Re: single function

[James Sherman Jr.]

I believe that is to be expected from single floating point precision.   (eps(single(1)) = 1.92e-7)  Because you're using single precision numbers, any calculations are only reliable up to +/- that number.  It has nothing to do with the function "single" but with floating point calculations.  The wikipedia article probably describes it better than I can.

http://en.wikipedia.org/wiki/Machine_epsilon

2010/1/26 Kuznetsov Roman <address@hidden>

I don't know is it normal but I've faced with this

DATA="" href="http://www.mathworks.com/access/helpdesk/help/techdoc/ref/struct.html" target="_blank">struct("T",[-2:0.3:4]');

#a little experiment

octave:3> (DATA.T(2)-DATA.T(1))-(DATA.T(3)-DATA.T(2))

ans = 0

octave:4> DATA.T=single(DATA.T);

octave:5> (DATA.T(2)-DATA.T(1))-(DATA.T(3)-DATA.T(2))

ans = -1.1921e-07

So you can see that "single" function don't work correct.

_______________________________________________
Help-octave mailing list
address@hidden
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave