<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Fri, Jun 1, 2018 at 9:21 AM, Samuel Lanthaler <span dir="ltr"><<a href="mailto:s.lanthaler@gmail.com" target="_blank">s.lanthaler@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<br>
<br>
I was wondering what the most efficient way to use MatPtAP would be in the following situation: I am discretizing a PDE system. The discretization yields a matrix A that has a band structure (with k upper and lower bands, say). In order to implement the boundary conditions, I use a transformation matrix P which is essentially the unit matrix, except for the entries P_{ij} where i,j<k and n-i,n-j<k, so<br>
<br>
P = [ B, 0, 0, 0, ..., 0, 0 ]<br>
[ 0, 1, 0, 0, ..., 0, 0 ]<br>
[ ]<br>
[ ]<br>
[ ..., 1, 0 ]<br>
[ 0, 0, 0, 0, ..., 0, C ]<br>
<br>
with B,C are (k-by-k) matrices.<br>
Right now, I'm simply constructing A, P and calling<br>
<br>
CALL MatPtAP(petsc_matA,petsc_matP,<wbr>MAT_INITIAL_MATRIX,PETSC_DEFAU<wbr>LT_REAL,petsc_matPtAP,ierr)<br>
<br>
where I haven't done anything to pestc_matPtAP, prior to this call. Is this the way to do it?<br>
<br>
I'm asking because, currently, setting up the matrices A and P takes very little time, whereas the operation MatPtAP is taking quite long, which seems very odd... The matrices are of type MPIAIJ. In my problem, the total matrix dimension is around 10'000 and the matrix blocks (B,C) are of size ~100.<br></blockquote><div><br></div><div>Are you sure this is what you want to do? Usually BC are local, since by definition PDE are local, and</div><div>are applied pointwise. What kind of BC do you have here?</div><div><br></div><div> Thanks,</div><div><br></div><div> Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Thanks in advance for any ideas.<br>
<br>
Cheers,<br>
Samuel<br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener</div><div><br></div><div><a href="http://www.caam.rice.edu/~mk51/" target="_blank">https://www.cse.buffalo.edu/~knepley/</a><br></div></div></div></div></div>
</div></div>