Current research confirms abnormalities in resting-state electroencephalogram (EEG) power and functional connectivity (FC) patterns in specific brain regions of individuals with depression. To study changes in the flow of information between cortical regions of the brain in patients with depression, we used 64-channel EEG to record neural oscillatory activity in 68 relevant cortical regions in 22 depressed patients and 22 healthy adolescents using source-space EEG. The direction and strength of information flow between brain regions was investigated using directional phase transfer entropy (PTE). Compared to healthy controls, we observed an increased intensity of PTE information flow between the left and right hemispheres in the theta and alpha frequency bands in depressed subjects. The intensity of information flow between anterior and posterior regions within each hemisphere was reduced. Significant differences were found in the left supramarginal gyrus, right delta in the theta frequency band and bilateral lateral occipital lobe, and paracentral gyrus and parahippocampal gyrus in the alpha frequency band. The accuracy of cross-classification of directed PTE values with significant differences between groups was 91%. These findings suggest that altered information flow in the brains of depressed patients is related to the pathogenesis of depression, providing insights for patient identification and pathological studies.
Citation: Zhongwen Jia, Lihan Tang, Jidong Lv, Linhong Deng, Ling Zou. Depression-induced changes in directed functional brain networks: A source-space resting-state EEG study[J]. Mathematical Biosciences and Engineering, 2024, 21(9): 7124-7138. doi: 10.3934/mbe.2024315
[1] | Mogtaba Mohammed, Mamadou Sango . Homogenization of nonlinear hyperbolic stochastic partial differential equations with nonlinear damping and forcing. Networks and Heterogeneous Media, 2019, 14(2): 341-369. doi: 10.3934/nhm.2019014 |
[2] | Catherine Choquet, Ali Sili . Homogenization of a model of displacement with unbounded viscosity. Networks and Heterogeneous Media, 2009, 4(4): 649-666. doi: 10.3934/nhm.2009.4.649 |
[3] | Alexander Mielke, Sina Reichelt, Marita Thomas . Two-scale homogenization of nonlinear reaction-diffusion systems with slow diffusion. Networks and Heterogeneous Media, 2014, 9(2): 353-382. doi: 10.3934/nhm.2014.9.353 |
[4] | Liselott Flodén, Jens Persson . Homogenization of nonlinear dissipative hyperbolic problems exhibiting arbitrarily many spatial and temporal scales. Networks and Heterogeneous Media, 2016, 11(4): 627-653. doi: 10.3934/nhm.2016012 |
[5] | Erik Grandelius, Kenneth H. Karlsen . The cardiac bidomain model and homogenization. Networks and Heterogeneous Media, 2019, 14(1): 173-204. doi: 10.3934/nhm.2019009 |
[6] | Hakima Bessaih, Yalchin Efendiev, Florin Maris . Homogenization of the evolution Stokes equation in a perforated domain with a stochastic Fourier boundary condition. Networks and Heterogeneous Media, 2015, 10(2): 343-367. doi: 10.3934/nhm.2015.10.343 |
[7] | Junlong Chen, Yanbin Tang . Homogenization of nonlinear nonlocal diffusion equation with periodic and stationary structure. Networks and Heterogeneous Media, 2023, 18(3): 1118-1177. doi: 10.3934/nhm.2023049 |
[8] | Martin Heida, Benedikt Jahnel, Anh Duc Vu . Regularized homogenization on irregularly perforated domains. Networks and Heterogeneous Media, 2025, 20(1): 165-212. doi: 10.3934/nhm.2025010 |
[9] | Didier Georges . Infinite-dimensional nonlinear predictive control design for open-channel hydraulic systems. Networks and Heterogeneous Media, 2009, 4(2): 267-285. doi: 10.3934/nhm.2009.4.267 |
[10] | Xavier Blanc, Claude Le Bris . Improving on computation of homogenized coefficients in the periodic and quasi-periodic settings. Networks and Heterogeneous Media, 2010, 5(1): 1-29. doi: 10.3934/nhm.2010.5.1 |
Current research confirms abnormalities in resting-state electroencephalogram (EEG) power and functional connectivity (FC) patterns in specific brain regions of individuals with depression. To study changes in the flow of information between cortical regions of the brain in patients with depression, we used 64-channel EEG to record neural oscillatory activity in 68 relevant cortical regions in 22 depressed patients and 22 healthy adolescents using source-space EEG. The direction and strength of information flow between brain regions was investigated using directional phase transfer entropy (PTE). Compared to healthy controls, we observed an increased intensity of PTE information flow between the left and right hemispheres in the theta and alpha frequency bands in depressed subjects. The intensity of information flow between anterior and posterior regions within each hemisphere was reduced. Significant differences were found in the left supramarginal gyrus, right delta in the theta frequency band and bilateral lateral occipital lobe, and paracentral gyrus and parahippocampal gyrus in the alpha frequency band. The accuracy of cross-classification of directed PTE values with significant differences between groups was 91%. These findings suggest that altered information flow in the brains of depressed patients is related to the pathogenesis of depression, providing insights for patient identification and pathological studies.
Homogenization theory has become an important tool in the investigation of processes taking place in highly heterogenous media ranging from soil to the most advanced aircraft the construction of which uses composite materials. So far, the problems solved by means of homogenization have mainly involved deterministic partial differential equations (PDEs) and the homogenization of PDEs with randomly oscillating coefficients; the great wealth of results obtained over several decades on problems of diverse classes and methodologies can be found for instance in [9,6,40,41,23,34,22,49,31,17,4,32,36,46,50,33], for the deterministic case and [13,14,18,20,24,37,19,47,48]. for the random case. Fundamental methods were subsequently developed such as the method of asymptotic expansions ([9], [6], [40], [41]), the two scale-convergence ([4], [32]), Tartar method of oscillating test functions and H-convergence ([49]), the asymptotic method for non periodically perforated domains ([23], [46]), G-convergence ([36]) and
However physical processes under random fluctuations are better modelled by stochastic partial differential equations (SPDEs). It was therefore natural to consider homogenization of this very important class of PDEs. Research in this direction is still at its infancy, despite the importance of such problems in both applied and fundamental sciences. Some relevant interesting work have recently been undertaken, mainly for parabolic SPDEs; see for instance [3,8,10,11,21,43,44]. We also note the closely related work [3,25,15,16] dealing with stochastic homogenization for SPDEs with small parameters. The list of references is of course not exhaustive, but a representation of the main trends in the field.
The homogenization of hyperbolic SPDEs was initiated in [27], [28,29], [30] where the authors studied the homogenization of Dirichlet problems for linear hyperbolic equation with rapidly oscillating coefficients using the method of the two-scale convergence pioneered by Nguetseng in [32] and developed by Allaire in [4] and [5]; they also dealt with the linear Neumann problem by means of Tartar's method and obtained the corresponding corrector results within these settings; [30] deals with a semilinear hyperbolic SPDE by Tartar's method.
In the present work, following the two-scale convergence method, we investigate the homogenization of a non-linear hyperbolic equation with nonlinear damping, where the intensity of the noise is also nonlinear and is assumed to satisfy Lipschitz's condition. Our investigation relies on crucial compactness results of analytic (Aubin-Lions-Simon's type) and probabilistic (Prokhorov and Skorokhod fundamental theorems) nature. It should be noted that these methods extend readily to the case when Lipschitz condition on the intensity of the noise is replaced by a mere continuity. In contrast to the linear and the semilinear cases considered in previous papers, the type of nonlinear damping and nonlinear noise in the present paper leads to new challenges in obtaining uniform a priori estimates as well as in the passage to the limit. It should be noted that the process of damping in mechanical systems is a crucial stabilizing factor when the system is subjected to very extreme tasks; mathematically this translates in some regularizing effects on the solution of the governing equations.
We are concerned with the homogenization of the initial boundary value problem with oscillating data, referred to throughout the paper as problem
duϵt−div(Aϵ(x)∇uϵ)dt+B(t,uϵt)dt=f(t,x,x/ε,∇uϵ)dt+g(t,x,x/ε,uϵt)dW in (0,T)×Quϵ=0 on(0,T)×∂Q,uϵ(0,x)=aϵ(x), uϵt(0,x)=bϵ(x) in Q, |
where
Few words about the difference between the current work and previous works by the authors on homogenization of SPDEs. Compared to [27,28,29,30], the structure of problem (
We now introduce some functions spaces needed in the sequel.
For
W1,p(Q)={ϕ:ϕ∈Lp(Q),∂ϕ∂xj∈Lp(Q),j=1,...,n}, |
where the derivatives exist in the weak sense, and
For a Banach space
||ϕ||Lp(0,T;X)=(∫T0||ϕ||pXdt)1p,0≤p<∞. |
When
‖ϕ‖L∞(0,T;X)=esssup[0,T]‖ϕ‖X<∞. |
For
||ϕ||Lq(Ω,F,P;Lp(0,T;X))=(E||ϕ||qLp(0,T;X))1/q. |
When
||ϕ||Lq(Ω,F,P;L∞(0,T;X))=(E||ϕ||qL∞(0,T;X))1/q. |
It is well known that under the above norms,
We now impose the following hypotheses on the data.
n∑i,j=1ai,jξiξj≥αn∑i=1ξ2i for, ξ∈Rn,ai,j∈L∞(Rn),i,j=1,…,n. |
(ⅰ)
(ⅱ) There exists a constant
(ⅲ) There exists a positive constant
(ⅳ)
(ⅴ) The map
(A3) We assume that
||f(t,x,xε,w)||L2(Q)≤C||w||L2(Q), |
for any
(A4)
(A5)
●
●
||gj(t,x,y,ϕ)||L2(Q)≤C(1+||ϕ||L2(Q)), | (1) |
and
●
|gj(t,x,y,s1)−gj(t,x,y,s2)|≤L|s1−s2|, | (2) |
with the constant
If
From now on we use the following oscillating functions
fϵ(t,x,w)=f(t,x,xε,w), gεj(t,x,ϕ)=gi(t,x,xε,ϕ). |
We now introduce our notion of solution; namely the strong probabilistic one.
Definition 1.1. We define the strong probabilistic solution of the problem
uϵ:Ω×[0,T]⟶H10(Q), |
satisfying the following conditions:
(1)
(2)
uϵ∈L2(Ω,F,P;C(0,T;H10(Q)))uϵt∈L2(Ω,F,P;C(0,T;L2(Q)))∩Lp(Ω,F,P;Lp(0,T;W1,p0(Q))), |
(3)
(uϵt(t,.),ϕ)−(uϵt(0,.),ϕ)+∫t0(Aϵ∇uϵ(s,.),∇ϕ)ds+∫t0⟨Bϵ(s,uϵt),ϕ⟩ds=∫t0(fϵ(s,.,∇uϵ),ϕ)ds+(∫t0gϵ(s,.,uϵt)dW(s),ϕ),∀ϕ∈C∞c(Q). |
The problem of existence and uniqueness of a strong probabilistic solution of
Theorem 1.2. Suppose that the assumptions
Our goal is to show that as
(P){dut−divA0∇udt+B(t,ut)dt=˜f(t,x,∇u)dt+˜g(t,x,ut)d˜W in Q×(0,T),u=0 on∂Q×(0,T),u(0,x)=a(x)∈H10(Q),ut(0,x)=b(x)∈L2(Q), |
where
A0=1|Y|∫Y(A(y)−A(y)χ(y))dy, |
{divy(A(y)∇yχ(y))=∇y⋅A(y)inYχisYperiodic, |
for any
˜f(t,x,∇u)=1|Y|∫YF(t,x,y)⋅[∇xu(t,x)+∇yu1(t,x,y)]dy,˜g(t,x,ut)=1|Y|∫Yg(t,x,y,ut)dy, |
Here and in the sequel,
Lemma 2.1. Under the assumptions
Esup0≤t≤T‖uϵ(t)‖2H10(Q)≤C,Esup0≤t≤T‖uϵt(t)‖2L2(Q)≤C, | (3) |
and
E∫T0‖uϵt(t)‖pW1,p0(Q)≤C. | (4) |
Proof. The following arguments are used modulo appropriate stopping times. It
d[‖uϵt‖2L2(Q)+(Aϵ∇uϵ,∇uϵ)]+2⟨B(t,uϵt),uϵt)⟩dt=2(fϵ(t,x,∇uϵ)),uϵt)dt+2(gϵ(t,x,uϵt),uϵt)dW+m∑j=0‖gϵj(t,x,uϵt)‖2L2(Q)dt. |
Integrating over
‖uϵt(t)‖2L2(Q)+(Aϵ∇uϵ(t),∇uϵ(t))+2∫t0⟨B(s,uϵt(s)),uϵt(s))⟩ds=‖uϵ1‖2L2(Q)+(Aϵ∇uϵ0,∇uϵ0)+2∫t0(fϵ(s,x,∇uϵ),uϵt)ds+2∫t0(gϵ(s,x,uϵt),uϵt)dW+m∑j=0∫t0‖gϵj(s,x,uϵt)‖2L2(Q)ds. |
Using the assumptions
E[sup0≤t≤T‖uϵt(t)‖2L2(Q)+sup0≤t≤T‖uϵ(t)‖2H10(Q)+2γ∫t0‖uϵt(s)‖pW1,p0(Q)ds]≤C[C1+∫t0‖uϵt(t)‖2L2(Q)dt+2∫t0|(fϵ(s,x,∇uϵ),uϵt)|ds+2sup0≤s≤t|∫s0(gϵ(σ,x,uϵt),uϵt)dW|], | (5) |
where
C1=C(T)+‖uϵ1‖2L2(Q)+‖uϵ0‖2H10(Q). |
Using assumptions (A3), thanks to Cauchy-Schwarz's and Young's inequalities, we have
E∫T0|(fϵ(s,x,∇uϵ),uϵt)|dt≤E∫T0‖∇uϵ‖L2(Q)‖uϵt‖L2(Q)dt≤Esup0≤t≤T‖uϵt(t)‖L2(Q)∫T0‖∇uϵ‖L2(Q)dt≤ϱEsup0≤t≤T‖uϵt(t)‖2L2(Q)+C(ϱ)T(∫T0‖∇uϵ‖2L2(Q)dt), | (6) |
where
Esup0≤s≤t|∫s0(gϵ(σ,x,uϵt(σ)),uϵt(σ))dW(σ)|≤CE(∫t0(gϵ(σ,x,uϵt(σ)),uϵt(σ))2dσ)12≤CE(sup0≤s≤t‖uϵt(s)‖L2(Q)∫t0‖gϵ(σ,x,uϵt(σ))‖2L2(Q)dσ)12. |
Again using Young's inequality and the assumptions
2Esup0≤s≤t|∫s0(gϵ(σ,x,uϵt(σ)),uϵt(σ))dW|≤ϱEsup0≤s≤t‖uϵt(s)‖2L2(Q)+C(ϱ)∫T0‖gϵ(σ,uϵt(σ))‖2L2(Q)dσ≤ϱEsup0≤s≤t‖uϵt(s)‖2L2(Q)+C(ϱ)(T)+C(ϱ)∫T0‖uϵt(σ)‖2L2(Q)dσ, | (7) |
for
Esup0≤t≤T‖uϵt(t)‖2L2(Q)+Esup0≤t≤T‖uϵ(t)‖2H10(Q)+CE∫t0‖uϵt(s)‖pW1,p0(Q)ds≤C(T,C1,C2)+CE∫t0[‖uϵt(s)‖2L2(Q)+‖uϵ(s)‖2H10(Q)]dt, | (8) |
Using Gronwall's inequality, we have
E[sup0≤t≤T‖uϵt(t)‖2L2(Q)+sup0≤t≤T‖uϵ(t)‖2H10(Q)]≤C, |
and subsequently
E∫t0‖uϵt(s)‖pW1,p0(Q)ds≤C. |
The proof is complete.
The following lemma will be of great importance in proving the tightness of probability measures generated by the solution of problem
Lemma 2.2. Let the conditions of Lemma 2.1 be satisfied and let
Esup|θ|≤δ∫T0‖uϵt(t+θ)−uϵt(t)‖p′W−1,p′(Q)dt≤Cδp′/p, |
for any
Proof..
Assume that
uϵt(t+θ)−uϵt(t)=∫t+θtdiv(Aϵ∇uϵ)ds−∫t+θtB(s,uϵt(s))ds+∫t+θtfϵ(s,x,∇uϵ)ds+∫t+θtgϵ(s,uϵt(s))dW(s). |
Then
‖uϵt(t+θ)−uϵt(t)‖W−1,p′(Q)≤‖∫t+θtdiv(Aϵ∇uϵ)ds‖W−1,p′(Q)+‖∫t+θtB(s,uϵt(s))ds‖W−1,p′(Q)+‖∫t+θtfϵ(s,x,∇uϵ)ds‖W−1,p′(Q)+‖∫t+θtgϵ(s,uϵt(s))dW(s)‖W−1,p′(Q). | (9) |
Firstly, thanks to assumption
‖∫t+θtdiv(Aϵ∇uϵ)ds‖W−1,p′(Q)≤supϕ∈W1,p0(Q):‖ϕ‖=1|⟨∫t+θtdiv(Aϵ∇uϵ)ds,ϕ⟩W−1,p′(Q),W1,p0(Q)|=supϕ∈W1,p0(Q):‖ϕ‖=1∫Q∫t+θtAϵ∇uϵ∇ϕdxds≤Csupϕ∈W1,p0(Q):‖ϕ‖=1∫t+θt‖∇uϵ‖Lp′(Q)‖∇ϕ‖Lp(Q)ds≤C∫t+θt‖∇uϵ‖L2(Q)ds≤Cθ1/2(∫t+θt‖∇uϵ‖2L2(Q)ds)1/2, | (10) |
where we have used the fact that
Secondly, we use assumption
‖∫t+θtB(s,uϵt(s))ds‖W−1,p′(Q)≤supϕ∈W1,p0(Q):‖ϕ‖=1|⟨∫t+θtB(s,uϵt(s))ds,ϕ⟩W−1,p′(Q),W1,p0(Q)|≤supϕ∈W1,p0(Q):‖ϕ‖=1∫t+θt‖B(s,uϵt(s))‖W−1,p′(Q)‖ϕ‖W1,p0(Q)ds≤Cθ1/p(∫t+θt‖uϵt‖pW1,p0(Q)ds)1/p′. | (11) |
Thirdly,
‖∫t+θtfϵ(s,x,∇uϵ)ds‖W−1,p′(Q)≤‖∫t+θtfϵ(s,x,∇uϵ)ds‖L2(Q)≤C∫t+θt‖∇uϵ‖L2(Q)≤θ1/2(∫t+θt‖∇uϵ‖2L2(Q)ds)1/2, | (12) |
where we have used assumption (A3).
Using 10, 11 and 12 in 9 raised to the power
Esup0<θ≤δ∫T0‖uϵt(t+θ)−uϵt(t)‖p′W−1,p′(Q)dt≤CEsup0<θ≤δθp′/2∫T0(∫t+θt‖∇uϵ‖2L2(Q)ds)p′/2dt+CEsup0<θ≤δθp′/p∫T0∫t+θt‖uϵt‖pW1,p0(Q)dsdt+Esup0<θ≤δ∫T0‖∫t+θtgϵ(s,uϵt(s)dW(s)‖p′W−1,p′(Q)dt. | (13) |
We now estimate the term involving the stochastic integral.
We use the embedding
W1,p0(Q)↪L2(Q)↪W−1,p′(Q) |
to get the estimate
Esup0<θ≤δ∫T0||∫t+θtgϵ(s,uϵt(s)dW(s)||p′W−1,p′dt≤Esup0<θ≤δ∫T0||∫t+θtgϵ(s,uϵt(s)dW(s)||p′L2(Q)dt. | (14) |
Thanks to Fubini's theorem and H
E∫T0sup0<θ≤δ||∫t+θtgϵ(s,uϵt(s)dW(s)||p′L2(Q)dt≤∫T0(∫QEsup0<θ≤δ(∫t+θtgϵ(s,uϵt(s))dW(s))2dx)p′/2dt≤∫T0(E∫t+δt||gϵ(s,uϵt(s)||2L2(Q)ds)p′/2dt, | (15) |
where we have used Burkholder-Davis-Gundy's inequality. We now invoke assumption
Esup0<θ≤δ∫T0||∫t+θtgϵ(s,uϵt(s)dW(s)||p′W−1,p′dt≤∫T0[E∫t+δt(1+||uϵt(s)||2L2(Q))ds]p′/2dt≤CTδp′/2. | (16) |
For the first term in the right-hand side of 13, we use Fubini's theorem, H
Esup0<θ≤δθp′/2∫T0(∫t+θt‖∇uϵ‖2L2(Q)ds)p′/2≤δp′/2∫T0(E∫t+δt‖∇uϵ‖2L2(Q)ds)p′/2≤CTδp′. | (17) |
The second term on the right hand side of 13 is estimated using 4 and we get
Esup0<θ≤δθp′/p∫T0∫t+θt‖uϵt‖pW1,p0(Q)dsdt≤δp′/p∫T0E∫T0‖uϵt‖pW1,p0(Q)dsdt≤Cδp′/p. | (18) |
Combining 13, 16, 17 and 18, and taking into account the fact that the similar estimates hold for
Esup|θ|≤δ∫T0‖uϵt(t+θ)−uϵt(t)‖p′W−1,p′(Q)dt≤Cδp′/p. |
This completes the proof.
The following Lemmas are needed in the proof of the tightness and the study of the properties of the probability measures generated by the sequence
We have from [45]
Lemma 3.1. Let
Ns,p(0,T;B1)={v∈Lp(0,T;B1):suph>0h−s‖v(t+h)−v(t)‖Lp(0,T−θ,B1)<∞}. |
Then
The following two lemmas are collected from [12]. Let
Lemma 3.2. (Prokhorov) A sequence of probability measures
Lemma 3.3. (Skorokhod) Suppose that the probability measures
limn→∞ξn=ξ,P−a.s.; |
the symbol
Let us introduce the space
Z1={ϕ:sup0≤t≤T‖ϕ(t)‖2H10(Q)≤C1,sup0≤t≤T‖ϕ′(t)‖2L2(Q)≤C1}, |
and
Z2={ψ:sup0≤t≤T‖ψ(t)‖2L2(Q)≤C3,∫T0‖ψ(t)‖pW1,p0(Q)dt≤C4,∫T0‖ψ(t+θ)−ψ(t)‖p′W−1,p′(Q)≤C5θ1/p}. |
We endow
‖(ϕ,ψ)‖Z=‖ϕ‖Z1+‖ψ‖Z2=sup0≤t≤T‖ϕ′(t)‖L2(Q)+sup0≤t≤T‖ϕ‖H10(Q)+sup0≤t≤T‖ψ(t)‖2L2(Q)+(∫T0‖ψ(t)‖pW1,p0(Q)dt)1p+(supθ>01θ1/p∫T0‖ψ(t+θ)−ψ(t)‖p′W−1,p′(Q))1p′. |
Lemma 3.4. The above constructed space
Proof. Lemma 3.1 together with suitable arguments due to Bensoussan [7] give the compactness of
We now consider the space
Ψϵ:ω↦(W(ω),uϵ(ω),uϵt(ω)). |
Define on
Πϵ(A)=P(Ψ−1ϵ(A))for allA∈B(X). |
Lemma 3.5. The family of probability measures
Proof. We carry out the proof following a long the lines of the proof of [27,lemma 7]. For
Wδ⊂C(0,T;Rm),Dδ⊂L2(0,T;L2(Q)),Eδ⊂L2(0,T;L2(Q)) |
such that
Πϵ{(W,uϵ,uϵt)∈Wδ×Dδ×Eδ}≥1−δ. |
This is equivalent to
P{ω:W(⋅,ω)∈Wδ,uϵ(⋅,ω)∈Dδ,uϵt)(⋅,ω)∈Eδ}≥1−δ, |
which can be proved if we can show that
P{ω:W(⋅,ω)∉Wδ}≤δ,P{uϵ(⋅,ω)∉Dδ}≤δ,P{uϵt)(⋅,ω).∉Eδ}≤δ. |
Let
Wδ={W(⋅)∈C(0,T;Rm):supt,s∈[0,T]n|W(s)−W(t)|≤Lδ:|s−t|≤Tn−1}. |
Using Arzela's theorem and the fact that
P(ω:η(ω)≥α)≤E|η(ω)|kαk, | (19) |
where
P{ω:W(⋅,ω)∉Wδ}≤P[∞⋃n=1(supt,s∈[0,T]|W(s)−W(t)|≥Lδn:|s−t|≤Tn−1)]≤∞∑n=0P[n6⋃j=1(supTjn−6≤t≤T(j+1)n−6|W(s)−W(t)|≥Lδn)]. |
But
E(Wi(t)−Wi(s))2k=(2k−1)!!(t−s)k,k=1,2,3,…, |
where
For
P{ω:W(.,ω)∉Wδ}≤∞∑n=0n6∑j=1(nLδ)4E(supTjn−6≤t≤T(j+1)n−6|W(t)−W(jTn−6)|4)≤C∞∑n=0n6∑j=1(nLδ)4(Tn−6)2=CT2(Lδ)4∞∑n=0n−2. |
Choosing
P{ω:W(.,ω)∉Wδ}≤δ3. |
Now, let
Dδ={z:sup0≤t≤T‖z(t)‖2H10(Q)≤Kδ,sup0≤t≤T‖z′(t)‖2L2(Q)≤Mδ}. |
Lemma 3.4 shows that
P{uϵ∉Dδ}≤P{sup0≤t≤T‖uϵ(t)‖2H10(Q)≥Kδ}+P{sup0≤t≤T‖uϵt(t)‖2L2(Q)≥Mδ}. |
Markov's inequality 19 gives
P{uϵ∉Dδ}≤1KδEsup0≤t≤T‖uϵ(t)‖2H10(Q)+1MδEsup0≤t≤T‖uϵt(t)‖2L2(Q)≤CKδ+CMδ=δ3. |
for
Similarly, we let
Bδ={v:sup0≤t≤T‖v(t)‖2L2(Q)≤K′δ,∫T0‖v(t)‖pW1,p0(Q)dt≤L′δ,supθ≤μn∫T0‖v(t+θ)−v(t)‖p′W−1,p′(Q)dt≤νnM′δ}. |
Owing to Proposition 3.1 in [7],
P{uϵt∉Bδ}≤P{sup0≤t≤T‖uϵt(t)‖2L2(Q)≥K′δ}+P{∫T0‖uϵt(t)‖pW1,p0(Q)dt≥L′δ}+P{supθ≤μn∫T0‖uϵt(t+θ)−uϵt(t)‖p′W−1,p(Q)dt≥νnM′δ}. |
Again thanks to 19, we obtain
P{uϵt∉Bδ}≤1K′δEsup0≤t≤T‖uϵt(t)‖2L2(Q)+1L′δE∫T0‖uϵt(t)‖pW1,p0(Q)dt+∑∞n=01νnM′δE{supθ≤μn∫T0‖uϵt(t+θ)−uϵt(t)‖p′W−1,p(Q)dt}≤CK′δ+CL′δ+CM′δ∑μp′/pnνn=δ3δ, |
for
From Lemmas 3.2 and 3.5, there exist a subsequence
Πϵj⇀Π |
weakly. From lemma 3.3, there exist a probability space
(Wϵj,uϵj,uϵjt)→(˜W,u,ut)inX,˜P−a.s.. | (20) |
Let us define the filtration
~Ft=σ{˜W(s),u(s),ut(s)}0≤s≤t. |
We show that
In this section, we state some key facts about the powerful two-scale convergence invented by Nguetseng [32].
Definition 4.1. A sequence
limϵ→0∫T0∫Qvϵψϵdxdt=1|Y|∫T0∫Q×Yv(t,x,y)ψ(t,x,y)dydxdt, | (21) |
where
The following result deals with some of the properties of the test functions which we are considering; it is a modification of Lemma 9.1 from [17,p.174].
Lemma 4.2. (i) Let
‖ψ(⋅,⋅,⋅ϵ)‖Lp(0,T;Lp(Q))≤‖ψ(⋅,⋅,⋅)‖Lp((0,T)×Q;Cper(Y)) | (22) |
and
ψ(⋅,⋅,⋅ϵ)⇀1|Y|∫Yψ(⋅,⋅,y)dyweakly inLp(0,T;Lp(Q)). |
Furthermore if
limϵ→0∫T0∫Q[ψ(t,x,xϵ)]2dxdt=1|Y|∫T0∫Q×Y[ψ(t,x,y)]2dtdxdy. | (23) |
(ii) If
1r+1s=1p, |
then
ψ(⋅,⋅,⋅ϵ)⇀ψ1(⋅,⋅)|Y|∫Yψ2(y)dyweakly inLp(0,T;Lp(Q)). |
The following theorems are of great importance in obtaining the homogenization result; for their proofs, we refer to [4], [17] and [26].
Theorem 4.3. Let
‖uϵ‖L2(0,T;L2(Q))<∞. | (24) |
Then up to a subsequence
Theorem 4.4. Let
‖uϵ‖L2(0,T;H10(Q))<∞. |
Then, up to a subsequence, there exists a couple of functions
uϵ→u 2−s inL2(0,T;L2(Q)), | (25) |
∇uϵ→∇xu+∇yu1 2−s inL2(0,T;L2(Q)). | (26) |
The following lemma is crucial in obtaining the convergence of the stochastic integral in the next section
Lemma 4.5. The oscillating data given in (A5) satisfies the following convergence
g(t,x,xε,uεt)⇀˜g(t,x,ut)=:1|Y|∫Yg(t,x,y,ut)dy weakly in L2((0,T)×Q), ˜P−a.s.. | (27) |
Proof. Test with
∫T0∫Qg(t,x,xε,uεt)ψ(t,x,xε)dxdt=Iε1+Iε2, |
where
Iε1=∫T0∫Q[g(t,x,xε,uεjt)−g(t,x,xε,ut)]ψ(t,x,xε)dxdt,Iε2=∫T0∫Qg(t,x,xε,ut)ψ(t,x,xε)dxdt. |
Then
Iε1≤||ψ(t,x,xε)||L2((0,T)×Q)||g(t,x,xε,uεt)−g(t,x,xε,ut)||L2((0,T)×Q)≤C||uεt−ut||L2((0,T)×Q), |
thanks to the Lipschitz condition on
Now we can apply 2-scale convergence for the limit of
limε→0Iε2=∫T0∫Q∫Yg(t,x,y,ut)ψ(t,x,y)dxdt,˜P−a.s. |
Therefore
g(t,x,xε,uεt)2−s→g(t,x,y,ut), ˜P−a.s. | (28) |
and this implies the result.
Remark 1. From the assumption (A5), 28 and 23, we have the following strong convergence
limϵ→0∫T0∫Q[g(t,x,xϵ,uϵt)]2dxdt=1|Y|∫T0∫Q×Y[g(t,x,y,ut)]2dtdxdy. | (29) |
We will now study the asymptotic behaviour of the problem
Theorem 5.1. Suppose that the assumptions on the data are satisfied. Let
aϵj⇀a,weakly inH10(Q), | (30) |
bϵj⇀b,weakly inL2(Q). | (31) |
Then there exist a probability space
Proof. From estimates 3 and 4 and assumption
uϵj⇀uweakly inL∞(0,T;H10(Q))ˆP−a.s, | (32) |
uϵjt⇀utweakly inL∞(0,T;L2(Q))ˆP−a.s, | (33) |
uϵjt⇀utweakly inLp(0,T;W1,p0(Q))ˆP−a.s, | (34) |
B(t,uϵjt)⇀χweakly inLp′(0,T;W−1,p′(Q))ˆP−a.s.. | (35) |
Now let us identify the limit in 35. By arguing as in [38,Lemma 2.6,p. 51], we get
∫t0⟨B(s,uϵjt),uϵjt⟩ds→∫t0⟨χ,ut⟩ds,weakly inL1(Ω), ∀t∈[0,T]. | (36) |
Having this in hand, let
χϵj=ˆE∫T0⟨B(t,uϵjt)−B(t,v),uϵjt−v⟩dt. | (37) |
From the monotonicity assumption
ˆE∫T0⟨χ−B(t,v),ut−v⟩dt≥0. |
For
ˆE∫T0⟨χ−B(t,ut(t)−λw(t)),w(t)⟩dt≥0. | (38) |
Using the hemicontinuty assumption
⟨χ−B(t,ut(t)−λw(t)),w(t)⟩⟶⟨χ−B(t,ut(t)),w(t)⟩, as λ⟶0, ˆP−a.s.. |
Now, from assumptions
ˆE∫T0⟨χ−B(t,ut(t)),w(t)⟩dt≥0. | (39) |
But the inequality 39 is true for all
χ=B(t,ut(t),ˆP−a.s.. |
Testing problem
−∫T0∫QuϵjtΦt(t,x)dxdt+∫T0∫QAϵj∇uϵj∇Φdxdt+∫T0∫Q⟨Bϵj(t,uϵjt),Φ⟩dxdt=∫T0∫Qfϵj(t,x,∇uϵj)Φdxdt+∫T0∫Qgϵj(t,x,uϵjt)ΦdxdWϵj, | (40) |
Using estimate 3, the convergence 20 and Theorems 4.3 and 4.4, we show the two-scale convergence
∇uϵj→∇xu+∇yu1 2-s in,L2(0,T;L2(Q)). |
Let
−∫T0∫Quϵjt(t,x)[ϕt(t,x)+ϵjϕ1t(t,x,xϵj)]dxdt+∫T0∫QAϵj(x)∇uϵj(x,t)[∇xϕ(t,x)+ϵj∇xϕ1(t,x,xϵj)+∇yϕ1(t,x,xϵj)]dxdt+∫T0∫Q⟨B(t,uϵjt),[ϕt(t,x)+ϵjϕ1t(t,x,xϵj)]⟩dxdt=∫T0∫Qfϵj(t,x,∇uϵj)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdt+∫T0∫Qgϵj(t,uϵjt)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdWϵj. | (41) |
Let us deal with these terms one by one, when
limϵj→0∫T0∫Quϵjt(t,x)[ϕt(t,x)+ϵjϕ1t(t,x,xϵj)]dxdt=limϵj→0∫T0∫Quϵjt(t,x)ϕt(t,x)dxdt+limϵj→0ϵj∫T0∫Quϵjt(t,x)ϕ1t(t,x,xϵj)dxdt=∫T0∫Qut(t,x)ϕt(t,x)dxdt,˜P−a.s.. |
The second term can be written as follows,
limϵj→0∫T0∫Q∇uϵj(x,t)Aϵj[∇xϕ(t,x)+∇yϕ1(t,x,xϵj)]dxdt+limϵj→0ϵj∫T0∫QAϵj∇uϵj(x,t)∇xϕ1(t,x,xϵj)dxdt. | (42) |
Since
limϵj→0∫T0∫Q∇uϵj(x,t)Aϵj[∇xϕ(t,x)+∇yϕ1(t,x,xϵj)]dxdt=1|Y|∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)][∇xϕ(t,x)+∇yϕ1(t,x,y)]dydxdt. |
Thanks to H
limϵj→0ϵj∫T0∫QAϵj∇uϵj(x,t)∇xϕ1(t,x,xϵj)dxdt=0,˜P−a.s.. |
Again, thanks to estimate 22 and convergence 35, we have
limϵj→0∫T0∫Q⟨B(t,uϵjt),[ϕt(t,x)+ϵjϕ1t(t,x,xϵj)]⟩dxdt=limϵj→0∫T0∫Q⟨B(t,uϵjt),ϕt(t,x)⟩dxdt+limϵj→0ϵj∫T0∫Q⟨B(t,uϵjt),ϕ1t(t,x,xϵj)⟩dxdt=∫T0∫Q⟨B(t,ut),ϕt(t,x)⟩dxdt,˜P−a.s.. |
Let us write
limϵj→0∫T0∫Qfϵj(t,x,∇uϵj)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdt=limϵj→0∫T0∫QFϵj(t,x)⋅∇uϵj[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdt=limϵj→0∫T0∫QFϵj(t,x)⋅∇uϵjϕ(t,x)dxdt+limϵj→0ϵj∫T0∫QFϵj(t,x).∇uϵjϕ1(t,x,xϵj)dxdt, | (43) |
where we have used the assumption (A3). It is easy to see that the second term in 43, converges to zero. For the first term in the right-hand side of 43, we readily have
limϵj→0∫T0∫QFϵj(t,x)⋅∇uϵjϕ(t,x)dxdt=1|Y|∫T0∫Q×YF(t,x,y)⋅[∇xu+∇yu1]ϕ(t,x)dxdydt,˜P−a.s.. | (44) |
Concerning the stochastic integral, we have
˜E∫T0∫Qgϵj(t,x,uϵjt)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdWϵj=˜E∫T0∫Qgϵj(t,x,uϵjt)ϕ(t,x)dxdWϵj+˜Eϵj∫T0∫Qgϵj(t,x,uϵjt)ϕ1(t,x,xϵj)dxdWϵj. | (45) |
We deal with the term involving
˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)dWεt=˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)d(Wεt−˜Wt)+˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)d˜Wt. | (46) |
In view of the unbounded variation of
gελ(uε)(t)=1λ∫T0ρ(−t−sλ)g(s,x,xε,uεs(s))ds for λ>0, | (47) |
where
We have that
˜E∫T0||gελ(uε)(t)||2L2(Q)dt≤˜E∫T0||g(t,x,xε,uεt(t))||2L2(Q)dt, for any λ>0, | (48) |
and for any
gελ(uε)(t)→gε(t,x,uεt(t)) strongly in L2(˜Ω,˜F,˜P,L2((0,T)×Q)) as λ→0. | (49) |
We split the first term in the right-hand side of 46 as
˜E∫T0∫Qϕ(t,x)gε(t,x,uεt(t))dxd(Wεt−˜Wt)=˜E∫T0∫Qϕ(t,x)gελ(uε)(t)dxd(Wεt−˜Wt)+˜E∫T0∫Qϕ(t,x)[gε(t,x,uεt(t))−gελ(uε)(t)]dxd(Wεt−˜Wt). | (50) |
Owing to 49, and Burkholder-Davis-Gundy's inequality, it readily follows that the second term in 50 is bounded by a function
˜E∫T0∫Qϕ(t,x)gελ(uε)(t)d(Wεt−˜Wt)=˜E∫T0∫Q(Wεt−˜Wt)∂∂t[ϕ(t,x)gελ(uε)(t)]dt+˜E∫Qϕ(T,x)gελ(uε)(T)(WεT−˜WT). | (51) |
Thanks to the conditions on
Wεt→˜Wt uniformly in C([0,T]), ˜P−a.s., | (52) |
we get that both terms on the right-hand side of 51 are bounded by the product
|˜E∫T0∫Qϕ(t,x)gε(t,x,uεt(t))dxd(Wεt−˜Wt)|≤σ1(λ)+σ2(λ)η1(ε). | (53) |
Thus, we infer from 46 that
|˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)dxdWεt−˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)d˜Wt|≤σ1(λ)+σ2(λ)η1(ε) | (54) |
Taking the limit in 54 as
limε→0|˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)dxdWεt−˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)d˜Wt|≤σ1(λ); |
but the left-hand side of this relation being independent of
limε→0˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)dxdWεt=limε→0˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)d˜Wt. | (55) |
Owing to 27; that is
g(t,x,xε,uεt)⇀˜g(t,x,ut) weakly in L2((0,T)×Q), ˜P−a.s., |
we can call upon the convergence theorem for stochastic integrals due to Rozovskii [39,Theorem 4,p. 63] to claim that
˜E∫T0∫Qϕ(t,x)g(t,x,xε,uεt)dWt→˜E∫T0∫Qϕ(t,x)˜g(t,x,ut)d˜Wt. |
Hence, we deduce from 55 that,
∫T0∫Qϕ(t,x)g(t,x,xε,uεt)dWεt→∫T0∫Qϕ(t,x)˜g(t,x,ut)d˜Wt, ˜P−a.s.. | (56) |
For the second term in 45, thanks to Burkholder-Davis-Gundy's inequality, the assumptions on
limϵj→0ϵj˜Esupt∈[0,T]|∫t0∫Qϕ1(t,x,xε)g(t,x,xε,uεt)dxdWϵjt|≤Climϵj→0ϵj˜E(∫T0(∫Qϕ1(t,x,xε)g(t,x,xε,uεt)dx)2dt)12≤Climϵj→0ϵj˜E(∫T0‖g(t,x,xε,uεt)‖L2(Q)‖ϕ1(t,x,xϵj)‖L2(Q)dt)12≤Climϵj→0ϵj(∫T0‖g(t,x,xε,uεt)‖L2(Q)dt)12→0,˜P−a.s. |
Combining the above convergences, we obtain
−∫T0∫Qut(t,x)ϕt(t,x)dxdt+1|Y|∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)]⋅[∇xϕ(t,x)+∇yϕ1(t,x,y)]dydxdt+∫T0∫Q⟨B(t,ut),ϕ(t,x)⟩dxdt=1|Y|∫T0∫Q×YF(t,x,y).[∇xu(t,x)+∇yu1(t,x,y)]ϕ(t,x)dxdydt+∫T0∫Q˜g(t,x,ut)ϕ(t,x)˜Wdx. | (57) |
Choosing in the first stage
∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)][∇yϕ1(t,x,y)]dydxdt=0, | (58) |
and
−∫T0∫Qut(t,x)ϕt(t,x)dxdt+∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)][∇xϕ(t,x)]dydxdt+∫T0∫Q⟨B(t,ut),ϕ(t,x)⟩dxdt=1|Y|∫T0∫Q×YF(t,x,y).[∇xu(t,x)+∇yu1(t,x,y)]ϕ(t,x)dxdydt+∫T0∫Q˜g(t,x,ut)ϕ(t,x)d˜Wdx. | (59) |
By standard arguments (see [17]), equation 58 has a unique solution given by
u1(t,x,y)=−χ(y)⋅∇xu(t,x)+~u1(t,x), | (60) |
where
{divy(A(y)∇yχ(y))=∇y⋅A(y),inY,χisYperiodic. | (61) |
As for the uniqueness of the solution of 59, we prove it as follows. Using 60 in 59, one obtains that 59 is the weak formulation of the equation
dut−A0Δudt+B(t,ut)dt=˜f(t,x,∇u)dt+˜g(t,x,ut)d˜W, | (62) |
where
A0=1|Y|∫Y(A(y)−A(y)∇yχ(y))dy,˜f(t,x,∇u)=1|Y|∫YF(t,x,y)⋅[∇xu(t,x)+∇yu1(t,x,y)]dy, | (63) |
and
˜g(t,x,ut)=1|Y|∫Yg(t,x,y,ut)dy. |
But the initial boundary value problem corresponding to 62 has a unique solution by [38]. It remains to show that
−∫T0∫Quϵjt(t,x)[ϕt(t,x)+ϵjϕ1t(t,x,xϵj)]dxdt+∫T0∫QAϵj(x)∇uϵj(x,t)⋅[∇xϕ(t,x)+ϵj∇xϕ1(t,x,xϵj)+∇yϕ1(t,x,xϵj)]dxdt+∫T0∫Q⟨B(t,uϵt),[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]⟩dxdt=∫T0∫Qfϵj(t,x,∇uϵj)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdt+∫T0∫Qgϵj(t,x,uϵt)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdWϵj+∫Quϵjt(x,0)v(x)dx, |
where we pass to the limit, to get
−∫T0∫Qut(t,x)ϕt(t,x)dxdt+∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)]⋅[∇xϕ(t,x)+∇yϕ1(t,x,y)]dydxdt+∫T0∫Q⟨B(t,ut),ϕ(t,x)⟩dxdt=1|Y|∫T0∫Q×YF(t,x,y)⋅[∇xu(t,x)+∇yu1(t,x,y)]ϕ(t,x)dxdydt+∫T0∫Q˜g(t,x,ut)ϕ(t,x)˜Wdxdt+∫Qb(x)v(x)dx. |
The integration by parts, in the first term gives
∫T0∫Qdut(t,x)ϕ(t,x)dx+∫Qut(x,0)v(x)dx+∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)]⋅[∇xϕ(t,x)+∇yϕ1(t,x,y)]dydxdt+∫T0∫Q⟨B(t,ut),ϕ(t,x)⟩dxdt=1|Y|∫T0∫Q×YF(t,x,y)⋅[∇xu(t,x)+∇yu1(t,x,y)]ϕ(t,x)dxdydt+∫T0∫Q˜g(t,x,ut)ϕ(t,x)˜Wdxdt+∫Qb(x)v(x)dx. |
In view of equation 57, we deduce that
∫Qut(x,0)v(x)dx=∫Qb(x)v(x)dx, |
for any
∫T0∫Quϵj(t,x)[ϕtt(t,x)+ϵjϕ1tt(t,x,xϵj)]dxdt+∫T0∫QAϵj(x)∇uϵj(x,t)⋅[∇xϕ(t,x)+ϵj∇xϕ1(t,x,xϵj)+∇yϕ1(t,x,xϵj)]dxdt+∫T0∫Q⟨B(t,uϵt),[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]⟩dxdt=∫T0∫Qfϵj(t,x,∇uϵj)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdt+∫T0∫Qgϵj(t,x,uϵt)[ϕ(t,x)+ϵjϕ1(t,x,xϵj)]dxdWϵj−∫Quϵj(x,0)v(x)dx. |
Passing to the limit in this equation, we obtain
∫T0∫Qu(t,x)ϕtt(t,x)dxdt+∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)]⋅[∇xϕ(t,x)+∇yϕ1(t,x,y)]dydxdt+∫T0∫Q⟨B(t,ut),ϕ(t,x)⟩dxdt=1|Y|∫T0∫Q×,YF(t,x,y)⋅[∇xu(t,x)+∇yu1(t,x,y)]ϕ(t,x)dxdydt+∫T0∫Q˜g(t,x,ut)ϕ(t,x)˜Wdxdt−∫Qa(x)v(x)dx. |
We integrate by parts again to obtain
−∫T0∫Qut(t,x)ϕt(t,x)dxdt−∫Qu(x,0)v(x)dx+∫T0∫Q×YA(y)[∇xu(t,x)+∇yu1(t,x,y)]⋅[∇xϕ(t,x)+∇yϕ1(t,x,y)]dydxdt+∫T0∫Q⟨B(t,ut),ϕ(t,x)⟩dxdt=1|Y|∫T0∫Q×YF(t,x,y)⋅[∇xu(t,x)+∇yu1(t,x,y)]ϕ(t,x)dxdydt+∫T0∫Q˜g(t,x,ut)ϕ(t,x)˜Wdxdt−∫Qa(x)v(x)dx. |
Using the same argument as before, we show that
Let us introduce the energies associated with the problems (
Eϵj(uϵj)(t)=12˜E‖uϵjt(t)‖2L2(Q)+12˜E∫QAϵj∇uϵj(x,t)⋅∇uϵj(x,t)dx+˜E∫t0⟨B(s,uϵjt),uϵjt⟩dsE(u)(t)=12˜E‖ut(t)‖2L2(Q)+12˜E∫QA0∇u(x,t)⋅∇u(x,t)dx+˜E∫t0⟨B(s,ut),ut⟩ds. |
But from It
12˜E‖uϵjt(t)‖2L2(Q)+12˜E∫QAϵj∇uϵj(t)⋅∇uϵj(t)dx+˜E∫t0⟨B(s,uϵjt),uϵjt⟩ds=˜E[12‖uϵj1‖2L2(Q)+12∫QAϵj∇uϵj0⋅∇uϵj0dx+∫t0(fϵj(s,x,∇uϵj),uϵjt)ds+12∫t0‖gϵj(s,uϵjt)‖2L2(Q)ds+∫t0(gϵj(s,uϵjt),uϵjt)dWϵj]. |
Thus
Eϵj(uϵj)(t)=12˜E‖uϵj1‖2L2(Q)+12˜E∫QAϵj∇uϵj0⋅∇uϵj0dx+˜E∫t0(fϵj(s,x,∇uϵj),uϵjt)ds+12˜E∫t0‖gϵj(s,uϵjt)‖2L2(Q)ds, | (64) |
E(u)(t)=12˜E‖u1‖2L2(Q)+12˜E∫QA0∇u0⋅∇u0dx+˜E∫t0(˜f(s,x,∇u),ut)ds+12˜E∫t0‖˜g(s,x,ut)‖2L2(Q)ds. | (65) |
The vanishing of the expectation of the stochastic integrals is due to the fact that
Theorem 6.1. Assume that the assumptions of Theorem 5.1 are fulfilled and
−div(Aϵj∇aϵj)→−div(A0∇a),strongly inH−1(Q), | (66) |
bϵj→b, strongly inL2(Q). | (67) |
Then
Eϵj(uϵj)(t)→E(u)(t)inC([0,T]), |
where
Proof. Thanks to the convergences 20, 44, 29, 66 and 67, we show that
Eϵj(uϵj)(t)→E(u)(t),∀t∈[0,T]. |
Now we need to show that
|Eϵj(uϵj)(t)|≤12˜E‖bϵj‖2L2(Q)+α2˜E‖aϵj‖H10+˜E∫t0|(fϵj(s,x,∇uϵj),uϵjt)|ds+12∫t0‖gϵj(s,uϵjt)‖2L2(Q)ds. |
Thanks to the assumptions on the data
|Eϵj(uϵj)(t)|≤C,∀t∈[0,T]. |
For any
|Eϵj(uϵj)(t+h)−Eϵj(uϵj)(t)|≤˜E∫t+ht|(fϵj(s,x,∇uϵj),uϵjt)|ds+12˜E∫t+ht‖gϵj(s,uϵjt)‖2L2(Q)ds. |
Again assumptions (A3), (A5) and Cauchy-Schwarz's inequality, give
|Eϵj(uϵj)(t+h)−Eϵj(uϵj)(t)|≤C(h+h12). |
This implies the equicontinuity of the sequence
In this section, we establish a corrector result stated in the following
Theorem 7.1. Let the assumptions of Theorems 5.1 and 6.1 be fulfilled. Assume that
1r+1s=12. |
Then
uϵjt−ut−ϵju1t(⋅,⋅,⋅ϵj)→0 strongly inL2(0,T;L2(Q))˜P−a.s., | (68) |
uϵj−u−ϵju1(⋅,⋅,⋅ϵj)→0 strongly inL2(0,T;H1(Q))˜P−a.s.. | (69) |
Proof. It is easy to see that
limϵj→0ϵju1t(⋅,⋅,⋅ϵj)→0inL2(0,T;L2(Q))˜P−a.s.. |
Then convergence 20 gives
uϵjt−ut−ϵju1t(⋅,⋅,.ϵj)→0inL2(0,T;L2(Q))˜P−a.s.. |
Thus 68 holds. Similarly we show that
uϵj−u−ϵju1(⋅,⋅,⋅ϵj)→0strongly inL2(0,T;L2(Q))˜P−a.s.. |
It remains to show that
∇(uϵj−u−ϵju1(⋅,⋅,⋅ϵj))→0strongly inL2(0,T;[L2(Q)]n)˜P−a.s.. |
We have
∇(uϵj−u−ϵju1(⋅,⋅,⋅ϵj))=∇uϵj−∇u−∇yu1(⋅,⋅,⋅ϵj))−ϵj∇u1(⋅,⋅,⋅ϵj)). |
Again
limϵj→0ϵj∇u1(⋅,⋅,⋅ϵj)→0inL2(0,T;[L2(Q)]n),˜P−a.s.. |
Now from the ellipticity assumption on the matrix
αE∫T0‖∇uϵj−∇u−∇yu1(⋅,⋅,⋅ϵj)‖2L2(Q)dt≤E∫T0∫QA(xϵj)(∇uϵj−∇u−∇yu1(⋅,⋅,⋅ϵj))⋅(∇uϵj−∇u−∇yu1(⋅,⋅,⋅ϵj))dxdt=E∫T0∫QAϵj∇uϵj⋅∇uϵjdxdt−2E∫T0∫Q∇uϵjA(xϵj)⋅(∇u+∇yu1(⋅,⋅,⋅ϵj))dxdt+E∫T0∫QA(xϵj)(∇u+∇yu1(⋅,⋅,⋅ϵj))⋅(∇u+∇yu1(⋅,⋅,⋅ϵj))dxdt. | (70) |
Let us pass to the limit in this inequality. We start with
E∫QAϵj∇uϵj⋅∇uϵjdx. |
From the convergence of the energies in Theorem 6.1 and using 63 and 60, we have
limϵj→0E∫QAϵj∇uϵj⋅∇uϵjdx=E∫Q×YA(y)⋅[∇xu(t,x)+∇yu1(t,x,y)]⋅[∇xu(t,x)+∇yu1(t,x,y)]dydx. | (71) |
Next, using the two-scale convergence of
limϵj→0∫T0∫Q∇uϵj(t,x)⋅A(xϵj)⋅(∇u+∇yu1(t,x,xϵj))dxdt=∫T0∫Q×Y(∇u(t,x)+∇yu1(t,x,y))⋅A(y)⋅(∇u(t,x)+∇yu1(t,x,y))dxdydt. | (72) |
Now, let us write
ψ(t,x,y)=A(y)(∇u(t,x)+∇yu1(t,x,y))⋅(∇u(t,x)+∇yu1(t,x,y))=A(y)∇u(t,x)⋅∇u(t,x)+2A(y)∇u(t,x)⋅∇yu1(t,x,y)+A(y)∇yu1(t,x,y)⋅∇yu1(t,x,y). |
For
ψ(t,x,y)=A(y)∇u(t,x)⋅∇u(t,x)−2A(y)∇u(t,x)⋅∇y[χ(y)⋅∇xu(t,x)]+A(y)∇y[χ(y)⋅∇xu(t,x)]∇y[χ(y)⋅∇xu(t,x)]. |
Now using
limϵj→0∫T0∫QA(xϵj)(∇u(t,x)+∇yu1(t,x,xϵj))⋅(∇u(t,x)+∇yu1(t,x,yϵj))dxdt=∫T0∫Q×YA(y)(∇u(t,x)+∇yu1(t,x,y))⋅(∇u(t,x)+∇yu1(t,x,y))dxdydt. | (73) |
Combining 71, 72 and 73 with 70, we deduce that
limϵj→0E∫T0‖∇uϵj−∇u−∇yu1(.,.,.ϵj)‖2L2(Q)dt=0˜P−a.s.. |
Thus the proof is complete.
As a closing remark, we note that our results can readily be extended to the case of infinite dimensional Wiener processes taking values in appropriate Hilbert spaces; for instance cylindrical Wiener processes.
The authors express their deepest gratitude to the reviewers for their careful reading of the paper and their insightful comments which have improved the paper. Part of this work was conducted when the first author visited the African Institute for Mathematical Sciences (AIMS), South Africa, he is grateful to the generous hospitality of AIMS.
[1] |
E. Brinda, A. Rajkumar, J. Attermann, U. Gerdtham, U. Enemark, K. Jacob, Health, Social, and Economic Variables Associated with Depression Among Older People in Low and Middle Income Countries: World Health Organization Study on Global AGEing and Adult Health, Am. J. Geriatr. Psychiatry, 24 (2016), 1196–1208. https://doi.org/10.1016/j.jagp.2016.07.016 doi: 10.1016/j.jagp.2016.07.016
![]() |
[2] |
A. Leuchter, I. Cook, D. Debrota, A. Hunter, W. Potter, C. McGrouther, et al., Changes in brain function during administration of Venlafaxine or Placebo to normal subjects, Clin. EEG Neurosci., 39 (2008), 175–181. https://doi.org/10.1177/155005940803900405 doi: 10.1177/155005940803900405
![]() |
[3] |
R. Wichers, J. Findon, A. Jelsma, V. Giampietro, V. Stoencheva, D. Robertson, et al., Modulation of brain activation during executive functioning in autism with citalopram, Transl. Psychiatry, 9 (2019). https://doi.org/10.1038/s41398-019-0641-0 doi: 10.1038/s41398-019-0641-0
![]() |
[4] |
J. Somberg, American Journal of Therapeutics, Am. J. Ther., 9 (2002), 1–1. https://doi.org/10.1097/00045391-200201000-00001 doi: 10.1097/00045391-200201000-00001
![]() |
[5] |
Y. Noda, Potential neurophysiological mechanisms of 1Hz-TMS to the Right Prefrontal Cortex for depression: An exploratory TMS-EEG study in healthy participants, J. Pers. Med., 11 (2021). https://doi.org/10.3390/jpm11020068 doi: 10.3390/jpm11020068
![]() |
[6] |
T. Hutton, S. Aaronson, L. Carpenter, K. Pages, W. Scott, C. Kraemer, et al., The anxiolytic and antidepressant effects of transcranial magnetic stimulation in patients with anxious depression, J. Clin. Psychiatry, 84 (2023). https://doi.org/10.4088/JCP.22m14571 doi: 10.4088/JCP.22m14571
![]() |
[7] |
N. Ichikawa, G. Lisi, N. Yahata, G. Okada, M. Takamura, R. Hashimoto, et al., Primary functional brain connections associated with melancholic major depressive disorder and modulation by antidepressants, Sci. Rep., 10 (2020). https://doi.org/10.1038/s41598-020-73436-y doi: 10.1038/s41598-020-73436-y
![]() |
[8] |
H. Li, W. Yan, Q. Wang, L. Liu, X. Lin, X. Zhu, et al., Mindfulness-based cognitive therapy regulates brain connectivity in patients with late-life depression, Front. Psychiatry, 13 (2022). https://doi.org/10.3389/fpsyt.2022.841461 doi: 10.3389/fpsyt.2022.841461
![]() |
[9] |
C. Greco, O. Matarazzo, G. Cordasco, A. Vinciarelli, Z. Callejas, A. Esposito, Discriminative power of EEG-based biomarkers in major depressive disorder: A systematic review, IEEE Access, 9 (2021), 112850–112870. https://doi.org/10.1109/access.2021.3103047 doi: 10.1109/access.2021.3103047
![]() |
[10] |
A. Damborska, E. Honzirkova, R. Bartecek, J. Hořínková, S. Fedorová, S. Ondruš, et al., Altered directed functional connectivity of the right amygdala in depression: High-density EEG study, Sci. Rep., 10 (2020). https://doi.org/10.1038/s41598-020-61264-z doi: 10.1038/s41598-020-61264-z
![]() |
[11] |
E. Olejarczyk, A. Jozwik, V. Valiulis, K. Dapsys, G. Gerulskis, A. Germanavicius, Statistical analysis of graph-theoretic indices to study eeg-tms connectivity in patients with depression, Front. Neuroinform, 15 (2021). https://doi.org/10.3389/fninf.2021.651082 doi: 10.3389/fninf.2021.651082
![]() |
[12] |
F. Hasanzadeh, M. Mohebbi, R. Rostami, Graph theory analysis of directed functional brain networks in major depressive disorder based on EEG signal, J. Neural Eng., 17 (2020). https://doi.org/10.1088/1741-2552/ab7613 doi: 10.1088/1741-2552/ab7613
![]() |
[13] |
H. Lee, G. Mashour, G. Noh, S. Kim, U. Lee, Reconfiguration of Network Hub Structure after Propofol-induced Unconsciousness, Anesthesiology, 119 (2013), 1347–1359. https://doi.org/10.1097/ALN.0b013e3182a8ec8c doi: 10.1097/ALN.0b013e3182a8ec8c
![]() |
[14] |
R. Pascual-Marqui, D. Lehmann, M. Koukkou, K. Kochi, P. Anderer, B. Saletu, et al., Assessing interactions in the brain with exact low-resolution electromagnetic tomography, Philos. Trans. R. Soc. A, 369 (2011), 3768–3784. https://doi.org/10.1098/rsta.2011.0081 doi: 10.1098/rsta.2011.0081
![]() |
[15] |
F. Tadel, S. Baillet, J. Mosher, D. Pantazis, R. Leahy, Brainstorm: A user-friendly application for MEG/EEG analysis, Comput. Intell. Neurosci., 10 (2011). https://doi.org/10.1155/2011/879716 doi: 10.1155/2011/879716
![]() |
[16] |
A. Gramfort, T. Papadopoulo, E. Olivi, M. Clerc, OpenMEEG: Opensource software for quasistatic bioelectromagnetics, Biomed. Eng. Online, 9 (2010). https://doi.org/10.1186/1475-925x-9-45 doi: 10.1186/1475-925x-9-45
![]() |
[17] |
R. Desikan, F. Segonne, B. Fischl, B. Quinn, B. Dickerson, D. Buckner, et al., An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest, Neuroimage, 31 (2006), 968–980. https://doi.org/10.1016/j.neuroimage.2006.01.021 doi: 10.1016/j.neuroimage.2006.01.021
![]() |
[18] |
J. Waller, T. Miao, I. Ikedionwu, K. Lin, Reviewing applications of structural and functional MRI for bipolar disorder, Jpn. J. Radiol., 39 (2021), 414–423. https://doi.org/10.1007/s11604-020-01074-5 doi: 10.1007/s11604-020-01074-5
![]() |
[19] | L. Orgo, M. Bachmann, K. Kalev, Resting EEG functional connectivity and graph theoretical measures for discrimination of depression, in 2017 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI), Orlando, FL, USA, (2017), 389–392. https://doi.org/10.1109/BHI.2017.7897287 |
[20] |
V. Grin-Yatsenko, I. Baas, V. Ponomarev, J. Kropotov, EEG Power Spectra at Early Stages of Depressive Disorders, J. Clin. Neurophysiol., 26 (2009), 401–406. https://doi.org/10.1097/WNP.0b013e3181c298fe doi: 10.1097/WNP.0b013e3181c298fe
![]() |
[21] |
W. Peng, Z. Jia, X. Huang, X. Huang, S. Lui, W. Kuang, et al., Brain structural abnormalities in emotional regulation and sensory processing regions associated with anxious depression, Prog. Neuro-Psychopharmacol. Biol. Psychiatry, 94 (2019). https://doi.org/10.1016/j.pnpbp.2019.109676 doi: 10.1016/j.pnpbp.2019.109676
![]() |
[22] |
C. Dell'acqua, S. Ghiasi, S. M. Benvenuti, A. Greco, C. Gentili, G. Valenza, Increased functional connectivity within alpha and theta frequency bands in dysphoria: A resting-state EEG study, J. Affect Disord., 281 (2021), 199–207. https://doi.org/10.1016/j.jad.2020.12.015 doi: 10.1016/j.jad.2020.12.015
![]() |
[23] |
P. Cristancho, N. Trapp, S. Siddiqi, D. Dixon, J. Miller, E. Lenze, Crossover to Bilateral repetitive transcranial magnetic stimulation a potential strategy when patients are not responding to unilateral left-sided high-frequency repetitive transcranial magnetic stimulation, J. ECT, 35 (2019), 3–5. https://doi.org/10.1097/yct.0000000000000500 doi: 10.1097/yct.0000000000000500
![]() |
[24] |
J. Maller, R. Thomson, J. Rosenfeld, R. Anderson, Z. Daskalakis, P. Fitzgerald, Reply: Occipital bending in depression, Brain, 138 (2015). https://doi.org/10.1093/brain/awu199 doi: 10.1093/brain/awu199
![]() |
[25] |
H. Zhou, X. Chen, Y. Shen, L. Li, N. Chen, Z.Zhu, Rumination and the default mode network: Meta-analysis of brain imaging studies and implications for depression, Neuroimage, 206 (2020). https://doi.org/10.1016/j.neuroimage.2019.116287 doi: 10.1016/j.neuroimage.2019.116287
![]() |
[26] |
H. Wu, T. Li, C. Peng, C. Yang, Y. Bian, X. Li, The right prefrontal cortex (PFC) can distinguish anxious depression from non-anxious depression: A promising functional near infrared spectroscopy study (fNIRS), J. Affect. Disord., 317 (2022), 319–328. https://doi.org/10.1016/j.jad.2022.08.024 doi: 10.1016/j.jad.2022.08.024
![]() |
[27] |
Y. Gao, X. Yao, Q. Yu, Y. An, Z. Chen, J. Yi, et al., Resting-state functional magnetic resonance study of the brain's network of the temporal lobe epilepsy patients with depression, Zhonghua yi xue za zhi, 96 (2016), 1696–1698. https://doi.org/10.3760/cma.j.issn.0376-2491.2016.21.017 doi: 10.3760/cma.j.issn.0376-2491.2016.21.017
![]() |
[28] |
Y. Hou, S. Zhang, N. Li, Z. Huang, W. Li, Y. Wang, Neurofeedback training improves anxiety trait and depressive symptom in GAD, Brain Behav., 11 (2021). https://doi.org/10.1002/brb3.2024 doi: 10.1002/brb3.2024
![]() |
[29] |
G. Northoff, How do resting state changes in depression translate into psychopathological symptoms? From 'Spatiotemporal correspondence' to 'Spatiotemporal Psychopathology', Curr. Opin. Psychiatry, 29 (2016), 18–24. https://doi.org/10.1097/yco.0000000000000222 doi: 10.1097/yco.0000000000000222
![]() |
1. | Hermano Frid, Kenneth H. Karlsen, Daniel Marroquin, Homogenization of stochastic conservation laws with multiplicative noise, 2022, 283, 00221236, 109620, 10.1016/j.jfa.2022.109620 | |
2. | Mogtaba Mohammed, Well-Posedness for Nonlinear Parabolic Stochastic Differential Equations with Nonlinear Robin Conditions, 2022, 14, 2073-8994, 1722, 10.3390/sym14081722 | |
3. | Mogtaba Mohammed, Homogenization of nonlinear hyperbolic problem with a dynamical boundary condition, 2023, 8, 2473-6988, 12093, 10.3934/math.2023609 | |
4. | Mogtaba Mohammed, Noor Ahmed, Homogenization and correctors of Robin problem for linear stochastic equations in periodically perforated domains, 2020, 120, 18758576, 123, 10.3233/ASY-191582 | |
5. | Chigoziem Emereuwa, Mogtaba Mohammed, Homogenization of a stochastic model of a single phase flow in partially fissured media, 2022, 129, 18758576, 413, 10.3233/ASY-211735 | |
6. | Mogtaba Mohammed, Waseem Asghar Khan, Homogenization and Correctors for Stochastic Hyperbolic Equations in Domains with Periodically Distributed Holes, 2021, 12, 1756-9737, 10.1142/S1756973721500086 | |
7. | Hakima Bessaih, Mogtaba Mohammed, Ismail M. Tayel, Homogenization and corrector results for a stochastic coupled thermoelastic model, 2024, 0, 1078-0947, 0, 10.3934/dcds.2024168 |