Solution diverging on using double precision

qrie · January 22, 2013, 03:11

Hi,

I have an incompressible Navier Stokes code that was working fine as long as I was using single precision. When I converted to double precision, the residues are very high. Why is this happening?

Thanks!
Qrie

Martin Hegedus · January 22, 2013, 11:29

A bug in your code?

Sometimes issues like the following may occur

if (xx == 0.0) {
then do something
}

Under single precision this might work but double does not if xx is very small but not equal to 0.0, for example xx=1.0e-12. Single precision would round value to zero. Error tolerances should be used.

if (abs(xx) < 1.0e-8) {
then do something
}

qrie · January 22, 2013, 23:11

That helps! Thank you!

January 22, 2013, 03:11	Solution diverging on using double precision	#1
qrie New Member Join Date: Jun 2010 Posts: 20 Rep Power: 15	Hi, I have an incompressible Navier Stokes code that was working fine as long as I was using single precision. When I converted to double precision, the residues are very high. Why is this happening? Thanks! Qrie

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Parallel rasInterFoam	openfoam_user	OpenFOAM Running, Solving & CFD	4	November 1, 2008 04:14
When to use double precision	Mark Render	Main CFD Forum	10	January 21, 2008 12:16
re: Double precision and grid quality	Wooster	CFX	5	July 26, 2007 10:10
UDF & Double precision; Please help??	Asghari	FLUENT	0	January 31, 2007 13:14
HELP! Diverging Solution for Incompressible Prob.	Bob	FLUENT	3	April 16, 2006 19:33

January 22, 2013, 11:29		#2
Martin Hegedus Senior Member Martin Hegedus Join Date: Feb 2011 Posts: 500 Rep Power: 19	A bug in your code? Sometimes issues like the following may occur if (xx == 0.0) { then do something } Under single precision this might work but double does not if xx is very small but not equal to 0.0, for example xx=1.0e-12. Single precision would round value to zero. Error tolerances should be used. if (abs(xx) < 1.0e-8) { then do something }

January 22, 2013, 23:11		#3
qrie New Member Join Date: Jun 2010 Posts: 20 Rep Power: 15	That helps! Thank you!