Polynomial Chaos and Scaling Limits of Disordered Systems 4. Free - - PowerPoint PPT Presentation

polynomial chaos and scaling limits of disordered systems
SMART_READER_LITE
LIVE PREVIEW

Polynomial Chaos and Scaling Limits of Disordered Systems 4. Free - - PowerPoint PPT Presentation

Pinning models Weak disorder regime The marginally relevant regime Polynomial Chaos and Scaling Limits of Disordered Systems 4. Free energy estimates. Introduction to marginal relevance. Francesco Caravenna Universit` a degli Studi di


slide-1
SLIDE 1

Pinning models Weak disorder regime The marginally relevant regime

Polynomial Chaos and Scaling Limits of Disordered Systems

  • 4. Free energy estimates. Introduction to marginal relevance.

Francesco Caravenna

Universit` a degli Studi di Milano-Bicocca Levico Terme ∼ September 30 - October 2, 2015

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 1 / 26

slide-2
SLIDE 2

Pinning models Weak disorder regime The marginally relevant regime

Overview

In the previous lecture we constructed continuum partition functions ZW and we used them to define a continuum disordered model PW

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 2 / 26

slide-3
SLIDE 3

Pinning models Weak disorder regime The marginally relevant regime

Overview

In the previous lecture we constructed continuum partition functions ZW and we used them to define a continuum disordered model PW In this lecture we show how the continuum objects ZW and PW yield quantitative information on the discrete model (free energy estimates) We will focus on Pinning models (rather than DPRE)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 2 / 26

slide-4
SLIDE 4

Pinning models Weak disorder regime The marginally relevant regime

Overview

In the previous lecture we constructed continuum partition functions ZW and we used them to define a continuum disordered model PW In this lecture we show how the continuum objects ZW and PW yield quantitative information on the discrete model (free energy estimates) We will focus on Pinning models (rather than DPRE) In the last part we will introduce marginally relevant models (Pinning for α = 1

2, DPRE for d = 2, 2d Stochastic Heat Equation)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 2 / 26

slide-5
SLIDE 5

Pinning models Weak disorder regime The marginally relevant regime

Outline

  • 1. Pinning models
  • 2. Weak disorder regime
  • 3. The marginally relevant regime

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 3 / 26

slide-6
SLIDE 6

Pinning models Weak disorder regime The marginally relevant regime

Outline

  • 1. Pinning models
  • 2. Weak disorder regime
  • 3. The marginally relevant regime

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 4 / 26

slide-7
SLIDE 7

Pinning models Weak disorder regime The marginally relevant regime

Ingredients: renewal process & disorder

0 = τ0 τ1 τ2 τ3 τ4 τ5 τ6 Discrete renewal process τ = {0 = τ0 < τ1 < τ2 < . . .} ⊆ N0 Gaps (τi+1 − τi)i≥0 are i.i.d. with polynomial-tail distribution: Pref(τ1 = n) ∼ cK n1+α , cK > 0, α ∈ (0, 1)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 5 / 26

slide-8
SLIDE 8

Pinning models Weak disorder regime The marginally relevant regime

Ingredients: renewal process & disorder

0 = τ0 τ1 τ2 τ3 τ4 τ5 τ6 Discrete renewal process τ = {0 = τ0 < τ1 < τ2 < . . .} ⊆ N0 Gaps (τi+1 − τi)i≥0 are i.i.d. with polynomial-tail distribution: Pref(τ1 = n) ∼ cK n1+α , cK > 0, α ∈ (0, 1) τ = {n ∈ N0 : Sn = 0} zero level set of a Markov chain S = (Sn)n≥0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 5 / 26

slide-9
SLIDE 9

Pinning models Weak disorder regime The marginally relevant regime

Ingredients: renewal process & disorder

0 = τ0 τ1 τ2 τ3 τ4 τ5 τ6 Discrete renewal process τ = {0 = τ0 < τ1 < τ2 < . . .} ⊆ N0 Gaps (τi+1 − τi)i≥0 are i.i.d. with polynomial-tail distribution: Pref(τ1 = n) ∼ cK n1+α , cK > 0, α ∈ (0, 1) τ = {n ∈ N0 : Sn = 0} zero level set of a Markov chain S = (Sn)n≥0 Disorder ω = (ωi)i∈N: i.i.d. real random variables with law P λ(β) := log E[eβω1] < ∞ E[ω1] = 0 Var[ω1] = 1

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 5 / 26

slide-10
SLIDE 10

Pinning models Weak disorder regime The marginally relevant regime

Bessel random walks

For α ∈ (0, 1) the α-Bessel random walk is defined as follows:

prob.

1 2

x prob.

1 2

  • 1 + cα

x

  • Sn

prob.

1 2

prob.

1 2

  • 1 − cα

x

  • cα := 1

2 − α

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 6 / 26

slide-11
SLIDE 11

Pinning models Weak disorder regime The marginally relevant regime

Bessel random walks

For α ∈ (0, 1) the α-Bessel random walk is defined as follows:

prob.

1 2

x prob.

1 2

  • 1 + cα

x

  • Sn

prob.

1 2

prob.

1 2

  • 1 − cα

x

  • cα := 1

2 − α ◮ (α = 1 2) no drift (cα = 0)

  • simple random walk

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 6 / 26

slide-12
SLIDE 12

Pinning models Weak disorder regime The marginally relevant regime

Bessel random walks

For α ∈ (0, 1) the α-Bessel random walk is defined as follows:

prob.

1 2

x prob.

1 2

  • 1 + cα

x

  • Sn

prob.

1 2

prob.

1 2

  • 1 − cα

x

  • cα := 1

2 − α ◮ (α = 1 2) no drift (cα = 0)

  • simple random walk

◮ (α < 1 2) drift away from the origin (cα > 0)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 6 / 26

slide-13
SLIDE 13

Pinning models Weak disorder regime The marginally relevant regime

Bessel random walks

For α ∈ (0, 1) the α-Bessel random walk is defined as follows:

prob.

1 2

x prob.

1 2

  • 1 + cα

x

  • Sn

prob.

1 2

prob.

1 2

  • 1 − cα

x

  • cα := 1

2 − α ◮ (α = 1 2) no drift (cα = 0)

  • simple random walk

◮ (α < 1 2) drift away from the origin (cα > 0) ◮ (α > 1 2) drift toward the origin (cα < 0)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 6 / 26

slide-14
SLIDE 14

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N

rewards ωn > 0 penalties ωn < 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-15
SLIDE 15

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N Free renewal

rewards ωn > 0 penalties ωn < 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-16
SLIDE 16

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N Pinning model

rewards ωn > 0 penalties ωn < 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-17
SLIDE 17

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N Pinning model

rewards ωn > 0 penalties ωn < 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-18
SLIDE 18

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N Pinning model

rewards ωn > 0 penalties ωn < 0

N ∈ N (system size) β ≥ 0, h ∈ R (disorder strength, bias)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-19
SLIDE 19

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N Pinning model

rewards ωn > 0 penalties ωn < 0

N ∈ N (system size) β ≥ 0, h ∈ R (disorder strength, bias)

The pinning model

Gibbs change of measure Pω

N = Pω N,β,h of the renewal distribution Pref

dPω

N

dPref (τ) := 1 Z ω

N

exp

  • N
  • n=1

(βωn + h − λ(β)

  • 1{n∈τ}
  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-20
SLIDE 20

Pinning models Weak disorder regime The marginally relevant regime

Disordered pinning model

N Pinning model

rewards ωn > 0 penalties ωn < 0

N ∈ N (system size) β ≥ 0, h ∈ R (disorder strength, bias)

The pinning model

Gibbs change of measure Pω

N = Pω N,β,h of the renewal distribution Pref

dPω

N

dPref (τ) := 1 Z ω

N

exp

  • N
  • n=1

(βωn + h − λ(β)

  • 1{Sn=0}
  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 7 / 26

slide-21
SLIDE 21

Pinning models Weak disorder regime The marginally relevant regime

The phase transition

How are the typical paths τ of the pinning model Pω

N?

Contact number CN :=

  • τ ∩ (0, N]
  • = N

n=1 1{n∈τ} = N n=1 1{Sn=0}

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 8 / 26

slide-22
SLIDE 22

Pinning models Weak disorder regime The marginally relevant regime

The phase transition

How are the typical paths τ of the pinning model Pω

N?

Contact number CN :=

  • τ ∩ (0, N]
  • = N

n=1 1{n∈τ} = N n=1 1{Sn=0}

Theorem (phase transition)

∃ continuous, non decreasing, deterministic critical curve hc(β):

◮ Localized regime: for h > hc(β) one has CN ≈ N ◮ Deocalized regime: for h < hc(β) one has CN = O(log N)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 8 / 26

slide-23
SLIDE 23

Pinning models Weak disorder regime The marginally relevant regime

The phase transition

How are the typical paths τ of the pinning model Pω

N?

Contact number CN :=

  • τ ∩ (0, N]
  • = N

n=1 1{n∈τ} = N n=1 1{Sn=0}

Theorem (phase transition)

∃ continuous, non decreasing, deterministic critical curve hc(β):

◮ Localized regime: for h > hc(β) one has CN ≈ N

∃µ = µβ,h > 0 : Pω

N

  • CN

N − µ

  • > ε

− − − →

N→∞

ω–a.s.

◮ Deocalized regime: for h < hc(β) one has CN = O(log N)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 8 / 26

slide-24
SLIDE 24

Pinning models Weak disorder regime The marginally relevant regime

The phase transition

How are the typical paths τ of the pinning model Pω

N?

Contact number CN :=

  • τ ∩ (0, N]
  • = N

n=1 1{n∈τ} = N n=1 1{Sn=0}

Theorem (phase transition)

∃ continuous, non decreasing, deterministic critical curve hc(β):

◮ Localized regime: for h > hc(β) one has CN ≈ N

∃µ = µβ,h > 0 : Pω

N

  • CN

N − µ

  • > ε

− − − →

N→∞

ω–a.s.

◮ Deocalized regime: for h < hc(β) one has CN = O(log N)

∃A = Aβ,h > 0 : Pω

N

CN log N > A

− − − →

N→∞

ω–a.s.

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 8 / 26

slide-25
SLIDE 25

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-26
SLIDE 26

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-27
SLIDE 27

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-28
SLIDE 28

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

◮ (α < 1 2) disorder is irrelevant: hc(β) = 0 for β > 0 small [Alexander] [Toninelli] [Lacoin] [Cheliotis, den Hollander]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-29
SLIDE 29

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

◮ (α < 1 2) disorder is irrelevant: hc(β) = 0 for β > 0 small [Alexander] [Toninelli] [Lacoin] [Cheliotis, den Hollander] ◮ (α ≥ 1 2) disorder is relevant: hc(β) > 0 for all β > 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-30
SLIDE 30

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

◮ (α < 1 2) disorder is irrelevant: hc(β) = 0 for β > 0 small [Alexander] [Toninelli] [Lacoin] [Cheliotis, den Hollander] ◮ (α ≥ 1 2) disorder is relevant: hc(β) > 0 for all β > 0

◮ (α > 1)

hc(β) ∼ C β2 with explicit C =

α 1+α 1 2E(τ1)

[Berger, C., Poisat, Sun, Zygouras]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-31
SLIDE 31

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

◮ (α < 1 2) disorder is irrelevant: hc(β) = 0 for β > 0 small [Alexander] [Toninelli] [Lacoin] [Cheliotis, den Hollander] ◮ (α ≥ 1 2) disorder is relevant: hc(β) > 0 for all β > 0

◮ (α > 1)

hc(β) ∼ C β2 with explicit C =

α 1+α 1 2E(τ1)

[Berger, C., Poisat, Sun, Zygouras]

◮ ( 1

2 < α < 1)

C1 β

2α 2α−1 ≤ hc(β) ≤ C2 β 2α 2α−1

[Derrida, Giacomin, Lacoin, Toninelli] [Alexander, Zygouras]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-32
SLIDE 32

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

◮ (α < 1 2) disorder is irrelevant: hc(β) = 0 for β > 0 small [Alexander] [Toninelli] [Lacoin] [Cheliotis, den Hollander] ◮ (α ≥ 1 2) disorder is relevant: hc(β) > 0 for all β > 0

◮ (α > 1)

hc(β) ∼ C β2 with explicit C =

α 1+α 1 2E(τ1)

[Berger, C., Poisat, Sun, Zygouras]

◮ ( 1

2 < α < 1)

C1 β

2α 2α−1 ≤ hc(β) ≤ C2 β 2α 2α−1

[Derrida, Giacomin, Lacoin, Toninelli] [Alexander, Zygouras]

◮ (α = 1

2) hc(β) = e − c+o(1)

β2

[Giacomin, Lacoin, Toninelli] [Berger, Lacoin]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-33
SLIDE 33

Pinning models Weak disorder regime The marginally relevant regime

Estimates on the critical curve

For β = 0 (homogeneous pinning, no disorder) one has hc(0) = 0 What is the behavior of hc(β) for β > 0 small ?

Theorem ( P(τ1 = n) ∼

cK n1+α )

◮ (α < 1 2) disorder is irrelevant: hc(β) = 0 for β > 0 small [Alexander] [Toninelli] [Lacoin] [Cheliotis, den Hollander] ◮ (α ≥ 1 2) disorder is relevant: hc(β) > 0 for all β > 0

◮ (α > 1)

hc(β) ∼ C β2 with explicit C =

α 1+α 1 2E(τ1)

[Berger, C., Poisat, Sun, Zygouras]

◮ ( 1

2 < α < 1)

hc(β) ∼ ˆ C β

2α 2α−1

using continuum model!

[C., Torri, Toninelli]

◮ (α = 1

2) hc(β) = e − c+o(1)

β2

[Giacomin, Lacoin, Toninelli] [Berger, Lacoin]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 9 / 26

slide-34
SLIDE 34

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-35
SLIDE 35

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-36
SLIDE 36

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s.

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-37
SLIDE 37

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-38
SLIDE 38

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-39
SLIDE 39

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅) ∼ (const.)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-40
SLIDE 40

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅) ∼ (const.)

Nα ◮ Critical curve

hc(β) = sup{h ∈ R : F(β, h) = 0} non analiticity!

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-41
SLIDE 41

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅) ∼ (const.)

Nα ◮ Critical curve

hc(β) = sup{h ∈ R : F(β, h) = 0} non analiticity! (convexity) ∂F(β, h) ∂h = lim

N→∞ Eω N

CN N

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-42
SLIDE 42

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅) ∼ (const.)

Nα ◮ Critical curve

hc(β) = sup{h ∈ R : F(β, h) = 0} non analiticity! (convexity) ∂F(β, h) ∂h = lim

N→∞ Eω N

CN N > 0 if h > hc(β) = 0 if h < hc(β)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-43
SLIDE 43

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅) ∼ (const.)

Nα ◮ Critical curve

hc(β) = sup{h ∈ R : F(β, h) = 0} non analiticity! (convexity) ∂F(β, h) ∂h = lim

N→∞ Eω N

CN N > 0 if h > hc(β) = 0 if h < hc(β) F(β, h) and hc(β) depend on the law of τ and ω Universality as β, h → 0 ?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-44
SLIDE 44

Pinning models Weak disorder regime The marginally relevant regime

Discrete free energy and critical curve

Partition function Zω

N := E

  • eHN(τ)

= E

  • e

N

n=1(h+βωn−Λ(β))1{n∈τ}

  • Consider first the regime of N → ∞ with fixed β, h

◮ Free energy

F(β, h) := lim

N→∞ 1 N log Zω N ≥ 0

P(dω)-a.s. Zω

N ≥ E

  • eHN(τ) 1{τ∩(0,N]=∅}
  • = P(τ ∩ (0, N] = ∅) ∼ (const.)

Nα ◮ Critical curve

hc(β) = sup{h ∈ R : F(β, h) = 0} non analiticity! (convexity) ∂F(β, h) ∂h = lim

N→∞ Eω N

CN N > 0 if h > hc(β) = 0 if h < hc(β) F(β, h) and hc(β) depend on the law of τ and ω Universality as β, h → 0 ? YES, connected to continuum model

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 10 / 26

slide-45
SLIDE 45

Pinning models Weak disorder regime The marginally relevant regime

A word on critical exponents

The free energy F(β, h) is non analytic at the critical point h = hc(β) F(β, h) = 0 (h < hc(β)) F(β, h) > 0 (h > hc(β)) What is the behavior of F(β, h) as h ↓ hc(β) ?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 11 / 26

slide-46
SLIDE 46

Pinning models Weak disorder regime The marginally relevant regime

A word on critical exponents

The free energy F(β, h) is non analytic at the critical point h = hc(β) F(β, h) = 0 (h < hc(β)) F(β, h) > 0 (h > hc(β)) What is the behavior of F(β, h) as h ↓ hc(β) ? For β = 0 the model is exactly solvable: hc(0) = 0 and F(0, h) − F(0, hc(0)) ∼ C (h − hc(0))

1 α

(α ∈ (0, 1))

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 11 / 26

slide-47
SLIDE 47

Pinning models Weak disorder regime The marginally relevant regime

A word on critical exponents

The free energy F(β, h) is non analytic at the critical point h = hc(β) F(β, h) = 0 (h < hc(β)) F(β, h) > 0 (h > hc(β)) What is the behavior of F(β, h) as h ↓ hc(β) ? For β = 0 the model is exactly solvable: hc(0) = 0 and F(0, h) − F(0, hc(0)) ∼ C (h − hc(0))

1 α

(α ∈ (0, 1))

Smoothing inequality

[Giacomin, Toninelli]

F(β, h) − F(β, hc(β)) ≤ C β2 (h − hc(0))2

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 11 / 26

slide-48
SLIDE 48

Pinning models Weak disorder regime The marginally relevant regime

A word on critical exponents

The free energy F(β, h) is non analytic at the critical point h = hc(β) F(β, h) = 0 (h < hc(β)) F(β, h) > 0 (h > hc(β)) What is the behavior of F(β, h) as h ↓ hc(β) ? For β = 0 the model is exactly solvable: hc(0) = 0 and F(0, h) − F(0, hc(0)) ∼ C (h − hc(0))

1 α

(α ∈ (0, 1))

Smoothing inequality

[Giacomin, Toninelli]

F(β, h) − F(β, hc(β)) ≤ C β2 (h − hc(0))2

◮ For α > 1 2 disorder makes phase transition smoother!

Also hc(β) = hc(0) for every β > 0 disorder relevance

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 11 / 26

slide-49
SLIDE 49

Pinning models Weak disorder regime The marginally relevant regime

A word on critical exponents

The free energy F(β, h) is non analytic at the critical point h = hc(β) F(β, h) = 0 (h < hc(β)) F(β, h) > 0 (h > hc(β)) What is the behavior of F(β, h) as h ↓ hc(β) ? For β = 0 the model is exactly solvable: hc(0) = 0 and F(0, h) − F(0, hc(0)) ∼ C (h − hc(0))

1 α

(α ∈ (0, 1))

Smoothing inequality

[Giacomin, Toninelli]

F(β, h) − F(β, hc(β)) ≤ C β2 (h − hc(0))2

◮ For α > 1 2 disorder makes phase transition smoother!

Also hc(β) = hc(0) for every β > 0 disorder relevance

◮ For α < 1 2 and for β > 0 small F(β, h) ≈ F(0, h) irrelevance

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 11 / 26

slide-50
SLIDE 50

Pinning models Weak disorder regime The marginally relevant regime

Outline

  • 1. Pinning models
  • 2. Weak disorder regime
  • 3. The marginally relevant regime

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 12 / 26

slide-51
SLIDE 51

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Build continuum partition functions for Pinning Model with α ∈ ( 1

2, 1)

(disorder relevant) following “usual” approach [C, Sun, Zygouras 2015+]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 13 / 26

slide-52
SLIDE 52

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Build continuum partition functions for Pinning Model with α ∈ ( 1

2, 1)

(disorder relevant) following “usual” approach [C, Sun, Zygouras 2015+] We need to rescale β = βN = ˆ β Nα−1/2 h = hN = ˆ h Nα

(Note that hN ≈ β

2α 2α−1

N

≈ hc(βN))

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 13 / 26

slide-53
SLIDE 53

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Build continuum partition functions for Pinning Model with α ∈ ( 1

2, 1)

(disorder relevant) following “usual” approach [C, Sun, Zygouras 2015+] We need to rescale β = βN = ˆ β Nα−1/2 h = hN = ˆ h Nα

(Note that hN ≈ β

2α 2α−1

N

≈ hc(βN))

One has Zω

N d

− − − − →

N→∞

ZW with ZW := 1 + C

  • 0<t<1

dW

ˆ β,ˆ h t

t1−α + C 2

  • 0<t<t′<1

dW

ˆ β,ˆ h t

dW

ˆ β,ˆ h t′

t1−α(t′ − t)1−α + . . . where W

ˆ β,ˆ h t

:= ˆ βW t + ˆ h t and C = α sin(απ)

π cK

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 13 / 26

slide-54
SLIDE 54

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Exercise

Recalling that Pref(n ∈ τ) = Pref(Sn = 0) ∼ c n1−α check that βN and hN are the correct scaling (polynomial chaos)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 14 / 26

slide-55
SLIDE 55

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Exercise

Recalling that Pref(n ∈ τ) = Pref(Sn = 0) ∼ c n1−α check that βN and hN are the correct scaling (polynomial chaos) Like for DPRE we build constrained partition functions: 0 ≤ s < t < ∞ ZW s, t

  • = scaling limit of Eref

eHω

[Ns,Nt]1{Nt∈τ}

  • Ns ∈ τ
  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 14 / 26

slide-56
SLIDE 56

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Exercise

Recalling that Pref(n ∈ τ) = Pref(Sn = 0) ∼ c n1−α check that βN and hN are the correct scaling (polynomial chaos) Like for DPRE we build constrained partition functions: 0 ≤ s < t < ∞ ZW s, t

  • = scaling limit of Eref

eHω

[Ns,Nt]1{Nt∈τ}

  • Ns ∈ τ
  • We show that they satisfy continuity, strict positivity, semigroup

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 14 / 26

slide-57
SLIDE 57

Pinning models Weak disorder regime The marginally relevant regime

Continuum partition functions

Exercise

Recalling that Pref(n ∈ τ) = Pref(Sn = 0) ∼ c n1−α check that βN and hN are the correct scaling (polynomial chaos) Like for DPRE we build constrained partition functions: 0 ≤ s < t < ∞ ZW s, t

  • = scaling limit of Eref

eHω

[Ns,Nt]1{Nt∈τ}

  • Ns ∈ τ
  • We show that they satisfy continuity, strict positivity, semigroup

Theorem

[C., Sun, Zygouras 2015+b]

We can build a continuum disordered Pinning model PW as a random probability law on the space of closed subsets of [0, 1]

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 14 / 26

slide-58
SLIDE 58

Pinning models Weak disorder regime The marginally relevant regime

Continuum free energy

In analogy with the discrete model, define Continuum free energy F(ˆ β, ˆ h) := lim

t→∞

1 t log ZW

ˆ β,ˆ h(0, t)

a.s. (existence and self-averaging need some work)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 15 / 26

slide-59
SLIDE 59

Pinning models Weak disorder regime The marginally relevant regime

Continuum free energy

In analogy with the discrete model, define Continuum free energy F(ˆ β, ˆ h) := lim

t→∞

1 t log ZW

ˆ β,ˆ h(0, t)

a.s. (existence and self-averaging need some work) Again F(ˆ β, ˆ h) ≥ 0 and define Continuum critical curve Hc(ˆ β) := sup ˆ h ∈ R : F(ˆ β, ˆ h) = 0

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 15 / 26

slide-60
SLIDE 60

Pinning models Weak disorder regime The marginally relevant regime

Continuum free energy

In analogy with the discrete model, define Continuum free energy F(ˆ β, ˆ h) := lim

t→∞

1 t log ZW

ˆ β,ˆ h(0, t)

a.s. (existence and self-averaging need some work) Again F(ˆ β, ˆ h) ≥ 0 and define Continuum critical curve Hc(ˆ β) := sup ˆ h ∈ R : F(ˆ β, ˆ h) = 0

  • Scaling relations

∀c > 0 : ZW

ˆ β,ˆ h(c t) d

= ZW

cα− 1

2 ˆ

β,cαˆ h(t)

(Wiener chaos exp.)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 15 / 26

slide-61
SLIDE 61

Pinning models Weak disorder regime The marginally relevant regime

Continuum free energy

In analogy with the discrete model, define Continuum free energy F(ˆ β, ˆ h) := lim

t→∞

1 t log ZW

ˆ β,ˆ h(0, t)

a.s. (existence and self-averaging need some work) Again F(ˆ β, ˆ h) ≥ 0 and define Continuum critical curve Hc(ˆ β) := sup ˆ h ∈ R : F(ˆ β, ˆ h) = 0

  • Scaling relations

∀c > 0 : ZW

ˆ β,ˆ h(c t) d

= ZW

cα− 1

2 ˆ

β,cαˆ h(t)

(Wiener chaos exp.)

F

  • cα− 1

2 ˆ

β, cα = c F(ˆ β, ˆ h)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 15 / 26

slide-62
SLIDE 62

Pinning models Weak disorder regime The marginally relevant regime

Continuum free energy

In analogy with the discrete model, define Continuum free energy F(ˆ β, ˆ h) := lim

t→∞

1 t log ZW

ˆ β,ˆ h(0, t)

a.s. (existence and self-averaging need some work) Again F(ˆ β, ˆ h) ≥ 0 and define Continuum critical curve Hc(ˆ β) := sup ˆ h ∈ R : F(ˆ β, ˆ h) = 0

  • Scaling relations

∀c > 0 : ZW

ˆ β,ˆ h(c t) d

= ZW

cα− 1

2 ˆ

β,cαˆ h(t)

(Wiener chaos exp.)

F

  • cα− 1

2 ˆ

β, cα = c F(ˆ β, ˆ h) Hc(ˆ β) = Hc(1) ˆ β

2α 2α−1 Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 15 / 26

slide-63
SLIDE 63

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-64
SLIDE 64

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-65
SLIDE 65

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Assuming uniform integrability of log Zω (OK) F(ˆ β, ˆ h) = lim

t→∞

1 t E

  • log ZW

ˆ β,ˆ h(t)

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-66
SLIDE 66

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Assuming uniform integrability of log Zω (OK) F(ˆ β, ˆ h) = lim

t→∞

1 t E

  • log ZW

ˆ β,ˆ h(t)

  • = lim

t→∞

1 t lim

N→∞ E

  • log Zω

βN,hN(Nt)

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-67
SLIDE 67

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Assuming uniform integrability of log Zω (OK) F(ˆ β, ˆ h) = lim

t→∞

1 t E

  • log ZW

ˆ β,ˆ h(t)

  • = lim

t→∞

1 t lim

N→∞ E

  • log Zω

βN,hN(Nt)

  • Assuming we can interchange the limits N → ∞ and t → ∞

F(ˆ β, ˆ h) = lim

N→∞

lim

t→∞

1 t E

  • log Zω

βN,hN(Nt)

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-68
SLIDE 68

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Assuming uniform integrability of log Zω (OK) F(ˆ β, ˆ h) = lim

t→∞

1 t E

  • log ZW

ˆ β,ˆ h(t)

  • = lim

t→∞

1 t lim

N→∞ E

  • log Zω

βN,hN(Nt)

  • Assuming we can interchange the limits N → ∞ and t → ∞

F(ˆ β, ˆ h) = lim

N→∞ N lim t→∞

1 Nt E

  • log Zω

βN,hN(Nt)

  • Francesco Caravenna

Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-69
SLIDE 69

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Assuming uniform integrability of log Zω (OK) F(ˆ β, ˆ h) = lim

t→∞

1 t E

  • log ZW

ˆ β,ˆ h(t)

  • = lim

t→∞

1 t lim

N→∞ E

  • log Zω

βN,hN(Nt)

  • Assuming we can interchange the limits N → ∞ and t → ∞

F(ˆ β, ˆ h) = lim

N→∞ N lim t→∞

1 Nt E

  • log Zω

βN,hN(Nt)

  • =

lim

N→∞ N F(βN, hN)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-70
SLIDE 70

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Can we relate continuum free energy to the discrete one? By construction of continuum partition functions ZW

ˆ β,ˆ h(t) d

= lim

N→∞ Zω βN,hN(Nt)

Assuming uniform integrability of log Zω (OK) F(ˆ β, ˆ h) = lim

t→∞

1 t E

  • log ZW

ˆ β,ˆ h(t)

  • = lim

t→∞

1 t lim

N→∞ E

  • log Zω

βN,hN(Nt)

  • Assuming we can interchange the limits N → ∞ and t → ∞

F(ˆ β, ˆ h) = lim

N→∞ N lim t→∞

1 Nt E

  • log Zω

βN,hN(Nt)

  • =

lim

N→∞ N F(βN, hN)

Setting δ = 1

N for clarity, we arrive at. . .

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 16 / 26

slide-71
SLIDE 71

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Conjecture

F(ˆ β, ˆ h) = lim

δ→0

F ˆ βδα− 1

2 , ˆ

hδα δ

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 17 / 26

slide-72
SLIDE 72

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Conjecture

F(ˆ β, ˆ h) = lim

δ→0

F ˆ βδα− 1

2 , ˆ

hδα δ

Theorem

[C., Toninelli, Torri 2015]

For all ˆ β > 0, ˆ h ∈ R and η > 0 there is δ0 > 0 such that ∀δ < δ0 F(ˆ β, ˆ h − η) ≤ F ˆ βδα− 1

2 , ˆ

hδα δ ≤ F(ˆ β, ˆ h + η)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 17 / 26

slide-73
SLIDE 73

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Conjecture

F(ˆ β, ˆ h) = lim

δ→0

F ˆ βδα− 1

2 , ˆ

hδα δ

Theorem

[C., Toninelli, Torri 2015]

For all ˆ β > 0, ˆ h ∈ R and η > 0 there is δ0 > 0 such that ∀δ < δ0 F(ˆ β, ˆ h − η) ≤ F ˆ βδα− 1

2 , ˆ

hδα δ ≤ F(ˆ β, ˆ h + η) This implies Conj. and hc(β) ∼ Hc(β) ∼ Hc(1) β

2α 2α−1

For any discrete Pinning model with α ∈ ( 1

2, 1), the free energy F(β, h)

and the critical curve hc(β) have a universal shape in the regime β, h → 0

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 17 / 26

slide-74
SLIDE 74

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Very delicate result. How to prove it?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 18 / 26

slide-75
SLIDE 75

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Very delicate result. How to prove it?

◮ Assume that there is a continuum Hamiltonian:

Zω = E

  • eHω

Nt

ZW = E

  • eHW

t Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 18 / 26

slide-76
SLIDE 76

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Very delicate result. How to prove it?

◮ Assume that there is a continuum Hamiltonian:

Zω = E

  • eHω

Nt

ZW = E

  • eHW

t

◮ Couple Hω Nt and HW t

  • n the same probability space in such a way

that the difference ∆N,t := Hω

Nt − HW t

is “small”

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 18 / 26

slide-77
SLIDE 77

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Very delicate result. How to prove it?

◮ Assume that there is a continuum Hamiltonian:

Zω = E

  • eHω

Nt

ZW = E

  • eHW

t

◮ Couple Hω Nt and HW t

  • n the same probability space in such a way

that the difference ∆N,t := Hω

Nt − HW t

is “small”

◮ Deduce that

E

  • log Zω

≤ E

  • log ZW

+ log EE

  • e∆N,t

and show that the last term is “negligible”

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 18 / 26

slide-78
SLIDE 78

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Very delicate result. How to prove it?

◮ Assume that there is a continuum Hamiltonian:

Zω = E

  • eHω

Nt

ZW = E

  • eHW

t

◮ Couple Hω Nt and HW t

  • n the same probability space in such a way

that the difference ∆N,t := Hω

Nt − HW t

is “small”

◮ Deduce that

E

  • log Zω

≤ E

  • log ZW

+ log EE

  • e∆N,t

and show that the last term is “negligible” Problem: there is no continuum Hamiltonian!

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 18 / 26

slide-79
SLIDE 79

Pinning models Weak disorder regime The marginally relevant regime

Interchanging the limits

Very delicate result. How to prove it?

◮ Assume that there is a continuum Hamiltonian:

Zω = E

  • eHω

Nt

ZW = E

  • eHW

t

◮ Couple Hω Nt and HW t

  • n the same probability space in such a way

that the difference ∆N,t := Hω

Nt − HW t

is “small”

◮ Deduce that

E

  • log Zω

≤ E

  • log ZW

+ log EE

  • e∆N,t

and show that the last term is “negligible” Problem: there is no continuum Hamiltonian! Solution: perform coarse-graining and define an “effective” Hamiltonian

(drawing!)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 18 / 26

slide-80
SLIDE 80

Pinning models Weak disorder regime The marginally relevant regime

The DPRE case

What about the DPRE? We can still define discrete F(β) and continuum F(ˆ β) free energy Since F(ˆ β) ∼ F(1) β4 we can hope that F(β) ∼ F(1) β4 as β → 0 provided the “interchanging of limits” is justified

  • N. Torri is currently working on this problem. A finer coarse-graining is

needed, together with sharper estimates on continuum partition functions

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 19 / 26

slide-81
SLIDE 81

Pinning models Weak disorder regime The marginally relevant regime

Outline

  • 1. Pinning models
  • 2. Weak disorder regime
  • 3. The marginally relevant regime

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 20 / 26

slide-82
SLIDE 82

Pinning models Weak disorder regime The marginally relevant regime

The marginal case

We consider simultaneously different models that are marginally relevant:

◮ Pinning Models with α = 1 2 ◮ DPRE with d = 2 (RW attracted to BM) ◮ DPRE with d = 1 (RW with Cauchy tails: P(|S1| > n) ∼ c n ◮ Stochastic Heat Equation in d = 2

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 21 / 26

slide-83
SLIDE 83

Pinning models Weak disorder regime The marginally relevant regime

The marginal case

We consider simultaneously different models that are marginally relevant:

◮ Pinning Models with α = 1 2 ◮ DPRE with d = 2 (RW attracted to BM) ◮ DPRE with d = 1 (RW with Cauchy tails: P(|S1| > n) ∼ c n ◮ Stochastic Heat Equation in d = 2

All these different models share a crucial feature: logarithmic overlap RN =         

  • 1≤n≤N

Pref(n ∈ τ)2

  • 1≤n≤N
  • x∈Zd

Pref(Sn = x)2 ∼ C log N

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 21 / 26

slide-84
SLIDE 84

Pinning models Weak disorder regime The marginally relevant regime

The marginal case

We consider simultaneously different models that are marginally relevant:

◮ Pinning Models with α = 1 2 ◮ DPRE with d = 2 (RW attracted to BM) ◮ DPRE with d = 1 (RW with Cauchy tails: P(|S1| > n) ∼ c n ◮ Stochastic Heat Equation in d = 2

All these different models share a crucial feature: logarithmic overlap RN =         

  • 1≤n≤N

Pref(n ∈ τ)2

  • 1≤n≤N
  • x∈Zd

Pref(Sn = x)2 ∼ C log N More generally: RN diverges as a slowly varying function

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 21 / 26

slide-85
SLIDE 85

Pinning models Weak disorder regime The marginally relevant regime

The marginal case

Analogy between Pinning model with α = 1

2 and DPRE with d = 2

Pin = 1 + N

  • n=1

Pref(n ∈ τ) X n + . . . Zω

DPRE = 1 + N

  • n=1

x∈Z2

Pref(Sn = x) X n,x

  • + . . .

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 22 / 26

slide-86
SLIDE 86

Pinning models Weak disorder regime The marginally relevant regime

The marginal case

Analogy between Pinning model with α = 1

2 and DPRE with d = 2

Pin = 1 + N

  • n=1

Pref(n ∈ τ) X n + . . . Zω

DPRE = 1 + N

  • n=1

x∈Z2

Pref(Sn = x) X n,x

  • + . . .

Note that Pref(Sn = x) ∼ 1

n g1( x √n) (recall that d = 2)

Then the random variable in parenthesis has variance

  • x∈Z2

Pref(Sn = x)2 ∼ 1 n 1 n

  • x∈Z2

g1 x

√n

2 ∼ g12

2

n hence we can replace it by g12

√n X n Pinning! (Pref(n ∈ τ) ∼ c √n)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 22 / 26

slide-87
SLIDE 87

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

For computation we focus on Pinning model (for simplicity h = 0) Look at relevant case α > 1

2 (Pref(n ∈ τ) ∼ 1 n1−α )

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 23 / 26

slide-88
SLIDE 88

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

For computation we focus on Pinning model (for simplicity h = 0) Look at relevant case α > 1

2 (Pref(n ∈ τ) ∼ 1 n1−α )

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

k

  • i=1

Pref(ni − ni−1 ∈ τ) X ni

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 23 / 26

slide-89
SLIDE 89

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

For computation we focus on Pinning model (for simplicity h = 0) Look at relevant case α > 1

2 (Pref(n ∈ τ) ∼ 1 n1−α )

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

k

  • i=1

Pref(ni − ni−1 ∈ τ) X ni = 1 +

N

  • k=1
  • 0<n1<...<nk≤N

k

  • i=1

X ni (ni − ni−1)1−α

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 23 / 26

slide-90
SLIDE 90

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

For computation we focus on Pinning model (for simplicity h = 0) Look at relevant case α > 1

2 (Pref(n ∈ τ) ∼ 1 n1−α )

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

k

  • i=1

Pref(ni − ni−1 ∈ τ) X ni =

N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk n1−α

1

(n2 − n1)1−α · · · (nk − nk−1)1−α

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 23 / 26

slide-91
SLIDE 91

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

For computation we focus on Pinning model (for simplicity h = 0) Look at relevant case α > 1

2 (Pref(n ∈ τ) ∼ 1 n1−α )

β ∼

ˆ β Nα− 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

k

  • i=1

Pref(ni − ni−1 ∈ τ) X ni =

N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk n1−α

1

(n2 − n1)1−α · · · (nk − nk−1)1−α =

N

  • k=0

β √ N N1−α k

  • 0<n1<...<nk≤N

1 √ N X n1 1 √ N X n2 · · · 1 √ N X nk

( n1

N )1−α( n2 N − n1 N )1−α · · · ( nk N − nk−1 N )1−α

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 23 / 26

slide-92
SLIDE 92

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

For computation we focus on Pinning model (for simplicity h = 0) Look at relevant case α > 1

2 (Pref(n ∈ τ) ∼ 1 n1−α )

β ∼

ˆ β Nα− 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

k

  • i=1

Pref(ni − ni−1 ∈ τ) X ni =

N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk n1−α

1

(n2 − n1)1−α · · · (nk − nk−1)1−α =

N

  • k=0

β √ N N1−α k

  • 0<n1<...<nk≤N

1 √ N X n1 1 √ N X n2 · · · 1 √ N X nk

( n1

N )1−α( n2 N − n1 N )1−α · · · ( nk N − nk−1 N )1−α d

− − − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dW t1 dW t2 · · · dW tk t1−α

1

(t2 − t1)1−α · · · (tk − tk−1)1−α For α = 1

2 last step breaks down 1 √t ∈ L2([0, 1])

How to make sense?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 23 / 26

slide-93
SLIDE 93

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

Always in the relevant case α > 1

2

(setting χ = 2(1 − α) < 1 ) Var[Zω

N] −

− − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 24 / 26

slide-94
SLIDE 94

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

Always in the relevant case α > 1

2

(setting χ = 2(1 − α) < 1 ) Var[Zω

N] −

− − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ

  • k=0

ˆ βk Γ(1 − χ)k+1 Γ((1 − χ)(k + 1))

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 24 / 26

slide-95
SLIDE 95

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

Always in the relevant case α > 1

2

(setting χ = 2(1 − α) < 1 ) Var[Zω

N] −

− − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ

  • k=0

ˆ βk Γ(1 − χ)k+1 Γ((1 − χ)(k + 1)) ≤

  • k=0

ˆ βk ck

1

(c2k)! < ∞

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 24 / 26

slide-96
SLIDE 96

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

Always in the relevant case α > 1

2

(setting χ = 2(1 − α) < 1 ) Var[Zω

N] −

− − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ

  • k=0

ˆ βk Γ(1 − χ)k+1 Γ((1 − χ)(k + 1)) ≤

  • k=0

ˆ βk ck

1

(c2k)! < ∞ The k! makes the series converge for all ˆ β > 0 It arises from the constraint 0 < t1 < . . . < tk ≤ 1

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 24 / 26

slide-97
SLIDE 97

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

Always in the relevant case α > 1

2

(setting χ = 2(1 − α) < 1 ) Var[Zω

N] −

− − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ

  • k=0

ˆ βk Γ(1 − χ)k+1 Γ((1 − χ)(k + 1)) ≤

  • k=0

ˆ βk ck

1

(c2k)! < ∞ The k! makes the series converge for all ˆ β > 0 It arises from the constraint 0 < t1 < . . . < tk ≤ 1

Exercise

Prove “by bare hands” that

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ ≤ e−Ck log k

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 24 / 26

slide-98
SLIDE 98

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

Always in the relevant case α > 1

2

(setting χ = 2(1 − α) < 1 ) Var[Zω

N] −

− − − →

N→∞ ∞

  • k=0

ˆ βk

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ

  • k=0

ˆ βk Γ(1 − χ)k+1 Γ((1 − χ)(k + 1)) ≤

  • k=0

ˆ βk ck

1

(c2k)! < ∞ The k! makes the series converge for all ˆ β > 0 It arises from the constraint 0 < t1 < . . . < tk ≤ 1

Exercise

Prove “by bare hands” that (probabilistic argument!)

  • 0<t1<...<tk≤1

dt1 dt2 · · · dtk tχ

1 (t2 − t1)χ · · · (tk − tk−1)χ ≤ e−Ck log k

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 24 / 26

slide-99
SLIDE 99

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

In the marginal regime α = 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk √n1 √n2 − n1 · · · √nk − nk−1

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 25 / 26

slide-100
SLIDE 100

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

In the marginal regime α = 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk √n1 √n2 − n1 · · · √nk − nk−1 = 1 + β

  • 0<n≤N

X n √n + β2

  • 0<n<n′≤N

X n X n′ √n √ n′ − n + . . .

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 25 / 26

slide-101
SLIDE 101

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

In the marginal regime α = 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk √n1 √n2 − n1 · · · √nk − nk−1 = 1 + β

  • 0<n≤N

X n √n + β2

  • 0<n<n′≤N

X n X n′ √n √ n′ − n + . . . Goal: find the joint limit in distribution of all these sums

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 25 / 26

slide-102
SLIDE 102

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

In the marginal regime α = 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk √n1 √n2 − n1 · · · √nk − nk−1 = 1 + β

  • 0<n≤N

X n √n + β2

  • 0<n<n′≤N

X n X n′ √n √ n′ − n + . . . Goal: find the joint limit in distribution of all these sums Linear term is easy (X n ∼ N(0, 1) by Lindeberg): asympt. N(0, σ2) σ2 = β2

0<n≤N

1 n ∼ β2 log N

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 25 / 26

slide-103
SLIDE 103

Pinning models Weak disorder regime The marginally relevant regime

Relevant vs. marginal regime

In the marginal regime α = 1

2

N = N

  • k=0

βk

  • 0<n1<...<nk≤N

X n1X n2 · · · X nk √n1 √n2 − n1 · · · √nk − nk−1 = 1 + β

  • 0<n≤N

X n √n + β2

  • 0<n<n′≤N

X n X n′ √n √ n′ − n + . . . Goal: find the joint limit in distribution of all these sums Linear term is easy (X n ∼ N(0, 1) by Lindeberg): asympt. N(0, σ2) σ2 = β2

0<n≤N

1 n ∼ β2 log N We then rescale β = βN ∼ ˆ β √log N Other terms converge?

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 25 / 26

slide-104
SLIDE 104

Pinning models Weak disorder regime The marginally relevant regime

References

◮ F. Caravenna, F.L. Toninelli, N. Torri

Universality for the pinning model in the weak coupling regime Preprint (2015)

◮ G. Giacomin

Random Polymer Models Imperial College Press, London (2007)

◮ G. Giacomin

Disorder and critical phenomena through basic probability models

  • St. Flour Lecture Notes, Springer (2010)

◮ F. den Hollander

Random Polymers

  • St. Flour Lecture Notes, Springer (2007)

Francesco Caravenna Scaling Limits of Disordered Systems Sep 30 - Oct 2, 2015 26 / 26