Dear Jonas Latt and all,
Does OpenLB solves for entropy estimate in entropic LB models (by Ansumali and Karlin) ? I am learning about ELBM. I need info regrading how to solve for parameter $\alpha$ in ELBM formulation ?
Thanks and regards,
Dhiraj.
Hello,
yes it does. You will find the entropic dynamics in the complexDynamics/entropicDynamics.* files.
Yes I looked at subroutines. Thank you. I could understand that the entropyGrowth and entropyGrowthDerivative being used with Newton’s method with an initial arbitrary guess to iterate the root. Can’t we do the bracketting of root between maximum \alpha (\alpha_{max}) and minimum \alpha as suggested by Ansumali (Ansumali Thesis, ETH) and then use either bisection method/Newton’s method ?
What value of \alpha being used if the Newton’s method fails to converge ? And, what value should be used when Entropy growth failes to converge ?
Thanks and regards,
Dhiraj.
The choice made here was to use the initial guess being alpha=2 because the final solution must be close to this value (except maybe if the resolution is really too small and therefore the physics are not correctly represented anymore). If the entropy fails to converge, then one may take the value alpha=2 (which is equivalent to BGK by the way) but I’m not sure if it is a good choice. Usually the non convergence of the Newton scheme will not happen when you are resolving your system enough.