Author

Topic: Rene's Puzzles (Read 120646 times)

Werbad



« Last Edit: September 07, 2009, 05:43:14 PM by Werbad »

Logged




Rene

Excellent solution, Werbad. I am surprised that you managed to fool the generator in a reliable way. Here is my solution: For Crypto III ( rogilus): fazylugI guess you are ready for the ultimate Crypto now. NEW PUZZLEIn this one you cannot manipulate the generator in any way. You just need to derive information from the generated stream. Easy, isn't it? Crypto IV: livadul


« Last Edit: September 10, 2009, 01:38:18 AM by Rene »

Logged




Werbad

I thought I had a solution, but then I noticed that the machine sometimes adds two crates instead of one. I can't seem to find any way of detecting this by looking at the output stream, is it even possible?



Logged




Rene

I thought I had a solution, but then I noticed that the machine sometimes adds two crates instead of one. I can't seem to find any way of detecting this by looking at the output stream, is it even possible?
It is possible to make a solution, though not easy. I am not sure whether it is possible to detect in some direct way whether 2 or 1 crates are added, but my solution does not depend on that.



Logged




Werbad

Almost working solution: sabekok (Takes forever if the first crate fails...) I overlooked that my crate evaluator fails to recognize 0crates and picks 1crates instead. It is rarely a problem, but it will break the solution in some cases... I might redo it using the same idea, or a different one if I figure out how you might have solved it.


« Last Edit: September 10, 2009, 07:22:34 PM by Werbad »

Logged




Rene

Almost working solution: sabekok (Takes forever if the first crate fails...) I overlooked that my crate evaluator fails to recognize 0crates and picks 1crates instead. It is rarely a problem, but it will break the solution in some cases... I might redo it using the same idea, or a different one if I figure out how you might have solved it. Good attempt, Werbad. It took me some time to understand how the mechanism on the left works, but I think I got it now. You could actually speed that up by using a copy of the first solution crate as the seed, instead of the second. You are still encouraged to solve the 0/1 problem in your solution, but if you have given up on it, then have a look at my solutions. The first one uses the same principle as you do: kixuzucFor each train of crates, it determines the minimum and the maximum that the hidden value can have. If these are equal, then it has determined the value of the hidden crate. When the hidden crate is determined for two consecutive trains, then the difference is the solution or its double. To account for the possibility of finding the double, if another solution comes out that is half the previous one, then it replaces the previous.The second one is slightly more complicated, and uses a different principle: kamicozFor each train of crates, it determines the range of values that the hidden value can. If the ranges of two consecutive trains are not overlapping, it can calculate the range that the solution can have. This range is then used to validate the correctness of an assumed solution or its double. If neither the solution nor its double fall within the range, then it is replaced with the next higher solution.Which one is faster depends on the situation. I believe the second is faster for numbers <8 and the first is faster for numbers >8. BTW, I discovered a fault in the puzzle: it is not possible to distinguish between an 8 and a 0 target if the hidden crate is 8


« Last Edit: July 14, 2010, 09:47:54 PM by Rene »

Logged




ais523

My solution to Crypto IV: fozofavThis works purely by brute force: it feeds initially deterministic, then (after tens of minutes!) random data into a replica of the crypto generator, and looks to see if they produce the same output or not. I expect it'll normally take hours or days to run; it's optimized for the case where the target's an odd number, in which case there's a decent chance it'll discover what it is some time this millenium, but in theory it should always work eventually.



Logged




Rene



« Last Edit: January 01, 2010, 04:53:13 PM by Rene »

Logged




Werbad

For Glass Floor Static ( mogafur): zavidigI invented a new way of checking the floor as the one I've used in similar puzzles takes 16*40+ ticks to check a single crate. This one takes a total of 76*16 ticks to check the value of all crates.



Logged




Rene

Brilliant, Werbad! I thought something like this should be possible, but I did not work out a practical solution for it myself. My solution still checks each crate separately: For Glass Floor Static ( mogafur): teluder


« Last Edit: July 14, 2010, 09:49:47 PM by Rene »

Logged




jf

Good puzzle. Werbad's comment about a new way of checking the floor inspired me to tackle this puzzle differently. I came up with something that resembles Werbad's solution in the opposite direction. I really like Rene's clever solution too. It's a long way since the days of Timberline, isn't it? For Glass Floor Static ( mogafur): cevumob  it's a little disorganized, but I won't spend time to optimize it.



Logged




Rene

It's a long way since the days of Timberline, isn't it?
Darn, I knew I saw it somewhere before, just did not remember where. I just reread my comments back then (see http://kevan.org/rubicon/forums/index.php/topic,62.msg1168.html#msg1168). I thought it would not be possible with 8 crates or more lol. And now we have 11. With Werbad's and your solutions, we could easily go up to the full 15 crates with only minimal extra time and space required. But then again, Timberline was more than 3 years ago... Well done, guys!



Logged




Rene

NEW PUZZLETry to catch the snake in the stream of water. Water Snake: jiduvid



Logged




Werbad

For Water Snake ( jiduvid): ranurulIt catches the first two crates who's sum matches the target crate and that's part of the same stream.



Logged






