May 12, 2017

Friday Puzzles #300

How appropriate to celebrate a nice round number by being nearly-but-not-quite late!

Clues outside the grid indicate the "property" of the first two numbers placed in the corresponding direction.  Odd=1,3,5,7; Even=2,4,6,8; Big=5,6,7,8; Small=1,2,3,4.

    #341 Odd-Even-Big-Small Sudoku – rated medium
All puzzles © Tom Collyer 2009-17.

May 09, 2017

2017 Sudoku GP: Round 5 Estonia

So hopefully there will be no further barrage of comments this time round, nothing too controversial going on here! The authors this time round were the world champion, Tiit Vunk, and Rauno Parnits. I can't personally recall any more than the occasional puzzle from either author, but this time round there was a pretty heavy chess theme to the round. And although a theme is nice to have, more important in my eyes is a good solid set of fun puzzles to solve, and that's what we got with this set. The vital stats according to me were:

Enjoyment: 7/10
Classics: 7/10
Favourite Puzzle: Chess Kings Sudoku

Anyhow, on to the review.

1-6 Classic Sudoku (18, 20, 16, 43, 20, 40 points)
I tend to like the classics being arranged in the order of difficulty, but I suppose the gimmick here were the patterns of givens, which were respectively a King, Queen, Rook, and a Bishop. This was followed by "b1c3" which I don't quite get (a Knight's opening?), and then "1/2" which I suppose is a draw. Nothing particularly difficult, nothing overly easy (despite some fairly low points totals) - I think the last one had some colouring going but generally a solid set of classics

7 Diagonal Sudoku (43 Points)
Another bishop shaped set of givens, I suppose this makes a lot of sense for a Diagonal puzzle.

8 Fortress Sudoku (39 points)
One of my favourite Fortress puzzles of recent memory. I don't much like the type, but this was remarkable firstly for (1) the clever use of a checkerboard pattern as the fortress with row numbering to the side, and (2) solving cleanly without me making a stupid mistake.

Scratch that, it's another fortress puzzle I've cocked up. This is utterly bizarre.  I need to go away and practice these for a week until I stop cocking them up.

9 Queen Sudoku (69 points)
I definitely prefer the plural "Queens". This was nice enough for the points, but I do think this type suffers a little bit because as soon as you try and make one with any kind of difficulty it's very tempting just to guess the placement of the queens. The non-attacking constraint is pretty strong, which pretty much guarantees a quick contradiction if you happen to make a bad guess, and easy enough to double check if you make a good guess sans contradiction. Which I did.

10 Battenburg Sudoku (73 points)
Nice to see this type make an appearance. This is generally quite a puzzle-y type as the key to solving these is by first determining whether a particular cell contains an odd or even placement, with the converse rule also needing to be used heavily to make any headway.

I am not a fan of the phrasing "all such patterns are marked".  I think it is far more friendlier to newer solvers to explicitly say "where there is no marking, there can be no pattern".  It feels like I'm banging the same old drum here, but really we do ourselves no favours as a community when we make things harder than they need to be.

11 Antiknight Sudoku (85 points)
I made the strategic error of leaving this puzzle until last, and my placing suffered massively for it.  It didn't feel particularly remarkable to me, just a tough version of a type I personally don't enjoy all that much.

12 Sum 10 Twin Sudoku (37 points)
Well done me for solving this without submitting an answer.  An otherwise pleasant clone style variant.

13 Chess Kings Sudoku (44 points)
My favourite of the round, and a surprisingly elegant extension of the "No Touch" variant.  The example for this novelty had the gimmick that you could isolate the pair of kings by considering the centres of the 3x3 boxes.  This logic left me with 1,3,4,7,8 as the possible kings, and although this still leaves a possible 10 different pairs, actually these were eliminated almost instantly leaving the rest of the puzzle to solve nicely.

14 Unicorn Sudoku (53 points)
The constraint feels very very strong for this novelty, and I only used it around three of the placed 9's to solve the puzzle.  I think it's probably stretching the chess theme a little thinly, and I don't think we'll see too many more of this type, but it solved pleasantly enough.

I had to take this test last on Monday in a bit of a rush, and my solving performance was down.  It definitely wasn't helped by not submitting an answer to one of the puzzles.  The preliminary results have now been released and it does looks like there have been a fair few mistakes.  I ended up with 12/14 puzzles correctly solved, 11 submitted, 1 stupid error, and perhaps a couple of minutes extra required to finish the anti-knight.

Despite a disappointing result, and a mix of variants which aren't necessarily to my advantage, I did enjoy this round a lot, and hope that the remaining rounds can match this standard.

Top 3 (preliminary results):
1. Jakub Ondrousek (882.8 points)
2. Jan Mrozowski (872.3)
3. Sinchai Rungsangrattanakul (847.7)
I had to scroll down much further than I cared for to find my result, and I was surprised to see that I was just two places behind Dr. Snyder, who had 2 errors costing over 25% of the points plus bonus.  I suppose I shouldn't complain too much.  Commiserations also to Tantan Dai, who informed me of further computer problems which meant her initial submission of under one hour (a score in excess of 900) was not received by the server.  Even then, re-submitting the answers via tapping them out on a mobile phone in an extra 10 minutes was still good for an incredible 6th place, and the current overall lead of the GP (see the link below).

As for the Brits:
35. Mark Goodliffe (561)
51. Heather Golding (483)
68. Tom Collyer (439)
Submitting my extra puzzle would have pushed me to 476 points, and 54th place, highlighting the magnitude of the blunder of leaving the anti-knight until last.  Mark's performance is the best British score in any of the rounds this year, and represents a really strong solve - congratulations!

What is definitely interesting is this page - which highlights that Mark and Heather are separated by a single point over 5 rounds.  Granted the dropping of the two worst scoring rounds will start coming into play here, but it's fascinating to see who will come out on top here.

I'm pretty confident I'm going to need to score bonus time on at least 2 of the remaining 3 rounds if I'm going to end up where I'd set my eyes upon at the start of the season.  I'm not sure this is looking very likely at the moment!

May 05, 2017

Friday Puzzles #299

I've got a backlog of puzzles that will hopefully be unleashed either later today or else next week.

This one is x-sums.  The clues outside the grid indicate the sum of the first X placed numbers in the corresponding direction.

So for example a clue of 6 could have X=2, in which case the second placed number would have to be 4.  Or, it could have X=3, in which case the second and third placed numbers would have be to 1 and 2 in some order.

    #340 X-Sums Sudoku – rated easy
All puzzles © Tom Collyer 2009-17.

April 11, 2017

2017 Sudoku GP: Round 4 United Kingdom

First off - the last couple of weeks of Friday Puzzles have been delayed for one reason or another.  Hopefully I can catch back up this week and post 3 puzzles this coming Friday.

Secondly, although I have been quite free with my opinions about this year's Sudoku GP thus far, if you think I have the temerity to rate my own puzzles - then you have another thing coming, dearest reader.  Nevertheless, I thought it'd be worth offering an author's (rather than ex-director's) view on this round's puzzles, which I co-authored with my good friend/bitter solving rival, David McNeill.

1-6 Classic Sudoku (25, 17, 20, 25, 47, 30 points)

So I hope this represented a decent selection of classics.  Certainly no puzzle was what I'd call a guessing puzzle.  David's were puzzle 2 and 6, the rest were mine.  Puzzle 1 is probably a lot easier if you start by placing all the 1s, then the 2s and so on.  This is a standard nikoli gimmick for easy puzzles that I like to throw out every now and again.  It works better if you aren't necessarily expecting it; over using it wears thin very quickly.  5 was the hardest, perhaps notable because it needed a few naked singles to get going.

7 Diagonal Sudoku (31 points)

David's puzzle - not sure there's too much to say other than a pleasant enough Diagonal puzzle

8 Antidiagonal Sudoku (33 points)

Mine (and I prefer Anti-Diagonal) - This was nearly a very good puzzle of 16 givens, but I couldn't quite get it to solve uniquely.  I'd have persisted with this a lot longer had this been say, for a world championships.  This would have fitted in well with a couple of other 16 given puzzles in Round 2 of the 2014 WSC, and might be something I end up revisiting.  Watch this space.

9 Windoku (58 points)

Mine - I really quite like Windoku as a type, with its 4 visible extra regions and additional 5 implied extra regions.  At first glance you might think I was going for another 16 digit puzzle and had to add another given in the middle to get to 17, but no, that's not what I was going for this time.  Perhaps the more observant of you will have noticed the centre 9 givens form a magic square?  Anyhow, a reasonably tough windoku, as far as these things go, but I suppose if you've done a few of my puzzles you'll have got through this steadily enough.

10 Antiknight Sudoku (36 points)

David's (and I prefer Anti-Knight) - I suspect he was going for a 17 given puzzle here and had to add the corner givens to get this out uniquely.  Other than that, a pleasant enough Anti-Knight solve.

11 Disjoint Groups Sudoku (56 points)

David's - I always think that half the issue with disjoint groups sudoku on paper is visualising each of the 9 disjoint groups.  When solving I often have to draw shapes in the cells to help with that.  This puzzle has a nice UK theme, and is potentially a tricky solve if you struggle to visualise the groups, but if you are comfortable with the type I think solves quite nicely.

12 Irregular Sudoku (40 points)

David's - the temptation with nicely designed irregular puzzles is to push the difficulty to an extreme, or else to go to extremes using the grid's geometry to shape the solve.  I think David hit a nice medium here - the first comment I gave when testing this was that I assumed a couple of the givens weren't strictly needed (in the sense that removing them would still give a puzzle that solved uniquely) and that it felt like an easy solve - but checking my time it was clear that this wasn't a trivial puzzle and instead the feeling that it had gone quicker is a reflection on how nicely the puzzle solved.

13 Renban Killer Sudoku (63 points)

David's - normally I'm sceptical about combining together too many rules, but the combination of Renban groups and Killer cages combines really nicely and I think David is really on to something with this variant, which he has made his own.  Another nice example

14 Cloned Strands Sudoku (33 points)

David's - I'm pretty sure David invented this type primarily to draw funny pictures of animals.  And a nice illustration that novelties don't have to (a) have overly complicated rules and (b) have overly difficult solving logic.

15 Total Blackout Sudoku (86 points)

David's - blackouts tend to have quite fiddly solving paths, so the gimmick with the sums of the cells around the cells helps to give this type a bit more flavour, as well as providing a nice pun for the variant.  I think most of the points value here comes from firstly the novelty value of the variant, and second of all the slightly fiddly bit of logic needed to resolve the top right of the grid.

Top 3 (preliminary results):
1. Kota Morinishi (850.3 points)
2. Tantan Dai (845.5)
3. Tiit Vunk (790.2)
It looks as if there is a strange result for 11th (Huxuan Yu) who has a lower points total than 10th, despite the fact he managed to solve the puzzles nearly 3 and half minutes faster, going by submission time.  It's my strong suspicion that this is another case of the failure of the claim bonus button, an issue that has been plaguing the GP for years, particularly for Chinese solvers.  I hope an appeal is made or someone notices this before the results are made final.

As for the Brits:
45. Heather Golding (478)
68. Mark Goodliffe (380)
99. Michael Collins (334)
Also good to see there was a relative strong UK turnout of 18

March 24, 2017

Friday Puzzles #298

I've run out of time to add the final polish, but there should still be enough here to keep you busy.  Cages are clued with the sums of numbers to be placed inside them; no repeats within cages.

    #339 Killer Sudoku – rated medium
All puzzles © Tom Collyer 2009-17.