Talk Pham - BCAM - Basque Center for Applied

Transcription

Talk Pham - BCAM - Basque Center for Applied
PHYSICAL CHARACTERISATION OF THE
GENERALISED GRAY BROWNIAN MOTION
Pham Minh Tuan
Belgorod National Research University, Russia
in collaboration with
Daniel Molina-Garcı́a and Gianni Pagnini
Ikebasque and Basque Center for Applied Mathematics
Bilbao 2015
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Outline
1
Motivation
2
Definition of the ggBm with stationary increments
3
Definition of the ggBm with nonstationary increments
4
Physical characteristics of ggBm
a. Ensemble average mean square displacement
b. Ergodicity breaking parameter
c. Time average mean square displacement
5
Conclusion
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Motivation
Anomalous diffusion is observed in various physical systems like:
photonics, plasma physics, biological systems, etc.
1
Normal diffusion:
⟨X 2 (t)⟩ = Ct
2
Anomalous diffusion:
⟨X 2 (t)⟩ = Ct 𝛾
𝛾 ̸= 1
In order to provide a stochastic process that well describes an experiment
exhibiting anomalous diffusion, here we present an in-depth study of the
so-called ”generalised grey Brownian process with nonstationary
increments”.
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Definition of ggBm with stationary increments
Mura A. Pagnini G. 2008 Jour. of Phys. A: Math. & Theor.
√︀
X𝛽,H (t) = Λ𝛽 XH (t)
where Λ𝛽 is a positive random variable distributed according to the
one-side M-Wright/Mainardi function with 𝛽 ∈ (0, 1]
M𝛽 (x) =
∞
∑︁
(−1)k
k=0
k!
xk
,
Γ(1 − 𝛽(k + 1))
x ≥0
and XH (t) is the fBm with Hurst index H ∈ (0, 1) and
2 (t)⟩ = 2D t 2H
⟨X𝛽,H
H
[︂∫︁ t
1
)︀
XH (t) := (︀
(t − 𝜏 )H−1/2 dB(𝜏 )+
Γ H + 12
0
∫︁ 0 {︁
}︁
+
(t − 𝜏 )H−1/2 − (−𝜏 )H−1/2 dB(𝜏 )
−∞
⇒ the process has stationary increments
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
PDF evolution equation
Pagnini G. 2012 Fractional Calculus and Applied Analysis
The one-point one-time PDF of the X𝛽,H (t) is
(︂ )︂
|x|
1
.
P(x, t) = H M𝛽/2
2t
tH
This PDF obeys the following evolution equation
2
𝜕P
2
𝛽−1,1−𝛽 𝜕 P
= t 2H−1 D2H/𝛽
,
𝜕t
𝛽
𝜕x 2
where D𝜂𝜎,𝜇 is the Erdélyi-Kober fractional derivative defined as
)︂
n (︂
∏︁
)︀
1 d (︀ 𝜎+𝜇,𝜂−𝜇
𝜎,𝜇
D𝜂 𝜙(t) =
𝜎+𝜇+ t
I𝜂
𝜙(t)
𝜂 dt
j=1
where
I𝜂𝜎,𝜇
is the Erdélyi-Kober fractional integral operator
∫︁
t −𝜂(𝜎+𝜇) t 𝜎𝜂 𝜂
𝜎,𝜇
𝜏 (t − 𝜏 𝜂 )𝜇−1 𝜙(𝜏 )d(𝜏 𝜂 )
I𝜂 𝜙(t) =
Γ(𝜇)
0
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Definition of ggBm with nonstationary increments
Molina, Pham, Pagnini, 2015 Proceedings of the NOLASC 15
X𝛼,𝛽,H (t) =
√︀
t 𝛼 Λ𝛽 XH (t) ,
𝛼 ∈ [−1, ∞)
where Λ𝛽 is a positive random variable distributed according to the
one-side M-Wright/Mainardi function with 𝛽 ∈ (0, 1]
M𝛽 (x) =
∞
∑︁
(−1)k
k=0
k!
xk
,
Γ(1 − 𝛽(k + 1))
x ≥0
and XH (t) is the fBm with Hurst index H ∈ (0, 1) and
2 (t)⟩ = 2D t 2H .
⟨X𝛽,H
H
⇒ this process has nonstationary increments
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Trajectories and PDF of ggBm with nonstationary
increments H = 0.3 , 𝛽 = 0.3, 𝛼 = 0.3
150
histogram (N=10000)
100
x(t)
50
0
-50
-100
-150
0
100 200 300 400 500 600 700 800 900 1000 0
t
0.005
0.01
0.015
0.02
0.025
P(x, t=1000)
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Trajectories and PDF of ggBm with nonstationary
increments H = 0.3 , 𝛽 = 0.3, 𝛼 = −0.3
20
histogram (N=10000)
15
10
x(t)
5
0
-5
-10
-15
-20
0
100 200 300 400 500 600 700 800 900 1000 0
t
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
P(x, t=1000)
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Trajectories and PDF of ggBm with nonstationary
increments H = 0.3 , 𝛽 = 0.3, 𝛼 = 0
histogram (N=10000)
40
f2H,β(x,t=1000)
x(t)
20
0
-20
-40
0
100 200 300 400 500 600 700 800 900 1000 0
t
0.01
0.02
0.03
0.04
0.05
0.06
P(x, t=1000)
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Physical characteristics
From physical point of view, the second moment of the process X𝛼,𝛽,H (t)
is one of the most crucial observables
EAMSD (Ensemble Average Mean Square Displacement)
⟨︀ 2
⟩︀
X𝛼,𝛽,H (t) =
2DH
t 2H+𝛼 ,
Γ(1 + 𝛽)
𝛼 ∈ [−1; ∞)
Remark:
In comparison with the well-known fBm, the power law of time is
perturbed by 𝛼.
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Ergodicity Breaking parameter
A parameter characterising the ergodicity of the studied process is the EB
(ergodicity breaking parameter).
Definition
)︁2 ⟩
𝛿X2 (ta , T , Δ)
⟨
⟩2 − 1
𝛿X2 (ta , T , Δ)
⟨(︁
EB = lim
T →∞
where T , Δ, ta stand for measurement time, lag time and aging time,
respectively. Remark:
Since this limit does not depend on the value of ta , so to simplify
notations, we consider 𝛿X2 (T , Δ) ≡ 𝛿X2 (ta , T , Δ).
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Temporal-Average MSD
The denominator of EB
TAMSD
∫︁
𝛿X2 (ta , T , Δ) =
⟨
⟩
𝛿X2 (ta , T , Δ)
≃
ta +T −Δ [︁
]︁2
X𝛼,𝛽,H (𝜏 + Δ) − X𝛼,𝛽,H (𝜏 ) d𝜏
ta
Δ≪ta ≪T
T −Δ
⎧
T𝛼
⎪
⎪
K
⎪
⎨ 1 + 𝛼 , −1 < 𝛼 < ∞
⎪
⎪
⎪
⎩ K lnT
T
,K =
2DH Δ2H
Γ(1 + 𝛽)
, 𝛼 = −1
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Comparison with numerical data H = 0.3, 𝛽 = 0.3,
𝛼 = −0.3
〈δ2 (T,∆) 〉
10
1
∆=1
∆=10
0.1 0
10
∆=20
∆=30
∆=40
∆=50
1
10
10
2
Tα
3
10
4
10
T
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Calculation for the numerator of the EB
From the definition of the TAMSD, we have
∫︁
⟨(︁
)︁2 ⟩
=
𝛿X2 (T , Δ)
T −Δ
∫︁
d𝜏1
0
T −Δ
d𝜏2 G (𝜏1 , 𝜏2 )
0
(T − Δ)2
where
⟨[︁
]︁2 [︁
]︁2 ⟩
G (𝜏1 , 𝜏2 ) = X (𝜏1 + Δ) − X (𝜏1 ) X (𝜏2 + Δ) − X (𝜏2 )
= G0 + G1 + G2 + G3
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Expressions for G0 and G1
⟨︀ ⟩︀
G0 = 4DH2 Δ4H Λ2𝛽 [𝜏1 𝜏2 (𝜏1 + Δ)(𝜏2 + Δ)]𝛼/2
⟨︀ ⟩︀ {︁
𝛼/2
+ 2DH2 Λ2𝛽 |𝜏1 + Δ − 𝜏2 |2H 𝜏2 (𝜏1 + Δ)𝛼/2
𝛼/2
+ |𝜏2 + Δ − 𝜏1 |2H 𝜏1 (𝜏2 + Δ)𝛼/2 − |𝜏1 − 𝜏2 |2H
[︁
]︁ }︁2
× (𝜏1 𝜏2 )𝛼/2 + (𝜏1 + Δ)𝛼/2 (𝜏2 + Δ)𝛼/2
𝛼/2
G1 = Δ2H 𝜏2
𝛼/2
+Δ2H 𝜏1 (𝜏1
[︁
]︁[︁
]︁
𝛼/2
2H+𝛼/2
−(𝜏1 +Δ)2H+𝛼/2 𝜏1 −(𝜏1 +Δ)𝛼/2 +
(𝜏2 +Δ)𝛼/2 𝜏1
+ Δ)
𝛼/2
[︁
]︁[︁
]︁
2H+𝛼/2
𝛼/2
2H+𝛼/2
𝛼/2
− (𝜏2 + Δ)
𝜏2
𝜏2 − (𝜏2 + Δ)
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Expressions for G2 and G3
{︁[︁ 2H+ 𝛼
]︁[︁ 𝛼
]︁
𝛼
𝛼
2
G2 = 𝜏1
− (𝜏1 + Δ)2H+ 2 𝜏22 − (𝜏2 + Δ) 2 +
[︁ 2H+ 𝛼
]︁[︁ 𝛼
]︁
𝛼
𝛼
2
+ 𝜏2
− (𝜏2 + Δ)2H+ 2 𝜏12 − (𝜏1 + Δ) 2 ×
{︁
(︁
)︁ 𝛼
(︁
)︁ 𝛼
2
2
2H
2H
× |𝜏1 + Δ − 𝜏2 |
+ |𝜏2 + Δ − 𝜏1 |
𝜏2 (𝜏1 + Δ)
𝜏1 (𝜏2 + Δ)
]︁}︁
[︁
𝛼
𝛼
𝛼
−|𝜏1 − 𝜏2 |2H (𝜏1 𝜏2 ) 2 + (𝜏1 + Δ) 2 (𝜏2 + Δ) 2
[︁
]︁ [︁
]︁
2H+𝛼/2 2
𝛼/2 2
G3 = (𝜏1 + Δ)2H+𝛼/2 − 𝜏1
(𝜏2 + Δ)𝛼/2 − 𝜏2
[︁
]︁ [︁
]︁
2H+𝛼/2 2
𝛼/2 2
+ (𝜏2 + Δ)2H+𝛼/2 − 𝜏2
(𝜏1 + Δ)𝛼/2 − 𝜏1
[︁
]︁[︁
]︁
2H+𝛼/2
𝛼/2
+ 4 𝜏1
− (𝜏1 + Δ)2H+𝛼/2 𝜏1 − (𝜏1 + Δ)𝛼/2
[︁
]︁[︁
]︁
2H+𝛼/2
𝛼/2
× 𝜏2
− (𝜏2 + Δ)2H+𝛼/2 𝜏2 − (𝜏2 + Δ)𝛼/2
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Approximation of G (𝜏1 , 𝜏2 )
It can be shown that
G0 ∼ T 4H+2𝛼 (Δ/𝜏 )4H
G1 , G2 ∼ T 4H+2𝛼 (Δ/𝜏 )2H+2
and
G3 ∼ T 4H+2𝛼 (Δ/𝜏 )4 .
This means that, due to 4H < 2 + 2H < 4, the following inequalities are
met
G0 ≫ G1 , G2 ≫ G3 .
Using the approximation Δ ≪ 𝜏1 , 𝜏2 , we obtain
[︁
⟨︀ 2 ⟩︀ 𝛼 𝛼 {︁ 2 4H
G (𝜏1 , 𝜏2 ) ≈
Λ𝛽 𝜏1 𝜏2 4DH Δ + 2DH2 |𝜏1 + Δ − 𝜏2 |2H +
Δ≪𝜏1 ,𝜏2
+|𝜏2 + Δ − 𝜏1 |2H − 2|𝜏1 − 𝜏2 |2H
]︁2 }︁
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Final result
Now, EB is given by
EB = lim
T →∞
⟨Λ2𝛽 ⟩
⟨Λ𝛽 ⟩2
[︂
]︂
(1 + 𝛼)2 Ψ
−1
1+
2T 2𝛼+2 Δ4H
Here, the integral Ψ is defined as
∫︁ T −Δ ∫︁ T −Δ
{︁
Ψ=
d𝜏1 d𝜏2 𝜏1𝛼 𝜏2𝛼 |𝜏1 − 𝜏2 + Δ|2H +
0
0
+ |𝜏2 − 𝜏1 + Δ|2H − 2|𝜏1 − 𝜏2 |2H
}︁2
Depending on the value of H,
⎧ 4H+1 2𝛼+1
T
, 0 < H < 43 ,
⎨ Δ
Ψ∼
⎩ 4 4H+2𝛼−2 3
Δ T
, 4 < H < 1.
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Final result
EB depends solely on 𝛽
EB =
⟨Λ2𝛽 ⟩
⟨Λ𝛽 ⟩2
−1=
𝛽Γ2 (𝛽)
−1
Γ(2𝛽)
We have presented here the analysis of the Ergodicity breaking parameter
of the defined ”nonstationary” ggBm. Our finding is that this parameter
depends only on 𝛽 in the same way like the standard ggBm.
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Comparison with numerical data
10
EB (T,∆)
Vn(2)(t)
2
n=9
n=8
n=7
n=6
5·10
2
4·10
3·102
2·102
1·102
0
0
2·103
4·103
t
6·103
1
∆=1
∆=10
101
∆=20
∆=30
102
∆=40
∆=50
103
104
T
[H = 0.3, 𝛽 = 0.3, 𝛼 = −0.3]
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Conclusion
1
The variance of the process X𝛼,𝛽,H (t) is mainly governed by H and
affected by 𝛼
2
The process is non-ergodic and 𝛽 is the only parameter that controls
the ergodic/non-ergodic transition. This mean that in this aspect,
there is no difference from ggBm with stationary and nonstationary
increments.
3
𝛼 is the only parameter that controls aging, driving its slope
according to its sign.
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR
Thank you for your attention!
Pham Minh Tuan Belgorod National Research University, Russia inPHYSICAL
collaboration
CHARACTERISATION
with Daniel Molina-Garcı́a
OF THE
and GENERALISED
Gianni Pagnini Ikeb
GR

Similar documents