Thread Subject:

numbers output

Hi All,
I have a little code in MATLAB, whose success depends on MATLAB “correctly” reading certain numbers, that is the way I want them to be read. The issue can be described as follow. In my MATLAB window, I type
>> a=4123
a =
        4123
>> b=a*0.01
b =
  41.230000000000004
I want MATLAB to "get" b=41.23, not something else, and I can’t see why multiplying 4123 by 0.01 should mathematically give me 41.230000000000004 and not 41.23. Can you please help me to deal with this issue? I have 7.10.0 (R2010a) version. (I am not so much worried how the number is “displayed”, I need to make sure they are “stored/read” they way I need).
Thanks a lot.
Anna.

"Anna Kaladze" <anna.kaladze@gmail.com> wrote in message
news:i65h20$fdo$1@fred.mathworks.com...
> Hi All,
> I have a little code in MATLAB, whose success depends on MATLAB
> “correctly” reading certain numbers, that is the way I want
> them to be read. The issue can be described as follow. In my MATLAB
> window, I type
>>> a=4123
> a =
> 4123
>>> b=a*0.01
> b =
> 41.230000000000004
> I want MATLAB to "get" b=41.23, not something else, and I can’t see
> why multiplying 4123 by 0.01 should mathematically give me
> 41.230000000000004 and not 41.23.

See question 6.1 in the newsgroup FAQ. 0.01 is NOT exactly one
one-hundredth; it is the closest representable floating-point number to one
one-hundredth.

--
Steve Lord
slord@mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
To contact Technical Support use the Contact Us link on
http://www.mathworks.com

"Steven_Lord" <slord@mathworks.com> wrote in message <i65h96$109$1@fred.mathworks.com>...
>
>
> "Anna Kaladze" <anna.kaladze@gmail.com> wrote in message
> news:i65h20$fdo$1@fred.mathworks.com...
> > Hi All,
> > I have a little code in MATLAB, whose success depends on MATLAB
> > “correctly” reading certain numbers, that is the way I want
> > them to be read. The issue can be described as follow. In my MATLAB
> > window, I type
> >>> a=4123
> > a =
> > 4123
> >>> b=a*0.01
> > b =
> > 41.230000000000004
> > I want MATLAB to "get" b=41.23, not something else, and I can’t see
> > why multiplying 4123 by 0.01 should mathematically give me
> > 41.230000000000004 and not 41.23.
>
> See question 6.1 in the newsgroup FAQ. 0.01 is NOT exactly one
> one-hundredth; it is the closest representable floating-point number to one
> one-hundredth.
>
> --
> Steve Lord
> slord@mathworks.com
> comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
> To contact Technical Support use the Contact Us link on
> http://www.mathworks.com
Thank you for your answer Steve.
Does this mean that I never will be able to "get" 41.23 in the above example? Is there a simple way to deal with my problem?

"Anna Kaladze" <anna.kaladze@gmail.com> wrote in message <i65l7l$17$1@fred.mathworks.com>...

> Does this mean that I never will be able to "get" 41.23 in the above example? =======

No. Due to floating point approximation, the number 41.23 isn't even representable as the following shows

>> format long, 41.23

ans =

  41.229999999999997


>Is there a simple way to deal with my problem?
==================

Possibly. First, describe why it's a problem, i.e., why you can't accept a floating point approximation of this number.

"Matt J " <mattjacREMOVE@THISieee.spam> wrote in message
news:i65lqn$be7$1@fred.mathworks.com...
> "Anna Kaladze" <anna.kaladze@gmail.com> wrote in message
> <i65l7l$17$1@fred.mathworks.com>...
>
>> Does this mean that I never will be able to "get" 41.23 in the above
>> example? =======
>
> No. Due to floating point approximation, the number 41.23 isn't even
> representable as the following shows
>
>>> format long, 41.23
>
> ans =
>
> 41.229999999999997

Matt's correct; 41.23 is not exactly representable. The way 41.23 is
represented is smaller than forty-one and twenty-three one-hundredths; the
next largest representable number after that is larger than forty-one and
twenty-three one-hundredths.

format long
x = 41.23
next = x+eps(x)

and if you print out those numbers' IEEE representations in hex, you can see
that only the last bit of the mantissa changed, so there are no
representable numbers in-between.

format hex
x
next

Don't forget to change the display format back to your favorite after doing
this, or you may confuse yourself the next time you display something.

>>Is there a simple way to deal with my problem?
> ==================
>
> Possibly. First, describe why it's a problem, i.e., why you can't accept a
> floating point approximation of this number.

--
Steve Lord
slord@mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
To contact Technical Support use the Contact Us link on
http://www.mathworks.com

"Anna Kaladze" <anna.kaladze@gmail.com> wrote in message <i65h20$fdo$1@fred.mathworks.com>...
> Hi All,
> I have a little code in MATLAB, whose success depends on MATLAB “correctly” reading certain numbers, that is the way I want them to be read. The issue can be described as follow. In my MATLAB window, I type
> >> a=4123
> a =
> 4123
> >> b=a*0.01
> b =
> 41.230000000000004
> I want MATLAB to "get" b=41.23, not something else, and I can’t see why multiplying 4123 by 0.01 should mathematically give me 41.230000000000004 and not 41.23. Can you please help me to deal with this issue? I have 7.10.0 (R2010a) version. (I am not so much worried how the number is “displayed”, I need to make sure they are “stored/read” they way I need).

You can find a useful tool num2strexact to help visualize this on the FEX:

>> num2strexact(4123)
ans =
4.123e3
>> num2strexact(0.01)
ans =
1.000000000000000020816681711721685132943093776702880859375e-2
>> num2strexact(4123*0.01)
ans =
4.12300000000000039790393202565610408782958984375e1
>> num2strexact(4123*0.01-eps(4123*0.01))
ans =
4.122999999999999687361196265555918216705322265625e1

As you can see, there are two floating point numbers close to 41.23, but the exact value of 41.23 cannot be represented in IEEE double precision floating point format.

num2strexact can be found here:

http://www.mathworks.com/matlabcentral/fileexchange/22239-num2strexact-exact-version-of-num2str


James Tursa

Hi,
I can't find how display numbers without power e. For example:

0.003923^2

I would like to print:
0.000015389929
instead of:
1.5389929e-05

Thanks in advance

Grzegorz

"Grzegorz Knor" <gknor@o2.pl> wrote in message <i9p2rb$ilk$1@fred.mathworks.com>...
> Hi,
> I can't find how display numbers without power e. For example:
>
> 0.003923^2
>
> I would like to print:
> 0.000015389929
> instead of:
> 1.5389929e-05
>
> Thanks in advance
>
> Grzegorz

Ok, I found that i can do this in this way:
fprintf('%.10f',n)

But is there any possibilty, to set it as a default Matlab format?

"Grzegorz Knor" <gknor@o2.pl> wrote in message
news:i9p3e3$pd8$1@fred.mathworks.com...
> "Grzegorz Knor" <gknor@o2.pl> wrote in message
> <i9p2rb$ilk$1@fred.mathworks.com>...
>> Hi, I can't find how display numbers without power e. For example:
>>
>> 0.003923^2
>>
>> I would like to print:
>> 0.000015389929
>> instead of:
>> 1.5389929e-05
>>
>> Thanks in advance
>>
>> Grzegorz
>
> Ok, I found that i can do this in this way:
> fprintf('%.10f',n)
>
> But is there any possibilty, to set it as a default Matlab format?
>

help format ?? e.g., format short or format long - set it in your
preferences if you want to always use it.

--
Loren
http://blogs.mathworks.com/loren/
http://matlabwiki.mathworks.com/MATLAB_FAQ