Difference between revisions of "Archived talk:RaikoMX 20040227"

From Robowiki
Jump to navigation Jump to search
(migrating old talk)
 
 
(No difference)

Latest revision as of 04:51, 8 October 2009

RaikoMX Sub-pages:
RaikoMXVersion History - Archived Talk 20040227 - Archived Talk 20090525
       Archive        This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.     

Irrelevant stuff now, about 0.1:

An experimental movement which I wanted to test, but couldn't fit into Raiko (yet). The gun is very similar to Raiko's for now, though I have something up my sleeve once I have a baseline score. ;-) It's designed to make DT-style adaptive movement easy, so if it works out I'll probably start on a megabot with adaptive movement, virtual guns, etc... in which case I'll put a proper page here shortly. - Jamougha

Comments / Suggestions for a proper name? :-)


RaikoMX is a good name. -- PEZ

I concur. -- Tango

Sounds good to me, scary even. :) -- ABC

Lol, OK, I'll start a Raiko family. And ABC, if you think that's scary, wait until I hook it up to my anti-Shadow gun. ;-) - Jamougha

Interesting. So far it only loses to 5 bots. Of which 3 are mine and 1 half-mine. =) It's a very strong bot though! -- PEZ

Funny how you didn't include saved data on my bots. :) -- ABC

Looks like it's doing really poorly against classic style pattern matchers, it barely beats Fermat. Stat guns seem to all be getting burned really badly, though. -- Kuuran

Wow, way better than I'd hoped! Ah, if I can just find a few more bytes in Raiko...

ABC, yeah :-) saved data is weird with Shadow. Sometimes it helps and sometimes it makes things go *really* wrong. I suspect that if you were to reset your movement system every 20 round, say, then things would get... problematical.

Pez, it's that lingering loyalty in the Tityus code. ;-) Seriously, I think it's the good old time-since-velocity-change segment, at least in Gloomy. DT seems to lock onto this movement crazily well, too. I'll work out something.

Kuuran, yeah, I kind of expected that. There's quite a predictable pattern to it. Possibly I should alternate with Raiko's movement, they have roughly opposing peaks. - Jamougha

Wow 1908 points! You could be right about the segmentation. Now it's only Swiffer and GloomyDark that has an edge against RMX (veeery slight edge in the case of Gloomy). They have exactly the same gun and segmentation if I remember correctly. I think it's an excellent idea to switch movements. You have two excellent, but completely different movements there. Time to go MultiMode. Feel free to use Swiffer's system for rating and alternating movements. It should be pretty straight forward to understand I hope. If you also use its CribSheet data saving you can fit 1000+ opponents in your data dir and pack the movement rating data with it (like Swiffer does). I would say that #1 spot is within reach! If you send me an e-mail we can plan this throne attack together if you like. =) pez@pezius.com -- PEZ

PEZ, thanks, I wouldn't get too excited yet. :-) My experiments with alternating the two movements haven't been impressive so far, probably since this does better than Raiko's against all but 3 bots. However, I have another experimental movement which seems to do well against stat guns. If I can refine this then I might try it. Right now I'm just going to see if I can push up another 11 points. ;-) - Jamougha