The Nineteenth Byte

vzn

00:32

hello doorknob/codegolfers...

we meet again

TheDoctor

Hi

Doorknob

Greetings

vzn

saw your tcs.se community ads. nice

Doorknob

9

Q: Let's make an ad for our site! (And now vote for our newly posted ads!)

UPDATE: Ads have been posted! Vote for us! There was a tie, with two answers at 10 votes, so I just decided to post them both. Ads have been posted on Programmers (1, 2), Mathematica (1, 2), Game Development (1, 2), and Theoretical Computer Science (1, 2)! Vote for ...

vzn

have you passed ads on sites other than codegolf?

Doorknob

00:34

No.

vzn

ok.

+1'd your ad.

over on Computer Science & Theoretical Computer Science dont recall any community ads have been passed, ever... it is possibly the same story on some other se sites...

Doorknob

Thanks :D We have an ad displaying on Programmers.SE, but that's the only one so far

vzn

so you have +6 on the codegolf ad on programmers.se? for 2014?

Doorknob

Erm, on most sites there's at least 7 or 8 ads in rotation....

Yes

vzn

many sites seem not to vote any community ads...

anyway am gonna inline those tcs.se ads here & lets see if anyone here votes on em eh? afaik everyone here can just sign up to Theoretical Computer Science & get +100 & then immediately vote on meta right?

sound good to you?

Doorknob

00:38

If people wanted to vote on them, they would have already.

vzn

tcs.se/cs.se metas are generally comatose

so, whaddya say? seems like your motley crew around here could fix that a bit at least in this case?

Doorknob

Erm. ?

vzn

easy right? a lot easier than a lot of the coding challenges on this site right? :p

3

A: Community Promotion Ads - 2014

2

A: Community Promotion Ads - 2014

Doorknob

Ahem, I think the term for that is called "spam"

vzn

you guys seem to be more about teamwork around here...?

@#%& dude, wtf

its called an idea

Doorknob

00:42

I don't particularly like having half my screen covered with oneboxes.

vzn

?!? you make no sense. whatever

Doorknob

You could just post a link to the community ad post. People will click on it and vote if they want to.

vzn

dont think you understand what the idea here is. nevermind! let me know if you wanna talk about it further sometime

← prob just wasted 10m... whatever

Doorknob

The idea is that you want people to vote on the TCS community ads?

vzn

kinda dude... the idea is that you want people to vote on those ads....

Doorknob

00:45

Okay, but there's no need to cover up our screens with those oneboxes in order to do that.

Just link to the community ad post and people will follow the link if they so desire.

vzn

dude, you're awfully hyper about what shows up in your chat room...

Doorknob

(And I've already linked them in the meta post)

vzn

the ads have been around awhile already. nobody is voting on them.

Doorknob

Actually, at least 3 people have voted on them in less than 24 hours....

vzn

so, chat is a place to motivate the motley crew...

you dont know how many of those votes are from codegolfers. some may be from tcs.se users

Doorknob

00:47

Sigh Alright, I'm going to go eat dinner now.

vzn

sure. anyway, dont know how this will go. just noticed you guys seem to work almost as a team sometime in chat.

which is ... worthwhile at times.

ps does Boy Scouts have a programming badge?

heard awhile back they introduced a brand new robotics badge.

anyway re ads, if you look at old cs.se/tcs.se ads last few yrs, they get a dribble of votes & then never hit the "easy" 6v threshhold.

so, seems like a nice opportunity waiting to be tapped.

ah here it is

boy scout programming badge

what if Boy Scouts showed up on codegolf & cs.se with their programming challenges/questions? that could be neat

(another idea that occurred to me)

lots of good links on that badge, maybe will blog on this sometime.

ah here it is

boy scout robotics badge workbook

good stuff!

anyone here thinks any of this is a good idea, star it!

←worked in lego mindstorms. built/coded 2 biped walkers. =D

do you guys have any code challenges in robotics? wouldnt that be awesome

saw that meta post asking for new tags.... there ya go!

robotics

Chris Jester-Young

01:10

I am going to nuke that spam barrage, just for the record. If you want to read it, save it up now. ;-)

vzn

@#$&

ProgrammerDan

That would be a difficult challenge to objectively score, without a significant physical investment

I <3 robots, but this is definitely not the forum for it :)

vzn

there are many programming challenges in robotics....

ProgrammerDan

@ChrisJester-Young :) sounds good

vzn

its a natural crossfit with the site.

a significant number of users probably have robotics bkg

Chris Jester-Young

01:11

@ProgrammerDan Actually, it sounded like you actually read those messages, so it probably had value to at least one person. I'll leave it around.

ProgrammerDan

@ChrisJester-Young Only the last 5-7 messages...

vzn

oh yeah let me go scroll thru all your old chat room msgs looking for value

am sure its quite ubiquitous

ProgrammerDan

@vzn I could see doing a challenge around a particularly interesting subproblem in robotics. Perhaps a navigation problem, or a kinematics problem

DoorknobChatbot

New post! Sudoku to the power of 21

ProgrammerDan

In fact, challenges of that type I've seen here ... they often come up as king-of-the-hill

vzn

01:13

hi programmerdan thx for savin me. escaped the guillotine for 3sec

at least

is it just me or is it lord of the flies around here? =(

Chris Jester-Young

@vzn Nobody is calling you Piggy.

vzn

yeah see how you treat your guests in this particular room

but, not much worse than elsewhere on se

ProgrammerDan

@vzn I've only been here a few days, generally the only time there's an issue is if someone monologues :P

vzn

the @#%& chat room was idle/empty until wrote something

is it like, a restriction to speak at 5th grade level in here? sorry

ProgrammerDan

I actually should pull out my mindstorms at some point and build something. It's been quite a while, though.

vzn

01:17

PD did you hear about cubestormer? lego rubik cube solver

ProgrammerDan

Yeah, watched the vids with interest. Quite a feat, it's very impressive

vzn

did you write any code for it?

=D

TheDoctor

funny how active this chat becomes when people ac**tually** write st*uff*

ProgrammerDan

@TheDoctor typos intentional? :P

@vzn Yeah, but not in years now. Last program I wrote for mindstorms was probably in 2006 or 2007.

TheDoctor

yes

vzn

01:18

lego has been improving the coding system... its visual programming as you know...

TheDoctor

For some reason i can't see the text while i'm writing it

vzn

there also was a C/java compiler years ago, figure its better now

ProgrammerDan

Oh, I've got mindstorms v1/1.5 -- can't afford the new one. The visual programming aspect was the worst part of it for me.

@TheDoctor That's truly unusual!

vzn

ah yeah it might not be desirable for old guys like us who learned with text coding

yeah the price has gone up quite a bit.

it started out $199 in ~1999. think its now around $350...

but, also major improvements in sensors, cpu, etc

ProgrammerDan

Yeah, it looks great, but frankly -- I've got bills and a family to support.

vzn

01:21

right. you can support your kids by teaching em mindstorms wink

Doorknob

I return

Chris Jester-Young

@Doorknob You returned nil? :-P

vzn

exactly :p

Geobits

compiler error

luser droog

I think he means IRET

vzn

01:22

lol

NPE

Chris Jester-Young

I like the IRET comment. The rest of you don't know Ruby. :-P

Doorknob

Oh come on can I not say two words without programmers barraging me? ;) Fine:

Doorknob.returnFrom(Doorknob.currentLocation);

luser droog

I liked it.

s.b. come from :)

Geobits

Eww, hard-coded string parameters? (ninja edited away...)

vzn

uh oh. syntax highlighting in this chat. bzzzzt

←ruby geek

Doorknob

01:23

@Geobits Edited? What are you talking about? ;)

What syntax highlighting...?

vzn

you wrote in a different font. excessive use of that is frowned on.

Geobits

Code font is acceptable. For code, especially.

Chris Jester-Young

@vzn Backticks are not "different font" when it's used to mark code.

luser droog

Yes, but "excessive" is subjective.

Doorknob

@vzn I'm not sure how using proper formatting (on one line) is excessive.

ProgrammerDan

01:25

Wow, this just took a turn

vzn

← _RULEBREAKER_

Chris Jester-Young

@Doorknob They're just nitpicking because people hassle them about their excessive usage of bold and italics.

@vzn Hangman? Nice game, not sure of suitability for CGSE. ;-)

Doorknob

@ChrisJester-Young By "people" you're referring to me, of course. ;)

That's only when he was doing this

On every single message

Chris Jester-Young

@Doorknob You're not the only one. I hassle them about that too.

vzn

aieee not that!!!

there is only so much syntax highlighting possible in a chat room. it runs out like memory. 640K. be conservative

Chris Jester-Young

01:27

@vzn That's so 30 years ago.

Doorknob

So yeah, off-topic ftw. :D

vzn

exactly dude

Doorknob

Okay you can stop now

@Chris Mod powarz, activate! :P

Chris Jester-Young

Inorite.

vzn

crazy man. talk about micromanaging. fixing fonts of comments in chat? dude, surely you have better things to do

Chris Jester-Young

01:29

Aren't you happy that that's all I'm doing...for now?

vzn

thx so much man

← capable of real gratitude

Chris Jester-Young

You're the reason freenode introduced channel mode +c. ;-)

vzn

← used to think that chat rooms on se were less restrictive than the sites by nature ... & then found this one....

Doorknob

Try spamming this in any other chatroom....

vzn

← heart racing

most chat rooms are empty & nobody cares, have visited many...

Science Fiction & Fantasy was almost as edgy as this one....

Chris Jester-Young

01:33

@vzn So you just like to hear the sound of your own voice, is that it? ;-)

vzn

find this the edgiest chat room of all visited so far...

Doorknob

Empty > Has people?

Heh, "edgy"

vzn

like to chat. is that ok?

whatever bye guys.

Chris Jester-Young

@vzn Dialogue > soliloquy. ;-)

ProgrammerDan

seeya @vzn

Chris Jester-Young

01:34

@ProgrammerDan ...don't let the door hit your back on the way out. ;-)

vzn

have a good one programmer dan try Computer Science chat hope to see ya & chat somewhere more open

Doorknob

Erm... what was that random spam flag?

Chris Jester-Young

@Doorknob I didn't catch it in time. :-(

Doorknob

Anyway, they somehow started talking about us over at the Tavern for some reason :P

@ChrisJester-Young I didn't either :P "You can only counterflag messages that have been flagged"

DoorknobChatbot

New post! Draw an ant line in mobile

Doorknob

01:36

Ant line??? What is that

Ah, nvm, ^ cv-plz

Chris Jester-Young

Okay, migrated.

Doorknob

Ah, this is how the PPCG discussion started in the Tavern :P

Chris Jester-Young

Hope the gamedev mods don't kill me for doing an unsolicited migration.

Doorknob

Heh

:D

ProgrammerDan

It's mildly amusing to me ... if you open your question with "I might not be in the right place" -- perhaps you should spend a few more minutes finding the right place.

Chris Jester-Young

01:38

@ProgrammerDan Well, to be fair, there are >100 SE sites, so that could be challenging.

Doorknob

Okay, I'm going to do some Texas History homework now (:/). Brb

ProgrammerDan

@ChrisJester-Young I suppose so.

Chris Jester-Young

Heck, even I don't know the correct site for something sometimes.

just because so many new sites get launched all the time.

ProgrammerDan

E.g. the fundamental searching dilemma-- I have a problem, but I don't know what to search for to solve it...

I ran into that a lot when entering a new research field

Trying to find "related works" became more a game of "collecting keyword combinations that relate to my work" either superficially or directly, depending on how broad the field.

I suppose it would be unfair to expect others to be familiar with that process

/end monologue

Chris Jester-Young

I read it. :-) (But mostly, I'm doing work.)

ProgrammerDan

01:49

:) Thanks for the validation, hehe

Chris Jester-Young

:-D

ProgrammerDan

I suppose back on topic, does anyone have suggestions on how to improve this new question? I might be tripping, but it seems like there might be a valid way to make this interesting. It's in the same vein as the Aristotle's number problem, but what made that interesting is when solving, you discover there are actually only a very small number of possible answers, etc. Here, we know a-priori that there are billions of answers.

And writing a program that could enumerate all of those, just seems ... meh. There's no mystery here. But I'm struggling to find a way to suggest to the OP a concrete improvement.

Doorknob

@ProgrammerDan That's not new.

ProgrammerDan

Sorry, linked the wrong one

@Doorknob Meant to link to this new one. I had both open, oops!

undergroundmonorail

03:02

do you guys think "Output the base for which an equation is correct" has potential as a code golf question here? for example, ./program <<< '6*9=42' should print 13 because 6 * 9 in base 13 = 42

i haven't put it in the sandbox because it's late and i don't feel like writing spec

Chris Jester-Young

@undergroundmonorail Do you actually have an algorithm for solving it?

undergroundmonorail

i have an idea of how i'd attack it but i haven't formalized it yet

Chris Jester-Young

@undergroundmonorail It's most ideal if you can actually solve it before posting on the sandbox, but of course you can ask for help from other people too.

undergroundmonorail

yeah, for sure, i'll get a shitty python script or something running before actually posting anything

part of the reason i want to post it is because i'm curious if there are any cool ways to do it that i hadn't considered :P

DoorknobChatbot

06:38

New post! Taking Beal's $1,000,000

New post! Anonymize names - sort of

Peter Taylor

07:49

@ProgrammerDan That question is just above the "Count from 1 to N" that I mentioned recently in a comment in the "'Too trivial' close reason" meta thread, because it requires you to count from 1 to N and filter.

luser droog

08:36

This paper looks pretty awesome: Implementing Primitive Datatypes for Higher-Level Languages, 1988. citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.385.485

luser droog

10:53

Tornado warning.

But I still have power and internet. :)

ProgramFOX

Ah! That means that my jQuery Tornado plugin is working. :-)

mniip

Isn't Tornado a python websocket library?

luser droog

I don't know. :( I'm so behind the times.

I still haven't bothered to learn python or javascript.

But I did try reading some Python not too long ago.

2

Q: Python raytracing distorts object shapes when close to the camera

The Problem I recently found someone's awesome little pure-Python raytracing script from this link, and extended it a little bit for more convenient functions. However, sometimes it distorts the shapes of the objects and I'm wondering if someone with raytracing/3d experience might have any clue ...

python graphics 3d raytracing

ProgramFOX

@mniip Yes, it is:

> Tornado is a Python web framework and asynchronous networking library, originally developed at FriendFeed. By using non-blocking network I/O, Tornado can scale to tens of thousands of open connections, making it ideal for long polling, WebSockets, and other applications that require a long-lived connection to each user.

Source: tornadoweb.org/en/stable

Rusher

14:08

Omg I have the Stack Exchange bug.

DoorknobChatbot

New meta post! URL shortener policy

Rusher

I just sent an email, and noticed something I could have worded better, and I immediately looked for an edit button...

@ProgrammerDan Grats!

ProgrammerDan

@Rusher Thanks! 'twas fun. Good thing is, I've tested and double tested, the latest version of the solver is correct. Had to gut a lot of things I initially thought improved performance, but actually just reduced the ability to find solutions. It was always a valid solver, just not a very efficient one at first :D

Rusher

14:29

I just tested it as well. I had left it running overnight actually, but came to find an updated solution

I basically have to beat you before I post another challenge

ProgrammerDan

Hahaha, great! Are you still planning to port it, or are you devising your own solver method?

Rusher

I am slowly building my own solver. When I said port, I meant port my own. I will port it to Python after I build it in Java

For some reason, I feel like I can write fluidly in one language, but am better at minifying it in another.

ProgrammerDan

Gotcha! Awesome. I look forward to seeing it in action. I'm tempted to make a direct adaptation of MT0's solution sketch -- his principle is sound, and it might be possible to make a more general solver using it, although I think you'd still need to efficiently search for a route to the lowest cost path between nodes, and that basically resolves to my solver.

A simple look-ahead, lowest cost heuristic will fail to reach the lowest cost path node if the route is cleverly devised. You might be able to implement a shortest-path algorithm between either v1 or v1 of path p and the outpost 1, perhaps -- just minimize in terms of cost (ammo lost).

Or ... if you really want golfing goodness, use a genetic algorithm ... although then you cannot claim it will always find the winning solution, or that it will actually terminate

but it would be fun, and I imagine you could make it concise.

/end monologue

Rusher

I've never implemented a genetic algorithm or studied them. Maybe I'll try. Working on the brute force version first (as usual)

ProgrammerDan

They're a pretty awesome search technique, if your optimization function is very well defined. Less so, if that condition is not met.

The most fun I've had with programming, probably ever, was writing a program that used GA's to search the possible constructions of neural networks, which I used as "brains" for simulated bugs in a maze.

The optimization function was complex, took many iterations to refine, and I probably couldn't explain it again now if I needed to, but the best was watching bugs slowly get better at finding their way through mazes. The inputs to the NN were two "rays" that indicated the presence of a wall and its distance. The outputs controlled rotation and speed.

Rusher

14:47

Oh man... you are going to have a blast when I post my KotH challenge

ProgrammerDan

:D :D

Rusher

I have high expectations for you. Build a hive minded wolf XD

ProgrammerDan

Oh yesssss

I am so looking forward to that. I wasn't planning on using NNs for that, but hey -- why not?

Rusher

I have two problems keeping me from posting. One is due to laziness (building an interface to make it language agnostic). The other you might be able to help with.

I wrote that the size of the map will grow with respect to the number of submissions being tested

ProgrammerDan

I'll have to read the challenge carefully to see how I can subvert it, so that I can appropriate refine the brain before and/or during the simulation itself. The issue with GA's and NN's, is that getting a better NN is very slow (generally) due to the immense complexity of a NN solver large enough to handle complex inputs.

Ah yes, the map size issue

Rusher

14:49

That means that if someone wrote an algorithm that attempted to reproduce the map, they would have more and more trouble as the map grows

So the conditions for testing are constantly changing, and that is not ideal

ProgrammerDan

I suppose. I'm curious why you are using a cell-based map, instead of an object-centric map

Rusher

What does that mean?

ProgrammerDan

Basically, just represent the map as a set of looping bounds, and the location of each object in real-space.

discrete cells aren't as interesting, although they are easier to represent as ASCII :)

lemme bring up your meta post again and refresh myself on your map construction, brb

RIGHT, okay. So you plan to pass just a local neighborhood to each animal. And your concern is that an algorithm to reconstruct the map based on the local neighborhood of all animals is hard?

Rusher

It's not that it is hard or easy. It's that it gets hardER as the map gets larger

On a tiny map, you can easily gain information about the map and possibly form a wolf pack. On a large map, you may be lost forever

Someone may build an awesome wolf that gets progressively worse as the number of submissions increases

ProgrammerDan

Right, right. Good point. One issue -- how many Wolves of each type are spawned? How can a wolf know that another 'W' it see is infact, itself?

Last thought ftm -- the win condition is survival. This would mean that the winning solution won't be a wolf, but a sheep in wolf's clothing. It would be relatively simple (and fun) to construct a "Wolf" that worked together to avoid everything, and survive as long as possible.

Rusher

14:57

Hold on your first question. To your second question, I suggest "dancing" with any wolves you see. If you both move up and then down, it might be friendly.

Ok so for your first question: Simulation begins at ~25% board capacity with equal amounts of each Animal pseudorandomly distributed across the board.

The board is a square with sides of length sqrt(n+3)*20 where n is the number of submissions.

So you begin with sqrt(n+3)*20 divided by 4 divided by (n+3)

the +3 is from the Stones, Lions, and Bears

ProgrammerDan

Hmm, okay. So you're using the length of the side as your numerator?

not # of cells in the map?

Rusher

Well, if there are 6 submissions, we get a 60x60 size board. That will be filled to 25%, which is 3600 Animals. 3600 cells will be divided among the the 9 different Animals, so that is 400 per.

ProgrammerDan

b/c if you use # of cells, you get (sqrt(n+3)*20)^2 / 4 / (n+3) = 400*(n+3)/4/(n+3) = 100*(n+3)/(n+3) = 100 ... unless I'm still too sleepy for this.

Rusher

Hmmm... I guess you can reduce it that way.

ProgrammerDan

which is just a constant, and that seems wrong to me, so I'm going to run the math in python

Rusher

15:04

How did I end up with 400 then

ProgrammerDan

well, 25% of 60x60 is not 3600

you didn't actually do the division

Rusher

hahahahah

ProgrammerDan

;P

Rusher

Alright. You begin with 100 Animals! Thank you for that. I wish I could edit my earlier comment.

ProgrammerDan

Hahaha, no problem. The nice thing about that, is when building your animal, you don't have to deal with a varying number of animals. That's actually kind of nice, although you are right -- a hive-minded animal that works well when it can basically search the whole map will likely do worse when the map is larger. Although the map will be "just as full" when you add more animals. :D

I love how the bear protects her territory, the rock just sits there, and the lion just roams in a straight line.

Rusher

15:09

Yea, I got the population density down pat. I wonder if I should post the Lions, Stones, and Bears as community wiki answers to my own question.

ProgrammerDan

That would probably be helpful, at least to have them posted in some way.

Rusher

Did you see the gif I posted with it? drive.google.com/…

ProgrammerDan

I still think that the winning strategy will be general avoidance, perhaps Loki'ing it up by removing all the rocks and any other fixed navigation items to bugger the other wolves ...

No I hadn't seen that, that is freaking awesome :D

Rusher

Remove rocks? I'm gonna sit right next to any rock I find lol

Geobits

Once you can find some teammates, I think a basic flocking/boids-like pattern would work well.

ProgrammerDan

15:13

@Geobits Yeah, that's a good point too.

The challenge is finding teammates, as all submissions will be labelled 'W' in the surroundings map, and there is no information that fixes you in the world, so determining that you've encountered a teammate could be difficult.

Rusher

If I gave passed in the number of submissions, would that help?

Geobits

Agreed. I think if you can get a pack together, things are much easier from there. 3x3 is pretty unhelpful for that, though, while anything larger might make it too easy.

Rusher

I don't want to go 5 x 5, because you could avoid all collisions completely. 3x3 forces some deaths.

Geobits

Yeah, that's more-or-less what I meant, came out wrong.

Rusher

What will be interesting to watch is if someone comes and wipes out a wolf you thought was friendly, you won't know and you may continue following an enemy

Or if someone submitsa wolf that actually attacks wolves and encounters itself, they will swap places endlessly

Geobits

15:19

One option is to just assume all Ws are friendly when you see them. If your pack moves consistently, there's then no chance of wiping out teammates. If it's actually hostile, then you're still better off because you'd probably have had to fight it anyway.

ProgrammerDan

@Rusher So friendly wolves can't attack each other?

Rusher

They can, but if I attack you and you attack me, guess what happens on a discrete board?

DoorknobChatbot

New meta post! Can we change the skin to make code tags in spoilers more readable

ProgrammerDan

Ah, I see. Because an attack is actually just a move.

Rusher

Right, we would swap places and never get to play rock paper scissors

ProgrammerDan

15:28

This is why I generally only work with continuous maps :D

@Rusher regarding that problem, I'm guessing you haven't settle on a way to make the challenge language agnostic?

Rusher

Well, I'm not sure where to begin really. Peter Taylor said something about pipes and stdin and stdout that I am not familiar with.

Peter Taylor

@Rusher I don't see how you gain useful information about the map. Each of my wolves is a separate instance of the class, right?

ProgrammerDan

@PeterTaylor within java, you could use static methods to emulate a hivemind

Rusher

@PeterTaylor You can persist data, so that all of your separate wolves can, together, build a complete map.

ProgrammerDan

or that

Peter Taylor

15:40

That sounds like it has the potential to be really slow

ProgrammerDan

@Rusher I'd suggest just creating an API based on command line arguments. For instance, to "initialize" a new external program, execute it and pass it a number. This number indicate which "Wolf" of that type the external program should init. Then, when you invoke the external program in the future, just pass that number along to indicate which Wolf you are interacting with. Rounds can conclude with any number of Attacks, so initially you start a round by invoking a Wolf with '<id><mapdata>'

Peter Taylor

My suggestion was based on the assumption that each wolf would be independent

ProgrammerDan

where <mapdata> is a 9-char string describing the Wolf's surroundings. You expect back from the process on STDIN the move to take, in the format R,L, etc. If within the round attacks need to be processed, invoke the process again using `<id>A<attackerchar>' and expect back some enum char value indicating Rock, paper or scissors.

External programs would have to store state on disk, and would need to keep track of when a round has begun in case that impacts their attack strategy. It would be up to the external program to keep track of which wolf was "acting" as well.

Geobits

After looking over the spec again, I have to wonder how you get any real benefit from the extra work of building a map or emulating a pack.

ProgrammerDan

@Geobits I still believe the winning strategy is avoidance, as much as possible.

Rusher

15:44

@PeterTaylor Each wolf is independent. I don't understand what that has to do with persistence. You and I are independent, but you are perfectly capable of sharing your knowledge with me (and you do, quite often)

If 100 wolves persist all of their knowledge to the same file, you could draw a half decent map and say "Maybe I am lost, but I know where I am relative to my buddy"

Geobits

My possibly-naive startegy: Do nothing. If a lion is up|left of you, move out of its way(they might kill you). If a wolf is near, just sit there. It will either be one of yours(which will just sit there), or an enemy. If the enemy's strategy is to full-on chase, it will eventually catch up to you, so avoidance really doesn't help.

Peter Taylor

If we're sharing knowledge then we're interdependent, not independent

ProgrammerDan

@Rusher that is a very difficult challenge, as the surroundings sent to each wolf at each round individually do not contain enough data to "combine"

Geobits

Otherwise, it's just rolling the dice whether avoiding will run you into another creature.

ProgrammerDan

@Geobits Actually yes, that sounds like a better strategy based on the current rules

Geobits

15:47

Boring as hell, but probably pretty effective.

ProgrammerDan

Another fun strategy might be to "latch on" to a lion

not sure exactly, but I think it would be possible, due to discrete cells, to follow the lion. Allowing it to "attack", and you can wait safely behind it.

Geobits

Yea, if you can save state you'll know exactly where it's going after the first move.

Rusher

I've half a mind to post this and require Java and accept the downvotes that I will inevitably get

ProgrammerDan

Extending to non-java wouldn't be terribly hard, just slow

(considering 10k iterations, with 100 of each Wolf to invoke each iteration)

Rusher

I agree, but I promised someone that I wouldn't post it until I made it language agnostic, and asked for upvotes on meta assuming everything else was OK.

Geobits

15:53

@ProgrammerDan Aw come on, it's only a million invocations per submission :p

ProgrammerDan

@Geobits :P True, what was I thinking. That shouldn't take too long! :P Oh -- and a million invocations is just a lower bound. If attacks occur, that number increases quite a lot!

Rusher

1000 iterations * (100 moves + 100 attacks) maximum

Geobits

Well, attacks between wolves would lower it, but other than that...

Rusher

Oh, times 3. Because of 3 trials

ProgrammerDan

there we go

Or ... you could keep the process alive

and just maintain 100 processes per wolf class

instead of invoking each time, just send more data to STDIN of the process ...

^.^;

Rusher

15:56

Oh I know.

What if Wolf is a static class, and I pass in a location as an arg

ProgrammerDan

wait. Just construct the scheme so that you use exactly 1 ProcessBuilder per external invocation. You use the <id> based scheme I mention above, and it's up to the external process to determine which wolf you are invoking when you send data along STDIN to the process.

Rusher

So rather than 100 wolves, you only need 1 that responds accordingly

ProgrammerDan

@Rusher meh ...

Geobits

If you pass in a location, that somewhat defeats the purpose of only giving a 3x3 FOV, right? You could build a full, accurate map without much trouble.

Rusher

You only know your own location though. What does that say about the rest of the map?

ProgrammerDan

15:59

So, you would need to make exactly 1 invocation of the external program, and could just keep the pipes open and interact with that external process throughout the course of the 10k iterations.

@Rusher everything, b/c you can share knowledge between wolves.

Rusher

But you have no idea where everyone else is. Passing in a full map would be different...

Geobits

You don't even have to save state. The static wolf would have its own map, just updated with the local surroundings on each run.

ProgrammerDan

@Rusher If you get 100 3x3 blocks, over the course of 1 round, by the second round you'll know a goodly portion of the map, since when you pass in location, you can fix those 3x3s in an absolute sense against the complete map

Geobits

And you'll know exactly which wolves are yours

Rusher

Yea ok I'm trashing that idea lol. Will look into ProcessBuilder and see what it takes to communicate with Python

ProgrammerDan

16:01

@Rusher it's super easy

Rusher

I've used it to open Notepad lol

But still, how does this solve my problem if someone submits an answer with Mathematica?

I'm not going to buy it lol.

Geobits

Just restrict it to languages with freely available compilers/interpreters.

I can't see anyone really complaining about that.

Rusher

That sounds reasonable.

DoorknobChatbot

16:55

New post! Php divide a input into 3 digits

Transcript for