| « home | Browsing all comments (spoilers: on | off) (info: on | off) | next 25 |
 |
<noname> (SPOILER) Added speedrun: 59 moves (old: 69). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 65 moves (old: 90). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 195 moves (old: 197). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 92 moves (old: 94). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 146 moves (old: 159). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 158 moves (old: 162). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 55 moves (old: 60). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 98 moves (old: 106). Optimal. |
 |
<noname> (SPOILER) Added speedrun: 697 moves (old: 709). By assuming that each red block is to be pushed at most once, I formulated this as an instance of the assymmetric traveling salesman problem, and used PySCIPOpt [1] to solve it optimally (under the assumption). In fact, I basically just imported the atsp.py file from the Examples and work from there. Note that PySCIPOpt is based on the SCIP Optimization Suite [2]. References [1] S. J. Maher, M. Miltenberger, J. P. Pedroso, D. Rehfeldt, R. Schwarz, and F. Serrano, 'PySCIPOpt: Mathematical Programming in Python with the SCIP Optimization Suite', in Mathematical Software - ICMS 2016, 2016, vol. 9725, pp. 301-307. [2] S. Bolusani et al, 'The SCIP Optimization Suite 9.0', Optimization Online, Feb. 2024. |
 |
<noname> (SPOILER) Added speedrun: 201 moves (old: 240). Optimal. |
 |
<noname> (SPOILER) Added solution "Optimal bot killed": Maybe interesting to future players. <noname> (SPOILER) Added speedrun: 74 moves (old: 86). Optimal. Turns out not too bad for my solver to solve unrestricted. <noname> (SPOILER) Added solution "Optimal without bot": Perhaps this would still be interesting for future players... <noname> (SPOILER) Added speedrun: 86 moves (old: 103). This should be optimal among those solutions which kill the hugbot. I was able to keep the number of positions small by the above assumption, since this implies that all the panels controling the path to exit are covered by some block at the end, which means that blocks can't be, say, at an edge or form a 4x4 pattern with the wall (with at least one of the block being grey). I don't know if it's possible to solve without all blocks on a panel at the end; the total number of possible configurations without restriction seems too large for my solver to handle in reasonable time, so I'm not going to have my solver attempt this until I can come up with some new idea for this part. |
 |
<mark> (SPOILER) Added speedrun: 487 moves (old: 534). I always wanted to know if this way was possible. |
 |
<noname> (SPOILER) After seeing the original solution, I agree with Mark that the concept of the original version is more or less flawed. Without that realization, I ended up solving this with some 400+ moves of bot manipulation after taking the heart framer... |
 |
<mark> (SPOILER) Added solution "Untitled": the 604 way tidied up a bit. liked your solves espcially the dull way. Ive got a new version of this coming soon that gets rid of every known solution so far. <noname> (SPOILER) Added speedrun: 522 moves (old: 571). And this is the clever way. <noname> (SPOILER) Added solution "The Old Dull Way": This is the old way to solve this level I mentioned. <noname> (SPOILER) It turns out that I already tried the second exit when "Magic Trick bot kiss" came out, and obtained solutions much faster than what I did a few days ago. The solution done in the old dull way is slower than the solution done in the clever way, though. I have video-captured the solutions and then input them by hand into this level, and I'm going to upload both solutions for you to see. <mark> (SPOILER) Nice one for solveing Ive also been playing this for hours got another way and a speedrun of 604 so far. Wont post it yet to see if you can clean up your solution into a speedrun. Been in ver4 to see original solution and wow thats where all my solves was.I got one bookmark where Ive put a grey block on the imaginary button. Obvious really because i always play the original loads before making mod.Its been that long Ive forgot so much. Thanks for question replys. <noname> (SPOILER) Without killing the hugbot near the beginning, of course. <noname> (SPOILER) Took me eight or nine hours to solve, but that's because I solved it the old way, like in Magic Trick (v4)... |
 |
<noname> (SPOILER) By the way, I agree that this is way easier than the original. <noname> For the speedrun desappears thing, I have reported this in the feedback thread back in 2016. As a workaround, if, after downloading speedruns, you immediately solve the level, e.g. by playing one of the speedruns downloaded, or playing your own solution, and press enter at the end, then the downloaded speedruns will no longer disappear. According to the source repository, this bug has been fixed, but a new version incorporating the fix is yet to be released. As for the enter thing, I definitely can enter a level by pressing enter, so I'm not sure what happened. I did notice that with my laptop, the arrow key in the numpad does not work with Escape: I have to use the main arrow key instead. So perhaps you tried to use the enter key in the numpad? |
| « home | Browsing all comments (spoilers: on | off) (info: on | off) | next 25 |