wrong comparison result

[Andy Buckle]

Ido,

this may help you

 > printf('%.18f \n',[[47:49]*[1/49]])
0.959183673469387710
0.979591836734693850
0.999999999999999890

 > x = [[47:49]*[1/49]];
 > abs(x - 1) < 0.00001
ans =

   0   0   1

the small difference that you compare against is often called epsilon.

Andy

On Mon, Mar 2, 2009 at 3:52 PM, Sergei Steshenko <sergstesh at yahoo.com> wrote:
>
>
>
>
> --- On Sun, 3/1/09, Jaroslav Hajek <highegg at gmail.com> wrote:
>
>> From: Jaroslav Hajek <highegg at gmail.com>
>> Subject: Re: wrong comparison result
>> To: idohlp at neto.net.il
>> Cc: bug at octave.org
>> Date: Sunday, March 1, 2009, 11:42 PM
>> On Sat, Feb 28, 2009 at 11:51 PM, Ido Halperin
>> <idohlp at neto.net.il> wrote:
>> > >From ido at login.modem Sun Mar 01 00:45:43 2009
>> > Return-path: <ido at login.modem>
>> > Envelope-to: ido at login.modem
>> > Delivery-date: Sun, 01 Mar 2009 00:45:43 +0200
>> > Received: from ido by login.modem with local (Exim
>> 4.69)
>> >        (envelope-from <ido at login.modem>)
>> >        id 1LdXwF-000126-SO; Sun, 01 Mar 2009
>> 00:45:43 +0200
>> > To: bug at octave.org, ido at login.modem
>> > Subject:  wrong comparison result
>> > Message-Id: <E1LdXwF-000126-SO at login.modem>
>> > From: idohalperin <ido at login.modem>
>> > Date: Sun, 01 Mar 2009 00:45:43 +0200
>> >
>> > To: bug at octave.org
>> > Cc: ido
>> > Subject: wrong comparison result
>> >
>> > Bug report for Octave 3.0.1 configured for
>> i486-pc-linux-gnu
>> >
>> > Description:
>> > -----------
>> >
>> > The next code returns a right result:
>> >
>> > octave:4> [[47:49]/49]==1
>> > ans =
>> >
>> >   0   0   1
>> >
>> >
>> > but this code does not:
>> >
>> > octave:3> [[47:49]*[1/49]]==1
>> > ans =
>> >
>> >   0   0   0
>> >
>> > it is obvious that the 3rd cell should be
>> "1".
>> >
>> >
>>
>> No, it is not. Please read something about floating point
>> numbers.
>> Wikipedia may be a good start:
>> http://en.wikipedia.org/wiki/Floating_point
>>
>> cheers
>>
>> --
>> RNDr. Jaroslav Hajek
>> computing expert
>> Aeronautical Research and Test Institute (VZLU)
>> Prague, Czech Republic
>> url: www.highegg.matfyz.cz
>>
>
> If one needs infinite precision numbers, he/she should consider other
> programs from the CAS (Computer Algebra System) category - like Maxima, Axiom, GAP and many others.
>
> These programs support rational fractions, i.e. numbers in the
> nominator/denominator form.
>
> At the same time infinite precision arithmetic is typically slower.
>
> ...
>
> 'octave-forge' has 'symbolic' package based on GINAC - I think GINAC itself
> supports infinite precision numbers, but I'm not sure how complete
> 'octave-interface' to GINAC is.
>
> Regards,
>  Sergei.
>
>
>
>
> _______________________________________________
> Bug-octave mailing list
> Bug-octave at octave.org
> https://www-old.cae.wisc.edu/mailman/listinfo/bug-octave
>



-- 
/* andy buckle */