Chat:World/2020-05-23
cegprakash: for those who missed my stream today : https://www.twitch.tv/videos/629219816
jacek: happy Caturday
cegprakash: Saelyos
cegprakash: how does Saelyos GA's on opponent :o
cegprakash: isn't nested GA too slow?
abt8601: On the coordination system, I remember seeing somewhere in the contribution guideline writes that this is a convention is CodinGame.
abt8601: *in
CyberLemonade: It's a convention usually in graphics too
CyberLemonade: 0,0 is top left
Csipcsirip: stupid convention tho imo
Csipcsirip: confuses every beginnners
eulerscheZahl: for me that coord system feels natural
eulerscheZahl: i guess because i saw it so often before CG
Csipcsirip: yeah same
cegprakash: http://chat.codingame.com/pastebin/7d822d69-82bb-46b2-a643-0544656822d5
cegprakash: does ur PacMan collision code similar to mine eulerscheZahl
cegprakash: also what do u think about https://github.com/Saelyos/Spring-Challenge-2020
cegprakash: he mentioned he used GA vs GA vs GA vs GA ...
cegprakash: isn't that too slow?
cegprakash: I mean it can work if he has a depth of 1 at each GA
cegprakash: but I wonder what depth he finally achieved with this strat
cegprakash: I think what he does is random vs enemy GA vs his GA vs enemy GA vs his GA (finally maximize this score)
cegprakash: which can be really good against greedy bots
Ryao: hi
TheFlyingStopwatchAndHisDog_b771: hi
Ryao: i am knew to this chat
BenWo: hello
TheFlyingStopwatchAndHisDog_b771: same :)
Ryao: so i am not so lonely
BenWo: yeah better to chat occasionally
BenWo: also good for wasting time :p
Doju: Heya
Doju: BenWo whats up
Csipcsirip: this new integrated ide is better than most ide in linux
BenWo: heya Doju
BenWo: not too much
BenWo: hopping between puzzles here :)
BenWo: how 'bout you?
Doju: alright
Doju: uhh, just procrastinating on doing important stuff :P
Doju: the usual
BenWo: heh, indeed :)
fangskills: anybody know how to solve the Only Negative test case in the temperatures puzzle
BenWo: don't recall
BenWo: can you give the jist of it?
fangskills: "Write a program that prints the temperature closest to 0 among input data. If two numbers are equally close to zero, positive integer has to be considered closest to zero (for instance, if the temperatures are -5 and 5, then display 5)."
fangskills: Everything else works, its just the "Only Numbers" test case
BenWo: compare absolute values
Marchete: sort data by abs(value), then by positive/negative
fangskills: like this? negativeTemps.sort((a, b) => Math.abs(b) - Math.abs(a))
positiveTemps.sort((a, b) => Math.abs(a) - Math.abs(b))
Marchete: I think not
Marchete: sort must be at the same time
Marchete: like first sort by abs(value) and if they are the same then by positive/negative
fangskills: Math.abs(negativeTemps.sort((a, b) => b - a))
Marchete: maybe abs(x-0.1) or something similar, I can't recall
Marchete: the thing is that positive numbers must be sorted in first place
Marchete: or sort by (abs(x),-x) is python allows it
fangskills: i didn't write a if function lol im so stupid
jacek: :o
Marchete: :anguished:
eulerscheZahl: :confused:
DaNinja: sigh :( https://www.codingame.com/replay/468603470
kovi: bugged?
DaNinja: yep, it was working better 30 minutes ago
kovi: illedan fixed it wrongly. should consider order
kovi: when touching multiple cps in the same turn
Marchete: rushed optim....
DaNinja: yes, I noticed that on cp 0 and 1
kovi: yeah, i meant turn36
kovi: hmm, actually my current solver also relies on that bug :o
kovi: (unknowingly)
pb4: Doesn't the engine require the the center of the pod touches the CP ?
kovi: it does but probably checks each cp separately
DaNinja: yes, but the car can cross two cps in one turn, and they dont have to be in the right order
kovi: not comparing time of touch
pb4: ohhhhhh
pb4: I see how it could be implemented wrong :)
kovi: sure, i have the similar bug (my code went wrong yesterday)
kovi: hmm, i got an interesting insight even improved my csb bot
kovi: (not seriously as per 2^n, but still)
amol: Does anyone know how to crack the game: There is no Spoon i did not even get what exactly is expected
H.S.H: hello how can i make 1110 to 1,1,1,0 ?
DollarAkshay: @H.S.H You will have to be more precise with that question
H.S.H: its will take me a nuber phone and i need to havethem 1 by 1 to translate it
H.S.H: like 0913254 to (0,9,1,2,...)
H.S.H: than use dict an tanslate it to morce
NapTown: covert to string then iterate over characters
H.S.H: th
reCurse: A night of training and -5 improvement... I think I'm done.
eulerscheZahl: nice job
eulerscheZahl: do you have the checkpoint order bug included?
reCurse: No
Mimino_Cj: hi
reCurse: Probably doesn't matter and will get fixed
reCurse: Alright it's moving time
reCurse: Bye
Astrobytes: Good luck and enjoy!
jrke: hey guys trying fantastic bits how to prevent from bludgers
WINWINWIN: jrke obliviate
eulerscheZahl: most useless spell of the game
CyberLemonade: is creating a class containing 1 integer equally fast as just having an integer
CyberLemonade: meaning: class Test{int value} object vs int value
Csipcsirip: if you use c# class will be probably sower because it will be created on the heap and you will access it by reference
CyberLemonade: hmmmm
CyberLemonade: thanks
struct: What do you think?
struct: https://i.imgur.com/NV1MSs3.png
eulerscheZahl: now it looks like the original :)
ZarthaxX: gooooooooooood struct
Astrobytes: hell yeah struct, much better
Snef: noice
struct: thanks
Snef: did you tried the black circle without white border ?
struct: Ill do a few more touches to UI then Ill finish statemnt
struct: yeah, I think it looks better without white border
struct: But I left it on the pic :p
Snef: ok cool
Snef: seems perfect then
struct: without white border on stones
struct: https://i.imgur.com/ERMTF4P.png
struct: Snef do you also mean no white border on rings?
struct: Or its fine on rings?
Snef: i'd say remove on ring too
struct: https://i.imgur.com/jV0f8s2.png
Snef: mmmmh
struct: I think border on ring looks better
Snef: well the problem is that with black border they are much thicker
struct: they look more thick
struct: They are the same thickness
struct: but with white it doesnt look as thick
Snef: could you try without any border ?
struct: Sure
Snef: i mean is it possible just on black circle ?
struct: to remove the border just on black?
Snef: yea
struct: yeah
struct: https://i.imgur.com/xoNYyHI.png
struct: Might be good like this
Snef: yeap but maybe others won't agree idk :p
darkhorse64: Sorry if someone already posted about it on the chat but there is a bug in Search Race referee: while accounting for multiple checkpoint intersections, it ignores their order. In Hold the line, when the car returns, it moves frequently on CP 1 then 0, it counts as if it moves over 0 then 1. CP is then turned on
darkhorse64: CP 2
Snef: I'm pretty sure it has already been posted
Snef: not sure if Illedan saw it tho
darkhorse64: Ah yes a few hours ago. Dunno if is related to latest changes but my bot is now unable to reach 100%. I get the test s right but one or two validators fail
Snef: sames validator each time ?
darkhorse64: No, but GA is not deterministic
Snef: i have that too
Snef: i have to do like 5 submit to get 100%
darkhorse64: I try to avoid captchas or worse being locked so I have tried to analyse my sim. That's how I found the pb.
darkhorse64: Gotta go
eulerscheZahl: regarding Search Race: IlledanToday at 5:21 PM I will fix it in a few hours when I get back home.
Alshock: I'm curious though, how can that bug be a disadvantage to you darkhorse64? If it's not an idustrial secret ofc
Alshock: from my naive point of view, it's a little bonus at worse
Alshock: -e+t
Snef: if you reuse your last turn solution for seed you will not optimize as fast i guess
Alshock: oh yeah, if you notice your paths are out of date and start again from scratch then yeah it can hurt
WINWINWIN: Hi,
WINWINWIN: I have this statement in python3
WINWINWIN: if (color_a in rotated_grid[i]) == True:
WINWINWIN: and even though the color is very clearly in the array
WINWINWIN: 2 ['.', '.', '.', '.', '.', '.', '.', '.', '.', '.', '2', '2']
WINWINWIN: (From Stderr)
WINWINWIN: itt returns False why?
Counterbalance: int vs string?
WINWINWIN: No, checked that
WINWINWIN: I printed the type of color_a to stderr as well
WINWINWIN: it was <class, st>
WINWINWIN: str
Aesop: http://chat.codingame.com/pastebin/2c21e622-0f0f-4b2c-ab7c-871503954e39
Aesop: DEFIB is a string, by the way
Aesop: Thankful for any help I can get
Counterbalance: j = DEFIB.length is an assignment
Counterbalance: j <
Counterbalance: WINWINWIN maybe the problem is elsewhere? I tested this script in python3 and it works as expected:
Counterbalance: http://chat.codingame.com/pastebin/4f15dbc2-74ac-4d1f-a089-93b6ba3df351
Counterbalance: (i'm a python noob though)
WINWINWIN: Hmm Counterbalance,
WINWINWIN: (b in a)
WINWINWIN: I did not use the brackets
WINWINWIN: is that important?
Counterbalance: I think you can get away with if color_a in rotated_grid[i]
anid: WINWINWIN, I don't know why your error is happening, so better if you write your own contains() method
anid: Just do it via for loop and you should not be able to go wrong, it will not have any performance impact
WINWINWIN: Will try that ty anid
pb4: "Snef 17H14 I'm pretty sure it has already been posted not sure if Illedan saw it tho"
pb4: he knows
struct: Might need to change some things, due to the making player 2 black
WINWINWIN: contains method doesnt work as well
WINWINWIN: def contains(arr, elmt): http://chat.codingame.com/pastebin/9d06c997-e282-4568-9a12-ef58283bb51d
WINWINWIN: Is this a correct implementation of checking whether there the element is there in py3?
eulerscheZahl: if element in collection: ...
WINWINWIN: Tried that, it didnt work either
WINWINWIN: no idea wth is wrong with it
MoMaT: can you paste in a bit of your exact code?
MoMaT: it is really x in why that you need.
MoMaT: * "x in y"
WINWINWIN: http://chat.codingame.com/pastebin/21990f52-6184-4c39-8c67-08a3e7f10b89
WINWINWIN: This is where I use the contains metod
Counterbalance: ahh
Counterbalance: you did if (color_a in rotated_grid[i])
Counterbalance: you should lose the [i], I think
WINWINWIN: But it is a 2D array
WINWINWIN: rotated_grid[i] is that specific column
Counterbalance: hm ok.. weird then!
WINWINWIN: Completely confused, no idea wtf is going on
MoMaT: you only need "if color_a in rotated_grid[i]:" as a test.
MoMaT: you might be storing different types that looks the same when you print.
WINWINWIN: but I compare str(i) == str(elmt)
WINWINWIN: so types will be irrelevant
MoMaT: perhaps try "print(repr(color_a), repr(rotated_grid[i]))"
WINWINWIN: repr?
MoMaT: yes, for "representation".
darkhorse64: Alshock: not sure if recent changes are the cause of my bot's woes, although I can't help noticing than of Hold the line, it has a hard time finding a path when returning. Still, the game does not work as expected and I thought it was a good thing to report. Once it is fixed, I'll be sure the problem is between the chair and the keyboard
darkhorse64: *that on*
WINWINWIN: '3' ['.', '.', '.', '.', '.', '.', '.', '.', '3', '3', '2', '2']
WINWINWIN: this is the result of the repr print line
Alshock: Oh ok, I thought your bot suddenly got worse and by investigating it you found that bug
WINWINWIN: but still got false
Alshock: @WINWINWIN wait wait wait
Alshock: @WINWINWIN your colour is a number, but your list contains strings
WINWINWIN: No but I am comparing str(i) == str(elmt)
MoMaT: this works. there must be something else going on there.
WINWINWIN: :(
Alshock: Nvm I said that looking back in the history of the chat, but your last post says otherwise
MoMaT: maybe try: if str(color_a) in rotated_grid[i]
WINWINWIN: will try that now
Alshock: or `if any(str(c) == str(color_a) for c in rotated_grid[i]):`
WINWINWIN: in did not work either
WINWINWIN: will try any now
Alshock: the problem is probably elsewhere
WINWINWIN: None of them are working, code is anyway only like 30 lines, will do a full rewrite
MoMaT: my guess would be that you change color_a after the first match. try printing it together with thr grid.
WINWINWIN: Am doing a rewrite probably some unknown error
MoMaT: there is no such thing :)
Alshock: The question is: how do you know for sure that you don't match?
WINWINWIN: :)
Alshock: MoMaT of course there is: unknown to WINWINWIN at least, and for now unknown to us as well
WINWINWIN: ty for ur help MoMaT
MoMaT: I meant, there's a bug to be found, not an error in python...
Alshock: who knows? Maybe it's a bug of the python interpreter, with an odd of 0.0000001
MoMaT: yeah, but you answered yourself how likely it is...
Daher: python is a joke
Alshock: yes it's completely neglictable, but no one claimed that's what's happening :P
WINWINWIN: Yeah I like C# better, but Im just starting Smash the Code I began with python3
Alshock: @Daher and Java is a bad joke
Alshock: C++ is a drunk racist uncle
Alshock: and C# is an unsustainable I-know-all
Alshock: dunno how that last one translates to proper english
WINWINWIN: :D
WINWINWIN: Rust ftw
Alshock: if you need to choose between Rust and brainfuck, by all means pick brainfuck?
WINWINWIN: :/
Alshock: why? Cause I just decided I'd trash all languages, yours won't be any exception ok?!?
struct: MSmits here by any chance?
Alshock: NullPointerException... OK Java you might be an exception
WINWINWIN: AGAIN!!!! -_-
WINWINWIN: damn annoying, will try it out in ruby
Alshock: participants says no, but idk if it only counts people who posted or all online people
Alshock: @WINWINWIN do you mind pasting all the part of the code that leads ot that error?
AntiSquid: if brainfuck was on CG and it gave stupidly huge advantage compared to any other language, i would pick brainfuck
AntiSquid: so yeah do pick rust
Alshock: BTW Ruby is life, Ruby is love
AntiSquid: "ruby users are the children of sata" - a clasher who codes in python
Alshock: @AntiSquid No, if that was the case, you would use a transpiler just like that of Cotr
jacek: struct 3 2 is best opening move in Yavalath
jacek: now lets wait for Him
struct: ty
WINWINWIN: Alshock already pasted that code in there
WINWINWIN: now switching to C# and praying it works
ArianPunk: hard puzzles doesn't make sense.
eulerscheZahl: so the puzzle is wrong? and not you?
Quidome: he euler, question on Bender 4
Astrobytes: Obviously a bugged puzzle.
Quidome: I solved the thing, i heard the testcases on submit are available
Quidome: What is the nature of the puzzl, are you allowed to solve it offline and hardcode the results?
Quidome: Or only use the available time on the server?
struct: euler why does this happen?
struct: java.lang.RuntimeException: Error reading Referee command. Buffer capacity: 32756 / 100000
struct: It only happens sometimes, on same seed
struct: Like 1 / 20 times
eulerscheZahl: i've seen that before
eulerscheZahl: might be that you create too much data for a single turn?
eulerscheZahl: but i don't see what could be that expensive
struct: I guess, but is strange that it shows a lopwer number
eulerscheZahl: there are 2 size limits per turn
eulerscheZahl: 100k on first turn
eulerscheZahl: 30k for later turns
eulerscheZahl: json size in the replay that is
eulerscheZahl: Quidome you are allowed to solve offline and hardcode
eulerscheZahl: personally my solution is found online
struct: I see, but seems strange it only happens sometimes
struct: on same map and play
struct: So game is =
eulerscheZahl: that's really weird
Quidome: eulerscheZahl ok, thnxs, that makes it a little bit easier :D
struct: Ill try to reproduce, ill also export, if it happens again ill send it to CG
fangskills: in Coders Strike Back, why does my AI output X and Y distance and then it output X and Y as NaN
struct: fangskills what do you mean?
struct: can you put the code here?
eulerscheZahl: division by 0?
jacek: oh my
fangskills: console.log(nextCheckpointX, nextCheckpointY, thurst) => response: 3974 7807 100. 2nd response: NaN NaN 100
Snef: gig you just changed league fangskills ?
Snef: did*
fangskills: Nevermind i figured it out lol i always keep forgetting my if statements
Counterbalance: struct do you get that on the CG servers? I got it locally, figured it was a windows I/O thing. Didn't happen on the first turn, and the mirrored version of the game had 115kb on that frame (and worked fine).
struct: yeah locally
WINWINWIN: Got my Smash the Code working in C#, is basic code enough to get out of Wood2?
struct: WINWINWIN to get to bronze all you need is to play a legal move
WINWINWIN: Trying that
cegprakash: will stream oware abapa minimax after having dinner
WINWINWIN: right now I make a sequence if they are in a column, else an empty column else random
WINWINWIN: Still only #2 in Wood2
WINWINWIN: @cegprakash 11:20 dinner?
WINWINWIN: Man
cegprakash: it's supposed to be my breakfast
cegprakash: I just woke up
WINWINWIN: :D
cegprakash: GA can work amazing on Smash the code
cegprakash: or even a recursion will do
cegprakash: without considering enemy
cegprakash: it's afterall 6^8 possibilities
WINWINWIN: Yes
Snef: isn't it 11^8 ?
Snef: with rotation
struct: 22^8
cegprakash: I didn't include rotation
Snef: right
cegprakash: :o
cegprakash: why 22?
Snef: 6+6+5+5
Snef: 6 when they are in "I" shape
Snef: 5 when they are in "_" shape
cegprakash: oh there are L, T shapes in higher leagues?
struct: No
Snef: no
struct: ceg
Snef: vertical horizontal
struct: Pieces can have 2 colors
ZarthaxX: 6^8 is pretty horrible already tho
eulerscheZahl: "GA can work amazing on Smash the code" I don't see how GA should be awesome here
struct: if it was 6^8 you preety much could simulate all
cegprakash: 6^8 runs in 1 ms in my computer
struct: Sure
struct: go sim smash the code
ZarthaxX: right
ZarthaxX: :P
struct: 1.5mill sims in 1ms
struct: Do you have nano computer?
ZarthaxX: he has the quantum
Snef: lel
eulerscheZahl: now this is suddenly on topic: https://codeforces.com/blog/entry/77614
struct: I can get more than 1.5mill sure, but not on 1ms, or when board is almost full
ZarthaxX: wtf
eulerscheZahl: 100 tshirts in total
eulerscheZahl: in kotlin heroes soon. not announced yet. but last time it was 100 tshirts as well
Snef: can't we do this in D instead of Q# ?
cegprakash: https://ideone.com/rORadz
eulerscheZahl: there is no D only contest
eulerscheZahl: kotlin heroes on may 29th
ZarthaxX: what with that cegprakash?
MostComplicatedUsername: I just realized I missed a whole contest
cegprakash: 0 secs
struct: lol
struct: dont troll
struct: please
ZarthaxX: ceg that's not a way of testing
ZarthaxX: and you know it
cegprakash: well u need a sim with O(1)
ZarthaxX: funny
eulerscheZahl: i have
eulerscheZahl: but higher constant
cegprakash: see..
cegprakash: it's possible
cegprakash: eulerscheZahl has done it
ZarthaxX: i will just stop talking
struct: I will pay you
struct: if you get 1.5mill sims
struct: in 1 ms
struct: with depth 8
eulerscheZahl: when we define everything as constant (taking the upper limit, e.g. board size), it is constant
cegprakash: go watch my stream for pallankuzhi I was getting easily million sims at depth 10
ZarthaxX: well said toad
cegprakash: I streamed today morning
struct: pallankuzhi is not STC
cegprakash: it's still 6^depth
struct: STC is not 6^depth
struct: its 22^depth
cegprakash: Oware Abapa
struct: and you need to floodfill
Snef: and one turn cost way more
struct: You need to drop pieces
cegprakash: I see 24^depth struct
cegprakash: how 22?
Snef: just think lel
struct: ^
eulerscheZahl: scroll up
eulerscheZahl: there's an explanation for 22
cegprakash: I don't get it.. we have 4 possible rotations * 6 positions to place
cegprakash: that's 24
Snef: look the rotation
Snef: for all of them you have 6 places ?
struct: Should I keep both?
struct: https://i.imgur.com/fp3AiTW.png
struct: Currently is optional, viewer can choose
cegprakash: oh yes we can't keep it outisde board
cegprakash: gotcha
struct: viewer* player
eulerscheZahl: both sounds good
eulerscheZahl: you already coded both
struct: yeah
struct: Now im gonna add mouse over, to display coordinates
eulerscheZahl: then players will share screenshots and we will be confused that it looks completely different :D
struct: I have coordinates optional for hex board
struct: but I cant find a good color for it
struct: Due to black pieces
struct: I add black text before
ArianPunk: whos done The Bridge puzzle?
MostComplicatedUsername: euler im above you in td :)
MostComplicatedUsername: Why does my code look different?
MostComplicatedUsername: It looks like keywords are in different colors
Astrobytes: They changed the IDE from ACE to Monaco MCU
MostComplicatedUsername: oh
Astrobytes: struct, that Yinsh board with the original design is sooo much better. Great job.
struct: Thanks, stil giving some finishing touches
jrke: hey i wanna know how to predict enemy pod in csb for attacking i am using x+vx and y+vy is this ok
MostComplicatedUsername: My rank fell 900 places in the few months that I wasn't on
MostComplicatedUsername: jrke that only predicts one turn ahead though
ArianPunk: have u encountered buggy puzzles?
struct: jrke check csb magu s guide
Astrobytes: ^
eulerscheZahl: congrats MostComplicatedUsername
MostComplicatedUsername: :)
Astrobytes: now fix it MCU
MostComplicatedUsername: ?
MostComplicatedUsername: Fix what
Astrobytes: Your rank :)
struct: I have a hard decision, should I add the coordinates or not, they are optional on viewer, you can choose if you see them or not
struct: https://i.imgur.com/VOIWI66.png
struct: Also the color, which one?
eulerscheZahl: that looks like a mess
struct: Yeah :(
struct: I guess ill remove this optioin
struct: lol
Astrobytes: That font is shit
MostComplicatedUsername: I'm 6th place! That's good!
struct: Give me a font
struct: :D
Astrobytes: What do I look like, a designer? :P
eulerscheZahl: a ca
eulerscheZahl: cat
Astrobytes: A font with no outline will be better, for starters
Astrobytes: Thank you toad
struct: I need outline because of white pieces
struct: or I can change the color of text
Astrobytes: black font on white, white font on black
struct: hmm, a bit messy to code now
Astrobytes: Also, perhaps a sans-serif font would be more readable?
struct: Ill just remove
struct: :p
struct: or leave it, its optional
struct: you can just disable
struct: someone might like it
eulerscheZahl: options are always good
jacek: comic sans
Astrobytes: No one will like it in the current state :P
Snef: jacek => []
Astrobytes: YES Jacek has the answer
eulerscheZahl: that font might be what gets you towards the 30k limit
eulerscheZahl: lengthy name used many times i json
eulerscheZahl: in json
eulerscheZahl: for CGFunge i show one line after another not to exceed the per-frame limit
struct: Arial looks good
struct: might be euler
struct: but for me its strange that it happens only sometimes
struct: It is not happening for a while though
struct: so I guess im safe
AntiSquid: hey the new design looks good
struct: Thanks
AntiSquid: maybe even the right one (grey colored board) could work with different background color
Astrobytes: struct, also try Verdana
AntiSquid: use an Apple font
struct: Soon I have too remake yinsh
struct: and it will be become customizable viewer
Illedan: Why? Did you fail?
AntiSquid: Myriad, Lucida Grande and Helvetica Neue or even Avenir
AntiSquid: Jobs sat in on Palladino's calligraphy class at Portland's Reed College, which eventually inspired the elegance for which Apple computers are renowned, the tech icon recalled in his famous 2005 commencement address at Stanford University. "It was the first computer with beautiful typography,"
Astrobytes: Come on, Arial and Helvetica Neue are practically interchangeable
AntiSquid: that's why you're not a world renowned businessman Astrobytes
Astrobytes: Or dead.
AntiSquid: he died of cancer, are you saying the font gave him cancer?
Astrobytes: I'm no oncologist...
Astrobytes: Speaking of medical things, I have to go for a brain and spine MRI in a couple of weeks.
struct: why brain?
Astrobytes: It's attached to my spine ;)
Snef: he has only 16 404 at search race
Snef: that's why
Astrobytes: lol shuddup snef :D
Snef: :D
AntiSquid: "only" ? hard to tell what's a good score
AntiSquid: is the game released?
struct: y
Snef: yes
Astrobytes: 10,641.950 is a good score
Snef: well i think that without a search sub 15000 is good
Snef: and with a search sub 11k is good
Astrobytes: Yeah I have no search yet
AntiSquid: since today or what
Snef: hmm 2 day i think
Astrobytes: Was either this morning or last night, not 100% sure
AntiSquid: no cover pic
AntiSquid: also many community multis have no cover
AntiSquid: should be made mandatory
AntiSquid: even the avatarless guy agrees
Astrobytes: They don't call JBM Jean-Baptiste the Mysterious for no reason you know
eulerscheZahl: stop doxxing
JBM: that's a pseudo for hiding my real "java bad mystery" id
Astrobytes: We already established that it's J'aime. Bof. Merde eulerscheZahl
WINWINWIN: :) any point going for #1 in multi`s like search race? reCurse always dominates with his NN
Astrobytes: Of course. See how far you can get
AntiSquid: Jack Beanstalk the Mysterious
Snef: go for #2 WINWINWIN
Astrobytes: *Jacques
eulerscheZahl: search race is not a multi
JBM: i'm really going to have to open a wiki for this topic
Astrobytes: :D
struct: Just to think I still have to write statement :/
JBM: i've already forgotten what my new favorite was
struct: JBM can you do it?
AntiSquid: WINWINWIN yes ofc, bragging rights and you get to annoy the guy you dropped, bonus points if it's someone like recurz
eulerscheZahl: saw beanstalk only in context of AWS so far. not even sure if i should pronounce it bean-stalk or beans-talk
Astrobytes: It'll end up like the Ubuntu dev code names that euler posted the other day
WINWINWIN: :D
JBM: you'd pronounce both differently?
Astrobytes: The former euler
eulerscheZahl: so like secretly observing beams. creepy
Astrobytes: *beans
eulerscheZahl: right, type
eulerscheZahl: typo
eulerscheZahl: (╯°□°)╯︵ ┻━┻
Astrobytes: :D
AntiSquid: was reference to "Jack and the Beanstalk" btw, just in case AutomatonNN missed it
AutomatonNN: you still have a lose and no one does
eulerscheZahl: https://www.codingame.com/forum/t/performances/184883 that's some serious misunterstanding of how CG works
eulerscheZahl: oh, now i got the name. thanks AntiSquid
Astrobytes: Just a bit.
JBM: or just next-level trolling
Astrobytes: Notice, eulerscheZahl, that's it's not Jack and the Beans Talk. That would be a show about Jack and some Mexicans presumably.
JBM: whereas Jack and the Bean stalk is a PI agency name?
Astrobytes: ofc, Jack Haricot, P.I
Astrobytes: Yes, I realised my mistake and waited for the correction
Astrobytes: I'm immensely grateful
AntiSquid: he should reply with "So when i a CG server update to be expected?"
AntiSquid: is *
Aesop: why does it have to be so damn hard to find a minimum value's index in a List in C#? I just can't figure it out.
Aesop: Scourged the net for answers and it's all about writing your own damn methods, why isn't this standardized?
kovi: nice neumann
Neumann: thks
AntiSquid: https://stackoverflow.com/questions/16684769/get-minimum-and-maximum-value-using-linq
AntiSquid: Aesop
Illedan: Neumann \o/
Astrobytes: Oof great score Neumann, grats
Aesop: Thanks AntiSquid, just figured it out
Neumann: Wait for when we'll be able to use full 50ms :P
Aesop: List.IndexOf(List.Min()) works just fine
Illedan: ^^
eulerscheZahl: but that is slower than necessary Aesop
Aesop: could you please elaborate?
eulerscheZahl: you iterate twice over the list
Aesop: does Linq mitigate that slowdown?
eulerscheZahl: 1st to find the min and then again to find the index
eulerscheZahl: i don't know what you are doing. maybe it doesn't even matter in your case
kovi: +30% time will give no more than a 1-2% with SA
Aesop: it's a small list, but i'd rather keep it clean and nice from the beginning so as to avoid future mistakes of that kind
Aesop: never used Linq though so I'll have a closer look at AntiSquid's link
kovi: (actually only +25% for me, im at 36ms)
eulerscheZahl: LINQ doesn't have the greatest performance. but it's awesome to express some thoughts in very little code
darkhorse64: kovi: Shall I infer that I should learn SA for SR
darkhorse64: ?
kovi: well, actually i dont use real SA either
eulerscheZahl: read the tags: you need beam search
AntiSquid: obey
Illedan: My buggy beam search gets 22k and 90 % :D
Astrobytes: You need Simulated 'Anhealing'
eulerscheZahl: Unhealing
eulerscheZahl: let me unheal you quickly :rifle:
eulerscheZahl: :pistol:
eulerscheZahl: really?
eulerscheZahl: :weapon:
eulerscheZahl: :gun:
AntiSquid: how do you decide what to use for speed racer? code them all or what
eulerscheZahl: :sword:
eulerscheZahl: come on...
Astrobytes: 'Simulated Anhealing' is Illedan's secret algo
Illedan: -3 vel AntiSquid ezpz
Illedan: fix it
Astrobytes: :knife: eulerscheZahl
AntiSquid: i mean how to decide which meta heuristic to use
eulerscheZahl: thanks Astrobytes. was looking for something more powerful
eulerscheZahl: :car: that will do the job
AntiSquid: :fist:
eulerscheZahl: ??
Astrobytes: In the right hands a knife can be very powerful indeed
AntiSquid: :truck:
eulerscheZahl: i can't remember failing that many emojis in a row
Illedan: :bomb:
Astrobytes: :bow_and_arrow:
AntiSquid: :toad:
AntiSquid: :frog:
Astrobytes: :squid:
Astrobytes: :motorcycle:
AntiSquid: not me
eulerscheZahl: :musical_score: hurting your ears
Astrobytes: :no_entry_sign: :motorcycle:
Astrobytes: Closest I can get AntiSquid
AntiSquid: :no_entry_sign: :squid:
AntiSquid: whichever lol
Astrobytes: Both covered nicely there
AntiSquid: or i need to chain all 3 ?
Neumann: Beam Search sucks on SR
eulerscheZahl: but it's tagged
Neumann: It's a trap
AntiSquid: but what about SEARCHO NO CHOKUDAI ?
eulerscheZahl: so from my point of view as an outstander not player beam search must be a save bet
Neumann: Didn't tried Chokudai but I'm guessing it'll suck for the same reasons
Astrobytes: But Chokudai is the Only Way. Other than the One True AI.
Illedan: How do you know beam search sucks Neumann? :P
Neumann: Have I missed some memes about Chodukai ?
Neumann: Because I tried it Illedan :D Among other things
Illedan: :D
eulerscheZahl: like people tried minimax and saw it fail on CSB? realizing that GA is the best way
Neumann: haha
AntiSquid: then why no GA here? why someone mentions SA?
Neumann: My guess is that every search that uses intermediate evaluations as a way to chose which branch to explore sucks
Neumann: I might be wrong
AntiSquid: or solo player smitsi
eulerscheZahl: MCTS
Astrobytes: ^
RoboStac: I can say that solo player smitsi isn't working very well for me
Neumann: Solo player Smitsi is called "MCTS"
Snef: Neumann i think that too
Illedan: haha robo
RoboStac: even if it still has 3 other pods that just can't collide and never move in it? :)
Astrobytes: Are you not planning some kind of NN-based solution RoboStac?
Neumann: Capped-depth MCTS worked "ok" for me
Illedan: Damn, you have tried a lot already :o
Astrobytes: How deep Neumann?
Neumann: MCTS, Hill Climbing, SA, Beam Search
Neumann: Pick one
RoboStac: I've not really planned anything Astrobytes, just converted my csb code with as little effort as possible
Neumann: Astrobytes: not that deep
Neumann: About half as deep as Snef
Astrobytes: Simplest way first Robo ;)
Snef: you got this score with half my depth Neumann ?? or just for your mcts try ?
Astrobytes: What's your depth Snef?
Neumann: Both
Snef: wut
Snef: around 20 Astrobytes
Astrobytes: k
Alshock: What's the advantage of a capped depth mcts over a mcts with an eval?
Snef: capped depth mcts do have eval no ?
Neumann: Yup
Astrobytes: yes, EPT
Alshock: which depth is then capped only by its rollouts in the branch
AntiSquid: leaderboard is sorted in wrong order, i have most points
Neumann: lol
Snef: congratz AntiSquid
Astrobytes: I never thanked you Neumann for introducing me to that concept during Oware
Neumann: It's not mine
Astrobytes: No but you mentioned it
Neumann: What does EPT stands for ?
Astrobytes: And I didn't know about it previoously
Astrobytes: EArly playout termination
Neumann: Ok
jacek: but you still need eval for it :c
Astrobytes: Yes.
Neumann: Astrobytes = Astroshytes ?
jacek: Astro8bits
Astrobytes: Yes Neumann, I should probably change it on Discord
Neumann: Ok, I can stop to wonder that everytime now
Astrobytes: :D
MostComplicatedUsername: I feel like Clash of Code doesn't test your knowledge of programming concepts, it tests your knowledge of obscure libraries
eulerscheZahl: here's another one for you Neumann: i'm an issue = BenjaminUrquhart
Neumann: I knew that one
Astrobytes: Doesn't he have another alias too?
Astrobytes: (changed my discord name to astrobytes btw)
Astrobytes: creepersbane or something
YurkovAS: Neumann do you have fails in SR with ttl 45-49ms? don't understand why 2 distinc sim work only with ttl 30-40ms.
eulerscheZahl: congrats, now you confused those who are mostly active on discord
Neumann: YurkovAS: Yes, happends to everyone trying to use more than 35-38ms
Neumann: happens *
Astrobytes: Meh. I always assumed having the kitten version of my CG avatar was enough anyway
eulerscheZahl: i even get timeouts with 50ms for a 100ms multiplayer (connect4)
eulerscheZahl: used to work with 90ms
YurkovAS: Neumann thanks
Astrobytes: C# issue eulerscheZahl or general?
eulerscheZahl: dunno
eulerscheZahl: i almost exclusively have c# bots
Astrobytes: true
struct: I was having this issue to withc++
Astrobytes: On SR or others too?
struct: sr
Astrobytes: Yeah that's happening for all on SR using > x time limit. euler's one is the odd one here
struct: oh god
struct: white color was a mistake
struct: https://i.imgur.com/9tzuBjJ.png
Astrobytes: hehehe
struct: oh
struct: no
struct: I uploaded with yavalath stament
struct: ahahah
eulerscheZahl: is it a win or a draw?
eulerscheZahl: :D
struct: same on viewer is hard to tell
struct: I guess ill leave player colors
struct: and change it in viewer instead of on config
Astrobytes: you're up late eulerscheZahl
MSmits: I'm getting tired alreadt
MSmits: is the world upside down?
Astrobytes: Must be :) How are you MSmits
Illedan: MSmits, not on SR yet?
struct: Hello MSmits
struct: Ive somethig to ask, ill draw it
MSmits: hi
MSmits: not on SR no
MSmits: someone's gotta try and kick karlis o off nr 1 uttt :)
MSmits: draw it struct
Astrobytes: UTTT is almost a way of life for you now eh :)
MSmits: yeah
MSmits: but i am mostly doing it because i can do it on the side, while doing some work. Not much coding
Astrobytes: Any of your students playing it?
MSmits: Dont think so, but I have a student that wants to do his last year project on search algorithms. He will pick a CG game i think
MSmits: possibly uttt, though it's probably not the best pick
Astrobytes: The one who participated in the contest?
MSmits: its better to write a report on a game that makes more sense when you look at a board
MSmits: yes
MSmits: that one
Astrobytes: Very true, and awesome. That's really cool
MSmits: yeah
Astrobytes: Seemed a smart kid
MSmits: yeah he is quite good. He did the labyrinth puzzle without any help
MSmits: we dont teach at that level really, noone is supposed to be able to solve the labyrinth
ZarthaxX: i couldnt sovlve it even
ZarthaxX: what a beast
Astrobytes: Nice. He's got the interest, that will take him far.
MSmits: I had a lot of problems with it too ZarthaxX
MSmits: he did it in python too, i think
Astrobytes: Hey Zarthachloride
MSmits: you're going through chemical formulas now?
Astrobytes: Just whatever comes to mind
MSmits: I see
ZarthaxX: i saw it once and left it
ZarthaxX: then came another day and was easy
jacek: why dont he do paper soccer then
ZarthaxX: the fuck Astrobytes LOL
jacek: :s
MSmits: you should pose this question more generally jacek
jacek: hm?
MSmits: why don't people do paper soccer?
jacek: exactly!
MSmits: honestly, I dont know, it seems fine
Astrobytes: Multi-overload
MSmits: its about as interesting as twixt, it's on my list, but below a few others, not so many
MSmits: i want to do breakthrough first
MSmits: yinsh if it comes out probably
struct: I dont know how I should draw this
struct: Its a MCTS thingy
MSmits: oh
MSmits: maybe i understand it from your words
jacek: breakthrough eh? mcts again eh
Astrobytes: what about that online graph too lthat euler posted
MSmits: some form probably yes jacek
Astrobytes: *tool
struct: Do you have the link Astrobytes?
Astrobytes: gimme a sec, I might
struct: In my head this makes sence and it doesnt
MostComplicatedUsername: lol sence
struct: I just wanna know if this idea is flawed
MSmits: is it a superposition of sense and nonsense struct?
struct: Sorry english is not my mother language
MSmits: isn't mine either
MostComplicatedUsername: Me neither
struct: Mas se quiseres falo português e não me vês a errar
jacek: omelette du fromage
ZarthaxX: ban
Astrobytes: struct http://www.webgraphviz.com/
struct: I got no idea what that is MSmits :p
MSmits: what is>
MSmits: you're making less sense than usual struct
Astrobytes: Impossible! :P
ZarthaxX: he is high
Astrobytes: On a mountain? :P
MSmits: of meth
MostComplicatedUsername: ...
Astrobytes: lol, I'm rewatching Breaking Bad so I like that reply
MostComplicatedUsername: Children these days...
Astrobytes: Maybe if you smoke some DMT struct, it will become clear
MSmits: watched breaking bad two times. All of it
struct: You started it
struct: Dont come with the child play
Astrobytes: Great isn't it MSmits
MSmits: best series ever
Astrobytes: I concur
MSmits: I am also watching better call saul. It can't be as good as breaking bad, but it's an unfair comparison. I like it too
Astrobytes: I was thinking about watching that when I finish this, I've not seen it but it's got a good reputation
MSmits: it takes a while to get into it, but then again so did breaking bad
Astrobytes: I can live with that
MSmits: I remember watching the first 2 episodes or so, right around the bath thing and then quit for months
JBM: breaking bad took me no time to get into
MSmits: when i got back i watched all of them in like a week or two
JBM: that 1st episode was outstanding
Astrobytes: Did you ever get around to watching Picard MSmits?
Astrobytes: Yes I was hooked instantly too JBM
MSmits: yes, but you;ve got to realize the genre is very far from what i normally watch. But breaking bad is just so good, even if it's not your genre you have to see it
MSmits: Astrobytes i havent seen it no, because of the streaming service
Astrobytes: Oh just torrent it already :P
Astrobytes: (I never said that)
MSmits: i'm patient
MSmits: i will see it for sure
MSmits: I watched all star trek ever made
MSmits: most of it multiple times
Astrobytes: Can't be seen to be advocating smoking DMT and downloading torrents in the same chat stream JBM
MSmits: wait except the original series btw
Astrobytes: You will love Picard MSmits
MSmits: no doubt
Astrobytes: JBM: http://chat.codingame.com/pastebin/0ab970de-ad1c-4b66-bed2-b3363a4097bf
Astrobytes: ikr
Astrobytes: Left? As in position or alive?
jacek: so did struct said what he wanted to say?
Astrobytes: Right
JBM: just about nothing in this transcript makes sense
struct: What do you mean jacek?
Astrobytes: Yes, that's why I sent it to you JBM ;)
MSmits: struct PM-ed me jacek
MSmits: it's fine, no worries
JBM: could you gather the entire I/O? :/
struct: Yes, very secret algorithm
MSmits: just some mcts questions
struct: This will make me rank 1
MSmits: sorry, ruined the mystery :P
JBM: i can't deal with it now, but I kind of have to investigate that level of bugginess
struct: Anyone want to write statement, last chance
struct: I have good deal
Astrobytes: Can I just give you the seed JBM?
MSmits: struct, people can suggest edits as well cant they?
struct: suggest edits?
JBM: can i run both your arena bots?
struct: of the statement?
Astrobytes: Yes
MSmits: yeah
Astrobytes: seed=-3777575510120129500
struct: Yeah
struct: Should I take design one of the board or design2 for the statement?
Astrobytes: And both bots are deterministic it seems
MSmits: didnt see the pictures struct
struct: https://i.imgur.com/o1xMaRQ.png
struct: Both are available, so player can choose
Astrobytes: Much better font struct
MSmits: I like the 2nd one. The lines confuse me.
MSmits: (the lines on the first one)
struct: Line takes time to get used to it
Astrobytes: I prefer the lines, I see it as a graph
struct: But for the statement
MSmits: mmh ok
struct: might be easier to see on 2nd
Astrobytes: *see it more obviously as a graph
JBM: hey at least the viewer is correct
Astrobytes: Cool
struct: viewer is correct but referee isnt?
JBM: viewer is output by ref so...
JBM: not sure what ref gave the gamemanager as a result
Astrobytes: The wrong info? (just guessing there)
JBM: was it in your lastbattles?
JBM: it's the correct viewer + ranking
JBM: just the game text is wrong it seems
JBM: gameManager.addToGameSummary(GameManager.formatErrorMessage("Troll destroys " + gameManager.getPlayer(winner).getNicknameToken()));
JBM: yeah tjhat's kind of obviously wrong
Astrobytes: hahahaha
Astrobytes: No worries
JBM: you'll get nice git commit message credits
Astrobytes: Yay, fame, at last.
Daher: python is for the dumb
Zenoscave: What Multi's y'all doing?
struct: everyone doing SR Zenoscave
Zenoscave: SR?
struct: https://www.codingame.com/multiplayer/optimization/search-race
Astrobytes: Search RAce optim
Zenoscave: Nice an optim that isn't from euler ;)
Zenoscave: gz Illedan
Astrobytes: And testing JBMs Troll vs Castles. Well, I am.
Astrobytes: Also, pewpewpew
Zenoscave: pewpew
struct: also you are not allowed to use Sim on SR
struct: you must use true ai
Zenoscave: eh...
Zenoscave: how is that enforced
Illedan: thx, it is not :P
Zenoscave: Good to hear! I will try it
JBM: (I tried that with TvC, it kind of didn't work long)
Astrobytes: What, the One True AI?
Astrobytes: Or honour?
Zenoscave: honour
JBM: in context, in was honneur
Zenoscave: I'll try no sim AI
Zenoscave: does RL count as sim?
JBM: but no can do, in 5 minutes we had celebrity-account AIs throwing negative amounts of stones
Astrobytes: "celebrity-account AIs " :D
Zenoscave: I'm losing track of these acronyms
Zenoscave: tvc is what now?
JBM: Astrobytes: your bug is fixed in "production"
JBM: https://www.codingame.com/ide/demo/84238477bff184d6523d9b7b6a0dd0790496d8 Zenoscave
Astrobytes: Nice
struct: At least my ported games dont have acronyms
JBM: i'll probably change the title at the last minute just to troll everyone
struct: Do I need to cleanup Yinsh code?
struct: answer is no rigt?
JBM: depends if you intend to publish it
JBM: we'll make fun of the referee no matter what
struct: Noone will read :D
JBM: the code can either distract us or make us find more fun stuff to poke at
Astrobytes: We wil lread it at some point
Astrobytes: *will read
Astrobytes: fking codeine
struct: Should be over 2k liens of mess
struct: Maybe not
struct: I think its less than 1k?
struct: wtf
Astrobytes: It's all in one file?
struct: Good idea
Astrobytes: What? 1 file or splitting it up?
struct: 1 file
struct: around 1.2k lines
struct: How did I take this long
Astrobytes: Because of stuff. And also things.
MSmits: gonna get some sleep and stuff
MSmits: gn
struct: gn
Astrobytes: gn
struct: I must finish today
struct: then ill be gonne for another 3 months
Astrobytes: Why
struct: Im joking i wont run this time
Astrobytes: Yeah yeah, we'll see :P
struct: Longest user name is here?
struct: I need him to teach me english
struct: so i can write the statement
Astrobytes: Most complicated username?
struct: that
Astrobytes: MCU is a she iirc
Zenoscave: Astrobytes PEBCAC?
Zenoscave: not PEBCAK?
Astrobytes: Both valid, this one is between Computer and Chair/Chair and Computer
Zenoscave: The keyboard is between those!!! You need a new keyboard huh
Zenoscave: ;)
JBM: what's pebkak in that contexT?
struct: Is this US vs UK thing?
Zenoscave: problem exists between keyboard and keyboard
Astrobytes: Nah. The keyboard is only the abstraction of the real problem
JBM: Zenoscave: welcome to TvC!
Zenoscave: Thanks!
Zenoscave: It's pretty simple
Zenoscave: not gonna share
Zenoscave: Just watch a few replays. it's obvious
JBM: you tell me your secret, i give you the boss's alg
Zenoscave: There's a boss?
struct: JBM whats your secret to have people play your game without writing a statement?
Astrobytes: Hand drawn troll
Alshock: struct wonderful graphics
Zenoscave: Promotion
Zenoscave: and a hand drawn troll
struct: thanks
struct: im writing statement Alshock
struct: dont read that one, that one is from yavalath
struct: :p
Zenoscave: JBM who is the boss?
Alshock: who read statements anyway?
Zenoscave: I just shoot one more rock than the opp did last time.
JBM: it's actually rather long
Zenoscave: lol
JBM: the current boss sends random(0,stones)
JBM: the former boss just sent 1
Zenoscave: ...
struct: true ai
Alshock: it wasn't when we started playing your game so his point is valid
Alshock: when you told us about that game the statement was 3 lines long
pb4: including one line complaining that nobody would read it anyway
Zenoscave: That's great
struct: statements overrated
Zenoscave: I think that line should be brought back
struct: just read the referee
Alshock: @pb4 I didn't read up to there :P
struct: let me se
JBM: i don't know how the java people ever ship anything
Alshock: shipping? https://www.youtube.com/watch?v=nCeFbK-WEVE
struct: you even have comments
AntiSquid: "one line complaining that nobody would read it anyway" the alternative is to write something cheeky at the end like "this is best game ever, glad you like it" and you will know when people read the statement because they will react to that line
struct: on your code
struct: pro
struct: How am I supposed to write all this in the statement?
struct: http://www.gipf.com/yinsh/rules/rules.html
struct: copy paste?
AntiSquid: the page you linked is more wordy than it needs to be
AntiSquid: you could definitely trim quite a bit of text
struct: no JBM
AntiSquid: remove pictures and just add the link at the end as "more detailed explanations"
struct: Its the first link of my statement
struct: lol
struct: first or second line of the statment
struct: is this
AntiSquid: didn't check your game recently
AntiSquid: just saw the fancy new pic
struct: I shouldnt make my default AI random should I?
struct: that is bad for players that want to test right?
Zenoscave: Never make the default random
Alshock: that depends, is it functional?
Zenoscave: never!
struct: ah its not
struct: int r = new Random(412).nextInt(actionCount)
Alshock: ideally, give your random a fixed seed
Alshock: then it's ok
Alshock: Idk if that's what the 412 does, but if so I'm affraid you have only one number
struct: yeah its bad
struct: but actionCount changes depending on turn
Alshock: you should make randomgenerator only once in your programme
Alshock: and then call it
struct: you are right
AntiSquid: i am 16/32 in the troll game, with D lang, maybe if i add 5 lines i get #1
AntiSquid: using default code btw
struct: I have been training NN for that game for 5 days
struct: but still no t good
AntiSquid: try different model
pb4: struct ?
struct: Sorry I missed /s
struct: Today is my troll day
pb4: Thought so
pb4: though
Zenoscave: pb4 yw
pb4: https://github.com/pb4git/Nash-DQN-CSB-Article
pb4: could apply
Zenoscave: bumped you to 1st in TvC
pb4: Nice Zenoscave :)
pb4: Thanks !
AntiSquid: seriously, with 4 inputs each turn why not try to train a perceptron? :D
AntiSquid: ok 5 inputs, 1 output
pb4: You'll notice I didn't add the /s ;)
Astrobytes: I'm attempting an alphabeta type thing in TvC but I fucked it up because it's late and codeine
Zenoscave: ... perceptrons were used ever?
AntiSquid: you should get something that performs statistically well
Alshock: AntiSquid what's the fifth output?
Alshock: input*
AntiSquid: simple game, doesn't need much more does it Zenoscave?
Alshock: I see aStone/aDist/bStones/bDist
AntiSquid: roadlength
Zenoscave: Fair
Alshock: it's aDist+bDist
Astrobytes: Since I'm now wasted mentally and in TvC I shall retire to the sofa
Astrobytes: gn/bn to you all
AntiSquid: didn't look properly at the while loop
AntiSquid: yes 4
AntiSquid: roadlength, trolldist, stones, opstones
AntiSquid: bye Astrobytes
Alshock: would a NN fair better with a fixed roadlength than with the opponentDist ?
Alshock: fare*
Alshock: I must admit my total lack of knowledge in that area
struct: I have no idea too, never tried NN
AntiSquid: https://miro.medium.com/max/645/0*LJBO8UbtzK_SKMog
AntiSquid: trying to find an article
Illedan: TvC?
Zenoscave: the troll game
AntiSquid: troll vs castle
fangskills: anybody know how to do the temperatures puzzel
struct: use abs()
struct: absolute value
struct: |-3| == 3
struct: you need to find the closest value to 0
fangskills: i only need the "Only Negative" test case to be completed
struct: What are you outputing on that test?
struct: Share the code
fangskills: it suppose to console log it but idk why it doesn't do it
struct: Ill try to see whats wrong
fangskills: want all of the code?
struct: yeah
struct: just paste it here
fangskills: http://chat.codingame.com/pastebin/f62a5e1d-5884-4ade-bf41-7cbac3d93ec5
struct: fangskills you want the solution
struct: or want me to tell you whats wrong?
fangskills: tell me whats wrong
struct: So the problem is
struct: that if positive temps is empty
fangskills: and give me the solution by giving me the link to the code
struct: you will never have a if for that
fangskills: thanks for that hint!
fangskills: i been using if statements all the time lol
struct: http://chat.codingame.com/pastebin/56fd52bb-fbbe-46ff-9208-0278155c0a06
Alshock: if is love, if is life
struct: But check other players solutions
struct: There are better ways to solve this problem
Alshock: js just returns undefined when you access an out-of-range element I guess?
struct: yeah Alshock
struct: at least i think
struct: Javascript is a mess
Alshock: http://chat.codingame.com/pastebin/de283467-0126-4f05-851e-c9ad35e6385f
Alshock: undefined < x => false, so we can slightly rework that
Alshock: I just realized pasting a working solution in cg chat might not be the most clever move ever
Alshock: hey, I realized before actually doing it :P
Alshock: don't throw stones at me or I'll turn your castle upside down
Illedan: Do you guys want debug lines for SR? Like print some special message and I draw red lines on the canvas for whatever you want to debug?
Alshock: hummmm, that'd be useful, but only to draw dicks on the screen IMHO
Illedan: You can still do that with skidmarks..
Alshock: why do you think so many people play your game?
Illedan: DaNinja around?
Illedan: It's not many players yet
Alshock: We Are Billions!
struct: Are there sites that make writing html easier?
Alshock: Sorry I let you ask real questions to real player and I stop that noise xD
struct: I guess that can save me a bit of time on statement
Illedan: Any of top 3 on SR around?
Neumann: Sure
Neumann: Is the CP "collision" calculated like regular collisions ? Or once per turn only ? My code think he got the CPs but doesn't, sometimes .. https://www.codingame.com/share-replay/468778260
Illedan: It is multiple times per turn
Illedan: I rewrote the collision part because of DaNinja's bug: Not live yet, but in this Arena for testing: https://www.codingame.com/contribute/view/5075a2f7cec5d95972367597b8ae3fe7c79f
Neumann: What was his bug ?
Illedan: https://www.codingame.com/replay/468603470
Illedan: Frame 35
Illedan: I didn't move the car when checking for multiple collisions
fangskills: On the "Only Negative Numbers" why does it send multiple numbers but expect one negative number
Alshock: errrr, can you link the puzzle page? ^^
Illedan: Neumann, I use the same collision check as in Bit Runner
Alshock: oh ok you're on temperatures
struct: you mean on temperatures fangskills?
struct: If there are multiple negative
Alshock: well you need to output the temperature closest to 0, there's only one
struct: only*
struct: you must output the one closest to 0
struct: yeah
fangskills: okay i got it then lol i was confused for a sec
Neumann: Yeah I've got the same
Illedan: Do you truncate positions or round them?
Neumann: round
Illedan: They are truncated
Neumann: /flip
struct: :D
Neumann: Weren't they rounded in CSB/BR ?
Alshock: ah ah ah
struct: yes on CSB at least
Illedan: CSB = rounder
Alshock: position is rounded and speed is truncated
Illedan: BR = truncated
Alshock: iirc
Illedan: Else the game wasn't 100% equal
Zaphus: Does it bug anyone else that in many of the Bot (and contest) competitions, the timeline bar is the same yellow as player 1, so you can't easily see where the actions took place at the end of the playthrough?
Illedan: truncation helped for rounding errors
struct: Illedan you dont touch angle do you?
Illedan: Angle is rounded
struct: (╯°□°)╯︵ ┻━┻
Illedan: but that is no problem with EXPERT movement
Alshock: @Illedan on SR your expert output takes a floating point?
struct: it is because I do one thing
Illedan: I'm not going to have an internal double angle that you don't know about :P
struct: I take the angle to next checkpoint
struct: at start
Illedan: You can't select the angle at start
Illedan: Your angle are selected
Neumann: Ah much better https://www.codingame.com/share-replay/468780613
Alshock: sorry the question is a little lazy, but that one is on my todo list for now :P
Illedan: You are forced to be towards the next CP with a rounded angle at start
struct: at turn 0 the angle is rounded right?
Illedan: Nice Neumann :)
Illedan: ye
Illedan: before you get it
Illedan: Did it help on submit Neumann?
Neumann: Idk, I'm out of them
Illedan: lol
Neumann: ^^
Alshock: Correct me if I'm wrong, but having an internal double angle would be perfectly ok since position and speed are rounded. No?
Illedan: I want players to 100 % know what their state is.
Illedan: It is double internal, but rounded.
Alshock: yeah but then they need to round their own angle while if it wasn't rounded they could not get a different position or speed from yours, but wouldn't have to round their angle
Alshock: that's very weird if you target points to have to round the angle actually
Illedan: Updated the bugfix now and added a cover photo from Neumann driving :D
Neumann: :sunglasses:
Illedan: Can you try to run some tests Neumann, just to see if something looks off? should be good, but nice to have 1 extra little test :)
Neumann: Sure. I just need to F5 the IDE ?
Illedan: yeah
Neumann: Seems fine
Illedan: Thx :)
Neumann: What was the submit limit already ?
Neumann: 25/5h or so ?
Snef: i heard 25
Snef: not sure about 5h
Illedan: I'm off gn.
struct: gn
Illedan: Ping me on Discord if anything crashes and I'll wake to fix xD
struct: Time to hunt bugs
Snef: SUB 11K finally
struct: grats
BigBadStopwatch_4678: im noticing a lot of languages look very similar to java
BigBadStopwatch_4678: makes me think i should focus on java which would make it easy to branch out later on
BigBadStopwatch_4678: i tried Rust, i got creamed
Snef: you should try D, the best language to create real AI
BigBadStopwatch_4678: all the AI stuff i know is in python atm
struct: C++ master race
Snef: python isn't the best to advanced stuff
Snef: except for machine learning
BigBadStopwatch_4678: yeah i know ML, i dont know other areas outside that so much
Snef: try to train a model for Search Race puzzle
BigBadStopwatch_4678: well main goal now is just exposure to other languages, i might look into getting back into stats eventually
struct: c++, C#, java
struct: rist
struct: rust*
Snef: then do search race in c++
BigBadStopwatch_4678: i was trying to pick a language that will bring the most utility, a lot of langs look super close to Java and they use java at my company
struct: I feel like c++ is a must
struct: So pick java
Snef: the thing is that you won't learn things you'd you at work or not much
Snef: still good to learn the basics if you're at this stage
Snef: you'd use*
BigBadStopwatch_4678: yeah im just now tending towards it after seeing how it resembles so many other langs. I tried rust and i got crushed by the syntax quickly, managed 2 puzzles but i think it assumes a lot more low level knowledge
Snef: to start i'd also recommend c++
BigBadStopwatch_4678: seems like C background would help undersatnd rust
BigBadStopwatch_4678: well, im already pretty fluent in python
Snef: i'd say C is close to c++ which are close to c# java..
BigBadStopwatch_4678: really i was looking for a strong complement to python. C was a contender. Also rust due to the FFI. Java as well.
BigBadStopwatch_4678: with C++ im worried its a bit too specialized these days
BigBadStopwatch_4678: for like high performance environments
Snef: sure
Snef: i use c++ here i don't use it anywhere else
struct: Also the c++ code i write here i would not write anywhere else
BigBadStopwatch_4678: how long have you been using it here
Snef: 2 - 2.5 years
BigBadStopwatch_4678: that's enough time to be ready for an interview i imagine
Snef: not really
Snef: because as i said, the things you learn here aren't the ones required for work
BigBadStopwatch_4678: well, not if you only do one puzzle a year. lol
Snef: it's not about number of puzzle
Snef: it's about what you're doing
BigBadStopwatch_4678: i was thinking more about fluency with the language more than specific libraries, even third party
BigBadStopwatch_4678: if youre fluent in the language and you have some subject matter expertise, in machine learning for instance, then its easy to go from there
Snef: hmm if you know a bit C then you're good for c++
BigBadStopwatch_4678: but surely after 2.5 years you're pretty fluent in the language
BigBadStopwatch_4678: maybe you havent solved all the various problem types, but you should be able to read most code in that language after that time period i'd guess
BigBadStopwatch_4678: whats your main language since you only do C++ here?
Snef: well i haven't really a main langage
Snef: i have done much java at first
Snef: and c
BigBadStopwatch_4678: best language i guess
Snef: then mostly javascript for work and c++ here
BigBadStopwatch_4678: ah interesting..
BigBadStopwatch_4678: i spend like 8-10 hours per day doing python for the past yr
struct: thats a lot of time
BigBadStopwatch_4678: not a developer but my team lets me code since ive built some good stuff
BigBadStopwatch_4678: yes, after all that time though i have an itch for another language. happy to find this place.
BigBadStopwatch_4678: python is too quick to write, i dont see myself using any other language except to implement some slow process in that language for instance, would be a nice way to get started
BigBadStopwatch_4678: like i said was trying Rust, but that syntax kicked my butt quickly
struct: Site has plenty of stuff, many puzzles, but the best part for me is mutliplayers
struct: where you can code your own ai
struct: for plenty of games
Snef: yea
BigBadStopwatch_4678: that sounds pretty fun
Snef: and i guess you use more of a language in multi than in puzzle
struct: you can even create your own multiplayer games
struct: Also the site hosts contests
struct: usually 12(?) days long
struct: Snef what do you use on SR?
Snef: I think it look like a ga but as i've done it myself without really looking at paper... it's kinda heavily customised
struct: I never tried GA
struct: I guess I can try it here
Snef: i finished to setup my local arena today so i'll try to play with it for the next days
struct: Should be easy to mix it with avx
Snef: i never did avx
Snef: seems hard
struct: It seems harder than it is
struct: After I finish the statement Ill write avx sim for SR
struct: if it works ill send it to you so you can take a look if you want
struct: Just to get an idea
Snef: why not
struct: Do you use doubles or floats on your sim?
Snef: floats
Snef: and int
struct: works fine?
Snef: what do you mean by that
struct: referee uses doubles
struct: Its not the same precision
Snef: i'm not sure how much it affect my sim
struct: I have avx already built for floats
struct: not for doubles
Snef: as you round every turn i guess there isn't much time to get big offset between double and float
struct: if number is 1.4999999-1.50000000
struct: they are truncated right?
struct: not rounded
Snef: i'm not sure but i think so
struct: yeah, its truncated, for speed and velocity i checked refered
struct: referee
Snef: but they round x and y
struct: on csb
Snef: and trunc vx and vy
struct: on sr illedan trunc pos and vel
Snef: oh
Snef: is truncate from java doing same thing on negative value
Snef: as a (int) cast in c++
struct: should be
struct: yeah its the same
struct: intel intrisics page is broken again for me :/
Snef: 10863 yay
struct: What depth?
Snef: i'm trying differents depth i'm at 18 here
struct: damn
struct: seems a lot
Snef: still not sure which is the best
Snef: Neum ann said he has half of my initial so around 10-12 i guess
struct: I want to try depth 8 and try all possible moves
Snef: why bruteforce ?
struct: I mean out of a move of 6
struct: 6^8
struct: not very high number
struct: should be plausible
Snef: better have ga with unlimited sets of moves
Snef: 7400^18 lel
Snef: 7437^18 *
struct: lol
Snef: you can just adapt the probability of each mvoe being selected
Snef: but still if one turn you wanna do angle 6 with thrust 176 it can be good
struct: is anyone hardcoding yet?=
Snef: i don't think so
Snef: i guess soon there will be more and more sub 11k and some of them should be hardcoded
Snef: i could also like run my ga locally for 100*time and use those moves but it's not really fun :p
struct: hardcode is true ai
Snef: 50 validator => 50 if => true ai
struct: Writing the statement is so boring
Snef: it should be quick tho no ?
struct: kinda, but I usually add images
struct: explaining
struct: https://www.codingame.com/ide/puzzle/yavalath
Snef: yea i see
struct: how should I phrase this
struct: you have the option to disable getting legal actions
Snef: what is the benefit of disabling it ?
Snef: more time per turn ?
struct: yeah
struct: Sometimes you wouild get 200+ actions
struct: in 1 turn
Snef: oh
Snef: i see
Snef: what about something like that:
Snef: You are free to choose whether you want to receive the legal moves or not by sending yes or no....
struct: thanks, sounds good ;)
Snef: i'm not sure if you have to keep "the" beofre legal mvoes tho
Snef: before * moves *
struct: On the first turn you are free to choose whether you want to receive legal moves or not, to do this you simply have to output "no".
struct: He can output anything else, doesnt need to be yes
struct: so i guess ill leave it like this :pç
Snef: sure
Snef: and in stub you print "yes" right ?
struct: yeah
Snef: seems good then
struct: "Once a ring moves it will leave a marker of his color behind."
struct: his doesnt sound right :/
Snef: Once a ring as moved, his previous position will be colored ?
Snef: has*
struct: Thanks Snef
struct: I cant sleep until I finish this
Snef: its instead of his also
struct: wall of text
Snef: time to sleep
Snef: cya struct
struct: gn