Imprecise addition in matlab R14SP2 under fedora 13, library mismatch?

Hi. I am running a 32 bit matlab R14 SP2 on 64 bit fedora 13 (I use the -glnx86 architecture flag). When I do a simple calculation I am getting what looks like roundoff or significant digit errors. Example:

 >> a = .43 + .33 + .52 + .58
a = 1.8600e+00      : so good so far!
>> a - 1.86
ans = -2.2204e-16   :  should be 0

Obviously I shouldn't be having these errors. I have used matlab a long time without any such issues. But I did upgrade to fedora 13 a while ago so I'm wondering if there is some library mis-match. Has anyone seen such a problem? what is the solution? thanks. Larry

[Jan]

This behaviour does not concern only your Matlab version and Linux. It appears in a C-program under Windows also.

Answer 1

[the cyclist]

I suggest you take a look at the answers (especially the accepted one) here:

http://www.mathworks.com/matlabcentral/answers/69-why-does-1-2-3-1-3-not-equal-zero

I'm a little surprised that you have not seen this behavior before, if you have been using MATLAB a long time.

[Walter Roberson]

http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F