Anush Krishnan <[email protected]> writes:

> Hi all,
>
> I have two 2D distributed arrays U and V, and a vector created from the
> DMComposite of both. If I know just the Natural Ordering index of an
> element of U  or V, can I calculate the PETSc Ordering index of that
> element in the composite vector?

That direction is fragile and/or expensive unless it is within the
subdomain overlap/ghost region.  Where are you getting this index from?

Attachment: pgpIhTWFibbAl.pgp
Description: PGP signature

Reply via email to