I don't believe so. Not without extreme measures.

On Fri, Jan 7, 2022 at 8:39 AM Gong Ding <[email protected]> wrote:

> Dear All,
>
> I had a question. For petsc configured with 128bit float, Is it possible
> to use 64bit direct solver as mixed precision preconditioner for krylov
> subspace iterative method?
>
> If so , such as MUMPS or Pardiso can be used as efficient parallel LU 
> preconditioner
> I think.
>
> Regards
>
> Gong Ding
>
>
>

Reply via email to