<div dir="ltr">These functions are for eliminating the variables corresponding to active sets when performing matrix, vector operations. I don't think they were meant to be in the public interface, only for utilities in active set algorithms.<div>
<br></div><div>Jason</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Feb 18, 2014 at 9:22 AM, Jed Brown <span dir="ltr"><<a href="mailto:jed@jedbrown.org" target="_blank">jed@jedbrown.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="">Lisandro Dalcin <<a href="mailto:dalcinl@gmail.com">dalcinl@gmail.com</a>> writes:<br>
<br>
> Barry, I've pushed a quick fix (missing PETSC_EXTERN in TaoMonitor())<br>
> in your branch.<br>
><br>
> The following prototypes in petsctao.h are also missing PETSC_EXTERN,<br>
> and I think they should be more properly namespaced.<br>
><br>
> PetscErrorCode VecGetSubVec(Vec, IS, PetscInt, PetscReal, Vec*);<br>
> PetscErrorCode MatGetSubMat(Mat, IS, Vec, TaoSubsetType, Mat*);<br>
><br>
> Perhaps you should move these prototypes (as well as TaoSubsetType) to<br>
> taoimpl.h ? I'm not sure these routines should be part of the public<br>
> interface.<br>
<br>
</div>I would rather not, but if nothing else changes, the PetscInt argument<br>
should be TaoSubsetType. I'm afraid I don't understand exactly what<br>
role this function is serving, but the only TAO reference is the enum,<br>
so it can probably be made generic and moved down to the Vec module.<br>
</blockquote></div><br></div>