Hi,

Another reason that we utilize 64 bit precision is in developing
algorithms.  In particular, we rely heavily on diagnostics that detect
whether tracer is conserved over a single time step.  Numerous times one
will code a new algorithm and make a mistake, and this mistake typically
shows up clearly with the tracer time step diagnostics.  The difference
between precision of 1e-8 and 1e-16, or thereabouts, is the difference
between a properly coded scheme and one that is not coded properly. 
Similar arguments hold for energetic consistency diagnostics derived
from the momentum equation. 

This argument for using full 64 bit precision is not an argument for
always using such, since one could conceivably develop algorithms with
64 bit and turn it off for production runs.   I believe the strongest
arguments for 64 bits is what Pacanowski said in his email.  We need to
maintain negligible levels of spurious roundoff errors in the
discretization of transport operators, especially where those operators
act on a background meanstate with tight gradients (such as the oceanic
pycnocline).  

Hope this helps to explain why we remain faithful to the 64 bit
precision in mom4. I suspect similar arguments are used for other models
as well. 

-Stephen 

-- 
Dr. Stephen M. Griffies                phone: +1-609-452-6672
Geophysical Fluid Dynamics Laboratory  FAX:   +1-609-987-5063
P.O. Box 308, Forrestal Campus Site B  email: [EMAIL PROTECTED]
Princeton, NJ 08542  USA               http://www.gfdl.noaa.gov/~smg


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0
_______________________________________________
FMS-mom4 mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fms-mom4

Reply via email to