黎曼 ζ ( 2 ) ζ(2) ζ(2)的导数: ζ ′ ( 2 ) = − 1 ζ'(2)=-1 ζ′(2)=−1吗?
http://mathworld.wolfram.com/Glaisher-KinkelinConstant.html
( n − x ) ′ = − ln n n x , (n^{-x})'=-\frac{\ln n}{n^x} , (n−x)′=−nxlnn,
− ζ ′ ( 2 ) = ( − ∑ n = 1 ∞ 1 n 2 ) ′ = ln 2 2 2 + ln 3 3 2 + ln 4 4 2 + . . . + ln n n 2 + . . . -\zeta'(2)=\left(-\sum_{n=1}^{\infty}{\frac{1}{n^2}} \right)'=\frac{\ln2}{2^{2}}+\frac{\ln3}{3^{2}}+\frac{\ln4}{4^{2}}+...+\frac{\ln n}{n^{2}}+... −ζ′(2)=(−n=1∑∞n21)′=22ln2+32ln3+42ln4+...+n2lnn+...
= π 2 6 ( 12 ln A − γ − ln 2 − ln π ) =\frac{\pi^{2}}{6}(12\ln A-\gamma-\ln2-\ln\pi) =6π2(12lnA−γ−ln2−lnπ)
= 0.93754825431584... ≈ 1 , ( n → ∞ ) =0.93754825431584...\approx1,(n\rightarrow\infty) =0.93754825431584...≈1,(n→∞)
ζ ′ ′ ( 2 ) = ∑ n = 1 ∞ ln 2 n n 2 = 1.989280234... ≈ 2 ! = 2 \zeta''(2)=\sum_{n=1}^{\infty}{\frac{\ln^2 n}{n^2}}=1.989280234...\approx2!=2 ζ′′(2)=n=1∑∞n2ln2n=1.989280234...≈2!=2
− ζ ′ ′ ′ ( 2 ) = ∑ n = 2 ∞ ln 3 n n 2 = 6. 0 0 0 145802... ≈ 3 ! = 6 -\zeta'''(2)=\sum_{n=2}^{\infty}{\frac{\ln^3 n}{n^2}}=6.{\color{red}0}{\color{red}0}{\color{red}0} 145802...\approx3!=6 −ζ′′′(2)=n=2∑∞n2ln3n=6.000145802...≈3!=6
ζ ( 4 ) ( 2 ) = ∑ n = 1 ∞ ln 4 n n 2 = 24. 0 0 1486393... ≈ 4 ! = 24 \zeta^{(4)}(2)=\sum_{n=1}^{\infty}{\frac{\ln^4 n}{n^2}}=24.{\color{red}0}{\color{red}0}1486393...\approx4!=24 ζ(4)(2)=n=1∑∞n2ln4n=24.001486393...≈4!=24
− ζ ( 5 ) ( 2 ) = ∑ n = 1 ∞ ln 5 n n 2 = 120. 000 8243332... ≈ 5 ! = 120 -\zeta^{(5)}(2)=\sum_{n=1}^{\infty}{\frac{\ln^5 n}{n^2}}=120.{\color{red}000}8243332...\approx5!=120 −ζ(5)(2)=n=1∑∞n2ln5n=120.0008243332...≈5!=120
ζ ( 6 ) ( 2 ) = ∑ n = 1 ∞ ln 6 n n 2 = 720. 000 12472831... ≈ 6 ! = 720 \zeta^{(6)}(2)=\sum_{n=1}^{\infty}{\frac{\ln^6 n}{n^2}}=720.{\color{red}000}12472831...\approx6!=720 ζ(6)(2)=n=1∑∞n2ln6n=720.00012472831...≈6!=720
− ζ ( 7 ) ( 2 ) = ∑ n = 1 ∞ ln 7 n n 2 = 5039. 999 783274... ≈ 7 ! = 5040 -\zeta^{(7)}(2)=\sum_{n=1}^{\infty}{\frac{\ln^7 n}{n^2}}=5039.{\color{red}999}783274...\approx7!=5040 −ζ(7)(2)=n=1∑∞n2ln7n=5039.999783274...≈7!=5040
ζ ( 8 ) ( 2 ) = ∑ n = 1 ∞ ln 8 n n 2 = 40319. 999 74686... ≈ 8 ! = 40320 \zeta^{(8)}(2)=\sum_{n=1}^{\infty}{\frac{\ln^8 n}{n^2}}=40319.{\color{red}999}74686...\approx8!=40320 ζ(8)(2)=n=1∑∞n2ln8n=40319.99974686...≈8!=40320
− ζ ( 9 ) ( 2 ) = ∑ n = 1 ∞ ln 9 n n 2 = 362879. 999 8677... ≈ 9 ! = 362880 -\zeta^{(9)}(2)=\sum_{n=1}^{\infty}{\frac{\ln^9 n}{n^2}}=362879.{\color{red}999}8677...\approx9!=362880 −ζ(9)(2)=n=1∑∞n2ln9n=362879.9998677...≈9!=362880
太神奇了! ≈ \approx ≈ 我怀疑是 e , π , A 与 γ e,\pi, A与\gamma e,π,A与γ 的误差造成的!与收敛的快慢有关。
∣ − ζ ( k ) ( 2 ) ∣ = ∣ ( − ∑ n = 1 ∞ 1 n 2 ) ( k ) ∣ = ∑ n = 1 ∞ ( ln n ) k n 2 = k ! |-\zeta^{(k)}(2)|=\left| \left(-\sum_{n=1}^{\infty}{\frac{1}{n^2}} \right)^{(k)} \right|=\sum_{n=1}^{\infty}{\frac{(\ln n)^{k}}{n^2}}=k! ∣−ζ(k)(2)∣=∣∣∣∣∣∣(−n=1∑∞n21)(k)∣∣∣∣∣∣=n=1∑∞n2(lnn)k=k!
意味着:
∣ − ζ ( k ) ( 2 ) ∣ = ∑ n = 1 ∞ ( ln n ) k n 2 = ∫ 1 ∞ ( ln x ) k x 2 d x = ∫ 0 ∞ x k e x d x = k ! |-\zeta^{(k)}(2)|=\sum_{n=1}^{\infty}{\frac{(\ln n)^{k}}{n^2}}=\int_{1}^{\infty}\frac{(\ln x)^{k}}{x^2}dx=\int_{0}^{\infty}\frac{x^{k}}{e^{x}}dx=k! ∣−ζ(k)(2)∣=n=1∑∞n2(lnn)k=∫1∞x2(lnx)kdx=∫0∞exxkdx=k!
lim n → ∞ ∑ n = 1 ∞ ( ln n ) k n 2 Δ x = ∫ 1 ∞ ( ln x ) k x 2 d x = k ! \lim_{n \rightarrow \infty}{}\sum_{n=1}^{\infty}{\frac{(\ln n)^{k}}{n^2}}\Delta x=\int_{1}^{\infty}\frac{(\ln x)^{k}}{x^2}dx=k! n→∞limn=1∑∞n2(lnn)kΔx=∫1∞x2(lnx)kdx=k!
Δ x = b − a n = ∞ − 1 ∞ = 1 \Delta x=\frac{b-a}{n}=\frac{\infty-1}{\infty}=1 Δx=nb−a=∞∞−1=1
− ζ ′ ( 2 ) = 0.93754825431584... ≈ 1 -\zeta'(2)=0.93754825431584...\approx1 −ζ′(2)=0.93754825431584...≈1 误差大是因为收敛得慢计算机算不到无穷远处?后面的误差很小是因为 k k k 较大时收敛得快速?还是我想多了? − ζ ′ ( 2 ) ≠ 1 ? -\zeta'(2)\ne1 ? −ζ′(2)=1?
还有:
( − ∑ x = 1 ∞ 1 e x ) ′ = ∫ 0 ∞ 1 e x d x = 1 , \left( -\sum_{x=1}^{\infty}{\frac{1}{e^{x}}} \right)'=\int_{0}^{\infty}\frac{1}{e^{x}}dx=1, (−x=1∑∞ex1)′=∫0∞ex1dx=1,
− ζ ’ ( 2 ) = ∫ 0 ∞ ln 2 2 x d x = ∫ 1 ∞ ln x x 2 d x = ∫ 1 ∞ 1 x 2 d x = 1 , -\zeta’(2)=\int_{0}^{\infty}\frac{\ln 2}{2^{x}}dx=\int_{1}^{\infty}\frac{\ln x}{x^{2}}dx=\int_{1}^{\infty}\frac{ 1}{x^{2}}dx=1, −ζ’(2)=∫0∞2xln2dx=∫1∞x2lnxdx=∫1∞x21dx=1,
这里有一个重要问题:什么情况下级数和等于积分?即什么情况下当 a n = f ( n ) a_{n}=f(n) an=f(n)时, ∑ n = 1 ∞ a n = ∫ 1 ∞ f ( x ) d x ? \sum_{n=1}^{\infty}{a_{n}}=\int_{1}^{\infty}f(x)dx? ∑n=1∞an=∫1∞f(x)dx?
我在《托马斯微积分》第10版p665页找到了答案:
原文讲积分判别法
设 { a n } \left\{ a_{n} \right\} {an} 是一个正数项序列,假定对 x ≥ N x\geq N x≥N (正整数), a n = f ( n ) , f a_{n}=f(n),f an=f(n),f 是 x x x 的一个连续,正的,递减函数,则级数 ∑ n = N ∞ a n 和 ∫ N ∞ f ( x ) d x \sum_{n=N}^{\infty}{a_{n}} 和 \int_{N}^{\infty}f(x)dx ∑n=N∞an和∫N∞f(x)dx 同时收敛或同时发散。
我们对 N = 1 N=1 N=1 证明,对一般的 N N N 证明是类似的。
我们首先假设 f f f 是减函数并对所有 n n n 使 f ( n ) = a n , f(n)=a_{n} , f(n)=an,图 ( a ) , (a), (a),矩形 a 1 + a 2 + ⋯ + a n a_{1}+a_{2}+\cdots +a_{n} a1+a2+⋯+an面积大于 从 x = 1 x=1 x=1 到 x = n + 1 x=n+1 x=n+1 曲线 y = f ( x ) y=f(x) y=f(x) 下的面积,于是
∫ 1 n f ( x ) d x ≤ ∫ 1 n + 1 f ( x ) d x ≤ a 1 + a 2 + ⋯ + a n \int_{1}^{n}f(x)dx\leq\int_{1}^{n+1}f(x)dx\leq a_{1}+a_{2}+\cdots +a_{n} ∫1nf(x)dx≤∫1n+1f(x)dx≤a1+a2+⋯+an
现在,矩形从往右画改成往左画,图 ( b ) (b) (b)。如果暂时忽略第一个矩形 a 1 , a_{1} , a1,我们看到
a 2 + a 3 + ⋯ + a n ≤ ∫ 1 n f ( x ) d x a_{2}+a_{3}+\cdots +a_{n}\leq\int_{1}^{n}f(x)dx a2+a3+⋯+an≤∫1nf(x)dx
算上 a 1 , a_{1} , a1,我们有 a 1 + a 2 + ⋯ + a n ≤ a 1 + ∫ 1 n f ( x ) d x a_{1}+a_{2}+\cdots +a_{n}\leq a_{1}+\int_{1}^{n}f(x)dx a1+a2+⋯+an≤a1+∫1nf(x)dx
组合这些结果得
( 1 ) : ∫ 1 n f ( x ) d x ≤ a 1 + a 2 + ⋯ + a n ≤ a 1 + ∫ 1 n f ( x ) d x (1):\,\,\,\int_{1}^{n}f(x)dx\leq a_{1}+a_{2}+\cdots +a_{n}\leq a_{1}+\int_{1}^{n}f(x)dx (1):∫1nf(x)dx≤a1+a2+⋯+an≤a1+∫1nf(x)dx
若 ∫ 1 n f ( x ) d x \int_{1}^{n}f(x)dx ∫1nf(x)dx 有限,右面的不等式表明 ∑ a n \sum_{}^{}{a_{n}} ∑an 有限,若 ∫ 1 n f ( x ) d x \int_{1}^{n}f(x)dx ∫1nf(x)dx 无限,左面的不等式表明 ∑ a n \sum_{}^{}{a_{n}} ∑an 无限。
因此,级数和积分同时有限或无限。
收敛的级数与积分并不一定收敛到同一值。
比如 ∑ n = 1 ∞ 1 / n 2 = π 2 / 6 , \sum_{n=1}^{\infty}{1/n^2}=\pi^{2}/6 , ∑n=1∞1/n2=π2/6, 而 ∫ 1 ∞ 1 / x 2 d x = 1 \int_{1}^{\infty}1/x^2dx=1 ∫1∞1/x2dx=1
问题来了,什么情况下收敛的级数与积分收敛到同一值?
由 ( 1 ) (1) (1) 式知: a 1 = 0 a_{1}=0 a1=0 时,且从 n = 2 n=2 n=2 开始, a n a_{n} an与 f ( n ) f(n) f(n) 递减,满足
∫ 1 n f ( x ) d x ≤ 0 + a 2 + ⋯ + a n ≤ ∫ 1 n f ( x ) d x \int_{1}^{n}f(x)dx\leq 0+a_{2}+\cdots +a_{n}\leq\int_{1}^{n}f(x)dx ∫1nf(x)dx≤0+a2+⋯+an≤∫1nf(x)dx
即
∑ n = 1 ∞ a n = ∫ 1 ∞ f ( x ) d x , a 1 = 0 \sum_{n=1}^{\infty}{a_{n}}=\int_{1}^{\infty}f(x)dx,a_{1}=0 n=1∑∞an=∫1∞f(x)dx,a1=0
敛散性一致,同时收敛或同时发散。
因此, a 1 = 0 a_{1}=0 a1=0 时, ∑ n = 1 ∞ a n = ∫ 1 ∞ f ( x ) d x \sum_{n=1}^{\infty}{a_{n}}=\int_{1}^{\infty}f(x)dx n=1∑∞an=∫1∞f(x)dx
(从 n = 2 n=2 n=2开始递减)
这就证明了:
∣ − ζ ( k ) ( 2 ) ∣ = ∑ n = 1 ∞ ( ln n ) k n 2 = ∫ 1 ∞ ( ln x ) k x 2 d x = k ! |-\zeta^{(k)}(2)|=\sum_{n=1}^{\infty}{\frac{(\ln n)^{k}}{n^2}}=\int_{1}^{\infty}\frac{(\ln x)^{k}}{x^2}dx=k! ∣−ζ(k)(2)∣=n=1∑∞n2(lnn)k=∫1∞x2(lnx)kdx=k!
证毕
同理得到: − ζ ′ ( s ) = ∑ n = 1 ∞ ln n n s = ∫ 1 ∞ ln x x s d x = 1 ( s − 1 ) 2 , s > 1 -\zeta'(s)=\sum_{n=1}^{\infty}{\frac{\ln n}{n^{s}}}=\int_{1}^{\infty}\frac{\ln x}{x^{s}}dx=\frac{1}{(s-1)^{2}},s>1 −ζ′(s)=n=1∑∞nslnn=∫1∞xslnxdx=(s−1)21,s>1
ζ ( 2 ) = ∑ n = 1 ∞ 1 n 2 = ∫ 0 ∞ x e x − 1 d x = π 2 6 \zeta(2)=\sum_{n=1}^{\infty}{\frac{1}{n^2}}=\int_{0}^{\infty}\frac{x}{e^{x}-1}dx=\frac{\pi^{2}}{6} ζ(2)=n=1∑∞n21=∫0∞ex−1xdx=6π2
1 e x − 1 = ∑ n = 1 ∞ 1 e n x \frac{1}{e^x-1}=\sum_{n=1}^{\infty}{\frac{1}{e^{nx}}} ex−11=n=1∑∞enx1
∫ 0 ∞ x e x − 1 d x = ∫ 0 ∞ ∑ n = 1 ∞ x e n x d x \int_{0}^{\infty}\frac{x}{e^{x}-1}dx=\int_{0}^{\infty}\sum_{n=1}^{\infty}\frac{x}{e^{nx}}dx ∫0∞ex−1xdx=∫0∞n=1∑∞enxxdx
换元,令 x = ln y x=\ln y x=lny
∫ 0 ∞ x e x − 1 d x = ∫ 1 ∞ ∑ n = 1 ∞ ln y e n ln y d ( ln y ) = ∫ 1 ∞ ∑ n = 1 ∞ ln y y n + 1 d y \int_{0}^{\infty}\frac{x}{e^{x}-1}dx=\int_{1}^{\infty}\sum_{n=1}^{\infty}\frac{\ln y}{e^{n\ln y}}d(\ln y)=\int_{1}^{\infty}\sum_{n=1}^{\infty}{\frac{\ln y}{y^{n+1}}}dy ∫0∞ex−1xdx=∫1∞n=1∑∞enlnylnyd(lny)=∫1∞n=1∑∞yn+1lnydy
ζ ( 2 ) = ∑ n = 1 ∞ 1 n 2 = ∫ 1 ∞ ∑ n = 1 ∞ ln y y n + 1 d y = ∑ n = 1 ∞ ∫ 1 ∞ ln y y n + 1 d y \zeta(2)=\sum_{n=1}^{\infty}{\frac{1}{n^2}}=\int_{1}^{\infty}\sum_{n=1}^{\infty}{}\frac{\ln y}{y^{n+1}}dy=\sum_{n=1}^{\infty}{\int_{1}^{\infty}}\frac{\ln y}{y^{n+1}}dy ζ(2)=n=1∑∞n21=∫1∞n=1∑∞yn+1lnydy=n=1∑∞∫1∞yn+1lnydy
又因为, ∫ 1 ∞ ln y y n + 1 d y = 1 n 2 , ∫ 1 ∞ 1 n 2 d n = 1 \int_{1}^{\infty}\frac{\ln y}{y^{n+1}}dy=\frac{1}{n^2} , \int_{1}^{\infty}\frac{1}{n^2}dn=1 ∫1∞yn+1lnydy=n21,∫1∞n21dn=1
所以有
∑ y = 1 ∞ ln y y k + 1 = 1 k 2 , \sum_{y=1}^{\infty}{}\frac{\ln y}{y^{k+1}}=\frac{1}{k^2} , y=1∑∞yk+1lny=k21,
k = 1 k=1 k=1时, − ζ ′ ( 2 ) = ∑ n = 1 ∞ ln n n 2 = 1 -\zeta'(2)=\sum_{n=1}^{\infty}{}\frac{\ln n}{n^{2}}=1 −ζ′(2)=n=1∑∞n2lnn=1
− ζ ′ ( s ) = ∑ n = 1 ∞ ln n n s = ∫ 1 ∞ ln x x s d x = 1 ( s − 1 ) 2 , s > 1 -\zeta'(s)=\sum_{n=1}^{\infty}{\frac{\ln n}{n^{s}}}=\int_{1}^{\infty}\frac{\ln x}{x^{s}}dx=\frac{1}{(s-1)^{2}},s>1 −ζ′(s)=n=1∑∞nslnn=∫1∞xslnxdx=(s−1)21,s>1
− ζ ′ ( 2 ) = ( − ∑ n = 1 ∞ 1 n 2 ) ′ = ∑ n = 1 ∞ ln n n 2 -\zeta'(2)=\left(-\sum_{n=1}^{\infty}{\frac{1}{n^2}} \right)'=\sum_{n=1}^{\infty}{}\frac{\ln n}{n^{2}} −ζ′(2)=(−n=1∑∞n21)′=n=1∑∞n2lnn
∑ n = 1 ∞ 1 n 2 = ∫ 0 ∞ x e x − 1 d x \sum_{n=1}^{\infty}{\frac{1}{n^2}} =\int_{0}^{\infty}\frac{x}{e^{x}-1}dx n=1∑∞n21=∫0∞ex−1xdx
定积分的先积后导与先导后积是一样的结果
( − ∑ n = 1 ∞ 1 n 2 ) ′ = − ∫ 0 ∞ ( x e x − 1 ) ′ d x = ∫ 0 ∞ ( x − 1 ) e x + 1 ( e x − 1 ) 2 d x = 1 \left(-\sum_{n=1}^{\infty}{\frac{1}{n^2}} \right)'=-\int_{0}^{\infty}\left(\frac{x}{e^{x}-1} \right)'dx=\int_{0}^{\infty}\frac{(x-1)e^x+1}{(e^{x}-1)^2} dx=1 (−n=1∑∞n21)′=−∫0∞(ex−1x)′dx=∫0∞(ex−1)2(x−1)ex+1dx=1
所以有: ∑ n = 1 ∞ ln n n 2 = ln 2 2 2 + ln 3 3 2 + ln 4 4 2 + ⋯ + ln n n 2 + ⋯ = 1 \sum_{n=1}^{\infty}{}\frac{\ln n}{n^{2}}=\frac{\ln2}{2^{2}}+\frac{\ln3}{3^{2}}+\frac{\ln4}{4^{2}}+\cdots+\frac{\ln n}{n^{2}}+\cdots=1 n=1∑∞n2lnn=22ln2+32ln3+42ln4+⋯+n2lnn+⋯=1