[comp.theory.cell-automata] wireworld on CAM and Cellsim

hiebeler@TURING.CS.RPI.EDU (Dave Hiebeler) (03/02/90)

  Here's Wireworld for CAM and for Cellsim.  I only spent a few
minutes writing them, so I won't say they're great, but they seem to
work.  Actually, I'd like to issue a challenge for CAM-users: if you
have a better (which I guess means "more interesting and/or compact")
way of implementing this, I'd like to hear about it!

  The Cellsim version of the rule is easier to read, but the nice
thing about CAM is you can run 60 generations/sec on a 256x256 array,
so you can do large circuits pretty quickly if you want.  Cellsim will
go pretty fast for smaller arrays.  You can't run this rule on the
Connection Machine using Cellsim, because the lookup-table would be
too large for a CM with 8Kbytes/processor (that's what I use now).  Of
course, if you were to re-write it in Paris, it would run at several
hundred generations/sec, if you turn off the display!  I'm not really
motivated to write it in Paris at the moment though..  (Paris is the
Parallel Instruction Set, a semi-low-level language on the CM that you
can call from C, Lisp and Fortran, for those who don't know.)

=====
CAM version of Wireworld
 Wires are on plane 1, electron heads on plane 0, and I store electron
tails on CAM-B.  To non-CAM users, this might look a little scary.
It's written in Forth, a stack-based language.  If you've used an HP
calculator, Forth is nothing fundamentally new.  Plus, you only have
to know a little Forth to use CAM.  So here's the rule.  I hope I
didn't make a typo; I am not able to transfer files from my PC to Sun
right now, so I just scribbled it down on paper from the PC screen and
typed it in here.

\ Wireworld, simple logic-circuits as described in
\ A.K. Dewdney's "Computer Recreations" column in the
\ January 1990 Scientific American.

NEW-EXPERIMENT N/MOORE &/CENTERS
CAM-A
: 8SUM  NORTH SOUTH EAST WEST N.WEST N.EAST S.WEST S.EAST
    + + + + + + + ;
: EH 1 ;  \ Electron Head
: W 2 ;   \ Wire
: WIRERULE
    8SUM { W EH EH W W W W W W } ;
: WIREWORLD  \ If not E-Tail, do normal rule, else become wire
    &CENTER IF W ELSE CENTERS { 0 W WIRERULE 0 } THEN >PLNA ;
MAKE-TABLE WIREWORLD

CAM-B N/MOORE &/CENTERS
: ET-RULE
    &CENTERS EH = IF 1 ELSE 0 THEN >PLN2 ;
MAKE-TABLE ET-RULE


=====
Cellsim version of Wireworld

  Here it is for Cellsim.  It uses the "m4" neighborhood, so it
generates a lookup-table rather than running in computed-function
mode.  This form of the rule is fairly portable, and should be very
easy to plug into CA Lab and other packages, I think.  Of course, this
rule is so simple, it's no big feat to rewrite it for other systems.

/*
 * File: wireworld.m4.c
 *   By: Dave Hiebeler
 *       hiebeler@heretic.lanl.gov
 *       February 1990
 *
 * Wireworld CA rule, as described in A.K. Dewdney's Jan 1990 "Computer
 * Recreations" column, in Scientific American.
 */


#include "nborhood.h"

byte wireworld_rule();

#define EMPTY 0
#define EHEAD 1  /* electron heads on plane 0 */
#define WIRE  2  /* wires on plane 2 */
#define ETAIL 3  /* electron tail is a 1 in both planes */

/* Little boolean macro, equals 1 if cell is e-head, otherwise equals 0 */
#define ehead(cell) ((cell==EHEAD) & 0x01)

void
init_function()
{
    update_function = wireworld_rule;
}


byte
wireworld_rule(nbors)
moore_nbors *nbors;
{
    int eightsum;

    Get_moore_nbors;

    eightsum = ehead(tl) + ehead(l) + ehead(bl) + ehead(t) + ehead(b) +
	ehead(tr) + ehead(r) + ehead(br);
    switch (center) {
      case EMPTY:
	return EMPTY;
	break;
      case EHEAD:
	return ETAIL;
	break;
      case ETAIL:
	return WIRE;
	break;
      case WIRE:
	if ((eightsum == 1) || (eightsum == 2))
	    return EHEAD;
	else
	    return WIRE;
	break;
    }
}


  If someone out there using CAM or Cellsim would like me to crank out
a little program to convert those text-diagrams we've been seeing into
CAM or Cellsim image-format, let me know.  It's easy enough to draw
stuff in both systems, that I haven't bothered writing a program to
convert.  It would only take a couple of minutes to do, though.

Dave Hiebeler                       hiebeler@turing.cs.rpi.edu
Computer Science Dept               hiebeler@heretic.lanl.gov
Amos Eaton Bldg.                    hiebeler@rpitsmts.bitnet
Rensselaer Polytechnic Institute
Troy, NY 12180-3590