120 lines
3.5 KiB
QBasic
120 lines
3.5 KiB
QBasic
1 rem Tic Tac Toe solving app that learns what WOPR learned: you can't win
|
|
2 rem Only three starting positions are examined. Others are just reflections of these
|
|
3 rem b% -- The board
|
|
4 rem al% -- Alpha, for pruning
|
|
5 rem be% -- Beta, for pruning
|
|
6 rem l% -- Top-level loop iteration
|
|
7 rem wi% -- The winning piece (0 none, 1 X, 2, O )
|
|
8 rem re% -- Resulting score of 4000/minmax board position. 5 draw, 6 X win, 4 Y win
|
|
9 rem sx% -- Stack array for "recursion" X can be P, V, A, or B for those variables.
|
|
10 rem v% -- Value of a board position
|
|
11 rem st% -- Stack Pointer. Even for alpha/beta pruning Minimize plys, Odd for Maximize
|
|
12 rem p% -- Current position where a new piece is played
|
|
14 rem rw% -- Row in the Winner function (2000)
|
|
15 rem cw% -- Column in the Winner function (2000)
|
|
18 rem mc% -- Move count total for debugging. Should be a multiple of 6493
|
|
19 rem Note: Can't use real recursion with GOSUB because stack is limited to roughly 5 deep
|
|
20 rem BASIC doesn't support goto/gosub using arrays for target line numbers
|
|
21 print "date: "; date$
|
|
30 dim b%(9)
|
|
32 dim sp%(10)
|
|
34 dim sv%(10)
|
|
36 dim sa%(10)
|
|
37 dim sb%(10)
|
|
38 mc% = 0
|
|
39 print "start time: "; time$
|
|
40 for l% = 1 to 10
|
|
41 mc% = 0
|
|
42 al% = 2
|
|
43 be% = 9
|
|
44 b%(0) = 1
|
|
45 gosub 4000
|
|
58 al% = 2
|
|
59 be% = 9
|
|
60 b%(0) = 0
|
|
61 b%(1) = 1
|
|
62 gosub 4000
|
|
68 al% = 2
|
|
69 be% = 9
|
|
70 b%(1) = 0
|
|
71 b%(4) = 1
|
|
72 gosub 4000
|
|
73 b%(4) = 0
|
|
74 rem print "mc: "; mc%; " l is "; l%
|
|
80 next l%
|
|
85 print elap$ ; " for "; l% - 1; " iterations"
|
|
86 print "end time: "; time$
|
|
87 print "final move count "; mc%
|
|
90 system
|
|
100 end
|
|
|
|
2000 wi% = b%( 0 )
|
|
2010 if 0 = wi% goto 2100
|
|
2020 if wi% = b%( 1 ) and wi% = b%( 2 ) then return
|
|
2030 if wi% = b%( 3 ) and wi% = b%( 6 ) then return
|
|
2100 wi% = b%( 3 )
|
|
2110 if 0 = wi% goto 2200
|
|
2120 if wi% = b%( 4 ) and wi% = b%( 5 ) then return
|
|
2200 wi% = b%( 6 )
|
|
2210 if 0 = wi% goto 2300
|
|
2220 if wi% = b%( 7 ) and wi% = b%( 8 ) then return
|
|
2300 wi% = b%( 1 )
|
|
2310 if 0 = wi% goto 2400
|
|
2320 if wi% = b%( 4 ) and wi% = b%( 7 ) then return
|
|
2400 wi% = b%( 2 )
|
|
2410 if 0 = wi% goto 2500
|
|
2420 if wi% = b%( 5 ) and wi% = b%( 8 ) then return
|
|
2500 wi% = b%( 4 )
|
|
2510 if 0 = wi% then return
|
|
2520 if wi% = b%( 0 ) and wi% = b%( 8 ) then return
|
|
2530 if wi% = b%( 2 ) and wi% = b%( 6 ) then return
|
|
2540 wi% = 0
|
|
2550 return
|
|
|
|
4000 rem minmax function to find score of a board position
|
|
4010 rem recursion is simulated with gotos
|
|
4030 st% = 0
|
|
4040 v% = 0
|
|
4060 re% = 0
|
|
4100 mc% = mc% + 1
|
|
4102 rem gosub 3000
|
|
4104 if st% < 4 then goto 4150
|
|
4105 gosub 2000
|
|
4106 if 0 = wi% then goto 4140
|
|
4110 if wi% = 1 then re% = 6: goto 4280
|
|
4115 re% = 4
|
|
4116 goto 4280
|
|
4140 if st% = 8 then re% = 5: goto 4280
|
|
4150 if st% and 1 then v% = 2 else v% = 9
|
|
4160 p% = 0
|
|
4180 if 0 <> b%(p%) then goto 4500
|
|
4200 if st% and 1 then b%(p%) = 1 else b%(p%) = 2
|
|
4210 sp%(st%) = p%
|
|
4230 sv%(st%) = v%
|
|
4245 sa%(st%) = al%
|
|
4246 sb%(st%) = be%
|
|
4260 st% = st% + 1
|
|
4270 goto 4100
|
|
4280 st% = st% - 1
|
|
4290 p% = sp%(st%)
|
|
4310 v% = sv%(st%)
|
|
4325 al% = sa%(st%)
|
|
4326 be% = sb%(st%)
|
|
4328 b%(p%) = 0
|
|
4330 if st% and 1 then goto 4340
|
|
4331 if re% = 4 then goto 4530
|
|
4332 if re% < v% then v% = re%
|
|
4334 if v% < be% then be% = v%
|
|
4336 if be% <= al% then goto 4520
|
|
4338 goto 4500
|
|
4340 if re% = 6 then goto 4530
|
|
4341 if re% > v% then v% = re%
|
|
4342 if v% > al% then al% = v%
|
|
4344 if al% >= be% then goto 4520
|
|
4500 p% = p% + 1
|
|
4505 if p% < 9 then goto 4180
|
|
4520 re% = v%
|
|
4530 if st% = 0 then return
|
|
4540 goto 4280
|
|
|