Rounding error in program

[Dol]

Hi all,

When I execute the following program, $d is set to 18 instead of 19 as
I would expect. Any ideas why this happens?

Thanks,
Dol

#!/yourpathtoperl/perl

$a = 50;
$b = 0.37;

$c = $a*$b
$d = sprintf("%.0f",$c);

print "$c\t$d\n";

 

[Sherm Pendley]

When I execute the following program, $d is set to 18 instead of 19 as
I would expect. Any ideas why this happens?

Have a look in 'perldoc -q trig' for a discussion of rounding errors and
why this happens.

sherm--

 

[Helgi Briem]

When I execute the following program, $d is set to 18 instead of 19 as
I would expect.

Why would you expect that? It is the wrong expectation.

Any ideas why this happens?

Yes, because 18.5 should be rounded to 18, not 19.

Always rounding X.5 *up* as you probably learned in grade school
introduces a systematic bias. This was changed by the International
Standards Organisation in 1992, to rounding to the nearest even
integer.

#!/yourpathtoperl/perl

$a = 50;
$b = 0.37;

$c = $a*$b
$d = sprintf("%.0f",$c);

print "$c\t$d\n";

for (1..30)
{
$c = $_+0.5;
$d = sprintf("%.0f",$c);

print "$c\t$d\n";
}

--
Helgi Briem hbriem AT simnet DOT is

Never worry about anything that you see on the news.
To get on the news it must be sufficiently rare
that your chances of being involved are negligible!

 

[Mark Clements]

When I execute the following program, $d is set to 18 instead of 19 as
I would expect. Any ideas why this happens?
$a = 50;
$b = 0.37;

$c = $a*$b
$d = sprintf("%.0f",$c);

print "$c\t$d\n";

from

perldoc -f sprintf

You can specify a precision (for numeric
conversions) or a maximum width (for string
conversions) by specifying a "." followed by a
number. For floating point formats, this
specifies the number of decimal places to show
(the default being 6), eg:

so you've specified zero decimal places and got zero decimal places.
*but* (breaking the problem down further)

redwood 23780 $ perl -le '$a=shift;$f=shift;printf "%.${f}f\n",$a' .5 0
0
redwood 23781 $ perl -le '$a=shift;$f=shift;printf "%.${f}f\n",$a' .6 0
1

which, while being intuitive, seems to contradict the documentation. I'm
obviously missing something.

Mark


Mark

 

[Chris Mattern]

Hi all,

When I execute the following program, $d is set to 18 instead of 19 as
I would expect. Any ideas why this happens?

Your error was in expecting to get rounding. Using sprintf like that
doesn't result in the your number being rounded; the fractional part
is simply truncated. If you want rounding, add 0.5 to the number
first before sending it to sprintf.

Thanks,
Dol

#!/yourpathtoperl/perl

$a = 50;
$b = 0.37;

$c = $a*$b
$d = sprintf("%.0f",$c);

print "$c\t$d\n";

--
Christopher Mattern

"Which one you figure tracked us?"
"The ugly one, sir."
"...Could you be more specific?"

 

[Shawn Corey]

for (1..30)
{
$c = $_+0.5;
$d = sprintf("%.0f",$c);

print "$c\t$d\n";
}

My output is:
1.5 2
2.5 2
3.5 4
4.5 4
5.5 6
6.5 6
7.5 8
8.5 8
9.5 10
10.5 10
....

Why?

--- Shawn
--
perl -v

This is perl, v5.8.0 built for darwin

Copyright 1987-2002, Larry Wall

Perl may be copied only under the terms of either the Artistic License
or the
GNU General Public License, which may be found in the Perl 5 source kit.

Complete documentation for Perl, including FAQ lists, should be found on
this system using `man perl' or `perldoc perl'. If you have access to the
Internet, point your browser at http://www.perl.com/, the Perl Home Page.

 

[Paul Lalli]

Your error was in expecting to get rounding. Using sprintf like that
doesn't result in the your number being rounded; the fractional part
is simply truncated. If you want rounding, add 0.5 to the number
first before sending it to sprintf.

Please check your assumption before posting a reply to a question like
this. You are regrettably wholly incorrect.

printf "%.0f\n", 3.8;

outputs "4".

Paul Lalli

 

[Helgi Briem]

My output is:
1.5 2
2.5 2
3.5 4
4.5 4
5.5 6
6.5 6
7.5 8
8.5 8
9.5 10
10.5 10
...

Why?

????

I just told you why. Didn't you read my post?

perldoc -q trig

a.k.a. Does Perl have a round() function? What about ceil() and
floor()? Trig functions?

--
Helgi Briem hbriem AT simnet DOT is

Never worry about anything that you see on the news.
To get on the news it must be sufficiently rare
that your chances of being involved are negligible!

 

[A. Sinan Unur]

My output is:
1.5 2
2.5 2
3.5 4
4.5 4
5.5 6
6.5 6
7.5 8
8.5 8
9.5 10
10.5 10
...

Why?

--- Shawn

You snipped the relevant part of Helgi's post. Reading what you are
responding to helps:


Sinan.

 

[Helgi Briem]

Have a look in 'perldoc -q trig' for a discussion of rounding errors and
why this happens.

It's not a rounding "error". It is correct rounding.

--
Helgi Briem hbriem AT simnet DOT is

Never worry about anything that you see on the news.
To get on the news it must be sufficiently rare
that your chances of being involved are negligible!

 

[Jeff 'japhy' Pinyan]

You snipped the relevant part of Helgi's post. Reading what you are
responding to helps:

Does anyone know the ISO's reasoning behind it? It seems to create a new
bias, which doesn't seem to make sense to me, at least.

--
Jeff "japhy" Pinyan % How can we ever be the sold short or
RPI Acacia Brother #734 % the cheated, we who for every service
Senior Dean, Fall 2004 % have long ago been overpaid?
RPI Corporation Secretary %
http://japhy.perlmonk.org/ % -- Meister Eckhart

 

[A. Sinan Unur]

Does anyone know the ISO's reasoning behind it? It seems to create a new
bias, which doesn't seem to make sense to me, at least.

It is in

ISO 1995 Guide to the Expression of Uncertainty in Measurement 1st edn
(Geneva: ISO)

but the following ought to give you a good enough idea:

http://www.crystalkeen.com/articles/crystalreports/rounding.htm

Hope that helps.

Sinan.

 

[ctcgag]

Does anyone know the ISO's reasoning behind it?

So that you round up just about as often as you round down, so they
cancel each other out--rather than rounding up more often than down.

It seems to create a new
bias,

Normally, no one cares what the even/odd distribution of the least
siginificant digit of a column of numbers are, so the bias it introduces is
harmless.

which doesn't seem to make sense to me, at least.

None of it makes sense to me. If you are going to be summing a column
of numbers, don't round them first.

Xho