Evaluation of calculated influences in parabolic Systems. L1-approach (Q4396202)

It is considered the initial-boundary problem \[ {\partial\xi_{0}\over \partial t}=L_{t,x}\xi_{0}+A(t,x,\xi_{0}),\quad \xi_{0}(0,x)=\varphi(x), x\in D \subset {\mathbb{R}}^{n},\quad \xi_{0}(t,x)| _{x\in\partial D}=\Phi(t,x), t\in [0,T], \] where \[ L_{t,x}=\sum\limits_{i,j=1}^{n}a_{ij}(t,x){\partial^{2}\over \partial x_{i}\partial x_{j}}+\sum\limits_{i=1}^{n}b_{i}(t,x){\partial\over\partial x_{i}}. \] The observed motion is described by the equation \[ {\partial\xi_{\varepsilon}\over \partial t}=L_{t,x}\xi_{\varepsilon}+A(t,x,\xi_{\varepsilon})+ \sigma(t,x,\xi_{\varepsilon})\eta(t/\varepsilon), \] \[ \xi_{\varepsilon}(0,x)= \varphi(x), \;x\in D\subset {\mathbb{R}}^{n},\quad \xi_{\varepsilon}(t,x)| _ {x\in\partial D}=\Phi(t,x),\;t\in [0,T], \] where \(\eta(t)\) is a stationary stochastic process with zero mean. The author investigates the large deviations of non-parametric estimate \(A_{\varepsilon}(t,x)\) of \(A(t,x,\xi_{0})\), where \[ A_{\varepsilon}(t,x)={1\over h_{\varepsilon}^{n+1}}\int\limits_{D} \int \limits_{0}^{T}\left[{\partial \xi_{\varepsilon}(s,y)\over\partial s}- L_{s,y}\xi_{\varepsilon}(s,y)\right]K_{t}^{\varepsilon}\left((t-s)/ h_{\varepsilon}\right)K_{x}^{\varepsilon}\left((x-y)/ h_{\varepsilon}\right)dsdy, \] with some kernels \(K_{t}, K_{x}\).