[CLN-list] floating point underflow

Richard B. Kreckel kreckel at ginac.de
Sat Oct 27 21:53:29 CEST 2007


Hi!

Michael Schuster wrote:
> I'm facing a problem using cöln (1.1.9 and 1.1.13) with my 32 Bit Linux-Athlon 
> System here: 
> I have expression like b+(a*a*a-a)*h/2
> Some "a" are rather small, so I retrieve an underflow. I did inhibit this 
> then, but the older version seems to stop calculation in a correct manner 
> (results of other expressions in my programms can not be correct) and the 
> newer says: "Internal error: statement in File ./float/elem/cl_F_mul.cc line 
> 25 has been reached."

In order for anyone to be able to help you it is essential that you 
write a reduced test case where the problem appears and can be investigated.

Best wishes
    -richy.
-- 
Richard B. Kreckel
<http://www.ginac.de/~kreckel/>


More information about the CLN-list mailing list