<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<tt>The problem most likely is loss of accuracy in calculation of
block coordinates. Vector of coordinates tested in the
interpolation package is a direct copy of FLASH mesh coordinates
via<br>
<br>
Grid_getCellCoords<br>
<br>
You can find those calls in amr_prolong_gen_unk1_fun. I suggest
you check if </tt><tt><tt>Grid_getCellCoords </tt>returns
monotonically increasing values when you use 30 refinement levels.
If not, the problem is with computing block coordinates and not
with the interpolation package.<br>
<br>
Tomek<br>
--</tt><br>
<div class="moz-cite-prefix">On 04/23/15 13:00, Joshua Wall wrote:<br>
</div>
<blockquote
cite="mid:CAHnUEowCn4CLPjwr-wNxykDDXbAC=W=6MtWF1inzHwsxfjoNgQ@mail.gmail.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
<div dir="ltr">
<div>
<div>
<div>Klaus and Tomasz,<br>
<br>
</div>
It was indeed corrected by reducing the refinement. I
had intended to not push up to 30 levels on this particular
machine, and may have just run out of memory/procs. I am
going to be using 30 levels soon however, and I'm assuming
that the umap subroutines can handle these levels of
refinement (when I have the requisite memory/procs)?<br>
<br>
</div>
Thanks,<br>
<br>
</div>
Josh<br>
<br>
<div class="gmail_quote">On Thu, Apr 23, 2015 at 12:53 PM Klaus
Weide <<a moz-do-not-send="true"
href="mailto:klaus@flash.uchicago.edu">klaus@flash.uchicago.edu</a>>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">On Thu, 23
Apr 2015, Joshua Wall wrote:<br>
<br>
> > I'm receiving an odd error from FLASH and I'm
not sure how to<br>
> > interpret it. Perhaps someone can enlighten me?
Its:<br>
> ><br>
> > **ERROR** - ierr from UMAPn is 421 ,
isg= 4<br>
> > DRIVER_ABORT: ERROR - ierr from UMAPn is not 0<br>
<br>
The message indicates that the error occurs when
interpolating data<br>
for block 4 (in some proc), and it is complainign about the
X coordinate.<br>
<br>
As Tomek wrote, this may all be because of the high
refinement level.<br>
Or...<br>
<br>
* Check your parfile, maybe xmin is the same as xmax?
(forgotten /<br>
duplicate sign ?)<br>
<br>
* Check that umap.F is compiled correctly, the compiler may
somhow act<br>
differently, this is about the only file that has a .F
suffix instead<br>
of .F90. In particular, check that flags promoting real to
double<br>
precision are there.<br>
<br>
Just some ideas.<br>
<br>
Klaus<br>
</blockquote>
</div>
</div>
</blockquote>
<br>
</body>
</html>