From: Phil Goetz Subject: Re: Genetic Boggle alg.? Date: Mon, 5 Jun 2000 19:29:12 -0400 (EDT) > I saw your comments in http://www.cs.buffalo.edu/~goetz/interests.html > about having experimented with genetic algorithm for Boggle. I'd be > interested in seeing that code (and a report if you wrote one?) and > adding it to the archive of sources that I keep for the wordgame-programmers > mailing list. (http://www.gtoal.com/athome/scrabble/ - see the very end of > the rather long page) > > Just out of interest - did you think of generating dense boggle boards > independently, or was this an entry for the Programmer of the Month > competition, which did a boggle puzzle like this some time ago? > (you can see the winning entry at > http://www.gtoal.com/athome/scrabble/potm/inverse-boggle/ > ) By coincidence almost the same problem was posed in one of > the MacTech competitions: > http://www.mactech.com/articles/mactech/Vol.14/14.08/Aug98ProgChallenge/ > > regards > > Graham Toal No, I have never heard of either of those competitions. I just played a lot of online Boggle. Where can I learn about the POTM contest? There was no info about the winning entry at the page. I never did write a report. Below is the code. Here are some good boards it found (the 4 rows, followed by the number of words in that board). You can find better boards by setting BIRTHS and GENS higher. I used my computer's existing "boggle" program to actually find the words. I don't have the source code for that. Note that the source code makes a Unix call to that boggle program. ektt oras pyow aotn 117 gush taiv inde elop 113 xydb retu asic ipeh 115 ipyt sand tora icej 122 ugiu ohts bere dacv 101 obkc soen dpat ieru 129 etes tcia avdr lege 122 zetb laoo pont sice 150 psdl oeie eras zpte 170 adyi lerv apea dtsf 171 ecte spao lidr vate 193 oeav dite mart pcso 194 -------------------------------------------------------------- ADDENDUM: I have modified Phil's original source (the copy here is the modified version) to include an internal boggle score evaluator. This is at least 10 times faster than calling an external program. Otherwise his code is unchanged. If you fetch this code, you can either use the words.dwg file here (/usr/dict/web2 in dawg format) or you can go to ../spell and get the token program which lets you build your own dawg files. You'll also need the library utilities from ../spell/SpellLib to link this code and the dyntrie.c snippet from ../gtboggle (which should soon be making its way into the ../spell library I hope) Graham