Velocity Reviews > floating point arithmetic

# floating point arithmetic

fred8865
Guest
Posts: n/a

 08-26-2008
Hi all,

I understand that due to different arithmetic used in floating points
they are just approximations. Hence, 180/100=1 in my python interpreter.
How can I tackle this problem of inaccurate floating point numbers?
thank you

regards
xtd

Mensanator
Guest
Posts: n/a

 08-26-2008
On Aug 26, 4:11 pm, fred8865 <(E-Mail Removed)> wrote:
> Hi all,
>
> I understand that due to different arithmetic used in floating points
> they are just approximations. Hence, 180/100=1 in my python interpreter.
> How can I tackle this problem of inaccurate floating point numbers?

Try actually using floating point numbers, not integers:

>>> 180.0/100

1.8

or

>>> float(180)/100

1.8

> thank you
>
> regards
> xtd

John Machin
Guest
Posts: n/a

 08-26-2008
On Aug 27, 7:11 am, fred8865 <(E-Mail Removed)> wrote:

> I understand that due to different arithmetic used in floating points
> they are just approximations. Hence, 180/100=1 in my python interpreter.

It's not "hence". What you are seeing is truncating integer division.

> How can I tackle this problem of inaccurate floating point numbers?

>>> 180 / 100

1
>>> 180 / 100.

1.8
>>> 180 / float(100)

1.8
>>> from __future__ import division
>>> 180 / 100

1.8

Rob Clewley
Guest
Posts: n/a

 08-26-2008
> I understand that due to different arithmetic used in floating points
> they are just approximations. Hence, 180/100=1 in my python interpreter.

No, that's not the reason you get 1, it's because the current version
of python does integer division by default. Try doing 180.0/100 or
including

from __future__ import division

> How can I tackle this problem of inaccurate floating point numbers?

There are few occasions for "regular" users to worry about the
inaccuracy of floating point, unless you are doing very technical
calculations at high precision. If you need decimals to be represented
"perfectly" in python (e.g. you are writing scripts for financial
applications), try importing the decimal package (look it up in the
python docs).

-Rob

fred8865
Guest
Posts: n/a

 08-26-2008
thanks guys

fred8865 wrote:

> Hi all,
>
> I understand that due to different arithmetic used in floating points
> they are just approximations. Hence, 180/100=1 in my python interpreter.
> How can I tackle this problem of inaccurate floating point numbers?
> thank you
>
> regards
> xtd

Terry Reedy
Guest
Posts: n/a

 08-27-2008

John Machin wrote:
> On Aug 27, 7:11 am, fred8865 <(E-Mail Removed)> wrote:
>
>> I understand that due to different arithmetic used in floating points
>> they are just approximations. Hence, 180/100=1 in my python interpreter.

>
> It's not "hence". What you are seeing is truncating integer division.
>
>> How can I tackle this problem of inaccurate floating point numbers?

>
>>>> 180 / 100

> 1
>>>> 180 / 100.

> 1.8
>>>> 180 / float(100)

> 1.8
>>>> from __future__ import division
>>>> 180 / 100

> 1.8
> --

Or start using 3.0, which fixed this

>>> 180/100

1.8

Gabriel Genellina
Guest
Posts: n/a

 08-27-2008
En Tue, 26 Aug 2008 18:11:30 -0300, fred8865 <(E-Mail Removed)> escribi�:

> I understand that due to different arithmetic used in floating points
> they are just approximations. Hence, 180/100=1 in my python interpreter.
> How can I tackle this problem of inaccurate floating point numbers?
> thank you

In the current Python versions (2.2 and up), 180/100 means integer
division (because both operands are integer).
If you want a floating point result, use 180.0/100 or
float(some_variable)/100

Starting with Python 3.0, the / operator will return a floating point
result ("true division"). So in that Python version, 180/100 gives 1.8
To enable that behavior on Python 2.x, execute "from __future__ import
division":

>>> 180/100

1
>>> from __future__ import division
>>> 180/100

1.8

In any Python version, 180//100 always means integer division:

>>> 180//100

1

--
Gabriel Genellina