Pages: 1 ... 3 4 [5] 6



Author

Topic: ais523's puzzles (Read 41336 times)

Rene

I was inspired by the very smart mechanism that Werbad has in his solution ( xadepip), and used it to make a solution without barrels: For My Timer, Your Timer ( gydogah): baxymel



Logged




curly



« Last Edit: March 25, 2012, 05:07:09 AM by curly »

Logged




ais523

New puzzle: The Crate Trap ( juhucuk). I haven't solved it myself yet, but I'm pretty sure it's possible; it's just an issue of getting all the moving parts to synchronize correctly.



Logged




curly

New puzzle: The Crate Trap ( juhucuk). I haven't solved it myself yet, but I'm pretty sure it's possible; it's just an issue of getting all the moving parts to synchronize correctly. Ummm......? ragicobI guess those 8's should be ?'s.


« Last Edit: May 17, 2012, 09:47:02 AM by curly »

Logged




Rene

New puzzle: The Crate Trap ( juhucuk). I haven't solved it myself yet, but I'm pretty sure it's possible; it's just an issue of getting all the moving parts to synchronize correctly. Ummm......? ragicobI guess those 8's should be ?'s. If you replace the 8's with ?'s, then the puzzle becomes theoretically impossible, if I judged it right. If you add another up/down pipe, then I think it might be theoretically possible. Wether you can build a machine for it, would still be questionable. This would still be an extremely hard puzzle. Extend the up/down pipes to 8, and it becomes more easy. BTW, I am still working on your last two puzzles, Curly. They are hard!



Logged




curly

If you replace the 8's with ?'s, then the puzzle becomes theoretically impossible, if I judged it right. If you add another up/down pipe, then I think it might be theoretically possible. Wether you can build a machine for it, would still be questionable. This would still be an extremely hard puzzle. Extend the up/down pipes to 8, and it becomes more easy.
Yes I was lying in bed last night trying to work out if ?'s could be done with just 5 pipes but I don't know how to do it with less than 8 since the read is destructive. BTW, I am still working on your last two puzzles, Curly. They are hard!
He he. I wasn't sure. I figured either that or they were too easy and nobody could be bothered :)



Logged




ais523

Err, the 8s are meant to be ?s, indeed. The Crate Trap  fixed: baxikugFive is the minimum number of crate traps required to solve it with ?s; I sat down and worked out the maths before actually setting the problem. I already have a working mechanism that communicates both numbers to both sides, but ran of space in the puzzle trying to actually construct the mechanism to put the crate in the solution. I'll keep working on solving it, in case nobody else can. (@Rene: It's pretty easy with 8. If you want to work out the algorithm, you could try working out how to do it in 7 and working it out from there. A hint: communicating the bits in the individual numbers one at a time isn't going to work; you need to communicate the bits in something else.)


« Last Edit: May 19, 2012, 07:33:49 PM by ais523 »

Logged




ais523

Whew, solved it! That took hours of effort, but it's so satisfying to make it work. And it proves that the level most definitely is possible. (I love it when a very simple level requires a very complex solution, such as in this case; this isn't the only level with this property, but it's probably the first of mine that works like that.) For baxikug: jebigexSerious spoilers on how to solve this puzzle and on my solution, don't read this unless you've solved it yourself or aren't planning to solve it: The mechanism I ended up using over the right is rather simpler than some of the ones I'd tried before. The trick turned out to be the use of halfcycle furnace draws in order to allow me to choose a "default" crate when no crate is present; that way, I could send either an F (to wait the maximum possible length of time before next sending the dozer out) or some other lettter (e.g. E for a minimum wait, 7 for a wait exactly in between those, etc.) This allowed me to do a binary search and communicate all 5 bits of the difference between the two values, in such a way that both sides knew what it was. Previously, I'd been using gates that acted like left conveyors on half the cycle, by putting Fs connected to winches underneath them, to give packers time to add the crate I wanted and the crate I didn't together; but that both used more space, and delayed the crates a crucial 2 ticks, which turned out to be just enough to make the F/0 case indistinguishable from F/1.
I also like the mechanism for communicating from right to left; because I had to use the entire range of crate values (and F in particular) for controlling the timing machinery, I was stuck with having to use ? crates for counting the number of crates that go past, as they needed to move to the right on top of a gate. Once six crates have passed (five that measure the results of the comparisons, one that is always F but is used so that the timing impact of the previous crate is taken into account), the ? crate falls off the gates with a timing that jams the timing machinery.
Over the left, the trick was to convert the numbers into timings at the rate of 4 ticks per value, rather than 2 ticks like I was trying earlier. This simplifies the communication from right to left somewhat, because it doesn't have to use awkward singlecycle timings or an alternate communication path to signal the last bit, and means I only have one case over the left rather than two. This is done using a "reservoir" of crates, that's filled with a number of crates representing the number to encode, and with one crate removed every four cycles until it runs out.
Finally, at the far left there's a loop that cycles through all the possible values over time, and from there we can work out the difference between the crates mod 16; and from that, it's a case of simple arithmetic to calculate the value of the remaining crate. (Not that simple, though, because I was running seriously short on space; top and bottom use entirely different tricks to make it work, such as reusing the unpacker at the bottom, and sending a dozer into the middle region of the program and back again at the top.)



Logged




ais523

Bleh, nobody's got it yet? Meanwhile, here's a much easier puzzle: Narrow Window: fosapom (my solution: fecoteg)



Logged




Werbad

For Narrow Window ( fosapom): sasadetI've looked a bit at The Crate Trap but I haven't been able to find a working algorithm for it yet. I do want to solve it eventually, as with many other puzzles on this site.



Logged




Rene

Whew, solved it! That took hours of effort, but it's so satisfying to make it work. And it proves that the level most definitely is possible. (I love it when a very simple level requires a very complex solution, such as in this case; this isn't the only level with this property, but it's probably the first of mine that works like that.)
Darn.... That means that I actually need to sit down and solve it . And I was pretty sure of my theoretical line of thought. Haven't looked at your spoilers yet, I will have to if I cannot work out a possible solution in theory



Logged




Rene

Bleh, nobody's got it yet? Meanwhile, here's a much easier puzzle: Narrow Window: fosapom (my solution: fecoteg) Easier indeed. For Narrow Window: dekelap



Logged




ais523

Another tough one, which hopefully I'll solve eventually (I've already worked out the algorithm, the problem is fitting it onto the screen). Summing Station: sexekar



Logged




Rene

What an amazing, inventive and original puzzle! I honestly was completely convinced that it was impossible. And not impossible in the sense that there is not enough space to work with. No, impossible in the sense that the fundamental laws of the universe need to change. And then I thought about it again, and..... no, totally impossible. And then I thought about it again... For The Crate Trap ( baxikug): raharymDouble thumbs up, Ais! EDIT: I have now studied Ais's solution, and although we use the same solution principle, we actually use different algorithms. We both convert the crate values into time, in order to detect which of the crates is larger: whomever drops a barrel in the pipe first, is the smaller. Both sides can reliably determine that. Then its a binary search to find the difference between the values. But Ais also uses the timing for determining the difference eventually: how much time it took determines the difference. But he needs to double the difference between them at the start. In my solution, I keep track of which comparisons were larger and which smaller, and get the difference value from that.


« Last Edit: July 02, 2012, 02:21:01 PM by Rene »

Logged




ais523

Wow, you used so much less space than I did. I like what you did with the winches over the right hand side; I think that's the main thing responsible for the savings, as it means you only have one dozer to have to deal with on top and bottom rather than two. It turns out I still had a couple of columns spare in my solution though – I can see where to save them – but it's nothing on the many tens you have. Incidentally, with my solution, I was originally trying to make top and bottom exact mirrors in order to get the timing right, but it didn't quite work out.



Logged





Pages: 1 ... 3 4 [5] 6



