0% found this document useful (0 votes)
49 views15 pages

Apery

Uploaded by

Me Shaad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
49 views15 pages

Apery

Uploaded by

Me Shaad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

数理解析方法論

Irrationality of ζ(3)
1.Riemann’s zeta function –1–
Riemann’s zeta function
∑∞
1
ζ(s) =
n=1
ns
converges in ℜs > 1.
Definition The Bernoulli numbers Bn are defined by the power series expan-
sion ∞ ∞
x ∑ xn x ∑ x2n
= Bn = 1 − + B2n
ex − 1 n=0 n! 2 n=1 (2n)!
x x
Remark. x + is an even function.
e −1 2
The first few Bernoulli numbers:
B1 = −1/2, B2 = 1/6, B4 = −1/30, B6 = 1/42

Lemma 0 ∞ ( )
∏ x2
sin πx = πx 1− 2
n=1
n
Theorem (Euler) For n = 1, 2, 3, ..., –2–
(−1)n−1 22n−1
ζ(2n) = B2n π 2n
(2n)!
when n is an integer ≥ 1.
Proof. By the definition of the Bernoulli numbers,
∑∞
eix + e−ix 2ix 22n x2n
x cot x = ix ix = ix + 2ix =1− n+1
(−1) B2n
e − e−ix e −1 n=1
(2n)!

and Lemma 0 gives the expansion



∑ ∑∞ ∑∞
x2 x2n
x cot x = 1 + 2 =1−2
x2 − k 2 π 2 n=1
k 2n π 2n
k=1 k=1

Now equate the coefficients of x2n in the two series for x cot x to complete the
proof.

Remark This showed that ζ(2n) is irrational.

In 1978 R. Apéry proved that ζ(3) is irrational.


2. Criterion of irrationality –3–
p0
A number α is irrational if it is not of the form ; where p0 , q0 are integers.
q0
A rational number β is characterized by the property that for p, q ∈ Z (q > 0)
p
and β ̸= , there exists an integer q0 > 0 such that
q
p 1
β− ≥ .
q qq0
On the other hand, for an irrational number α there are always infinitely many
p
such that
q
p 1
α − < 2.
q q
Plainly this yields a criterion for irrationality.
Proposition
{ } (criterion for irrationality) It there is a δ > 0 and a sequence
pn pn
of rational numbers such that ̸= α and
qn qn
pn 1
α− < 1+δ , n = 1, 2, ...
qn qn
then α is irrational.
3. Apéry’s Theorem –4–
Lemma 1 There exist two sequences of integers {An } and {Bn } such that
( )n
9
0 < |An + Bn ζ(3)| < 3 .
10
This immediately implies the theorem.

Theorem [Apéry] ζ(3) is irrational.

Proof. If ζ(3) = p/q, where p and q are integers, then the sequence of nonzero
rational numbers |An + Bn ζ(3)| ≥ 1/q.
The second inequality in Lemma 1, however, implies that this sequence be-
comes arbitrarily small as n increases. This contradiction implies that ζ(3) is
irrational.

The proof of Lemma 1 depends on the following lemmas [Beukers].


Lemma 2 Assume ∫ 1that
∫ 1 r and s are nonnegative integers. –5–
log xy r s
(1) When r > s, − x y dxdy is a rational number whose denom-
0 0 1 − xy
inator divides d3r . Here dr = lcm(1, 2...., r).
(2) { }
∫ 1∫ 1 ∑ r
log xy r r 1
− x y dxdy = 2 ζ(3) −
0 0 1 − xy k3
k=1
Proof. (1) Note that for σ > 0 and r > s,
∫ 1 ∫ 1 r+σ s+σ ∫ 1∫ 1∑ ∞
x y
dxdy = xr+σ+n y s+σ+n dxdy
0 0 1 − xy 0 0 n=0

∑ 1
=
n=0
(n + r + σ + 1)(n + s + σ + 1)
∞ { }
1 ∑ 1 1
= −
r − s n=0 n + s + σ + 1 n + r + σ + 1
{ }
1 1 1
= + ··· + .
r−s s+σ+1 r+σ
Differentiate –6–
∫ 1∫ 1 { }
xr+σ y s+σ 1 1 1
dxdy = + ··· +
0 0 1 − xy r−s s+σ+1 r+σ
with respect to σ. Set σ = 0:
∫ 1∫ 1 { }
log xy r s 1 1 1
x y dxdy = − + ··· + 2
0 0 1 − xy r − s (s + 1)2 r
Thus (1) is proved.
(2) Differentiate
∫ 1∫ 1 ∑∞
xr+σ y r+σ 1
dxdy = .
0 0 1 − xy n=0
(n + r + σ + 1) 2

with respect to σ. Set σ = 0:


∫ 1∫ 1 ∞
{ }
log xy r r ∑ 2 ∑
r
1
x y dxdy = − = −2 ζ(3) − .
0 0 1 − xy n=0
(n + r + 1) 3 k 3
k=1
4. Integral of the Legendre polynomials –7–
Let
1 dn
pn (x) = n
{xn (1 − x)n } .
n! dx
which is essentially the Legendre polynomial on (0, 1).
Lemma 3 There exist integers An and Bn such that
∫ 1∫ 1
log xy
0 ̸= − pn (x)pn (y) dxdy = (An + Bn ζ(3))d−3
n .
0 0 1 − xy
Here dn = lcm(1, 2.....n).
Proof.
Since pn (x) is a polynomial with integer coefficients, the equality follows
from Lemma 2. Now observe that
∫ 1
log xy 1
− = dz.
1 − xy 0 1 − (1 − xy)z

Then the integral in the lemma can be written as


∫ 1∫ 1∫ 1
pn (x)pn (y)
dxdydz.
0 0 0 1 − (1 − xy)z
5. Triple Integral –8–
∫ 1∫ 1∫ 1
pn (x)pn (y)
dxdydz.
0 0 0 1 − (1 − xy)z
Integrate by parts n times with respect to x to see that this integral is equal
to ∫ 1∫ 1∫ 1
(xyz)n (1 − x)n pn (y)
dxdydz.
0 0 0 (1 − (1 − xy)z)n+1
Set
1−z
w=
1 − (1 − xy)z
and rewrite the last integral as
∫ 1∫ 1∫ 1
pn (y)
(1 − x)n (1 − w)n dxdydw
0 0 0 1 − (1 − xy)w
∫ 1∫ 1∫ 1
(x(1 − x)y(1 − y)w(1 − w))n
= dxdydw
0 0 0 (1 − (1 − xy)w)n+1
The last equality uses integration by parts n times. It is clear that the final
integral is nonzero and so the lemma is proved.
Lemma 4 For An and Bn as in Lemma 3, –9–

0 < |An + Bn ζ(3)|d−3
n < 2( 2 − 1)4n ζ(3).

Proof. The integral in Lemma 3 is equal to the integral


∫ 1∫ 1∫ 1
(x(1 − x)y(1 − y)w(1 − w))n
I= dxdydw.
0 0 0 (1 − (1 − xy)w)n+1
We find the maximum value of the integrand in I = |An + Bn ζ(3)|. Let
x(1 − x)y(1 − y)w(1 − w)
f (x, y, w) =
1 − (1 − xy)w
By solving the equations
∂f ∂f ∂f
= = = 0,
∂x ∂y ∂w
1
it is easy to see that at the maximum, x = y and w = . Thus f is
1+x
x2 (1 − x)2 √
bounded by , which is maximized at x = 2 − 1.
(1 + x)2
Thus integral I is bounded above by –10–
√ ∫ 1∫ 1∫ 1
1
( 2 − 1) 4n
dxdydz
0 0 0 1 − (1 − xy)w
√ ∫ 1∫ 1
log xy
= ( 2 − 1) 4n
− dxdy
1 − xy
√ 0 0
= 2( 2 − 1)4n ζ(3)

This proves the lemma.

Proposition (Tchebyshev) [Weak version of the prime number theorem]


Let π(n) denote the number of primes less than n. Then
n
π(n) < 1.06 .
log n
6. Proof of Lemma 1 –11–
Observe that
∏ ∏
dn = p[log n/ log p] < plog n/ log p = nπ(n) .
p≤n,p=prime p≤n,p=prime

By Tchebyshev’s proposition, it follows that

dn < n1.06n/ log n < e1.06n < 3n .

An upper bound for ζ(3) is given by


∫ ∞
dx 3
ζ(3) < 1 + = .
1 x3 2
Therefore, Lemma 4 gives
( )n ( )n
27 9
0 < |An + Bn ζ(3)| < 3 √ <3 .
( 2 + 1)4 10

Remark. ζ(3) = 1.202056903 · · ·


Exercise –12–
Assume that r and ∫ 1s∫are nonnegative integers.
1
xr y s
(1) When r > s, dxdy is a rational number whose denominator
0 0 1 − xy
divides d2r . Here dr = lcm(1, 2.....r).
(2)
∫ 1∫ 1 r r ∑r
xy 1
dxdy = ζ(2) −
0 0 1 − xy k2
k=1
(3) Let
1 dn
pn (x) = n
{xn (1 − x)n } .
n! dx
Consider the integral
∫ 1∫ 1
(1 − y)n pn (x)
I= dxdy
0 0 1 − xy
Show that
I = (Cn + Dn ζ(2))d−2
n

for some integers Cn , Dn .


(4) Show that
∫ 1∫
n
1
y n (1 − y)n xn (1 − x)n
I = (−1) dxdy
0 0 (1 − xy)n+1
and that (√ )5
y(1 − y)x(1 − x) 5−1
≤ ,
1 − xy 2
for all 0 ≤ x ≤ 1, 0 ≤ y ≤ 1.
(6) Show that
(√ )5n ( )n
5−1 5
0 < |Cn + Dn ζ(2)| < d2n ζ(2) <
2 6

(7) Prove that ζ(2) is irrational.

The dead line of report is 1st February 2018


References –14–
[A] R. Apéry, Irrationalité de ζ(2) et ζ(3), Astérisque 61 (1979), 11–13.
[B] F. Beukers, A note on the irrationality of ζ(2) and ζ(3), Bull. London
Math. Soc. 11 (1979), 268–272.
[vdP] A. van der Poorten, A proof that Euler missed...Apéry’s proof of the
irrationality of ζ(3), The Mathematical Intelligencer, 1(1979), 195–203,

You might also like