Phase transitions on the scaling limits of the
symmetric slowed exclusion
Patr´ıcia Gon¸calves PUC - Rio de Janeiro
University of Nice 17 October 2013
joint work with
♣ Tertuliano Franco (Federal University of Bahia),
Outline
1. Slowed exclusion process: generator and invariant states.
2. Scaling limits for the density of particles: Hydrodynamics and Fluctuations.
3. Central limit theorem for the current and tagged particle.
4. Occupation time.
Slowed exclusion process
• ηt is a Markov process with space stateΩ = {0, 1}Tn.
• for x ∈ Tn,η(x) = 1 if the site is occupied, otherwise η(x) = 0.
• The generator is given on f : Ω → R by: Lnf(η) =
x∈Tn
anx,x+1f(ηx,x+1) − f (η),
where ηx,x+1 is the configuration obtained fromη by exchanging the occupation variables η(x) and η(x + 1).
• Let νρbe the Bernoulli product measures of constant parameter
ρ ∈ [0, 1]: the occupation variables (η(x))x are independent and
Slowed exclusion process
Above for β ∈ [0, ∞] and α > 0, we consider
anx,x+1 =
αn−β, if x = −1 ,
1, otherwise . Then:
Hydrodynamic Limit
For each configuration η we denote by πn(η; du) the empirical measure: πn(η; du) = 1 n x∈Tn η(x)δx/n(du)
and πtn(η, du) := πn(ηtn2, du).
Assume a L.L.N. for {π0n}n∈N to ρ0(u)du under the initial
distribution of the system, then for any t > 0 the L.L.N. holds for
{πn
t}n∈N to ρ(t, u)du under the corresponding distribution of the system at time t, where ρ(t, u) evolves according to the
Hydrodynamic Limit: initial state
Definition
Fixρ0: T → [0, 1]. A sequence of probability measures {μn}n∈N on Ω is said to be associated to a profile ρ0 : T → [0, 1] if, for every
δ > 0 and every continuous function H : T → R, it holds that
lim n→∞μn η : 1 n x∈Tn H(xn) η(x) − TH(u) ρ0(u)du > δ = 0 .
Hydrodynamic Limit: Law of Large Numbers
Theorem
Fix β ∈ [0, ∞] and ρ0 : T → [0, 1], and let {μn}n∈N be associated
to ρ0. Then, for any t ∈ [0, T ], for every δ > 0 and every
continuous function H : T → R: lim n→∞Pμn η.: 1n x∈Tn Hxnηtn2(x)− TH(u)ρ(t, u)du > δ = 0 , where:
for β ∈ [0, 1), ρ(t, ·) is the unique weak solution of (1); for β = 1, ρ(t, ·) is the unique weak solution of (2); for β ∈ (1, ∞], ρ(t, ·) is the unique weak solution of (3).
Hydrodynamic Limit: Law of Large Numbers
Theorem
Fix β ∈ [0, ∞] and ρ0 : T → [0, 1], and let {μn}n∈N be associated
to ρ0. Then, for any t ∈ [0, T ], for every δ > 0 and every
continuous function H : T → R: lim n→∞Pμn η.: 1n x∈Tn Hxnηtn2(x)− TH(u)ρ(t, u)du > δ = 0 , where:
for β ∈ [0, 1), ρ(t, ·) is the unique weak solution of (1);
for β = 1, ρ(t, ·) is the unique weak solution of (2); for β ∈ (1, ∞], ρ(t, ·) is the unique weak solution of (3).
Hydrodynamic Limit: Law of Large Numbers
Theorem
Fix β ∈ [0, ∞] and ρ0 : T → [0, 1], and let {μn}n∈N be associated
to ρ0. Then, for any t ∈ [0, T ], for every δ > 0 and every
continuous function H : T → R: lim n→∞Pμn η.: 1n x∈Tn Hxnηtn2(x)− TH(u)ρ(t, u)du > δ = 0 , where:
for β ∈ [0, 1), ρ(t, ·) is the unique weak solution of (1); for β = 1, ρ(t, ·) is the unique weak solution of (2);
Hydrodynamic Limit: Law of Large Numbers
Theorem
Fix β ∈ [0, ∞] and ρ0 : T → [0, 1], and let {μn}n∈N be associated
to ρ0. Then, for any t ∈ [0, T ], for every δ > 0 and every
continuous function H : T → R: lim n→∞Pμn η.: 1n x∈Tn Hxnηtn2(x)− TH(u)ρ(t, u)du > δ = 0 , where:
for β ∈ [0, 1), ρ(t, ·) is the unique weak solution of (1); for β = 1, ρ(t, ·) is the unique weak solution of (2); for β ∈ (1, ∞], ρ(t, ·) is the unique weak solution of (3).
Heat equation with periodic boundary conditions:
∂tρ(t, u) = Δρ(t, u) , t ≥ 0, u ∈ T ,
ρ(0, u) = ρ0(u), u∈ T . (1)
Definition
ρ : [0, T ] × T → [0, 1] is a weak solution if for any t ∈ [0, T ] and
any H ∈ C1,2([0, T ] × T),
Tρ(t, u)H(t, u) − ρ(0, u)H(0, u)du
− t 0 Tρ(s, u) ∂sH(s, u) + ΔH(s, u) du ds = 0 .
Heat equation with a type of Robin’s boundary conditions:
⎧ ⎪ ⎨ ⎪ ⎩ ∂tρ(t, u) = Δρ(t, u) , t ≥ 0, u ∈ (0, 1) , ∂uρ(t, 0)=∂uρ(t, 1) =α(ρ(t, 0)−ρ(t, 1)), t ≥ 0, ρ(0, u) = ρ0(u), u ∈ (0, 1) . (2) Definitionρ : [0, T ] × T → [0, 1] is a weak solution if ρ ∈ L2(0, T ; H1) and for all t ∈ [0, T ] and for all H ∈ C1,2([0, T ] × [0, 1]),
Tρ(t, u)H(t, u) − ρ(0, u)H(0, u)du − t 0 Tρ(s, u)∂sH(s, u)duds − t 0 Tρ(s, u)ΔH(s, u)duds − t 0(ρs(0)∂u Hs(0)−ρs(1)∂uHs(1))ds + t 0 α(ρs(0) − ρs(1))(Hs(0) − Hs(1))ds = 0.
Heat equation with Neumann’s boundary conditions:
⎧ ⎪ ⎨ ⎪ ⎩ ∂tρ(t, u) = Δρ(t, u) , t ≥ 0, u ∈ (0, 1) , ∂uρ(t, 0) = ∂uρ(t, 1) = 0 , t ≥ 0 , ρ(0, u) = ρ0(u), u ∈ (0, 1) . (3) Definitionρ : [0, T ] × T → [0, 1] is a weak solution if ρ ∈ L2(0, T ; H1) and for all t ∈ [0, T ] and for all H ∈ C1,2([0, T ] × [0, 1]),
Tρ(t, u)H(t, u) − ρ(0, u)H(0, u)du
− t 0 Tρ(s, u) ∂sH(s, u) + ΔH(s, u) du ds − t 0(ρs(0)∂u Hs(0) − ρs(1)∂uHs(1)) ds = 0 .
Equilibrium fluctuations
Take the process with state space {0, 1}Z and fixρ ∈ [0, 1]. For
t ∈ [0, T ], define the density fluctuation field on functions f ∈ Sβ(R) as Yn t(f ) := √1n x∈Z f x n {ηtn2(x) − ρ}. Theorem
The sequence of processes {Ytn}n∈N converges in distribution, as n goes to ∞, with respect to the Skorohod topology of
D([0, T ], S
β(R)) to a gaussian process Yt in C([0, T ], Sβ(R)),
which is the stationary solution of the Ornstein-Uhlenbeck equation
dYt = ΔβYtdt +
2χ(ρ)∇βdBt, where Bt is aSβ(R)-valued Brownian motion.
Equilibrium fluctuations
This means that the trajectories of Yt are in C([0, T ], Sβ(R)), Y0 is a white noise of variance χ(ρ), namely if for any f ∈ Sβ(R), the real-valued random variable Y0(f ) has a normal distribution of mean zero and variance χ(ρ) f 2.
Definition
For β ∈ [0, ∞], we define the operators Δβ and∇β onSβ(R) by
∇βH(u) = H(1)(u), if u = 0 , H(1)(0+), if u = 0 , and ΔβH(u) = H(2)(u), if u = 0 , H(2)(0+), if u = 0 ,
Equilibrium fluctuations
Definition
Let S(R\{0}) be the space of functions H ∈ C∞(R\{0}), that are continuous from the right at x = 0, with
H k, := supx∈R\{0}|(1 + |x|) H(k)(x)| < ∞ , for all integers
k, ≥ 0, and H(k)(0−) = H(k)(0+), for all k integer, k ≥ 1. Let Sβ(R) be the subset of S(R\{0}) composed of functions H
satisfying
For β ∈ [0, 1), H(0−) = H(0+) .
For β = 1, H(1)(0+) = H(1)(0−) = α(H(0+) − H(0−)) . For β ∈ (1, +∞], H(1)(0+) = H(1)(0−) = 0 .
Equilibrium fluctuations
Definition
Let S(R\{0}) be the space of functions H ∈ C∞(R\{0}), that are continuous from the right at x = 0, with
H k, := supx∈R\{0}|(1 + |x|) H(k)(x)| < ∞ , for all integers
k, ≥ 0, and H(k)(0−) = H(k)(0+), for all k integer, k ≥ 1. Let Sβ(R) be the subset of S(R\{0}) composed of functions H
satisfying
For β ∈ [0, 1), H(0−) = H(0+) .
For β = 1, H(1)(0+) = H(1)(0−) = α(H(0+) − H(0−)) . For β ∈ (1, +∞], H(1)(0+) = H(1)(0−) = 0 .
Equilibrium fluctuations
Definition
Let S(R\{0}) be the space of functions H ∈ C∞(R\{0}), that are continuous from the right at x = 0, with
H k, := supx∈R\{0}|(1 + |x|) H(k)(x)| < ∞ , for all integers
k, ≥ 0, and H(k)(0−) = H(k)(0+), for all k integer, k ≥ 1. Let Sβ(R) be the subset of S(R\{0}) composed of functions H
satisfying
For β ∈ [0, 1), H(0−) = H(0+) .
For β = 1, H(1)(0+) = H(1)(0−) = α(H(0+) − H(0−)) . For β ∈ (1, +∞], H(1)(0+) = H(1)(0−) = 0 .
Equilibrium fluctuations
Definition
Let S(R\{0}) be the space of functions H ∈ C∞(R\{0}), that are continuous from the right at x = 0, with
H k, := supx∈R\{0}|(1 + |x|) H(k)(x)| < ∞ , for all integers
k, ≥ 0, and H(k)(0−) = H(k)(0+), for all k integer, k ≥ 1. Let Sβ(R) be the subset of S(R\{0}) composed of functions H
satisfying
For β ∈ [0, 1), H(0−) = H(0+) .
For β = 1, H(1)(0+) = H(1)(0−) = α(H(0+) − H(0−)) . For β ∈ (1, +∞], H(1)(0+) = H(1)(0−) = 0 .
The current
For u ∈ R define the current through the bond {un − 1, un}, as Jun(t), that is Jun(t) counts the total number of jumps from the site un − 1 to the site un minus the total number of jumps from the site un to the site un − 1 in the time interval [0, tn2].
+
-The current
TheoremFor every t ≥ 0 and every u ∈ R, Ju√n(t)
n −−−→n→∞ Ju(t) in the sense of
finite-dimensional distributions, where Ju(t) is a gaussian process with mean zero and variance given by
• for β ∈ [0, 1), Eνρ[(Ju(t))2] = 2χ(ρ) t π; • for β = 1, Eνρ[(Ju(t))2] = 2χ(ρ) t π +Φ2t(2u+4αt) e 4αu+4α2 t−Φ2 t(2u) 2α ; • for β ∈ (1, +∞], Eνρ[(Ju(t))2] = 2χ(ρ) t π 1− e−u2/t + 2u Φ2t(2u) , where Φ2t(x) := +∞ x e−u2/4t √ 4πt du.
A tagged particle
Start {ηt : t ≥ 0} from νρ conditioned to have a particle at the site
un. If Xn
u(t) is the position at the time tn2 of that tagged
particle, then {Xun(t) ≥ k} =Jun(t) ≥ un+k−1 x=un ηtn2(x) .
A tagged particle
Theorem
For allβ ∈ [0, ∞], every u ∈ R and t ≥ 0 X√un(t)
n −−−−→n→+∞ Xu(t) in
the sense of finite-dimensional distributions, where Xu(t) = Ju(t)/ρ in law. In particular, the variance of Xu(t) is given by
• for β ∈ [0, 1), Eνρ[(Xu(t))2] = 2 χ(ρ) ρ2 t π; • for β = 1, Eνρ[(Xu(t))2] = 2 χ(ρ) ρ2 t π + Φ2t(2u+4αt) e 4αu+4α2 t 2α ; • for β ∈ (1, +∞], Eνρ[(Xu(t))2] = 2 χ(ρ) ρ2 t π 1− e−u2/t + 2u Φ2t(2u) .
Occupation time
Definition
The occupation time of the origin is defined as Γn(t) := 1
n3/2 tn2
Occupation time
Theorem
As n goes to infinity, the sequence of processes
{Γn(t) : t ∈ [0, T ]}n∈N converges in distribution with respect to
the uniform of C([0, T ], R) to a mean-zero gaussian process {Γ(t) : t ∈ [0, T ]}, with variance given by:
• for β ∈ [0, 1), EΓ(t)2 = 4 3χ(ρ)√πt3/2. • for β = 1, EΓ(t)2= 4 3 χ(ρ)√ π t3/2+ 2√χ(ρ) 4πt t 0 s 0 Fα(s − r)drds where Fα(t) = 1 2t +∞ 0 z e−z2/4t−2αz dz • for β ∈ (1, +∞), EΓ(t)2= 8 3χ(ρ)√π t3/2.
Hydrodynamic Equations
Theorem
For α > 0, let ρα: [0, T ] × [0, 1] → [0, 1] be the unique weak solution of the heat equation with Robin’s boundary conditions:
⎧ ⎪ ⎨ ⎪ ⎩ ∂tρα(t, u) = Δρα(t, u) , ∂uρα(t, 0) = ∂uρα(t, 1) = α(ρα(t, 0) − ρα(t, 1)) , ρα(0, u) = ρ0(u).
Then, limα→0ρα = ρ0 and lim
α→∞ρα = ρ∞, in
L2([0, T ] × [0, 1]), where ρ0 is the unique weak solution of the heat equation with Neumann’s boundary conditions and ρ∞ is the unique weak solution of the heat equation with periodic boundary conditions.
References
Franco, T.; Gonc¸alves, P.; Neumann, A. Hydrodynamical
behavior of symmetric exclusion with slow bonds, Annales de
l’Institut Henri Poincar´e: Probability and Statistics, 49, no. 2, 402–427 (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase Transition
of a Heat Equation with Robin’s Boundary Conditions and Exclusion Process, to appear in Transactions of the American
Mathematical Society (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase transition
in equilibrium fluctuations of symmetric slowed exclusion,
Stochastic Processes and their Applications, 123, no. 12, 4156–4185, (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Occupation times
References
Franco, T.; Gonc¸alves, P.; Neumann, A. Hydrodynamical
behavior of symmetric exclusion with slow bonds, Annales de
l’Institut Henri Poincar´e: Probability and Statistics, 49, no. 2, 402–427 (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase Transition
of a Heat Equation with Robin’s Boundary Conditions and Exclusion Process, to appear in Transactions of the American
Mathematical Society (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase transition
in equilibrium fluctuations of symmetric slowed exclusion,
Stochastic Processes and their Applications, 123, no. 12, 4156–4185, (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Occupation times
References
Franco, T.; Gonc¸alves, P.; Neumann, A. Hydrodynamical
behavior of symmetric exclusion with slow bonds, Annales de
l’Institut Henri Poincar´e: Probability and Statistics, 49, no. 2, 402–427 (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase Transition
of a Heat Equation with Robin’s Boundary Conditions and Exclusion Process, to appear in Transactions of the American
Mathematical Society (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase transition
in equilibrium fluctuations of symmetric slowed exclusion,
Stochastic Processes and their Applications, 123, no. 12, 4156–4185, (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Occupation times
References
Franco, T.; Gonc¸alves, P.; Neumann, A. Hydrodynamical
behavior of symmetric exclusion with slow bonds, Annales de
l’Institut Henri Poincar´e: Probability and Statistics, 49, no. 2, 402–427 (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase Transition
of a Heat Equation with Robin’s Boundary Conditions and Exclusion Process, to appear in Transactions of the American
Mathematical Society (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Phase transition
in equilibrium fluctuations of symmetric slowed exclusion,
Stochastic Processes and their Applications, 123, no. 12, 4156–4185, (2013).
Franco, T., Gonc¸alves, P. and Neumann, A.: Occupation times