From news-rocq!jussieu.fr!fr.usenet-edu.net!usenet-edu.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.cwix.com!sjc-peer.news.verio.net!phl-feed.news.verio.net!news.verio.net!news.ttsg.com!ayanthegreat@crosswinds.net Thu Mar 22 14:12:00 2001
Article: 12401 of rec.games.corewar
Path: news-rocq!jussieu.fr!fr.usenet-edu.net!usenet-edu.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.cwix.com!sjc-peer.news.verio.net!phl-feed.news.verio.net!news.verio.net!news.ttsg.com!ayanthegreat@crosswinds.net
From: Ayan Chakrabarti <ayanthegreat@crosswinds.net>
Newsgroups: rec.games.corewar
Subject: Warrior Improvement
Date: 21 Mar 2001 15:30:13 -0500
Organization: TTSG
Lines: 105
Distribution: world
Message-ID: <5.0.2.1.2.20010321190330.00a24050@pop.crosswinds.net>
Reply-To: ayanthegreat@crosswinds.net
NNTP-Posting-Host: odin.ttsg.com
X-Trace: odin.ttsg.com 985206613 24691 216.231.111.29 (21 Mar 2001 20:30:13 GMT)
X-Complaints-To: abuse@news.ttsg.com
NNTP-Posting-Date: 21 Mar 2001 20:30:13 GMT
Originator: corewar-l@koth.org
X-Mailer: QUALCOMM Windows Eudora Version 5.0.2
Xref: news-rocq rec.games.corewar:12401



Hi,

I've been trying to make a scanning vampire called Mohini for quite some 
time now and I seem to have hit a brick wall in terms of improving it. So 
I'm posting the source code here. I would greatly appreciate it if somebody 
gave me directions on improving it. Please tell me whether I should work on 
optimizing my step size or if there is something else that I am not doing 
right or if the flaw is in the basic strategy  itself.

Mohini is basically a scanner which lays down a carpet of traps as it 
scans. Whenever it finds anything, it just changes the destination where 
the carpet is being laid (after having made appropriate changes to the 
trap). If a process is captured, it switches from scanning to a core-clear.

I've been trying to optimize this thing for a long time. It does especially 
badly against scanners (as far as I could make out from the Wilkies 
scores).  Any further improvements that I try to make seem to backfire 
(something like a rain dance causing an earthquake ;) like trying to make 
captured processes implement a gate.

Again, I would be very grateful for any suggestions (including criticisms).

;redcode
;name Mohini
;author Ayan Chakrabarti
;strategy Scan + Vamp -> Clear
;strategy Captured processes stunned.
;assert CORESIZE == 8000

;Wilkies Points = 122.92

init    equ kill+100
inscan  equ kill+2*step
step    equ 1226


;--- Scan

kill     mov.i  trap,>ptr
          add.f  stinc,trap
check    jmz.f  kill,@trap
          mov.b  trap,ptr
          sub.ab #1,ptr
          mov.a  #pit-trap,trap
          sub.ba trap,trap
          jmp kill,<-400

stinc    dat    #-1,#step

;------
for     3
         dat     0
rof
;------

;--- Clear (something captured)

ptr2            dat     0,pit-ptr2
                 dat     0,pit-ptr2
                 dat     -10,pit-ptr2
                 dat     -10,pit-ptr2

done            spl     #4000,pit-ptr2
                 mov.i   @2,>ptr2
                 mov.i   @1,>ptr2
                 djn.b   -2,{done

                 dat     0,0
                 dat     0,0


;--- The trap which doubles up as scanning pointer.

trap     jmp    pit-init,inscan

;--- The pointer at which traps are being dropped. Also used to switch
;    from scan to clear

ptr      jmp    done-check,init

;------
for 9
dat 0
rof
;------

;--- The pit

pit      mov    ptr,check               ; Process captured. Switch to clear.

i        for 9
          spl     #0
          rof
          jmp    -9
;--- The end :)

end check


---------------
Ayan Chakrabarti
" A bird in the hand is better than two overhead "