Extreme Behaviour - How big can things get? - WarwickConjecture (Farmer, Gonek, Hughes) max t2[0;T]...
Transcript of Extreme Behaviour - How big can things get? - WarwickConjecture (Farmer, Gonek, Hughes) max t2[0;T]...
-
Extreme BehaviourHow big can things get?
Christopher Hughes
Venice, 9 May 2013
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 1 / 24
-
Motivation: The Riemann zeta function
The Riemann zeta function is of fundamental importance in numbertheory.
Understand the distribution of prime numbers.Prototypical L-function (Dirichlet, elliptic curve, modular form, etc).An interesting and challenging function to understand in its ownright.Universality (approximates any holomorphic function arbitrarilywell).Can be used as a “experimental” test-bed for certain physicalsystems.One of the most important open problems in modern mathematicsconcerns it.It’s popular!
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 2 / 24
-
The Riemann zeta function
2 4 6 8 10
-10
-5
5
10
A real version of the Riemann zeta function,plotted for 104 ≤ t ≤ 104 + 10
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 3 / 24
-
The Riemann zeta function
2 4 6 8 10
-10
-5
5
10
A real version of the Riemann zeta function,plotted for 1010 ≤ t ≤ 1010 + 10
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 3 / 24
-
Extreme values of zeta
Conjecture (Farmer, Gonek, Hughes)
maxt∈[0,T ]
|ζ(12 + it)| = exp(( 1√
2+ o(1)
)√log T log log T
)
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 4 / 24
-
Bounds on extreme values of zeta
Theorem (Littlewood; Ramachandra and Sankaranarayanan;Soundararajan; Chandee and Soundararajan)Under RH, there exists a C such that
maxt∈[0,T ]
|ζ(12 + it)| = O(
exp(
Clog T
log log T
))
Theorem (Montgomery; Balasubramanian and Ramachandra;Balasubramanian; Soundararajan)There exists a C′ such that
maxt∈[0,T ]
|ζ(12 + it)| = Ω
(exp
(C′√
log Tlog log T
))
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 5 / 24
-
Characteristic polynomials
Keating and Snaith modelled the Riemann zeta function with
ZUN (θ) := det(IN − UNe−iθ)
=N∏
n=1
(1− ei(θn−θ))
where UN is an N × N unitary matrix chosen with Haar measure.
The matrix size N is connected to the height up the critical line T via
N = logT2π
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 6 / 24
-
Characteristic polynomials
-10 -5 5
0.05
0.10
0.15
0.20
0.25
Graph of the value distribution of log |ζ( 12 + it)| around the 1020th zero (red),
against the probability density of log |ZUN (0)| with N = 42 (green).
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 7 / 24
-
An Euler-Hadamard hybrid
Theorem (Gonek, Hughes, Keating)A simplified form of our theorem is:
ζ(12 + it) = P(t ; X )Z (t ; X ) + errors
where
P(t ; X ) =∏p≤X
(1− 1
p12+it
)−1and
Z (t ; X ) = exp
(∑γn
Ci(|t − γn| log X )
)
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 8 / 24
-
An Euler-Hadamard hybrid: Primes only
Graph of |P(t + t0; X )|, with t0 = γ1012+40,with X = log t0 ≈ 26 (red) and X = 1000 (green).
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 9 / 24
-
An Euler-Hadamard hybrid: Zeros only
Graph of |Z (t + t0; X )|, with t0 = γ1012+40,with X = log t0 ≈ 26 (red) and X = 1000 (green).
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 10 / 24
-
An Euler-Hadamard hybrid: Primes and zeros
Graph of |ζ( 12 + i(t + t0))| (black) and |P(t + t0; X )Z (t + t0; X )|,with t0 = γ1012+40, with X = log t0 ≈ 26 (red) and X = 1000 (green).
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 11 / 24
-
RMT model for extreme values of zeta
Simply taking the largest value of a characteristic polynomial doesn’twork.
Split the interval [0,T ] up into
M =T log T
N
blocks, each containing approximately N zeros.Model each block with the characteristic polynomial of an N × Nrandom unitary matrix.Find the smallest K = K (M,N) such that choosing M independentcharacteristic polynomials of size N, almost certainly none of them willbe bigger than K .
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 12 / 24
-
RMT model for extreme values of zeta
Simply taking the largest value of a characteristic polynomial doesn’twork.Split the interval [0,T ] up into
M =T log T
N
blocks, each containing approximately N zeros.
Model each block with the characteristic polynomial of an N × Nrandom unitary matrix.Find the smallest K = K (M,N) such that choosing M independentcharacteristic polynomials of size N, almost certainly none of them willbe bigger than K .
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 12 / 24
-
RMT model for extreme values of zeta
Simply taking the largest value of a characteristic polynomial doesn’twork.Split the interval [0,T ] up into
M =T log T
N
blocks, each containing approximately N zeros.Model each block with the characteristic polynomial of an N × Nrandom unitary matrix.
Find the smallest K = K (M,N) such that choosing M independentcharacteristic polynomials of size N, almost certainly none of them willbe bigger than K .
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 12 / 24
-
RMT model for extreme values of zeta
Simply taking the largest value of a characteristic polynomial doesn’twork.Split the interval [0,T ] up into
M =T log T
N
blocks, each containing approximately N zeros.Model each block with the characteristic polynomial of an N × Nrandom unitary matrix.Find the smallest K = K (M,N) such that choosing M independentcharacteristic polynomials of size N, almost certainly none of them willbe bigger than K .
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 12 / 24
-
RMT model for extreme values of zeta
Note that
P{
max1≤j≤M
maxθ|Z
U(j)N(θ)| ≤ K
}= P
{maxθ|ZUN (θ)| ≤ K
}M
TheoremLet 0 < β < 2. If M = exp(Nβ), and if
K = exp(√(
1− 12β + ε)
log M log N)
then
P{
max1≤j≤M
maxθ|Z
U(j)N(θ)| ≤ K
}→ 1
as N →∞ for all ε > 0, but for no ε < 0.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 13 / 24
-
RMT model for extreme values of zeta
Recallζ(12 + it) = P(t ; X )Z (t ; X ) + errors
We showed that Z (t ; X ) can be modelled by characteristic polynomialsof size
N =log T
eγ log X
Therefore the previous theorem suggests
ConjectureIf X = log T , then
maxt∈[0,T ]
|Z (t ; X )| = exp(( 1√
2+ o(1)
)√log T log log T
)
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 14 / 24
-
RMT model for extreme values of zeta
Recallζ(12 + it) = P(t ; X )Z (t ; X ) + errors
We showed that Z (t ; X ) can be modelled by characteristic polynomialsof size
N =log T
eγ log X
Therefore the previous theorem suggests
ConjectureIf X = log T , then
maxt∈[0,T ]
|Z (t ; X )| = exp(( 1√
2+ o(1)
)√log T log log T
)
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 14 / 24
-
RMT model for extreme values of zeta
TheoremBy the PNT, if X = log T then for any t ∈ [0,T ],
P(t ; X ) = O
(exp
(C
√log T
log log T
))
Thus one is led to the max values conjecture
Conjecture
maxt∈[0,T ]
|ζ(12 + it)| = exp(( 1√
2+ o(1)
)√log T log log T
)
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 15 / 24
-
Distribution of the max of characteristic polynomials
Fyodorov, Hiary and Keating have recently studied the distribution ofthe maximum of a characteristic polynomial of a random unitary matrixvia freezing transitions in certain disordered landscapes withlogarithmic correlations. This mixture of rigorous and heuristiccalculation led to:
Conjecture (Fyodorov, Hiary and Keating)For large N,
log maxθ|ZUN (θ)| ∼ log N −
34
log log N + Y
where Y has the density P {Y ∈ dy} = 4e−2yK0(2e−y )dy
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 16 / 24
-
Distribution of the max of characteristic polynomials
-2 2 4
0.1
0.2
0.3
0.4
The probability density of Y
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 17 / 24
-
Distribution of the max of characteristic polynomials
This led them to conjecture that
maxT≤t≤T+2π
|ζ(12 + it)| ∼ exp(
log log(
T2π
)− 3
4log log log
(T2π
)+ Y
)with Y having (approximately) the same distribution as before.
This conjecture was backed up by a different argument of Harper,using random Euler products.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 18 / 24
-
Distribution of the max of characteristic polynomials
This led them to conjecture that
maxT≤t≤T+2π
|ζ(12 + it)| ∼ exp(
log log(
T2π
)− 3
4log log log
(T2π
)+ Y
)with Y having (approximately) the same distribution as before.This conjecture was backed up by a different argument of Harper,using random Euler products.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 18 / 24
-
Distribution of the max of characteristic polynomials
Distribution of −2 log maxt∈[T ,T+2π] |ζ( 12 + it)| (after rescaling to get theempirical variance to agree) based on 2.5× 108 zeros near T = 1028. Graph
by Ghaith Hiary, taken from Fyodorov-Keating.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 19 / 24
-
Distribution of the max of characteristic polynomials
Note thatP {Y ≥ K} ≈ 2Ke−2K
for large K .
However, one can show that if K/ log N →∞ but K � N� then
P{
maxθ
log |ZUN (θ)| ≥ K}
= exp(− K
2
log N(1 + o(1))
)Thus there must be a critical K (of the order log N) where theprobability that maxθ |ZU(θ)| ≈ K changes from looking like linearexponential decay to quadratic exponential decay.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 20 / 24
-
Distribution of the max of characteristic polynomials
Note thatP {Y ≥ K} ≈ 2Ke−2K
for large K .However, one can show that if K/ log N →∞ but K � N� then
P{
maxθ
log |ZUN (θ)| ≥ K}
= exp(− K
2
log N(1 + o(1))
)
Thus there must be a critical K (of the order log N) where theprobability that maxθ |ZU(θ)| ≈ K changes from looking like linearexponential decay to quadratic exponential decay.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 20 / 24
-
Distribution of the max of characteristic polynomials
Note thatP {Y ≥ K} ≈ 2Ke−2K
for large K .However, one can show that if K/ log N →∞ but K � N� then
P{
maxθ
log |ZUN (θ)| ≥ K}
= exp(− K
2
log N(1 + o(1))
)Thus there must be a critical K (of the order log N) where theprobability that maxθ |ZU(θ)| ≈ K changes from looking like linearexponential decay to quadratic exponential decay.
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 20 / 24
-
Moments of the local maxima
Theorem (Conrey and Ghosh)As T →∞
1N(T )
∑tn≤T
∣∣ζ(12 + itn)∣∣2 ∼ e2 − 52 log Twhere tn are the points of local maxima of |ζ(12 + it)|.
This should be compared with Hardy and Littlewood’s result
1T
∫ T0|ζ(12 + it)|
2dt ∼ log T
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 21 / 24
-
Moments of the local maxima
Recently Winn succeeding in proving a random matrix version of thisresult (in disguised form)
Theorem (Winn)As N →∞
E
[1N
N∑n=1
∣∣ZUN (φn)∣∣2k]∼ C(k)E
[∣∣ZUN (0)∣∣2k]where φn are the points of local maxima of
∣∣ZUN (θ)∣∣, and where C(k)can be given explicitly as a combinatorial sum involving Pochhammersymbols on partitions.
In particular,
E
[1N
N∑n=1
∣∣ZUN (φn)∣∣2]∼ e
2 − 52
N
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 22 / 24
-
Summary
We looked at the rate of growth of the Riemann zeta function onthe critical line.The Riemann zeta function can be written in terms of a product ofits zeros times a product over all primes.The product over zeros can be modelled by ZUN (θ).One argument required knowing the large deviations ofmaxθ |ZUN (θ)|.The distribution of maxθ |ZUN (θ)| is now known.Its moderate deviations are still under study.The moments of the local maxima are not much bigger thanordinary moments.Extreme behaviour is rare!
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 23 / 24
-
Bibliography
Growth of zeta function D. Farmer, S. Gonek and C. Hughes “The maximum size of L-functions” Journal für die reine undangewandte Mathematik (2007) 609 215–236
Upper bound on growth rate V. Chandee and K. Soundararajan “Bounding |ζ(1/2 + it)| on the Riemann hypothesis” Bull.London Math. Soc. (2011) 43 243–250
Lower bound on growth rate K. Soundararajan “Extreme values of zeta and L-functions” Mathematische Annalen (2008) 342467–486
Modelling zeta using characteristic polynomials J. Keating and N. Snaith “Random matrix theory and ζ(1/2 + it)” Comm. Math.Phys. (2000) 214 57–89
Euler-Hadamard product formula S. Gonek, C. Hughes and J. Keating, “A hybrid Euler-Hadamard product for the Riemann zetafunction” Duke Math. J. (2007) 136 507–549
Distribution of max values Y. Fyodorov, G. Hiary and J. Keating, “Freezing transition, characteristic polynomials of randommatrices, and the Riemann zeta-function” Phys. Rev. Lett. (2012) 108 170601
Y. Fyodorov and J. Keating, “Freezing transitions and extreme values: random matrix theory, ζ(1/2 + it),and disordered landscapes” (2012) arXiv:1211.6063
A. Harper, “A note on the maximum of the Riemann zeta function, and log-correlated random variables”(2013) arXiv:1304.0677
Moments at the local maxima J. Conrey and A. Ghosh, “A mean value theorem for the Riemann zeta-function at its relativeextrema on the critical line” J. Lond. Math. Soc. (1985) 32 193–202
B. Winn, “Derivative moments for characteristic polynomials from the CUE” Commun. Math. Phys. (2012)315 531–563
Christopher Hughes (University of York) Extreme Behaviour Venice, 9 May 2013 24 / 24