Averaging principle for slow-fast stochastic partial differential equations with Hölder continuous coefficients (Q2208449): Difference between revisions

The authors consider a system of semilinear stochastic evolution equations in a Hilbert space \(H\) driven by additive noises \begin{align*} dX^\varepsilon&=[AX^\varepsilon+B(X^\varepsilon,Y^\varepsilon)]\,dt+\sqrt{Q_1}\,dW^1,&X^\varepsilon(0)&=x\in H, \\ dY^\varepsilon&=\frac{1}{\varepsilon}[AY^\varepsilon+F(X^\varepsilon,Y^\varepsilon)]\,dt+\frac{1}{\sqrt{\varepsilon}}\sqrt{Q_2}\,dW^2,&Y^\varepsilon(0)&=y\in H, \end{align*} where \(A\) is a strictly negative self-adjoint operator on \(H\), \(B,F:H\times H\to H\) are bounded and satisfy \begin{align*} |B(x_1,y_1)-B(x_2,y_2)|&\le C(|x_1-x_2|^\alpha+|y_1-y_2|^\beta) \\ |F(x_1,y_1)-F(x_2,y_2)|&\le C|x_1-x_2|^\gamma+L_F|y_1-y_2| \end{align*} for some \(\alpha,\beta,\gamma\in(0,1]\), \(Q_1\) and \(Q_2\) are non-negative self-adjoint bounded operators on \(H\), \(W_1\) and \(W_2\) are independent cylindrical Wiener processes on \(H\) and additional assumptions are made on \(L_F\) and spectral parameters of \(A\), \(Q_1\) and \(Q_2\). Then \[ \lim_{\varepsilon\to 0+}\mathbb E\left(\sup_{t\in[0,T]}|X^\varepsilon(t)-\bar X(t)|^p\right)=0 \] for every \(p>0\) and \(T>0\), where \(\bar X\) satisfies \[ d\bar X=[A\bar X+\bar B(\bar X)]\,dt+\sqrt{Q_1}\,dW^1,\qquad \bar X(0)=x \] for \(\bar B:H\to H\) defined by the formula \[ \bar B(z)=\int_HB(z,\cdot)\,d\mu^z \] where \(\mu^z\) is the unique invariant probability Borel measure on \(H\) of the transition semigroup associated with the frozen system \[ dY=[AY+F(z,Y)]\,dt+\sqrt{Q_2}\,dW^2. \]