Difference between revisions of "Wintermute/Version History"

Latest revision as of 07:49, 24 December 2013

0.8

0.7: APS: 85.07 Survival: 92.29 PL: 26 losses

Just one small tweak to the danger fuction, which may (or may not) have a significant impact. If it works next priority will be speeding Wintermute up, it's become quite slow.

0.6: APS: 83.76 Survival: 91.46 PL: 21 losses

Revert to 0.4
Begin branching when the wave just starts to pass over, instead of after it has passed completely
Fix bug in the second wave stop option which just froze the position instead of simulating a stop.
- This slowed the bot down significantly, so I had to modify the danger function to not take 'safe spots' into account on waves in order to speed it up

0.5: APS: 83.94 PL: 1386 - 710 pairings

0.4: APS:83.86 PL: 1356 - 698 pairings

Revert to 0.2
Fix the wave-generation direction bug
Remove LT and HOT weighting schemes - they help only minimally, and hurt everything else
Speeded up surfing by terminating branches which have exceeded the previous minimum danger
Tuned the danger function to be less likely to stop as a wave passes over
Orbits enemy instead of center of next wave to hit

0.3: APS:83.03 PL:1360 - 702 pairings

Found bug in wave generation that used direction from now instead of 2 ticks ago
Added a 'very advanced guns' weighting
Changed the way the weightings are normalized to make it more correct - LT and HOT no longer need to be handicapped
Waves are weighted by their 'time-until-hit' at each point

0.2: APS:83.19 PL:1366 - 699 pairings

Added HOT and LT to the weighting schemes (acts as pre-loaded HOT and LT shots)
Changed the weighting system to give the most accurate scheme all the score, the others 0 (with a rolling average, so it evens out)
Now surfs the second wave, and differently I believe than to other TrueSurfing algorithms: at each tick along the second wave I check what the danger would be if I decelerated at that point, and then take the minimum danger of all those decelerations, as well as continuing, as the actual danger. Also, the second wave is weighted equally to the first. (This may be changed at a later date.)
Takes ram damage into account for enemy energy drop

0.1: APS:82.92 PL:1350 - 698 pairings

@@ Line 1: / Line 1: @@
-.5:
+.8
+* Move from Simonton's KD-Tree to my KD-Tree
+.7: {{GTStats|link=http://literumble.appspot.com/BotDetails?game=roborumble&name=jk.precise.Wintermute%200.7|rumble=RoboRumble|scorelabel=APS|score=85.07|rank=13th|win=1048|loss=26|plrank=22nd|glicko2=N/A|pwin=97.58|vote=0.66|anpp=89.79|score2label=Survival|score2=92.29}}
+* Just one small tweak to the danger fuction, which may (or may not) have a significant impact. If it works next priority will be speeding Wintermute up, it's become quite slow.
+.6: APS: 83.76 Survival: 91.46 PL: 21 losses
+* Revert to 0.4
+* Begin branching when the wave just starts to pass over, instead of after it has passed completely
+* Fix bug in the second wave stop option which just froze the position instead of simulating a stop.
+** This slowed the bot down significantly, so I had to modify the danger function to not take 'safe spots' into account on waves in order to speed it up
+.5: APS: 83.94 PL: 1386 - 710 pairings
 * Weight gun scans by inverse Manhattan Distance, as in DrussGT