View source for Talk:Tomcat/Version History

Now not all pairings for v3.67a completed, but after 782 pairings this version get - ~0.4 APS and it's strange, because the change is fix for used data for surfing. Tomcat for a long time use data from two ticks before fire detection for range search. For example, if Tomcat detect a energy drop at 10th tick, then it use for range search data from 8th tick, while, theoretically, enemy uses data from 9th tick for aiming. And v3.67a try to use data from 9th tick. Without success. Anyone has any suggestions why?

If you detect it on the 10th tick, the bullet was fired (setFire) on the 9th tick, from the enemy's location on the 9th tick, using data from the 8th tick. There's no tick delay between the data they're using and calling setTurnGun, so the setTurnGun on tick 8 is using data from the 8th tick, not the 7th tick.

And yeah, gun heat waves can create this fireTime=9 wave pretty accurately at time=7 or time=8 instead of waiting for time=10. If there are no other waves or all your movement options on the nearest wave are about the same, 1-3 ticks of extra surfing can really help. (Diamond would only create the gun heat wave at time=8 or 9, not at 7, because I didn't find benefit 3 ticks out, but maybe it could help if implemented right.)

By "using data from the 8th tick, not the 7th tick" I assume you mean their data internally on the 8th tick right?

By 7th tick I meant that their scan of you on the 8th tick is your actual location on the 7th tick due to scan delay.

There's no delay as far as I know. On each tick, everyone moves, then everyone receives events based on the current state of the battle field.

Ahh right, nevermind. For some reason I was incorrectly thinking scans occured before movement. To revise, what it is then is... to construct the search data for an energy drop you detected at the 10th turn, you should use:

1. The enemy scan that you received on the 9th tick (they turned their gun at the 8th, and 1 tick delay in your scan)
2. Your robot data from the 8th tick (Last one they would have got scan data of before turning their gun)

As a note, it is fun how you have to use different ticks for where the bullet fires from, and the search point :P

O, sorry for stupid question. Thanks, guys:)

Not a stupid question, off-by-1 errors can creep up all over the place in Wave Surfing!

Accidentally lxx.Tomcat 3.67 became PL King again. How to hurry VoidBot to take snapshot of this insignificant, but very pleasant moment?:)

Wow, you're really abandoning Tomcat for ConceptA? So now Diamond will have to worry about beating TWO of your bots?! :-)

Best of luck! A fresh start can make a big difference. Especially with a big bot that climbs its way up the ranks, you learn and change so much along the way - it's easy for the code to end up a huge mess.

Thanks for luck:) Actually, this night i understand that i have stuck with gun attributes tuning, but there're many other areas to improvement in Tomcat. So Tomcat's abadoning is delayed a little, but in the long term i will switch to ConceptA. And, i think, it's better for you, because current version of Diamond beats current version of Tomcat and totally annihilate current version of ConceptA:)

Looks like, I hastened to write off Tomcat - + ~0.65 APS for v. 3.58, some new ideas for v. 3.59 & 3.60 and Tomcat again in game:) But ConceptA is still in future plans:)

Diamond 1.6.17 gets 51.39 over 105 battles (so far) against Tomcat 3.48c. Losing with 49.36 against the latest DrussGT, so it may be a 3-way tie if Tomcat beats DrussGT.

I'll work on resolving that "problem" soon. ;)

Good luck, i think (or hope?) you need it, because i have two ideas, which must hit and brings me behind 88% APS:P But there're a little problems - both ideas need in God's blessing:)

I've made advances in PL for XanderCat, and think with a little more work I could land a solid 4th place in PL, but I'll have to do something significant to challenge any of the top 3. Nevertheless, despite averaging around 45, I did beat Diamond in one match for the latest version with a score of 57; I can't help but wonder if that gave Voidious a scare. :-)

I did notice (nice work!), but honestly I'm too busy being scared of DrussGT and Tomcat right now to worry about anyone else. =) And when it takes a few hundred battles vs DrussGT to eke out gains a tiny bit at a time, I just don't have the CPU power to focus on too many opponents at once, even if I wanted to...

Wow, that's a lot of changes in one update. You are brave. =) Looks like it worked out though...

I hate develop movement's mechanics, but eventually my tests show +0.5 APS and i quick release this version:) but i already find a bugs in this version:)

But developing movement is fun. :)

No, fun - it's "Destroy the rest", but movement's development is suffering:)

Non-sense. Isn't the phrase, "dodge the best"? ;)

For me it were better if "predict enemy bullets the best":)

I thought it was "Destroy DrussGT".

And Diamond:) And Shadow And Hydra And so on And so on:)

the knack of rhythm is lost to you both.

Hmm, for me it's more like: 'Get slaughtered by the best, destroy the rest' , but I still have ideas only hardly any time. Now you have only 1.5 APS left to get to #2, but with your pace, it will cost you 3 to 4 weeks to accomplish that. Congrats with your astonishing appearance in the rumble!

I don't think, that Voidious so easy give up second place and only two jokers left in my arms. So i'm not so optimistic:) But lets see where is my limit:)

I'm down to 0 jokers and currently looking for a new deck.

I'd say 0 jokers and looking for a new deck is a nice relaxed state to be in, and a good time for the idle pondering of new ideas. It can be frustrating having 5 or 6 jokers but not enough time to get any into play :P

I have like 6 but they are secured to the ceiling with fishing line and I lack a ladder.

Chase do not use so difficult sentences with english lamer:) I understood you like you know 6 jokers, but can not take them some reason

More or less. English is full of strange idioms.

Basically I know a few things that could work out pretty well, but not sure how to go about implementing them at the moment.

Including a potential solution to the multiple wave surfing problem.