nguyenletienbao: how to loop through every two element in array? help..
jacek: which language
Klient: Use a simple for loop with i += 2
jacek: or [::2] with python
Klient: Don't know python so said the most general thing xD
Klient: If that works it works tho
jrke: [::2] slice is enough for python
Uljahn: "for a, b in zip(arr, arr[1:])" gives you all pairs of elements i.e. 0, 1; 1, 2; 2, 3 etc. and "for a, b in zip(arr[::2], arr[1::2]):" gives pairs with step 2: 0, 1; 2, 3 etc.
jacek: arr :pirate:
Uljahn: that's what you get being addicted to numpy :/
Uljahn: should be of list type ofc not array
Uljahn: or 1D array, though could work with ND
Astrobytes: How do you annoy a pirate? Kick him in the arrr-se...
Astrobytes: (sorry)
Cedrikus: hello, world
WINWINWIN: Hi
Herdre: hi
Klient: That was perfect, Astro :"D
jacek: arrstrobytes
Astrobytes: lol
Astrobytes: R - language of choice for pirates.
DJBrutalCarpet_85b3: 1
something_ok: sr. how long to ranking be reloaded?
WINWINWIN: :D Astrobytes
Astrobytes: ;)
JoeEhab12: does anyone know how to join a clash with multiple rounds?
WINWINWIN: dont understand you... multiple rounds in what sense?
JoeEhab12: i mean not only one 5 min challenge
JoeEhab12: like 3 rounds of different problems
JoeEhab12: or that needs a private clash?
WINWINWIN: Surely not in a regular clash, but I am unsure with the private clash mechanics.
TrapRNotGay: integrals are fun + C
ad99526: https://www.codingame.com/clashofcode/clash/1295176ee79f147b4c73379735e5e4c58c51d94
ad99526: join
ad99526: challenge
Hanifu: tes
ad99526: https://www.codingame.com/clashofcode/clash/12952079e3788ee1f564eed77de13de6cfa2e60
darkhorse64: Please don't post CoC invites on the chat
proscripterlol: hi
AggYzz: hello peoples
AggYzz: my name is aggyzz
AggYzz: no
jacek: good for you
AggYzz: no u
Washier: <vuvuzela>
omer-chetin: why people not sharing thier code
AIDRI: hi
omer-chetin: soo annoying
omer-chetin: hi
proscripterlol: guys why not share code please
proscripterlol: im here to learn
proscripterlol: ;(
proscripterlol: it make me sad
StockBot: lol, we are here to WIN !
AIDRI: ok
AIDRI: i m sad for you
Washier: some of the opponents are bots. they dont share afaik
K4Mx: eulerscheZahl, can you paste again the sample game output of Onitama per turn, you sent a sample once here , forgot to save it , its for a pres , thx
K4Mx: cant find it on git, is it there ?
K4Mx: or ill just hack it in the code , lazy me xD
jacek: oO
omer-chetin: neredesin turk
Laminator: Pirate favorite language: "You may think it's R but me only love is the C"
Washier: haha.
Astrobytes: lol, well played
Washier: I know R quite well. its not the tool for this place, unless maybe if you're training\evolving something. it is beautiful tho when you work with data\stats
Astrobytes: yeah, it's common in my field (biology), still haven't got around to it myself (left it for the bioinformatics guys) but I should
Laminator: Yeah, I have a stats background so same. I could probably hold my own in a clash of code using R. I'd say it's as easy as Python. Plus it uses a base-1 system.
Washier: noice. yes, it evolved from S. biologists and the like were doing data science\ai in the 80's already. Ihe big diff is vectorisation, so you could declare a vector like (1,2,3) and add 1 to that, without a loop.
Munchkin: Hey, guys, is it norm way to use shield? print(str(next_check_x) + " " + str(next_check_y) + " " + "SHIELD")
Uljahn: seems good
Washier: R is a script , so really slow. for puzzles, and especially golf, yes i would like the option, but it doesnt fit imo
Munchkin: ty
omer-chetin: HELLO WORLD
Washier: invite me to a clash moer-chetin
Washier: *omer-chiten. apologies
omer-chetin: how
omer-chetin: clash not found
eulerscheZahl: Washier do you know that you can tab-complete usernames?
Washier: no didnt, ty
Washier: noice
Washier: omer-chetin (it works) we missed each other.
Washier: I will follow you and invite you incessantly
omer-chetin: do it
Washier: promise.
Washier: i also need some learning on CoC. i wield C# and ruby
omer-chetin: so you dont know python
Washier: no. the lack of scoping brackets pisses me off ;)
omer-chetin: I have simple experience in Python and web devtools Except javascript
eulerscheZahl: i like python :(
eulerscheZahl: + some experience with Django
Washier: i'm just full of it big boss euler. python is amazing
Laminator: ruby is an absolute monster at shortest mode. I can pull out all the stops on python and still lose by 10%
omer-chetin: thats true
Washier: yes, it is so short. i like that
Laminator: We need some language balancing for shortest mdoe
Washier: haha. being a c# guy i would say yes. but i would be wrong.
Washier: shortest mode forced me to start learning ruby. so good imho. for fastest or reverse i will stick to C#, because i know it best, more about the problem solving
eulerscheZahl: that's why we have 15min on shortest: you can google your way
Washier: exactly. how i'm learning.
Washier: i know i sound like a stuck gramaphone, but i really love codingame. my advice to anybody would be, keep going
Laminator: I learned Ruby for the exact same and only reason lol
Astrobytes: Agreed. If you put the effort in you can learn a lot.
DiL: I'm also learning ruby just to compete on shortest mode lol
DiL: and as a python dev, ruby just looks crazy :-)
Laminator: I did learn a ton of tricks in python. But it still wasn't enough. 15 min is still plenty of time to solve it in Python and then try to google my way through Ruby.
Ripkite: whats up hombres
Washier: deep discussions bud
tomatoes: poor ruby
Washier: haha. ouch
Washier: tomatoes, you obviously a top coder. what do you think?
tomatoes: im not. think about what?
Washier: dont believe that. but ok, about languages. what lang would you lear if you started now
Washier: *learn
tomatoes: same i know. python and rust
Washier: ty
tomatoes: python is everywhere. rust is nice and fast. js maybe, but depends on tasks
jacek: as for "dont believe that" Dunning–Kruger effect
Washier: haha, my surname is Kruger
Washier: sharp point though, absolutely.
McCheng: Rust is better or C++ is better?
jacek: yes
eulerscheZahl: only correct answer :D
JBM: Nice underhanded plug for D
JBM: +1 wouldn't have done better
eulerscheZahl: throw new DNotMentionedException()
AbundantPuddle: if Rust > C++{
gotoAndPlay(D);}
else{
gotoAndPlay(D);}
Astrobytes: http://chat.codingame.com/pastebin/5316480c-1191-4342-81b6-e8193cf88576
Astrobytes: Because everyone should use the D once.
proscripterlol: hey guys please be nice and share your code :(
Astrobytes: I did
DiL: yes you D-d Astrobytes
Astrobytes: lol DiL :)
eulerscheZahl: proscripterlol are you clashing vs the bots that don't share?
eulerscheZahl: also: nice one DiL :D
DiL: :-)
DomiKo: wow
DomiKo: first win vs jacek
DomiKo: https://www.codingame.com/replay/477182396
Washier: wd
DomiKo: i guess that's good direction
Washier: has to be
Astrobytes: gg
Astrobytes: Are you MCTS or alphabeta-based DomiKo?
Washier: would like to know also
DomiKo: alphabeta
Washier: chest bump
Washier: how deep?
DomiKo: like 10 in first move
Washier: very noice
DomiKo: then like 7/8
DomiKo: in worst moment
DomiKo: mid game so expensive
Astrobytes: I get 14-16 currently but I have other issues
DomiKo: full depth?
Washier: wow
Washier: 14-16!
Astrobytes: Do you use a TT?
DomiKo: or some cuts
Astrobytes: Cuts
DomiKo: TT?
Astrobytes: sorry, transposition table
Astrobytes: Also, aspiration windows
DomiKo: no
DomiKo: i don't really know what could be use of TT
DomiKo: i just use bitmasks and currently trying to make good eval
Astrobytes: Well, whatever you're doing keep doing it, seems to be working good ;)
DomiKo: maybe i will try to do some better cuts
DomiKo: yours 14-16 is so impressiv
Astrobytes: Just have to fix my other bugs, currently rewriting. Just finished my offline arena, seems to be working nicely.
DomiKo: yea offline areana is really useful
DomiKo: i will try MCTS in some days
Astrobytes: Yeah I was thinking about trying MCTS variant too
Astrobytes: But I hope to get multi probcut working with my alphabeta at some point (one day lol)
DomiKo: i have to read about some varations of MCTS
DomiKo: because i guess that fully random mcts woudn't be so cool
Astrobytes: Yeah, heuristic-guided works well for a lot of things. Early playout termination w/ eval etc
DomiKo: i heard that
DomiKo: but first i have to find some blogs
DomiKo: that will expail that
Astrobytes: The EPT sometimes just becomes a UCB guided tree search, no random
Astrobytes: I'd say experiment with lower depths and eval when you terminate, experiment with what you backprop score-wise
DomiKo: interesting
Astrobytes: Depending on the game, MCTS solver might help, sometimes maybe not so much
DomiKo: i have to spend some time reading about MCTS variations
DomiKo: and maybe i would improve on UTTT too
Astrobytes: There's some good papers out there (and some not so good), just ask in here if you get stuck with anything.
DomiKo: i found some
DomiKo: will read them tomorrow
DomiKo: and definitely will ask if i needed
tomatoes: there are a lot
Astrobytes: Cool. Watch out for the pseudocode in the original MCTS solver paper, it's awful, just read the text.
DomiKo: oh yea
DomiKo: pseudocodes and really oful sometimes
Astrobytes: Aw you saw that already I think
Astrobytes: Or maybe was someone else
tomatoes: text is a bit awful too :grinning:
Astrobytes: lol true
DomiKo: old papers have that thing
DomiKo: (new ones too...)
Astrobytes: hehe indeed
DomiKo: or maybe
DomiKo: have you guys some paper
tomatoes: dont know why they mixed nega/mini in one paragraph
DomiKo: on EPT?
DomiKo: i guess about MCTS solver i will read the original one
tomatoes: i just translated idea to big if tree
tomatoes: it works and i dont touch it
Astrobytes: EPT I don't think I used a paper for it, I just did it and then found out what it was after
Astrobytes: Whilst some of us were working on Oware Abapa
Astrobytes: Actually, it was Neumann who mentioned it to me. I didn't get it at first then just tried it out and it worked.
Astrobytes: In that case it was depth 0, just select -> expand -> eval -> backprop
DomiKo: so when i backprob
DomiKo: i make it like in minmax?
Astrobytes: You can, bunch of ways to do it, it would seem
DomiKo: ohh yea
tomatoes: im propagating in mcts style
DomiKo: i found
DomiKo: http://chat.codingame.com/pastebin/b76bd23e-7405-4e16-af33-f546655ed729
tomatoes: i think its better with weak eval
DomiKo: i guess yes
DomiKo: https://www.cs.du.edu/~sturtevant/papers/im-mcts.pdf
Astrobytes: In Oware I tanh to keep it sensible
DomiKo: in this paper they try to do two things at once
DomiKo: mcts style is like taking avreage of eval values?
finnn: when i use lua
finnn: it prints 2.0 instead of 2
DomiKo: convert it to integer
finnn: but why should i have to do that
Astrobytes: brb, afk 5 mins
finnn: it should already be an integer
DomiKo: idk what you did
tomatoes: yes
finnn: just doing print(2/4)
finnn: i mean
finnn: print(4/2)
finnn: prints 2.0 instead of 2
tomatoes: try //
DomiKo: yap
remi6: '{:g}'.format(4/2)
DomiKo: operator / return float
finnn: it's lua
tomatoes: yes, and?
RvveSplit: oh this is a surpirse
RvveSplit: hello
RvveSplit: oh ok bye
finnn: print(4/2) in lua should print 2
finnn: not 2.0
jacek: ohai
finnn: lua doesn't have integer/number types
finnn: i mean float not number
tomatoes: print(4//2)
Astrobytes: ^
finnn: that's a thing??
finnn: strange
Astrobytes: Yes. You perhaps should spend some time learning more about the language ;)
finnn: i come from using luau (roblox's lua)
finnn: not plain lua
finnn: so i guess you're right
Astrobytes: Well, plenty resources out there to get up to speed. Keep going and don't give up
finnn: i think it's a version thing
Astrobytes: Alright I'm out for tonight, brain's frazzled. Goodnight/good day guys
finnn: dunno which version // was added or removed in but it's definitely specific to a version of lua
DomiKo: goodnight
DomiKo: see ya
finnn: literally
finnn: if you do
finnn: 5^5
finnn: it prints 25.0
finnn: ???
finnn: i have to add extra lines & spend extra time to correct that
finnn: not a normal lua thing
omer-chetin: 5**5 in python
omer-chetin: use round func
Laminator: I remember seeing python code where instead of a for loop, they used the exec() function and multiplied the actions as a string by the iteration amount or something, I can't figure it out. Do anyone know what I'm talking about and can give me a working example?
Ripkite: yerrrrrrrrrr
bartekkowalski: YERRRRR
Ripkite: lit
Ripkite: we need a more hype chat
bartekkowalski: ikr
Ripkite: this is why we get no biddies in this field
MarkAdell: hey, this is my first time here, is there a way to compete with a random user?
MarkAdell: like a 1 on 1 contest
bartekkowalski: not sure but if you just do clash of code you play against randoms but it is not a 1v1
MarkAdell: alright
MarkAdell: yea i tried clash of code, it's fun
Laminator: you could do a private clash of codes
recreationx: wsup
Ripkite: anyone else smacked
Ripkite: bro coding is so much fun
Ripkite: what you guys use in clash?
Ripkite: been switching between c++ and javascript
Laminator: python all the way
Ripkite: off that
Ripkite: only good for machine learning
Ripkite: and data mining
Cedrikus: hello, world
