Editing GloomyDark/Comments20031213

Robo Home | Changes | Preferences | AllPages

Nice bot!  You are quiet close to DT:

1st: pe.SandboxDT 2.11	        86696	34900	6980	38839	5976	1	0	699	302	0
 2nd: pez.GloomyDark 0.1	61457	15100	3020	39716	3552	4	64	302	698	0

GD beat DT on bullet damage, it's just the survival that's not too good.  A little tweaking and you might force paul into an update. -- [[Tango]]

Is that from a 1000 round battle? I was going to say it was just luck, but not with that many rounds. Cool, but still not close enough. DT is a survivalist bot and thus sacrifices bullet damage for survival, and does it extremely well. [[Marshmallow]] used to always beat DT clearly in bullet damage, but over 1000 rounds always lost in total score. Of course that was earlier DTs. M doesn't have much of a say against DT 2.11. I'm glad GD looses with some face left. =) Thanks for posting those results. I have been running much shorter matches and only have tested long battles against [[Kawigi]]'s and [[iiley]]'s bots. DT feels a bit unreachable for me at the time. 2.11 is such a strong bot that I almost doubt I can do anything about it. Almost. I '''will''' try! -- [[PEZ]]

I wonder what drm.Magazine does that DT doesn't like... And [[GB]], the strangest bot ever I think, seems to score better than Gloomy against DT too. I wonder if GB could beat DT if I give it a mirroring movement. Wait, I can try that. =) -- [[PEZ]]

Yes, it was 1000 rounds.  I was also wondering what was happening with Magazine... -- [[Tango]]

I'm just running 1000 rounds of DT vs Magazine... it doesn't look like it was a fluke... DT doesn't seem to be able to hit it... -- [[Tango]]

1st: pe.SandboxDT 2.11	42576	17400	3480	18630	3065	0	0	348	97	0
 2nd: drm.Magazine 0.39	24906	4850	970	17718	1347	20	0	98	348	0

Those are very SlowBots... that was less than 500 rounds and took ages...  it looks like DT takes a long time to learn how to hit Magazine, but once it does, it is fine.  That's the problem with small battles. -- [[Tango]]

That's interesting. In RoboRumble@Home teh chances that DT and Magazine meet twice on the same client are minimal. Which means, that with a movement like Magazine, which DT seems to have trouble figuring out quick enough you can beat DT. Magazine obviously does. =) My test with using MirrorMovement in [[GB]] didn't work out all that well though. It seems like its old movement was what DT has problems with, not its guns. Nevertheless I'll release this mirroring GB since it is a crap bot anyways. =) Funny that if you use a GoToBot mirroring is as easy as:<pre>goTo(new Point2D.Double(centerX - (enemyLocation.getX() - centerX), centerY - (enemyLocation.getY() - centerY)));</pre>I stole the code from Marshmallow's AntiMirroringTargeting gun. (Which uses M's future position instead of enemyLocation of course). -- [[PEZ]]

Wow... Already someone using my graphing utility. :D - That's an excellent looking curve PEZ, it's very flat and spread out; curve flattening targeting bots (Fractal especially, quite obviously from the graph) will have a lot of trouble hitting it. Very nice. -- [[Vuen]]

Thanks for those words. I had actually hoped it would make a bigger difference with this tuned movement. But it hasn't exactly soared through the rankings yet. The automatic movement adapter (DT style) isn't in operation yet, since I can't get it to not wreak havoc. Maybe with that in place I can do some real kick-ass curve flattening. -- [[PEZ]]

Summary:
Consciousness check: 3 - 2 =
This change is a minor edit. Send email notification that GloomyDark/Comments20031213 has been changed.
(Visit Preferences to set your user name.)

View other revisions
Robo Home | GloomyDark | Changes | Preferences | AllPages