SameGame

Rate this contribution

+45

Make the contribution public (lvl 20+)

Comments

45 Comments

Illedan Unnamed CodinGamer

over 5 years ago

https://www.codingame.com/... The dropped tiles on the last round suddenly change colour for a brief moment.

aCat Unnamed CodinGamer

over 5 years ago

Yeah, I've noticed this before but given I'm so happy it already worked after multiple iterations of much weirder stuff happening I didn't found a fix for this one.

jrke Unnamed CodinGamer

over 5 years ago

Though its approved but i found a bug fix it. The game has crashed. Please contact the author and enclose the following error: java.lang.ArrayIndexOutOfBoundsException: Index 1 out of bounds for length 1 at com.codingame.game.engine.GameState.validate(GameState.java:59) at com.codingame.game.Referee.gameTurn(Referee.java:69) at com.codingame.gameengine.core.GameManager.start(GameManager.java:122) at com.codingame.gameengine.core.RefereeMain.start(RefereeMain.java:67) at com.codingame.gameengine.core.RefereeMain.main(RefereeMain.java:50) its happening when user is not giving space between output like giving "00" instead of "0 0"

aCat Unnamed CodinGamer

over 5 years ago

thanks for reporting, I will catch this one

hoggi Unnamed CodinGamer

over 5 years ago

✅ Approved the contribution. Comment: Original et bien réalisé. Un certain challenge pour performer.

darkhorse64 Unnamed CodinGamer

over 5 years ago

Because the validators are standard test cases, there are known records for each of them or for the sum of them. Wouldn't it be worth to add these numbers in the statement ?

aCat Unnamed CodinGamer

over 5 years ago

You have some 'bestscore' in the human-playing link, I'm pretty sure someone mentioned somewhere that they were obtained programmatically... On the other hand, it is actually hard to say, as I know some unpublished papers that already beat what is currently 'best score'. So just search the literature at the time of writing.

Flodnes Unnamed CodinGamer

over 5 years ago

Cool, looking forward to this one!

[CG]Maxime Unnamed CodinGamer

over 5 years ago

I was a bit confused by this: "First 15 lines: rows of the board, top (row 14) to bottom (row0). " because it forced me to reverse the rows of my grid. But it's not a big deal. (I'm also a fan of (row, column) instead of (x, y) but it's easy to adapt :).) Apart from that, it seems to be working and I'm glad we'll have this puzzle on CodinGame, it's a must have.

aCat Unnamed CodinGamer

over 5 years ago

I know - but I really think converging to 0, 0 is natural here, so I'm not about to change it until really forced to ;p. Good point with (x,y) to (row,column), silly cat I didn't think of this. I've changed it of course - It's much more clear that way - and I have to remember the case for the future. And I'm really glad you like it :-).

[CG]Maxime Unnamed CodinGamer

over 5 years ago

That's actually on this game that I learned SP-MCTS (and had to learn MCTS)! I'll give it a shot when I'll have some time. Thanks for the contribution!

Thyl Unnamed CodinGamer

over 5 years ago

You think make a contest ?

aCat Unnamed CodinGamer

over 5 years ago

No, I don't think so. There will be exercise list for my course based on this puzzle, with some "try this and that algorithm, report how they behave, bonus for obtaining score >= something". I may publish it, but that's probably all ;-).

eulerscheZahl Unnamed CodinGamer

over 5 years ago

I like the contribution and intend to approve it. I'd just like to see different testcases: you have 20 validators and then the same again with different colors. For my program they are still the same. I think just random would do better here. The first few testcases are a bit boring and give most of the points. But as testcases I guess they are fine. They aren't validators after all. Y starting at the bottom is a bit unconventional for CodinGame standards but no big deal ;) "Indices must start at 0, not 1. The origin is always the top left pixel/tile." https://www.codingame.com/...

aCat Unnamed CodinGamer

over 5 years ago

Actually I wanted to make the same testcases not repeated 2 times but at least 10 but cannot make so many validators. Tascases are standard research benchmark, and allow you to compare with other positions in literature. About Y at the bottom... it seems natural as you converge to 0,0 when all the tiles are consistently disappearing.

aCat Unnamed CodinGamer

over 5 years ago

OK - I updated the puzzle. Important changes: - now you can make your action turn-by-turn (color -1 for empty tile) - no more warnings - selecting illegal action causes hard error.

Skrakout Unnamed CodinGamer

over 5 years ago

Pretty cool game !

Thyl Unnamed CodinGamer

over 5 years ago

It's very cool !

aCat Unnamed CodinGamer

over 5 years ago

Thank you for your approval so far - but I agree that the CG community will gain if I simplify the problem and allow solutions not requiring writing the engine. I should make an update in the next few days.

Astrobytes Unnamed CodinGamer

over 5 years ago

I recommend switching to WIP, lest the contribution be approved as it is.

Zylo Unnamed CodinGamer

over 5 years ago

✅ Approved the contribution. Comment: Nice simple problem: various number of possible complex solutions.

VilBoub Unnamed CodinGamer

over 5 years ago

Why not have created an infinite loop to make the game easier for beginners?

aCat Unnamed CodinGamer

over 5 years ago

Mostly - to be consistent with the standard benchmark scenarios in publications. I was thinking about that, but this creates this weird choice of time-spending. Either there will be insignificant time per turn to just make non-optimal strategies 'work'. Or this time will be significant and you have to tune your algorithm to handle planning and searching.

jrke Unnamed CodinGamer

over 5 years ago

Oups An error occurred (#510): "An error has occurred during compilation/execution.". Is this error is with me only?

aCat Unnamed CodinGamer

over 5 years ago

When I'm updating, from time to time the package I send causes this error. No idea why - the build is nearly the same. Thanks for pointing this out. It should be working again.

igolus Unnamed CodinGamer

over 5 years ago

Nice game. +1 Would it be possible to see in the intro how the column moves ? to illustrate : "Then, for every empty column, all columns that are on the right are moved maximally to the left, so that the row 0 has no gaps. "

aCat Unnamed CodinGamer

over 5 years ago

Ouh I was not aware that intro stops after some frames. Nide idea I will try to find something better, thanks.

Scarfield Unnamed CodinGamer

over 5 years ago

Havent tried much yet, but great idea for a game here :) I think the player should either "lose" or at least stop the sequence of moves, when doing an illegal move, instead of just a warning. And 20sec, even for 1 turn only, seems to much? Thats 16 minutes just for the testcases, if you use all 20sec for 50 cases. Any way, take my upvote :)

DomiKo Unnamed CodinGamer

over 5 years ago

I agree. Illegal move => instant lose, its a lot easier to debug

radekmie Unnamed CodinGamer

over 5 years ago

✅ Approved the contribution. Comment: I find it interesting and really well made. The visualization is great and looks like an actual game I played a few years back.