Chat:World/2021-07-24
MaliciouslyCrypticUsername: Can someone tell me if I typed out step 2 in the "expert rules" of CSB correctly?
MaliciouslyCrypticUsername: vx *= throttle*(cos(angle))
vy *= throttle*(sin(angle)) x += vx y += vy
MaliciouslyCrypticUsername: (I'm making a function to predict position ahead of time)
dbdr: I think there might be rounding involved
dbdr: the best is to test it
DaNinja: angle is in radians?
DaNinja: x = round(x); vx = (int)(vx * FRICTION);
DaNinja: you should add to vx not multiply: vx += cos(angle) * thrust;
MaliciouslyCrypticUsername: Oh
MaliciouslyCrypticUsername: No the friction comes after
MaliciouslyCrypticUsername: ok thx for the last part!
DaNinja: yep, friction and rounding come after
MaliciouslyCrypticUsername: http://chat.codingame.com/pastebin/c38fb52c-f2e2-4f79-ae70-d2da480d2728
DaNinja: looks ok
MaliciouslyCrypticUsername: Ty!
DaNinja: :thumbsup:
MaliciouslyCrypticUsername: I also made the input a list and the output a list so I can return the value into the function again and again to predict successive turns
MaliciouslyCrypticUsername: Inside a loop presumable
MaliciouslyCrypticUsername: presumably*
.Z: ;o
jacek: happy Caturday
Marchete: :smirk_cat:
Illedan: dbdr, what level was crashing?
dbdr: 30 (val 4) for instance
dbdr: can you reproduce Illedan?
Illedan: ye
Illedan: -.-
dbdr: :+1:
Illedan: thx
jacek: so can i, whats the deal
Marchete: deal or not deal?
struct: (╯°□°)╯︵ ┻━┻
IcyStar: Hello, I have a question about Coders Strike back. I noticed that the position of the pod passed as input (x,y) is not the current pos on the playscreen, but rather it is the previous pos of the pod. Is this normal ?
dbdr: IcyStar, I think it's just that the screen shows the state after applying your action for this turn
IcyStar: oh really ? I find that really disturbing..
IcyStar: I'm not sure if that impacts my code or not. I use previous positions of the pod, and so it makes me confused as to what is the previous..
Marchete: or maybe your are printing before reading the new data...
jacek: or servers are lagging
jacek: ohayo
DaNinja: ohayo~
Uljahn: konnichiwassup
jacek: oh my
dtiwari: Howdy everyone!
dtiwari: :slight_smile:
MSmits: such good news
MSmits: I wake up to read Illedan can reproduce. The world needs more little Illedans
dbdr: :D
MSmits: morning :)
dbdr: \o
Illedan: Lol
Illedan: I will fix stuff tonight dbdr. So keep finding things :)
Illedan: I'll move it WIP
Illedan: dbdr, do you solve the 1 food testcase within 25000 moves?
dbdr: nope
dbdr: not 1 and 2
Illedan: Good :)
dbdr: you? all but 1?
Illedan: Not 1,2 or 3 :( But my code is 80 lines long and I know what to do. Just gotta fix that viewer for some demanding user first :P
dbdr: 3 I solve with a bad score (7k)
Illedan: What is your total score on submit?
dbdr: 569074
Illedan: 60k more than me :D
dbdr: for 2 I had 508. so it's almost surely solvable
dbdr: 1 I have doubts
Illedan: I can increase the move count, but I think it is nice with a challenge like this
dbdr: no, it's definitely better if it's a challenge to solve some
dbdr: bigger sense of progression
Illedan: True
Illedan: Maybe lower it? xD
dbdr: why not?
Illedan: True
Illedan: 20k?
Illedan: 15k?
dbdr: yeah
Illedan: How many steps do you have on 3 now?
dbdr: 17850
Illedan: 18k?
Illedan: ok
Illedan: I can set limit to 17800 :smirk:
dbdr: :P
dbdr: deal
Illedan: I'll fix later. bye
dbdr: ttyl
seshoumara: how is my CoC score adjusted after one play?
jacek: probably trueskill algorithm
dGrayCoder: is there any tutorial for any turn based challenge?
dGrayCoder: I want to know how I can represent a game state in form of node so I would be able to apply some search algorithm.
jacek: try https://www.codingame.com/learn
jacek: like https://www.codingame.com/playgrounds/48392/bitboard-for-the-3x3-tic-tac-toe-game
dbdr: class State { ... } and a method State::update(action)
jrke: snake game is easily solvable but decreasing move is main motive
dbdr: yes, it's an optim :)
jrke: but i think atleast it shouldn't be that easy to solve
jrke: i can solve many just under simp logicle
jacek: nah, 2048 is easier to 'solve'
jrke: atleast width and height both should be odd to prevent that very easy solving
dbdr: you can't solve 1 this way ;)
dbdr: not enough steps
jrke: yeah 1, 2, 3
dbdr: number of steps will be reduced further a bit
jrke: cool then
Gabbinetto: i use python3 and this variable n=[int(input()) for i in range(3)] doesn't take the inputs and says its equal to none
Gabbinetto: what do i do
dGrayCoder: so for a turn based game like the 2021 spring challenge, how should I apply a search algorithm for finding the next move? Should I simulate opponents moves too?
MSmits: did you try search algorithms on simple board games before?
dGrayCoder: I have only done minimax on tictactoe
jacek: yes you need to simulate the game, make move and undo move
dGrayCoder: I have used a*, dijkstra on some other problems
MSmits: well spring challenge 2021 is not hard to simulate, but has some complications using search
MSmits: I used beamsearch, some used a form of mcts
MSmits: you can get to legend with either
MSmits: but it's not that easy to do
MSmits: thats why i ask about the simple boardgames.
MSmits: Beamsearch is doable for spring 2021 I think, but if you've never done mcts, dont start there
MSmits: (better on uttt)
dGrayCoder: utt?
MSmits: oh i thought you knew, you said TTT, but you meant normal TTT
MSmits: https://www.codingame.com/multiplayer/bot-programming/tic-tac-toe
dGrayCoder: can you apply beam search on ttt?
MSmits: it's one of the easiest places to practice getting a mcts working. The only issue with it, is that it's hard to see what a good move is (for a human).
MSmits: dGrayCoder no
MSmits: beamsearch is mostly used on optims
MSmits: it involves ignoring the opponent mostly
MSmits: and finding the best sequence of moves for yourself
MSmits: and since optims are single player, it makes more sens there
MSmits: sense
MSmits: in spring 2021 you can to some degree, ignore opponent. It's not perfect by a long shot, but it works somewhat
jrke: my spring 2021 is still without any opponent interference
dGrayCoder: what is optims?
MSmits: optimization
MSmits: get the most score out of the problem
dGrayCoder: I see
MSmits: https://www.codingame.com/multiplayer/optimization
dGrayCoder: thank you very much it helped.
MSmits: np
MSmits: if you see people talk about snake here, that's an optim in development
jacek: or it could be python
Illedan: jrke, true. I can make the width/height randomized with some odd maps?
Illedan: Would actually give something to the game
Illedan: Just +-1 of the current
Westicles: Anyone know the story on thor golf? Looks like new validator and maybe a bash update or something?
BlaiseEbuth: Another mean trick from the French...
Westicles: Yes, very devious :P
BlaiseEbuth: :3
seshoumara: where can I find the weekly puzzles?
BlaiseEbuth: Via the notifications, or by clicking the coresponding item on the quest map.
geppoz: question from noob to experts in ruby
geppoz: about if conditional
geppoz: example:
geppoz: { puts n; exit} if n>1
geppoz: that gives a big error
geppoz: puts n if n>1
geppoz: that works fine
geppoz: question: how is defined a block of code that needs to be tratead as a single entity?
JohnPotter895: anh em việt nam ơiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
BlaiseEbuth: No.
BlaiseEbuth: Speak english on this chan.
seshoumara: the quest map directs me to the current puzle of the week, how do I access the previous ones?
BlaiseEbuth: Puzzles of the week are puzzles already present on the site, so you can find them back by browsing CodinGame. But there's no history of which ones was pow...
seshoumara: there was a list some 1-2 years ago, guess they refactored it, ok, thanks
Pythonista_TN: hey guys, If a clash competition is complete can I have access to the problem statement?
frozen_flames: after you submit your code
KiwiTae: Westicles you got targetted again :p
Pythonista_TN: @frozen_flames yes
Pythonista_TN: after submission
Pythonista_TN: can I do that ?
**Pythonista_TN slaps around a bit with a large fishbot
pyprog: there is a bug running around inside my computer
pyprog: should i run a debugger?
Illedan: Yo
MrAnderson: yo-yo
struct: hi
Illedan: So, how should I create this API from Snake to JS :thinking:
Illedan: Send all the moves?
struct: Yes
Illedan: Send coordinates with food and snake?
struct: you can check chess or try angle catch
Illedan: Then I need the engine both places
struct: I would say send moves only
struct: yes
Illedan: -.-
Illedan: Chess has it both places?
struct: you dont need to check for legal moves on js
struct: because the java one will handle that
Illedan: Indeed
jacek: oO
Illedan: But I need to send the spawned food too
Illedan: Or should I keep the seed logic in JS too?
struct: its your call
struct: if you send the food its more data
Illedan: I guess I can send all moves as: N10W3ESS
struct: that should help yeah
struct: max moves is currently 25k right?
Illedan: Yeah
Illedan: Will be 17000
Illedan: And I wonder about having some boards with odd sizes
struct: if you just send the moves to the js module the replay size will be way under 40k
struct: with the module you can also have all the frames if you want
struct: instead of teleporting the snake
Illedan: I want to speed move :D
Illedan: When you do many each frame
Illedan: 50 moves? => 1 sec / 50 each move
eulerscheZahl: still waiting for approval Illedan?
jacek: lets approve without making the bot
struct: once i finish breakthrough i might make a bot
struct: I dont know what to play next anyways
jacek: still obsessed with breakthrough eh
struct: yes, its a fun game
eulerscheZahl: > I dont know what to play next anyways the private contest?
struct: whats the link?
eulerscheZahl: https://www.codingame.com/hackathon/sogeti-codefest-2021
eulerscheZahl: probably it's just fantasitc bits
eulerscheZahl: maybe some minor changes
eulerscheZahl: like one can only do ACCIO and the other wizard only FLiPPENDO? we'll see
struct: not eligible for prizes :/
eulerscheZahl: you have so many smurfs
eulerscheZahl: ask nulte
struct: I do some people in spain I guess it can be delivered there
struct: i do know*
eulerscheZahl: but first you have to reach top3
eulerscheZahl: that's not easy ;)
jacek: nulte is noob
struct: higher chances than on a multi
eulerscheZahl: you mean on a regular contest?
struct: yes
struct: sorry i havent rested properly due to breakthrough
eulerscheZahl: maybe the next contest will scramble chances of winning
struct: hope not
eulerscheZahl: ++
jacek: next contest? the non-multi one?
struct: 20 shirts top 20 + 20 shirts for top 100 would not be that bad
eulerscheZahl: the undetermined one
struct: but if its all randomized i would say no
eulerscheZahl: what are you talking about?
eulerscheZahl: i'm not even thinking about tshirts or whatever. just the format (bot vs escape)
struct: ah
struct: my answer remains the same :)
eulerscheZahl: fantastic bits had tshirts for top10 + 50 random winners in top250
martinpapa69: ye that was bad
eulerscheZahl: https://www.codingame.com/forum/t/fantastic-bits-random-t-shirts/2292
martinpapa69: i finished like 13 or something. no t shirt
eulerscheZahl: you got your fair share at this point :)
eulerscheZahl: when did you change your country from Hungary to Spain? but still Budapest
martinpapa69: I moved to spain so i can participate in the private contest
eulerscheZahl: really? :D
martinpapa69: might move back to hungary later idk
struct: moved or changed the flag?
eulerscheZahl: CG will never be able to disprove one of them
martinpapa69: yo soy espanol
struct: I guess ill use nulte
eulerscheZahl: considering your slow response time this could be a machine translation
martinpapa69: no I actually know this sentence :D
martinpapa69: duolingo...
eulerscheZahl: no ablo espanol
struct: you need this char maybe ñ
eulerscheZahl: that's an N with a dash on top
eulerscheZahl: while I want an N with a ~
martinpapa69: gonna tell them to send a laptop with spanish keyboard if i win
struct: yes its pronounced as nh
struct: español
struct: in portuguese its espanhol
eulerscheZahl: don't I want the nj?
eulerscheZahl: like sen(j)or
Astrobytes: yes
struct: maybe, my spanish inst great
struct: maybe im confusing it with portuguese
struct: Didnt know that google translate had multiple voices for the same language
jacek: i see he changed budapest to barcelona eh
struct: darkhorse64 how do you know that breakthrough may only last up to 209 moves? I can't seem to find any info about it.
jacek: :thinking:
jrke: maybe i think an estimated
jrke: like (8*8 + 8*7)*2
jrke: something like tha
jrke: t
Illedan: (╯°□°)╯︵ ┻━┻
Illedan: Stupid reverse clashes -.-
struct: you are doing clashes?
jrke: nope
Illedan: *clash
jacek: oO
eulerscheZahl: so csipcsirip = Marchet e confirmed?
RoboStac: struct - every move is going forwards so 209 would be moving all pieces onto row 7/2 before one is forced to move onto the end row and end the game
struct: thanks robo
Marchete: lol
jacek: oh my
Marchete: he changed to spain
jrke: thats how i tried to estimate - 8pawns at bottom can go to top in 8 moves which is 8*8 and 8 at second row will take 7 moves to reach top so 8*7 and for players *2, maybe i am wrong
RoboStac: only one pawn can do 8 moves
RoboStac: but yeah, 16*7+16*6 + 1
jacek: lets ask those who approve breakthrough how did they test that
jrke: yeah robo is right i think 16*7 = 112 16*6 = 96 =208 +1 = 209 if it fits
martinpapa69: he *moved to Spain
martinpapa69: I thought if I work on Marchete's code, it could be beneficial if i moved closer to him
MSmits: good thinking
jacek: then we all should move to canada
MSmits: to be near RockyMullet!
BlaiseEbuth: Caribou powa!
reCurse: Why would they limit the countries that can participate? I can't find a single good reason
MSmits: maybe this is something they leave up to the company that is sponsoring
MSmits: they may be hiring in some
BlaiseEbuth: Participate to ?
reCurse: So they wouldn't be interested in expat candidates? lol
reCurse: Not to mention people with no interest in a job with the right country can still do it
reCurse: Like I said, not a single good reason
MSmits: no idea, but I think it's not CG making this decision
reCurse: I don't care who's making the decision, I think it's a dumb one regardless who makes it
MSmits: possibly you are correct :)
BlaiseEbuth: :(
struct: its not like having people from other countries will affect anything, if they want to recruit people from the countries listed they can still do it
reCurse: Maybe CG is charging per country now
BlaiseEbuth: Thanks, chat log.
reCurse: Someone with a twitter should @ them
MSmits: a twitter?
jacek: you dont do twitter?
reCurse: No
jacek: :O
reCurse: I'm off social media in general
MSmits: I dont do any social media
BlaiseEbuth: same..
MSmits: I don't do any social tbh
jacek: yet you are in cg chat
MSmits: meh, ok 1 social
Astrobytes: that's different
dbdr: Illedan clashing instead of snaking? :thinking:
BlaiseEbuth: It's nothing social, you're all bots...
struct: he is learning javascript from clashes
MSmits: mmh i miss benchmarking
MSmits: no idea how much improvement i am making vs mikla unless i submit repeatedly
reCurse: Glad that stupid change didn't happen when I was still active
MSmits: I guess you can still do it with a large time between games
Wontonimo: what's the change ? (i'm late to the conversation)
Wontonimo: that is preventing benchmarking?
reCurse: 3-4 times longer than before
Wontonimo: ah yes, managing server costs
reCurse: They never mentioned that
Wontonimo: what did they mention?
reCurse: They said "so people participating in contests don't get slow servers because of people benchmarking"
reCurse: And we know what happened
BlaiseEbuth: Benchmarking performance is a 5% concern...
martinpapa69: ye, how many ppl use cgbenchmark in a contest, like 30?
reCurse: Probably more than that
MSmits: that's not negligible though
reCurse: The real issue is that if you prevent them from doing that, what do you think is going to happen
reCurse: More submits
MSmits: yes, thats whats happening for me now
reCurse: Then things practically came to a halt because of all the submits
reCurse: :clap:
Astrobytes: There is logic, then there is CG logic
reCurse: Anyway enough negativity from me today
BlaiseEbuth: Yeah that's really surprising, since CG always took smart decisions until now... :thinking:
**Astrobytes brushes off the salt
BlaiseEbuth: Throw it over your shoulder.
Astrobytes: MSmits: refactoring done?
Wontonimo: let's talk about PID controllers
Wontonimo: arn't they wonderful
Wontonimo: so many uses
MSmits: kinda
reCurse: Never used one
MSmits: still looking for bugs
Astrobytes: No improvement then
MSmits: hard to say
MSmits: i only have 1 opponent to test against
Astrobytes: Yeah indeed
Wontonimo: :puzzled-that-distraction-worked-but-no-as-intended-face:
Wontonimo: ^^ MSmits, if this image doesn't work, try that refresh control sequence again
MSmits: :rage:
Wontonimo: bwahahahah
Astrobytes: refresh control sequence?
Wontonimo: ctrl-shift-W to load the alt emojis
Astrobytes: lol. No, no I won't do that.
Wontonimo: and refresh them
Wontonimo: haha, good for you
Astrobytes: Did you fall for that MSmits?
MSmits: :blush:
jacek: o: https://img-9gag-fun.9cache.com/photo/aV79OVv_460sv.mp4
MSmits: why the cubes
jacek: hexes are so last week
MSmits: i see
martinpapa69: I expected the girl from the circle at the end of the vid at least
jacek: girl?
MSmits: win first 55 games in submit, including 2 mikla, but i know it's luck
MSmits: 3 mikla now
MSmits: what a submit
martinpapa69: https://d13ezvd6yrslxm.cloudfront.net/wp/wp-content/images/Scariest-Scene-in-The-Ring-700x300.jpg
MSmits: the ring :P
MSmits: not circle
MSmits: dude
Astrobytes: Nice MSmits
martinpapa69: ye, *the ring
MSmits: and she's called Samara
MSmits: 75% still no loss, 5 wins mikla
MSmits: ah damn 1 mikla loss
MSmits: i have no clue what my real winrate is vs him :(
Astrobytes: Did you also add the long chains thing? Or this is the same bot?
MSmits: i did something different again with disjointed areas and nimsums
MSmits: to find more bad moves and exclude them
Astrobytes: Ah nice.
MSmits: I think it works, but no idea to what extent
Astrobytes: 1 loss v Mikla?
Astrobytes: Oh boom, 2 more
MSmits: 6-4 now yeah
MSmits: if i 6-5
MSmits: if i am even
MSmits: i need to submit spam
Astrobytes: Oh I didn't see the first one
struct: download contribution change time contrainst get rank 1
MSmits: http://cgstats.magusgeek.com/app/multi-dots-and-boxes/MSmits first time all green
struct: ezpz
IconicKid: i dont get this
dbdr: gg MSmits!
Astrobytes: You seem to have cleared up the losses due to score
MSmits: I think there is a pretty hard limit to what you can do with nim solving and i think I have reached it just like mikla
MSmits: hmm
Astrobytes: (against the non-miklas)
MSmits: oh right yeah, i think i had bugs
MSmits: but they were rare anyways
IconicKid: im literly so confused
MSmits: IconicKid if you have a question, you can ask it
Wontonimo: what are you working on IconicKid
MSmits: oh, btw thanks dbdr, but not at all confident yet. Think this submit was lucky
Wontonimo: if anyone cares to download yet another Caesar cipher https://www.codingame.com/contribute/view/7162f264cc6630afc2954404150b1f7856de
IconicKid: how do i do this
Wontonimo: do what?
BlaiseEbuth: this
Wontonimo: oh!
IconicKid: im working on coders strike back'
struct: http://chat.codingame.com/pastebin/56f40ce8-9211-4677-8855-8eabf8610c95
struct: (╯°□°)╯︵ ┻━┻
Wontonimo: oh, i'll PM/DM you, i'd be happy to help wih CSB
NovaGrindge: I am failing at this horribly
jacek: oO
antiwonto: you too eh NovaGrindge
Illedan: dbdr, I had 5 min and wanted a simple clash :P Fixing Snake now.
Wontonimo: lol, i didn't mean to txt from my alt account
Wontonimo: what are you having trouble with NovaGrindge ?
dbdr: you're forgiven then ;)
jacek: nyoro~n
Necromonkey: = "";
Illedan: dbdr, do you think odd sizes give any value?
dbdr: I think it would annoy me
dbdr: so yes ;)
Illedan: haha
Illedan: 29x19 always then?
Illedan: Or maybe 31x21 for more space :D
MSmits: for more annoyance: 33x21
Illedan: Why?
Illedan: 33 worse than 31?
Illedan: *in magnitudes that is
dbdr: power of 2
Illedan: ah lol
Illedan: ok 33
asdfg47: 3 playing Coding Escape live
dbdr: thanks MSmits :rage: ;)
MSmits: I know how to annoy
dbdr: Illedan size will be random?
Illedan: Nah, I'll make it 33x21
Illedan: hard enough
Illedan: Unless random is far better :thinking:
Illedan: but then I need to send dimensions to the damn viewer too
Illedan: hm
Illedan: no I don't
Illedan: I only need the seed
Illedan: What range should I make it?
Illedan: 29-33 x 19-21?
struct: range brings nothing but pain imo
dbdr: it forces to make a more general solution
Illedan: yeah
Illedan: I like it
Illedan: :evil:
Illedan: https://imgflip.com/i/5higx5
Wontonimo: what happened to hyperbole and a half?
struct: what does that mean Wontonimo?
jacek: sounds like 2 hipoboles
SlowedFever: is there a way to finish your code in a clash after time runs out?
jacek: no. when clash is over its over
SlowedFever: thats annoying i want to keep working on my code
jacek: try puzzles, you can go back to them whenever you like ~
SlowedFever: yeah clash is kinda dumb because how are you supposed to learn how to do stuff if it doesn't give you enought time to learn and finish your code
Astrobytes: It supposed to be for fun
Astrobytes: *It's
BlaiseEbuth: You're not supposed to
jacek: :head_bandage:
Astrobytes: You'll learn more doing puzzles as jacek said
SlowedFever: what's the purpose of fun
BlaiseEbuth: :thinking:
dwarfie: :popcorn:
MSmits: fun is selfpurposing
MSmits: the purpose of fun is fun
SlowedFever: correct answer but it's still useless
jacek: fun is to pull rebless down in checkers
MSmits: what do you need SlowedFever
Astrobytes: Bet you're the life of the party
MSmits: hi everyone, I hate fun!
jacek: you are not from germany, hmm
SlowedFever: I like fun but I thought this website was a learning tool
MSmits: it is both
BlaiseEbuth: It's not
MSmits: it's a learning tool and a funning tool
BlaiseEbuth: lier
jacek: it is more that clashes. it has puzzles, optims, multiplayer games
Astrobytes: It is if you want it to be. It also happens to be fun.
dwarfie: clashs are useless ....
SlowedFever: it's not that hard to let players finish their code after time runs out
SlowedFever: that would make them better
Astrobytes: DO PUZZLES THEN
BlaiseEbuth: I want it to be a pronographic site !
BlaiseEbuth: Now
BlaiseEbuth: I'm waiting
jacek: mods are asleep, postponies
MSmits: that's a 5% issue BlaiseEbuth
Astrobytes: Post prawns
BlaiseEbuth: -_-
SlowedFever: I'm just saying that they could improve the way clash works
BlaiseEbuth: Improve in your view
MSmits: SlowedFever improve how
dwarfie: clashs worked .... 5 years ago ....
BlaiseEbuth: But since clashs are made for fun/competition and not to learn they will not
SlowedFever: I don't see the downside of letting players fix their mistakes after the round has ending
KiwiTae: SlowedFever just do very easy puzzle categories instead
MSmits: SlowedFever yeah that would not be a bad idea
Astrobytes: Never clash [solved]
MSmits: but personally I feel that if you want to learn, do real puzzles
MSmits: as has been said
SlowedFever: yeah and now i know that
jacek: clashes or one time things. you can 'return' to it using unofficial search like https://eulerschezahl.herokuapp.com/codingame/puzzles/\
jacek: and i cant type properly
KiwiTae: SlowedFever like this one https://www.codingame.com/training/veryeasy/nintendo-sponsored-contest
BlaiseEbuth: CG like good ideas, you should propose it to them
Astrobytes: sigh
BlaiseEbuth: :speak_no_evil:
struct: This game is driving me crazy
Astrobytes: Take a rest from it for a few days man
struct: I cant
MSmits: he cant
Astrobytes: I noticed :)
struct: I still think my eval can improve by so much
struct: But im so bad with it
jacek: been there done that
BlaiseEbuth: Was about to say that's out of board...
MSmits: you cant be expected to understand every error, thankfully I am here
BlaiseEbuth: :thumbsup:
MSmits: mmh my D&B bot is now 3.7k lines
MSmits: waiting for the limit to hit :P
struct: o.o
MSmits: I put in an mcts
MSmits: for early game
MSmits: I was hoping it would beat mikla, but I think he's immune
struct: im also struggling to improve on bt now
MSmits: you'be been struggling all week
MSmits: like me
struct: tomorrow ill try one more thing or two then im done
struct: i improved it two days ago
MSmits: not sure i believe you :P
MSmits: I remember STC
struct: I havent spent much time on this game though
struct: I started it like 1 week ago
MSmits: only a week?
MSmits: ah I guess this seemed like a long week
struct: i would say 1 week and half maybe
struct: i tried to do the early win check with shifts only but its not easy
struct: I can do it easily if its the current player turn
MSmits: so you need flips then
struct: but if a player makes a move and opponent still has a move its harder to check for a win
MSmits: ahh ok
struct: currently I can find a win from 6th row for any player
struct: if its his current turn
struct: its quite simple
MSmits: ah I see
Drichie: Is it ok to ask questions here?
MSmits: it is ok
Drichie: Currently I am doing the Horse-Racing Duals practice puzzle. I have 2/3 of the testcases working. The last one has inputs 99,999 which makes my program time out. I'm guessing it is because I am doing too many comparisons and it is slowing it down too much yea?
MSmits: yes
MSmits: you need to make your algorithm smarter
MSmits: so that it has to do less comparisons
Astrobytes: even a pre-solve mcts didn't work eh
MSmits: preferring loops and short objects didnt
MSmits: i reversed it to prefer long chains now
MSmits: maybe that helps, i dunno
Astrobytes: that was the idea earlier
MSmits: thats what my current submit does
Astrobytes: looks OK but not your best
MSmits: the thing is, from checking all the lost games it's clear to me that he doesnt solve better than i do,
MSmits: not anymore
MSmits: the gamestate just favors him somehow, more often
MSmits: when it's time to sovle
MSmits: so he must do something in ply 1-35 that gives him some gain
Marchete: just spam submit and you are 1st
MSmits: no, i dont think so
MSmits: this is a lucky submit again i think
Astrobytes: Well yeah, definitely something going on there.
MSmits: but sure i'll spam a bit to be sure
Astrobytes: But what, I have no idea
MSmits: I think the submit is pretty good. Hamburger got me twice, but other than that its fine
Astrobytes: Yeah I saw that, was on score?
MSmits: 1 on score
Astrobytes: Didn't look at the replats
MSmits: 1 on nim
MSmits: he does a nim search
Astrobytes: *replays
Astrobytes: Aha
MSmits: even if i solve it sooner, if i solve it as a loss and he makes no mistakes...
MSmits: he still wins
Astrobytes: Captain Obvious enters the building :P
MSmits: yes
MSmits: btw, i just noticed his nim solver is really good
MSmits: he solves only 1 turn later
MSmits: so he's not that far from me and mikla
Astrobytes: Far enough it would seem
MSmits: but as we discussed, you need to know how to prefer splitting moves and such
Astrobytes: Indeed
MSmits: there we go again, another submit. Last one was 5-4. Keeping score now to see what my real WR is
Astrobytes: That's gonna be a bunch of games :P
MSmits: well only if it's close
MSmits: if i get to 10-30 i dont think I need more
Astrobytes: Fair
MSmits: I still have a solver search for pre-endgame on my list. Not sure how that'll be squeezed in my already 3.7k bot
MSmits: might have to use minifier
Astrobytes: Golf it
MSmits: mmh I guess i could rename some of those awfully long functions
MSmits: Link* ShortestGroundedLongStringWithNoShortAndNoDouble()
struct: You get no points from winning vs the other players
struct: it will be hard to pass miklla
Astrobytes: jesus effin christ
MSmits: its a very descriptive name :P
Astrobytes: You know you can just put comments and strip them for the submit right
MSmits: sure, I suppose
MSmits: struct if i have > 50% it's not hard. Just takes a lot of submit. But I probably do not have >50, thats the real problem
MSmits: I should just set up a cg bench with a long time between
MSmits: is more efficient
struct: i think the new version takes the current constraints into account
MSmits: oh didnt know there's a new version
struct: I think there is
struct: it was updated around when this change was made
Astrobytes: Still a lot slower though so...
MSmits: dont need more than 100 games
struct: maybe ill also use it for bt
struct: anyways im off for today, need some rest from this game
struct: gn
MSmits: good idea struct, gn
Astrobytes: gn struct, take it easy man
theycallmedavid: Today I learned that it is possible to place flags in minesweeper
MSmits: :cheer:
MSmits: :cheers:
MSmits: blah
Astrobytes: :beers:
Astrobytes: :raised_hands:
MSmits: I set it to use 60s between games
MSmits: hope that is ok
Astrobytes: I thiiiink that's about right, but I don't know what the limit is per hour
MSmits: if you happen to come across a mikla one of these days, ask him if he does anything special early game
Astrobytes: I will do. Didn't see him for a bit
MSmits: same
MSmits: I noticed that he's somewhat immune to my making loops and such
MSmits: so maybe he's also doing a mcts to make long chains
Astrobytes: Yes, it seems he's already thought of it one way or another
MSmits: yea
dzosz: any1 knows is CG already has a puzzle that is about finding the smallest integer that is NOT in the list?
Astrobytes: sounds clashy
Astrobytes: MSmits: I'm off man, gonna watch a movie or something. Good luck and take care
Marchete: Smits maybe he is OK to answer all that
MSmits: yeah might be
MSmits: gn Astrobytes
MSmits: WR seems around 40%
MSmits: 48 games in CG bench so far. 39% won
MSmits: submit: 16-19 so far
MSmits: mmh though on cg bench i had 2 crashes
MSmits: allright thats clear enough I think. 47-70 score
Marchete: who is who?
Marchete: 47 you?
MSmits: yes
Marchete: too bad
MSmits: yeah, but it's really close
MSmits: this nim game is a bit all or nothing
MSmits: I'm right between all or nothing
MSmits: in that very arrow space
MSmits: narrow
Marchete: can't you make some kind of endgame book?
Marchete: taking chain properties and such
MSmits: nah
MSmits: chains dont matter
MSmits: not in the nim game
MSmits: anything long is immediately removed from consideration
MSmits: all that matters is divide and conquer
MSmits: only safe moves and sacrifice moves (1 box and 2 box moves)
MSmits: you use a hashmap to store regions and their values
MSmits: best way to look at it is just pick a replay like this one: https://www.codingame.com/replay/571897632 and check when the solve happens, frame 35
MSmits: it looks like nothing you can easily use tables for
A0-poppy: 求带