comp.lang.idl-pvwave archive: archive

Home » Public Forums » archive » Re: A bug in MOD ?

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Re: A bug in MOD ? [message #41192 is a reply to message #41182]

Fri, 24 September 2004 10:06

Chris Lee
Messages: 101
Registered: August 2003

Senior Member

In article <cj1cqe$gut$1@canarie.caspur.it>, "sandrokan"
<mura@remove.ifsi.rm.cnr.it> wrote:

> I don't know much about libs, I only have IDl and another s/w: IDL>
> print, 1.0 mod 0.1
> 0.100000
> but:
>
>>> mod(1.0, 0.1)
> ans =
> 0
>>>

Hi,

Ah, matlab, wonderful matlab.I think Matlab uses arbitrary precision
math. where this answer is correct. I could be wrong of course.

The answer lies in the floating point representation of 1.0 and 0.1, or
any number. One of the numbers are really what they appear (not sure
which one) and the result is that ..

floor(1.0/0.1)=9
1.0 mod 0.1 = 0.1

;these may not work in any known language, but they do show what's
happening.

Calculating 0.8 mod 0.1, you get the correct answer, because
whatever representation error exists in 0.8 also exists in 0.1 .similarly
for 1.0 and 0.5|0.25|0.125 (powers of 2).

This is true of the IDL mod, the C++ fmod call (and probably the C
library fmodf call, as its used internallyin C++), the fortran mod function, the python
mod function, etc.

I'm not sure what the correct method would be. I can't really round a
value to zero when the value is comparable to the denominator in the
'mod' equation. It gets worse when I realize I've used 'mod' on a
floating point before, in FORTRAN code.

Chris.

Report message to a moderator

[Message index]

		Re: A bug in MOD ? By: Paolo Grigis on Mon, 27 September 2004 00:37
		Re: A bug in MOD ? By: R.Bauer on Sun, 26 September 2004 01:52
		Re: A bug in MOD ? By: Chris Lee on Fri, 24 September 2004 10:06
		Re: A bug in MOD ? By: sandrokan on Fri, 24 September 2004 08:00
		Re: A bug in MOD ? By: Chris Lee on Fri, 24 September 2004 03:49

Previous Topic:	Munich Adventures and other Coyote Stories
Next Topic:	Integration of 2-D irregularly gridded data

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Sat Oct 11 08:12:23 PDT 2025

Total time taken to generate the page: 0.48436 seconds