Sunday, 26th March 2017
Puzzles Solved Yesterday: 84
Home | Register | Login | Current Puzzle | Archives | Leaderboard | Forum | Tutorial | FAQ
Forum Index
 
Page 6 of 6« First<123456
Discussions on User Beast Puzzles
Darklady
Kwon-Tom Obsessive
Puzzles: 3283
Best Total: 9m 37s
Posted - 2012.08.13 14:08:19
I solved the first one (took me a while), and found plenty of highlanders along the way. For instance, if you put two Xs in the upper-left corner, then you can make a highlander deduction that quickly leads to a dead end, showing that the upper-left corner spaces must be lines.

They did mostly show up during trial-and-error exploration, however. And a lot of them weren't real obvious patterns, such as this double highlander situation (found on the top edge of the puzzle, near the right side) which ended one of my explorations:

pqg
Kwon-Tom Obsessive
Puzzles: 3595
Best Total: 15m 37s
Posted - 2012.08.15 14:46:04
Solved all 4. Lots of highlanders throughout, ranging from the simple (such as the first 2 up from the bottom on the left hand edge of #3), to more complex situations like the example above.

When solving, I find there are 2 situations that automatically have me looking for highlanders - empty space (which you've mostly eradicated, although there are still empty space highlanders available, such as about halfway up the right edge of #4), and any 2 with no numbers adjacent horizontally or vertically (which you have loads of, providing a rich seam of highlanders).

But even getting rid of these situations won't eliminate highlanders. If you remember Naivoj's "Highlander Patterns are Gone" series, he managed to remove most of the direct highlander deductions, but left many to be used in trial and error explorations. (in other words, you could't often simply say Highlander Rule=>Line/Cross, but you could frequently rule out a line/cross which after some deductions including one or more highlanders would lead to a contradiction) The same is broadly true of your puzzles here.

In any case, I don't think I agree with your premise that removing highlanders is the best way to make puzzles more difficult. If anything, complex highlanders are one of the hardest types of any deduction to spot.

Although you may not have succeeded in removing highlanders, these are definitely very tough, but that's more to do with them being made mainly of 1s and 2s. That style of puzzle usually requires extensive trial and error, so combining that with the larger size will always make for a challenging puzzle - look at Naivoj's #411 for an extreme example. I don't think any of your's were quite as hard as that one, but they were similarly laborious enough to make me not like the idea of a beast-size puzzle in the same style.

For what it's worth, #2 was the most difficult and #4 the easiest.
MondSemmel
Kwon-Tom Obsessive
Puzzles: 3592
Best Total: 7m 47s
Posted - 2012.08.15 15:35:30
Quote:
Originally Posted by pqg
Although you may not have succeeded in removing highlanders, these are definitely very tough, but that's more to do with them being made mainly of 1s and 2s. That style of puzzle usually requires extensive trial and error, so combining that with the larger size will always make for a challenging puzzle - look at Naivoj's #411 for an extreme example. I don't think any of your's were quite as hard as that one, but they were similarly laborious enough to make me not like the idea of a beast-size puzzle in the same style.

I agree completely. I tried puzzle #1 here, and quit way before coming anywhere close to finishing. For humans, there's no point in _just_ solving difficult puzzles - if all the puzzle really requires is trial and error, it becomes less of a puzzle and more like work...
lodenkamper
Kwon-Tom Fan
Puzzles: 21
Best Total: 47m 58s
Posted - 2012.08.16 03:48:51
Thanks to all for the feedback on the test cases.  I have posted some new beasts where the test case ideas were NOT implemented.
MondSemmel
Kwon-Tom Obsessive
Puzzles: 3592
Best Total: 7m 47s
Posted - 2012.08.27 23:03:53
Thanks for posting the new beast puzzles, lodenkamper! I tried the first three with the flash solver on this site, but I gave up #1 and #2 after >1h of solving time on each. The puzzles turned out to be more than difficult enough already - there certainly was no need to make them harder.
I don't think I'll try the remaining puzzles now. I started studying physics three years ago, and since then, time spent solving huge, complicated, abstract (e.g. slither link beast-sized) puzzles as a hobby must compete with time spent solving similarly difficult and abstract physics exercises during my studies. And my studies tend to (or at least ought to) win...
Before I became (semi-)serious about my studies, I solved tons of these huge beasts, but since then, huge beast puzzles seem more like "work" than like a "hobby" .

EDIT: That said, the puzzles I tried were certainly interesting. I think I managed to go ~30 min on both #1 and #2 before I had to resort to a highlander deduction.
Last edited by MondSemmel - 2012.08.27 23:06:12
pqg
Kwon-Tom Obsessive
Puzzles: 3595
Best Total: 15m 37s
Posted - 2012.08.28 01:12:01
I solved all 10 of these, and have to agree with Mondsemmel - they were very tough. Maybe it's just because I'm rusty at doing really hard beast size puzzles, or maybe I was doing too many other things while solving, but they seemed significantly more difficult than any of Lodenkamper's previous batches of puzzles, and in some cases even comparable to the hardest of Naivoj's (#99 etc.). As with all of the most difficult user beasts, I relied heavily on highlander techniques.

That said, they have quite a different style to the Naivoj beasts - for one thing they are more readily segregated into discrete sub-puzzles, but since the isolated areas are still very difficult that doesn't make them massively faster.

IMHO the most difficult one was #3, athough #8 and #10 pushed it close, and might have taken the title had I not become more accustomed to the techniques required by then. The most straightforward is probably #6.
Last edited by pqg - 2012.08.28 01:16:04
MondSemmel
Kwon-Tom Obsessive
Puzzles: 3592
Best Total: 7m 47s
Posted - 2012.08.28 11:35:03
pqg, congrats on solving all these puzzles. May I ask how much time it took you to solve them?
pqg
Kwon-Tom Obsessive
Puzzles: 3595
Best Total: 15m 37s
Posted - 2012.08.28 16:41:29
Hard to say really. I finished them last Thursday (a week after they were posted) and my recorded times for each one ranged from ~3 hours to well over 2 days. However, I was leaving the one in progress permanently open, though only actually working on it in spare time between work/sleep/girlfriend/life/etc. (and even then I wasn't aiming for speed, just wanted to maintain my record of having done every puzzle on the site, so I was often watching TV or something simultaneously, not fully concentrating on it) In terms of actual active solving time, I'd estimate the range as perhaps 2.5 - 7 hours? (with the exception of #3 which took quite a bit longer mainly because I ended up restarting after a mistake ~70% of the way through it)
MondSemmel
Kwon-Tom Obsessive
Puzzles: 3592
Best Total: 7m 47s
Posted - 2012.10.04 14:09:31
An update: I watched tons of Google Tech Talks and physics talks, and so had some time to do beasts in the meantime.

#05 done after ~3-5h. (Sadly, I didn't keep track.)
#06 done after ~3.5h.
#07 done after ~4.5h. Horribly difficult.
#08 still not done after 4h. I don't think I still want to solve it, either...

As I said, I tried the first four puzzles, but didn't solve any of them. I haven't tried #09 and #10 yet.
I like the puzzles, but as I said, they are really very difficult .
MondSemmel
Kwon-Tom Obsessive
Puzzles: 3592
Best Total: 7m 47s
Posted - 2014.09.22 13:57:06
Two years later (!), I tried to solve puzzle #8 again. (It's "Beast #8" in the last comment by lodenkamper in this thread.) In both attempts, I ran into logical contradictions, i.e. I couldn't find any valid solution.

Can someone post a valid solution?

Or alternatively, tell me where I went wrong in these save files (Simon Tatham's loopy.exe r7849; it seemed Tilps' LoopdeLoop could load though not solve them, and the most recent loopy.exe can't even load them): early state, almost "done"

Any help is much obliged. Thanks in advance.
Tilps
Kwon-Tom Obsessive
Puzzles: 3931
Best Total: 20m 22s
Posted - 2014.09.23 11:37:55
I wonder if the puzzle has multiple solutions but they have all been eliminated by use of highlander.  I've left my latest LoopDeLoop running on this puzzle for ages now, still going.

Edit:
I hacked together a special 'Solve From Here' mode and tested it on your 'almost done' point - it found no solutions, so I am reasonably confident that you've either made a mistake or used a highlander that eliminated the actual (multiple) solutions by that point. (Or there are no actual solutions...)
Last edited by Tilps - 2014.09.23 12:28:27
pqg
Kwon-Tom Obsessive
Puzzles: 3595
Best Total: 15m 37s
Posted - 2014.09.23 16:24:26
It's definitely possible. Just solved it again to make sure. Lots of highlander needed.
Tilps
Kwon-Tom Obsessive
Puzzles: 3931
Best Total: 20m 22s
Posted - 2014.09.23 22:30:10
LoopDeLoop solved it by brute forcing half the puzzle to ensure no multiple solutions - took ~16 hours.

Spoilers: Solution in loopy save file format.

Edit: by reversing the moves in your 'almost done' file it appears you deviate from the correct solution about 33 moves back.
Last edited by Tilps - 2014.09.23 22:38:15
MondSemmel
Kwon-Tom Obsessive
Puzzles: 3592
Best Total: 7m 47s
Posted - 2014.09.24 14:07:57
Thanks. Such a stupid mistake =(.

That said, I'm glad this doesn't happen more often to me.
Page 6 of 6« First<123456

Forum Index