d} \varphi(d) \left[ \frac{n}{d}\right] \left[ \frac{m}{d}\right] \end{array} f(n,m)=∑in∑jm∑d∣(i,j)φ(d)=∑dφ(d)∑d∣i∑d∣j1=∑dφ(d)[dn][dm]
eeeexxxx5555
设整系数多项式 P ( n ) \ P(n) P(n), S ( n ; P ( x ) ) \ S(n;P(x)) S(n;P(x))为满足 ( P ( d ) , n ) = 1 , 1 ≤ d ≤ n \ (P(d),n)=1,\,\,\,1 \le d \le n (P(d),n)=1,1≤d≤n的个数。
试证明 S ( n ) = S ( n ; P ( x ) ) \ S(n)=S(n;P(x)) S(n)=S(n;P(x))积性。
利用式(14)
S ( n ) = ∑ d = 1 , ( P ( d ) , n ) = 1 n 1 = ∑ d = 1 n ∑ k ∣ ( P ( d ) , n ) μ ( k ) = ∑ k ∣ n μ ( k ) ∑ d = 1 , k ∣ P ( d ) n 1 S(n)= \sum_{d=1,(P(d),n)=1}^{n}1=\sum_{d=1}^{n}\sum_{k \mid (P(d),n)}\mu(k)= \sum_{k \mid n} \mu(k) \sum_{d=1,k \mid P(d)}^{n}1 S(n)=d=1,(P(d),n)=1∑n1=d=1∑nk∣(P(d),n)∑μ(k)=k∣n∑μ(k)d=1,k∣P(d)∑n1
以 T ( k ) = T ( k ; P ( x ) ) \ T(k)=T(k;P(x)) T(k)=T(k;P(x))表示以下同余方程的解数:
P ( x ) ≡ 0 ( m o d k ) P(x) \equiv 0 \pmod{k} P(x)≡0(modk)
k ∣ n \ k \mid n k∣n时,有:
∑
d
=
1
,
k
∣
P
(
d
)
n
=
n
k
T
(
k
)
\sum_{d=1,k \mid P(d)}^{n}=\frac{n}{k}T(k)
d=1,k∣P(d)∑n=knT(k)
所以
S ( n ) = n ∑ k ∣ n μ ( k ) T ( k ) k S(n)=n \sum_{k \mid n}\frac{ \mu(k)T(k)}{k} S(n)=nk∣n∑kμ(k)T(k)
显然是积性的。
其他
h ( n ) = ∑ d ∣ n f ( d ) g ( n d ) h(n)= \sum_{d \mid n}f(d)g(\frac{n}{d}) h(n)=d∣n∑f(d)g(dn)
也写作:
h = f ∗ g h=f*g h=f∗g
我们称 h \ h h为 f \ f f和 g \ g g的狄利克雷卷积,为一种重要卷积形式。