<div dir="ltr">Seems those last few significant figures were significant after all. Now to figure out why getting rid of them gets a better answer than leaving them in...<div><br></div><div>Thanks for the help, I'll root around a bit more.<br>
</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On 9 July 2014 09:41, 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">
Andrew Cramer <<a href="mailto:andrewdalecramer@gmail.com">andrewdalecramer@gmail.com</a>> writes:<br>
<br>
> I get *significantly *different results from backslash and petsc using<br>
> the *same<br>
> matrix*.<br>
<div class="">><br>
> Deflection |Au-b|/|b|<br>
> Euler-Bernoulli: 4000<br>
> Petsc: 4386 0.3<br>
> Matlab: 4013 1.6e-10<br>
><br>
> I'm exporting to 6 significant figures. Condition number is rather poor<br>
> (~2e7) but I get similar or worse results with problems having lower<br>
> conditions numbers (~1e3).<br>
<br>
</div>Are you sure the matrices are the same? Use "-ksp_view_mat binary",<br>
then in MATLAB, A = PetscBinaryRead('binaryoutput') if you want to get<br>
both matrices in the same environment. Or explicitly compute the norms<br>
in PETSc.<br>
<br>
To debug, start with the simplest and most well-conditioned problem<br>
that exhibits the issue.<br>
</blockquote></div><br></div>