Hi guys! I have just subscribed to the mailing list. Recently I've
been using Maxima a lot for analyzing the performance of computer
algorithms. I'm surprised at how easy it is to master. There are,
however, some things that are non-obvious.
I have run into a problem which I cannot get past, though. You see, I
am using Maxima to do some curve fitting, and then count an integral.
However, sometimes the curve fits have parameters which turn out to be
"arbitrary". This isn't really so, but I guess the newton-raphson gets
confused around 0. For example it'll return things such as:
3^2 + 0.22e-40*%r21 where obviously any small changes to r21 will
not change the complete value at all (the changes to %r21 get
truncated).
The objective of the maxima code I am executing is to calculate those
integrals and put them out to stdout and have the Python script (which
does a lot of other work) take these numbers and continue analysis. Of
course, Python cannot interpret %r21 and throws exceptions. Therefore
I thought I would just plug something into those arbitrary values. I
did something like this:
arbitrary_values: [%r0 = 1, %r1 = 1, ..., %r30 = 1]
output: ev(calculated_integral,arbitrary_values,nouns,numer)
however, when I do that, I get lisp reporting errors - floating point overflow:
Maxima encountered a Lisp error:
arithmetic error FLOATING-POINT-OVERFLOW signalled
Automatically continuing. To enable the Lisp debugger set
*debugger-hook* to nil.
(note the formatting may be incorrect because of the way my program
gets this output from maxima through stdout)
What can I do?
I am on Ubuntu GNU/Linux 10.04 and I understand that maxima is using
the SBCL - Steel Bank Common Lisp - to run its computational engine.
If anyone has any ideas, I would be very thankful. You'd help not only
me but also the users of the software which will be released under
GPL.
Thanks for reading!