|
|
|
Hoola is back for more |
Sat 05 Oct 2019 10:59 |
|
<noname> (SPOILER) While I first solved this level by hand, it is just too tempting to use a computer program to solve this: only 4 movable objects! - which means the total state space is small enough for a computer to brute force. Since this level is quite young, I'll hold back the computer-found optimal solution until this is older, so that other players may have a chance to find it (perhaps by hand?) |
|
|
Conflicted |
Sat 05 Oct 2019 10:43 |
|
<noname> (SPOILER) Added speedrun: 123 moves (old: 299).
Did not use all the available blocks. |
|
|
Rotary Club Again |
Sat 05 Oct 2019 10:40 |
|
<noname> (SPOILER) Added speedrun: 246 moves (old: 259).
Solved differently. |
|
|
Bates' Motel Mission Impossible |
Sat 05 Oct 2019 10:34 |
|
<noname> (SPOILER) Added speedrun: 149 moves (old: 292).
Nice level. |
|
|
Bates' Motel - The Sequel |
Fri 04 Oct 2019 07:41 |
|
<mark> (SPOILER) Added speedrun: 175 moves (old: 180).
I did not solve it this way, speedrun only came after seeing your solution.
So I will throw a mod back your way.
Ps the level was great thanks. |
|
|
California Roll + Extra Hot Wasabi |
Sun 29 Sep 2019 02:02 |
|
<noname> (SPOILER) Added speedrun: 694 moves (old: 1018).
End up being faster than my old speedrun... |
|
|
Magic Trick bot kiss |
Mon 23 Sep 2019 10:36 |
|
<mm> (SPOILER) Added speedrun: 422 moves (old: 432).
Nice one, too, thank you Mark! |
|
|
Magic Trick (v4) |
Thu 19 Sep 2019 08:58 |
|
<mark> (SPOILER) Added speedrun: 245 moves (old: 355).
Maybe a little bit of a clue?
Over 100 steps of last run. |
|
|
Bates' Motel No Killing |
Sun 08 Sep 2019 15:39 |
|
<mm> Nice one! |
|
|
Reposition 1 (Basic ver.) |
Thu 05 Sep 2019 20:44 |
|
<ncrecc> (SPOILER) Cook: You only need two bombs. Push one near one of the gaps in the electric floor, then push another near the blocks surrounding the door. |
|
|
Assembly Square |
Sun 01 Sep 2019 12:40 |
|
<mark> Nice speedrun I solved this with brute force in the editor years ago. I never rated not sure why.
Ive never understood how to solve any level with a pc or program.
Thanks for info below solved my first sokoban level earlier,was having a problem with 3 of the handemade ones.
Not tried Daves Reassembly maybe oneday.
Ps only 98 levels left to go, thanks again for soko solver sure this will help with quite a few more when I get the hang of it. |
|
|
Reassembly Square |
Sun 01 Sep 2019 02:31 |
|
<noname> (SPOILER) Added speedrun: 201 moves (old: 216).
In the past, I thought that fixing the first bots before randomly placing the rest, as in what I did for Assembly Square, would not work because of the placement of Bot 15. Looking at Dave's original solution again and thinking about it, however, leads me to think that it might actually be okay to fix the first seven bots. I also limited the bots to the first five rows in the 6x6 square. Otherwise, I followed the procedure described in my recent comment to Assembly Square, and this speedrun is the result. |
|
|
I'm glad I can really swim. |
Fri 30 Aug 2019 16:33 |
|
<mark> (SPOILER) Added speedrun: 222 moves (old: 242).
PC has been broken for almost 2 years not played escape until last 3 weeks.Had a list of about 30 speedruns to add but have lost it.
Found about 10 of them so far will post when I find the rest.
Have made a few fun mods lately so watch out for them. |
|
|
Reassembly Square |
Fri 30 Aug 2019 01:20 |
|
<noname> (SPOILER) Added speedrun: 216 moves (old: 223).
Using the end position my computer generated when this level fisrt came up with my hand-solved solution for the first part as input to vicinity search gives an even faster solution.
<noname> (SPOILER) Added speedrun: 223 moves (old: 231).
It turns out that using Dave's original solution as input to my vicinity search variant gives faster solution.
<noname> (SPOILER) Added speedrun: 231 moves (old: 239).
Found by using a variant of vicinity search to shorten a hard-coded silly solution to the position given by Dave's original solution. The act of randomly generate positions failed to find good positions that seem to have short sokoban solution.
For details of my variant of vicinity search, see my comment 20365 for the level #2880 "Assembly Square". |
|
|
Assembly Square |
Wed 28 Aug 2019 04:17 |
|
<noname> (SPOILER) Added speedrun: 164 moves (old: 199).
Found with the help of computer. I first have my computer randomly generate one million positions inside the 6x6 square, filter out those positions which do not need to a solution, then used a crude heuristic to estimate the length of the sokoban part for each of the remaining positions. Then I chose a few of those with a small estimated length of solution, and used a variant of the vicinity search to shorten hard-coded silly solutions for those positions, and this solution is the result.
The details of vicinity search can be found at the sokoban wiki at <sokobano.de/wiki/> (look for the section "Optimizer development" -> "YASO Optimizer"). My variant is essentially the same as the setting infinity/0. |
|
|
Be on the Right Side III |
Wed 28 Aug 2019 03:50 |
|
<noname> (SPOILER) Added speedrun: 103 moves (old: 107).
Optimal among those solutions which do not release the bot. Releasing the bot seems to make the level too complicated for my computer to solve in reasonable amount of time, so I haven't tried that part. |
|
|
The Kansas City Shuffle Shuffled |
Wed 28 Aug 2019 03:46 |
|
<noname> (SPOILER) Added speedrun: 140 moves (old: 175).
Speedrun found with the help of computer. While the first room is separate from the rest, the rest turns out to be too complicated, so I need to separate into second and third room. And I need to simulate the return from the third room to the second room while keeping the rooms separate. And it turns out that returning to the second room saves a couple of moves, comparing with not returning to the second room after reaching the third room. As I'm not certain about my simulations, and I did not check the case with returning to the second room more than once, I don't claim optimality. |
|
|
The Kansas City Shuffle |
Wed 28 Aug 2019 03:41 |
|
<noname> (SPOILER) Added speedrun: 81 moves (old: 85).
Optimal solution found by separating the level into two parts and having my computer run each part separately. The two parts are: the first room, and the rest. Of course, one needs to show that the first room is indeed separate from the rest, and this is done by showing, using computer, that the player cannot reach the second room without a broken bot in the first room. |
|
|
Do Not Press |
Wed 28 Aug 2019 03:37 |
|
<noname> (SPOILER) Added speedrun: 76 moves (old: 87).
Optimal solution found by excluding positions with a stop sign flipped before the green light is hit. (The only way one can solve without the green light is to use bots to block laser, but this does not seem possible.) Fortunately, I allowed positions with stop sign flipped *after* the green light is hit, otherwise my computer would miss this solution. |
|
|
The Long Level |
Wed 28 Aug 2019 03:31 |
|
<noname> (SPOILER) Added speedrun: 490 moves (old: 510).
Optimal solution found by separating the level into three parts and having my computer run each part separately. |
|
|
Blide Srick? |
Wed 28 Aug 2019 03:29 |
|
<noname> (SPOILER) Added speedrun: 77 moves (old: 79).
Optimal. |
|
|
Sea of Blue |
Wed 28 Aug 2019 03:28 |
|
<noname> (SPOILER) Added speedrun: 51 moves (old: 59).
Optimal. |
|
|
Married to the Sea |
Wed 28 Aug 2019 03:28 |
|
<noname> (SPOILER) Added speedrun: 66 moves (old: 86).
Optimal. |
|
|
|
|
|