Automatic Plagiarism
Disclaimer
I don't approve of plagiarism, but I thought it would make for an interesting challenge
Challenge
Now we've got that over with, it's time for the challenge. What you have to do is write a script that takes a piece of work (whether it's from online or from a tex...
Some days I just really want to destroy the 1100 Visual Studio warnings we have in our solution. Goodbye unreachable code. Goodbye variable that was declared but never used. Goodbye conditional statement that always results in true.
> There were 3225 bots with equal flags
> FreezeTag had 1159 points
> FastMoveCloneDodge had 952 points
> SuperFreeze had 939 points
> FastForwardClone had 785 points
> TheCommonCold had 742 points
> RovingVirus had 711 points
> ForwardClone had 549 points
> HideBlockAttack had 478 points
> QuickFreezerbot had 402 points
> Randomcopier had 384 points
> Copycat had 270 points
> BlockFreezeAttack had 265 points
> RushAttackDodge had 260 points
> QuickFreeze had 251 points
> BoringCopybot had 244 points
> There were 3328 bots with equal flags
> FreezeTag had 1151 points
> RovingVirus had 938 points
> FastForwardClone had 864 points
> TheCommonCold had 724 points
> HideBlockAttack had 628 points
> QuickFreezerbot had 530 points
> ForwardClone had 524 points
> RandomCopier had 507 points
> BlockFreezeAttack had 423 points
> Copycat had 386 points
> RushAttackDodge had 282 points
> QuickFreeze had 281 points
> BoringCopybot had 265 points
> RepairAndProtect had 225 points
> Magus had 203 points
Is it worth creating a complete data dump (all bots at all steps) for identical runs in python and java in order to highlight any differences in what should be identical output? Having two implementations seems like a perfect opportunity for cross testing.
@NathanMerrill yes I do love the python %. It has some unusual cases but for positive modulus it behaves nearly always exactly how I'd want it to.
Chess king-of-the-hilljava
General
This is a chess-KOTH with simplified rules (because chess itself is already complicated, playing it via a simple program isn't making it easier). At the moment it is limited to java (version 8), but creating a wrapper-class isn't that difficult (in case someon...
@NathanMerrill yes you might need to use a non-standard PRNG to ensure they are both using the same one, but you could write a very basic one just for that purpose, and not use it in the real contest. A one liner should do it
Other than that the fact that it's all integers should ensure the results are identical (or that where they are not is evidence of a bug which would be useful to know about)
A home made PRNG could be added to both as a one line function - for testing purposes I imagine something as simple as the following would suffice: t += 1; return (t * p) % 24 for some constant p coprime with 24
I don't see that a difference in program layout should affect the order in which random numbers are requested from the PRNG unless one of the implementations is parallel, which I wouldn't expect since the bots are updated sequentially.
There are some interesting (though potentially useless) things you can do with in-line addition. For example, A+A+A+A==B+B+B+B returns true if A is congruent to B mod 6.
@NathanMerrill Ah, you want to sort bots by score?
List<Bot> sort(final Map<Bot, Integer> scores) {
List<Bot> bots = new ArrayList<Bot>(scores.keySet());
Collections.sort(bots, new Comparator<Bot>() {
public int compare(Bot a, Bot b) {
// Sorts ASC; to sort DESC negate the value or swap a and b
return scores.get(a).compareTo(scores.get(b));
}
});
return bots;
}
@NathanMerrill right now, something in one of the published bots is leading to arg==26 and an exception on this line: return person.actions[arg].name == "Flag"
almost done with a full run with cancer and insidious etc
> There were 2975 bots with equal flags
> Insidious had 1122 points
> FreezeTag had 1059 points
> RovingVirus had 1019 points
> SuperFreeze had 1018 points
> FastForwardClone had 844 points
> FastMoveCloneDodge had 822 points
> TheCommonCold had 774 points
> ForwardClone had 550 points
> HideBlockAttack had 500 points
> QuickFreezerbot had 452 points
> Randomcopier had 400 points
> BoringCopybot had 339 points
> BlockFreezeAttack had 332 points
> Copycat had 266 points
> QuickFreeze had 246 points
@overactor so far my best debugging approach has been to change the settings to run just one copy of my bot against one copy of flagbot, and print out their Actions every turn
@NathanMerrill __repr__ or __str__ for Argument, Action, Bot would be really handy for debugging purposes.
I golfed a line off of Insidious: Move If D #3 #0 Copy 0 C Copy 3 C Copy #11 *#*C Copy #12 *#*C Copy #13 *#*C Copy #14 *#*C Copy #15 *#*C Copy E D Copy 0 C Flag Copy C+3 A Copy #C+22 #A Copy A+1 A Copy C+21 C
@NathanMerrill I lost the stacktrace, but the gist is that return person.actions[arg].name == "Flag" was occurring where arg==26, and 26 is larger than len(actions) so it was bailing out
> There were 1371 bots with equal flags
> FreezeTag had 614 points
> Insidious had 570 points
> FastForwardClone had 557 points
> SuperFreeze had 480 points
> ForwardClone had 442 points
> TheCommonCold had 286 points
> FastMoveCloneDodge had 273 points
> QuickFreezerbot had 255 points
> HideBlockAttack had 244 points
> BoringCopybot had 215 points
> Randomcopier had 190 points
> RovingVirus had 183 points
> Copycat had 178 points
> RushAttackDodge had 172 points
> QuickFreeze had 138 points
> Flagbot had 128 points
Here is a version with a slightly faster reaction time if it rotates and sees another bot:
Move If D #3 #0 Copy 0 C Copy 3 C Copy #11 *#*C Copy #12 *#*C Copy #13 *#*C Copy #14 *#*C Copy #15 *#*C Copy E D If D #3 #2 Flag Copy C+3 A Copy #C+22 #A Copy A+1 A Copy C+21 C
> There were 3371 bots with equal flags
> SuperFreeze had 1137 points
> FreezeTag had 1099 points
> FastForwardClone had 947 points
> RovingVirus had 921 points
> FastMoveCloneDodge had 914 points
> Insidious had 810 points
> TheCommonCold had 671 points
> ForwardClone had 571 points
> HideBlockAttack had 488 points
> QuickFreezerbot had 391 points
> Randomcopier had 311 points
> Copycat had 274 points
> BoringCopybot had 265 points
> Magus had 256 points
> QuickFreeze had 252 points
> BlockFreezeAttack had 230 points
that's a full normal run, I think
Insidious does less well on long matches, since its thralls are mostly defenseless
I might try putting a Move or even a turn into the thrall loop
They aren't sorted, but do these look about right:
BlockFreezeAttack got a score of 435
ReproducingBot got a score of 58
BoringCopybot got a score of 214
DNAbot got a score of 19
FastMoveCloneDodge got a score of 141
HideBlockAttack got a score of 703
FastForwardClone got a score of 224
ForwardClone got a score of 193
Movebot got a score of 132
Attacker got a score of 27
QuickFreezerbot got a score of 158
TheCommonCold got a score of 1297
Parasite got a score of 71
RovingVirus got a score of 145
Blocker got a score of 250
Copycat got a score of 210
Oh, I don't have all of your bots
no, those don't look right....QuickFreeze isn't doing so well
@overactor maybe you need to write a non-standard to standard compiler, so you can use '-' in your code and then just run it through the compiler to convert them all correctly to combinations of '+' before you submit the code
If D #2 #1 Copy 23 C Copy #8 *#*C Copy 9 A Copy #A *#A+*C Copy A+1 A If A==24 #7 #24 Copy 23 C Copy C+23 C If D #C+2 #C+23 Copy C+21 C Copy C+2 C If D #C+6 #C+22 If D #C+5 #C+22 If D #C+4 #C+20 If D #C+3 #C+19 If D #C+2 #C+18 Copy E D Copy #C+7 *#*C Flag Copy C+3 A Copy #C+22 #A Copy A+1 A Copy C+21 C
This bot should turn opponents into Insidious bots
It's a terrible tactic though
And the Insidious bots are not by far as good as the original Insidious
Wait, sorry, I don't understand the parameters of this contest then. If you plant flags on another bot, what connects it to the flagger for scoring at the end?
Okay, sorry; I have only been paying marginal attention to this, because it seemed like everyone had the entire contest covered as soon as it was posted
@EricTressler yes there was a lot of sandbox activity for this one
I think that was one of the approaches, yes. The more flags you check the higher your probability of identifying an invader, but the less room you have for other code
@EricTressler we had a lot of ideas before the contest started, but the metagame has definitely evolved a lot in the last day and a half. early discussions focused on attack and block and self repair, but then we discovered freezing and self replication and other tactics
"Your changes were not checked into due to conflicting changes. These conflicting changes have been automatically resolved. Click dat button again, SUCKER!"
Would anyone here be interested in other problems like the "Find good ways to check for perfect Euler bricks"?
The problem as it was posed was bad, but someone in the comments linked to another one that was similar. I.e. the goal wasn't to solve the problem, but to exhaustively check potential solutions
the code bot challenge has me thinking of ants and state machines and trails
I remember doing an a-life sim where you wrote a state machine for an ant, which could walk around and leave trails and change behavior based on the trails it encountered
so, same concept, but with more overlapping trails and an arbitrarily complex state machine rather than the simple left/right decision in langton's ant
@Rainbolt have I ever told you about how CPAN on OSX installed the perl "HEAD" script which overwrote the standard "head" because the filesystem is case insensitive?
It took me about 4 hours to figure out that the problem was that two of the source files in the zip had a name which differed only in case, and when I extracted it on an OS X box with a case-insensitive filesystem the second overwrote the first.
@Geobits however, some pieces of software (cough*Adobe*cough) rely on the insensitivity to avoid having to consistently capitalize things. if you enable case sensitivity, many Adobe products will break because they named the file xxx.so and tried to link against XXX.so, etc.
Can anybody explain why my answer here would give results dependant on my PC? I get 28320 on one (Ubuntu,i7) and 28340 on another (Windows,i3). Same code and input on both.
I didn't think I was using anything platform specific.
There were 4696 bots with equal flags
SuperFreeze had 1245 points
FastMoveCloneDodge had 853 points
Insidious had 816 points
RovingVirus had 789 points
FastForwardClone had 781 points
FreezeTag had 760 points
ForwardClone had 478 points
QuickFreezerbot had 463 points
TheCommonCold had 439 points
HideBlockAttack had 392 points
Influenza had 368 points
BlockFreezeAttack had 310 points
BoringCopybot had 292 points
Copycat had 286 points
QuickFreeze had 218 points
FlagInjector had 204 points
Magus had 199 points
Terminals are so boring these days. They used to look like this:
Now they're just bland and dull and black-and-white. I want you to write me a program that will make my terminal all colorful again!
Description
Take this example Ruby code:
Most Linux terminals support these escape sequenc...
It has an option to write to file, so you could run it for several different values of N and see if one particular part of the code spikes as N increases, then you'll know where your O(n^2) problem is
Automatic Plagiarism
Disclaimer
I don't approve of plagiarism, but I thought it would make for an interesting challenge
Challenge
Now we've got that over with, it's time for the challenge. What you have to do is write a script that takes a piece of work (whether it's from online or from a tex...
Chess king-of-the-hilljava
General
This is a chess-KOTH with simplified rules (because chess itself is already complicated, playing it via a simple program isn't making it easier). At the moment it is limited to java (version 8), but creating a wrapper-class isn't that difficult (in case someon...
Terminals are so boring these days. They used to look like this:
Now they're just bland and dull and black-and-white. I want you to write me a program that will make my terminal all colorful again!
Description
Take this example Ruby code:
Most Linux terminals support these escape sequenc...
