Chat:World/2020-10-29
HistoriCode97: Bonjour les confinés
HistoriCode97: Meh
Trap-chan: good morning my little fellows :D
WidePlayer: sup
Aya_Noaman: hey
pascscha: Hello, BRETTRIVERBOAt
ToshiTuringMachine: hi
pascscha: Hello, ToshiTuringMachine
SydM107: Is there a quick way to somehow refresh the IDE? Sometimes it erroneously identifies errors and will no longer parse
ThunderbirdOne: submit the code and do a hard refresh on the page?
AttilaStein: Are there any ressources on this site wehre you can learn about the concepts and syntax needed for a puzzle in a specific programming language?
Light-Bringer: Hey I m new here I have a question, the input have to stay the same or can be change to manipulate them?
Light-Bringer: Ex the Bank Robber puzzle the input for the Vaults is a integer but would be easy to just make it an object
player_one: You can change the inputs. The system doesn't know anything about your code. Just what you output to stdout is important.
Light-Bringer: I m in C btw so no stdout for me
player_one: Umm... C has stdout. look into the args to sprintf()
player_one: printf() is to stdout by default.
Light-Bringer: it was light banter
player_one: sorry fprintf, not sprintf.
player_one: typo.
Light-Bringer: wait you dont even have to actually
Light-Bringer: can we talk about the puzzles in here or not?
Light-Bringer: not sure about the rules
ThunderbirdOne: you can, i think it's just mostly frowed upon to give straight solutions
pascscha: Are there solutions to old code-golf tasks? I cannot make my solution any shorter, yet there are some almost half its size in the same language on the leaderboard^^
pascscha: I mean code golf tasks from the past that don't accept any new submissions
jacek: clash?
jacek: ohai
player_one: What code golf tasks from the past don't accept solutions?
player_one: ^don't accept new submission?
Ola-Kitan: anyone on temperature?
pascscha: I don't know if there are any that don't accept solutions, but I get the option to "Publish On Codingame", so I figured the published solutions must be available somewhere
Astrobytes: They're not, that button does nothing. I think it's there for the Optimization games too.
pascscha: Oh ok, is there any more general way to learn about the tricks these top submissions might use?
Astrobytes: There are a few threads in the forum for language-specific golfing yeah. Also https://codegolf.stackexchange.com, and there's a few other sites around with tips
ThunderbirdOne: question, can i use the new ValueTuples in C# on CG ?
ThunderbirdOne: well... 'new'
Aya_Noaman: guys I want to make a contribution but I don't know how to draw a grid map that looks like this if printed:
Aya_Noaman: ####_####
Aya_Noaman: ####__###
Aya_Noaman: #####_##
Aya_Noaman: something like this
Aya_Noaman: help?
Astrobytes: ThunderbirdOne: I think so, C# 8.0 (.NET Core Runtime 3.1.3)
ThunderbirdOne: yeah, it works
ThunderbirdOne: for some reason i typed
ThunderbirdOne: function (int x, int y) Name() { ... }
ThunderbirdOne: no idea why i typed function and not just private static
Astrobytes: lol, switching langs a lot recently? :)
ThunderbirdOne: not at all actually :P
Astrobytes: lol, no excuses then
ThunderbirdOne: yeah, no clue
Aya_Noaman: hey anyone know an answer to my question up there
Astrobytes: Aya_Noaman: Have patience, was getting a link for you. https://www.codingame.com/playgrounds/40701/contribute---help/stub-generator-syntax
Aya_Noaman: ooh ok sorry
Aya_Noaman: and thanks alot
Drakkhonian: https://www.codingame.com/ide/puzzle/a-childs-play
Drakkhonian: test 3, wtf is that number of moves LOL
Drakkhonian: ok, time to think of a more efficient solution
wlesavo: hey, are validators for CodinGame Sponsored Contest is the same as test cases or there is a difference? and if they are different then are they known?
la_sacree_gueule: hello everyone
Yoann.V: Cette sacree gueule go le ban
la_sacree_gueule: who is this fcking french ?
Skynse: ngl captcha is annoying
RohanasaurusRex: yeah but at least sometimes you just have to check the box
RohanasaurusRex: but yea its still pretty annoying lol
Skynse: The ones without images are fine
ThunderbirdOne: have you tried asking nicely? using please etc?
pascscha: Sorry, I'm asking about coding golf again. How are these insanely low coding golf byte counts even possible? For temperatures, some people managed to do it in 54 chars with Python3, that's how much I need to read in the data. These scores seem impossibly low to me.
Uljahn: pascscha: what's your current score?
Uljahn: im at 68 in py3, no cheats just min() with lambda
wlesavo: also reading of inputs could probably be done better
pascscha: Ok, lambda shaved off my code to 78, thanks for that tip. The top of the leaderbaord still seems outlandishly short, I have a lot of respect for these guys.
pascscha: Usually if you can't make it any shorter, the problem is that you should approach the problem from a completely different way. It might be that a given way of solving it is optimized as good as possible, with no shorter way of doing it, but with another approach many more characters can be shaved off. Thank you for your help, I am new to this community and you seem quite welcoming.
Astrobytes: It's a great community :)
Monstrux: Hey guys, I have a question. Im struggling with the Puzzle of the week: A Game of Go. i dont have the technical knowledge in javascript to find an easy way to organise the data(matrix of dots, Ws and Bs)
Monstrux: any1 can give me some pointers? I tryed nested arrays, but I'm struggling to print them now
rohitkumar212: https://www.codingame.com/clashofcode/clash/1440412361eb5b5861067e61c21ac756d0e5b37
rohitkumar212: join
rohitkumar212: private
SPDene: use #clash for CoC invites please
Monstrux: gl in clash. Any1 can help me with my question guys? :D
Majeck: SURE
Majeck: *Sure
Majeck: What's the question?
Monstrux: Im struggling with the Puzzle of the week: A Game of Go. i dont have the technical knowledge in javascript to find an easy way to organise the data(matrix of dots, Ws and Bs) any1 can give me some pointers? I tryed nested arrays, but I'm struggling to print them now
Majeck: DM
AlPaca: is there a way to reset a language's default implementation that's presented when you open a problem for the first time?
AlPaca: for a puzzle/problem, that is
ZarthaxX: you have a reset button in the ide, top right
AlPaca: ahhhh! great, thanks
ZarthaxX: yw
Th1nhNg0: nom nom nom
eulerscheZahl: hey ZarthaxX lockdown is back in Germany too
kovi: so can we expect another unofficial challenge, before the fall-official one?
eulerscheZahl: the official one is in 2 weeks, so no
eulerscheZahl: i suggested a game in August. CG seemed interested but said they won't do it before the Fall challenge
eulerscheZahl: not decided if it will happen after or not at all
eulerscheZahl: the topcoder marathon looks interesting
IGNNE: Hey, I have a stupid question. I'm trying to go up the leagues in the "Coders Strike Back" game, but it somehow doesn't let me go to the next level
eulerscheZahl: are you above the boss on the leaderboard?
IGNNE: I'm beating the boss (and I'm #1 at the leaderboard ;) ) but there's no continue button
eulerscheZahl: did your submit reach 100% yet?
IGNNE: How do I check? I'm pretty new, sorry if this is a noob question
wlesavo: euler, do you know if maps for validators in CG sponsored contest are the same for validators?
eulerscheZahl: nope, you are at 66%
dbf: "not decided if it will happen after or not at all" - you have not decided, or CG team has not decided yet? :)
eulerscheZahl: wait until it finishes
ZarthaxX: eulerscheZahl yes i heard and thought of you when you told me it was coming :P
eulerscheZahl: no wlesavo, ask dbdr about it
ZarthaxX: im talking about lockdown
wlesavo: thx will do
eulerscheZahl: i'm waiting for the GO from CodinGame. they have to approve the game first
eulerscheZahl: no more soccer and working 100% remotely again - but i'm still better off than you i think
dbdr: wlesavo: honestly, I don't remember, I looked at Sponsored quite a long time ago
dbdr: maybe someone else remembers?
wlesavo: hm
eulerscheZahl: Neumann said it's the same iirc
eulerscheZahl: congrats IGNNE, you are promoted now
wlesavo: oh thx
dbdr: I think at the time it was possible to view validators replays, but now it's blocked?
eulerscheZahl: correct
eulerscheZahl: which sucks
dbdr: yeah, especially that it changes, creates unfairness
dbdr: I know I didn't completely reverse that optim, I guess a few people have?
dbdr: as in 100% deterministic prediction
eulerscheZahl: schubc invested some time
eulerscheZahl: generated 15k testcases and passes all of them
dbdr: wow
eulerscheZahl: as in: predicting the next step for AIs
wlesavo: nice
eulerscheZahl: not sure if he completely reversed the bonus score
dbdr: ah, 15k steps? I thought games
eulerscheZahl: turns
wlesavo: is there supper pellets?
eulerscheZahl: psst
dbdr: I *think* the score is not that hard, but I might misremember
dbdr: no spoilers
eulerscheZahl: i don't know what the game is about
dbdr: :D
wlesavo: anyway, thx, i ll try to recreate maps now =)
eulerscheZahl: you wanted to solve space maze before :P
dbdr: you might be able to extract info from tests, and use an assertion, to see if it crashes during validation
dbdr: to confirm
kovi: 3-5k is fair score
wlesavo: yeah, i gave it a couple of tries and failed misserably, so ill try again later :slight_smile:
wlesavo: dbdr yeah, thas my plan
eulerscheZahl: by bot has no prior knowledge about the maps
eulerscheZahl: and realizes too late that there is no 2nd exit
wlesavo: well yeah, it looks like 3-5k achievable without map knowledge, but i want a proper sim
dbdr: mine does, it detects the map from the first steps
dbdr: a few look almost the same
eulerscheZahl: i'm not surprised that you added that knowledge
dbdr: :angel:
eulerscheZahl: https://www.topcoder.com/challenges/71581555-b87a-4930-b37e-d78357073479?tab=details
eulerscheZahl: secret validators after the contest, no hardcoding
dbdr: payment of $0 :D
eulerscheZahl: https://www.topcoder.com/challenges/8440570d-d16b-43df-be4e-a720577626d5
eulerscheZahl: if you play for the money
wlesavo: 0 for first and second place, if you score lower you gonna owe money to the organisers :slight_smile:
dbdr: geolocation based on sound; interesting!
dbdr: wlesavo lol
eulerscheZahl: in Soviet Russia you pay for the contest
dbdr: I read a short story about a reverse lottery
dbdr: instead of everyone paying a little, and one person winning a lot
dbdr: everyone wins a little
dbdr: every one person who is killed / made a slave, something like that
dbdr: *except one person who is killed/enslaved
dbdr: I think it's by Jorge Luis Borges
ZarthaxX: eulerscheZahl yes you will always be beter than here :D
Majeck: I love Jorge Luis Borges' works
Majeck: He's a really good writer
ZarthaxX: we have people here occupying other's terrain
eulerscheZahl: so someone stole your pool?
ZarthaxX: and now the government as an incentive to stop them from doing this they will pay them
dbdr: https://en.wikipedia.org/wiki/The_Lottery_in_Babylon
ZarthaxX: approximately 1600 dollars in 6 months
ZarthaxX: so the deal right now is go occupy a terrain and you will get paid for that :D
eulerscheZahl: Argentinian dollars?
ZarthaxX: better salary than an actual job
ZarthaxX: i converted it to dollars
eulerscheZahl: ah
ZarthaxX: its 300k pesos here
ZarthaxX: we almost got a 200:1 conversion :D
ZarthaxX: and no, no one stole our pool lol
ZarthaxX: im talking about terrains like in the farm
ZarthaxX: for now at least
Majeck: Why does "if None:" in python register as True?
Majeck: Not complaining, it allowed me to remove 2 chars in a code golf
Majeck: Just want to know why
dbdr: wow, c4l is seriously random
dbdr: finished #45 in gold, resub and I'm #1, close to bose at 40% :)
dbdr: *boss
Illedan: That game is the definition of A > B > C > A :P
dbdr: yes? in what way?
dbdr: I thought it was the randomness of samples
dbdr: Illedan in gold, tsss ;)
Illedan: One of my early contests
Illedan: I was young
Illedan: And so very n00b
dbdr: multi is still open
Illedan: Not going to open that vound
dbdr: :D
Illedan: Working on my BS
dbdr: you're #23, with some spam submit yo might promote
dbdr: or maybe my bot is especially sensitive to samples
Illedan: Let me check
dbdr: I'm doing the "return undiagnosed samples" strat :)
Illedan: Efficient one
Illedan: Mine is just bruteforce, and a stupid one
Illedan: HOLY, the code I wrote :scream:
Illedan: Abandon shit
dbdr: bruteforce N next turns?
Illedan: There is no such thing as "next" in my bot
Illedan: Current.
dbdr: bruteforce depth 1?
Illedan: Yup
dbdr: how can that work??
dbdr: you need sime planning...
dbdr: *some
Illedan: I have 106 IFs in this code -.- Who writes this shit
Illedan: (╯°□°)╯︵ ┻━┻
Gabbek: Hello everyone :)
dbdr: :D MeanMax was my first contest, and only heuristic one
dbdr: but now I will be legend with heuristics in c4l
Gabbek: dbdr I'm really curious what's your approach in https://www.codingame.com/multiplayer/bot-programming/coders-of-the-realm . Would you like to talk about it a bit?
dbdr: 50 ifs Illedan
Illedan: Rookie numbers
dbdr: it's the quality, not the quantity ;)
dbdr: Gabbek, it's been a while, let me check...
dbdr: what do you do?
Zenoscave: dbdr what's the return undiagnosed strat
dbdr: e.g. if you have 4 xp + mols in each category, you can return a sample without knowing what it needs
dbdr: since 4 is the max for rank 1
Zenoscave: huh
Zenoscave: good to know
dbdr: Marchete mentions it in the forum
Gabbek: dbdr I only use the next two tiles and some heurestics when it comes to placing my next tiles; couldn't manage more with python ( talking about coders of the realm here, to avoid confusion :D )
ChastyXD: how to create private coding games??
Illedan: Clash ChastyXD?
Illedan: Bottom of the clash page
ChastyXD: Great I have seen that
Illedan: https://www.codingame.com/multiplayer/clashofcode => Start a private clash
dbdr: Gabbek, it looks like I'm doing a brute force search
jacek: looks like? so you dont know?
Gabbek: dbdr for how many tiles? 2/3/4 or more? :)
dbdr: quick read of the code I wrote 6 months ago, or more?
dbdr: eprintln!("My next tiles: {:?}", tiles);
tiles.truncate(2);
dbdr: so it looks like I only consider the next 2 tiles
dbdr: I don't remember the exact rules, we know more than that in advance?
Gabbek: only in case of 1vs1 variant
Gabbek: so about 30% of all matches
dbdr: isn't that a separate game?
eulerscheZahl: 125 if's in my code4life (including comments). did I win?
dbdr: but that's a MC, no?
eulerscheZahl: and why can't I use CTRL+Enter anymore to add a newline?
Gabbek: nope, the separate game is 1vs1 with all tiles available, while here it's just half of the tiles in case of 1vs1
eulerscheZahl: yes, monte carlo
eulerscheZahl: i mean SHIFT+ENTER
eulerscheZahl: where is my newline? :(
Gabbek: hello testing SHIFT+ENTER it works nice
Gabbek: seems to work eulerscheZahl
dbdr:
eulerscheZahl: used to work for me but broke recently
dbdr: in chat?
eulerscheZahl: yes
eulerscheZahl: let me try in private mode
dbdr: 1 2 3
eulerscheZahl: 1 2 3
eulerscheZahl: hm, some addon screwing up
dbdr: polizei
jacek: AutomatonNN will he succeed?
AutomatonNN: why did you say that ?
dbdr: arf you messed up my joke :D
Illedan: Give me that rank dbdr :D
Illedan: Still a little too far
Illedan: I see a bug
Scarfield: grenadier
Illedan: time to find the wrong IF
dbdr: Illedan: I succesfully tricked you to lose your rank?
eulerscheZahl: 1 2
eulerscheZahl: ah, vimium it is
Illedan: Lose?
Illedan: I was 24th? :P
Illedan: Now I ended 5th
dbdr: oh
dbdr: well
eulerscheZahl: i miss vimperator :cry:
Illedan: I changed a magic number though
Gabbek: :D
eulerscheZahl: the winrate number
dbdr: vimium here
dbdr: so you use vim mode in browser but not vim? original :D
eulerscheZahl: :D
eulerscheZahl: i'm too lazy to click links with my mouse
eulerscheZahl: but CG has poor support
dbdr: what happens if someone submits after being marked for promotion?
eulerscheZahl: the old chat allowed to switch chatrooms with vimium. the new doesn't
dbdr: the submit ends suddently?
Illedan: yes
dbdr: Illedan, is your rank stable? you need to resub
dbdr: you might promote
dbdr: ah, will mise the train in 2 minutes
dbdr: *miss
Illedan: :scream:
Illedan: Hold it!
dbdr: Hoddor!
dbdr: bing! :blush:
dbdr: Gabbek, are you limited by speed?
Gabbek: whoa, so many messages at once! Let me read quickly
Gabbek: dbdr mostly in 1vs1 variant - but in the 4 players one I could use a bit of extra time for sure; that's why I've started learning some C#
Gabbek: I was planning to incorporate remaining tiles pool somehow into my bot my after some quick tries I've gave up on it for now :D it sounds quite useful for sure
Gabbek: but after*
dbdr: Illedan, +0.05 :D
eulerscheZahl: going for #1 globally, dbdr?
dbdr: you won't let me ;)
Gabbek: hehe, the best kind of fight :P
dbdr: just taking my time, doing one multis when I feel like it, no rush
dbdr: *old multis
Gabbek: there's so many! I'm having issues on deciding which one I wanna start, hmm
eulerscheZahl: it's up to royale if you are allowed to take a lead
eulerscheZahl: he's playing onitama for several days already
dbdr: you guys stay close for a long time now, no?
eulerscheZahl: yes
dbdr: Royale has weird swings, +1000 on some days
eulerscheZahl: he's 1st when he doesn't resubmit at 4am
dbdr: but I guess he has big minuses too, since he's not running away
dbdr: lol
dbdr: 4am is a weird time to submit in europe
eulerscheZahl: i know
eulerscheZahl: maybe it's 3am... you got the idea
dbdr: Royale is a bot?
Gabbek: eulerscheZahl I see yours https://www.codingame.com/training/expert/space-maze ; scary success rate, how much harder it is compared to sokoban? :) Even the slightest comparison will be helpful, trying to plan out my weekend a bit :D
jrke: hope royale plays dots and boxes as fast as possible ;)
eulerscheZahl: sokoban is a child's birthday
dbdr: why jrke?
Illedan: (╯°□°)╯︵ ┻━┻
Illedan: Not enough :(
Illedan: 0.16
Illedan: -.-
Gabbek: eulerscheZahl so that means it will take me a year or so, noted :D
eulerscheZahl: sokoban can be done with BFS alone. space maze needs some heuristic on top, what to expand next
jrke: cause my latest code in D&B is pending i will finish it only when someone makes me push down in that
dbdr: oh, you were at 90% when you were above
dbdr: are you #1 jrke?
dbdr: last time I checked it was crazyremi
Gabbek: you guys are working hard :)
jrke: nope i am not 1st i am 2nd
dbdr: ah, above me then? :)
eulerscheZahl: by 0.01
dbdr: lol
Illedan: ezpz? :P
jrke: till dbdr resubmit i am 2nd
jrke: else 3rd cause someone pushed me above him by.01
dbdr: done
eulerscheZahl: :D
jrke: btw where is Ms research bot on D&B? still painting rooms?
jrke: lol i will be 3rd soon
dbdr: lost to jacek :/
Gabbek: off to read some stuff, have fun!
dbdr: so many more people on d&b
jrke: i think i will go back to D&B after fall challenge
dbdr: we need a new league
jrke: dbdr cause it became PotW once
dbdr: that's good
jrke: yup many CPs to me
dbdr: Illedan beating me too
dbdr: yay, 100% against CR :D
dbdr: 50%
jrke: 1/1 is not 100% you say dbdr ;)
dbdr: 1/1 is 100%, 1/2 is 50%
jrke: in that case i am having 100% winrate against you 1/1 ;)
jrke: sorry 3/3
dbdr: yep
dbdr: still 100% ;)
jrke: oke hoping good luck for furher resubmits (if you are going to do)i am off gn
dbdr: gn
starmaq: what challenge are you guys doing
dbdr: https://www.codingame.com/multiplayer/bot-programming/dots-and-boxes
starmaq: thanks
starmaq: wow
starmaq: the amount of pressure that this generates
Scarfield: Need some coffee? :p
starmaq: sure hehe
Scarfield: "Édouard Lucas, who called it "la pipopipette" which you will agree is a very pleasant name" Just started reading the statement xD
dbdr: pressure?
eulerscheZahl: peer pressure and next week he'll start smoking
dbdr: - what are you doing? - A - hey, you're pressuring me into doing A!
jacek: and im 6th just by being pushed by other people
MSmits: hey guys
Scarfield: oi
MSmits: working on a traveling salesman thingy: https://trinket.io/python/8a3db0ac8f?toggleCode=true
MSmits: t to make a new problem, r for random, i for best of 1000, h for hillclimbing
jacek: thats just some dots
MSmits: gonna add SA next and some code rewriting
MSmits: yes, travelling salesman :)
MSmits: try clicking r, i or h
MSmits: pressing
jacek: noice
MSmits: hillclimbing is super effective for this problem apparently
MSmits: except when N > 30
MSmits: I guess that's when you need SA or something
Scarfield: cool :)
jacek: each dots visited only once?
MSmits: thats the rule
MSmits: and it's cyclic
eulerscheZahl: there's a pretty good heuristic also: Christofides
MSmits: so you end where you start
jacek: lines overlapping looks like some dots are visited more than once
MSmits: yeah i know jacek, it's not easy to eliminate that
eulerscheZahl: proven do be at most 1.5x the best
MSmits: thanks euler, will check out that heuristic
MSmits: i did have a greedy heuristic
MSmits: but it's bad
jacek: Up until 2020, this was the best approximation ratio that has been proven for the traveling salesman problem
jacek: maybe 2020 isnt bad after all
MSmits: what happened in 2020
eulerscheZahl: i need a full southpark season to understand that
jacek: https://arxiv.org/abs/2007.01409
MSmits: ok that heuristic looks wayyyy too complicated for my class :P
MSmits: but good to know it exists
eulerscheZahl: no matching? :D
MSmits: matching?
eulerscheZahl: minimum-weight perfect matching
MSmits: no idea what that is
ZarthaxX: you are killing him :P
MSmits: btw, my definition of a neigbor solution in this problem is just swapping the order of two dots
MSmits: works wonders with hillclimbing
MSmits: i can do this with SA too, but I have no idea how to GA this. How do you combine two solutions?
eulerscheZahl: you can also take a chain of consecutive points and reverse the order
MSmits: to make babies and stuf
MSmits: eulerscheZahl ahh yeah
eulerscheZahl: the removes some crossing lines
MSmits: makes sense
MSmits: difficult to do this over the end of the array
MSmits: messy i mean, to code
ZarthaxX: combining solutions sounds weird
MSmits: i mean
eulerscheZahl: you mean lopping around while reversing?
MSmits: procreation
ZarthaxX: incompatbility in the paths
ZarthaxX: you mean crossover
MSmits: yes eulerscheZahl
Scarfield: for GA
eulerscheZahl: you don't have to. it's the same as reverting all other nodes
MSmits: yeha crossover sorry
ZarthaxX: im talking about ga yes
MSmits: never done a GA
ZarthaxX: it seems weird for this
MSmits: eulerscheZahl ok, i should think about it more then
Scarfield: yea
MSmits: i will probably not do a GA then
eulerscheZahl: you can split the path in 2 sections. it doesn't matter if you revert one or the other half
ZarthaxX: SA is good too anyway
MSmits: random / greedy / hillclimbing/ SA seems more than enough for a class
ZarthaxX: yes
MSmits: eulerscheZahl makes sense
ZarthaxX: there is smth called tabu search too
MSmits: is it called tabu because people feel uncomfortable discussing it?
ZarthaxX: it's interesting, but maybe too complex to implement, idk
ZarthaxX: no lmao
ZarthaxX: it tries to rmemeber explored solutions
ZarthaxX: and other things
ZarthaxX: to avoid looping over the same place agan
ZarthaxX: again*
MSmits: ahh i see
ZarthaxX: but again, maybe out of scope
Scarfield: i guess you could GA it. Have horisontal right as zero, then rotate around, sort the other unvisited points in order of apperance, have each gene represent the number chosen in the order, and so on
MSmits: seems more something i should look at for an optim arena, not for a class
MSmits: i know a lot more about adversarial algorithms used in arenas than that optim stuff
ZarthaxX: yes, it may escape your class objectives
MSmits: Scarfield the problem is mainly, crossover
MSmits: if a gene is an order of points
MSmits: how do you crossover?
MSmits: mutating is easy
jacek: ask your mum
jacek: :v
MSmits: lol
Scarfield: wouldnt be a problem here: AAAA X BBBB -> AABB and BBAA AAAA X BBBB -> ABBB and BAAA
rwilson: MSmits - did you consider varying the line colors to avoid the issue jacek mentioned? .. like this maybe: https://trinket.io/python/ce6626ae6c
MSmits: damn
MSmits: that's fast rwilson
MSmits: you just looked at my code and changed that
jacek: is he your student?
MSmits: well done
MSmits: i wish
rwilson: ;) good starting material = easy work
MSmits: thanks :)
MSmits: something like this is better yeah
MSmits: maybe I would change the colors less dramatically
MSmits: but the idea is sound
Scarfield: since the genes will just represent which unvisted point to go to from the current point
rwilson: but.. it's a quick hack ;) even if it's only a couple lines, i'd do it a little nicer normally.
MSmits: yeah
MSmits: i know
MSmits: Scarfield let me think about it a sec
Scarfield: NO! :p
MSmits: what is the A and the B?
MSmits: my orders are more like this 0234576198
MSmits: all unique indices
MSmits: i swap them to generate neighbors
IsThisWhereMyNameGoes: Aaaaave maríiiiiiia
Scarfield: better example: 1st 5th 2nd 1st X 3rd 4th 8th 2nd 1st 5th 8th 2nd & 3rd 4th 2nd 1st
Scarfield: 1st will be the first dot from the current counter clockwise from vertical right for example
Scarfield: and so on
MSmits: hmm
jacek: or be more modern and for traveling salesman problem use neural network, big data and blockchain
MSmits: i'm getting to NNs as well jacek, just not that advanced :P
MSmits: I want to maybe Q-learn hexapawn
jacek: hexapawn is like mini breakthrough
MSmits: yeah
MSmits: Scarfield what if the solution is all over the place?
MSmits: you can't have a clockwise or anti-clockwise situation then
MSmits: i am a bit confused about what they do with the matchboxes in hexapawn. It's unplugged q-learning i think, but I don't understand how they choose between moving forward or taking an opponent piece
MSmits: they use beads to select a pawn
jacek: matchboxes? beads? w00t
MSmits: yeah basically they use a bunch of matchboxes to create an ai
Scarfield: how do you mean all over the place?
MSmits: you play games repeatedly and every time it loses you remove the bead that made the losing move
MSmits: so it learns
MSmits: Scarfield lots of crossing lines
MSmits: jacek: https://www.youtube.com/watch?v=sw7UAZNgGg8&vl=no
MSmits: start at 2 min or so
MSmits: skip introductory and advertising crap
jacek: wtf
Scarfield: i dont think that would be a problem, as you go through the genome, you check which dots havent been visited yet, and go to the specified number of the current gene. Im not saying it would work well, just thinking aloud that it can be done :)
MSmits: ahh ok Scarfield, i might look at that as an extra assignment
Scarfield: the issue is to represent the genes in a way that can be crossed over, ie not a series of unique indentifiers
Scarfield: you would have to choose a starting point though for this approach to make any sense
jacek: MSmits does he mean this? https://en.wikipedia.org/wiki/Matchbox_Educable_Noughts_and_Crosses_Engine
MSmits: oh yeah
MSmits: but thats not what the video is about
MSmits: thats just background info
MSmits: Scarfield yeah thats what i thought, that my way of representing a gene is not helpful
MSmits: ohh i understand it now, the colors are actions, not pawns
MSmits: so if a pawn can do 2 actions, both will have a bead
MSmits: different colored bead I mean
jacek: so this matchbox thing is q-table?
ZarthaxX: that game concept is amazing
Scarfield: https://prnt.sc/v9mr7z MSmits a super pretty illustration of what i mean
jacek: its ugly
Scarfield: :'(
Scarfield: should have added a pony :thinking:
ZarthaxX: yes you should
MSmits: jacek yeah Q table mostly
MSmits: seems to me
MSmits: the guy speaks of genetic algorithm but thats misleading, it's definitely not GA
MSmits: got it Scarfield :)
MSmits: no ponies required
Scarfield: https://prnt.sc/v9n0ds
Scarfield: fixed
Scarfield: yea it has nothing to do with GA but the idea is cool!
MSmits: i try to have at least 1 unplugged activity for every subject
MSmits: TSP is easy, just let them find a good solution on paper and make a contest of it
jacek: Scarfield twilight OP, nerf plx
MSmits: oh right, i forgot jacek is a mlp expert
MSmits: he made his own pony and called it derjak
MSmits: dejack
MSmits: cross those over for me Scarfield gettin tired of my typos
Scarfield: glad you liked it :p wasnt there something about fans getting ponies added to the show?
MSmits: ohh, jacek please tell me you participated
ZarthaxX: TSP easy in what sense?
jacek: i believe some bronies were included as backgroung ponies in the show
MSmits: easy to make material
MSmits: just print it and let them puzzle it out
MSmits: the matchbox thing seems really complicated
ZarthaxX: ah
ZarthaxX: true
MSmits: bronies?
MSmits: bronies before honies?
Scarfield: xD
jacek: gah how uneducated you are
MSmits: I admit it
jacek: and i see youre back to mcts in oware
MSmits: ah yeah, i gave your algo a good try, it could get almost the same rank, but was still significantly weaker. I think I would need a stronger eval to make it work
MSmits: the problem is that you could go 10 deep, with an inaccurate eval and completely overwrite the whole branch
MSmits: wheras ept is fuzzy and one bad eval wont mess things up too badly
MSmits: what game?
jacek: :o
MSmits: yeah but it's not one game
MSmits: it's many games
simon-tiger: Hi there! :wave:
KamalCh: well it is sort of a big huge game with multiple little puzzles!
MSmits: puzzles is one thing
KamalCh: I'm actually stuck on the first one with the horses
MSmits: is that the first one?
jacek: dual races?
jacek: horses? ponies?
MSmits: this one is yours jacek, obviously
KamalCh: did a list with all horses
KamalCh: but still something seems wrong in my code OO
KamalCh: even tho I'm just comparing abs values in the lis
simon-tiger: What is the puzzle you're doing?
simon-tiger: Can you post a link?
MSmits: which language KamalCh?
MSmits: i mean link the puzzle
KamalCh: https://www.codingame.com/ide/puzzle/horse-racing-duals
MSmits: oh lol mine is in bash
KamalCh: is alright I mean I did a list then "sort" the list
MSmits: me too, thats what you're supposed to do
KamalCh: then compared i and i+1 in the list
KamalCh: and took the smalled value
MSmits: thats exactly right
KamalCh: but I believe something is wrong with my code -.-
simon-tiger: Huh?
jacek: you have error?
MSmits: it's ok to link your code here, for an introductory puzzle like that
MSmits: not ingeneral, but for this puzzle its ok
MSmits: if you link it, we can fix it
KamalCh: http://chat.codingame.com/pastebin/c9aec37d-b6a8-4512-82cc-8ee1c3bbe07b
simon-tiger: It's ok to share code unless it's an ongoing contest
simon-tiger: Ah...You forgot the i += 1
MSmits: yeah
MSmits: i never changes :)
simon-tiger: Also, I would use a for loop, not a while loop
MSmits: i would use for i in range(n-1)
simon-tiger: yeah, that's what I would use
MSmits: thats safer because the i++ happens automatically then
jacek: and the elements of harmony
MSmits: is this a ponly thing jacek?
MSmits: pony
jacek: maybe
MSmits: thought so
MSmits: is this you jacek:
KamalCh: that's me learning trying a while
MSmits: whiles are useful
MSmits: just not here
jacek: G3 ponies? eww
MSmits: is this a class of pony?
MSmits: G3 is a stellar classification
KamalCh: Lol wait what was that
MSmits: KamalCh jacek is obsessed about ponies
MSmits: so we need to discuss it every once in a while
MSmits: otherwise he's gonna go to a CG alternative and we lose him
KamalCh: that'S alright a pony a day keeps teh doctor away
MSmits: you got it
Scarfield: jacek is our little troll
jacek: and new game has been approved
Scarfield: there are MLP games?
MSmits: blocking?
MSmits: by tric trac?
Scarfield: oh lol
jacek: so far im 1st
Scarfield: and last
Scarfield: mwuhaha
jacek: usual guys of board games
Scarfield: seems like a cool game though, didnt see it in contributions
MSmits: i'm not sure if i'll like this one
MSmits: will be a long time before I try it I think
MSmits: do you think it should go below paper soccer on my list jacek?
MSmits: paper soccer is like 4th
jacek: everything should be below paper soccer
MSmits: problem is, games i am addicted to keep drawing me back in
MSmits: like uttt
darkhorse64: In this game ("penguins" is another example), you need to implement a n player algorithm. It cannot be a rehash of the usual MCTS code
MSmits: smitsimax is my go to algorithm for > 2p, but it doesnt work for sequential games. Only for simultaneous =/
MSmits: well i suppose it would work for sequential somewhat, but the error it makes will be greater
trictrac: you can use paranoid or max^n
MSmits: biggest problem is that the allowed moves depend on opponent actions
MSmits: yeah thats what i read before, like with tron, trictrac
MSmits: congrats on acceptance of multi btw
darkhorse64: There's also an mcts for n players
MSmits: yeah, but is it any good?
trictrac: thank you this is my first contribution, i will work now on hanabi
MSmits: hanabi is a board game?
trictrac: no cooperative game (like code a la mode)
MSmits: interesting
darkhorse64: You need to implement this flavor to know
trictrac: 4 players --> so 4 games of 3 players and sum for the score
trictrac: the best you cooperate with others the best score you have
MSmits: I see
darkhorse64: trictrac: you failed to beat your boss
MSmits: lol
trictrac: the boss is a kind of max^n depth 2, I have a version time limited but I wait someone in bois 1 before launching this bot
trictrac: for the moment, my bot is an oriented random
darkhorse64: Hmm, I may need to add a few more ifs to my code then
jacek: ifmax
jacek: and im above worstfield
dapOnEmHaters: i just did Thor
Skynse: How do you guys use the chat system lol?
Skynse: as in
Skynse: the chat rooms get jumbled up in clashofcode
Skynse: so Im not sure which one is the current coc chat
GeorgeKrastev: Can I play 1v1 with a friend in a game other clash of code ?
SPDene: yes: scroll down to the bottom of https://www.codingame.com/multiplayer/clashofcode and "start a private clash", then give the link to your friend
GeorgeKrastev: I meant other than clash of code
ChastyXD: Hi
ChastyXD: I want to test the live coding stream
ChastyXD: I have a channel
ChastyXD: I put there Codingame
ChastyXD: but I'm not appearing at the streams page on Codingame
ChastyXD: Do I need to do sth else
Allis: Might be case-sensitive, such that it needs to be "CodinGame" exactly?
Allis: If not that, then it might be the case that you need to be added to the list of streamers before showing up on the Live page.
Allis: @ChastyXD In which case it might be best to ask for support in the Discord.
ChastyXD: Hello Allis
ChastyXD: I have tried that many times
ChastyXD: but I dont appear there
Allis: ¯\_(ツ)_/¯
kstoqnov: https://www.codingame.com/blog/teach-others-code-streaming/
Jdools05: Hey world chat, I'm relatively new to this site. Is it ok to ask for advice to go about a task? I'm stumped right now but am unsure on how the community views asking for help. Thanks!
ChastyXD: https://www.twitch.tv/chastywi
Tsepeto: guyss
Tsepeto: whaat is entityId in ghost in the cell?
Tsepeto: i don't think it is explained
rwilson: Tsepeto - it's just a unique # that refers to that 'THING' . so on each turn you can know which thing WAS which thing last time, if you happen to care about that.. think of it kinda like a 'name' .. Joe is at x,y this turn ... next time Joe is at A,B
Tsepeto: we talking about factories
Tsepeto: so if a factory has id 1 in one round
Tsepeto: and another one has the same id the next one
Tsepeto: we are talking about the same factory?
CodingCarter: yep, because ids must be unique
Tsepeto: i see, thanks boys
Tsepeto: another question, for ghost in the cell which is not clear
Tsepeto: so in the initialization input you get factory1, factory2 and the distance between them
Tsepeto: factory# is the same as its id?
Tsepeto: it's not explained in the problem
DaNinja: thats right
Jdools05: Hey, does anyone know any techniques to optimize looping through large datasets to speed up the code?
rwilson: Jdools05 - for stuff on CodingGame? if so, that's mostly going to be by analysis, not so much about a particular techique per se. ( example: looking for prime number from 1 to 10 billion -> instead of testing each number for primeness, you KNOW you can skip at least every other, since they'd be divisible by 2 )
Jdools05: yeah i should have made myself more clear, sorry. I'm stuck on Stock Exchange Losses. I have working code that I feel is short, but on the Large Dataset test case, it times out... I don't know what specifically to look for without looping through all the points
rwilson: Jdools05 - for that one it's almost certainly how you're checking for the loss. my tip for timeouts is to pick some key areas in your code to do some debug output of timing information - i.e. poor man's performance profiler.. just to make sure you're targeting the slow parts if you can't spot it clearly from the code alone.
LastRick: jdools05: my guess is you are comparing each value in the index versus any other value in the index that comes later? That's so many checks for the large dataset. There is a smarter way to do it without have to check N vs N-1 so to speak
Jdools05: Thanks, LastRick, I got it all figured out now! :thumbsup:
Mr.Snow: hey guys im new and stuck with temperatures puzzle, i know what to do but im not getting any progress on how to do, my code is returning wrong numbers, i hope you could show me a soloution so i can understand where im wrong.
Mr.Snow: here's my code https://pasteboard.co/JxYrAhs.png
DaNinja: where are you checking which one is closer to zero?