I managed to solve all the puzzles, and I've been putting together a spoilery guide to the game, but I thought you all might want to try it out before I post any such thing. Also, I dare you to beat my high power score on 3-4! :twisted:
If interests you (and if you're still posting here), 1.5 years later, I am now the 4th person to get 81MJ on 3-4 (prior to that, I was rank 4 (98%) with 27MJ).
I knew about the "loophole" (my solution to 1-4 above also takes advantage of it, I mean the fact that
you don't need continuos production per cycle, only 3 cycles in a row
), so I knew what kind of approach was necessary.
It wasn't really hard, getting 27Mj with 7 tiles was harder, which I think is a prerequisite, even then space was at premium. I didn't even read your spoilers, though kudos for getting there first, because if I didn't see 81 was possible, I would've concluded 27 was the theoretical max and moved on.
I reran some of the others on level 3, someone managed to beat 36 with 6 tiles on 3-2 (current best is 243MJ/20T), I'm now 3rd for efficiency, so there's 2 of them. 3-3 is in more need of revisiting for me anyway...
I don't play this game often, usually during holidays when I'm away visiting family. I just started 4-1, I got 1 solution, rather lousy one, I didn't use the new tiles, I'm not even sure what's the use pattern for them...
Thanks for the bump - I wasn't an Android user back then, but now I am. I shall install this tonight - it'll keep me busy during Xmas #2 and Xmas #3 tomorrow.
Edit: Level 1-1 to 1-4, took no time at all (and didn't read the hints above) to get the max score. 2-1, now that's where the challenge started, and where my phone battery died.
If interests you (and if you're still posting here), 1.5 years later, I am now the 4th person to get 81MJ on 3-4 (prior to that, I was rank 4 (98%) with 27MJ).
Edit: Level 1-1 to 1-4, took no time at all (and didn't read the hints above) to get the max score. 2-1, now that's where the challenge started, and where my phone battery died.
I have everything on levels 1 and 2 at 100%. I think the challenge starts at 3-1 when the new 'alternating' tile is introduced, that one can really complicate flows where I usually have to make something rough and see what happens, it's difficult to see what will happen at a glance...
I have 3-1 and 3-4 at 100%, 3-2 is 99% for efficiency, others on level 3 need some improvement.
Yeah, the game is brutal on battery, 3-4 hours and I'll be looking for a charger.
Another bump, and the last one from me.
It seems I only play this game around holidays, this time I finished it. I was aiming for top 10 on everything, and I got that in all but 2 cases. I will revisit some and try improving them every now and then, perhaps try beating some high scores.
So, my semi-spoilery advanced guide and comments+my result, only levels 4 and up.
First a general note - some of the "most power" I didn't bother doing because they require what Ken called "delay loops". Basically you cannot have long-term continuous production. Max continuous production is usually easy to calculate, so my goal was hitting that. Efficiency tends to be the more interesting one.
E.g. in 4.1, one input has 7R + 1G, the other 1R+1G, so in total 8R+2G.
Max theoretical continuous production is converting 8R->7Y->6G (i.e. lining up particles so 1R can convert all other 7R into 7Y and so on). With 2G from input you now have 8G->7B->6P. 6 purples is 6x81 = 486. That's the max you can get without "delay loops". It also assumes there's 7 tiles in front of the laser, the calculation can be more complicated when you don't have enough space there.
4.1
Rank 1/1, power:486, efficiency: 243/6 tiles
I used this level as an example to calculate theoretical max, and it also pretty much tells you the approach and a general pattern: lining up large number of particles to get the most "bang-for-the-buck" since the particle entering the laser is destroyed.
It's too busy to add delay loops, that's also the top result.
Max eficiency is 3P with 6 tiles, using one input just to control a switch to send to output or into laser, I will say no more
4.2.
Rank 5/1, power 405, efficiency: 243/11 tiles
Once again, max continuous production is 5P, so that's 405. Line up all 6R and hit with R from the other output and then Y with 1st laser to convert to 6G, then 2nd laser to get 6B. Use the last laser to get 6B->5P and that's 405. 5 people have 6P, this requires basically having cycles of 4P+4P+4P+6P+6P+6P or some such, and hitting back button at the right time.
Efficiency is 3P with 11 tiles, that's still the best although I probably had that a year ago (I dropped from 4 to 5 in power when I reran it).
4.3
Rank 6/1, power 648, efficiency: 432/7 tiles
This one is not theoretical max, and one of the more complicated designs, lots of switches and tiles used. The problem is the limited number of tiles in front of the laser and lots of traffic due to too much input. Top result is 972.
Efficiency is 5P+1B with 7 tiles. Line up particles and use switch to control whether particles go into the laser or to output. Only one laser used.
4.4
Rank 1/9, power 405, efficiency: 405/14 tiles
This is very tough to line up everything efficiently. Max is 405 without delay loops (no one has done more). This is one on the list to revisit for efficiency, though I remember it was driving me crazy, I have no idea how this can be done in 11 tiles which is the best score, I'm using 14.
The first one where I didn't make top 10. Power is OK, >12 requires delay loops, only 6 have had the patience to do it.
For efficiency, those 6 people are ahead, but there's 9 more that are doing it with less than 13 tiles. I spent a lot of time on getting it to 13, and I can't see how I can get below that. 1 or 2 people getting some crazy solution wouldn't be surprising, but with 9 people, it makes me think I'm missing something because even 13 was very tough. Now that I've picked up a few more patterns, I will revisit this at some point.
5.2
Rank 5/1, power 15 (top 30), efficiency 15/6 tiles
30 requires delay loops, lots of space available, I may go for it at some point.
This is one of my favourite levels. For a while I was convinced there was some cheating going on. My reasoning was that having electrons coming in 3 levels, you need at least 3 tiles just to send them to the output, and that won't work, so 4 is theoretical min (and I can't do it without 6). Since top efficiency was 6 tiles, you're left with 2 to branch the input into 3 levels - clearly impossible. I was even thinking about writing a program to search all cases with 6 tiles, with some restrictions it may be doable. The problem with my logic is that only in 5.3. did I realize that electrons don't need to travel in the same direction, they just need to be in those spots. The description for the resonator is a bit misleading. (Ken said that as well, but I didn't want to read his post before doing the level...)
There's one other (mathematically) cool trick to get it working with only 6 tiles, but it was very satisfying getting it.
5.3
Rank 1/1, power 54, eff. 54/12 tiles
This is where it occured to me that maybe we can send particles through resonator in different directions. It's needed to get it with 12 tiles. It's not trivial, but I got this fairly quickly after the "revelation". I feel like <12 is possible.
And the second one where I'm out of top 10. Power requirement is very low. I was also switching between using laser only and trying to involve resonator. Getting one purple was easy, but then I got stuck in high 90s. Eventually I managed to find a way to get 2P using only the laser and using 3 electrons from one input and 3 from the other. One of the input goes straight to output, so the rest 3 particles just go to output, and that's 165. For efficiency, I just auto-destroy the unused 3 from the other input since sending them to get 168 involves efficiency hit even with 1 additional tile. My best power solution involves converting one of those reds into yellow, so that's 170. I'm not a fan of resonators, but I will get back to this level, I can get better power, I was only going for efficiency.
6.1
Rank 1/4, power 60, eff. 60/11 tiles
With 5.2, one of my favorite levels. Simply meeting the power requirement is rank 1, so that's not an issue.
At first, I was following the same approach as Ken - there are 5 particles in a cycle, in each cycle, take 4 to send through resonator, and use halting particle to gather 4 extra particles from each cycle to send them together. It's not hard to get power requirement this way. I was aware of another approach, but I kept switching between trying to make second approach work and optimizing this (you can play with which particle in the cycle you select as the extra). The best I got in this way was 21 tiles, and top efficiency is 7.
Ultimately I got the second approach working and my first (and only) solution uses 11 tiles, so almost half the previous best, and rank 4. So what's the second approach? It's best described as treating the input as a continuous stream of particles and having a setup such that when you get a group of 4, you let them go into the resonator. Basically, it doesn't matter if particles come in groups of 3, or 6, or 5 in a different pattern, it would still work (well, maybe 7 particles wouldn't...). If I mark the cycles with A,B,C, the sets that go together are: (A1,A2,A3,A4), (A5,B1,B2,B3), (B4,B5,C1,C2), (C3,C4,C5,D1), etc, it should be clear what I mean by continuous stream.
I need to make a first left turn and then up to create more "playing space" in front of the resonator. Without this, I managed to get a setup with 9 tiles that works for 4 cycles, but every 5th is wrong, so I get 52. Technically, 52/9 is better efficiency than 60/11, but 60 is the requirement. I still have no idea how someone found a way to do this with only 7 tiles. Maybe someone wrote a program to simulate this and do exhuastive search. You can put some obvious limitations - first tile has to be in the column with the input, and restricting it to the 2 rows with the resonator probably, it should cut down the search space to something manageable. It must be some black magic...
6.2
Rank 1/6, power 108, eff 108/22 tiles
You have to do Ken's "priming the pump", without "self-feeding" the resonator you won't be able to meet power requirement. My first solution was 29 and I spent time trying different approaches ending up with 22, best efficiency is 19, so it's within reach.
6.3
Rank 9/9, power 486, eff 396/22 tiles (global best: 972 and 396/15 tiles)
Another "open-ended" level in the sense that power requirement is really low (50, top result 972). Resonator has a weird shape, I feel it's too much effort setting up particles to make use of it. Probably required for top power score, in my attempts it was really just getting in the way since it takes away spots for tiles. I split the 16R from input into 2 groups of 8, each one producing 2P+1B+1G, so x2 that's 396. My best is 22 tiles vs top 15, still room for improvement, but getting from mid-high 20s to 22 was already difficult.
For best power, I'm using those 1B+1G from one group to get 2P out of the other two, so my best power is 6P = 486. Still only using laser and barely making it into top 10 in both categories.
6.4.
Rank 1/4, power 648, eff 648/14 tiles (best is 12 tiles)
Every R has to be converted to P, so meeting the requirement is automatic rank 1 since you can't do better than that, no delay loops can help.
A bit of a letdown level since it's not very hard. Clearly you need to "prime the pump". I was an idiot at first, because I wanted to get the first Y+G+B to come horizontally into the resonator, and even after realizing that I actually need 1Y+2G+3B, I was just married to that approach. It's not much trouble making that work, but efficiency was way off the mark, I got to like 20 tiles, perhaps not so bad, but it was mid-20s rank vs my top 10 target.
Eventually I figured out it's better to produce 1Y+1G+1B and line them up in front of the stream od reds. My best is 14, there's 3 people who got it with 12 (some of them maybe 13...), I'm pretty close here, I got a solution with 12 that produces Y+G+B and places them correctly, but reds to produce YGB have some extras and one of them hits the main stream or reds, thus breaking the cycle and killing everything.
One more update, or rather a brag - I am teh master of this game.
Besides clearing everything at rank 1, there are 5-6 cases where I beat the previous top score.
I doubt anyone here plays anymore, so I won't cover all levels again, just a select few. Though there are people still playing it - in the last 4-5 months I fell down 20-30% in rankings for some levels, and there are no guides/threads anywhere, other than Mersenne forum which was also started by Ken_g6 with even less info. So perhaps it'll be useful to people outside of AT.
Challenges start at level 3 if you want high rank, 3-2 was annoying because there are lots of possible ways of solving it, but you don't know beforehand how efficient you can make it. Someone got 243/23 tiles, I got to rank 2 with 27, then improved to 24. I only got back to it when I cleared levels 4 and 5 and decided to go for rank 1 everywhere and that was a sore spot. It's also annoying without switches from level 4. 4-4 also took a long time to get top efficiency.
Doing delay loops can also be pretty fun. Having a dedicated electron which just loops and serves as a timer is another useful concept.
6-1
I said earlier there must be some black magic to get this with 7 tiles. It was driving me crazy, at some point I started (semi)randomly placing tiles and let it go to see what patterns come out. One of them solved it, but the crazy thing is that the first 3 electrons were wrong, and after that it started working. I don't even know why, I left it for like 15 cycles, I can see why every particular cycle works, but every cycle is a bit different and I don't see why this works overall. There's probably some long cycle before you get one that is messed up again, but I don't know how someone can logically come up with it. I still think someone found it via computer program, or maybe just stumbled on it accidentally like me. It truly is black magic.
It also makes me think if there are some solutions with a long cycle of say 1000 and you get 3 working in a row only at like 647-649. No way a human could find it...
6-3
Top was 396 with 15, I was at 22. I found a good general pattern and got it with only 14, wohoo. Then I went for top power and my first attempt was 540/19 which was better efficiency than 396/14. And I wasn't even trying to be efficient. Though I couldn't improve it much, it's down to 540/17, that's the current top result.
I matched rank 1 in power by using like >80% tiles, pretty nasty. Still haven't found a way to use resonator, the shape is just not conducive to a good self-feeding design, and without it, I doubt you can do any better than what I have.
5-4
I was struggling with getting 2 purples by using just lasers, Ken said the same thing. There's actually a much better approach with resonator easily getting 3P+1B, and even 4P is not much harder. The trick is to use self-feeding pattern and notice that if you run both streams into the resonator in opposite directions, all you need to do is put a single yellow in front of each stream (and align the streams), and you will get a steady stream of greens on both end. And you only need this done once, so just select 4R to create 2Y and start it up (you could get away with 3, but it's initial setup, you don't care how many it takes). One stream of greens you just send to the output, theoretically you could split the stream and then reassemble to make a gap and allow both streams to go towards laser, but in practice probably not possible. The other stream of greens goes to the laser. Getting 3-4P out of 6G is a lot easier than getting 6R->2P with just laser. Even with the one-time resonator setup it's still much easier, so don't bother using just laser.
If you want bare minimum - just 2 streams of greens is enough (12x9 = 108), and this is really child play, barely tougher than some 2-x levels.
Getting top score requires some arcane one-time setups. Top efficiency is 540/39 - when I set it, it was simultaneously top score and top efficiency
I further improved top score to 558, though at the cost of like 20+ tiles, rather tough to work it out though... I said there are 5-6 cases where I beat the previous best score - I think this is the only level where I got both power and efficiency.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
Facebook
Twitter