SYNOPSIS: USER'S GUIDE NOW AVAILABLE ELECTRONICALLY FOR GENETIC
          ALGORITHM-BASED OPTIMIZATION SUBROUTINE PIKAIA

                                               Boulder, 20/01/96
Greetings PIKAIA Users;

   Well, it sure took longer than we were hoping, but the User's
   Guide for the genetic algorithm-based optimization subroutine
   PIKAIA (first public release, version 1.0) is finally ready.
   The User's Guide is officially known as NCAR Technical Note 418+IA.
   We also have available drivers, fitness functions and synthetic
   datasets for the examples discussed in Chapter 5 of the User's
   guide.

   All files are available on the HAO anonymous ftp site, where
   most of you already got the code from. For those who received
   the code directly from me, here is how you get in our
   anonymous ftp site; type

   ftp hao.ucar.edu 

   You will be prompted for username and password; use

   for username: anonymous

   for password: your e-mail address

   Welcome to Unix. You need to change to the pikaia directory; to
   do so type

   cd /pub/pikaia

   In that directory you will find the following files

   pikaia.f      <-- the source code, which you already have
   userguide.ps  <-- postscript file of the User's Guide
   examples      <-- directory containing codes for the examples
   examples.tar  <-- Tar file of the above directory; if you
		     do not operate on a Unix system don't
		     bother with this one.

   This is the first release of the User's Guide. Any comment,
   suggestions and/or criticism will be greatly appreciated.
   So would a detailed list of typos, with page numbers...

   There is one simple addition that should be made to the
   code that most of you got sometime in the last six months.
   Unless you are ABSOLUTELY convinced that neither you nor
   any friend/acquaintance that you may pass on the code to
   will NEVER manage to run PIKAIA with a zero value for the
   seed to the random number generator, we STRONGLY suggest
   that you add the following line to the subroutine RNINIT,
   as follows:

      subroutine rninit( seed )
      implicit none
      integer    seed
c=====================================================================
c     Initialize random number generator urand with given seed
c=====================================================================
      integer iseed
c     Common block to communicate with urand
      common /rnseed/ iseed
c     Set the seed value
      iseed = seed
      if(iseed.le.0) iseed=123456 <-- ****** ADD THIS LINE ******
      return
      end

   You may notice in the user's guide the lack of any statement
   to the effect that we decline any responsibility for whatever
   consequences resulting from the use or misuse of the routine,
   including but not limited to explosion and/or combustion of
   large pieces of machinery, loss of revenue from disabled computer
   system and/or network, divorce, layoffs, earthquakes, flood,
   mice in the basement, leaky roof, loud neighbours, plague and
   pestilence, and other acts of God/Allah/Vishnu/Buddha/Krishna,
   and so on and so forth till your ears hurt. We're all scientists
   here, not lawyers or politicians, so we'll take for granted that
   you assume responsibility for your actions.

   Having read this far you may start to be concerned about
   receiving a lot of trash e-mail from us: let us assure you
   that this will not happen. We have other things to do also.
   All you will hear from us are bug reports (hopefully not too
   many of those), and maybe, just maybe, upgrade news once or
   twice in the next 10 years.

   Of course if you absolutely do not want to ever hear anything
   from us under any circumstances, just send us e-mail to
   that effect and your address will be immediately removed
   from our users e-mail alias.

   Have Fun.

				     Paul Charbonneau
				     Barry Knapp
				     HAO/NCAR, Boulder.

==================================================================

   Keep reading ! we're not quite done yet...

   And now, our first BUG REPORT !! It's a rather minor one
   actually, some of you may have already encountered it and
   fixed it. It didn't cause problems with the Sun/Solaris
   and SGI/Irix Fortran compilers, but did under VAX/VMS and
   Microsoft/DOS. Our thanks to Adrian Webster, Royal Observatory
   Edinburgh, for promptly reporting it. Here it is:

-----------------------------------------------------------------
                  ***** BUG REPORT #1 *****

   SYMPTOM:    code failed to compile
   SUBROUTINE: report (called by pikaia)
   PROBLEM:    declaration error for ifit(np), caused by np
	       not yet declared as integer at that point

   CORRECTIVE ACTION: replace the line

      integer ifit(np),ivrb,ndim,n,np,nd,ig,nnew

   in the original code by

      integer np,ifit(np),ivrb,ndim,n,nd,ig,nnew

               ***** END OF BUG REPORT *****
------------------------------------------------------------------