From totzke@news.cs.TU-Magdeburg.DE Tue Jun 7 17:29:32 EDT 1994 Article: 3215 of comp.ai.genetic Xref: glinda.oz.cs.cmu.edu comp.ai.genetic:3215 Path: honeydew.srv.cs.cmu.edu!fs7.ece.cmu.edu!europa.eng.gtefsd.com!howland.reston.ans.net!EU.net!Germany.EU.net!netmbx.de!zib-berlin.de!news.cs.tu-magdeburg.de!news.cs.tu-magdeburg.de!not-for-mail From: totzke@news.cs.TU-Magdeburg.DE (Holger Totzke) Newsgroups: comp.ai.genetic Subject: ParaTSP - a parallel GA and SA to solve TSP's Date: 6 Jun 1994 16:07:05 +0200 Organization: Univ. Magdeburg, Germany Lines: 216 Message-ID: <2svai9$9o5@fichte.cs.tu-magdeburg.de> NNTP-Posting-Host: fichte.cs.tu-magdeburg.de Keywords: TSP, GA, SA, parallel computer Hi, I wrote a software package to compute Travelling Salesman Problems (TSP) with genetic algorithms (GA) and with simulated annealing (SA). The source code in C is available by email. The program "ParaTSP" uses source code from Baeck's "GeneSyS v1.0". ParaTSP is running under UNIX. A parallel extension for the massiv parallel computer Parsytec GC is available (operating system is PARIX). The C-source are about 20000 lines of code. Now the help text for ParaTSP (GA): ------ ParaTSP 1.0 is running ... ParaTSP 1.0 (c) 1994 by Holger Totzke Usage: paratsp [Options] GA-specific options: -A tspfi # TSP input file -B I,G,R,P,L,N,W,B # breeder selection scheme I # inverse roulette wheel G # gauss R # random P # proportional L # linear ranking N # inverse linear ranking W # Whitley's index calculation B # Boltzmann -C xprob # crossover probability -D mudim # mutation step dimension -E N,M,_ # elitist scheme N # normal M # mutate _ # no elitism -F L,E,_ # fitness scaling type L # linear E # exponential _ # no fitness scaling -G gapsz # generation gap size -H elhld # elite number (hold) -I R,T,N,C,F,W,A # population initialization scheme R # random T # tour from input file N # nearest neighbour C # cheapest insertion F # farthest insertion W # random scheme without input (R,N,C,F) A # random scheme (R,T,N,C,F) -L R,W,F # elite selection scheme (sacrifice select) R # random W # weakest F # first weaker -M xy # mutation scheme x={C,A,_} # possible settings: y={M,S,I,R} # C # constant mutation rates A # adapted mutation rates _ # no mutation M # move S # swap I # invert R # random (M,S,I) -N norno # normalization number (town index) -O C,N,T # other GA-specific options C # crossover creates two childs N # normalization of tour T # filter twins -P popsz # population size -Q crowf # crowding factor -R R,C # replace scheme (dispersal) R # random C # crowding -S R,P # mates selection mechanism R # random P # positional index -T L,T,Q,A,1,2,3,_ # local tour optimization scheme L # Lin's 2-opt T # 2-opt Q # 2-quick A # sequence of 2-opt and or-opt 1 # or-opt(1) 2 # or-opt(2) 3 # or-opt(3) _ # no local tour optimization -W wdwsz # window size for scaling -X E,I,P,O,C,U,N,H,F, # crossover scheme R,_ # E # edge recombination (ERX) I # interval partially matched (IPMX) P # partially matched (PMX) O # order (OX) C # cycle (CX) U # uniform order based (UOBX) N # nearest neighbour H # cheapest insertion F # farthest insertion R # random (E,P,O,C,U,N,H,F) _ # no crossover -Y mprob # mutation probability -ZA fitpa # fitness scaling parameter A -ZB fitpb # fitness scaling parameter B -ZC fitpc # fitness scaling parameter C -ZD fitpd # fitness scaling parameter D -ZI chlen # cooling generations for boltzmann selection -ZL lwidx # lower index for breeder selection -ZN etamx # maximum expected value for linear ranking -ZT tmpct # temperature control for boltzmann selection -ZU upidx # upper index for breeder selection -ZX xpoin # number of crossover intervals in IPMX Parallel options: -pi comin # communication interval in generations -pm i,t,p,n # parallel model i # island t # token p # pollen n # neighbour -pn indno # number of individuals to change -pp popno # number of populations -pr nproc # number of processors -ps spdir # spread direction of pollen -pt r,p,g,o,h,t # type of communication topology r # ring p # pipe g # 2d-grid o # 2d-torus h # hypercube t # tree -px lwwin # lower limit to wind force -py upwin # upper limit to wind force General options: -a tourf # tour input file -b tdfrq # tour dump interval in generations -c # save tour for XTSP -d # show results on display -e noexp # total number of experiments -f dfreq # dump interval in generations -g nogen # total number of generations -h # this help information -? # -i infil # input file -k # no output files -m minqu # minimal tour length (quality) to abort -n savno # number of best individuals to save -o a,l,n,p,s,t,v # additional general options -q # create directory name and exit -r m,p # random number generator type m # Marsaglia's random number generator p # rand() from programming language C -s seed # seed for random number generator -t notrl # total number of trials -u popno # population no. to show on display -v intvl # number of trials between data collections -w # without report -x suffx # file suffix -y maxsp # number of generations wo. evaluations Additional options (following the general option -o): a # evaluate all individuals l # dump last generation n # simplified termination p # dump population s # trace schema history t # trace ParaTSP run v # collect tour values (much data !) ------ And the help text for SATSP: ______ SATSP is running ... ParaTSP 1.0 (c) 1994 by Holger Totzke Usage: satsp [Options] Options: -A tspfi # TSP input file -F tmpfa # temperature control factor -J jmax # number of steps per temperature levels -N nmax # number of temperature levels -S A,T # algorithm A # simulated annealing T # threshold accepting -T tmpst # start temperature -d # show results on display -e noexp # total number of experiments -h # this help information -? # -i comin # communication interval in temperature steps -m minqu # minimal tour length (quality) to abort -p nproc # number of processors -s seed # seed for random number generator -t r,p,g,o,h,t # type of communication topology r # ring p # pipe g # 2d-grid o # 2d-torus h # hypercube t # tree ______ Sorry for my bad english... Holger. -- +-----------------------------------------------------------------------------+ | Holger Totzke // Uni \\ | | Moskauer Str. 17 // Magdeburg \\ | | 39218 Schoenebeck // Computer Science \\ | | Germany // \\ privat voice: +49-3928/66334 | |-----------------------------------------------------------------------------| | e-mail: totzke@sunpool.cs.TU-Magdeburg.DE | +-----------------------------------------------------------------------------+ Article 3238 of comp.ai.genetic: Xref: glinda.oz.cs.cmu.edu comp.ai.genetic:3238 Path: honeydew.srv.cs.cmu.edu!das-news.harvard.edu!noc.near.net!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!spool.mu.edu!umn.edu!zib-berlin.de!news.cs.tu-magdeburg.de!news.cs.tu-magdeburg.de!not-for-mail From: totzke@news.cs.TU-Magdeburg.DE (Holger Totzke) Newsgroups: comp.ai.genetic Subject: ParaTSP - upload on FTP servers Date: 10 Jun 1994 13:32:54 +0200 Organization: Univ. Magdeburg, Germany Lines: 29 Message-ID: <2t9j16$14s@fichte.cs.tu-magdeburg.de> NNTP-Posting-Host: fichte.cs.tu-magdeburg.de Keywords: GA, SA, ParaTSP, parallel computing, TSP Hi, i have upload ParaTSP on various FTP server: * ftp.cs.tu-magdeburg.de:/in.coming/ParaTSP/* [141.44.21.200] * ftp.uni-paderborn.de:/incoming/unix/ParaTSP/* [131.234.2.32] * ftp.mi.uni-koeln.de:/pub/incoming/ParaTSP/* [134.95.229.226] * alife.santafe.edu:/pub/incoming/paratsp.* [192.12.12.99] -> not readable ParaTSP is a parallel GA & SA for solving TSP. It is based on GeneSyS 1.0 from Baeck. Comments and problems to me via e-mail. Holger. -- +-----------------------------------------------------------------------------+ | Holger Totzke // Uni \\ | | Moskauer Str. 17 // Magdeburg \\ | | 39218 Schoenebeck // Computer Science \\ | | Germany // \\ privat voice: +49-3928/66334 | |-----------------------------------------------------------------------------| | e-mail: totzke@sunpool.cs.TU-Magdeburg.DE | +-----------------------------------------------------------------------------+