Higher approximations of the averaging method for quasilinear parabolic equations with fast oscillating principal part in the case of the Cauchy problem (Q1569760)

The author considers the Cauchy problem for a quasilinear parabolic equation of the form \[ \frac{\partial u}{\partial t}=\sum_{|\alpha|=2k} a_{\alpha}\left(x, \delta^{2k-1} u,t,wt\right) D^\alpha u+f\left(x,\delta^{2k-1}u,t,wt\right), \] \[ u(x,0)=\varphi(x), \] where \(x=(x_1,x_2,\ldots,x_n)\) is an arbitrary point in \(\mathbb{R}^m\) , \(w\) is a large parameter, \(\alpha=\)\break \((\alpha_1,\alpha_2,\ldots,\alpha_m)\) is a multi-index and \(\delta^{2k-1}u\) is the vector function composed of all possible derivatives of \(u(x,t)\) with respect to \(x\) of order less or equal to \(2k-1\). The functions \(a_\alpha(x,e,t,\tau)\), \(f(x,e,t,\tau)\) and their derivatives \(\frac{\partial a_\alpha}{\partial e_i}\), \(\frac{\partial f}{\partial e_i}\), with respect to the components of \(e\) are defined, continuous and uniformly bounded. These functions and their derivatives satisfy the Hölder condition with respect to the variables \(x,e,t,\tau\). The equation (1) is uniformly parabolic, i.e., for all real vectors \(\sigma=(\sigma_1,\sigma_2,\ldots,\sigma_m)\) the estimate \[ (-1)^{k+1} \text{Re} \sum_{|\alpha|=2k}a_\alpha(x,e,t,\tau)\sigma^\alpha \geqslant\delta|\sigma|^{2k} \] holds, where the constant \(\delta\) does not depend on \((x,e,t,\tau)\). An algorithm for constructing the higher approximations of the averaging method for the equation (1) with fast oscillating coefficients is suggested and justified.