A method of differential iteration for boundary value problems in extended thermodynamics (Q2371007)

Extended thermodynamics is a nonequilibrium thermodynamic theory where the fluxes are incorporated to the set of basic independent variables [\textit{I. Müller} and \textit{T. Ruggeri}, Rational extended thermodynamics. 2nd ed., New York, NY: Springer (1998; Zbl 0895.00005), \textit{D. Jou, J. Casas-Vázquez} and \textit{G. Lebon}, Extended Irreversible Thermodynamics. 3rd ed., Berlin: Springer (2001; Zbl 0974.74003)]. Since the number of variables is higher than in classical theories, the number of boundary conditions required to solve boundary values uniquely is also higher. How to formulate the additional boundary data is still one of the main problems in extended thermodynamics. Several physical criteria have been proposed for this problem, such as minimax entropy production, boundary layer variational principles, vanishing critical derivatives, minimum iterative differences, or fluctuation theories. In this paper it is proposed an iterative method based on decoupling the system into two subsystems and solving them alternatively with an iterative procedure. One of them (corresponding to the classical variables) can be solved uniquely with the classical boundary conditions, and the other (corresponding to higher-order fluxes) determines the remaining non-equilibrium field variables by direct evaluation. The shearing flow with heat conduction is considered as an illustrative numerical example of the proposed method. Afterwards, the condition for convergence based on the estimated error that can be easily checked in numerical iterations is proved. Stability od uniqueness of the numerical iterative solutions are also considered. Additional examples in heat conduction with higher-order moments, in BGK model and in Maxwellian molecules with slip boundary conditions, are also given to substantiate the main results on convergence, stability and uniqueness.