User talk:Positive

Also, welcome to the wiki. =) Feel free to make a page for yourself and your bots. Good luck in the rumble! --Voidious 21:25, 11 July 2009 (UTC)

I will! :) --Positive 23:39, 11 July 2009

Welcome! Always nice to see someone new from our small but well-known country being active, even if it means that I am not the highest rood-wit-blauw in melee anymore. --GrubbmGait 23:31, 13 July 2009 (UTC)

Well, it's all for king and country, right GrubbmGait :)? Thanks for the welcome. --Positive 17:09, 14 July 2009 (UTC)

Enterprise

Okay, I just created a new robot, Enterprise 1.0, that tries to do bullet shielding (trying to shoot down enemy bullets). People have already discussed it, but I haven't seen an actual bot trying to do it. It's opensource, and everyone is free to play with and modify it. I'm thinking it might actually be fun to make a contest out of it. What do you guys think? --Positive 22:09, 4 August 2009 (UTC)

Neato. There are a couple out there; one is User:Pedersen/Roland. Damage given + damage taken might be a good scoring system, or we could modify the reference bot(s) to actually track how much bullet damage was destroyed by the bullet collisions. It would be kind of like the RamBot Challenge 2K6 (still on the old wiki.) Feel free to write up the challenge page, if you'd like. =) --Voidious 14:16, 5 August 2009 (UTC)

To be honoust, I had expected a somewhat more enthoustiastic response, and I am somewhat less motivated for it now than when I wrote that message. So I think I'll drop/freeze this project for now. Thankyou for your great tips and link though, I'll remember those. :) --Positive 23:01, 6 August 2009 (UTC)

Well I think this is a pretty cool idea, I might participate if you were to write the challenge--CrazyBassoonist 23:11, 6 August 2009 (UTC)

Maybe in the future. :) --Positive 14:46, 7 August 2009 (UTC)

I have a competitor, it would be exciting to see someone do this better then me :) --Rsim 18:57, 7 September 2009 (UTC)

Feel free to draft up a challenge. :) --Positive 15:03, 11 September 2009 (UTC)

Videos

I really like the way the videos are turning out. :) How do you feel about (very selectively) adding videos on the pages of robots? For example, I was thinking the video I uploaded about headon targeting might also be cool on HawkOnFire's page. --Positive 20:05, 16 August 2009 (UTC)

I've been meaning to say how cool these videos are for a while; I think it would be a great idea to add videos to bot pages--CrazyBassoonist 20:10, 16 August 2009 (UTC)

Yeah, I say go for it. I really like the videos, too. I considered adding one to the Diamond page, actually, but I wasn't sure which one to use, then I forgot... --Voidious 20:12, 16 August 2009 (UTC)

Great, I'm glad you guys like 'em too. :) I'll go for it. Of course, feel free to use them in any way you like as well. --Positive 20:25, 16 August 2009 (UTC)

Those video's are an excellent idea, this way you can show people what Robocode looks like in just a few seconds. Next to that it can be used to show some speciality of a bot. Btw, what do you use to make those videos? --GrubbmGait 22:25, 16 August 2009 (UTC)

I use CamStudio (opensoure app) together with replays of fights. After twiddling with it I noticed the Lossless Codec works best, together with capture frames and keyframes every 25 ms, playback 75 ms, region size of 640x480, autopan turned off. You should record the fight at about 10 ticks/second. --Positive 22:45, 16 August 2009 (UTC)

I've been thinking about what robot could best illustrate Minimum Risk Movement, preferably with debugging graphics. I've been thinking about both Diamond and Portia, but it might also be cool to modify HawkOnFire or another opensource bot to use debugging graphics. Any thoughts? :) --Positive 19:36, 17 August 2009 (UTC)

One problem (for any bot) is that I'm not sure how to graphically display what is behind the risk calculations. I think the 3 main things are: distance-squared to enemies, orthogonality to enemies, and how many bots you'd be the closest to (targeted by). I'll try to think about how to display any of that (and maybe add some more graphics to Diamond). I actually haven't looked at Portia's much lately, do you show any of that stuff? --Voidious 19:49, 17 August 2009 (UTC)

Well, "how many bots you'd be the closest to" can be displayed by circles around opponent. The other two I'm not sure of. Portia only shows and uses those circles and simulated bullets, it doesn't really use orthogonality and/or distancesquared.

One way might be to let the robot paint the entire battlefield with a transparent color indicating danger there. Although you won't exactly see what it's derived from. --Positive 20:13, 17 August 2009 (UTC)

Okay, for now I think I'll try a HawkOnFire/Diamond hybrid (HawkOnFire using diamonds paint system :)). Mainly because: HawkOnFire doesn't reevaluate points every tick (like Portia, Diamond, and many others do), is reasonably simple, and has an "understanding HawkOnFire" page which the video might be helpfull for. I'll upload it shortly. --Positive 00:58, 18 August 2009 (UTC)

Awesome, it looks great! I hope you don't think I'm crazy now that you've seen how complex my color coding logic is. =) --Voidious 01:25, 18 August 2009 (UTC)

Nope, I like it. :) It's especially usefull in HawkOnFire, because the return of the evaluation function fluctuates quite a bit. --Positive 01:29, 18 August 2009 (UTC)

Chat

Hi Positive. I want to add you as one of the contributors for the new getNewVelocity() (together with Voidious and Skilgannon). Could I get your real name for the credits? If you don't want to add it here, I should like you to send a mail to me with your name. :-) --Fnl 19:55, 21 August 2009 (UTC)

Thanks for thinking of me. :) However, I'd prefer it if you don't add me (I might seem paranoid, but I try to keep my real name off internet so people can't google me). --Positive 21:27, 21 August 2009 (UTC)

Okay. I respect that. Could I use your username here instead? --Fnl 13:53, 22 August 2009 (UTC)

Yeah, that's fine. :) --Positive 13:56, 22 August 2009 (UTC)

Just to inform you, your client 'Positive2' is generating results with 'zero' scores for Shadow 3.84a. Maybe the mem-setting is not right for that client? --GrubbmGait 13:43, 30 August 2009 (UTC)

Yes, I noticed. That other client was running the superpack on Java 5, and maybe Shadow has some Java 6 code. I just finished updating that client to Java 6 and changed the name to Positive3, hopefully that will fix it. (Darkcanuck, could you please delete the uploaded results from Positive2?) --Positive 13:48, 30 August 2009 (UTC)

Only possibility is that Shadow use the Java 6 specific Red's kd-tree. » Nat | Talk » 14:02, 30 August 2009 (UTC)

I doubt that's the issue unless the version ABC used was an old version. Both the most recent version before 3.84a was released and the latest version should be Java 5 compatible (if they aren't, please let me know). I think that perhaps a more likely cause for the 0 scores may be the issue with rounding of split values causing infinite loops, which I just fixed last night some time after 3.84a was released. --Rednaxela 16:20, 30 August 2009 (UTC)

Then again, I'm not sure that's more likely, considering that I'd expect at least a little score to be had before that issue would kick in, plus that issue would tend to cause an OutOfMemory issue. --Rednaxela 16:41, 30 August 2009 (UTC)

Well, it's weird. Whatever the problem is, I don't have much access to that computer anyway, so I'll stick to my own pc for now. By the way, tommorow is my first day attending computer science at university, I'm excited! --Positive 21:03, 30 August 2009 (UTC)

That's cool, I think the university experience is quite nice, although knowledge-wise mine isn't a good university (the best in Venezuela, but that still quite low) the whole experience is very nice. Hope you have lots of fun. --zyx 22:02, 30 August 2009 (UTC)

Cool, best of luck! I also quite enjoyed my time studying Computer Science, I hope you do too. --Voidious 01:53, 31 August 2009 (UTC)

Thank you! --Positive 17:20, 31 August 2009 (UTC)

High gun cooling rate tournament

The past days, I've been running some battles with gun cooling rate = 2, and it's pretty fun to watch. Though most bots act rather silly and send streams of a high power bullets at opponents completely draining themselves. I think better strategies should be possible, and perhaps something could be learned about energy management with the default cooling rate as well. :) Any thoughts? --Positive 11:19, 4 October 2009 (UTC)

Okay, I guess noones interested. On to bigger things. :) --Positive 01:58, 9 October 2009 (UTC)

If you make tournament, I think someone will join it. --Nat Pavasant 02:45, 9 October 2009 (UTC)

No time, but I'd be interested if there was =) --Skilgannon 10:07, 9 October 2009 (UTC)

Thanks for your responses, I'm glad there is some interest. :) I'm currently buzy with university, but I'll get back on this. --Positive 19:54, 11 October 2009 (UTC)

To get back on this, I've been trying to get RoboResearch running to run a tournament, but I've kinda given up on it because I keep getting exceptions. I'm sorry if someone was secretely holding his breath on this. :) --Positive 21:29, 25 October 2009 (UTC)

It won't run multi-threaded like RoboResearch, but the automation I use for the Twin Duel should be easy to adapt, if you want to take a look: Twin Duel/Tourney Runner. (Note that the part that restarts the match if there was an error doesn't work, I'm not sure why.) --Voidious 21:49, 25 October 2009 (UTC)

To be honoust after I failed with RoboResearch I kinda lost motivation for that project. Your code is very interesting and I'm sure I'll try to use it the next time I have a 'brilliant idea' :). Thanks for sharing! --Positive 11:21, 27 October 2009 (UTC)