[Home]Stelokim

Robo Home | Changes | Preferences | AllPages

Showing revision 51
About me

To infinity and beyond!

I am a Korean Robocode player. ()

I am the first person who put the South Korean flag to robocode rankings tables.

English wikipedia user page: http://en.wikipedia.org/wiki/User:SteloKim

My bots (stelo.)

You can download my old robots in robocode repository(including source codes):

And in my homepage:


Chat
Hey, welcome to the wiki! That's a pretty good rating for a first entry - better than most of us have had, I think. I don't have time to look at your bot this morning, so I'll just ask: what is the meaning of the name "RamTrackSurfer"? Best of luck with your bots. -- Voidious


Hi, Voidious. I was originally planning to write a bot that rams, tracks and wavesurfs! But a wavesurfer that rams is impossible i think now. (Wavesurfing == Perpendicular to the target != Ramming) I ended up with a basic wavesurfer + linear + circular targeting and avoid-a-bot-who-hits-me. I think ramming is basically not a good strategy in robocode world. -- Stelokim

Ramming is indeed not a good strategy if you aim for the top-10, but it can be fun to do and to watch. You can check the RamBots page and the RambotChallenge2K6 for some more info. Success with your current and future bots! -- GrubbmGait

Pleeaaaase don't write a rammer. It's impossible for a rammer to score very high (because you're easy to hit), but it's also impossible for other bots to score very high against one. So basically, every rammer that gets added to the rumble drags down the ratings of the top bots. If you like the idea of ramming, consider a MultiMode bot that rams when its opponent is low on energy. For example, Cigaret will ram if it has a lot of energy and the opponent is almost dead. The idea is that the bonus for killing someone by ramming is higher than the bonus for killing someone by shooting them. It's really fun to watch Cigaret do it. Just my two cents. See you in the rumble! --David Alves

I think, that any limitation is not applicable at this case. And any idea has rights for life. What about ratings... Hmm, there are many bots with great strategy against ram bots I think. So, anybody can write bot with behavior that he likes. -- DemetriX

Well yeah, I'm not saying that he can't write a rammer... I'm just pointing out that it hurts the top bots (including mine) and that rammers don't rank very highly. Go ahead and write one if you want though. =P --David Alves

A few months ago I have been write one already. ;-) -- DemetriX

I created a rammer!(Lifestealer) -- Stelokim


Just for your information, you don't have to create a new bot on the repository every time you update one of yours. After you login on the repository, choose the bot you want to update (you can enter the number in the url if necessary) and then select 'Save & Upload'. Your bot keeps the same number.

If you want to compare versions of the same bot, you can do that quite easily. Select 'details' from your bot, then compare it with any other bot, then select your own bot as first AND second. Now you can edit the url to compare against an older version. Bookmark this url if you want to compare more often.
Congratulations by the way with entering the top-50 with IntrinsicVolatility ! -- GrubbmGait

Thank you very very very much. I have some questions: When I just update my robot in robocode repository, should I increment version number and update our wiki's participants list accordingly? If I just update the robot, previously downloaded my robot in other person's computer still remains? Is there some sort of checksum thing to check changed robots? -- Stelokim

When you update a bot, you should increment the version number and change the participants list accordingly. (like stelo.Mirror 1.0 => stelo.Mirror 1.1) If you just put a new jar with exactly the same name, the clients will not notice the change. The robots name including version number must be unique, so the ranking will be built up from scratch. -- GrubbmGait

Hey man, I have to ask - what makes Moojuk so slow? Maybe you have some bug that is making something loop way more than it should? I usually keep my RRAH clients' CPU constant set very high so that nobody skips turns, but it's probably still skipping a lot of turns on mine, and even more on a normal CPU constant setting. It takes my computer 5 minutes to run 2 Moojuk's against each other for 35 rounds, compared to 1:05 for 2 copies of Dookious or 15 seconds for 2 copies of Komarious. -- Voidious

Moojuk 1.1 pattern-matched every tick for some accuracy, I will fix it soon. -- Stelokim

Ah. You can safely just use HeadOnTargeting until your gun is 3-4 ticks from being cool, that should speed things up a lot. Lukious (and Shadow, I think) only aim once per shot fired, 2 ticks before the gun is cool; if you keep the gun aimed a few degrees ahead of HOT when the enemy is at full speed or so, you can guarantee that you can aim to all reachable aim offsets in 2 ticks. -- Voidious

It seems that MatchupMicro? 1.1 tries to compete with ad.last.Bottom, I don't think that was your intention. -- GrubbmGait

I don't know what's wrong with this. It is supposed that someone got a corrupted jar file. I downloaded it from my repository and tested it. It worked fine. I upload same file with another file name. Let's see. -- Stelokim

I got what it is. MatchupMicro? 1.1 used robocode.Rules.getBulletSpeed?(bulletPower). That class is since 1.1.4. I think some clients' version is less than that, causing NoSuchClassException? or something. -_-; -- Stelokim

-- Are you using Simontons SingleTick? Maybe try making your own implimentation of it, might work out the bugs. Generally, bulletSpeed is equal to (20 - 3 * bulletPower). -Chase-san

I don't think the Rules.* stuff would be the problem - I think your bot would get a 0 score and the result would be discarded. I am still running 1.1.3... I wanted to upgrade to something newer, but it had an issue with something in the RRAH client setup when I tried 1.2+. If 1.1.5 has Rules and works fine, I'll upgrade to that for sure. -- Voidious

@Chase-san: I'm using my own implementation of PM which uses array of double, matching closest match. -- Stelokim

MatchupMicro? 1.2 reverted to using (20 - 3 * bulletPower), entered the competition, and the problem went away. -- Stelokim

Umm, Stelokim, entering SoJNanoMelee in the rumble is probably not a good idea if its a Junior, mostly becuase very few people are using the latest 'Beta' robocode for rumble battles (as this has caused problems before) and will probably just error out or get 0 rating on the older clients. Perhaps it would be prudant till atleast the stable version comes out to submit it for the rumble. But Congratz on making the first (that I know of) non-sample Junior robot. (really, don't take this the wrong way, i'm just trying to save you some grief) --Chase-san

Wow. What prompted the mass exodus of your bots from the rumble? -- Simonton


Robo Home | Changes | Preferences | AllPages
Edit revision 51 of this page | View other revisions | View current revision
Edited November 4, 2007 13:13 EST by Stelokim (diff)
Search: