On 10/22/19 10:41 AM, Xiang Sun wrote:
>
> In step-31, why did you define the residual by multiplying T^(a-1)? What does
> "a" mean? Thank you very much.
I assume you mean here?
https://dealii.org/developer/doxygen/deal.II/step_31.html#Stabilizationweakformandspacediscretizationforthetemperatureequation
To understand the form stated there, you have to go back to the paper by
Guermond and Popov. In essence, this goes back to some deep theory about
hyperbolic equations that says that a weak solution is a physical solution
only if it satisfies an entropy inequality. There are many ways to define an
entropy pair (a pair of functions), and one of these is to choose E=T^a. The
differential inequality E then has to satisfy is obtained by multiplying the
equation for T by T^{a-1}.
The theory is not particularly simple, and requires substantial background of
hyperbolic equations. Going to the papers by Guermond and coauthors will give
you the necessary insight.
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email:
bang...@colostate.edu
www:
http://www.math.colostate.edu/~bangerth/