I_love_CoC: hello everyone
I_love_CoC: hi antiwonto
Automaton2000: but i still have a bug
eulerscheZahl: thanks ClockSort (didn't see it yesterday). in round 3 my journey will surely end, but the tshirt is nice
jacek: oh my
jacek: :taco: Automaton2000
Automaton2000: but how would you do it?
Drmewzih: i hate short mode
jacek: use python/ruby [solved]
Drmewzih: how the output can be 0
Cumpanasii: Hi! If anyone here solved the nuggets numbers problem, can you please tell me validator 6 numbers? I can't get past it
5DN1L: Even if one has solved the problem it doesn't mean one has access to the validators.
wlesavo: euler is it your first round 3?
eulerscheZahl: yes and no. i qualified before but had no time on that specific date to participate
Cumpanasii: it seems that 2 guys in the discussion page posted other 2 validators data, I figured that it would be possible. But thank you for your answer.
5DN1L: I have access to the validators
5DN1L: The 5 numbrs in Validator 6 are all less than 20
Cumpanasii: oh well, thanks for that
5DN1L: 2 are less than 10, 3 are more than 10
5DN1L: try some random numbers to test your code
Cumpanasii: Alright, thank you!
wlesavo: euler grats
eulerscheZahl: thanks. i had a hard time yesterday. took me 30min to convince myself that i should start coding
eulerscheZahl: hay fever and boring/tedious tasks
Mortis_666: is it normal that i have access to the best solutions of the event puzzles that i didnt solve?
wlesavo: and im storming codebusters legend, around 600 LoC ago it seemed to be quite close, and now it is still really close :smiley:
5DN1L: I think it's been the case since the first of such events, Mortis_666
Mortis_666: oh ok
Manchi_o6o7: Has someone a tip how to transform a regex into subexpressions, so I can form an NFA ?
eulerscheZahl: i was spamming submits for codebusters until i promoted. once i was 0.01 behind the boss but didn't have the patience to wait for someone else pushing me on a mostly inactive game
wlesavo: even without ejects i was in around top 5, but since then most improovements dont gave much, and i also had quite bad wr against boss
wlesavo: i even sometimes get above the boss on submit, but still
struct: did royale already have a NN for onitama?
_Royale: struct: yes :)
wlesavo: yaaay finaly got this codebusters legend, it took quite longer than it should have
eulerscheZahl: it took so long, the game even changed its name
wlesavo: my old python submits actualy were at the times when the game had different name, so that basically true :slight_smile:
Darleanow: Hello every1
Napstablook_: Hi Automaton2000
Automaton2000: im sure it will be different
Washier: evening peeps
Washier: I like C#. I know it is not C
Washier: Any tips making my C# faster? Mr EulersxheZahl?
jacek: get faster computer
Washier: Mr jacek, even you have to admit it is not all about language choice? Or is it. How deep do you search on Checkers for example?
jacek: well c/c++ can be few times faster than java. dunno how c# can be optimized
jacek: and my checkers bot is mcts-like, depth is irrelevant
jacek: depth around 11-12 of most visited move. but its different from alpha-beta
jacek: hmm you can use intrinistrics in c#
Washier: thanks. i search 6 deep in my C# checkers bot. Mr Frog does so well in board games, curious if it is lang tricks or better search. obsessed with board games again
Washier: i do
jacek: material eval only?
Washier: try my utmost to make game engine fast
Washier: no, in checkers a bit more complicated eval, but obviously not good enough, or searching too shallow
Washier: same thing in othello. guess it is just talent
jacek: even the same rank ;o
Washier: bandas and breakthrough also fun. yavalath next.
eulerscheZahl: checkers: in early game I get depth 8 for my alpha-beta (about 60-70k nodes visited)
eulerscheZahl: pretty standard implementation i would say. no bitboards or other fancy tricks, going the OOP route
eulerscheZahl: zobrist hashing for duplicate states
eulerscheZahl: even ignoring hash collisions entirely
Washier: zobrist is a good next step for me. sehr gut
MSmits: yavalath is pretty hard to sim
MSmits: minimax works well though
MSmits: mcts only after a lot of modification
Washier: MSmits is the yavalath expert everyone.
MSmits: the problem with yavalath is mostly the insane amount of branching early game
MSmits: yeah I spent too much time thinking about this game :P
jacek: thats what you have opening books for
MSmits: yeah, this helps a lot, but most of your opponents don't use them, or a very small one Washier
Washier: I have seen that yes. I branch very coservatively.
jacek: prune array eh
MSmits: as for language choice, the simpler the game is, the more important performance becomes
MSmits: it's only the simplest boardgames where it is the deciding factor I think
Washier: i worked hard on an book for othello, didn't help much, but that might just be me
MSmits: I got 1st with a book. Counterbooked 10 players, won every game
MSmits: in othello
jacek: and made recurse rage quit
MSmits: I took it off
eulerscheZahl: i got first at code keeper with a book
MSmits: I did not mean to ruin people's game. He had just trained a neural network that beat everyone and I counterbooked it in half a day
Washier: it seems having a book is one thing, having a good book is another. my books suck
jacek: and now nyannyan owned everyone
MSmits: with a book also?>
struct: I was also first on c4 with a book a long time ago
struct: this was my book
struct: 1 << 9;
jacek: yes. his twitter was/is filled with rubik cubes and othello's algorithms
Washier: <bow> struct
MSmits: ahh I might go back to othello some day. too busy with oware now
MSmits: 37 seed book almost done, fingers crossed for it not having errors
MSmits: minimax bot?
Washier: I just love getting the board game guys talking. I am not that good but find board games extremely interesting. wish i could operate on this level
MSmits: We're all good at different things though
MSmits: I can't do NN very well, haven't made a serious NN for one of these codingame arenas yet
jacek: probably n-tuples and mpc
MSmits: but I've gotten good at solving games and endgame books
struct: Im good at wasting time shaving nanoseconds
eulerscheZahl: i'm good at idling on chat
MSmits: yeah i can also do a performant bot, though struct is better at this
MSmits: he can use AVX better
MSmits: eulerscheZahl can basically do everything, though he does not specialize. He can make a bot for any game in a few hrs
jacek: i stopped caring about sim speed that much since NN takes over 90% time. though in onitama its still relatively small so i could use faster sim
MSmits: yeah i stopped caring about it mostly as well. But I have kept the habits from when I did, so most of them time the bots are fairly optimized
struct: I should stop caring too
struct: since my bots with 10% of the speed perform nearly the same
MSmits: if it's no longer fun for you. That's the most important part
Washier: I also really love evlutionary algo's. have had some success with the. anyone ever use them?
Washier: struct. you are a hero
MSmits: many used them here. I didn't. I tried SA though, it's simpler
MSmits: GA has more params
Washier: have used SA before too. also lovely
MSmits: yeah I like it
jacek: GA on search race :heart:
Washier: isn't gow you formulate(frame) the GA also very important?
struct: I dont know GA
Washier: you can do it struct
MSmits: Washier you mean what you consider a solution and how you mutate it, what you cross etc?
MSmits: SA has the same issue in deciding what to random
MSmits: only less choices to make
jacek: GA in action https://streamable.com/a52fbn
MSmits: looks cool
MSmits: are these visualizations easy to make?
jacek: its qt. i just used paintEvent loop and draw rects, circles and lines
j4at: google it lmao
MSmits: googling 2 characters is not advisable :)
jacek: c++ gui framework
j4at: its crasosplatform
jacek: lul, youll be surprised
MSmits: ah ok
MSmits: ohh it finds it on google with just qt
jacek: similarly how in java i used swing's paintComponent for this https://streamable.com/qgza5b
Wontonimo: antiwonto needed a reboot. it's back
Washier: can vouch for qt
antiwonto: [auto] ':taco:' was defined as ' A currency to some, sustenance for others '
jacek: :taco: MSmits
antiwonto: [auto] jacek has awarded MSmits 4 tacos. MSmits now has 7 taco. jacek now has 18 taco
MSmits: wow 7 already
MSmits: thats my endgame book 37 seeds progress
MSmits: ever 2 hrs 10-20 mins it outputs a new iteration. When 0 are different book is done
MSmits: not sure how long left. Maybe 30 or so
MSmits: so... half a week
MSmits: then 38 seeds ::P
MSmits: I am stopping there. Need a whole different method, as 39 is around 35 GB.
Washier: what is that please?
struct: MSmits on smitsimax for csb do the trees share the same eval?
MSmits: endgame book for oware. 37 seeds means 37 seeds left on the board. I calculate the net game result when played from each positition
MSmits: there are 22 billion different boards with 37 seeds on them
MSmits: stored at 7 bits per board, so 19 GB
MSmits: The same score is backpropagated for each tree, so in that sense they share, but the blocker contributes differently to the eval than the runner
MSmits: the reason i backpropagate the same score is that if the runner does something good, that may be because of something the blocker did
struct: I see
struct: I might try checkers
struct: It seems its easy enough to make a sim
Washier: doo eet
jacek: is it though, with loop jumping
Washier: yes, that is the tough bit
jacek: you may want to read on cake development http://www.fierz.ch/cake186.php he finally used some n-tuples into his engine
struct: The board fits in 32 bits right?
MSmits: 4* 32 bit
jacek: black, white and kings
MSmits: oh, I always used 4, did not occur to me to do it that way
Washier: you need 2 x 32-bit boards
Washier: 3 really to mark kings
MSmits: yeah, I'll remember that for when i revisit
jacek: well i use 3x64bits. actually nothing fancy in my bitboard, except it isnt 2d array
MSmits: I used so much fancy that I dont know how it works anymore. Did not help me much though
Washier: same. 3 uint32's
MSmits: my sim has lots of intrinsic stuff, pdep and pext and all that
MSmits: eval matters a lot more here
Washier: eval speed or eval quality?
MSmits: I just counted kings and men mostly
MSmits: Also, I used mcts with ept, maybe a plain minimax would be better.
struct: feels similar to bt
Washier: i used plain minimax.
struct: but on bt its easy to check if a player won
jacek: and before that
Washier: quite similar to bt for me, use same bitboard construct
MSmits: eulerscheZahl with the hungarian method, can you square a matrix with both lack of agents and lack of goals? What I mean is, can you add rows to make it square, but also add columns to make it square?
Washier: that is some high level maths i used to know
jacek: hexagonal matrix :drooling_face:
Washier: likr yavalath
MSmits: yavalath is high level math?
MSmits: oh hex
struct: jacek in which games have you implemented ntupples?
jacek: othello and yavalath
MSmits: it being hex gives it 12 symmetries, as opposed to 8 with a square game
jacek: and c4 at first, but now its nn
MSmits: nine mens morris has even more symmetry. 16 if i remember correctly
Washier: i'm the idtiota in this coversation, but yavalath works pretty well with the board representations discussed so far. its just funny
MSmits: whats funny?
Washier: not funny, diffreent
MSmits: yeah I don't know how to work it efficiently really. I just do pext stuff to extract winning hexes and such
Washier: your brain works differently Mr Smith.
Washier: very interesting
Washier: *Mr Smits
eulerscheZahl: yes MSmits you can also add columns
eulerscheZahl: if it's in the context of PR2: add cells multiple times as a target but with decreasing score for having more units
MSmits: yeah my brain is weird
MSmits: eulerscheZahl should you explore with multiple pods then?
MSmits: i was just thinking of sending a 1 pod to explore
MSmits: per target
eulerscheZahl: what's a target? direct neighbors only or a few steps ahead?
eulerscheZahl: but multiple units can still be beneficial for fighting
MSmits: yeah i am not sure yet, for now i will just consider anythign a target but those further away from base are worth less
MSmits: yeah but if you have multiple units left over, you can also send them toward opponent base
eulerscheZahl: not going to argue about that now, i feel sick
eulerscheZahl: codejam was a torture yesterday
MSmits: I am not arguing. No idea how to play this game yet. Just trying stuff
Drmewzih: go take a nap :)
MSmits: ahh sick from codign?
Washier: drink tea. dress warm for bed
Washier: good night
MSmits: yay working bot in pr2. I was worried the hungarian method would be too slow. But no timeouts.
MSmits: oh nvm spoke too soon, just spotted one =)
MSmits: 106 by 106 matrix probably not a good idea
struct: MSmits are you using bitboards on it?
antiwonto: [auto] Hey struct, here is a :taco: for loggin in today while it is quiet. You now have 25 tacos
MSmits: no struct
MSmits: I am also just using C#. Not really concerned with performance
Astrobytes: are you 1st yet?
MSmits: I am 792th
MSmits: I have a lot left to do
Astrobytes: hi Washier! :wave:
MSmits: the hungarian method works, but scoring function is crap
MSmits: i need defense vs rushes
Astrobytes: ah right, yeah the rushes
Astrobytes: there are ways of countering it but I don't remember much tbh
Astrobytes: I figured I would do codebusters since I'm fresh from that last contest and it's similar
MSmits: better algo for guessing where the opponent could have sent his bots. I am just using floodfill now, but if you record the entire vision history, you can at turn 100 confirm that between 50 and 100, no opponent bot has visited a certain hex, which means you should not floodfill there
MSmits: once they visit, the owner changes
struct: whats this hungarian method?
MSmits: how to assign X agents to Y tasks
MSmits: using a cost for each combination of X and Y
MSmits: never done codebusters
Astrobytes: it's not my favourite game, put it that way
struct: me neither
struct: and I never will
Astrobytes: but it's very similar to last contest
MSmits: did you not like last contest either?
struct: maybe you can reverse the seed there?
struct: no I did not
Astrobytes: neither did I
Astrobytes: I forced myself to play it
Astrobytes: since they don't come along so often
struct: yeah only reason
struct: ifs and fog are not for me
struct: also infinite search space im not good at that
MSmits: ah, I like fog when it's on a discrete playing field
MSmits: like in oceans. And I like it in pr2 also
MSmits: but yeah i dont like it in infinite seach space
Astrobytes: yeah it's OK in discrete spaces
Astrobytes: OOC was fog done right
MSmits: the whole game was about the fog
MSmits: pacman also had fog like this. I did not mind it there either, but the game was hard to play anyways
Astrobytes: didn't like that
MSmits: when you have fog, it has to be an important feature. If it is a nuisance, then it's better not to have it
struct: Did this game even need fog?
struct: maybe it needed, who knows
MSmits: you mean the contest?
Astrobytes: what , the spider game?
MSmits: I don't think it needed fog
Astrobytes: No it didn't
Astrobytes: usual CG attempt to make it heuristic-based
Wontonimo: The fog sure makes it easier for newer contestants to climb the leaderboard and limit old-timers who have a whole harness made from just destroying everyone quickly (i think, if I'm being generous)
Astrobytes: Yeah that's what I was getting at with my last comment
MSmits: I had to read that 3 times Wontonimo
MSmits: I thought, why would they make a harness from the corpses of other contestants?
Astrobytes: You have a distracted mind MSmits
Astrobytes: + disturbed, evidently :P
MSmits: maybe yeah
Astrobytes: ohhh btw MSmits: https://www.youtube.com/watch?v=FdqcUTNHwyo
Astrobytes: Stephen Colbert plays DnD
Astrobytes: (it's awesome)
MSmits: holy sht
MSmits: thank you!
Astrobytes: courtesy of Scarfield. The whole channel is amazing btw, long videos though but amazing
MSmits: nice find
Astrobytes: yeah for sure, I feel my dnd urges coming back :D
Wontonimo: why did they go away?
Astrobytes: time, lack of other interested people to motivate me, life, etc
MSmits: I still play dnd, 2 times a year or so. By the time we finish a campaign, there's a whole new version of dnd out there
MSmits: same guys that I played with 20+ yrs ago basically
MSmits: except back then we did it every 2 weeks
Astrobytes: you can do it in person now at least
MSmits: yeah i did a month or two ago
Astrobytes: remote is OK but so much better when you're all physically there
MSmits: was fun, though I would not mind just playing other types of games that don't require this much preparation
Astrobytes: my mate is really into Vulcania, sent me the manual and everything but I'm lacking motivation recently
Astrobytes: ^ from there
MSmits: looks fun, yeah thats the sort of game i meant
MSmits: I liked the Munchkin board game
Astrobytes: he got free stuff from them for something, I can't remember why
MSmits: ah nice
Astrobytes: the Plague Inc. board games are good fun
Astrobytes: hey Vulcania can get very deep
Astrobytes: Character creation is almost it's own manual
MSmits: Yeah, I like it to be a bit simpler if I am only going to play twice a year
0xhotdawg: anyone tried socket programming with c?
MSmits: I don't even know what socket programming is
0xhotdawg: network programming
MAWAAW: ye long time ago when was student
MSmits: ah, well I haven't
MAWAAW: TCP / UDP :D
jacek: good thing with UDP jokes is that i dont care if you get them or not
0xhotdawg: im interested in learning socket programming, which language u think is best for that? or hsould i just start with c for example?
jacek: socket sounds low level, but id start with python
struct: everything starts with python
0xhotdawg: oh interesting
0xhotdawg: u were able to do like a server app with python?
struct: I dont understand how people learned how to code before python was created
Astrobytes: MSmits: there's always Hero Quest!
Westicles: search for traps!
MSmits: whats that again?
MAWAAW: ((x^2)+(9*(x+y))) derivative x give 2x + 9y + 9 right ?
MAWAAW: to be sure
MAWAAW: then the test case 4 is wrong
MSmits: i did not try that puzzle.
jacek: i feel so indifferent
MAWAAW: lul UDP on you
Westicles: MAWAAW nope
MSmits: whats the forumula to calculate the derivative of MAWAAW?
MSmits: for testcase 4
MSmits: what do you think the answer should be?
MSmits: you have to evaluate it MAWAAW
MAWAAW: 2x + 9y + 9 cuz derivate x not y
MSmits: the output is an integer
MSmits: not a formula
MSmits: apparently the answer is 11 for that testcase
MSmits: I am guessing x = 1 and y = 0
MSmits: and you get this input
MSmits: which means 2 * 1 + 9 * 0 + 9 = 11
MAWAAW: no actually it's x 1 y 2
MAWAAW: result 29
MSmits: let me check
TINOUAINANI: clash empty
Astrobytes: If you join it they will come.
Astrobytes: (party time, excellent)
MSmits: MAWAAW i cant see what you are doing wrong, still looking
MSmits: Perhaps the testcase is indeed wrong and the validators aren't. Then you can still solve it
0xhotdawg: anyone of u guys doing ctfs?
Astrobytes: a few people on here do
0xhotdawg: im actually a beginner and i wanna join an active team
Astrobytes: discord might be the best place to get people who are into it tbh
0xhotdawg: yeah true
0xhotdawg: i know afew ppl that r doing it but you know, they either not active or actually too good to accept me
Astrobytes: yeah, best to ask for others near your level
Astrobytes: or if you're good in one particular aspect, then ask for people who need someone in that role (reversing, crypto, cracking, whatever)
MAWAAW: ye it's wrong seems like it's a new puzzle
Westicles: nobody knows how partial derivates work apparently...
MAWAAW: what do you mean
MAWAAW: it's a x derivative not x and y derivate
Astrobytes: Is this an old puzzle or it was just in approval state for a very, very long time? It looks familiar
MSmits: ah yes sorry i missed that MAWAAW, Westicles is right
MSmits: I read it as x*y for some reason
Astrobytes: the puzzle seems fine from looking at it
MSmits: derivative is 2x + 9
MAWAAW: oh ya I got it my bad
MSmits: 2*1 + 9 = 11
MAWAAW: it's like a constant -_-
Astrobytes: Kinda clear from the statement ;)
jacek: ask away
Again, this is for Ghost Legs (easy). Really just trying to understand why the console output gets screwed up on Test Case 06, because it's clean on all the other test cases.
Wontonimo: what is the purpose of line 89?
Wontonimo: seems sus
j4at: Well that's a definitely long answer for an easy puzzle
Wontonimo: do you know about the join feature?
Wontonimo: ["hi","there"].join("_") 'hi_there'
Wontonimo: yeah, it won't fix your problem but it will be way cleaner
Wontonimo: what are you trying to do here
let jsonResultObject = JSON.stringify(resultObject);
Wontonimo: looks like it is just for debugging
Wontonimo: you can just do this
j4at: yep he is using it for logging
j4at: you don't need a string
Wontonimo: or all in one line with your prefix message
console.error("First line result is: " + JSON.stringify(resultObject) )
Wontonimo: again, not going to solve your problem, but you came looking for feedback
j4at: I think I know i found the problem
Wontonimo: yes, js
j4at: Dicts are not ordered in js right ?
Wontonimo: that's right
Wontonimo: use lists for order
j4at: Well the problem that he stored the resut in a dict
j4at: then printed them unordered
Astrobytes: lol jacek
Wontonimo: i love js for small bits of code. not so much when there are 100,000 lines of code
j4at: Well, dicts outputs keys ordered (sometimes it doesn't)
j4at: And the first cases keys are ordered
j4at: and last case they are not
Ayza: I am assuming because the other testcases were alphabetically and numerically ordered?
Ayza: while the last one was kinda random
LinkCoder: random question , can i specify the kind of clash of code which i want to play ?
b0n5a1: only in private mode
antiwonto: [auto] Hey Dono1, here is a :taco: for loggin in today while it is quiet. You now have 1 tacos
Dynamolten.: hello lucazade
trulynameless.: dono1 use a python dictionary https://www.w3schools.com/python/python_dictionaries.asp
trulynameless.: its really helpful
antiwonto: [auto] Hey Dono1, here is a :taco: for loggin in today while it is quiet. You now have 2 tacos
trulynameless.: umm ill see
Astrobytes: Dono1: No, don't ask for things like that. You have an example, now find a way to do that in the language you are using.
Dono1: I'm not looking for a free lunch here, but I've been stuck trying array.join() and the best I can get is: http://chat.codingame.com/pastebin/a5d9fdab-078b-4b22-8595-e8b0c982af96
Wontonimo: what's the code that is generating this?
Wontonimo: oh, i see, you don't have code to do it. it's just a question
The output stream shows my issue at the end
Wontonimo: look at this https://tech.io/snippet/E613Bsg
Wontonimo: it may help
Dono1: It might help...not sure though. I previously was using an object instead of a map, but on Test case 06 of this level, the object kept becoming unordered, and it was recommended to me to use a Map so that it remembers the insertion order.
Wontonimo: and here is some more https://tech.io/snippet/IwVqyf8
Wontonimo: gotta go. have fun
antiwonto: [auto] Hey waterproofsodium, here is a :taco: for loggin in today while it is quiet. You now have 19 tacos
Wild_Python: why i can't run java program :(((
Wild_Python: javac HelloWorld.java in cmd but it do nothing :(((
Wild_Python: help me
Wild_Python: why i can't run java program :((( javac HelloWorld.java in cmd but it do nothing :((( help me
Mortis_666: Wild_Python after running javac HelloWorld.java run "java HelloWorld"
Wild_Python: i run it thuogh
Wild_Python: it does nothin
Wild_Python: Code: http://chat.codingame.com/pastebin/e01bb08e-ab47-49b9-927b-21f90ecb2586
Wild_Python: i'm ew :(((
Wild_Python: help me
Wild_Python: i can't run java :(((
smallPrick: fuck you
antiwonto: [auto] smallPrick be respectful and watch your language https://www.codingame.com/playgrounds/40701/help-center/code-of-conduct
System-LD: i'm so confused right now.
Wild_Python: javac HelloWorld.java
Wild_Python: then java HelloWorld
Wild_Python: nothing happen
Mortis_666: no errors either?
Wontonimo: is that the beginning of a joke?
wfournier: at least 2$ per year
Wontonimo: i don't know xNotAHackerx, how much does a programmer get paid?