WEIGHTED VALUE DISTRIBUTIONS OF THE RIEMANN ZETA

FUNCTION ON THE CRITICAL LINE

ALESSANDRO FAZZARI
arXiv:2101.08036v1 [math.NT] 20 Jan 2021

Abstract. We prove a central limit theorem for log |ζ(1/2 + it)| with respect to the
measure |ζ (m) (1/2 + it)|2k dt (k, m ∈ N), assuming RH and the asymptotic formula for
twisted and shifted integral moments of zeta. Under the same hypotheses, we also study
a shifted case, looking at the measure |ζ(1/2 + it + iα)|2k dt, with α ∈ (−1, 1). Finally
we prove unconditionally the analogue result in the random matrix theory context.

1. Introduction and statement of the main results

In 1946 Selberg [20] proved a central limit theorem for the real part of the logarithm of
the Riemann zeta function on the critical line, showing that the distribution of log |ζ(1/2+
it)| is approximately Gaussian with mean 0 and variance 21 log log T , i.e.
( ) Z
1 log |ζ( 21 + it)| ∞
x2 dx
(1.1) meas t ∈ [T, 2T ] : q ≥V ∼ e− 2 √
T 1
log log T V 2π
2

for any fixed V ∈ R, as T goes to infinity. The author [9] studied the distribution of
log |ζ(1/2 + it)| with respect to the weighted measure |ζ(1/2 + it)|2 dt and, assuming
the Riemann Hypothesis (RH), proved that it is asymptotically Gaussian with mean
log log T and variance 12 log log T . In this paper we investigate the value distribution of
log |ζ(1/2 + it)| with respect to the measure
(1.2) |ζ (m) (1/2 + it)|2k dt
for any fixed m, k non negative integers. The motivation is due to the study of the large
values of the Riemann zeta function, i.e. the uniformity in V in the central limit theorem
(1.1). In [21], Soundararajan speculates that an upper bound like
( )
1 log |ζ(1/2 + it)| 1  V2 
meas t ∈ [T, 2T ] : q ≥ V ≪ exp −
T 1
log log T V 2
2

holds in a large range for V , in particular we expect (see [19, Conjecture 2], see also [12])
that for any fixed k
1 n o 1 1
(1.3) meas t ∈ [T, 2T ] : log |ζ(1/2 + it)| ≥ k log log T ≪k √ 2.
T log log T (log T )k
Expressing the characteristic function of positive reals as a Mellin transform, the left hand
side of (1.3) can be written as
Z +∞ Z 2T
1 1 1 du
k 2 k 2 eiu(log |ζ(1/2+it)|−k log log T ) |ζ(1/2 + it)|2k dt
(log T ) 2π −∞ T (log T ) T 2k + iu
1
and is then related to the distribution of log |ζ(1/2 + it)| with respect to the weighted
measure (1.2), with m = 0.

In the case k = 1, 2, we can prove a central limit theorem for log |ζ(1/2 + it)| with
respect to the measure |ζ (m) (1/2 + it)|2k dt, assuming RH only.
Corollary 1. Assume the Riemann Hypothesis, let m be a non negative integer and
k = 1 or k = 2. As t varies in T ≤ t ≤ 2T , the distribution of log |ζ(1/2 + it)| is
asymptotically Gaussian with mean k log log T and variance 12 log log T , with respect to
the weighted measure |ζ (m) (1/2 + it)|2k dt.
In the case k > 2, since not even the moments of zeta are known, we cannot expect to
prove unconditionally a central limit theorem. However, assuming the asymptotic formula
[13, Conjecture 7.1] suggested by the recipe [6] for the twisted and shifted 2k-th moments
of the Riemann zeta function
Z 2T  −it
a
ζ(1/2 + α1 + it) · · · ζ(1/2 + αk + it)ζ(1/2 + β1 − it) · · · ζ(1/2 + βk − it)dt
T b
one can deal with the general case too. More precisely, we use the strategy of [4, Theo-
rem 1.2] in order to re-write Hughes and Young conjecture and we assume the following
statement.
Conjecture 1 (Hughes-Young). Let T be a large parameter, α1 , . . . , αk , β1 , . . . , βk ≪
(log T )−1 , Φj the set of subset of {α1 , . . . , αk } of cardinality j, for j = 0, . . . , k, and
similarly Ψj the set of subset of {β1 , . . . , βk } of cardinality j. If S ∈ Φj and T ∈ Ψj
then write S = {αi1 , . . . , αij } and T = {βl1 , . . . , βlj } where i1 < i2 < · · · < ij and
l1 < l2 < · · · < lj . Let (αS ; βT ) be the tuple obtained from (α1 , . . . , αk ; β1 , . . . , βk ) by
replacing αir with −βlr and replacing βlr with −αir for 1 ≤ r ≤ j. Consider
X f (a) X g(b)
A(s) = and B(s) =
θ
as θ
bs
a≤T b≤T

where f (a) ≪ aε and g(b) ≪ bε for any ε > 0. We then conjecture that there exists a
δ > 0, depending on k, such that if θ < δ then
Z 2T
A(1/2+it)B(1/2 + it)ζ(1/2+it+α1) · · · ζ(1/2+it+αk )ζ(1/2−it+β1 ) · · · ζ(1/2−it+βk )dt
T

equals
X Z 2T X X  t −S−T
(1.4) f (a)g(b) ZαS ;βT ,a,b (t) dt + O(T 1−η )
T 0≤j≤k S∈Φj
2π
a,b≤T θ
T ∈Ψj
P P
for some η > 0, where we have written (t/2π)−S−T for (t/2π)− x∈S x− y∈T y
and
X  
1 m1 · · · mk n1 · · · nk
Zα;β,a,b (t) := √ 1/2+α1 1/2+αk 1/2+β1 1/2+βk
V
am1 ···mk =bn1 ···nk ab m1 · · · mk n1 · · · nk tk
2
with Z
1 G(s)
V (x) := (2π)−ks x−s ds
2πi (1) s
where G(s) is an even entire function of rapid decay in any fixed strip |ℜ(s)| ≤ C satisfying
G(0) = 1.
With this assumption we can prove the main theorem.
Theorem 1. Let k, m ∈ N and assume the Riemann Hypothesis and Conjecture 1 for k.
As t varies in T ≤ t ≤ 2T , the distribution of log |ζ(1/2 + it)| is asymptotically Gauss-
ian with mean k log log T and variance 12 log log T , with respect to the weighted measure
|ζ (m) (1/2 + it)|2k dt.
In particular, being Conjecture 1 known in the cases k = 1 (see [2, 5] and e.g. [3] for
the easy modifications needed to account for the shifts) and k = 2 (see [13] and [4]), we
notice that Corollary 1 trivially follows from Theorem 1.

We notice that Theorem 1 shows that the m-th derivative has no effect in the weighted
distribution of log |ζ(1/2 + it)|. This is consistent with the conjecture (see [14, Conjecture
6.1] and [8])
Z 2T
2
|ζ(1/2 + it)|2k−2h |ζ ′ (1/2 + it)|2h dt ∼ c(h, k)T (log T )k +2h
T
which indicates that |ζ ′(1/2 + it)|2h amplifies the contribution coming from the large val-
ues in the same way as |ζ(1/2 + it)|2h would do (up to a normalization of (log T )2h ). More
generally, as far as moments are concerned, the m-th derivative of zeta should behave like
zeta itself (see e.g. [7]), up to a normalization of logm T , in accordance with Theorem
1. We also note that, while in Selberg’s classical case the mean is 0 because the contri-
bution of the small values and that of the large values of zeta balance out, tilting with
|ζ(1/2 + it)|2k the mean of log |ζ(1/2 + it)| moves to the right as k grows and this reflects
the fact that the measure |ζ(1/2 + it)|2k dt gives more and more weight to the large values
of the Riemann zeta function.

Moreover, we look at the shifted weighted measure |ζ(1/2 + it + iα)|2k dt with α a real
number such that |α| < 1. As we will see, the distribution of log |ζ(1/2 + it)| is quite
sensitive to the parameter α. Indeed in computing the integral
Z 2T
(1.5) log |ζ(1/2 + it)||ζ(1/2 + it + iα)|2k dt
T
one expects the same magnitude as in the unshifted case if |α| is smaller than (log T )−1 ,
which is the typical scale for the Riemann zeta function. On the other hand, if |α| is
larger then the two factors in the integral (1.5) start decorrelating, thus the size of the
integral decreases. This phenomenon is shown in the following result, in which we use the
notation (
log log T + O(1) if |α| log T ≤ 1
µ
eα =
− log |α| + O(1) if |α| log T > 1
for any α ∈ (−1, 1).
3
Theorem 2. Let k ∈ N and assume the Riemann Hypothesis and Conjecture 1 for k. As
t varies in T ≤ t ≤ 2T , for any fixed and real α such that |α| < 1, the distribution of
µα and variance 21 log log T , with
log |ζ(1/2 + it)| is asymptotically Gaussian with mean ke
respect to the measure |ζ(1/2 + it + iα)|2k dt.
This theorem shows that the shift has no effect if it is smaller than (log T )−1 . On the
contrary, for larger values of the shift the mean gets smaller, for instance if
(log T )δ
α=
log T
with δ ∈ (0, 1) then the mean is ∼ (1 − δ)k log log T . In this shifted case too, if k ≤ 2
then Theorem 2 holds assuming RH only.

Lastly, we show that in the random matrix theory setting, an analogous weighted central
limit theorem can be proved unconditionally. We consider the characteristic polynomials
Z = Z(U, θ) = det(I − Ue−iθ )
of N ×N unitary matrices U and we investigate their distribution of values with respect to
the circular unitary ensemble (CUE). It has been conjectured that the limiting distribution
of the non-trivial zeros of the Riemann zeta function, on the scale of their mean spacing,
is the same as that of the eigenphases θn of matrices in the CUE in the limit as N → ∞
(see e.g. [16]). Then we consider a tilted version of the Haar measure and we have the
following theorem.
Theorem 3. As N → ∞, the value distribution of log |Z| is asymptotically Gaussian
with mean k log N and variance 21 log N with respect to the measure |Z|2k dHaar .
As usual, the correspondence with Theorem 1 holds if we identify the mean density of
the eigenangles θn , N/2π, with the mean density of the Riemann zeros at a height T up
1 T
the critical line, 2π log 2π , i.e. if
T
N = log .
2π

2. Proof of Theorems 1 and 2

To prove both the theorems, we introduce a set of shifts α1 , . . . , αk , β1 , . . . , βk and we
denote for the sake of brevity
ζα,β (t) := ζ(1/2 + α1 + it) · · · ζ(1/2 + αk + it)ζ(1/2 + β1 − it) · · · ζ(1/2 + βk − it).
The general strategy of the proof is similar to the one of Theorem 1 in [9], but here
we avoid the detailed combinatorial analysis we performed in [9], by working with Euler
products instead of Dirichlet series, inspired by [1, Proposition 5.1]. The first step is
then approximating the logarithm of the Riemann zeta function with a suitable Dirichlet
polynomial. Let’s denote
X 1
(2.1) Pe(t) :=
p≤x
p1/2+it
4
where x := T ε , with ε := (log log log T )−1 . Now we show that log |ζ(1/2 + it)| has the
same distribution as ℜPe(t) with respect to the measure ζα,β (t)dt. This is achieved in the
following proposition by bounding the second moment of the difference.

Proposition 2.1. For k a non negative integer, assume Conjecture 1 and RH. Let T be a
large parameter and α1 , . . . , αk , β1 , . . . , βk ∈ C such that |αi |, |βj | < 1, |ℜ(αi )|, |ℜ(βj )| ≤
(log T )−1 and |αi − βj | ≪ (log T )−1 for all 1 ≤ i, j ≤ k. Then we have:
Z 2T
 
 log |ζ(1/2 + it)| − ℜPe(t)2 ζα,β (t)dt ≪k T (log T )k2 (log log log T )5/2 .
T

Proof. From Tsang’s work [22, equation (5.15)] we know that

(2.2) log ζ(1/2 + it) − Pe(t) = S1 + S2 + S3 + O(R) − L(t)

with
X
X p−r(1/2+4/ log x+it)
S1 := p−1/2−4/ log x − p−1/2 p−it , S2 :=
p≤x pr ≤x
r
r≥2

 X  
X Λ(n) 5  Λ(n) 
S3 := n−1/2−4/ log x−it , R :=   + log T
3
log n log x  3
n1/2+4/ log x+it 
x<n≤x n≤x

XZ 1/2+4/ log x  
1 4 1 1
L(t) := + −u du,
ρ 1/2 2 log x u + it − ρ 2 + log4 x − ρ
1

where the sum in the definition of L(t) is over all the non-trivial zeros of ζ, then we have
to bound the second moments of the terms on the right hand side with respect to weighted
measure ζα,β (t)dt, by using Conjecture 1 (note that we are allowed to apply the conjec-
ture, since the shifts are small up to a change of variable, being |αi − βj | ≪ (log T )−1 ).
R 2T
Let’s start with T |S1 |2 ζα,β (t)dt which is
 Z 2T 
 X X  t −S−T X 
≪  (p −4/ log x
− 1)(q −4/ log x
− 1)ZαS ,βT ,p,q (t)dt
T 0≤j≤k S∈Φ
2π p,q≤x j
T ∈Ψj
(2.3) Z  
2T X X X − log4 x − log4 x

≪  1 (a)1 (b)(a − 1)(b − 1)Z (t) dt
 x x α ,β
S T ,a,b 
T 0≤j≤k S∈Φj a,b≤x
T ∈Ψj

where 1x (·) is the indicator function of primes up to x. If we denote Ω(n) the function
which counts the number of prime factors of n with multiplicity, then we have
X X  
(2.4) 1x (n)f (n) = ∂z z Ω(n) f (n) z=0
n≤x n:
p|n =⇒ p≤x
5
hence we get
(2.5)    
Z 2T Z Z
2
2T X X 
∂z ∂w 1 G(s)  t ks j,S,T 
dt
|S1 | ζα,β (t)dt ≪  Iα,β (z, w; s)ds 
T T 0≤j≤k S∈Φj
2πi (1) s 2π z=0
w=0
T ∈Ψj

where
X
(2.6) j,S,T
Iα,β (z, w; s) := z Ω(a) w Ω(b) gx (a)gx (b)ZeαS ,βT ,a,b (s)
a,b:
p|ab =⇒ p≤x

with gx the multiplicative function defined by gx (pα ) = p−4α/ log x − 1 and

X 1
(2.7) Zeα,β,a,b (s) := √ 1/2+α1 +s 1/2+αk +s 1/2+β1 +s 1/2+βk +s
.
am1 ···mk =bn1 ···nk ab m1 · · · mk n1 · · · nk
0
We now analyze the first term Iα,β (z, w; s) and then we will see how to apply the method
to deal with all the others. Since the sum in the definition (2.6) is multiplicative we have
(2.8)
X z Ω(a) w Ω(b) gx (a)gx (b)
0
Iα,β (z, w; s) = √ 1
+α1 +s 1
+αk +s 1
+β1 +s 1
+βk +s
am1 ···mk =bn1 ···nk : ab m12 · · · mk2 n12 · · · nk2
p|ab =⇒ p≤x
Y X a b
z Ω(p ) w Ω(p ) gx (pa )gx (pb )
= a b 1 1 1 1
p≤x a+m1 +···+mk = p 2 + 2 +m1 ( 2 +α1 +s)+···+mk ( 2 +αk +s)+n1 ( 2 +β1 +s)+···+nk ( 2 +βk +s)
=b+n1 +···+nk
Y X 1
· m1 ( 12 +α1 +s)+···+mk ( 12 +αk +s)+m1 ( 12 +β1 +s)+···+nk ( 21 +βk +s)
p>x m1 +···+mk = p
=n1 +···+nk

and by putting in evidence the first terms in the Euler products, this is
(2.9)
Y zwgx (p)2 zgx (p) wgx (p)

= Aα,β (z, w; s) 1+ + 1+β1 +s + · · · + 1+α +s
p≤x
p p p k
Y 1 1
Y
1 1

· 1 + 1+α1 +β1 +2s + · · · + 1+α +β +2s 1 + 1+α1 +β1 +2s + · · · + 1+α +β +2s
p≤x
p p k k
p>x
p p k k
Y k X 
∗ zwgx (p)2 zgx (p) wgx (p)
= Aα,β (z, w; s) ζ(1 + αi + βj + 2s) · exp + 1+β1 +s + · · · + 1+α +s
i,j=1 p≤x
p p p k

where Aα,β (z, w; s) and A∗α,β (z, w; s) are arithmetical factors (Euler products) converging
absolutely in a half-plane ℜ(s) > −δ for some δ > 0, uniformly for |z|, |w| ≤ 1, such that
their derivatives with respect to z and w at 0 also converge in the same half plane. We
now have extracted the polar part, hence we are ready to shift the integral over s in (2.5)
to the left of zero. To do so, it is convenient to prescribe the same conditions as in [4,
α +β
Remarks after Lemma 2.1], assuming that G(s) vanishes at s = − i 2 j for all i, j, so that
6
the only pole we pick in the contour shift is at s = 0. Moreover we assume that the shifts
are such that |αi + βj | ≫ (log T )−1 for every 1 ≤ i, j ≤ k, so that
k
Y 2
(2.10) |ζ(1 + αi + βj )| ≪k (log T )k .
i,j=1

Hence by (2.8), (2.9) and (2.10), we get

 Z 
1 G(s)  t ks 0
∂z ∂w Iα,β (z, w; s)ds
2πi (1) s 2π z=0
   X  w=0  

k2  zwgx (p)2 zgx (p) wgx (p) 
≪k (log T ) ∂z ∂w exp + 1+β1 + · · · + 1+α 
p p p k z=0

p≤x w=0
X 2
k2 |gx (p)| 2
≪k (log T ) ≪k (log T )k (log log log T )2
p≤x
p
P
being |ℜ(αi )|, |ℜ(βj )| ≤ (log T )−1 and p≤x |p−4/ log x − 1|/p ≪ 1; this last bound can be
obtained by using the Taylor’s approximation e−z = 1 + O(z) for z ≪ 1, which yields
X |p−4/ log x − 1| X |e−4 log p/ log x − 1| 1 X log p
= ≪ ≪1
p≤x
p p≤x
p log x p≤x
p
j,S,T
by Mertens’ first theorem. All the other terms Iα,β (z, w) can be treated exactly in the
same way as Iα,β (z, w) by assuming that |αi ± βj | ≫ (log T )−1 for every i, j, since they
0

only differ from the first case by permutations and changes of signs of the shifts. Therefore
we get
 Z 
1 G(s)  t ks j,S,T 2
∂z ∂w Iα,β (z, w; s)ds ≪k (log T )k (log log log T )2
2πi (1) s 2π z=0
w=0
provided that
(2.11) |αi ± βj | ≫ (log T )−1 for every 1 ≤ i, j ≤ k.
Plugging this into (2.5), we prove that
Z 2T
2
(2.12) |S1 |2 ζα,β (t)dt ≪k T (log T )k (log log log T )2 .
T

Moreover, since the left hand side in (2.12) is holomorphic in terms of the shifts, although
we have proved the above for αi , βj such that (2.11) holds, the maximum modulus principle
can be applied to obtain the bound to the enlarged domain we need.
R 2TWe will
2
treat similarly also the other pieces. As regards the second one, we have that
T
|S2 | ζα,β (t)dt is
Z  − 1
4r
− 2
4r

2T X X  X p1 log x p2 log x 
≪  ZαS ,βT ,pr11 ,pr22 (t)dt.
 r1 r2
T 0≤j≤k S∈Φj r r
p11 ,p22 ≤x
T ∈Ψj r1 ,r2 ≥2
7
As before, we analyze the first term only since all the others are completely analogous.
The term for j = 0 is
4r1 4r2
Z − −
1 G(s)  t ks X p1 log x p2 log x 1
p 1 1 ds
2πi (1) s 2π r1 r2
r1 r2 r1 r2
p1 p2 m1
2
+α1 +s
· · · nk
2
+βk +s
p1 m1 ···mk =p2 n1 ···nk :
r r
p11 ,p22 ≤x, r1 ,r2 ≥2

and this time, because of the condition r1 , r2 ≥ 2, when we estimate the sum via the first
terms of its Euler product we just get that the above is

k
Y
≪k |ζ(1 + αi + βj )|
i,j=1

and, applying the same machinery as before, this yields

Z 2T
2
(2.13) |S2 |2 ζα,β (t)dt ≪k T (log T )k .
T

We use the same approach in order to bound the second moment of S3 as well, which
is
Z  Z 
2T  1
 G(s)  t ks X Λ(a)Λ(b) a−4/ log x b−4/ log x 
dt
≪k  2πi √
T (1) s 2π am
log a log b ab m 2 +α1 +s · · · n 2 +βk +s 
1 1

1 ···mk =bn1 ···nk 1 k

x<a,b<x3
 X 2 Y k
1 2
≪k T |ζ(1 + αi + βj )| ≪k T (log T )k
3
p i,j=1
x<p≤x

P
since the sum x<p≤x3 p−1 is bounded.

We deal with R in the same way and we get that

Z 2T
2
(2.14) |R|2 ζα,β (t)dt ≪k T (log T )k (log log log T )2
T

where the extra factor with the triple log comes from the second term in the definition of
R, while the first one can be treated analogously to S3 .

Finally we have to bound the second moment of ℜL(t). To do so, in view of equations
(2.8) and (2.9) in [9], it suffices to study
Z
1 2T  1 2  X Λ(n)n−4/ log x 
+
2
(2.15) log   + log T ζα,β (t)dt
(log x)2 T ηt log x 3
n1/2+it
n≤x
8
where ηt := minρ |t − γ| and log+ t := max(log t, 0), with the aim of proving that this is
2
≪k T (log T )k (log log log T )5/2 . By applying the Cauchy-Schwarz inequality, the above is
(2.16)
 Z 2T   12
1  X Λ(n)n−4/ log x  1 4 2 2
≤  + |ζ(1/2 + β1 − it)| · · · |ζ(1/2 + βk − it)| dt
T log x 3
n1/2+it ε
n≤x
Z   21
2T
+ 1 4 2 2
· log |ζ(1/2 + α1 + it)| · · · |ζ(1/2 + αk + it)| dt
T ηt log x
p
and the first term can be treated as R above and it is ≪k T (log T )k2 (log log log T )4 .
We now conclude the proof, bounding the second term in (2.16). If we denote τ :=
[T − log1 x , 2T + log1 x ] then we have
Z 2T 
1 4
log+ |ζ(1/2 + α1 + it)|2 · · · |ζ(1/2 + αk + it)|2 dt
T ηt log x
X Z 1/ log x  1 4 Y
k


≤ log+ ζ 1/2 + αj + i(w + γ) 2 dw
γ∈τ −1/ log x
|w| log x j=1
(2.17) XZ 1  4 Y k    2
+ 1

 t  dt
= log ζ 1/2 + αj + i γ + log x  log x
γ∈τ −1 |t| j=1

1
Z 1 Y  X  
k
t 2k
1/k
4
≪ (log |t|) ζ 1/2 + iγ + αj + i  dt
log x −1 j=1 γ∈τ
log x

by Hölder inequality. The remaining sum can be bounded under RH in view of Kirila’s
[17, Theorem 1.2] and Milinovich’s [18] works, which generalize the well known result due
to Gonek about sum over the non-trivial zeros of zeta [10, Corollary 1]. Indeed, since the
shifts αj +i logt x in the sum over zeros in (2.17) have modulus ≤ 1 and real part ≤ (log T )−1
in absolute value, then we have
X   t 2k 2
ζ 1/2 + iγ + αj + i  ≪k T log T (log T )k
γ∈τ
log x

for every j = 1, . . . , k and putting this into (2.17) we get

(2.18)
Z 2T 
1 4 T log T 2
log+ |ζ(1/2 + α1 + it)|2 · · · |ζ(1/2 + αk + it)|2 dt ≪k (log T )k .
T ηt log x log x
Plugging (2.18) into (2.16) we prove that
Z 2T q q
2
|ℜL(t)| ζα,β (t)dt ≪k T (log T ) (log log log T ) T (log T )k2 log log log T
k 2 4
(2.19) T
2
≪k T (log T )k (log log log T )5/2
concluding the proof of the proposition. 
9
 The second step is getting rid of the small primes, showing that their contribution does
not affect the distribution asymptotically. This simple fact will simplify the third and
last step of the proof, as we will see in the following. Let’s define
X 1
(2.20) P (t) :=
p∈X
p1/2+it

where X := (log T, x] (we recall that x = T ε and ε = (log log log T )−1 ).
Proposition 2.2. For k a non negative integer, assume Conjecture 1 and RH. Let T be a
large parameter and α1 , . . . , αk , β1 , . . . , βk ∈ C such that |αi |, |βj | < 1, |ℜ(αi )|, |ℜ(βj )| ≤
(log T )−1 and |αi − βj | ≪ (log T )−1 for all 1 ≤ i, j ≤ k. Then we have:
Z 2T
 
ℜPe(t) − ℜP (t)2 ζα,β (t)dt ≪k T (log T )k2 (log log log T )2 .
T

Proof. This can be proved with the same method used in Proposition 2.1. We recall that
1log T (·) denotes the indicator function of primes up to log T . We start by studying
Z
1 G(s)  t ks X 1log T (a)1log T (b)
(2.21) √ 1 1 ds
2πi (1) s 2π am1 ···mk = ab m 2
+α1 +s +β +s
· · · n2 k
1 k
=bn1 ···nk

and, as usual, we estimate the sum with the first terms of its Euler product, extract the
polar part, shift the integral over s to the left, getting that (2.21) is
k    X   
Y  zw z w 
≪k |ζ(1 + αi + βj )| · ∂z ∂w exp + 1+β1 + · · · + 1+α 

i,j=1 p≤log T
p p p k z=0
w=0
and by the same argument as in the proof of Proposition 2.1 the above is
 X 2
k2 1 2
≪k (log T ) ≪ (log T )k (log log log T )2
p≤log T
p
and this concludes the proof. 
Finally we investigate the distribution of the polynomial ℜP (t), which has the same
distribution as log |ζ(1/2 + it)| thanks to Propositions 2.1 and 2.2. The most natural
method to do so is studying the moments and this is achieved in the following result.
Proposition 2.3. For k a non negative integer, assume Conjecture 1 and RH. Let T be a
large parameter and α1 , . . . , αk , β1 , . . . , βk ∈ C such that |αi |, |βj | < 1,P|ℜ(αi )|, |ℜ(βj )| ≤
(log T )−1 and |αi − βj | ≪ (log T )−1 for all 1 ≤ i, j ≤ k. Denote L := p∈X p1 ∼ log log T
and µ ∈ R such that µ ≪ log log T . Then for every fixed integer n we have
Z 2T
(ℜP (t) − kµ)n ζα,β (t)dt
T
Z 2T X X  −S−T   X 
t n z2
L−kzµ z gp (S, T )
= MαS ,βT (0) ∂z e 4 exp dt
T 0≤j≤k S∈Φ
2π 2 p∈X
p z=0
j
T ∈Ψj
 2

+ Ok,n T (log T )k −1+ε
10
where
X X X X
gp (S, T ) := p−x1 + p−x2 + px 3 + px 4
x1 6∈S x2 6∈T x3 ∈S x4 ∈T

and
X 1
Mα,β (s) := 1
+α1 +s 1
+αk +s 1
+β1 +s 1
+βk +s
m1 ···mk =n1 ···nk m1
2
· · · mk
2
n12 · · · nk2
so that Mα,β (0) is the first term of the moment of ζα,β , predicted by the recipe [6] (the sum
which defines Mα,β (s) does not converge for s = 0, so we appeal to [6, Theorem 2.4.1] for
the analytic continuation).

Proof. Expanding out the powers, since 2ℜ(z) = z + z, we get

(2.22)
Z 2T X n X h Z 2T s
n j −h
(ℜP (t) − kµ) ζα,β (t)dt = (−kµ) 2 P (t)r P (t) ζα,β (t)dt
T j+h=n
h r+s=h
r T

and using Conjecture 1, ignoring the error term which is negligible in this context, the
inner integral in (2.22) is
Z 2T X X  −S−T X
t
(2.23) 1∗r ∗s
X (a)1X (b)ZαS ,βT ,a,b (t)dt
T 0≤j≤k S∈Φ
2π a,b
j
T ∈Ψj

where 1∗r
X (a) denotes the indicator function of primes in the interval X, self-convoluted r
times. If we define temporarily the multiplicative function g given by g(pn ) = 1/n!, then
the inner sum over a, b in (2.23) equals
 X 
r s Ω(a) Ω(b)
(2.24) ∂z ∂w z w g(a)g(b)ZαS ,βT ,a,b (t)
z=0
a,b: w=0
p|ab =⇒ p∈X

then plugging (2.23) and (2.24) into (2.22), recollecting together the powers we expanded
before, we get
Z 2T Z
1 G(s)  t ks X X  t −S−T n h −kµz j,S,T i
(2.25) ∂z e Iα,β (z; s) dsdt.
T 2πi (1) s 2π 0≤j≤k S∈Φ
2π z=0
j
T ∈Ψj

with
X  z Ω(a)+Ω(b)
(2.26) j,S,T
Iα,β (z; s) := eα ,β ,a,b (s)
g(a)g(b)Z S T
a,b:
2
p|ab =⇒ p∈X

(see (2.7) for the definition of Zeα,β,a,b (s)). Now study the first term in (2.25), i.e. j = 0,
since all the other terms can be understood from the first one with a slight modification.
11
Then we look at
(2.27)
X (z/2)Ω(a)+Ω(b) g(a)g(b)
0
Iα,β (z; s) = √ 1
+α1 +s 1
+αk +s 1
+β1 +s 1
+βk +s
am1 ···mk =bn1 ···nk ab m12 · · · mk2 n12 · · · nk2
p|ab =⇒ p∈X
Y X (z/2)a+b g(pa)g(pb ) Y X 1
= a
m1 ( 12 +α1 +s)+···+nk ( 12 +βk +s)
+ 2b +m1 ( 12 +α1 +s)+···+nk ( 12 +βk +s)
p∈X a+m1 +···+mk = p 2
p6∈X m1 +···+mk = p
=b+n1 +···+nk =n1 +···+nk
X 
(z/2)2 z/2 z/2 z/2 z/2
= exp + 1+α1 +s + · · · + 1+α +s + 1+β1 +s + 1+β +s FX,α,β (z; s)Mα,β (s)
p∈X
p p p k p p k

where FX,α,β (z; s) is an arithmetical factor (Euler product) converging absolutely in a

product of half-planes containing the origin, such that FX,α,β (z; 0) is holomorphic at z = 0,
FX,α,β (0, 0) = 1 and all its derivatives at z = 0 are small, i.e. ∂zc [FX,α,β (z, 0)]z=0 ≪
(log T )−1 for any positive integer c. Now we want to shift the integral over s in (2.25) to
the left of zero, picking the contribution of the (only) pole at s = 0. To do so, we appeal
to the meromorphic continuation of the function Mα,β (s), see [6, Theorem 2.4.1]; thus we
1
can shift the path of integration to the vertical line (say) ℜ(s) = − 10 , where the integral
1−1/10+ε
is trivially bounded by ≪ T for any positive ε. Moreover, the contribution from
the pole at s = 0 gives
  X −α1  
z2 z p + · · · + p−βk
T ∂zn −kµz
e e 4
L
exp FX,α,β (z; 0) Mα,β (0).
2 p∈X p z=0

Thanks to the bounds for µ, L and the derivatives of FX,α,β , being FX,α,β (0; 0) = 1 and
2
Mα,β (0) ≪ (log T )k (again this is due to a similar argument as in the proof of Proposition
2.1, when we assume extra conditions on the shifts and then we appeal to the maximum
modulus principle), we get that the term for j = 0 in (2.25) equals
  X −α1   
z2 z p + · · · + p−βk 2
T ∂zn −kµz
e e 4
L
exp Mα,β (0) + Ok,n T (log T )k −1+ε .
2 p∈X p z=0

Analogously, the general term turns out to be

Z 2T  t −S−T  z2  X
z gp (S, T )
  2

n L−kµz
∂z e 4 exp MαS ,βT (0)dt + Ok,n T (log T )k −1+ε
T 2π 2 p∈X p z=0

and putting this into (2.23), i.e. summing over j, S, T , we get the thesis. 

Proof of Theorem 1. This proof follows easily from the three propositions we have
proved above. If we take µ = L and all the shifts small enough, i.e. αi , βj ≪ (log T )−1 for
12
any i, j, then the exponent in the right hand side of the thesis in Proposition 2.3 becomes
 X −α1 
z2 z X gp (S, T ) z2 |p − 1| + · · · + |p−βk − 1|
L+ − zkL = L + O z
4 2 p∈X p 4 p∈X
p
 X −α1 
z2 |p − 1| z2
= L + Ok z = L + Ok (z)
4 p∈X
p 4

and does not depend on S and T asymptotically, in fact. Hence we can bring that factor
outside and reconstruct the moment of ζα,β , as follows
Z 2T
(ℜP (t) − kL)n ζα,β (t)dt
T
h z2 i Z 2T X X  t −S−T  2

n L+Ok (z)
= ∂z e 4 MαS ,βT (0)dt + Ok,n T (log T )k −1+ε
z=0 T
0≤j≤k S∈Φ
2π
j
T ∈Ψj
 h i   Z 2T  
z2 n−1 2
= ∂zn e 4 L + Ok,n (log log T ) 2 ζα,β (t)dt + Ok,n T (log T )k −1+ε .
z=0 T

The thesis follows by analyzing

h z2 i  X  m  m 
n L n! 2zL 1 1 L 2
(2.28) ∂z e 4 = .
z=0
m1 +2m2 =n
m1 !m2 ! 4 2! 2 z=0

z2
If n is odd then the coefficient ∂zn [e 4 L ]z=0 vanishes, while if n is even then only the term
for m1 = 0 and m2 = n2 survives and gives (n/2)! n!
( L4 )n/2 = (n − 1)!!( L2 )n/2 , i.e.
Z 2T (
1 (1 + ok,n (1))(n − 1)!!( L2 )n/2 if n even
R 2T (ℜP (t) − kL)n ζα,β (t)dt =
.
ζα,β (t)dt T ok,n (log log T )n/2 if n odd
T

This matches with the Gaussian coefficient then this proves that, in the limit T → ∞,
ℜP (t) has Gaussian distribution, with mean k log log T and variance 21 log log T and then
so does log |ζ(1/2 + it)|, in view of Proposition 2.1 and 2.2. Theorem 1 follows by taking
the derivatives with respect to the shifts.

Proof of Theorem 2. To derive Theorem 2, in Proposition 2.3 we set α1 = · · · = αk = iα

and β1 = · · · = βk = −iα, with α ∈ R, |α| < 1 and we take µ as
(
X cos(α log p) log log T + O(1) if |α| log T ≤ 1
(2.29) µα := =µeα + O(1) =
p∈X
p − log |α| + O(1) if |α| log T > 1

by partial summation. Then we get

Z 2T
(ℜP (t) − kµα )n |ζ(1/2 + iα + it)|2k dt
T
h z2 i Z 2T


= 1 + ok,n (1) ∂zn e4L |ζ(1/2 + iα + it)|2k dt
z=0 T
13
 P
since the quantity 2z p∈X gp (S,T
p
)
− zkµα vanishes for all S, T for this choice of the shifts.
The thesis follows as in the proof of Theorem 1.

3. Proof of Theorem 3
In the usual notations we set in the introduction, let us define the moment generating
function
X∞
s h(log |Z|)j i j
(3.1) MN (s) = h|Z| i = s
j=0
j!

(where the mean has to be considered over the group U(N) with respect to the Haar
measure) and the cumulants Qj = Qj (N) by
∞
X Qj
(3.2) log MN (s) = sj .
j=1
j!

In [16], among other things, Keating and Snaith studied the cumulants showing that


0 if n = 1
1
(3.3) Qn = 2 log N + O(1) if n = 2

O(1) if n ≥ 3
and deduced a central limit theorem proving that the limiting distribution of log |Z| is
Gaussian with mean 0 and variance 21 log N. Here, for any k ∈ N, we study the distribution
of random variable log |Z| with respect to the tilted measure |Z|2k dHaar . Before starting
with our analysis, we recall that the moments of |Z| are known also for non integer k (see
[16] equation (6) and (16)):
X ∞  Y N
2k (2k)j Γ(j)Γ(j + 2k) 2
k 2 G (1 + k)
(3.4) MN (2k) = h|Z| i = exp Qj = ∼ N
j=1
j! j=1
Γ(j + k)2 G(1 + 2k)
QN Γ(j)Γ(j+2k)
where k ∈ R and G denotes the Barnes G-function. We denote M2k := j=1 Γ(j+k)2
.

Now we are ready to consider the first moment

   N 
2k d 2k+x d Y Γ(j)Γ(j + 2k + x)
h|Z| log |Z|i = h|Z| i =
dx x=0 dx j=1 Γ(j + k + x2 )2 x=0

by (3.1) and (3.4). We compute the derivative by Leibniz’s rule, writing

YN XN n o
Γ(j)Γ(j + 2k + x)
x 2 = exp log Γ(j) + log Γ(j + 2k + x) − 2 log Γ(j + k + x/2)
j=1
Γ(j + k + 2
) j=1

and we get
N
Y N  
2k Γ(j)Γ(j + 2k) X Γ′ Γ′
(3.5) h|Z| log |Z|i = (j + 2k) − (j + k) .
j=1
Γ(j + k)2 j=1
Γ Γ
14
Moreover an application of Stirling’s formula yields
 
Γ′ Γ′ k 1
(3.6) (j + 2k) − (j + k) = + Ok 2
Γ Γ j j
hence, by (3.5) and (3.6), the weighted mean of the random variable log |Z| is
1
µ2k := h|Z|2k log |Z|i = k log N + Ok (1).
M2k
Then we study the weighted n-th moment of the random variable log |Z|:
X n
2k n
h|Z| (log|Z| − µ2k ) i = (−µ2k )j h|Z|2k (log |Z|)h i
h+j=n
h

X n dj     N
dh Y Γ(j)Γ(j + 2k + x)

−xµ2k
(3.7) = e
h+j=n
h dxj
x=0 dx
h
j=1
Γ(j + k + x2 )2 x=0
  XN  
dn Γ(j)Γ(j + 2k + x)
= n exp − xµ2k + log .
dx j=1
Γ(j + k + x2 )2 x=0

If we denote fj (x) = fN,k,j (x) := log Γ(j) + log Γ(j + 2k + x) − 2 log Γ(j + k + x2 ), then we
can carry on the computation in (3.7) by computing the derivative, getting
(3.8)
  X N  
dn Γ(j)Γ(j + 2k + x)
exp − xµ2k + log
dxn j=1
Γ(j + k + x2 )2 x=0

X PN Yn   X N  mi
n! fj (0) 1 di
= e j=1
i
− xµ2k + fj (x)
m1 ,...,mn
m 1 ! · · · mn ! i=1
i! dx j=1 x=0

X  X N m1  X N m2 Y n  X N mi

n! ′ 1 ′′ 1 (i)
= M2k − µ2k + fj (0) f (0) f (0)
m ,...,m
m1 ! · · · mn ! j=1
2 j=1 j i=3
i! j=1 j
1 n

where the sums in (3.8) are over the n-uple (m1 , . . . , mn ) such that
m1 + 2m2 + · · · + nmn = n.
Using Stirling’s approximation formula, one can easily estimate the derivatives of fj (x)
and prove
N
X N n
X o
k
fj′ (0) = + Ok (j −2
) = k log N + Ok (1);
j=1 j=1
j
N
X N n
X o 1
1 1/2
(3.9) fj′′ (0) = − + O(j −2 ) = log N + Ok (1);
j=1 j=1
j + 2k j + k 2
N
X N
X
(i)
fj (0) = O(j −2) = O(1) for all i ≥ 3.
j=1 j=1
15
Putting together (3.7), (3.8) and (3.9) one has
h|Z|2k (log |Z| − µ2k )n i
X  m2  m1 +m3 +···+mn
n! 1
= M2k log N + Ok (1) Ok (1)
m1 +2m2 +···+nmn =n
m1 ! · · · mn ! 4

then if n is even the asymptotic is given by m2 = n/2 and mi = 0 for i 6= 2, giving

n!  1 n/2 1 n/2
h|Z|2k (log |Z| − µ2k )n i ∼k,n M2k log N = M2k (n − 1)!! log N
(n/2)! 4 2
while if n is odd the n-th moment is surely ok,n (M2k (log N)n/2 ).

Acknowledgments. I am grateful to Sandro Bettin for his support and encouragement

as well as for many useful suggestions and to Jon Keating for helpful conversations and
for pointing me out interesting connections with the random matrix theory setting, which
inspired, among other things, the last section of this paper. I also wish to thank the
referee for a very careful reading of the paper and for indicating several inaccuracies.

References
[1] L.-P. Arguin, D. Belius, P. Bourgade, M. Radziwill, K. Soundararajan. Maximum of the Riemann
zeta function on a short interval of the critical line. Comm. Pure Appl. Math. 72, no. 3, pp 500-535,
2019.
[2] R. Balasubramanian, J. B. Conrey, D. R. Heath-Brown. Asymptotic mean square of the product of
the Riemann zeta-function and a Dirichlet polynomial. J. Reine Angew. Math. 357 (1985), 161–181.
MR 783539 (87f:11061).
[3] S. Bettin. The second moment of the Riemann zeta function with unbounded shifts. Int. J. Number
Theory 6 (2010), no. 8, 1933-1944.
[4] S. Bettin, H. M. Bui, X. Li, M. Radziwill. A quadratic divisor problem and moments of the Riemann
zeta-function (preprint). ArXiv: 1609.02539 [math.NT], 2016.
[5] S. Bettin, V. Chandee, M. Radziwill. The mean square of the product of the Riemann zeta function
with Dirichlet polynomials. J. Reine Angew. Math. 729 (2017), 51–79.
[6] J.B. Conrey, D.W. Farmer, J.P. Keating, M.O. Rubinstein, N.C. Snaith. Integral moments of L-
functions, Proc. Lond. Math. Soc. 91 (2005), 33–104.
[7] J.B. Conrey, A. Ghosh. Zeros of derivatives of the Riemann zeta-function near the critical line.
Analytic number theory (Allerton Park, IL, 1989), 95–110, Progr. Math., 85, 1990.
[8] J.B. Conrey, M.O. Rubinstein, N.C. Snaith. Moments of the derivative of characteristic polynomi-
als with an application to the Riemann zeta-function. Comm. Math. Phys., 267(3):611–629, 2006,
arXiv:math.NT/0508378.
[9] A. Fazzari. A weighted central limit theorem for log |ζ(1/2 + it)|. Mathematika, to appear.
[10] S.M. Gonek. Mean values of the Riemann zeta-function and its derivatives. Inventiones Math.
75(1984),123-141.
[11] S.M. Gonek. A formula of Landau and Mean Values of ζ(s). In Topics in Analytic Number Theory
(edited by Graham, S. W. and Vaaler, J. D.) 92–97.
[12] A. Harper. Sharp conditional bounds for moments of the Riemann zeta functions (preprint).
Arxiv.org/abs/1305.4618v1.
[13] C.P. Hughes, M.P. Young. The twisted fourth moment of the Riemann zeta function, J. Reine Angew.
Math. 641 (2010), 203–236. MR 2643931 (2011h:11092).
[14] C.P. Hughes. On the characteristic polynomial of a random unitary matrix and the Riemann zeta
function. PhD thesis, University of Bristol, 2001.
16
[15] A.E. Ingham. Mean-values theorems in the theory of the Riemann zeta-function”, Proc. Lond. Math.
Soc. 27 (1926) 273–300.
[16] J. Keating, N. Snaith. Random Matrix Theory and ζ(1/2 + it). Commun. Math. Phys. 214, 57–89
(2000).
[17] S. Kirila. An upper bound for discrete moments of the derivative of the Riemann zeta-function.
arXiv:1804.08826v2 [math.NT] 2020
[18] M. Milinovich. Upper bounds for moments of ζ ′ (ρ). Bull. Lond. Math. Soc. 42 (2010), no. 1, 28–44.
[19] M. Radziwill. Large deviations in Selberg’s central limit theorem. arXiv:1108.5092 (2011).
[20] A. Selberg. The Riemann zeta function on vertical arithmetic progressions. International Mathemat-
ics Research Notices, Volume 2015, Issue 2, 2015, Pages 325–354.
[21] K. Soundararajan. Moments of the Riemann zeta function. Annals of Math., 170 (2):981 – 993, 2009.
[22] K. M. Tsang. The distribution of the values of the Riemann zeta function. PhD thesis, Princeton
University, Princeton, NJ, 1984.

Università di Genova, Dipartimento di Matematica. Via Dodecaneso 35, 16146 Genova,

Italy
Email address: fazzari@dima.unige.it

17