Let's all play Spacechem

Spacechem fever has hit the forums once again, so let's engage in some friendly competition. Here's how we'll do this:

1. Sign up as puzzlemaster. There's no need to sign up to participate, but if you want to design a puzzle, then say so in-thread. Don't... don't make me the only puzzlemaster forever ;_;
2. Every two weeks, a new puzzle. The puzzlemaster describes the parameters of the puzzle and mention whether we're competing for fewest cycles or fewest symbols. (In either case, the other one is the tiebreaker.)
3. Submit your solution. Once a puzzle has been posted, you have 10 days to submit an "official" answer to the puzzlemaster. Your submission should include a picture of your solution as well as the number of cycles and symbols it took. You can resubmit if you come up with a better solution before time's up.
4. The puzzlemaster collects the results. Ideally, this should happen before the next puzzle goes up, but we're all busy. Minimally, the puzzlemaster will show the best solution, but it's also nice if any particularly interesting solutions get pointed out. If the puzzlemaster feels especially diligent, he might show a video of the various solutions.

Some guidelines for puzzlemasters -- these are not rules, but they are good ideas:

• Keep puzzles simple. Ideally, your puzzle should not be hard to solve. Even the very first puzzle in Spacechem offers some interesting optimization problems, and a low barrier to entry ensures that players of all skill levels can participate. Single-reactor puzzles are preferred, too, not least because multireactor puzzles are much more difficult to screenshot.
• Don't use campaign levels. SolutionNet has optimal solutions published for all the story missions. (However, those solutions are optimal based on the symbols the game allows you to use at that point in the game, so if you think the world 1 puzzles are enhanced by the availability of sensors and flip-flops, then go for it.)
• Limit the gimmicks. You can create a reactor with fusion, fission, and tunnel, but again... keep it simple. If you're the first one to use any nonstandard reactor features (anything more than bonders), then include a brief explanation of the mechanic in your puzzle description.
• Come up with a good title. This is make-or-break, really.

And some rules for entrants. The short version: I should be able to copy your puzzle and get it to work.

• No relying on bonding order. If your solution requires specific bonders to be in specific spots, then it's disqualified. Feel free to show it in-thread for us to admire, but for the purposes of the competition, we are not relying on a feature that's hard to reproduce.
• No relying on molecule order. Inputs always produce their molecules in a certain order, and it's possible to rely on that order to produce an optimal solution. Your solution should work even if that order gets shuffled around.

I'll get things started, but please sign up for puzzlemaster for future rounds.

Puzzlemaster Queue

1. Mogri [Air Filter - Results]
2. MrBlarney [Antifreeze - Results]
3. Ted [Oh Jerry! Oh... You! - Results]
4. DFalcon [Desalination - Results]
5. spineshark [Nucleosynthesis - Results]
6. Mogri [Atomic Arithmetic - Results]
7. MrBlarney [3-Hexene - Results]
8. Ted [Smelly Filled Donuts - Results]
9. DFalcon [Limited Warranty - Results]

#1: Air Filter

Take carbon monoxide (CO) and turn it into oxygen (O2) and carbon (C), saving valuable lives and producing diamonds. Or coal? It's not really clear.



You can import this into ResearchNet with the following code:
Note: ResearchNet officially unlocks after you finish world 6. You can still access it before then; the game will warn you that it contains advanced techniques that you haven't seen yet. That's true: you haven't been introduced to the handy flip-flop symbol.

It's clearly not required to complete this puzzle, but you might find it handy. The flip-flop does nothing the first time it's executed, but on the second time, it sends the waldo in a different direction (then deactivates again). In other words, every other pass, it will take you down a different path.

And here's a basic solution to start things off!



EDIT: Optimizing for cycles or symbols?

And here's my less efficient solution, though it doesn't use a FlipFlop



I like this idea!

Here's another one, because this is actually kind of fun!

Unless specified, assume cycles.

I have a 314-cycle solution I'll post in a bit. I think it could be improved, but I haven't had a lot of time to look at it.

Here's my first attempt.


EDIT: Argh, Dropbox formatting. Gimme a minute, I'll put it on Photobucket.

EDIT2: I guess I'll also sign up to make a puzzle.

EDIT3: Apologies for the perhaps-too-prideful tone of the originally-edited post.

Okay, SpaceChem is enough to get me to unlurk.
There's so much content in this game - I ought to go back and work on the ResearchNet puzzles sometime.

It's solutions like this that make me wonder what in the world I was thinking when I designed mine.

Do you mind adding your cycles/symbols to the spoiler?

I love this idea! I'm so excited to give it a shot later tonight! (Now back to somehow focusing on real coding until then.)

I'm a lot less proud of my solution given the recent posts, but it's at least a unique take on the problem.

Oops, done.

Here's my first attempt, but I've got another approach to try.



EDIT: And here's my second, faster, attempt. I'm sure I was working on it in my dreams last night because I jumped out of bed with a fully-baked solution in my head and just had to get it working first thing.


EDIT2: I managed to chop another cycle:

Attached Thumbnails

     

I'll volunteer for puzzlemeister. I've got a few ideas.

You've got less than 24 hours to submit your final solutions.

oops, didn't have time for this one! i'll try the next!

Challenge 1 complete!

We had six entrants, and half of them submitted very similar solutions.

6th: SpoonyBardOL (437/37)
5th: aturtledoesbite (307/29)
4th: Mogri (305/47)
3rd: Ted (148/25)
2nd: MrBlarney (130/26)
1st: DFalcon (130/25)

Note the symbol count on our last three entrants. DFalcon's is extremely impressive -- it may very well be a cycle-optimal solution. Take a look:



That's a very quick video, so here's an explanation of what's happening:

1. On the first cycle, red inputs as blue grabs. The order is important: red always activates first, so if you switched them, the grab would fail.
2. Cycle 2 has blue input so that red can grab.
3. The two molecules rotate, barely avoiding each other.
4. On cycle 8, red and blue simultaneously debond blue's carbon.
5. Red outputs the abandoned carbon on cycle 9 as blue drops the oxygen in the topmost bond square.
6. Red drops its molecule just below blue's dropped atom on cycle 10 and blue debonds.
7. On cycle 11, red debonds as blue outputs. Again, waldo priority is important here: switch them, and this wouldn't work.
8. On cycle 12, red and blue bond simultaneously, so the second O2 is ready to output.
9. On cycle 13, blue outputs as red reaches its starting square. Notice that blue's start is repositioned to make this work.
10. The reaction loops from there.

Each loop outputs two carbon atoms and one oxygen molecule. This is an extremely compact, efficient solution -- DFalcon absolutely deserves the win this round.

Of the two other 25-symbol solutions, MrBlarney's original solution circles in the opposite direction, assembling oxygen before dropping off the carbon. This is a nice optimization, since the oxygen is the bottleneck, and outputting it first shaves a few cycles. Ted's solution is very similar to DFalcon's, but his loop ends up being wider, costing him 18 cycles. MrBlarney's final solution is very similar to DFalcon's structurally, but he put his red start outside the loop, requiring him to add an extra arrow, which cost him first place.

The lower-ranking solutions ended up with more meandering solutions. The better solutions did all of their bonding/debonding in the output zones, which is a lesson to the rest of us.

MrBlarney is our next puzzlemaster.

I tip my hat to you, DFalcon! I just couldn't figure out how to shrink my loop.

The lesson I learned from this is to first try drawing my loops as small as possible and then expanding if needed, instead of starting big and trying to shrink.

Mogri, thanks for hosting the first round!

Whoa, whoa! I submitted a 130/26 to you as a private message while the thread was locked for a few hours - that should have counted, right? I mean, I would have come in second behind DFalcon due to symbol tiebreakers, but still... I didn't realize I had to post my solution in-thread for it to count.

In any case, I'll post my puzzle at midnight PDT, approximately two hours from now. Hopefully it'll be an appropriate step up - look forward to it!

#2: Antifreeze

Ethylene glycol is a compound that is used as an antifreeze agent in automobiles. Due to its moderate toxicity, it has been as of recent started to be replaced with other chemicals. Ethylene glycol is also a precursor to the plastic PET. Here, you'll synthesize ethylene glycol through a reaction involving ethylene and water, with a hydrogen byproduct.


You can import this into ResearchNet with the following code:
Remember: The rules state that you may not exploit bonder order to solve this puzzle. Stated differently, if I can move the bonders around to break your solution, it doesn't count.

Solution Criteria: Least cycles, tiebreaker least symbols.

Solution Deadline: Solutions are due to me via PM by Friday, May 29, 12:00am PDT / 3:00am EDT / 7:00am UTC. Since I've already had time to test and look over the puzzle and will need to verify solutions as they come in, I won't be counting myself in the rankings.

Oh, sorry! I assumed it was the same solution. I should have checked more carefully.

Thanks for the puzzle! I'll sign up for puzzlemaster too.

There's a lot I didn't appreciate until I started going through the game again for more optimal solutions. In my case I had tended to leave binders in the middle so I always had room to move around them, go through them at least twice if necessary, etc.

I've updated the OP as well as the first puzzle results. You might be interested to compare MrBlarney's final solution with DFalcon's. MrBlarney's red starts one square behind DFalcon's. Because blue does the final output, this doesn't cost him a cycle, but it does cost him a symbol to get the waldo running the right way.

It's a few days in and I've received a couple of solutions! If I haven't replied to your PM, it means that your solution was recorded as valid. Everyone else, use this weekend to try and come up with solutions, or improve the ones you've already got! You're free to submit multiple times if you find improvements.

About two more days left to get your submissions in. I've received a couple more solutions; check your notifications if I sent you a message and your solution needs to be corrected. I suggest sending me something tomorrow so that there is enough time to validate your solution. Should be exciting!

I shaved 180 cycles off my original solution last night, and I think I can optimize it a little more. I'm glad I came back to this; my original solution was terrible.

Looks like time is up on submissions. Won't be able to get all that I want to do with the solution summary until Sunday, but I hope you look forward to them.

Challenge 2 Results



Apologies for the quality of presentation, it's the first video I've stitched together like this. Created the video first, then did editing and post-commentary. I'll work on my pacing a bit better next time so that I can talk more about the details of the solutions - if that's what people want.



Ted is the next puzzlemaster!

Wow! I thought I had a pretty good solution here -- my initial solution was on par with Spoony's, and I managed to reduce it by 1/3 -- and then you two come along and blow mine out of the water. Very impressive.

I also love MrBlarney's comically loud keyboard.

Liked the commentary, I appreciated seeing an improvement.

Looking at the solutions on Photobucket brings up options for framed or canvas prints. Even though I'd much rather see a solution in motion, it's oddly attractive.

Isn't that the standard Spacechem clicky sound effect?