<div dir="ltr">Hello,<div><br></div><div>Is it possible to partition a parallel matrix in PETSc as follows:
</div><div><br></div><div><pre>A B C
D E F
G H I</pre><pre>The blocks A-D-G belong to processor 0 (A is the diagonal block, D and G are off-diagonal blocks)</pre><pre>The blocks B-E-H belong to processor 1 (<span style="font-family:arial">E is the diagonal block, B and H are off-diagonal blocks)</span></pre>
<pre>The blocks C-F-I belong to processor 2 <span style="font-family:arial">(</span><span style="font-family:arial">I is the diagonal block, C and F are off-diagonal blocks)</span></pre><pre><br></pre><pre>Or, is it possible to have nine processors and each has a block of the matrix above? Block A belongs to processor 0, block B belongs to processor 1, and so on...</pre>
<pre><br></pre><pre>As far as I read from documentation, PETSc always seems to be considering a rowwise partitioning, where blocks A-B-C belong to processor 0, D-E-F belong to processor 1 and G-H-I belong to processor 2. </pre>
<pre><span style="font-family:arial">Is there a way to obtain the partitioning schemes I described above?</span><br></pre><pre><span style="font-family:arial"><br></span></pre><pre><span style="font-family:arial">Thanks in advance.</span></pre>
</div></div>