itaps-parallel Comments on Tutorial Slides

[Devine, Karen D]

Lori:

The tutorial slides look quite good!  Thank you for your effort putting them
together.

I noticed a few details; I was looking at file
ITAPS_SC08_Tutorial_Draft-09-16.pdf.

-  In the figure on slide 7 (Unstructured Meshes on Parallel Computers), the
labels on the figure are slightly garbled in the PDF version.  In
particular, the "M" labels are hard to read.  Since we don't define them or
use them later, perhaps we can just remove all of the "M" labels.  The "P"
labels are OK; people will assume they are processes.

-  On slide 64 (The ITAPS data model abstracts...), the animation covers the
text in the PDF version, and it is covering key definitions.  Since we have
to submit PDF, perhaps we should have one slide without the animation and
one with it, so the definitions appear in the PDF.

-  On slide 65 (The ITAPS data model has four...), there is one of those odd
carriage return characters at the end of the "Entity" bullet.

-  In the example on slide 115 (Ghost entities), we use ITAPS terminology in
a natural but technically incorrect way.  Our example is 2D, so the
ghost-entity dim and bridge-entity dim are correct.  But the text talks
about regions being ghosted across faces.  In ITAPS terminology, we are
really ghosting faces across edges.  But it sounds odd to call 2D finite
elements "faces."  And a 3D example would look too complicated.  Probably
the best solution is to describe the example as "one layer of ghost
triangles for all boundary triangles sharing edges."

Thanks again!
Karen