You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The analytic Jacobian for the original SDC (enabled by setting the SDC preprocessor) is not correct. The Jacobian elements should be derivatives with respect to the conserved variables. The proper way to do this is to construct a second transformation Jacobian, d(rho, rho X, rho E)/d(rho, T, X) and use this to transform the derivatives that come from the burner. This is actually quite easy, and is implemented in the new SDC implementation in the sdc branch of Castro.
With the analytic Jacobian enabled (jacobian=1), this original SDC method surprisingly runs, but very slowly.
The text was updated successfully, but these errors were encountered:
At the moment, our state vector that the ODE integration sees is (rho X, rho E, rho e). We get the Jacobian from the net in terms of rho, T, X. I think in order to do the transformation properly, we should integrate only one of (rho E) or (rho e), and add rho to the ODE integration vector. Then we can do the same transformation as we are doing in the new SDC stuff.
The analytic Jacobian for the original SDC (enabled by setting the
SDC
preprocessor) is not correct. The Jacobian elements should be derivatives with respect to the conserved variables. The proper way to do this is to construct a second transformation Jacobian, d(rho, rho X, rho E)/d(rho, T, X) and use this to transform the derivatives that come from the burner. This is actually quite easy, and is implemented in the new SDC implementation in thesdc
branch of Castro.With the analytic Jacobian enabled (
jacobian=1
), this original SDC method surprisingly runs, but very slowly.The text was updated successfully, but these errors were encountered: