Chat:World/2020-06-09

From CG community
Jump to navigation Jump to search

MostComplicatedUsername: Fyslexic Duck

MostComplicatedUsername: Figure it out :)

AbundantPuddle: i[0][0]=i[0][1]&&i[0][1]=i[0][0]

WINWINWIN: Yes?

WINWINWIN: change = to ==

Default avatar.png hjhjhjhjhjhjhjhj: join clash plz

AbundantPuddle: It wasn't comparitive, it was assigning.

AbundantPuddle: It was in regards to MostComplicatedUsername's comment.

hydroglicerol: what is the highest level you can reah and is 11 considerd good for 1 week?

hydroglicerol: is noone up at 2:16?

WINWINWIN: Definitely hydroglicerol

hydroglicerol: ahh

WINWINWIN: lvl 11 in one week is excellent

hydroglicerol: what is the highest?

WINWINWIN: Not sure what the highest somone did in 1 week is, but eulerscheZahl is at lvl 52

hydroglicerol: ahh wel back to the grind thanks for the encouragement

LelouchVC2: C++ definitely runs slower on this site

LelouchVC2: I just did a CoC in C# and it ran through those tests like they were butter

hydroglicerol: what is the anime in your pic if it is anime?

WINWINWIN: LelouchVC2 even if it does run slower, it is not noticeable by running test cases

hydroglicerol: and yes C++ is definatley slower here

hydroglicerol: I feel notacible lag when I use it

hydroglicerol: but I never really use it

hydroglicerol: I only used it to give that nintendo sponcerd puzzle a try

hydroglicerol: which was meh

jacek: LelouchVC2 do you have optmization pragmas?

struct: It's a must for CoC

WINWINWIN: I still dont understand how he notices a 1 ms difference in test cases

struct: he probably is talking about compilation time

Default avatar.png NicatSadiqov: I am not a good speller. Sometimes I mix up the spelling of words and sometimes I even forget to put periods. Please count the number of words I have misspelled and the number of periods I have missed. Input

Default avatar.png NicatSadiqov: is it now hard question

WINWINWIN: CoC puzzle?

Default avatar.png NicatSadiqov: task require that find wrong spelling words and missin periods

Default avatar.png NicatSadiqov: missing periods

Default avatar.png NicatSadiqov: number of them

Default avatar.png NicatSadiqov: is it possible in c++

Default avatar.png NicatSadiqov: i can fund number of periods put wrongly

Default avatar.png NicatSadiqov: but i dont know that of words

Tiramon: omg old code is always a mess, looked a pr2 and shortend the code by 1/4 without changing anything on the result

Nerchio: :D nice

Tiramon: hm somehow it is even working better, seems like i had a bug and fixed it

Tiramon: is something broken on submit right now? it does show me alot of latest battles but no progress in latest battles or leaderboard ... i'm listed as stable while i just submitted [CG]Thibaud

Tiramon: currently submitted pr2

Nerchio: idk it was working for me but it broke a few times in the last days

Nerchio: so maybe it happened again ;p

Tiramon: you are listed as computing with progress ...

Nerchio: i have so many bugs in my code that are off-by-1 mistakes

Nerchio: :x

Nerchio: yeah i just submitted so its possible

Tiramon: also after 80 battles my rank still didn't change ...

Nerchio: sometimes it's slow ;p

Tiramon: never seen it that slow before

LelouchVC2: The anime in my picture is Code Geass

LelouchVC2: And Idk what to optimizations I can enable on Codingame

LelouchVC2: I used C# and those tests flew by. With C++, it's slower.

[CG]Thibaud: it has been reported on Discord too

[CG]Thibaud: I don't know what's happening. These random freeze/slowness of submits are not normal

Tiramon: that's what i expected, i started a forum thread about it

Tiramon: and as i said it is not just slow, it is just as i never submited...

jacek: this is what you get for organizing contests

Tiramon: oh nice now there are only 8 multiplayer games listed at bot-programming ...

Tiramon: only the ones havn't started yet

leojean890: I encountered the same problem recently : you can see new fights after submitting but you have to wait for 1 hour before the leaderboard update

leojean890: in bot-programming you have 2 categories : already and not already begun multiplayers are split in 2 distinct parts of the web page

Tiramon: 1h passed nothing changed

Nerchio: yea :9

leojean890: I think that it can be more sometimes depending on the situation.. Do many people submit in the same time in PR2 ? and there are no leagues .. Problems may be the same than those encountered at the beginning of the PAC contest

Nerchio: it updated i think

Tiramon: at least it changed

Tiramon: 1,5h

Tiramon: at least this submit it is updating

Astrobytes: I had to just give up on it last night

AntiSquid: give up what?

AntiSquid: leaderboard stuck again?

Nerchio: not for me i think

Tiramon: damn i have a bug in my new distance calculation

Astrobytes: Yeah it's been sticking a lot AntiSquid

Default avatar.png Ace-Code: :D

Default avatar.png Ace-Code: :D

Default avatar.png Ace-Code: :D

Default avatar.png Ace-Code: hii

JKSTUDIOS2020: hi

Tiramon: ok ... at least the current code isn't timing out anymore

Default avatar.png SamyErd: *

Nicolas.ElFegir: wsh SamyErd ca parle au monde ?

Default avatar.png SamyErd: Yes !

Nicolas.ElFegir: y a qui pour bouger sur annonay les asiat ?

Default avatar.png Matthieu_gorisse: coding is for bitch

Nicolas.ElFegir: all coders :

Nicolas.ElFegir: :rage:

Default avatar.png Cihan07: what the fuck man

clement-soubeyrand: nico ta réussi sudoku

clement-soubeyrand: ,,,,,????????

clement-soubeyrand: ??????

Nicolas.ElFegir: non impossible

Nicolas.ElFegir: dites vous dans tous le monde on est les seuls actuellement à parler sur le forum

clement-soubeyrand: bravo!!!!!!!!!

Nicolas.ElFegir: des mexicains chaud pour un tacos jeudi ?

clement-soubeyrand: :peanuts:

Default avatar.png Loic_Astruc: Do u need help? I have ever finished The descent in Python if you want

Default avatar.png Val.thlt: ouey de ouf

clement-soubeyrand: :taco:

Default avatar.png Val.thlt: :heart_eyes:

Astrobytes: Tiramon, AntiSquid, Magus... trash needs taking out

Default avatar.png Val.thlt: si il savais a quel point on s'en branle de son cours :poop:

Magus: funny, they use their real firstname + name

Astrobytes: schoolkids I guess

Magus: easier for the teacher to spot them :D

Astrobytes: Indeed!

Default avatar.png Loic_Astruc: what a mistake

Default avatar.png Ace-Code: :D

AntiSquid: sorry was away Astrobytes

AntiSquid: damn pacman is still random af, before first submit today #155, then #200 or so and now at #441 same bot? seriously? :D

wlesavo: fix it?

eulerscheZahl: "At the risk of sounding elitist (again), who cares about precision in lower ranks"

WINWINWIN: :D

Tiramon: oh spamers and i missed them ...

WINWINWIN: spammers?

eulerscheZahl: they got taken care of

WINWINWIN: Hey Tiramon is your NN going well?

Tiramon: did try it again by now, currently wasted a lot of time on social stuff and during my vacation i will spend most of the time on the nanodegree and renovate my son's room

Tiramon: maybe i find some time to train it again or try a different setup

Tiramon: *didn't

AntiSquid: "lower ranks" lol

WINWINWIN: Hope it works, would like to see how NN compares to minimax.

AntiSquid: same issue with top 100 even in legend, no precision

AntiSquid: if you only top 3 is taken into account, then might as well throw the ranking out the window

Tiramon: 'lower ranks' is for you everything > 10 ?

WINWINWIN: >5

AntiSquid: don't ask me Tiramon, i didn't post that quote

WINWINWIN: To train a NN dont you just need to identify a set of datapoints?

WINWINWIN: Are there any sites with papers on NNs

WINWINWIN: ?

inoryy: what are you training it on, Tiramon?

Tiramon: oware

Tiramon: has a simple state and action set

Tiramon: so my DQN should learn nicly

Tiramon: took some time to avoid late game invalid actions, but last weights learned it after i kicked it a few times

Tiramon: oh nice ... you did it to deepmind?

AntiSquid: deepmind should try to use their skills to find a cure for COVID :p kaggle has a COVID related challenge

Default avatar.png AlmightyComet_aaa: hello

inoryy: Tiramon I work at DeepMind, yep

inoryy: AntiSquid https://deepmind.com/research/open-source/computational-predictions-of-protein-structures-associated-with-COVID-19

Tiramon: nice =) my brother is stil hoping to get a job there ;)

inoryy: I definitely recommend at least applying, doesn't hurt to give it a try... well maybe after covid19

jimmyjansen93: Hey guys im trying to compete in CSB, but im running into a wall with my "AI". I am trying to look through the documentation provided in the challenge and both smitsimax and genetic algorithm sound very interesting. For both I understand the theory of how they are supposed to work, but I am having trouble understanding the implementation side. Without spoiling anything, can anyone help me understand the practical side of this?

WINWINWIN: jimmyjansen93

WINWINWIN: http://files.magusgeek.com/csb/csb_en.html is an excellent documentation on how to practically apply a GA

WINWINWIN: Feel free to ask if you have any doubts

CommanderCero: What do you mean. Both algorithms rely on an simulation, meaning you probably have a class or methods that mimic the original game. In the best case you have a method like "step(action)" that advances your game forward one step. You can then pass the actions to the step function, which are provided by the algorithm you use

jimmyjansen93: Thanks @WINWINWIN maybe going through that one again will help me more than the first time I read it

Astrobytes: Please be gentle with our poor bots MSmits :P

Illedan: oware? :P

WINWINWIN: jimmyjansen93 this is inoryy`s functional search bot

WINWINWIN: https://github.com/inoryy/csb-ai-starter

Astrobytes: Yeah :D

WINWINWIN: It is a fully functional bot written in C++

WINWINWIN: Astrobytes working on PR2?

Astrobytes: No, working on my Oware eval today

WINWINWIN: Msmits crushing everyone?

Illedan: Not 100%

Astrobytes: Not sure what changes he's made other than using tanh

jimmyjansen93: Thanks so much @WINWINWIN seems like that will already help me a lot

Astrobytes: jimmyjansen93 Just make sure you don't blindly submit the starter bot

jimmyjansen93: of course

WINWINWIN: jimmyjansen93 you cannot apply that yet, you are only in Silver

jimmyjansen93: I am anyway writing JS, so blind copy is not in it for me, but I am looking for practical examples of how to implement genetic algorithm as the theory is a bit to much theory for me and I cannot seem to grasp it

jimmyjansen93: yea, thats fine, I will get to gold

WINWINWIN: In gold, a lot of new info is given allowing for a GA.

WINWINWIN: In Silver you dont know all of the checkpoints

Astrobytes: the Magu s document will help you, there's also a GA tutorial (in python) on tech.io

jimmyjansen93: Ah I had plans to log the points and for the second and third round optimize my routes for them

WINWINWIN: jimmyjansen93 just get to gold in 10 mins using the targ - 3vel

WINWINWIN: dont think GA is possible in Silver without knowing CPs, you cannot evaluate a solution

jimmyjansen93: targ - 3vel? what is that?

WINWINWIN: "Magic formula" for CSB, takes all the way to gold

WINWINWIN: target.x - 3 * pod.vx

Default avatar.png Matrix007: heloo guys need some help

WINWINWIN: target.y - 3 * pod.vy

WINWINWIN: Yes Matrix007?

Default avatar.png Matrix007: do u know where i could find the problems for clash of code

Default avatar.png Matrix007: are they random

Default avatar.png Matrix007: i want a solution for a particular problem

WINWINWIN: They are given randomly in the IDE

Default avatar.png Matrix007: is there any collection of those where i cold check

WINWINWIN: Matrix007 are you in that clash now?

Default avatar.png Matrix007: no

Default avatar.png Matrix007: i did it a week ago

Default avatar.png Matrix007: it was the code to print a window

WINWINWIN: Ahh. It is not possible to find the solution

Default avatar.png Matrix007: oh

Default avatar.png Matrix007: ok guuys

Default avatar.png Matrix007: i just wanted to check whether i had understood the problem correctly since it was having only one example

struct: Do you remember keywords from the clash?

struct: or part of the statement

Astrobytes: Matrix007 http://eulerschezahl.herokuapp.com/codingame/puzzles/search/

Astrobytes: Search for it there, you can then work through the problem to find the answer in your own time

WINWINWIN: Wow

Default avatar.png Matrix007: thanks so much astrobytes

Astrobytes: Hey thank euler ;)

Default avatar.png Matrix007: is that project euler?or an username

Astrobytes: username, it's his app

WINWINWIN: https://www.codingame.com/contribute/view/3525dbd0ddde1375af211b2469bebc73fc7

WINWINWIN: Im guessing that this was the puzzle you wanted Matrix007

AntiSquid: project euler was there before euler's number, right Automaton2000 ?

Automaton2000: and i went to bed

Default avatar.png Matrix007: wait a bit guys the websites are not opening

Default avatar.png Matrix007: probably my internet

Default avatar.png Matrix007: yeah thats the one winwinwin

Default avatar.png Matrix007: thank u

WINWINWIN: np

Default avatar.png Matrix007: so i am bit new here i cant focus on a problem to think its logic or its takin me more time any tips?

Default avatar.png Matrix007: BTW i know its gonna be practice but i just cant get myself to sit

Default avatar.png Matrix007: that fire inside me

Default avatar.png Matrix007: how to kindle?

Default avatar.png Matrix007: dont think it as some motivational ques

eulerscheZahl: a random ping by Astrobytes

Astrobytes: Well, it wasn't random

eulerscheZahl: time to finally push a fix to that broken search for badly formatted HTML

eulerscheZahl: the problem is that for user contributions the uploader has control over the HTML. so many ways to screw it up :D

eulerscheZahl: and the search is part of the GET parameter now, i broke the link from above :imp:

Astrobytes: lol, and the CG HTML is weird too right?

Astrobytes: As in what's allowed etc

eulerscheZahl: yeah, some filtering

MSmits: hey astrobytes. I did a cg bench test 400 games vs my arena version, got 87% winrate. Then I submitted it and it was far weaker than Robo and Recurse

MSmits: meaning it was far weaker than my old version

MSmits: so i need to fit the param not against myself, but agains tthem

MSmits: I should have known

Astrobytes: Aha, that would explain it

MSmits: crazy overfitting in this game

Astrobytes: This was just your tanh right

MSmits: yes

eulerscheZahl: and then a 4th player appears and beats you because of your fitting

MSmits: with a param setting that gave 87% winrate vs the old version, losing only 1 in 30 games

Astrobytes: lol, definitely makes for a tempting submit eh

MSmits: yeah

MSmits: I think this game is like this because the statespace is so small

MSmits: so good bots will play nearly perfectly

Default avatar.png Matrix007: what are bots

MSmits: a small param change may lead your bot to make 1 extra mistake or prevent 1 and that could change everything

Default avatar.png Matrix007: npcs?

Astrobytes: Well, perfect CG play at least

MSmits: nah i mean near perfect play

MSmits: like, you make 50 moves and 3 of them are wrong

MSmits: the others are perfect

Astrobytes: Yeah, true

Astrobytes: And yeah, I am fully aware of the parameter sensitivity :P

MSmits: wrong meaning, the solved game value changes from win to draw or draw to loss

Astrobytes: Yeah I gotcha

MSmits: k

MSmits: I think maybe simple games that are near solvable have this property

Astrobytes: Sounds reasonable, my game theory isn't good enough to confirm your hypothesis though ;)

MSmits: yeah it's just an idea

MSmits: I am hoping I will be able to fit the param to get the same strength as my old version, because I really like the sigmoid

Astrobytes: It's useful yeah, but you do have to do some tweaking

MSmits: apparently so

MSmits: btw... Othello might be a good game for CG

MSmits: anyone suggest this yet?

AntiSquid: yes

AntiSquid: but who has time to make it ? :p

Astrobytes: When? I didn't see that

MSmits: We'll force it on struct

Astrobytes: lol, poor guy

AntiSquid: could use checkers as a base to quickly create it, but no mood

MSmits: nah, let someone do it who would like to create a game for CG and has fun doing it

MSmits: it does seem easy

Astrobytes: Othello or making a game?

MSmits: making othello

struct: Hello

MSmits: Othello is a hard game to play as a human

MSmits: hi struct

Astrobytes: *a wild struct appears*

AntiSquid: disagree with that MSmits

struct: isnt it solved though?

Astrobytes: Haven't played Othello in years

Astrobytes: Doesn't matter for CG imo struct

MSmits: It's hard to imagine future positions with Othello

MSmits: harder than chess

AntiSquid: more like machine and human approach are different

MSmits: blind chess is easier than blind othello

MSmits: because the board transforms a LOT

struct: well blind chess doesnt seem hard for top players though

struct: But I see what you are saying

MSmits: its not, but blind othello is

MSmits: it's like uttt in that it seems chaotic

struct: Do you think 8x8

AntiSquid: where did you get this idea from that blind othello has to be harder?

struct: or bigger?

MSmits: a paper

AntiSquid: 12x12 imo

MSmits: 8x8 hasnt been solved yet

MSmits: i think

MSmits: nearly though

AntiSquid: 8x8 is solved afaik and 10x10 is partially solved

MSmits: really?

struct: https://en.wikipedia.org/wiki/Computer_Othello#Othello_8_x_8

struct: Doesnt seem solved according to wiki

MSmits: no its not

MSmits: and the complexity is greater than checkers

AntiSquid: i checked multiple sources a while ago

MSmits: could be wiki is outdated

MSmits: but still if checkers works on CG, so will othello 8x8

MSmits: far more complex than oware as well

MSmits: the only reason oware abapa isnt solved is because of the loops

MSmits: but it has a tiny statespace

struct: I guess I can take a look

MSmits: it was just an idea though

MSmits: dont stress yourself :)

struct: I was porting this one https://boardgamegeek.com/boardgame/272380/shbu but had to stop due to no time

struct: Ill check it on the weekend

eulerscheZahl: but computers can beat humans according to this source: https://xkcd.com/1002/

MSmits: yeah that i knew

eulerscheZahl: which says that humans beat computers at Go

MSmits: thats definitely odl

MSmits: old

MSmits: alpha go lee beat the best human

eulerscheZahl: from january 2012

MSmits: then alpha go zero beat alpha go lee 100%

MSmits: but i guess most go bots cant beat good humans

MSmits: alpha zero is a pretty heavy machine

AntiSquid: "computers may never outplay humans" section - lol what?

MSmits: yeah i dont get that

eulerscheZahl: it's XKCD. thorough research mixed with total nonsense

eulerscheZahl: like https://xkcd.com/485/

AntiSquid: https://www.urbandictionary.com/define.php?term=7%20minutes%20in%20heaven ah just weight until the sex droid market lifts off

struct: So all pieces that are between 2 pieces flip?

AntiSquid: oh it says for kids ... nvm then lol

struct: if they belong to opoonent?=

eulerscheZahl: you

eulerscheZahl: and these flip can cause more flips

eulerscheZahl: you => yes

eulerscheZahl: i'm stupid

AntiSquid: yes struct lots of /flips

AntiSquid: (╯°□°)╯︵ ┻━┻

eulerscheZahl: recursive flipping

inoryy: flip reCurse?

Astrobytes: lol "Brian Greene knitting furiously" :D

struct: Game doesn't seem hard to port

struct: I'll see what I can do

struct: Not sure about board size though

struct: I need feedback

AntiSquid: 10x10 min !?

struct: 100x100

struct: ah wait cant

RoboStac: choose each dimension randomly between 2 and 65. Also don't tell the player how big the board is, so there is an element of fog of war

AntiSquid: lol why !

Astrobytes: And no statement.

eulerscheZahl: and change the board size during the game

Astrobytes: Could be an official CG contest.

AntiSquid: make the stones rock paper scissors types

AntiSquid: or crosswords ?

Astrobytes: Each square is a UTTT board

eulerscheZahl: now you are about to exceed the timelimit

Astrobytes: Hm, OK. Stones have PCR physics

eulerscheZahl: you have to trow colored pillows on a grid to place your actions. but there is some random deviation in your aiming so you can't be sure where you place your action

eulerscheZahl: https://www.youtube.com/watch?v=BXUfnJKLoNQ

AntiSquid: any RAIC contest around the corner or did i miss it?

Astrobytes: lol, nice how you thought of RAIC when random aiming was mentioned

eulerscheZahl: nothing announced yet

Astrobytes: Haha yes, that should be the CG version eulerscheZahl :D

Astrobytes: Why does that exist!

eulerscheZahl: why not?

Astrobytes: Fair

Astrobytes: They should have this in parks. Like shuffleboard

Astrobytes: And giant chess/checkers

AntiSquid: giant chess pieces are easier to grab ?

Default avatar.png JBM: and more fun to steal

eulerscheZahl: they have giant chess checkers where i live

Default avatar.png JBM: you live in an honest country

eulerscheZahl: they call it manhole cover

AntiSquid: steal and place them next to your garden dwarfs JBM ?

Default avatar.png JBM: i can't imagine it lasting long around here :(

Astrobytes: Nor here

Default avatar.png JBM: don't have the garden dwarfs

Default avatar.png JBM: i don't think they'd be compatible with *her*

AntiSquid: well right now the protesters would probably spray paint "this game is racist" on it

Astrobytes: Round here the kids would set fire to them, and breathe in the fumes. Beats the wheelie bins they usually use...

AntiSquid: i thought random fires were more specific to Ireland

AntiSquid: walk through the park and notice lots of random patches of burnt stuff

Astrobytes: No, our parks and general area are full of those

Astrobytes: *the area in general

Default avatar.png ZombieGhost_f5c9: you need to pay for hints??

Default avatar.png JBM: how much?

Default avatar.png JBM: mine are cheaper

Astrobytes: CG Premium Account

Default avatar.png ZombieGhost_f5c9: i dont have option "hints" anymore

Default avatar.png ZombieGhost_f5c9: ah okay

Uljahn: you need premium account, like Automaton2000 has

Automaton2000: but i don't know which one to use

Default avatar.png JBM: multi-premium

AntiSquid: filthy rich

Nerchio: report automaton to cg multiple account

Default avatar.png ZombieGhost_f5c9: where is this shit

AntiSquid: AutomatonNN must be his alt

Default avatar.png JBM: maybe that's one of the perks of going premium

AutomatonNN: what is the problem?

AntiSquid: your alts! delete them!

AbundantPuddle: What does Applied Laser Technologies have to do with CG?

AbundantPuddle: chiikoneko I'm not sure if you'll read this, but I like your answer there.

Nerchio: leaderboard dead again i think

eulerscheZahl: nah, just taking a short nap

AbundantPuddle: Leaderboards never die, they just decompile.

Astrobytes: Looking much better so far MSmits

Astrobytes: As in, like, "holy shit" better

MSmits: its the old version

MSmits: but i figured out why the sigmoid failed I think

Astrobytes: Oh, that explains why it's better

Astrobytes: What do you think?

MSmits: I didnt modifiy the leaderboard version with the sigmoid

MSmits: but an experimental version i never got to work

MSmits: apparently the experimental version + sigmoid was overtrained vs the leaderboard version in some inexplicable way

MSmits: the experimental version opened when i started VS

MSmits: so i thought that was my best

Astrobytes: lol, classic mistake

MSmits: yeah

MSmits: so I have some hope maybe the sigmoid will work with the leaderboard version.

MSmits: it's a lot less ugly than capping the scores

MSmits: will try and see if i can fit a param that works well

Astrobytes: Be more careful next time eh ;)

MSmits: yes :)

MSmits: damn i forgot how strong this was

MSmits: and it's really just some param magic

Astrobytes: Only you forgot. The rest of us have to face it every time we submit :P

MSmits: :)

Astrobytes: It is a beast of a bot though

MSmits: yeah, I am really surprised at it. It's hard to find a set of params that works against everyone, i just picked darkhorse, re curse, robo and agade and hoped it would be representative of "good opponents"

MSmits: you cant pick too many or it will be too random

MSmits: it's really top 3... then a huge gap, then the rest

Astrobytes: Yes indeed. I struggle to fit adequate parameters to cover the whole spectrum of opponents

Astrobytes: Yup

Astrobytes: You get halfway through a submit and think "oh, nice" then you get spanked continuously by you 3 for the remaining duration!

MSmits: yeah, doesnt hurt your rating much, but wastes games

MSmits: community games dont have that many games on submit, so every game against a player out of range, is a lost game

MSmits: happens on langtons ant vs redstrike

MSmits: ridiculously strong

Astrobytes: Yep. Kinda forces you to try and improve though

Astrobytes: That ever get fixed?

MSmits: bug supposedly got fixed, but redstrike doesnt need it to clobber us

Astrobytes: lol

Astrobytes: What does he use?

MSmits: no idea

MSmits: I just use mcts with a lot of pruning

Astrobytes: Still haven't got around to fixing mine yet

MSmits: I'm in no mans land in langtons ant

MSmits: player far below and one far above

Astrobytes: Lonely island

MSmits: ye

Default avatar.png RAksix: help

Default avatar.png RAksix: Given two numbers a and b, you have to count how many numbers within a range belong to the following categories: http://chat.codingame.com/pastebin/e6516dde-f470-45f5-a0a6-5c37f0f8ccc6

MSmits: i ran out of ideas for it, but thats good, i dont like it as much as some others

Astrobytes: *cough* UTTT *cough* :P

MSmits: uttt has too many ideas, but i would like to let it go for a while, it's just hard :P

MSmits: RAksix use a % 3 == 0 to check if a number is divisible by 3

MSmits: or in your case Number % a == 0

Astrobytes: That's a solution not a hint! :D

MSmits: or in python: Number / a == Number // a though I am not sure that works

MSmits: oh I guess so :)

MSmits: might be the type changes on //

MSmits: or doesnt it?

Astrobytes: No idea tbh

MSmits: % a == 0 is the usual solution?

AbundantPuddle: Some of these clashes man... makes me wonder how they passed peer review.

Astrobytes: Yeah

MSmits: AbundantPuddle some are crap, but they can be edited i think

Astrobytes: And yeah, there were issues with approval previously

MSmits: by someone who has done enough clashes

Hjax: if we do more unofficial contests it would be nice to give the community multis some more love

Astrobytes: and/or use a new contribution

Astrobytes: thibpat's got loads of suggestions from everyone for future contests now :)

Hjax: if we use a new contribution cant we just get it made into an official contest?

Hjax: like how ocean of code was?

Astrobytes: Depends, it's a lot of work

Astrobytes: And depends on the type of game I guess

Astrobytes: If it's got fog of war and 65 million inputs it should be fine :P

Hjax: why do they like fog of war so much

Astrobytes: No clue man, no clue

jacek: smits dropped in ranking?

Astrobytes: Resubmit after failed tanh

jacek: failed? i thought he was winning muchly

Astrobytes: No, "apparently the experimental version + sigmoid was overtrained vs the leaderboard version in some inexplicable way the experimental version opened when i started VS so i thought that was my best"

MSmits: leaderboard version + sigmoid does much better than experimental + sigmoid

MSmits: not yet better than old version, but maybe if i find a good param value

jacek: is CG lagging today?

Astrobytes: Chat? Yeah. Sometimes leaderboards freezing too

YurkovAS: leaderboard not work

Astrobytes: Again.

jacek: nyoro~n

Astrobytes: I have no idea what that means

jacek: quite old and obscure meme

struct: ok I was mixing reversi and othello

jacek: theyre the same o.O

struct: no

jacek: w00t

struct: https://bonaludo.com/2016/02/18/reversi-and-othello-two-different-games-do-you-know-their-different-rules/

jacek: oh my

JKSTUDIOS2020: what a link :)

WINWINWIN: Submits stuck again?

Default avatar.png Tuo: I just got into Coders Strike Back legend league but my submit is not doing anything :( cant get a rank

Astrobytes: Yes, stuck again

WINWINWIN: :) will come up in some time

Astrobytes: minutes... hours... days...

struct: cg broke again?

MostComplicatedUsername: struct didn't you do yavalath?

struct: yes

MostComplicatedUsername: Says by undefined

struct: Yeah I deleted that account

MostComplicatedUsername: ohhh

MSmits: struct, if you're going to do othello, do the "real" othello that is used in international competitions

MSmits: seems more interesting

Hjax: the leaderboard really hates fun

WINWINWIN: Anyone know euler`s website to find codingame puzzles?

Hjax: yay the leaderboard updated

Hjax: and i gained a ton of ranks

Hjax: rank 47 in pr2 now

WINWINWIN: nice Hjax :)

Hjax: my bot is really weak on the super large maps

Hjax: but i think its quite good on the medium sized ones

eulerscheZahl: WINWINWIN I do

WINWINWIN: :D euler what is it?

struct: yeah msmits dont worry, my only question is grid size. but that wont matter for now I guess

WINWINWIN: I need to retrieve my CoC puzzle to make a few changes

eulerscheZahl: http://eulerschezahl.herokuapp.com/codingame/puzzles/

jacek: what is "real" othello?

WINWINWIN: thanks euler

jacek: i think i mostly fixed my mcts with ept. still weaker but not by much

jacek: i had int score in node and i was putting float there...

Astrobytes: lol, I made that exact mistake early on :D

eulerscheZahl: "real othello" = start in the center, always place next to other stone, isn't it?

Astrobytes: Yeah, otherwise it's optional. I think

jacek: so it differs mostly with starting?

Astrobytes: *optional in reversi

Astrobytes: yeah

Astrobytes: There's a special place in Hell for people who reduce my Oware score jacek :P

Default avatar.png noobyy: anyone know some fortran?

eulerscheZahl: C at the beginning of a line to start a comment. that's all my fortran knowledge

Default avatar.png noobyy: welp you can get paid a lot if you know it :^)

eulerscheZahl: but maintaining old code is no fun

jacek: yet fortran is not on the list in CG :<

struct: What is fortran used for?

Astrobytes: You can get paid a lot digging the roads at night, what's your point?

Default avatar.png noobyy: used in banks. It's very close to assembly in that it's very efficient if it's written well

eulerscheZahl: or for blackmailing

Default avatar.png noobyy: you can get paid more doing fortran - I guarantee it

eulerscheZahl: and obviously D is much faster

Astrobytes: :D

struct: Maybe I can write othello in fortran

jacek: write go in go

MostComplicatedUsername: My td code is an if jungle

Default avatar.png noobyy: write a program that prints itself without taking in any arguments

eulerscheZahl: that's called a "Quine"

MostComplicatedUsername: THat's impossible

eulerscheZahl: it is possible

Default avatar.png noobyy: Quine

MostComplicatedUsername: Cus if you have like a print (i)

MostComplicatedUsername: then you have to do

Default avatar.png noobyy: https://en.wikipedia.org/wiki/Quine_(computing)

MostComplicatedUsername: print ("print (i)")

Hjax: theres quines in lots of languages

MostComplicatedUsername: THen you have to do

Hjax: python has a pretty short one

eulerscheZahl: there are awesome variations: https://github.com/mame/radiation-hardened-quine

Default avatar.png noobyy: I'm glad I'm able to come to another random community to start shit posting :' )

MostComplicatedUsername: print ("print (print(i))")

eulerscheZahl: https://github.com/mame/quine-relay

Hjax: pythons quine: _='_=%r;print _%%_';print _%_


Default avatar.png noobyy: eulerschezahl is a real computer ethusist

MostComplicatedUsername: noobyy you can press tab to autofill names

Hjax: the secret is to do some string substitution

Hjax: its pretty neat, i wouldnt have ever come up with it

Default avatar.png noobyy: Or - can you write a program that will be able to determine whether or not another program will exit? Time and processing power is no concern.

eulerscheZahl: i only wrote 1 quine so far. in this language: http://www.hacker.org/hvm/

eulerscheZahl: trying to solve the halting problem, noobyy?

Default avatar.png noobyy: :)

eulerscheZahl: i paid attention in the lecture ;)

Default avatar.png noobyy: it's a no for anyone curious. Since you feed the program into itself and it blows up the computer.

Default avatar.png Joseph123456743: Is there a way to always get shortest mode clash of code on private

struct: no

Astrobytes: "Is there a way to always get shortest mode clash of code on private" sounds like an AutomatonNN statement

AutomatonNN: no i find it interesting

jacek: stop pushing me down :c

Astrobytes: Sorry, testing stuff

Astrobytes: Is derjack your smurf jacek?

jacek: smurf?

Astrobytes: alt

jacek: maybe

jacek: one is minimax, other is mcts

Astrobytes: Cool, I did the same when I switched, the cricri one is mine if you didn't notice

derjack: cricri?

Astrobytes: cricri10000, I disabled it, bottom of the leaderboard-ish

Astrobytes: derjack is the MCTS?

derjack: eeyup

derjack: in my local arena mcts is superior to minimax, yet hes behind jacek :s

Astrobytes: Yeah, it's always different in the leaderboard :P

Astrobytes: It's doing OK though

Astrobytes: Depth 0 or increased depth for now?

derjack: depth 0

Astrobytes: cool

derjack: and uses simpler eval than minimax

derjack: the comlex one is worse in mcts but better in minimax

Astrobytes: Yeah my minimax had more params too iirc

Astrobytes: fucking leaderboard frozen *again* ?

derjack: well if you spam submit, why not

Astrobytes: We're not spam submitting, waiting for first submit to finish

MostComplicatedUsername: Great, now im assigned to teaching a family friend python :rolling_eyes:

MostComplicatedUsername: Who created pr2?

AlanSpacer: And level 3 unlocks the chat

Xlos: Nice I got a contribution accepted 3/0 despite having more downvotes than upvotes

Xlos: Just gotta pander to the guys who have voting power

AlanSpacer: Thats the way to go

Default avatar.png FlatPanda_cf5: ?