Re: [Bug-apl] Problem with modulo arithmetic on Gaussian integers

Juergen Sauermann Wed, 26 Apr 2017 09:31:26 -0700

Hi Jay et al,

thanks for the analysis. I hope this problem is fixed in SVN 934.

For those of you that can reproduce the problem on their machine, please let me know
if the problem has not disappeared.

/// Jürgen

On 04/26/2017 10:19 AM, Jay Foad wrote:

I see the wrong result on my machine, just like Fred. Jürgen, the problem seems to be in this code:

// if ⎕CT != 0 and B ÷ A is close to an integer within ⎕CT then return 0.

//

// Note: In that case, the integer to which A ÷ B is close is either

// floor(A ÷ B) or ceil(A ÷ B).

//

const APL_Float qct = Workspace::get_CT();

if (qct != 0)

{

const APL_Complex quot = cval() / a;

const APL_Float qfr = floor(quot.real());

const APL_Float qfi = floor(quot.imag());

const APL_Float qcr = ceil(quot.real());

const APL_Float qci = ceil(quot.imag());

if (quot.real() > (qcr - qct) && quot.imag() > (qci - qct))

return IntCell::z0(Z); // quot is close to its ceiling

if (quot.real() < (qfr - qct) && quot.imag() < (qfi - qct))

return IntCell::z0(Z); // quot is close to its floor

}

In this case quot is:

(gdb) p quot

$14 = {_M_value = 6.9999999999999991 + -1.9999999999999998 * I}

... which is close to the Gaussian integer 7J¯2. But you only check for it being close to 7J¯1 and 6J¯2, both of which tests fail.

As for why Fred and I see this problem and you don't: I guess on your machine the _expression_ cval() / a probably returns an exact (Gaussian) integer result? If so, I guess it's down to some small change in the compiler version, or glibc, or whatever provides the implementation of std::complex<double>::operator/.

Jay.

On 26 April 2017 at 06:44, Frederick Pitts <fred.pi...@comcast.net> wrote:

To all,

I have 3 machines running 64-bit Fedora 25 Workstation with g++
(GCC) 6.3.1 20161221 (Red Hat 6.3.1-1) and either gnu-apl svn version
889 or 933. Two of the machines are about 8 years old and one less
than a year old.

On all three platforms, gnu-apl gives:

3J1 | 23J1 25J25
3J1 0

Juergen and Xtian (on svn 933) report their setups give the
right answer:

0 0

Am I the only one seeing this problem? BTW, I have about 1500
more examples of the modulo operator failing out of 6765201 tests with
distinct argument values.

Regards,

Fred

On Tue, 2017-04-25 at 23:01 -0400, Christian Robert wrote:
> Same result as Juergen,
>
> Xtian.
>
> [xtian@centos-7:/home/xtian] $ apl
>
>                      ______ _   __ __  __    ___     ____   __
>                     / ____// | / // / / /   /   |   / __ \ / /
>                    / / __ /  |/ // / / /   / /| |  / /_/ // /
>                   / /_/ // /|  // /_/ /   / ___ | / ____// /___
>                   \____//_/ |_/ \____/   /_/  |_|/_/    /_____/
>
>                       Welcome to GNU APL version 1.7 / 933M
>
>                  Copyright (C) 2008-2016  Dr. Jürgen Sauermann
>                         Banner by FIGlet: www.figlet.org
>
>                  This program comes with ABSOLUTELY NO WARRANTY;
>                            for details run: apl --gpl.
>
>       This program is free software, and you are welcome to
> redistribute it
>           according to the GNU Public License (GPL) version 3 or
> later.
>
> SAVED 2017-03-30 22:33:13 (GMT-4)
>        23J1 25J25 ÷ 3J1
> 7J¯2 10J5
>        3J1 | 23J1 25J25
> 0 0
>
>
>

> On 2017-04-25 21:50, Frederick Pitts wrote:
> > Juergen,
> >
> >          I did a 'make clean' followed by 'make' and 'make
> > install'.  I
> > obtained the same result that caused me to report the problem.
> >
> >          The version of gnu-apl I'm using is svn rev 933.  From the
> > banner in your email, I see you're testing with code from your
> > personal
> > svn.  Is it possible the changes you recently made to
> > (ComplexCell.hh
> > and FloatCell.hh) are not yet in the svn from which I clone?
> >
> > Regards,
> >
> > Fred
> >
> > On Tue, 2017-04-25 at 22:05 +0200, Juergen Sauermann wrote:
> > > Hi Fred,
> > >
> > > actually it does on my machine:
> > >
> > >                       ______ _   __ __  __    ___     ____   __
> > >                      / ____// | / // / / /   /   |   / __ \ / /
> > >                     / / __ /  |/ // / / /   / /| |  / /_/ // /
> > >                    / /_/ // /|  // /_/ /   / ___ | / ____// /___
> > >                    \____//_/ |_/ \____/   /_/  |_|/_/    /_____/
> > >
> > >                    Welcome to GNU APL version 1.7 / 12784:12785M
> > >
> > >                   Copyright (C) 2008-2016  Dr. Jürgen Sauermann
> > >                          Banner by FIGlet: www.figlet.org
> > >
> > >                   This program comes with ABSOLUTELY NO WARRANTY;
> > >                             for details run: apl --gpl.
> > >
> > >        This program is free software, and you are welcome to
> > > redistribute it
> > >            according to the GNU Public License (GPL) version 3 or
> > > later.
> > >
> > >         23J1 25J25 ÷ 3J1
> > > 7J¯2 10J5
> > >
> > >         3J1 | 23J1 25J25
> > > 0 0
> > >
> > >
> > > However, if I remember correctly then some of the changes that I
> > > made
> > > lately were in
> > > header files (ComplexCell.hh and FloatCell.hh). If you did
> > > ./configure without options, then
> > > your build is probably is a 'fast' one, which may not have
> > > detected
> > > header file changes.
> > >
> > > Please try make clean at the top level and rebuild GNU APL to see
> > > if
> > > the problem persists.
> > >
> > > Best Regards,
> > > Jürgen Sauermann
> > >
> > >
> > >
> > > On 04/25/2017 09:30 PM, Frederick Pitts wrote:
> > > > Jeurgen,
> > > >
> > > >      A greatest common divisor of 23J1 and 25J25 is 3J1.
> > > > Complex division by of 23J1 and 25J25 by 3J1 yields Gaussian
> > > > integers
> > > >
> > > >         23J1 25J25 ÷ 3J1
> > > > 7J¯2 10J5
> > > >
> > > > but mod 3J1 of the same numbers does not consistently yield
> > > > zeroes
> > > >
> > > >         3J1 | 23J1 25J25
> > > > 3J1 0
> > > >
> > > > I can provide numerous other examples of this problem if
> > > > needed.
> > > >
> > > > Regards,
> > > >
> > > > Fred
> > > >
> > > >
> > > >
> > > >
> > >
> > >
>
>

Re: [Bug-apl] Problem with modulo arithmetic on Gaussian integers

Reply via email to