ZETA REGULARIZATION APPLIED TO THE PROBLEM OF RIEMANN

HYPOTHESIS AND THE CALCULATION OF DIVERGENT INTEGRALS

Jose Javier Garcia Moreta

Graduate student of Physics at the UPV/EHU (University of Basque country)
In Solid State Physics
Address: Address: Practicantes Adan y Grijalba 2 5 G
P.O 644 48920 Portugalete Vizcaya (Spain)
Phone: (00) 34 685 77 16 53
E-mail: josegarc2002@yahoo.es

MSC: 45G05, 47H30 , 03.65.-w , 11.10.Gh

ABSTRACT: In this paper we review some results of our previous papers involving Riemann
Hypothesis in the sense of Operator theory (Hilbert-Polya approach) and the application of the

 (1  s ) x
s 1
negative values of the Zeta function to the divergent integrals dx and to the
0

problem of defining a consistent product of distributions of the form D m ( x) D n ( x) , in this

paper we present new results of how the sums over the non-trivial zeros of the zeta function
 h(  ) can be related to the Mangoldt function  0 ( x) assuming Riemann Hypothesis.

 Keywords:Zeta regularization, Urysohn equation , exponential nonlinearity , Riemann

Hypothesis Hilbert-Polya operator, divergent integral

1.Spectral Zeta function  H ( s) and Riemann Hypothesis :

In case Riemann Hypothesis (RH) is true, in a previous paper [6] we give the physical
equivalence between the explicit formula for the Chebyshev function  0 ( x) and the
formula for the trace of the Unitary operator Uˆ  eiuH , where H is the Hamiltonian
ˆ

1 
operator    iHˆ  n  0 , that is H is precisely the Hilbert-Polya operator solution to
2 
Riemann Hypothesis , let be the integral representation

 x   (0) 1
1 ci  ( s ) x s  x     log(1  x 2 ) x  1
2 i ci  ( s ) s 
 0 ( x)   ds      (0) 2 (1.1)
 0 x 1

1
Letting x  eu , and differentitating with respect to ‘u’ we find the (trace) identity

d  0 (eu )
 

eu / 2
eu / 2  e  u / 2  3u u   eiuEn  Tr Uˆ  eiuH
ˆ
u >0 (1.2)
du e e n 


Using the semiclassical representation for the trace e
n 
iuEn
in terms of an integral over

Phase Space , we have that the potential V(x) inside Hamiltonian H can not be arbitrary
but must satisfy a kind of nonlinear Urysohn integral equation ( r > 1)


log(r )  d  0 (r ) 1   i / 4
r dx  1  3 r  eu
iV ( x )
1  e (1.3)

r  dr r r 

d  0 ( x) 1
The derivative of the Chebyshev function is defined as    ( x  p )
dx log( x) p , p
(sum taken over prime and prime powers ) .However (1.3) is too complex to have a
known analytic solution, a good method to solve would be to suppose that the Operator
proposed by Berry and Keating [2] plus an interaction is the correct Hilbert-Polya
operator, in that case H b  xp  W ( x) and we can linearize (1.3) at first order in the
coupling constant ‘  ’ as

  2  
Tr eiuH   iu  dpFˆ W ( x), u Fˆ W ( x), u   dxeiuxpW ( x)
ˆ
(1.4)
|u |  


Also, if we introduce the function Z u     dxe
iu (V ( x )  x )
, with continuos partial


derivatives  k Z u ( ) , then solving (1.3) is equivalent to finding a solution to the initial-

value problem

Z u ( )   k k 
Z u ( )    iu   d k Z ( )  0
k  u
  k 0 (iu )  
k

(1.5)

 u / 2  u / 2 d  0 (e ) u
e  u u/2
i
e e  3u u   Z u (0)
4
e
 du e e  

Expression (1.8) tells us that proving RH is equivalent to show that the ODE given in

1 d kV ( x)
(1.5) with d k   R and d k 
k ! dx k x 0 , V ( x )  
k 0
d k x k using (1.5) together with

a finite power expansion for V(x) , using (1.5) we could obtain the constants d k   R
to get an approximate solution for the potential V(x).

2
 1 
If RH is true and    iEn   0 , with En   E n being the eigenvalues of a certain
2 
operator H  p  V ( x) , using expression (1.2) and the functional equation
2

s 
 (1  s )  2(2 ) s Cos   ( s ) ( s) ,then for n  0 we can define an spectral Zeta
 2 
function , involving the nontrivial zeros of Zeta and primes and prime powers

s 
Sec   
 s  (s) t/2  1 
   t d  0 (e )
 t
1  2  
 s

2 ( s )  dtTr e itHˆ
  2  
 (1  s) 0
dte 1  e  3t t  t s 1
e e 
n 0 En
0
 dx
(1.6)

 d  (0)

The value E
n 0
n e ds
would be the regularized product of all the positive

‘Eigenvalues’  En  this expression can also be used to obtain a Dirac measure for the
En , let us introduce

 
  1  1   s1  (s) 

 En
s
  dt     t    t   dtK 0 (t )t s1
 (1  s ) 0
(1.7)
n 0 0  n0 En  En  

Using the properties of the Mellin transform applied to solve linear integral operators

I [ f ]   dtR ( xt ) f (t ) , if we combine (1.6) and (1.7) we get the result
0


1 
1  dt  e 1/ 2t d  0 (e1/ t ) e1/ 2t 
( x)     x     K 0 (2 xt )  e1/ 2t  2  3/ t 1/ t  (1.8)
n 0 En  En  0 t  t dt e e 


If we took the Mellin transform  dxx s 1 inside (1.8) together with the change of
0
variable xt=z we would recover equation (1.6) , note that the Mellin transform of the
1
Kernel K 0 (2 xt ) does not depend on the nontrivial zeros    it .
2

1 1 i
Using test functions h   inside (1.8) obtained from our Trace formula for
x 2 x
  we can relate the convergent sum  h(  ) to a sum over primes and prime
Tr eiuH
ˆ


powers

3
1
1 i 

K 0  2 xt   1/ 2 t e1/ 2 t d  0 (e1/ t ) e1/ 2 t 
0 dxh  2  x  0 dt xt  e  t 2  3/ t 1/ t   c.c   h(  )
dt e e  

(1.9)

a
Formula (1.9) and its result can be compared with sums   (explicit formula for
1
Chebyshev function) and Z (n)   n  N , that can be calculated exactly.
 n

  ˆ
o The Trace Tr eiuH and the sum  h( )

Even though we can not solve equation (1.3) we can use the Trace expression (1.2) to
find stimates for sums  h( ) . First we define a couple of function g(x) and h(x) with

the following properties

 Both g(x)=g(-x) and h(x)=h(-x) are even functions

g ( x)
 lim exists and it is finite
x 0 x
 The functions h(x) and g(x) are related by a Fourier Cosine transorm

1
 0
dxh( x)Cos ( x)  g ( )

 The function h(x) can be defined by analytic continuation to the region of

complex plane defined by Re (s) =0 , in particular h(i / 2) 1  i

If RH (Riemann Hypothesis) is true, then the Trace (1.3) is just a sum of cosines
 2cos( u ) , then if we take g(u) as a test function
 0

 

0
 
 dug (u )Tr e 
iuH ˆ h(i / 2)  h(i / 2)   (n)
2

n 1 n 0
e
g (u )
g (log n)   due  u / 2 2u
 1
 2 h( )
 0

(1.10)

In order to obtain (1.10) we have used the representation in terms of Dirac deltas of the

d  0 ( eu )   ( n )
derivative of Chebyshev function to get  dug (u )  g (log n) , and the
0
du n 1 n

Euler formula for cosine to represent the integral  dug (u )eu / 2 as the sum
0

1
 h(i / 2)  h(i / 2)  . An special case is whenever we choose
2

h( x)   ( s  x)   ( s  x)  and g (u )  cos(u ) (1.11)
2

4
Then we can use the functions in (1.11) and the formula (1.10) to get

1   '(1/ 2  is )  '(1/ 2  is )  1 
(2n  1/ 2)
  (s   )  2   (1/ 2  is)   (1/ 2  is)   1  4s  
 
2
n 0  2n  1/ 2  s 
2 2

(1.12)

(1.12) is the ‘density of states’ in QM , and can be used to know how many zeros of the
T
 
form ½+is are with imaginary part less that a given ‘T’ since N (T )   ds    ( s   ) 
0   0 

A formal derivation of (1.12) can be obtained considering the following indentities for
divergent series involving zeta regularization or analytic continuation


1 
 ( n)  '(1/ 2  is )
 ean 
n 0 1  ea
(linearity) n
n 0
1/ 2is

 (1/ 2  is )
(1.13)


 ( n) 1   (n) is   (n)  is 

n 0 n
cos( s log n )  
2  n0 n
n 
n 0 n
n 

(1.14)


And the Laplace transform of Cosine  dte st cos(at )  s ( s 2  a 2 )1 .The poles inside
0

(1.14) are of three kinds s   n from the Non-trivial zeros of Zeta function ,
s  i / 2 due to the divergent value  (1) and s  i 1/ 2  2n  n  N form the trivial
zeros of zeta function -2,-4,-6,..........

o Riemann-Weyl formula and a solution for the inverse of potential V(x):

A similar formula to (1.10) had been previously introduced by Weyl in 1972 [9]



 ( n) 1 ' 1 ir 
 h( )  h(i / 2)  h(i / 2)  g (0) log   2
 n 1 n
g (log n) 
2  h(r )   4  2  dr

(1.15)

Weyl summation formula can be used to solve equation (1.10) if we make inside this
integral equation the change of variable x  V 1 ( ) , then (1.10) is simply proportional
e iu i / 4iu
to the inverse Fourier transform of  u
on the interval  0,  which is just

1
proportional to another sum  
involving the imaginary parts of the Riemann

zeta zeros , to get rid off the sum we ca use (1.15) to express the inverse of potential

5
 A
V 1 ( )   BCos (c   / 4)  D 
2  i

 (n)cos( log n   / 4)     1 ir   1 1 
E  Fp.v   dr      
n log n
1/ 2
    4 2     r 1/ 2   r 1/ 2  
n 1
  
(1.16)

Where A, B, c, D, E , F  R are real parameters that define the potential , from formula
(1.16) the Hamiltonian would be self adjoint and its energies (imaginary part of zeros)
would be real numbers.

The sum involving  (n) can be treated using fractional calculus and zeta regularization


 (n)cos( log n   / 4) d 1/ 2   '(1/ 2  i )  d 1/ 2   '(1/ 2  i ) 
2  i     i   (1.
n 1 n log n
1/ 2
d 1/ 2   (1/ 2  i )  d 1/ 2   (1/ 2  i ) 
17)

At the points    the inverse of potential becomes  , as we can expect from

           
1/ 2 1/ 2 1/ 2

 0  0

2. Zeta regularization for divergent integrals:

Given the function f ( x)  x m , we can use the Euler-Maclaurin summation formula to


obtain a recurrence relation between an integral of the form I (m, )   p m dp m Z 
0
 1
with m  x m1dx   m and the series i
i 0
m
, ref [7]
0

1 
B2 r
I (m, )  (m / 2) I (m  1, )   i m   amr (m  2r  1) I (m  2r ,  )
i 0 r 1 (2 r )!

(2.1)

  
m m1 
B2 r m!(m  2r  1)! m2 r
 x dx 
m

20
x dx   (  m )    x dx
r 1 (2 r )!(m  2 r  1)! 0

0

(m  1)
The coefficients amr  vanish when m  2  2r , hence the sum inside
(m  2r  2)
(2.1) is finite if m is an integer , in the physical limit the cutoff    , this makes the

6
 1
series i i 0
m
to be divergent for m  1 , in this case we should use the Functional

equation for the Zeta function to obtain the (Regularized) value

1
lim  n m 1  2m  3m  ...   m   R (m)   (m) (2.2)

n 1

(2.2) is the Zeta-regularized value for the divergent sum envolved in (2.1) , using this
method we can compute the divergent integrals I (m,  )    , for m=1,2,3

I (0, )   (0)  1/ 2

I (0,  )
I (1,  )    (1)
2
(2.3)
 1  B
I (2,  )   I (0,  )   (1)   2 a21I (0, )
 2  2

3 1 B 
I (3,  )    I (0, )   (1)   2 a21I (0,  )    (3)  B2 a31I (0, )
2 2 2 

The case m=0 is just equal to the divergent series 1+1+1+1+1+1+1+1+1+... taking the
regularized value -1/2 evaluated from  (0)

For an arbitrary function f(x) so its integral would diverge as a power of the cutoff  N 1
we could expand f(x) into a Laurent series convergent for |x| <1 and |x| >1 so we find

 N a 

 dxf ( x)   cr I (r , )  c1I (a, 1, )  O( )    dx  ci x    c j a

N
1 i  j 1
(2.4)
a r 0 i 0 0 j 2

ci  
, taking    , and using (2.1) (2.2) (2.3) to regularize the divergent

integrals I (m, ) we could obtain a regularized (finite) value for the integral  dxf ( x) ,
0

dx
however the logarithmic divergent integral I (a, 1, )   can not regularized by our
a
x
formulae, the solution would be to use the Euler-Maclaurin summation to approximate
the divergent integral by a divergent Harmonic sum that can be attached a ‘Ramanujan
a
1
sum’    (  =Euler-Mascheroni constant)
n 1 n

7
 o Zeta regularized product of distributions:

x
s 1
Formulae (2.1-2.3) can be used to compute divergent integrals of the form dx , but
0

also could give an answer to the problem of multiplication of two distributions

involving Dirac delta and its derivatives D m ( x) , if we tried to define the product of
distributions involving delta functions we could use the ‘convolution theorem’ applied
to the Fourier transform ( A=normalization constant) :

(2 ) 2 i mn D m ( ) D n ( )  F  x m  x n   AF 


dtt m ( x  t )n  (2.5)

Unfortunately (2.5) makes no sense , the integral is divergent for every real or complex
value of ‘x’ , if m and n are positive integers using the Binomial expansion

n
n
i mn D m ( ) D n ( ) =    i m k AD nk ( )(1)k i nk D m k (0)[ R ] (2.6)
k 0  k 


n
   x mk dx
n
i mn D m ( ) D n ( ) =   
k 0  
k
AD nk
 ( )( 1) k n k
i (  1) m k
 1 (2.7)
0

‘R’ stands for regularization (regularized value) , the divergent integrals come now
from the dirac delta and its derivatives evaluated at x=0 , which are proportional to


 x dx for k=2r+1 (Odd) the integral considered in Principal Value

k
is 0 , for k=2r


(even integer) the integral can be written as i 2 r D 2 r (0)  2 I (2r , ) 


, I (2, r )   x 2 r dx (r=integer) and can be evaluated using (2.1) and (2.2) .
0

The expression (2.7) is real ,this is what one would expect since the product of two
distributions taking only real values must be real , however (2.6 ) is not still invariant
under the change m  n and n  m (this is a mistake we made in paper [7] ) so we
should take a more symmetrical product of distributions defined by

D   D  
m n
R
( ) 
2

1 m
D  ( ) D n ( )  D n ( ) D m ( )  (2.8)

The simplest case is m=n=0 so     R ( )   A ( )

For the case of ‘m’ and ‘n’ not being an integer or we have a shifted dirac delta
D k ( x  a ) , we could use the identies for the k-th power of ‘x’ or the traslation
d
operator e D and D  in the form
dx

8
 
a j r j 
r
e aD D r ( x)   (1) j D  ( x)   ( x  a) D r      D  1
k
(2.9)
j 0 j! k 0  k 


In case of integrals on Rd  dkF ( k ) , if the function F , is invariant under Lorentz
Rd
transformations, then making a Wick rotation to imaginary time t  it ,the metric
becomes ds 2  dx 2  dy 2  dz 2  dt 2 which is invariant under rotations, taking 4-
 d /2 
(d / 2) 0
dimensional polar coordinates our integral can be evaluated as drf (r )r d 1 , if

not we could replace the integral over the cross section (angles) d by a discrete sum


  drf (r ,  )r
i
i
d 1
, with ‘d’ equal to the dimension of space-time
0


x2
 Example:  dx
a
1 x
with    in this case the integral has a power-law

(quadratic) divergence  2 , a >1 and integer (this is not relevant since the
integral diverges only for big ‘x’ ) , the Laurent series for |x| bigger than 1 is

x  1  x 1   (1) j x1 j , if we approximate the logarithmic divergent integral
j 3

1
of 1/x by the divergent series na
n 0
(after a change of variable x=t+a) then,

the approximate ‘Zeta regularized’ value of the integral would be

 x2   (0)  '(a ) a 2  (1) j 2 j

 dx     (  1)   a   a (2.10)
a 1  x R 2  ( a ) 2 j 3 j  2


x4
Another example without a logarithmic divergence , would be  dx in this case
a
(1  x 2 )
n(1) 32 n  n
the regularized finite value is just  (0)  a   a , the logarithmic derivative
n2 2n  3
of Gamma function inside (2.10) is just the Ramanujan resummation of the Hurwitz
series  H (1, a) avoiding the pole at s=1

Appendix A: an integral Trace for the Green function


A formula for the sum   ( E  E ) in terms of the Trace of the ‘Resolvent’ (green
n 0
n

function ) of a Quantum Hamiltonian Hˆ n  Enn can be defined as:

9
 
1
Tr G ( x, x ', E )   4 d 4 xG ( x, x, E )     ( E  En ) G ( x, x ', E )  ¨ (A.1)
R
n 0 E  i  Hˆ

One of the easiest method to prove this , is to consider that given a convergent series

  a 
with sum S and its Borel transform B(s) defined by B( S , an )   dt   n x n e t then
0  n 0 n ! 

S=B(S) , S   an in this case if we take the series
n 0

1 E 1   (1) n (i  H ) n x n    t (1i  Hˆ )

  E 1      dte (A.2)
E  i  H 1  (i  H )  n 0 E n n!  0
E

Where  is an small number so   0 , then using the formula for the Principal value
1 1
P.V     ( x)  , in this case taking the trace of the operator inside (A.2) we
x x  i
can give a proof to (A.1) using the technique of Borel resummation.


Another example of the method of Borel resummation , let be P( x)   (1) n  (n) x n
n 0
the generating function of the coefficients  (n) , let be the function f(t) defined by

 ( s  1)   dtf (t )t s 1 then using again the Borel-generalized resummation
0

 
   
f (t )
P( x)   dt   (1) n ( xt ) n  f (t )   (1) n  (n) x n or P( x)   dt (A.3)
0  n 0  n 0 0
1  xt


If we took the Mellin transform on both sides  dxx s 1 we Would find
0

Pˆ ( s )  Kˆ ( s ) Fˆ (1  s ) , or in terms of improper integrals

 
    ( s )
0    (n)( x) n    dtf (t )t   ( s)
s
dt since (A.5)
n 0  sin( s ) 0

This last formula is known as ‘Ramanujan Master theorem’ , note that we have proved
this only using the fact that for a convergent series its sums and Borel transform must be
equal S=B(S).

10
 References:

[1] Apostol, T. M. “Introduction to Analytic Number Theory”. New York: Springer-

Verlag, 1976.

[2] Berry, M. V. and Keating, J. P. " and the Riemann Zeros. “ In

Supersymmetry and Trace Formulae: Chaos and Disorder” (Ed. I. V. Lerner, J. P
Keating, and D. E. Khmelnitskii). New York: Kluwer, pp. 355-367, 1999.

[3] Conrey, J. B. "The Riemann Hypothesis." Not. Amer. Math. Soc. 50, 341-353,
2003. available at http://www.ams.org/notices/200303/fea-conrey-web.pdf.

[4] Delabaere E., “Ramanujan's Summation, Algorithms Seminar” 2001–2002, F.

Chyzak (ed.), INRIA, (2003), pp. 83–88.

[5] Elizalde E. ; “Zeta-function regularization is well-defined”, Journal of Physics A

27 (1994), L299-304.

[6] Garcia J.J “Chebyshev Statistical Partition function : A connection between

Statistical Mechanics and Riemann Hypothesis “ Ed. General Science Journal
(GSJ) 2007 (ISSN 1916-5382)

[7] Garcia J.J “ A new approach to renormalization of UV divergences using Zeta

regularization techniques “ Ed. General Science Journal (GSJ) 2008
(ISSN 1916-5382)

[8] Polyanin, A. D. and Manzhirov, A. V., “Handbook of Integral Equations”, CRC

Press, Boca Raton, 1998.

[9] Weyl, A. "Sur les formules explicites de la théorie des nombres", Izv. Mat.
Nauk (ser. Mat.) 36 (1972)

11