GhOsT16790: i made it
GhOsT16790: havent done this for awhile
Uljahn: k bye, Automaton2000
Automaton2000: i wish there was a way to check for timeouts
chinaredclif: gregory ?
PatrickMcGinnisII: Automaton2000 advice on bender 4?
Automaton2000: you have a better bot
chinaredclif: Xedux ?
Xedux: what ?
Xedux: chinaredclif ?
chinaredclif: can u give ur resolution of the clash code ?
chinaredclif: '#' ...
chinaredclif: thx u so much
Xedux: see your private chat
jacek: how may we help you, Sir/Madam
jrke: which is current PotW
black_ghost: hello all
Xedux: SuperMuppet ?
Xedux: can you send the answer ?
Xedux: of recent clash
Xedux: nth sequence ?
AntiSquid: Gangplank are you talking about DotA 2 ?
AntiSquid: or did you mean botters of the galaxy ?
A_Potato: hello Steve
A_Potato: it's nice to meet you
Uljahn: pls stop flooding
AntiSquid: Gangplank well this isn't a dota 2 website so most people won't know how to help you. As for the pudge hook, practice. I am sure you realize the top players are practicing almost 24/7 It's a hard game
AntiSquid: what MMR are you at Gangplank ?
MSmits: mmh so yesterday it was clash of clans and now people are talking about dota 2
MSmits: methinks CG is advertising on the wrong platforms
jrke: hey Ms
MSmits: hey jrke
MSmits: working on transposition tables now for my solver
MSmits: when that is done I can start putting it into my real bot
MSmits: assuming i can get the solve time low enough
MSmits: currently I can do 200 endgames in a few seconds, but a single game can take a full second still
MSmits: TT might take that down below 100 ms, maybe even as low as 10 ms if I am lucky
jrke: yeah some are complicated
MSmits: I also am not sure if all my heuristics are correct and my solver is bugfree. At some point you cant check it by hand anymore
MSmits: sometimes i test a new heuristic and it works in 199 out of 200 games
jacek: so make moves to avoid that 1 game
jrke: yeah it can happen cause in some cases humans are better than bot and if you succeed to fix that then just greed more :)
MSmits: jacek I actually considered that, that might even be viable. But if you know exactly what causes the difference then you can also just improve the heuristic
MSmits: but sometimes the difference can be caused by many different thingst hat are all rare
jrke: my code in arena is taking maximum 17 ms for a turn
jrke: thats only in few turns else its < 8ms
MSmits: hmm yes, but it doesnt search to end of game when you're into the endgame does it?
MSmits: like at ply 55, does it always solve?
MSmits: ah ok, my idea is that it should always do that. So far I havent succeeded, but i am getting closer
MSmits: TT with graphs is complicated
jrke: my idea is bit different when i play a move in minimax i play that whole chain except last two boxes and then transfer it to next depth
MSmits: yes but sometimes it's better to take all boxes
jrke: with value addition of chain
MSmits: hmm also, when it's a loop it's 4 boxes that you cant take
MSmits: so thats why its sometimes better to just take them all
MSmits: especially when there are many loops
jrke: what does solver will give you move to play or just value?
jrke: or both
MSmits: for now just the value, but when i put it in my bot, i should adapt it to also give me the move that goes with the best value
MSmits: it's always a messy part of a minimax, it's just the top layer that needs the move
dbdr: MSmits, do you MCTS in Langton?
Marchete: Msmits do you have a good for loop definition for bitmaps?
Marchete: like for uint64_t bitmaps
Marchete: nvm probably you are logged off :D
darkhorse64: ctlz + ^ (1ULL<<bit) ?
Marchete: thx, I'm doing similar
Marchete: but I was looking for a more packed way
darkhorse64: I think MSmits does the same (base on his samples)
Marchete: for (int i = bE1.startLoop(); i >=0; i = bE1.Next(i) )
Marchete: I think I'm doing twice the count_zeroes_righy on Next()
Marchete: and I'm not sure about those +1 either
darkhorse64: Looks odd to me. I do this
Marchete: I'm trying to add all logic in the first line, to be similar to a for loop of a vector<int>
Marchete: I'm sure that leaving stuff on the end I'll forget about it
darkhorse64: I am not against abstractions provided they cost nothing
Marchete: I'll take that as a yes
jacek: like fizzbuzzEnterprise abstractions?
T1Anias: hey i completed the weekly challenge but didnt get the reward , is it because it had already ended that week?
dbdr: eulerscheZahl, you timeout in Langton
DomiKo: dbdr are you using MCTS?
dbdr: no, depth 2 minimax
eulerscheZahl: i'm not supposed to do that dbdr
eulerscheZahl: only on Bandas
eulerscheZahl: and ocean of code
dbdr: you still are
eulerscheZahl: IDE or arena?
dbdr: look at your last battles
eulerscheZahl: that's an early one..
eulerscheZahl: oh, lots of timeouts
eulerscheZahl: that's why val destroyed me
eulerscheZahl: CS1069: The type name 'Bitmap' could not be found in the namespace 'System.Drawing'. This type has been forwarded to assembly 'System.Drawing.Common, Version=18.104.22.168, Culture=neutral, PublicKeyToken=cc7b13ffcd2ddd51' Consider adding a reference to that assembly.
eulerscheZahl: my debug drawings don't work with .net core :(
eulerscheZahl: fixed. thanks for reporting dbdr
dbdr: what algo do you use eulerscheZahl?
eulerscheZahl: i think minimax, let me check
eulerscheZahl: minimax with iterative deepening
dbdr: do you ever get deep?
Illedan: Redstrike is using MinMax too
eulerscheZahl: did he say that?
Illedan: Unless he changed after I talked to him last time
Marchete: with Java?
Marchete: this is madness!!
dbdr: java can be very close to C speed, especially if you don't allocate
dbdr: is this any surprise? :D
dbdr: damn, it got harder to beat euler :D
eulerscheZahl: back to #8 where i was before fixing my crashing bot :D
dbdr: nah, you are #7, where you were before I pulled you down
eulerscheZahl: not much difference between crashing and non-crashing bot
eulerscheZahl: was a bit higher before, it's not only you who pulled me down
dbdr: well, you would have drifted to 90 eventually
dbdr: it just took time because of little activity and low variance
eulerscheZahl: down to 8 again :D
dbdr: maybe send back the crashing one? ;)
eulerscheZahl: nah, i'll send some food into me first
itzblinkzy: i am #1
itzblinkzy: say hi
dbdr: activity is back on Langton :)
DomiKo: yes yes
dbdr: looks like Illedan broke something
dbdr: TanishqMehrotra, no tone like that here
jacek: how're my female dogs?
dbdr: shut up jacek :D
dbdr: Illedan, look at this 1st half :) https://www.codingame.com/replay/485740988
eulerscheZahl: collecting blue cells
Illedan: My MinMax is only trying to use the first 9 empty cells in the path of the ant xD
eulerscheZahl: i have some pruning too. i guess we all have
Illedan: I guess my pruning is very bad
eulerscheZahl: no, there's not :(
jacek: huh?i have world all the time
eulerscheZahl: always shows the channel your joined last. but you can't leave and rejoin world
eulerscheZahl: i guess you could change your country in the profile settings :D
eulerscheZahl: the next contest will take place mid november
jacek: js is supported
eulerscheZahl: https://www.codingame.com/multiplayer/bot-programming most of these are old contests
eulerscheZahl: except for those "by XYZ"
jacek: so is ts
darkhorse64: You only promote if you are above the boss when all battles have been playe
darkhorse64: See, promoted right after end of submit
Ifthel: How do I tell CodinGame that I don't want my Java code slathered with the "final" keyword?
Ifthel: Actually, how do I tell CodinGame to stop giving me suggestions altogether?
Schwase: first contribution officially published as ready
Schwase: @GV doesnt show code :(
TheSpiffiest: What's the puzzle that's almost Coders Strike Back, but simpler?
DaNinja: Search Race
TheSpiffiest: ah that was it Thanks. I got to gold in CSB, but I need to work out some simple stuff to get higher I think