
Genetic Algorithms Digest   Monday, November 25 1991   Volume 5 : Issue 36

 - Send submissions to GA-List@AIC.NRL.NAVY.MIL
 - Send administrative requests to GA-List-Request@AIC.NRL.NAVY.MIL

Today's Topics:
	- Administrivia
	- RE: Long Genomes - an alternate explanation
	- Negative rewards and optimal ML length in LCS
	- PASE Workshop Zurich, december 9-10
	- New TCGA reports
	- GAs and very fast simulated re-annealing: A comparison

**********************************************************************

CALENDAR OF GA-RELATED ACTIVITIES: (with GA-List issue reference)

 First European Conference on Artificial Life (v5n10)         Dec 11-13, 1991
 Canadian AI Conference, Vancouver, (CFP 1/7)                 May 11-15, 1992
 COGANN, Combinations of GAs and NNs, @ IJCNN-92 (v5n31)      Jun 6,     1992
 10th National Conference on AI, San Jose, (CFP 1/15)         Jul 12-17, 1992
 FOGA-92, Foundations of Genetic Algorithms, Colorado (v5n32) Jul 26-29, 1992
 ECAI 92, 10th European Conference on AI (v5n13)              Aug  3-7,  1992
 Parallel Problem Solving from Nature, Brussels, (v5n29)      Sep 28-30, 1992

 (Send announcements of other activities to GA-List@aic.nrl.navy.mil)

**********************************************************************
----------------------------------------------------------------------

From: Alan C. Schultz (GA-List Moderator)
Date: Mon, 25 Nov 91
Subject: Administrivia

   About 100 people on the mailing list did not receive issues v5n31 - v5n35
   due to a problem with the mailing list.  The problem has been found and
   corrected, and the issues are being sent to those who missed it.  This
   issue might proceed those missed issues.

   Also, for those who are interested, we now have over 1000 subscribers to
   GA-List!  The newest country to get delivery is Yugoslavia.

   Alan C. Schultz

------------------------------

From: jrv@sdimax2.mitre.org
Date: Fri, 15 Nov 91 07:50:01 EST
Subject: RE: Long Genomes - an alternate explanation

   A few issues back I mentioned:
   > There was a very interesting story in Analog a few months back...

   Charles Palmer (cpalmer@watson.ibm.com) later wrote:
   > I would be VERY interested in reading this story you refer to.
   > Could you provide me with a little more info about the issue
   > of Analog in which it appeared?

   It took a while to find (it was longer ago than I thought, and other
   stories kept distracting me) but I finally did.  In case anyone else is
   interested, it was:

	   Charles Sheffield, "The Double Spiral Staircase", Analog, Jan 1990.

			       - Jim Van Zandt (jrv@mbunix.mitre.org)

------------------------------

From: dorigo@ipmel2.elet.polimi.it (Marco Dorigo)
Date: Fri, 22 Nov 1991 18:54:28 -0100
Subject: Negative rewards and optimal ML length in LCS

   I have a couple of general question for people in the LCS community:
   1) I have always seen people using reward functions where reward is a
   positive value. 
   Is there anyone that used punishments (negative rewards)?
   I suppose with negative rewards the bucket brigade is not working anymore
   (or at least it works
   in a different way). Nevertheless punishment is an important aspect of most
   (real) learning 
   strategies (for example with animals).
   I have preliminary (experimental) results that show that, at least for very
   simple problems, punishments can make learning faster. I would like to
   share my experience with someone else that 
   has been working on the same problem.
   2) It seems to me that nobody cares about message list optimal length.
   Again I have preliminary 
   (experimental) results that show that there is a lower limit for the ML
   length, and that this limit
   is a function of the number of rules. 
   Any comments?

   Marco Dorigo 
   Dipartimento di Elettronica
   Politecnico di Milano
   Via Ponzio 34/5
   20133 Milano
   Italia
   Tel.  +39-2-2399-3622
   Fax. +39-2-2399-3411
   e-mail:  dorigo@ipmel2.elet.polimi.it

------------------------------

From: Diethelm Wuertz <wuertz@ips.id.ethz.ch>
Date: Mon, 18 Nov 91 12:11:18 +0000
Subject: PASE Workshop Zurich, december 9-10

   *****************************************************
   ***     PASE - Workshop Zurich,  December 9-10    ***
   *****************************************************

   Due  to the letter  spread by Prof.  Schwefel in this
   digest, concering the PASE Workshop, I like to inform
   the readers of the genetic  algorithms digest on this
   event in Zurich.

   *** PASE ***  "Parallel Problem Solving from Nature -
   Applications in Statistics and  Economics" is a small
   international  workshop   (about 70 participants)  in 
   the field of statistics and  economics with interests 
   in new algorithms motivated by parallel concepts from
   nature.  The keywords  PASE, Statistics and Economics
   and also the list of members of the program committee
   should make clear  that PASE is different from PPSN I
   and PPSN II. (Nevertheless, if there was created some 
   confusion from PASE in the PPSN community I apologize
   it.) The announcement of the workshop was distributed
   to a limited circle, mainly to people from statistics
   and economics. Since I got the feeling that there may
   be interest in this digest on our  PASE workshop too,
   you can request more information via e-mail from:
   <wuertz@ips.id.ethz.ch>.

   Diethelm Wuertz 
   IPS - ETH Zentrum CLU B3, CH-8092 Zuerich
   Tel. 0041.1.256.5567

------------------------------

From: <@risc.ua.edu,@ua1ix.ua.edu:rob@galab2.mh.ua.edu>
Subject: New TCGA reports.
Date: Thu, 14 Nov 91 08:33:19 CST

   The Clearinghouse for Genetic Algorithms at The University of Alabama
   is pleased to announce the availability of the following two technical
   reports. These reports can be obtained by contacting TCGA by phone,
   email, or USmail:

	      The Clearinghouse for Genetic Algorithms (TCGA)
		   Department of Engineering of Mechanics
			 The University of Alabama
			      P. O. Box 870278
			 Tuscaloosa, Alabama 35487
		<<email>> @ua1ix.ua.edu:rob@galab2.mh.ua.edu
			  <<phone>> (205) 348-1618
			   <<fax>> (205) 348-8573

   The two new reports are Master's theses.
   TCGA requests $9.00 ($12.00 overseas surface mail) to defray the costs 
   of copying, binding, and shipping these large documents.

   
	 Strength-to-Weight and Stiffness-to-Weight Optimization of
		    Laminates Using a Genetic Algorithm
			     (Master's Thesis)

				     by

			     Kelvin J. Callahan
			 The University of Alabama

			   TCGA Report No. 91006
			     November 11, 1991

				  ABSTRACT

   The methodology for the minimum weight design of fiber-
   reinforced, laminated, composite structures using genetic
   algorithms (GAs) is investigated.  Specifically, the
   strength-to-weight and stiffness-to-weight ratios of
   symmetric and unsymmetric laminates are optimized under in-
   plane and flexural loading.  The laminate stresses and
   strains are determined using classical laminate theory.
   Laminate failure is assessed using the quadratic failure and
   the first-ply-failure criteria.  The stacking sequence, ply
   orientation angles, and number of plies are varied with
   strain or strength constraints and multiple loads adding to
   the dimensionality of the analysis.  Several test cases
   using T300/5208 graphite-epoxy laminates are compared to
   known optimal solutions as a verification of the GA results.
   Finally, a micro-GA (l-GA) is used in an attempt to minimize
   fitness function evaluations and the results are compared
   with the GA.  The results of this study indicate that the GA
   can be a superior method for optimizing strength-to-weight
   and stiffness-to-weight ratios of laminated composite
   structures.

   
	Flow Vectoring of Supersonic Exhaust Nozzles Using a Genetic
	       Algorithm to Define Optimally-Shaped Contours
			     (Master's Thesis)

				     by

			  Everett Gordon King, Jr.
			 The University of Alabama


			   TCGA Report No. 91007
				May 15, 1991

				  ABSTRACT

   This research presents a novel approach to exit plane flow
   vectoring which incorporates the use of varying nozzle
   contours to affect changes in net flow angles from that of a
   baseline nozzle design.  A genetic algorithm is used to
   optimize the nozzle contour in order to achieve the best
   overall flow turning capability.  The design of a minimum
   length nozzle is presented in order to gain an understanding
   of the implementation of a genetic algorithm to an
   engineering search and optimization problem.  For the flow
   vectoring nozzle analysis, various baseline nozzle contours
   are used to investigate the genetic algorithm's ability to
   optimize nozzle contours given varying exit plane flow angle
   distributions.  Placement and displacement limitations of
   the nozzle contour variations are discussed.  Two different
   objective function forms are employed as search criteria,
   one form which consists of the force angle and another which
   has a second term to gage the influence of the nozzle thrust
   coefficient on the flow turning capability of a nozzle.
   Analysis results indicate that the flow turning capabilities
   of nozzles with optimal contours are good and, that for the
   case of a divergent baseline design, vectoring of the exit
   plane flow increases the axial thrust coefficient.

------------------------------

From: ingber@umiacs.UMD.EDU (Lester Ingber)
Date: Thu, 14 Nov 1991 14:25:06 EST
Subject: GAs and very fast simulated re-annealing: A comparison

   Bruce Rosen and I have written the following paper.
   PostScript-compressed-uuencoded email copies may be
   obtained by contacting me.

   Genetic algorithms and very fast simulated re-annealing: A comparison

			     Lester Ingber
     Science Transfer Corporation, P.O. Box 857, McLean, VA 22101
			 ingber@umiacs.umd.edu

				  and

			      Bruce Rosen
   Department of Computer & Information Sciences, University of Delaware,
			   Newark, DE 19716
			  brosen@cis.udel.edu


	We compare Genetic Algorithms (GA) with a functional  search
   method,  Very Fast Simulated Re-Annealing (VFSR) that not only is
   efficient in its  search  strategy,  but  also  is  statistically
   guaranteed  to  find the function optima.  GA previously has been
   demonstrated to be competitive with other standard Boltzmann-type
   simulated  annealing techniques.  Presenting a suite of six stan-
   dard test functions to GA and VFSR codes from  previous  studies,
   without  any  additional fine tuning, strongly suggests that VFSR
   can be expected to be orders of magnitude more efficient than GA.

   Prof. Lester Ingber            
   Science Transfer Corporation             
   P.O. Box 857                703-759-2769 
   McLean, VA 22101   ingber@umiacs.umd.edu 

------------------------------
End of Genetic Algorithms Digest
******************************
