World Puzzle Championship 2023 - First of Hopefully Many

As I sit down to write my experiences on the 2023 WPC, it's the night of October 27th, almost 11 PM. The championship ended on the 22nd. While some of this delay is because Lunarch - the remote company I work for - was having a meetup/retreat the days after the event, the rest is because I just had too many thoughts to sort through. This was not only my first WPC, but my first time attending any sort of large convention-type event, let alone a gathering of an online community. This might be a bit surprising given how long I've been around the speedrunning scene, but it's true*. I am still yet to ever attend a GDQ. So, the 2023 WPC in Toronto contained many, many firsts for me, and adequately conveying that seemed impossible. Where could I even begin, especially if a standard competition report would leave all of that out?

*MIT Logic Puzzle Open arguably counts, but that both felt and was a lot smaller.

New Genre Puzzle Pack! (Puzzles #577 - #589)

One of the things I've been doing is accumulating potential rulesets for new genres to introduce. I'd been toying with a few different ways to potentially introduce these, anything from a contest to a book to just single puzzles. Ultimately, a combination of factors forced my hand on this one, so in the past couple days I made another half dozen instances to put together into a 13 puzzle pack. It's larger than an Inaba pack (4 puzzles + example) but not as large as a Palmer pack (30+ puzzles). I think any good genre introduction should really try to show the depth and breadth of that genre, and you really do need a variety of sizes and difficulties to do that. Even here, a lot of the easier puzzles were attempts at writing the example first!

Isowatari Puzzle Pack

Rules: Shade some cells such that all shaded groups are of the indicated size, all unshaded cells are orthogonally connected, and no 2x2 area is fully unshaded. Black circles must be shaded, while  white circles must remain unshaded.

Puzzle Getchu!

Quick post here that will probably go mostly unread, but I have another puzzle contest upcoming in a few days. Information, including the instructions booklet can be found at the contest page on puzcon.jp.

I'm pretty excited to see how this goes, I think this set has some of my best puzzles and is definitely the best contest set I've put together yet, even putting aside that it's themed on one of my favorite games.

So yeah, if this is before October 4th, check out the contest and give it a try! If it's after October 4th, I'll either have posted the puzzles on this blog or will be in the process of doing so.

USPC 2023 Post-Mortem

USPC 2023 results are now up, having run this past Saturday. Somehow, I won the US side, and almost by significantly more than the results show. But that's getting ahead of myself, as there's a lot of surrounding context for this event as well as how far back I had to start preparing to put myself in the best position possible. I was not about to miss my best shot at a WPC A-team for, potentially, a long while for any reason.

Puzzle #576 - Tapa (Line)

Serkan posed a challenge to create a 10x10 Tapa Line (Tapa rules + 4 shaded cells in a row is banned) puzzle with as large of an empty space as possible. After the obvious degenerate case of a single shaded cell clued with just one row and column, a couple people tried making serious puzzles. An 8x8 empty middle was found pretty quickly, but I felt 8x9 would be possible. And it was.

puzz.link interface

The tiebreaker was to be the lowest sum of clues, so a few values are lower than they otherwise need to be. I also showcased a degenerate case that would allow for an arbitrarily large empty region on arbitrarily large grids, even with the constraint that every row and column must have a shaded cell. Really, there is no good formulation of an optimization challenge like this that doesn't end up with a degenerate puzzle allowed!

Puzzles #574 & #575 - Tilted Squares (Akari, Choco Banana)

The djmathman was streaming one or two weekends ago, and people were jokingly making meme tilted square puzzles. I contributed one of those but then decided no, I'm going to make something legitimate. So I made an Akari (always a flexible choice) and a Choco Banana (one of dj's favorite genres) that I'd somehow never made one of before? It's a fun genre but man is it hard to get started with this theme.

puzz.link interface

puzz.link interface

Puzzle #573 - People Do Read These? (Akari)

I recently caught up on logic race in Puzzlers Club, after falling very far behind. In one of my comments I wondered if anyone was actually checking 2 months of backlog and said that the first person to send me a screenshot of a comment would get a puzzle. ft029 was very prompt on this and I didn't have a puzzle planned, and they thought that it was pre-made, so... asked for an Akari. So I made an FT Akari.

puzz.link interface

Puzzle #572 - Pairs (Slitherlink)

Lately, djmathman has been making some large Nikoli-style Slitherlink puzzles. I decided to try to make a puzzle that had a bit more flow and a bit less bite to it than my usual puzzles, and what easier starting point is there than a 33 in a corner? The rest of the puzzle made itself, though I did tweak it several times to reduce the difficulty of the middle and emphasize a logically flowing theme throughout.

puzz.link interface

Puzzle #571 - Outside the Box (Masyu)

I was in a call with some friends playing Viewfinder when one of them joked that they were going to make a non-unique Masyu in the game. So I made a unique Masyu puzzle.

puzz.link interface

I started from the central ring (which makes a unique puzzle on its own on 2x2, 4x4 or 6x6) and then tried to find interesting ways to break up the pattern. I really like what I ended up finding here, especially in terms of logic. I really should make more puzzles with denser cluing like this, when the denser cluing has a good reason to be there...

Puzzle #570 - Time for Slitherlink

After the Pentominous puzzles, a little while later I decided to make a small Slitherlink puzzle. I wanted to have a sparser kind of layout than my usual with no "free" openings - I didn't completely succeed there as there's still one common pattern present, but this probably ended up making the puzzle nicer overall.

puzz.link interface

Puzzles #568 & #569 - Bite-Sized Pentominous

A while back DireKrow was making 6x6 region division examples, and said they'd had trouble making a good one for Pentominous. That seemed really interesting to me as I hadn't worked with walls in the grid for Pentominous before, and I ended up finding a pretty cool layout that showcased every rule the type has, I think?

puzz.link interface

I then continued experimenting with 6x6 puzzles and decided to copy the style of Menderbug, with the slanted diagonal. This one is pretty tricky!

puzz.link interface

Puzzles #564 - #567 - Turing Test (Nurikabe)

chaotic_iak's definition of a puzzle came up again as a topic of discussion. I very, very strongly disagree with some of the points in here, especially about excluding puzzles that are generated that meet every other requirement. (I also disagree that something can subjectively be treated as puzzle or not-puzzle based on perceived quality, though I acknowledge that this critiques are much more leveled at logic puzzles than other sorts of puzzles.) The final point in the human authorship section is this: 

"I'm willing to call these computer-assisted puzzles puzzles."

This distinction feels almost entirely meaningless to me, and I set out with an experiment to try to prove it. The thing I really dislike about this is that the exact same puzzle is considered a puzzle if it was made one way, and not if it was considered another way. And if including such a "puzzle?" in a scenario like this makes it one, when in some cases all I did was generate a puzzle and include the first one I got with no assistance or curation... then how would that be different than someone generating that puzzle and solving it?

Puzzle #563 - Continuing a Trend (Light and Shadow)

This Light and Shadow puzzle actually continued three trends. One, me making Light and Shadow puzzles because they're fun. Two, Rever made a bunch of puzzles with this clue layout in a variety of types which were pretty neat. Three, while hanging out with both Rever and jkitty in VC we started toying with other genres for this layout and, since Light and Shadow had been a recent thing to try, jkitty tried and didn't think it would work too well. So I took a stab at it and made this in about 5 minutes. To be fair, though, I'd already made the previous 5 puzzles which really helped pinpoint a potential starting point, and I did get a bit lucky with the ending working within the constraints of the start.

puzz.link interface

If this post goes up and there's not a link to Rever's puzzles with the same layout then remind me to link it or bug them to archive their puzzles somewhere already they're too nice to lose to the harsh void of time.

Puzzles #561 & #562 - The Compromise and The Uncompromising (Light and Shadow)

Here are two Light and Shadow I made, one compromising on the theme (6 and ? antisymmetry) and one where I didn't (the only white clues being the six 7s in a hexagon). The latter puzzle is probably the best Light and Shadow I've ever made and I put a lot of effort into both finding a novel (to me) break-in with the layout, and ensuring the rest of the fill didn't trivialize it along the way. As often happens, I later found that my start would render the end unworkable, but by tweaking the start slightly in a way that expanded the logic I could affect the later parts to give new options that did end up working out.

puzz.link interface

puzz.link interface

Puzzles #558, #559 & #560 - Light and Shadow

In hanging out in VC with jkittykitkat in the days after the Logic Puzzle Open, a topic of what genres should be made more came up. I suggested Light and Shadow and then there were suddenly some 5x5 puzzles using question mark clues. So... I decided to take a stab at using them myself, and with antisymmetric placements of a given value of one color and a question mark of the other. I quickly found a 5x5 with three 3s almost worked, and then a 6x6 with four 4s did. 7x7 with five 5s felt extremely possible but it took me quite a while to find a layout that did. I tried continuing the pattern farther but 6s proved a bit untenable, though you'll see that tomorrow.

puzz.link interface

puzz.link interface

puzz.link interface

Puzzles #556 & #557 - Broken Dynasties (Statue Kurodoko, Domino Context)

Recently - and by recently I mean nearly 3 months ago by the time of posting - I submitted 2 puzzles to Puzzlers Clue Logic Showcase 46, with the prompt of taking a "dynasty" genre and changing only the rule about shaded adjacency. "Dynasty" is the jargonistic term for a structure where shaded cells may not be adjacent and all unshaded cells are connected. I actually submitted with first Kurodoko Statue Park about 90 minutes after the showcase prompt was posted, and it ended up being a tied win! Clues act as Kurodoko clues, and the shaded cells must form exactly one instance of each of the twelve pentominos as in Statue Park.

puzz.link interface      penpa interface

I'm pretty sure this puzzle is unique without the 5s on the right side and possibly without the 7 at the top, but since puzzles are meant to be solved... it makes a much better puzzle to not require ridiculous packing steps.

I also made a Domino Context, where the only rule change is that all shaded groups are dominos instead of monominos. This one ended up in 5th (behind two Menderbug puzzles with the same number of votes, losing the tiebreak to Menderbug as I voted for 2 puzzles instead of 3) which is pretty respectable. I think any of these could have ended up winning and honestly, that this variant is more interesting than the base genre.

puzz.link interface      penpa interface

Puzzles #554 & #555 - Three? (Tapa)

This pair of Tapa puzzles was brought on from playing 14 Minesweeper Variants, specifically Wall and Partition. I knew that 3? could very interesting, both forcing a bit with the 3 and giving some flexibility with the ? part of the clue. The puzzle ended up working out much nicer than I had expected, and I was able to keep clues off of the edge entirely.

puzz.link interface

Then someone - I don't remember who - said I shouldn't make a "3?" puzzle using only ??? clues. So I made a puzzle using only ??? clues.

puzz.link interface

Puzzle #553 - Smile (Sudoku)

A while back pzprrt added support for a lot more genres, including Sudoku. I spent a bit of time toying with it to see how it felt and turned up what is apparently a reasonably nice vanilla sudoku puzzle. I wouldn't know myself, I'm terrible at assessing these, though tracing why something was forced was certainly useful for making the puzzle work more nicely, without having to overclue like in the past.

puzz.link interface

Puzzle #552 - Penalty Heyawake

Something I've seen happen more often recent is genres to have some sot of "theory" developed for them, utilizing the underlying structure to prove something must be true within the bounds of a puzzle. One older example of this is Yin Yang never being able to have a checkerboard pattern, while the most well known recent examples are penalty and river Heyawake. These are pretty hard to explain and pretty hard to understand, though basic applications are fortunately approachable enough. Here are three explanations if you want to dig farther into the topic. I was pretty happy to find an antisymmetric construction for this grid size.

puzz.link interface

For this puzzle, the basic calculation will prove that there must be the maximum number of shaded cells on the edge and exactly zero loops of unshaded cells. In other words, 0 penalties or inefficiencies in the packing.

Puzzle #551 - Three of Threes (Slitherlink)

This kind of Slitherlink structure, with evenly spaced out 3s has some interesting properties that I don't fully understand. The consequences when interacting with 1s and 2s are pretty hard to track, but 3s are somewhat approachable. So, I made this puzzle to try to get a feel for the structure better.

puzz.link interface

Puzzle #550 - IPC 2023 Round 3 Giant (Shikaku)

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

My final contribution was a large puzzle for round 3. By the time I got around to this one most of the other giants were already made and there were no region division puzzles. The difficulties also seemed to be tending high so I went for Shikaku as it's usually a pretty easy region division puzzle, and in particular one that's fast to input. I still didn't want this to be trivial and so started from a symmetric set of numbers that didn't resolve, and then made sure the surrounding clues made them resolve in two different ways. Minor spoiler I guess, though any rigorous solve will find out anyway!

puzz.link interface

The fact that this ended up on 550 exactly - another milestone puzzle post - is purely coincidental. I still have a few more stray puzzles to queue up, so daily puzzles will continue a little longer.

Puzzles #548 & #549 - IPC 2023 Round 3 Extras

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

These two Double Choco puzzles weren't used because they were 6x9 instead of the 6x6 limit. I do really love the theming and solves on these though, and it was worth a shot since I was struggling to come up with more good 6x6 puzzles with some bite.

puzz.link interface                   puzz.link interface

Puzzles #542 - #547 - IPC 2023 Round 3 (Double Choco)

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

This year, round 3 had a wild structure: thirty-six small puzzles, 6 each of 6 types, and 6 giant/large puzzles. It was not intended to be finishable and even my testsolve times of the entire round would come in a bit over the 1 hour time limit, without accounting for my 6th. Insanity. Anyway I made a bunch of small Double Choco puzzles and don't really have comments on them individually. Puzzle links first, images after unlike my usual order.

Double Choco 1    Double Choco 2    Double Choco 3

Double Choco 4    Double Choco 5    Double Choco 6

The first two puzzles here led off the round with the classic "IPC" "2 23". From there, using only 2s and 3s just kind of happened - one antisymmetric puzzle that took a very long time to find, a numberless, and two that mostly focused on a clean chocolate layout. I made a couple more of these, but that's what tomorrow is for. :)

Puzzles #540 & #541 - IPC 2023 Rule Pool Extras

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

Round 2 also required a new "rule pool" for non-loop puzzles, with a list of possible rules that might be used where the final puzzle or puzzles would use some combination of those rules. The list of rules for shading I came up with were:

• Numbers may never be shaded.

1. Numbers give the area of the region they're in. (Nurikabe clues)
2. Numbers give the number of shaded cells in the up to 8 surrounding cells. (Minesweeper clues)
3. Black circles must be shaded, white circles must be unshaded. (Statue Park clues)
4. All unshaded cells are connected.
5. All shaded cells are connected.
6. No two shaded cells share an edge.
7. All shaded areas have exactly 4 cells.

The first bonus puzzle (meant to be used with the other two) uses rules 2, 4 and 5. I also tried writing this on the train, but scrapped the original version as I ran into some constraints that proved difficult to work with without larger backtracking available. A couple days later I made this puzzle where I tried for a logically flowing theme.

puzz.link interface      penpa interface

This second puzzle uses rules 2, 4 and 7. This set really seems to have the strongest potential of the rules I selected, and if it didn't solve like a mashup of Minesweeper and Statue Park I'd probably poke at it more. This one wasn't used for difficulty, as testsolve times would have put this as the hardest puzzle in the round!

puzz.link interface      penpa interface

Puzzles #538 & #539 - IPC 2023 Rule Pool

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

Round 2 also required a new "rule pool" for non-loop puzzles, with a list of possible rules that might be used where the final puzzle or puzzles would use some combination of those rules. The list of rules for shading I came up with were:

• Numbers may never be shaded.

1. Numbers give the area of the region they're in. (Nurikabe clues)
2. Numbers give the number of shaded cells in the up to 8 surrounding cells. (Minesweeper clues)
3. Black circles must be shaded, white circles must be unshaded. (Statue Park clues)
4. All unshaded cells are connected.
5. All shaded cells are connected.
6. No two shaded cells share an edge.
7. All shaded areas have exactly 4 cells.

I took this opportunity to experiment farther with some rule combinations that are less common, and aimed for two options of what to use in the contest. I made a set of three smaller/easier puzzles, and one harder puzzle, with no real preference as to what was used. It turned out, neither - the round ended up using these two from the set of three, and I'll post the bonus puzzles tomorrow.

And by that I mean they're queued up for tomorrow. I'm writing this in mid June.

First puzzle: 30 points, using rules 1 and 6. This isn't a particularly rich ruleset but it's nice for a couple one-offs. I wrote this puzzle on the train to the MIT Logic Puzzle Open too!

puzz.link interface      penpa interface

Second puzzle: 10 points, using rules 3, 4 and 7. While this puzzle is intentionally very breezy, this combination of rules is very close to an original genre I have lying around that I think has some real potential. I'll make some puzzles for that later, but it adds a no 2x2 white rule (giving a way to force more shaded cells) and allows varying the shaded area sizes allowed in a puzzle.

puzz.link interface      penpa interface

Puzzles #535, #536 & #537 - IPC 2023 Round 2 (Akichiwake, Symmarea)

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

Round 2 is traditionally puzzles from the same categories as Puzzle Ramayan that year, but of different types. I picked Made In India (since Akichiwake was recently gaining prominence and it's a Prasanna invention) and Regions (since I like making region division). I started out by making the second puzzle here with a construction that got away from me, in the pursuit of a restrictive theme and layout. I then took that knowledge and used it to make another puzzle that also got away from me a bit, though since I started smaller the final puzzle ended up being much more reasonable, though still one of the harder puzzles in the round at 50/450 points.

puzz.link interface

puzz.link interface

For region division I went with a Symmarea because it's a fun type and I'd never made one before. It gave me a lot more trouble than I anticipated and I almost switched genres to something easier to make. I'm glad I persevered though, since I ended up with a really slick 90 point puzzle at the end of the process. Lesson learned, don't try a restrictive theme (cloned rings) for a first construction of a genre that dead-ends easily.
Over a month after the IPC, I realized there were two ways to resolve the last value. The original version of the puzzle is linked below - the image and first link are the fixed version, the second link is the original.

puzz.link interface

original, two solution version (penpa)

Puzzles #533 & #534 - IPC 2023 Round 1 (Product Cave, Curve Data)

I was lucky enough to be given the opportunity to construct puzzles for the Indian Puzzle Championship this year. Before getting into some of the puzzles I prepared, you can acquire the full set of puzzles at the link above. The IPC (and LMI, a phenomenal puzzling site) operate at a loss, so if you want to throw some support their way it would definitely be appreciated.

Round 1 is traditionally composed of puzzle genres that appeared in the Puzzle Ramayan events that year. Product Cave showed up as a "classic variant" and I jumped on the opportunity to write one. I love a good Cave puzzle! I spent some time on the train ride to Boston for the MIT Logic Puzzle Open thinking about themes and deductions I could use for this (as well as some stuff I'll post later...) and ended up hitting on some nice steps I could use. This puzzle ended up being worth 45/450 points in the round. As a reminder, the rules are the same as standard Cave, except the number gives the product of the horizontally and vertically visible cells.

puzz.link interface      penpa interface

There were also Curve Data puzzles. This has started to become another signature genre I write, oops. Prasanna placed 2nd overall in the 2022 WPC, so I wanted to honor that with this theming and make sure the puzzle was non-trivial. I, uh, ended up with a 60 pointer that was the highest value puzzle in the round.

puzz.link interface

SO MANY BEES (the full set)

For secret solver last year, I was assigned to write a puzzle for dylanamite. I did not write a puzzle, I wrote a 26 puzzle set that I posted over the previous 8 days, ending today. When Jamie signed Dylan up (Dylan would have anyway) he did so with the comment "please send me so many bees". And so, I sent so, so, so many bees.

Here's the full PDF.

Even if you've already solved everything before this point, the PDF is still worth a look through for some of the jokes and presentation. Each puzzle has a short blurb underneath it, and a couple puzzles shown already have something extra attached.

Now, why were J, M and R missing from puzzle posts prior? The answer is pretty simple: those puzzles weren't just puzzles with memes, they were meme puzzles. A parody of Jeopardy, the legendary 2x2 Masyu (with a few twists, especially in the PDF) and one of the best logic race memes in Rooms of Fours Bees. None of these are real puzzles which is why they aren't numbered, but I still wanted to call them out and share the entire PDF of puzzles at the end. Next up: IPC puzzles!

I also assembled a full set of solutions that can be found here.

Puzzles #530, #531 & #532 - So Many Bees Part 8

For secret solver last year, I was assigned to write a puzzle for dylanamite. I did not write a puzzle, I wrote a 26 puzzle set that I'll be posting the full PDF for tomorrow. When Jamie signed Dylan up (Dylan would have anyway) he did so with the comment "please send me so many bees". And so, I sent so, so, so many bees.

Just so many bees.

X Sums with Bees

Standard X Sums apply, except the number range is 0-5 instead of 1-6. Adding a 0 to the range always changes how a latin square puzzle feels a lot in interesting ways, I'm surprised I hadn't seen this done before. Though naturally, in the pack it was described as 1-5 and a bee, because bees. And bees outside the puzzle that do nothing, because bees.

puzz.link interface      penpa interface

Yajisan Kazusan

Standard Yajisan-Kazusan rules apply. This was actually the first puzzle I made for So Many Bees, and I really tried to make one that Jamie would be happy with. Not only do I think I succeeded, I know I did. Now the only question here is, is this a variant puzzle?

puzz.link interface      penpa interface

Zigzag

Standard Password Path rules apply. Puzzle types beginning with Z are rare, so I went with one final alternate name to get it right. The background color here was deliberately chosen to be the exact shade of the "Kappa" emote in Puzzlers Club. Why? Because Jamie suggested puce and no one else had a better idea, so that's what it is. This was also the only puzzle from my first drafts that needed to be revised for non-uniqueness - trying to make this one hard was quite hard.

penpa interface

Puzzles #527, #528 & #529 - So Many Bees Part 7

For secret solver last year, I was assigned to write a puzzle for dylanamite. I did not write a puzzle, I wrote a 26 puzzle set that I'll be posting over the next 2 days, ending on the 10th with the full PDF that was sent. When Jamie signed Dylan up (Dylan would have anyway) he did so with the comment "please send me so many bees". And so, I sent so, so, so many bees.

Just so many bees.

Uso-One but the clues are obscured by a faint buzzing sound

Standard Uso-One rules apply. Additionally, the clued are all ciphered: each distinct symbol stands for a different value, and all instances of the same symbol have the same value. Liar puzzles are still something I'm not comfortable with and always struggle. The region layout here definitely proved challenging to work with.

puzz.link interface      penpa interface (nice)      penpa interface (bees)

Vertex Slitherlink (Bees and Crabs)

Standard Vertex Slitherlink rules apply. Additionally, all bees must be inside the loop and all crabs must be outside the loop. Why crabs? Because the year prior, Dylan received a room escape game set in Jamie's basement, with so many crabs in it. Vertex Slitherlink as a type is really prone to getting uniqueness issues if you're not careful with setting up areas with all vertices used, it's pretty weird.

penpa interface

Word Search for Bees

Standard Word Search rules apply. Additionally, some cells are empty and their contents must be determined during solving. All instances of a bee emoji stand for either "BE" or "BEE".

I probably spent longer picking the word list than building the puzzle, honestly.

penpa interface