WPF Sudoku GP 2020

Information and discussion on Puzzle Competitions
Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

WPF Sudoku GP 2020

Post by Feadoor » Wed 01 Jan, 2020 9:44 am

The 2020 edition of the Sudoku GP kicks off this weekend with a round from Croatia. The instructions are available at https://gp.worldpuzzle.org/content/instruction-booklet

Have fun everyone!

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Tue 07 Jan, 2020 11:58 pm

Round 1 results. I thought that was pretty tough, and that seems to be backed up by only 12 solvers getting a full 13/13. I hope the GP's aren't going the same way as the USPC and the so-called "Snyder Effect" with the difficulty ratcheting up year on year.

Top 3 UK preliminary results are:

39. Heather Golding 479
46. Tom Collyer 436
58. Sam Cappleman-Lynes 406

Nice to see Heather back in competition with a good result!

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Sun 26 Jan, 2020 10:33 am

The instructions for my round are now live at https://gp.worldpuzzle.org/content/instruction-booklet

Nikola has introduced a linguistic quirk into the instructions for Puzzle 13 which I mentioned to him a while ago but doesn't appear to have been fixed. I've asked for it to be updated, so there will hopefully be a new version of the instructions soon.

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Mon 27 Jan, 2020 2:19 pm

What's the quirk Sam? I think I understand things OK?

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Mon 27 Jan, 2020 3:18 pm

My "for example" became "for the example" in Nikola's wording which I was worried would lead people to believe that something about the rules applied only to the example puzzle and not to the competition puzzle.

A new version has been produced using the phrase "for instance" instead - just waiting on Karel to upload it.

GarethMoore
Posts: 1245
Joined: Thu 24 Jun, 2010 9:27 pm
Contact:

Re: WPF Sudoku GP 2020

Post by GarethMoore » Wed 29 Jan, 2020 2:32 am

I see what you mean - that extra "the" completely changed the meaning!

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 04 Feb, 2020 7:44 am

The preliminary results are available, with the UK top 3 as follows:

19. Tom Collyer - 509
50. Jeremy Kong - 384
73. Heather Mitchell - 336

A strong performance by Tom, slotting in just above the 8 people who solved everything but the Hidden Clone.

The overall winner was Tiit Vunk, who at 63 minutes was fastest among the 9 all-correct solvers, nearly 13 minutes clear of the rest of the field.

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Tue 04 Feb, 2020 8:39 am

yeah I suspect that hidden clone caused more than a few people a spot of bother. I’m not at all sure I properly tweaked all of the logic. I just about tweaked that the regions might overlap, tried the top left shift and that just about worked with a bit of experimentation. Not sure how you’d have ruled out the bottom right shift though... I vaguely get the sense that it was roughly giving you similar looking initial pencilmarks. I don’t think I’d have included it if I were still in charge - even if it doesn’t seem to have done me much harm this time round :roll: I’ll probs use this puzzle in particular for some of this what is a sudoku discussion, as it’s a good example of a puzzle where you have to a fair amount of non-standard head-scratching with the constraint before you can start placing anything.

The classics were very nice, and the anti/diagonal and anti/windoku doppelgängers were great - the 16 given diagonals in particular were a brilliant construction. Great job Sam!

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 04 Feb, 2020 9:28 am

I'm glad you enjoyed the round! I knew the Hidden Clone was a bit unusual, but ultimately I liked it too much not to submit it, and Nikola also felt reluctant not to include it. I at least made sure to construct it in such a way that it was still a decent Sudoku after the start, even if the start itself was overly gimmicky.

Something I wanted from the round was to put the Classics back on a pedestal. In most rounds there is maybe one Classic that is on the medium-hard end of the spectrum, and some of the easier ones have a tendency to be how-fast-can-you-write all-singles puzzles. So I ended up with the easiest Classics requiring supposedly "harder" techniques than the others (3 pairs straight in a row for #1, an X-Wing for #2) and aimed for half of them to be worth 30+ points which just about worked out. I think most computer solvers would point to the 40-pointer as being one of the easiest because it is, in fact, solvable with only singles - but we humans know that isn't the whole story!

I also made the easiest puzzle in the round a variant (the Antiwindoku) which I think is fairly unusual. The points values don't point to it as being the easiest but I think that's just because it's a rare variant.
Last edited by Feadoor on Tue 04 Feb, 2020 11:26 am, edited 1 time in total.

tamz29
Posts: 62
Joined: Wed 01 Jun, 2011 1:51 pm

Re: WPF Sudoku GP 2020

Post by tamz29 » Tue 04 Feb, 2020 11:05 am

I thought the Classics were very very difficult for 20-30 points a grid.
And after guessing multiple layers deep through the first 5 puzzles, I ditched the 6th only to find out after the test that it was, indeed, the easiest (at least for me). The Japanese community in the twittersphere seem to more or less agree.

The Hidden Clone was very nice and am glad to have solved it in time.
Re: Tom, the bottom-right overlap is ruled out because it only leaves 3 numbers for a 4-number shaded bank.
I too think this particular puzzle is a bit puzzly for a sudoku GP. Though, Tiit certainly didn't seem too distracted.

The theme of twins and the givens in the variants resembling the classics was elegant.
Amazing how you pulled the Diagonal / Anti-diagonal twin off.

Thanks for the puzzles!

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 04 Feb, 2020 11:22 am

tamz29 wrote:
Tue 04 Feb, 2020 11:05 am
And after guessing multiple layers deep through the first 5 puzzles, I ditched the 6th only to find out after the test that it was, indeed, the easiest (at least for me). The Japanese community in the twittersphere seem to more or less agree.
I was surprised by the order that the Classics ended up in - when I submitted them, my own order, from easiest to hardest, was #3, #6, #2, #1, #4, #5 - but the test-solvers obviously had a different experience.
detuned wrote:
Tue 04 Feb, 2020 8:39 am
the 16 given diagonals in particular were a brilliant construction
There was also a 16-given pair of Windokus in a first draft, but Nikola rightly ditched them in favour of an easier pair to introduce a bit more balance to the round. You can see the originals at https://logic-masters.de/Raetselportal/ ... &id=00037F

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Tue 04 Feb, 2020 12:27 pm

In general the classics points have been a bit weird. At a guess I’d say for any given puzzle one of the testers had a great time, perhaps after guessing or even restarting. I doubt that was the preferring for any one individual tester either!

My strategy has been to ignore the lower point classics until I’ve done the higher point ones given that I seem to solve the higher points classics with a better PPM compared to the lower points.

I’ll have to give the windokus a go too... funnily enough I had exactly that pair of puzzles as 16 givens as the first puzzles I wrote for the 2014 WSC!

[Edit - how interesting, you used exactly the same pattern of givens I did for the windoku. Getting a windoku down to 16 certainly was not easy]

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 04 Feb, 2020 1:12 pm

detuned wrote:
Tue 04 Feb, 2020 12:27 pm
[Edit - how interesting, you used exactly the same pattern of givens I did for the windoku. Getting a windoku down to 16 certainly was not easy]
It would seem so! If you look closely, I used the same "boxes" pattern for the diagonals too. But I thought it would look silly if there were four puzzles all with the same layout, so I swapped some rows and columns on the diagonals to enlarge the boxes and make it less obvious. Of course, then the Windoku pair was replaced anyway, so I could have stuck with the original pattern on the diagonals which I think looks a bit nicer - but it wasn't worth the hassle because by that time the official files had been created and the solution keys were already decided.

pinkagape
Posts: 92
Joined: Mon 30 Apr, 2012 4:11 pm

Re: WPF Sudoku GP 2020

Post by pinkagape » Tue 04 Feb, 2020 6:57 pm

With the hidden clone, I managed to work out where the clone was, but couldn’t place a single number. Could you give me a hint to start it now?

I found the classics very difficult too. For so few points, I thought they were fairly nasty. I don’t mind hard classics being there, but for those points, I should have steered well clear.

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 04 Feb, 2020 7:24 pm

I sent you a PM, Adam.

pinkagape
Posts: 92
Joined: Mon 30 Apr, 2012 4:11 pm

Re: WPF Sudoku GP 2020

Post by pinkagape » Tue 04 Feb, 2020 7:44 pm

Got it. Thank you! Turns a frustrating mess into a lovely solution :)

Nilz
Posts: 277
Joined: Fri 02 Sep, 2011 5:21 pm

Re: WPF Sudoku GP 2020

Post by Nilz » Tue 04 Feb, 2020 9:29 pm

I felt like I solved more puzzles than I did which is an odd feeling. But anyway, those that I did solve were (mostly ;-) ) nice, particularly the diagonals.

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Fri 21 Feb, 2020 7:39 pm

The instructions for Hungary's Round 3 are live: https://gp.worldpuzzle.org/content/instruction-booklet

Looks to be a pleasant mix of well-known variants.

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Sat 22 Feb, 2020 7:00 pm

Well known is probably stretching things a bit :roll: personally I think that would apply to Extra Regions, Fortress and maybe Renban. I much preferred the balance of types in round 2 to be honest :D

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 03 Mar, 2020 7:06 am

Round 3 preliminary results are out:

26. Tom Collyer - 518
37. Sam Cappleman-Lynes - 464
61. Mark Goodliffe - 402

Tom follows up his strong Round 2 with another solid performance here. Well done!

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Tue 03 Mar, 2020 8:35 am

I'd be curious to know what people thought of the hardest Classic. The pattern on 5s, (whether you want to call it simple colouring, finned X-Wing, skyscraper, 2-string kite...) is something that I have a lot of practice at spotting and is very much part of my repertoire. But I normally steer clear of them as an author because I'm not sure how the wider base of solvers perceives the technique...

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Tue 03 Mar, 2020 1:21 pm

I'll have to go back and have another look at the classics. I made a very conscious decision to avoid quad max and more or less work backwards. There was a point where I wasn't sure I was going to get back through all the classics, but I seemed to get through them OK, but my focus was very much on getting through them quickly. I don't really remember having too much bother, aside from number 4 where I made a guess to speed things up.

Feadoor
Posts: 178
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Sudoku GP 2020

Post by Feadoor » Mon 23 Mar, 2020 6:14 pm

Round 4 instruction booklet: https://gp.worldpuzzle.org/content/instruction-booklet

Crazy looking round which throws the 40/40/20 split for classic/familiar/novel completely out of the window. Puzzle 8 seems inspired by one of my creations.

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Mon 23 Mar, 2020 6:32 pm

I wouldn't be totally convinced that's not just a coincidence Sam - at any rate this tic tac toe puzzle seems a bit friendlier and less reliant on second order logic than yours, although some of the same hidden gimmicks still apply (e.g. orientations of even winning lines).

I wonder what this Kropki thing is going to be like? I saw the points and thought this might be something hideously complicated. In theory restricting the constraint to just lines feels a bit nicer, as you aren't always on the lookout for the negative constraint all the time. On the other hand those negative constraints are so strong that it can often mean that once you have a third to half the puzzle complete that you are more or less done. I think I'm expecting to see plenty of C's, Z's and 20's!

detuned
Posts: 1957
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Sudoku GP 2020

Post by detuned » Wed 01 Apr, 2020 9:54 pm

I did not enjoy that round for many reasons - and looking at those results I don't think I was the only one!

54. Sam Cappleman-Lynes 450
71. Jeremy Kong 404
77. Tom Collyer 394

Post Reply