Emil Constantinescu emconsta at anl.gov
Tue Sep 11 12:32:42 CDT 2018

On 9/11/18 12:16 PM, Matthew Knepley wrote:
>     Gauss Runge-Kutta (including implicit midpoint) is symplectic and does
>     not require any splitting.
> How do you talk about conservation of the symplectic form? The default 
> model is on R^{2n} and
> you get 1 on one half and -1 on the other half. What do you mean no 
> splitting?
>    Matt

Matt, apply any method from above to R^N, N=2n; i.e., no splitting.


More information about the petsc-dev mailing list