suggestion for solving the Infinity model output?

[changks888]

Hi all modeler,

I'm up to here to ask about how to solve the model output as "Infinity".

Some of the calculations are"Infinity" and Matlab can not recognize it when I load the output in Matlab.

I already look at this document:

http://www.intel.com/software/products/compilers/flin/docs/main_for/mergedProjects/bldaps_for/using_the_floating_point_exception_handling_(-fpe)_option.htm

Thx,

Michael

[TimP · Accepted solution]

Is -Infinity a valid result of your analysis? According to search engine queries, Matlab should accept -Inf. So, where you WRITE values which may take this value, you could modify your WRITE along the line of
if(x < -huge(x)then
write(*,*)'-Inf'
else
write(*,*)x
endif

If necessary, you could
write(*,*) max(-huge(x),x) so as to replace -Infinity with the closest numeric value.
or, should we take your reference to exception handling to imply that you consider this erroneous, and want suggestions for debugging?

[changks888]

Hi Tim,

-Infinity is a valid result in the numerical model. Matlab accept -Inf after loading the file, however the it's such painful to load a file contains strings. Your tips is helpful. Right now, I generally parse the -Infinity as NaN in another text editor.

-Michael