0% found this document useful (0 votes)
35 views

Frank Thorne, University of South Carolina: Dartmouth College

The document describes an elementary method for computing values of the Riemann zeta function ζ(s) at negative integers. It involves defining polynomials Pn(t) whose integrals over [0,1] are zero, and relating their coefficients to values of ζ(s). This allows computing values like ζ(-1)=−1/12, ζ(-2)=0, and ζ(-3)=1/120 without tools like Poisson summation. The method also applies to finite sums.
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)
35 views

Frank Thorne, University of South Carolina: Dartmouth College

The document describes an elementary method for computing values of the Riemann zeta function ζ(s) at negative integers. It involves defining polynomials Pn(t) whose integrals over [0,1] are zero, and relating their coefficients to values of ζ(s). This allows computing values like ζ(-1)=−1/12, ζ(-2)=0, and ζ(-3)=1/120 without tools like Poisson summation. The method also applies to finite sums.
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/ 59

1 + 2 + 3 + 4 + ···

Frank Thorne, University of South Carolina

Dartmouth College

May 9, 2013

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Introduction.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Srinivasa Ramanujan (1887-1920)

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Ramanujan’s second letter to Hardy

“Dear Sir, I am very much gratified on perusing your letter of


the 8th February 1913. I was expecting a reply from you
similar to the one which a Mathematics Professor at London
wrote asking me to study carefully Bromwich’s Infinite Series
and not fall into the pitfalls of divergent series. I told him
that the sum of an infinite number of terms of the series:
1 + 2 + 3 + 4 + · · · = −1/12 under my theory. If I tell you this
you will at once point out to me the lunatic asylum as my
goal. I dilate on this simply to convince you that you will not
be able to follow my methods of proof if I indicate the lines on
which I proceed in a single letter. . . . ”

(S. Ramanujan, 27 February 1913)

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Warmup.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Ramanujan’s proof

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Q.E.D.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Q.E.D.

“The divergent series are the invention of the devil, and


it is a shame to base on them any demonstration
whatsoever.”

(N. Abel, 1832)

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


The Riemann zeta function.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Analytic continuation

Theorem (Riemann, 1859)


The zeta function has analytic continuation to all complex
numbers s 6= 1, with
1−s − 1−s

Γ 2 π
2
ζ(s) = ζ(1 − s) s
 −s .
Γ 2 π 2

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Analytic continuation

Theorem (Riemann, 1859)


The zeta function has analytic continuation to all complex
numbers s 6= 1, with
1−s − 1−s

Γ 2 π
2
ζ(s) = ζ(1 − s) s
 −s .
Γ 2 π 2

Therefore,

Γ(1)π −1 π2 1 × π −1 1
ζ(−1) = ζ(2) 1
= · √ =− .
Γ(− 2 )π 1/2 6 (−2 π)π 1/2 12

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Poisson summation

The usual proof is by Poisson summation.


When you first see it, it looks like a piece of magic.

(anonymous, MathOverflow comment)

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Poisson summation

The usual proof is by Poisson summation.


When you first see it, it looks like a piece of magic.

(anonymous, MathOverflow comment)


1
Can compute ζ(−1) = − 12 using elementary methods?

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Integration by parts.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


It keeps going...
Defining P3 (t) in the same manner, where
Z 1
P3 (t)dt = 0,
0

we have (on [0, 1])

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


It keeps going...
Defining P3 (t) in the same manner, where
Z 1
P3 (t)dt = 0,
0

we have (on [0, 1])

1 1 1
P3 (t) = t 3 − t 2 + t,
6 4 12

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


It keeps going...
Defining P3 (t) in the same manner, where
Z 1
P3 (t)dt = 0,
0

we have (on [0, 1])

1 1 1
P3 (t) = t 3 − t 2 + t,
6 4 12
similarly
1 4 1 1 1
P4 (t) = t − t3 + t2 − ,
24 12 24 720

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


It keeps going...
Defining P3 (t) in the same manner, where
Z 1
P3 (t)dt = 0,
0

we have (on [0, 1])

1 1 1
P3 (t) = t 3 − t 2 + t,
6 4 12
similarly
1 4 1 1 1
P4 (t) = t − t3 + t2 − ,
24 12 24 720
so
Z ∞
s 1 s s(s + 1)(s + 2) P4 (t)
ζ(s) = − + − −s(s +1)(s +2) .
s − 1 2 12 720 1 t s+4

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


1 + 8 + 27 + 64 + ...

Z ∞
s 1 s s(s + 1)(s + 2) P4 (t)
ζ(s) = − + − −s(s +1)(s +2) .
s − 1 2 12 720 1 t s+4

This implies that

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


1 + 8 + 27 + 64 + ...

Z ∞
s 1 s s(s + 1)(s + 2) P4 (t)
ζ(s) = − + − −s(s +1)(s +2) .
s − 1 2 12 720 1 t s+4

This implies that

ζ(−2) = 1 + 4 + 9 + 16 + 25 + · · · = 0,

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


1 + 8 + 27 + 64 + ...

Z ∞
s 1 s s(s + 1)(s + 2) P4 (t)
ζ(s) = − + − −s(s +1)(s +2) .
s − 1 2 12 720 1 t s+4

This implies that

ζ(−2) = 1 + 4 + 9 + 16 + 25 + · · · = 0,

1
ζ(−3) = 1 + 8 + 27 + 64 + 125 + · · · = .
120

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


1 + 8 + 27 + 64 + ...

Z ∞
s 1 s s(s + 1)(s + 2) P4 (t)
ζ(s) = − + − −s(s +1)(s +2) .
s − 1 2 12 720 1 t s+4

This implies that

ζ(−2) = 1 + 4 + 9 + 16 + 25 + · · · = 0,

1
ζ(−3) = 1 + 8 + 27 + 64 + 125 + · · · = .
120
and we can compute any value of ζ(−n) similarly.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A variant on analytic continuation of ζ(s)

This also works for finite sums.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A variant on analytic continuation of ζ(s)

This also works for finite sums. For example:


N−1
X N 1−s 1 1
n−s = ζ(s) + − N −s − sN −s−1 + Os (N −s−2 ).
1−s 2 12
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A variant on analytic continuation of ζ(s)

This also works for finite sums. For example:


N−1
X N 1−s 1 1
n−s = ζ(s) + − N −s − sN −s−1 + Os (N −s−2 ).
1−s 2 12
n=1

Taking s = −1,

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A variant on analytic continuation of ζ(s)

This also works for finite sums. For example:


N−1
X N 1−s 1 1
n−s = ζ(s) + − N −s − sN −s−1 + Os (N −s−2 ).
1−s 2 12
n=1

Taking s = −1,
N−1
X N2 N 1
n = ζ(−1) + − + + O(N −1 ).
2 2 12
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A variant on analytic continuation of ζ(s)

This also works for finite sums. For example:


N−1
X N 1−s 1 1
n−s = ζ(s) + − N −s − sN −s−1 + Os (N −s−2 ).
1−s 2 12
n=1

Taking s = −1,
N−1
X N2 N 1
n = ζ(−1) + − + + O(N −1 ).
2 2 12
n=1

1
We see again that ζ(−1) = − 12 .

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Some standard terminology

I The polynomial Bn (t) := n!Pn (t) is called the nth Bernoulli


polynomial;

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Some standard terminology

I The polynomial Bn (t) := n!Pn (t) is called the nth Bernoulli


polynomial;
I The constant term Bn := Bn (0) is called the nth Bernoulli
number.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Some standard terminology

I The polynomial Bn (t) := n!Pn (t) is called the nth Bernoulli


polynomial;
I The constant term Bn := Bn (0) is called the nth Bernoulli
number.
I If n is odd, then Bn = 0 (except B1 = − 12 ).

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···
Some Bernoulli polynomials

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


The Euler-Maclaurin sum formula

Theorem
If f ∈ C ∞ [0, ∞), then for all integers a, b, k we have
b Z b
X 1 
f (n) = f (t)dt + f (a) + f (b)
n=a a 2
k
X Bk  (k−1) 
+ f (b) − f (k−1) (a)
k!
`=2
Z b
1
+ Bk (x)f (k) (t)dt.
k! a

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Example

Stirling’s formula: Take f (n) = log(n):


x Z x
X 1
log(x!) = log n → log t dt + C + log x.
1 2
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Euler-Maclaurin: a special case

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


The Ramanujan constant CR
We have
x Z x ∞
X 1 X Bk
f (n) = f (t)dt + CR + f (x) + f (k−1) (x),
0 2 k!
n=1 k=2

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


The Ramanujan constant CR
We have
x Z x ∞
X 1 X Bk
f (n) = f (t)dt + CR + f (x) + f (k−1) (x),
0 2 k!
n=1 k=2
where the Ramanujan constant CR is defined by

1 X Bk (k−1)
CR = − f (0) − f (0).
2 k!
k=2

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


The Ramanujan constant CR
We have
x Z x ∞
X 1 X Bk
f (n) = f (t)dt + CR + f (x) + f (k−1) (x),
0 2 k!
n=1 k=2
where the Ramanujan constant CR is defined by

1 X Bk (k−1)
CR = − f (0) − f (0).
2 k!
k=2

The Ramanujan constant of ∞ 1


P
n=1 1 is − 2 , because
x Z x
X 1
1= 1 dt + CR + · 1.
0 2
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


The Ramanujan constant CR
We have
x Z x ∞
X 1 X Bk
f (n) = f (t)dt + CR + f (x) + f (k−1) (x),
0 2 k!
n=1 k=2
where the Ramanujan constant CR is defined by

1 X Bk (k−1)
CR = − f (0) − f (0).
2 k!
k=2

The Ramanujan constant of ∞ 1


P
n=1 1 is − 2 , because
x Z x
X 1
1= 1 dt + CR + · 1.
0 2
n=1

The Ramanujan constant of ∞ 1


P
n=1 n is − 12 , because
x Z x
X 1 1
n= t dt + CR + n + .
0 2 12
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A broader definition of Ramanujan sums?

Definition P∞
(???) We define the value of any infinite sum n=1 f (n) to be the
Ramanujan constant CR .

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum

Consider

X 1 π2
= − 1.
(n + 1)2 6
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum

Consider

X 1 π2
= − 1.
(n + 1)2 6
n=1

The Ramanujan constant is



1 X B2k
CR = − + · (2k)!
2 (2k)!
k=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum

Consider

X 1 π2
= − 1.
(n + 1)2 6
n=1

The Ramanujan constant is



1 X B2k
CR = − + · (2k)!
2 (2k)!
k=1

Can we speed up the convergence?

Warning: I am lying on this slide.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum (cont.)

Consider instead

X 1 π2  1 1 1 1
= − 1 + + + + .
(n + 5)2 6 4 9 16 25
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum (cont.)

Consider instead

X 1 π2  1 1 1 1
= − 1 + + + + .
(n + 5)2 6 4 9 16 25
n=1

Now the Ramanujan constant is



1 1 X B2k (2k)!
CR = − · + ·
2 25 (2k)! 52k+1
k=1
1 1 1 1
=− + − + − ···
50 750 93750 3281250
= −0.018677028 . . .

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum (cont.)

1 1 1
36 + 49 + 64 + · · · is not −0.018677028.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum (cont.)

1 1 1
36 + 49 + 64 + · · · is not −0.018677028. However,
∞ Z ∞
X 1 1
2
≈ −0.018677028+ dt = 0.181322971 . . .
(n + 5) 0 (t + 5)2
n=1

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum (cont.)

1 1 1
36 + 49 + 64 + · · · is not −0.018677028. However,
∞ Z ∞
X 1 1
2
≈ −0.018677028+ dt = 0.181322971 . . .
(n + 5) 0 (t + 5)2
n=1

and
X 1 π2  1 1 1 1
= − 1 + + + +
(n + 5)2 6 4 9 16 25
n=1
= 0.181322955.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


A convergent sum (cont.)

1 1 1
36 + 49 + 64 + · · · is not −0.018677028. However,
∞ Z ∞
X 1 1
2
≈ −0.018677028+ dt = 0.181322971 . . .
(n + 5) 0 (t + 5)2
n=1

and
X 1 π2  1 1 1 1
= − 1 + + + +
(n + 5)2 6 4 9 16 25
n=1
= 0.181322955.

π2
This calculation convinced Euler that ζ(2) = 6 .

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Rate of convergence

Question: Does the infinite series



1 1 X B2k (2k)!
CR = − · + ·
2 25 (2k)! 52k+1
k=1

converge?

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Rate of convergence

Question: Does the infinite series



1 1 X B2k (2k)!
CR = − · + ·
2 25 (2k)! 52k+1
k=1

converge?
No,
2(2n)!
|B2n | ∼ ,
(2π)2n

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Rate of convergence

Question: Does the infinite series



1 1 X B2k (2k)!
CR = − · + ·
2 25 (2k)! 52k+1
k=1

converge?
No,
2(2n)!
|B2n | ∼ ,
(2π)2n
but this can be fixed rigorously.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


How to get the correct constant?

P∞ π2
The ‘Ramanujan sum’ n=1 is not equal to 6 − 1.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


How to get the correct constant?

P∞ π2
The ‘Ramanujan sum’ n=1 is not equal to 6 − 1.
Hardy: Introduce another parameter a.
“The introduction of the parameter a allows more
flexibility and enables one to always obtain the “correct”
constant; usually, there is a certainP
value of a which is
more natural than other values. If f (k) converges,
then normally we would take a = ∞. Although the
concept of the constant of a series has been made
precise, Ramanujan’s concomitant theory cannot always
be made rigorous.”

(B. Berndt)

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Other connections

I “Ramanujan constant” in terms of eigenvalues of a shift


operator (Venkatesh).

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Other connections

I “Ramanujan constant” in terms of eigenvalues of a shift


operator (Venkatesh).
I Congruences for Bernoulli numbers (Kummer, von
Staudt-Clausen).

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Other connections

I “Ramanujan constant” in terms of eigenvalues of a shift


operator (Venkatesh).
I Congruences for Bernoulli numbers (Kummer, von
Staudt-Clausen).
I p-adic zeta functions (Kubota-Leopoldt).

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Other connections

I “Ramanujan constant” in terms of eigenvalues of a shift


operator (Venkatesh).
I Congruences for Bernoulli numbers (Kummer, von
Staudt-Clausen).
I p-adic zeta functions (Kubota-Leopoldt).
I Many cases of Fermat’s Last Theorem.

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···


Other connections

I “Ramanujan constant” in terms of eigenvalues of a shift


operator (Venkatesh).
I Congruences for Bernoulli numbers (Kummer, von
Staudt-Clausen).
I p-adic zeta functions (Kubota-Leopoldt).
I Many cases of Fermat’s Last Theorem.
I ... and more ...

Frank Thorne, University of South Carolina 1 + 2 + 3 + 4 + ···

You might also like