GunDuel KOTH is posted

New Main Posts

Dec 28, 2016 09:17

0

The Background Future In the year 2017, you and your opponent will face each other off in a futuristic gun battle where only one player may survive. As a player, are you experienced enough to defeat your opponent? The Futuristic Gun Duel General rules Each player starts with an unloaded gun ...

Geobits

@Downgoat Oh, and guns ;)

ATaco

C++

Geobits

Haven't used that in a long time, and don't plan to for this, as much as I like KotHs.

ATaco

Most KotHs use either Java or STDOUT, this is strange to me.

Kritixi Lithos

But the idea is interesting

Geobits

Dec 28, 2016 09:20

A few use python, but yeah.

It's similar to one we've had before, but probably not so close that it would be a dupe imo.

ais523

it's basically an advanced version of rock paper scissors

but being advanced means it breaks the symmetry, which is a good thing

Geobits

Hmm. Actually, I don't see the point of defending much. You can't win a point with defense at all, no matter what your opponent does. Yes, you can block plasma, but you could also do that by shooting plasma. If they're shooting a bullet instead, you'd be better off shooting plasma than defending.

ais523

you can do it if you're out of ammo

and the opponent isn't

(hoping that the opponent shoots and loses their ammo)

Geobits

True, but you also have to hope you choose the right shield. The odds are just bad for defense in general.

And then you're still out of ammo.

ais523

actually, probably the best time to use it is if both players have 1 ammo

you know you don't need to shield plasma that turn, the opponent can't shoot it

and if the opponent shoots and you shield, you have a huge advantage

meanwhile, if you both shoot, the best you can do is a draw

let's see: at 0 ammo each, you build ammo, no point in doing anything else

at 1 ammo each, the rock-paper-scissors is build beats shield beats shoot beats build

so if you neglect shielding there, it becomes a game of rock-scissors which… isn't very interesting

zyabin101

Dec 28, 2016 09:29

@NewMainPosts DumbBot: on odd turns, output 0, on even, 1.

Geobits

Meanwhile, if you shield at 1 each, the opponent can get a free reload and have plasma next turn.

zyabin101

RandomBot, on turn 0, 0, then randomly in 012-=.

ais523

@Geobits yes, that's where the rock-paper-scissors comes from

ATaco

Plasmabot, 002002002...

Geobits

@ATaco 01

ATaco

Dec 28, 2016 09:31

Yes, that would win, but only against bots that assume I'm going straight for plasma.

Geobits

I can almost guarantee there will be at least one "DumbBot".

ATaco

It seems a better idea for most bots to play somewhat defensively first few turns.

zyabin101

@Geobits Yes.

ATaco

0-=1

zyabin101

And my "standard koth bot set" for this koth will be half done.

ATaco

Dec 28, 2016 09:34

Hmm, 0-=1002002002002002...

Geobits

This seems like a good toy project to throw at a genetic algorithm for a few days, tbh. Simple rules and limited choices.

Fitness is easy to measure.

zyabin101

How?

Geobits

You either die or live.

ATaco

Better, Fitness = Turns until win.

With Losing being an Ultra High Value.

Geobits

If a bot wins more often at turn 47, I don't really care that another one can win less often at turn 4.

ATaco

Dec 28, 2016 09:35

Or simply, Fitness = Turns untill game end, with a bonus to winning, giving preference to longer games.

Geobits

A win is a win in this case.

ATaco

There's no reason to let it roam freely though, a variable fitness value for winning and punishing losing gives the same benifits as just +1 for win -1 for lose.

Geobits

But why make it complicated? There's no reason to select for longer/shorter games either.

ATaco

Good point.

Although, you may want to keep it way from the 100 limit.

Geobits

Oh sure, you count the 100-turn-limit as a loss.

Maybe even cut it off at 90 or so.

ATaco

Dec 28, 2016 09:38

I'm writing a lua interpreter for this challenge and a simple genetic algorithm to put it up against.

Geobits

Keep us posted :D

Ah, this is the one I was thinking of. Like I said, similar, but not really a dupe.

I tied for 7th in that one :/

ATaco

Lookin' good so far.

zyabin101

Dec 28, 2016 09:54

I declare: I absolutely know nothing about making angular apps.

mınxomaτ

Dec 28, 2016 10:06

@ATaco Is that Sublime? (pun not intended)

ATaco

Yeah.

zyabin101

0

A: The Futuristic Gun Duel

The GunClubPlayer Cut out from the original question. This is an example player that will participate in the gun duel tournament. The GunClubPlayer likes to go to the gun club. During each duel, he would first load ammo, then fire a bullet, and repeat this process until the end of the w...

Seriously, this one is not called DumbBot, even though it is a dumb bot.

Also, I had another Koth idea in there.

Let me write it in a notepad and copy it here.

Geobits

failed to survive is a polite way to phrase that :P

ATaco

I do try.

Alright, Running 90000 games on two bots.

They are really not very smart.

They find that just using shields makes them REALLY safe.

Qwerp-Derp

Dec 28, 2016 10:22

Should I learn Assembly?

flawr

I think it would be nice to have this koth challenge translated to Java or another more accessible language, any objections?

zyabin101

A number is awesome over another number if it is at least twice as larger as the other number.

Each bot has a signpost and a hammer.
  [9]
 O/
 |-#
/ \

The signpost shows a number from 0000 to 9999 (four digits).

The hammer gets power to hit another bot if its owner's number is awesome over the other bot's number.

Each fight consists of n + 2 rounds, where n is an unspecified number, and 3 phases.

=== Start of Fight
 |
 === Phase #1: Regular
  |
 --- Each bot sets their number, shows their number *simultaneously* (as will be the case in all other rounds.)

^ There, my thing.

ATaco

The bots aren't very smart, even after 90000 cycles.

zyabin101

@Qwerp-Derp If you want to not know what you're doing, yes. :D

ATaco

2:0	=
1:0	0
0:2	=
0:0	0
1:1	1
3:1	=
0:1	-

I've removed the bots abilities to think doing 1 or 2 without the valid ammo to ever be a good idea.

Qwerp-Derp

Dec 28, 2016 10:31

@zyabin101 Have you played around with Assembly?

zyabin101

A bit.

Qwerp-Derp

What did you use for Assembly?

ATaco

At a random interval, one bot was only doing 01010101 whilst the other just did 0-----------------

I have lost faith in my bots abilities.

They're really shield happy...

Running 1000000 games, Hoping the bots learn to not just idle for 90 turns...

...Lowering that to 100000

flawr

Does anyone here know Java?

ATaco

Kinda.

Kritixi Lithos

Dec 28, 2016 10:43

^ same

flawr

Ok problem solved, thank you for your help!

:D

Geobits

@zyabin101 I don't understand why any bot would ever pick anything other than 9999

@flawr No problem ;)

ATaco

Bots are now a bit smarter, but still fail to get much of a brain going at all.

I admire their courage.

Kritixi Lithos

What are the numbers?

zyabin101

@Geobits IDK

ATaco

Dec 28, 2016 10:46

MyAmmo:TheirAmmo Response

zyabin101

This idea is just an idea, and has to be improved.

ATaco

The bots are just Genetic Algorithms for Lookup tables as to what to do for each ammo.

Geobits

@zyabin101 Well, yeah. "Pick the biggest number" isn't going to be a very interesting KotH.

@ATaco Ah. My idea was to just generate random strings for the bots, consisting of the allowed characters, and pit them against each other.

ATaco

Well, that -would- work, but it's not very smart.

Then again, neither are my bots...

Geobits

You don't really need smart. You need winners. Doing it either way could work for that though.

betseg

Dec 28, 2016 10:49

@Geobits was your hat's tilt intentional

ATaco

I'm probably going to need to run mine for a while.

1:2	0
0:0	0
1:1	-

That is not a clever idea, bot.

...I want to automatically name them now...

Geobits

@betseg The control line is straight. If it's tilted, it's because the hat itself is.

ATaco

Their brains are getting pretty smart...

For dumb robots, that is.

Going back to the 1000000 test. Let's see what we generate!

Geobits

How many bots do you have in each pool? I'd think having each one tested against a bunch would work better than 1:1 for each generation.

ATaco

Right now it's just 1:1, I could easily add any number of bots to the pool though.

Geobits

Dec 28, 2016 10:54

Maybe get a group of good 1:1s, and pit them against each other :D

ATaco

Which I'll do, because after 1000000 tests, the bots didn't seem much smarter.

I could do teams

5 vs 5

Geobits

A round robin between the whole pool each generation would probably be the most comprehensive. Slower, but...

ATaco

25 * 100000 cycles, let's see what complexity is born.

Maybe I should start them on 10000...

Geobits

Sadly, the best ones probably won't be all that complex, but you never know.

ATaco

I don't think they'll map to any more than say 5:5.

New Main Posts

Dec 28, 2016 10:58

0

Q: Shrink your language's character set

We often get restricted-source questions here on PPCG that favour submitting solutions using only a subset of characters (sufficiently so that there's a tag printable-ascii that identifies a particular subset of those challenges). Wouldn't it be nice if we could automatically cut languages down t...

code-challenge source-layout

Geobits

And of course, these are all static bots, so can't respond to what the other is actually doing. So it won't spot patterns that more sophisticated ones might be able to.

ATaco

The bots actually did get a bit smarter.

DrNewBrainBot
2:0	2
1:0	0
2:2	2
1:2	-
2:4	1
0:0	0
1:1	0
2:1	1
0:5	-
0:2	0
2:3	=
0:1	-

It's hard to decide which bot would be best, but the 10 to pick from are all round pretty smart.

Geobits

@ais523 I don't really want to write the implementation, but want to know if I'm reading this right: So if I took in an input Java program, for example, I could convert it all to /uxxxx codes for a character set of size 18?

ais523

@Geobits yes, I think that works (Java allows \uxxxx in any context, doesn't it?)

Geobits

Yeah, it gets converted before it compiles iirc.

ais523

Dec 28, 2016 11:02

of course, your score there will be limited by the fact that a) you're still using 18 characters, and b) you have to write the program in Java :-P

ATaco

I think this is a duplicate, sadly.

Geobits

Sure, just making sure I read it right, like I said. Of course the output will be huge, too, since it's 6 chars for every 1 input.

ATaco

2

A: Encode a program with the fewest distinct characters possible,

RProgN, 7 Distinct Characters, Noncompeting RProgN is newer than this question. ' do.Lc Convert the program to 'oooo...' L c 'ooooo...' L c 'oo...' L c . . . do, where each 'oo...' represents the character code in o's, surrounded by apostrifies to make a string out of them. L c then turns them...

Disregard particular answer.

I just grabbed it from my answers...

ais523

I checked for duplicates but didn't find that one

the new challenge is probably better though, the old one's very vague

ATaco

Better normally isn't allowed, but I'm no-one's boss.

ais523

Dec 28, 2016 11:04

not sure whether to close as duplicate, reverse close as duplicate, or leave both open

ATaco

I've vote to closed the new one.

ais523

bleh, Lenguage postdates the challenge :-(

but Unary doesn't

let me go exploit that

(actually, Unary is easier than Lenguage…)

ATaco

Some bots that were AUTOMATICALLY GENERATED!

ais523

but in a very vague way

or, hmm, it tried to ban Unary

Geobits

> TacoGeobots

ATaco

Dec 28, 2016 11:08

Are you judging my bots randomly generated names?

Geobits

Totally

I used to go by Geobot many years ago.

ATaco

Alright, It's now a 10vs10, This may take like, 10 minutes to load, but hopefully the bots will be hella smat.

Geobits

Maybe even wicked smat

ATaco

Now to just learn C++ so I can submit one of these bad boys.

Geobits

If all you need is a lookup table, you shouldn't need to learn much at all.

ATaco

Dec 28, 2016 11:13

I should mention, these bots have not been taught to play games any higher than 10 rounds.

As such, their brains cannot get more complex than "9:9"

Geobits

Push that to 20 and you'll probably cover most situations. 10 might be a bit short, but maybe not.

ATaco

Ooh, Brains are done cooking!

Some more bots, Very smart, Would recommend.

Geobits

I may try something similar, but tomorrow. I've put off sleep long enough.

ATaco

I admire 'RobotGeorge' for only shooting when he has the advantage.

And, good night friendo.

zyabin101

Team Red's ` ` is kinda good.

ais523

Dec 28, 2016 11:18

I have my own plans for the KotH but I think I'll wait for the hill to get populated first

also I'm not sure I really want to enter in C++

zyabin101

@ATaco Your bot name generator is kinda borked >_>

ATaco

Itsafeature.

t.name = ""
	for i=1, math.random(0,2) do
		t.name = t.name .. nameStarts[math.random(#nameStarts)]
	end
	for i=1, math.random(0,2) do
		t.name = t.name .. nameBodies[math.random(#nameBodies)]
	end
	for i=1, math.random(0,1) do
		t.name = t.name .. nameEnds[math.random(#nameEnds)]
	end

zyabin101

Require at least one name body.

ATaco

I may change how the fitness works to give better thinking bots.

I will allow it, But I was keen for Mr but FASTER

ATaco

Dec 28, 2016 11:32

Bots now, instead of instantly receiving punish and reward calls after each game, total their wins across all opposing bots. If they win more on average then lose, they receive reward, otherwise, punish. As before there were 10 bots on each team, it was possible to have a 50/50 split, so botsperteam has been lowered to 9.

Of course, now each of the 100000 games these bots get to play for my amusement takes a little longer, but the reward is harder better faster stronger bots.

The new wave of robots has arrived.

zyabin101

Bot would be a nice name for a bot.

RobotRobot could be a good name for a bot.

ATaco

It's amazing these bots did as well as they did, noting their Fitness was broken and set to a permanent 9xGames

Kritixi Lithos

Dec 28, 2016 11:56

Holy cow, I got 50 rep in just 30 minutes in Puzzling

ais523

that's five answer upvotes, which isn't really a lot

rep is very easy to get on SE if you want it

especially if you're willing to use underhanded tactics to do so

(like submitting really easy questions to PPCG)

Kritixi Lithos

It's the fastest (and easiest) rep I ever got

At least I'm closer to Running Ragged now

Falcon

I got duplicate marked on my question, I edit it but the person who marked duplicate didn't remove duplicate tag... and I got downvoted for that... What to do? :(

Kritixi Lithos

Which question is it?

Falcon

0

Q: Tips for posting interesting challenges in ppcg

I'm looking for some interesting ideas which I can apply to code-golf problems which keep track of programmers interest as well as some what tricky to solve. It can be of any language (e.g Python, R etc). The task should rely on algorithm rather than languages. The purpose is to capture programme...

code-golf tips

ais523

Dec 28, 2016 12:09

you edited an old closed answer into an entirely different question

you probably shouldn't be doing that? it makes it very hard to follow what's going on

(also that's a duplicate of a post on Meta)

Falcon

So, how can I change my question as that other person said its a complete duplicate. So, I have to change it.... I'm sorry I'm new to this so I don't know much...

Should I remove my question?

ais523

yes, better to delete it and post a new one

you should be able to delete your own question

although, with duplicates, sometimes the duplicate question can help people searching for the question

Falcon

but doesn't the duplicate gets downvoted?

ais523

if it gets downvoted, that's normally a sign that people think it's worth deleting

Falcon

Doesn't a duplicate eventually gets deleted by people who got more badges then non-bagers

Isn't this something wrong btw, its actually use of power (kind of)

ais523

Dec 28, 2016 12:15

deletions aren't really related to badges, but to reputation

and actually, most deletions are done by Community (which is a bot) on posts that are downvoted, closed, and have no answers

Falcon

Its' written that code golf users with badges can single handedly close questions as duplicates and re-open them

this is not some bot closing or deleting it... right?

ATaco

I do not love C++

No sir

ais523

@Falcon that's possible but it's not the normal reason; the vast majority of closures on PPCG happen as a result of five close votes

and most of the rest happen as the result of moderators "shortcutting" a vote that's inevitably going to go one way or the other

Falcon

Ahaaa.... thanks @ais523.... :)

ATaco

My C++ just swore at me.

muddyfish

Dec 28, 2016 12:34

When I try and get the turn number and opponent ammo, it gives me not the correct values: Turn -858993460 ammo: 0 opponent's ammo: -858993460

ATaco

Dec 28, 2016 12:49

The winner of my messy Gene Pool is BotRobot, a name that I could not have made better myself.

zyabin101

@ATaco BotRobot might be a great name for a bot.

ATaco

BotRobot is his name. I don't plan to rename my victor...

The Nineteenth Byte

Participants