Wrong output on linux, ok on Win32

[Melly]

Hi,
The following script works on Win32 (produces a crude mandelbrot), but fails on Linux (produces zeroes only).

I've put the output under the readmore, but here's the code... (oh, and I should add that this is the end result of some golf, hence no strict, etc.)

map{
  $a=1-$_/10;
  map{
    $d=$a;
    $e=$b=$_/20-2;
    map{
      ($d,$e)=(2*$d*$e+$a,$e**2-$d**2+$b);
      $c=$d**2+$e**2>4?_:0
    }1..99;
    print$c
  }0..59;
  print$/
}0..20;

Win32 Output:

________________________________________0___________________
____________________________________________________________
____________________________________0000____________________
____________________________________0000____________________
______________________________0__0000000000_________________
______________________________00000000000000000_____________
____________________________0000000000000000000_____________
___________________________000000000000000000000____________
_________________0000000__0000000000000000000000____________
________________000000000_0000000000000000000000____________
0000000000000000000000000000000000000000000000______________
________________000000000_0000000000000000000000____________
_________________0000000__0000000000000000000000____________
___________________________000000000000000000000____________
____________________________0000000000000000000_____________
______________________________00000000000000000_____________
______________________________0__0000000000_________________
____________________________________0000____________________
____________________________________0000____________________
____________________________________________________________
________________________________________0___________________

Linux:

000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000

[shmem]

Animate it ;-)

#!/usr/bin/perl
use Term::Cap;
$t= Tgetent Term::Cap;
$t=$t->Tgoto('cm', 0, 0);
for$i(1..shift){
  print $t;
  map{
    $a=1-$_/10;
    map{
      $d=$a;
      $e=$b=$_/20-2;
      map{
        ($d,$e)=(2*$d*$e+$a,$e**2-$d**2+$b);
        $c=$d**2+$e**2>4?_:0
      }1..$i;
      print$c
    }0..59;
    print$/
  }0..20;
}

Try perl mandel.pl 13. If I pass 99, there'll be a block of zeroes at the end.

<update>

It's not an overflow issue. As the area delimiting the mandelbrot converges to a line with increased iterations over the inner loop, there's not enough resolution to draw it. You can skip the outer area exiting the inner loop as soon as $d**2+$e**2>4

map{
  $a=1-$_/10;
  map{
    $d=$a;
    $e=$b=$_/20-2;
    for(1..99){
      ($d,$e)=(2*$d*$e+$a,$e**2-$d**2+$b);
      $c=$d**2+$e**2>4?_:0;
      last if $c eq _
    }
    print$c
  }0..59;
  print$/
}0..20;

I wonder how that works on your windows box, though.

</update>

--shmem

[liverpole]

Very slick, ++shmem.

It's times like this I wish that replies to nodes could be front-paged on their own!

[Melly]

LOL, nice, and demonstrates the problem neatly...

[Melly]

Indeed, but then I lose at golf ;)

Here's my solution that avoids a last and only gains me a couple of strokes...

map{
  $a=1-$_/10;
  map{
    $d=$a;
    $e=$b=$_/20-2;
    map{
      ($d,$e)=(2*$d*$e+$a,$e**2-$d**2+$b);
      $c=$d**2+$e**2>4?$d=8:_
    }1..50;
    print$c
  }0..59;
  print$/
}0..20

[almut]

...probably some differences in handling of NAN / INF, etc. The algorithm makes your $d and $e values overflow rather quickly.

By reducing the number of iterations from 99 to something like 9 you should get a coarse approximation of the desired shape -- on both platforms.

[gellyfish]

Printing out the values of $d and $e on the respective platforms might give you a clue. The rendering of the floating point over/underflow on windows gives rise to a value that can be coerced to a number on Windows (or at least the version of Perl I have.

/J\

[rinceWind]

Gives me a rectangle of zeros on Windows.

[Melly]

Odd - I'm running activeperl build 810 (5.8.4) on Win2k, what about you?

[rinceWind]

ActivePerl build 633 (5.6.1) on Win2k.