WPF Puzzle GP 2020

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
WPF Puzzle GP 2020
The first round of the 2020 Puzzle GP starts this weekend. The instructions booklet can be found here: https://gp.worldpuzzle.org/content/inst ... booklet2.
Overall, this seems to be a pretty good mix of puzzles. A bit disappointed however to see a 97point hideout fences: I disliked the genre and did terribly on it at the WPC.
Happy puzzling!
Overall, this seems to be a pretty good mix of puzzles. A bit disappointed however to see a 97point hideout fences: I disliked the genre and did terribly on it at the WPC.
Happy puzzling!

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
Preliminary results for round 1:
11. Neil Zussman 619
27. James McGowan 493
76. Tom Collyer 335
This felt like a really tough round to me (loads of difficult high point puzzles) and indeed, nobody was able to finish the round, not even Ken Endo, who sent in submissions for all puzzles but made two mistakes.
Good to see Neil putting his ‘startfromthebackwherethepointsare’ strategy to effective use!
11. Neil Zussman 619
27. James McGowan 493
76. Tom Collyer 335
This felt like a really tough round to me (loads of difficult high point puzzles) and indeed, nobody was able to finish the round, not even Ken Endo, who sent in submissions for all puzzles but made two mistakes.
Good to see Neil putting his ‘startfromthebackwherethepointsare’ strategy to effective use!
Last edited by Puzzle_Maestro on Wed 22 Jan, 2020 3:50 pm, edited 1 time in total.
Re: WPF Puzzle GP 2020
There were lots of points for this round wich kind of meant puzzle selection was key. I'm a bit annoyed I missed the fillomino variant, and the tera XX as they were both relatively easy points compared to what I did do.
Bad luck with answer keys Freddie, I assume 10/14 isn't all wild speculation and/or mistakes on your part.
Bad luck with answer keys Freddie, I assume 10/14 isn't all wild speculation and/or mistakes on your part.
Re: WPF Puzzle GP 2020
Similar story to Tom for me  I ignored some puzzles during the round which it turned out afterwards would have been easy points in the bank (looking at Fillomino and Nanro). My two mistakes were both me entering the row above the keyed row, but luckily on lowpointers so I only lost ~40 points overall.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
Results have now been updated to include partial points for puzzle 2  the word search (count). I was hoping to pip Tom to the post with that small addition, but it turns out Tom did the same thing!

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
The IB for round 2 (with points) can be found here: https://gp.worldpuzzle.org/sites/defaul ... IB_v01.pdf.
Perhaps surprisingly, there are no unfamiliar puzzles in this round: I have frequently encountered all of the puzzle types (except nonconsecutive futoshiki, but this is only a small twist on a well known type), so I suspect there will be a lot of finishers this round. The 25point fill in the blanks is slightly intimidating though...
Perhaps surprisingly, there are no unfamiliar puzzles in this round: I have frequently encountered all of the puzzle types (except nonconsecutive futoshiki, but this is only a small twist on a well known type), so I suspect there will be a lot of finishers this round. The 25point fill in the blanks is slightly intimidating though...
Re: WPF Puzzle GP 2020
Re the GP, there is a certain limit built into the balance of the rounds, as stated on the rules page: https://gp.worldpuzzle.org/content/rules
Each round will consist of a variety of puzzles, of different difficulties.
~30% of the round will consist of puzzles that have very simple rulesets that any solver can understand and approach, without requirement of a WPC background. The puzzles must still observe culture and language neutrality as is the case in a WPC.
~40% of the round will consist of some selections from the 70 or so most commonly encountered “WPCstyle” puzzle types. These puzzle types are usually called “classics” or “evergreens” at the World Puzzle Championship. The puzzles are usually on a grid and require the solver to write information in the grid subject to given information and logical constraints. The rules for the puzzle types will be familiar to most WPC veteran competitors.
~30% will consist of puzzles that a competitor might encounter at a modern World Puzzle Championship (“WPCstyle”). The puzzles are usually on a grid and require the solver to write information in the grid subject to given information and logical constraints. They can often consist of new types or new variations, and may be harder than the puzzles comprising the remainder of the round.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
I am aware of that, but the latter 30% seems to be almost nonexistent!
Re: WPF Puzzle GP 2020
I’m not very impressed with the fill in the blanks. How is that a puzzle? I had a quick look while it was printing out but assumed some level of thought, not just processing, would be involved. I’m not sure how that warrants a space in the mix.
Otherwise I did like the range of puzzles available, but it’s not like they were short of puzzles for this round.
Otherwise I did like the range of puzzles available, but it’s not like they were short of puzzles for this round.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
Preliminary results for round 2:
13. Neil Zussman 605
29. James McGowan 508
34. Freddie Hand 495
Slightly disappointed not to have done the first puzzle: like Adam, I assumed the pattern would be slightly hidden (being a 25point puzzle) so skipped it, but it turned out to just be a cubic sequence (x^3+4x^2+3x+1 for anyone interested). Also, got stuck for several minutes on the scrabble puzzle (since I couldn't find 'ar' in the set of given words), and only realised afterwards that it was in the first word 'car'. Doh!
Congratulations to Neil on another excellent result. Finishing only 7 places behind Ken Endo is something to be proud of, even if Ken was only 6th!
13. Neil Zussman 605
29. James McGowan 508
34. Freddie Hand 495
Slightly disappointed not to have done the first puzzle: like Adam, I assumed the pattern would be slightly hidden (being a 25point puzzle) so skipped it, but it turned out to just be a cubic sequence (x^3+4x^2+3x+1 for anyone interested). Also, got stuck for several minutes on the scrabble puzzle (since I couldn't find 'ar' in the set of given words), and only realised afterwards that it was in the first word 'car'. Doh!
Congratulations to Neil on another excellent result. Finishing only 7 places behind Ken Endo is something to be proud of, even if Ken was only 6th!
Re: WPF Puzzle GP 2020
Or, if that looks ugly to you (it does to me)  how about (x + 1)^3 + x^2Puzzle_Maestro wrote: ↑Tue 18 Feb, 2020 9:07 amit turned out to just be a cubic sequence (x^3+4x^2+3x+1 for anyone interested)
Re: WPF Puzzle GP 2020
Quite annoyed to miscount 7 as 6 for the 59point nurikabe  that would have moved me up to 40th. I think it’s going to be pretty competitive behind Neil and James this year!
Re: WPF Puzzle GP 2020
Also note that for whatever values you put for the sequence, there will be a 6th order polynomial you can construct that fits the sequence. It’s a terrible puzzle for the GP.
Re: WPF Puzzle GP 2020
I'm not a fan of number sequence puzzles in general, but this one was nowhere near as bad as the one from last year's Mind Sports Olympiad that was posted in another thread at the time. The difference between successful numbers is 22, 42, 68... The difference between *those* numbers is 20, 26, 32... which increases by 6 each time. It's not what I'd really call a 'logic' puzzle, but I've seen far worse sequences.
I was *only* behind EIGHT Japanese solvers (including Ken)...Puzzle_Maestro wrote: ↑Tue 18 Feb, 2020 9:07 amCongratulations to Neil on another excellent result. Finishing only 7 places behind Ken Endo is something to be proud of, even if Ken was only 6th!

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
The instructions booklet for Round 3 of the Puzzle GP can be found here: https://gp.worldpuzzle.org/sites/defaul ... nd3_IB.pdf.
Some infection puzzles can be found here: https://www.gp.worldpuzzle.org/sites/de ... Round6.pdf. They may be useful for practice if you have not done them already.
Some infection puzzles can be found here: https://www.gp.worldpuzzle.org/sites/de ... Round6.pdf. They may be useful for practice if you have not done them already.
Re: WPF Puzzle GP 2020
I remember that round... Definitely the worst I've ever done on an online contest. Can anyone tell me how to actually solve Infection puzzles? I seem to remember finding even the supposedly easier ones very difficult.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
You're not the only one Sam... I didn't participate in that round (I hadn't discovered the puzzle GP yet!) but when I tried it, I found it equally difficult. Although I am still quite slow at infection puzzles, there are a few tips I can offer:
Diagonal adjacencies are powerful. Use them fully!
The rule that no two adjacent cells can contain the same number is crucial: a lot of logical steps involve small lookaheads to reach a contradiction.
For the 2018 round 6 puzzles, I can offer the following:
16. Perform all trivial steps. R1C5, R4C6, R5C1, R6C4 all must be shaded for the same reason (if they contained a number, they would have to be 3 so there would be neighbouring 3's). The rest is mostly easy, but note the recurring pattern on the edges of the grid
17. Pretty simple. Bottom right configuration is worth noting.
18. This is where it gets interesting. After performing all trivial steps (there are very few in this case), you can deduce that R4C2 contains a number, as if it was shaded, then whichever way the 2 clue in R1C5 'escapes', it will be adjacent to another 2. Now, if R3C5 and R4C4 both contain a number, then there will be adjacent 4's in R3C4 and R4C4. (proving that two cells next to a 3 clue cannot both contain numbers is a very useful idea in general). This means there is a 4 in R5C5, and the rest is fairly easy.
19. This has a few harder deductions. Using diagonal adjacencies, you find that R1C3, R4C6 and R3C7 all contain numbers. Now there are two important logical steps: the first is to conclude that R1C3 must be a 3 (if it was a 1, the 3 clue in R2C2 wouldn't work) and the second is that if R5C6 and R4C7 both contain numbers, there would be adjacent 4's in R4C6 and R4C7. The next hard step is a small lookahead to conclude that R2C3 cannot contain a number. The rest is simple, bar a small lookahead to conclude that R6C4 cannot have a number in (would cut off bottom left region).
20. This is a tough cookie. Shade R1C8, conclude R24C2 contain numbers, and conclude that R8C7 and R7C8 must both be 3. Tough lookahead to conclude that R5C1 cannot contain a number (if it is a 3, then R6C1 must be 2 or 3, and if it is a 1, then R4C2 is a 3 and so R3C2 must also be a 3). Then conclude that R2C5 must be shaded, and later on R5C6 in a similar fashion. The rest is easy.
Obviously, there is no need to solve only using logical methods, but do not make mad guesses! These will often be unhelpful: just make small bifurcations in constrained areas.
Diagonal adjacencies are powerful. Use them fully!
The rule that no two adjacent cells can contain the same number is crucial: a lot of logical steps involve small lookaheads to reach a contradiction.
For the 2018 round 6 puzzles, I can offer the following:
16. Perform all trivial steps. R1C5, R4C6, R5C1, R6C4 all must be shaded for the same reason (if they contained a number, they would have to be 3 so there would be neighbouring 3's). The rest is mostly easy, but note the recurring pattern on the edges of the grid
17. Pretty simple. Bottom right configuration is worth noting.
18. This is where it gets interesting. After performing all trivial steps (there are very few in this case), you can deduce that R4C2 contains a number, as if it was shaded, then whichever way the 2 clue in R1C5 'escapes', it will be adjacent to another 2. Now, if R3C5 and R4C4 both contain a number, then there will be adjacent 4's in R3C4 and R4C4. (proving that two cells next to a 3 clue cannot both contain numbers is a very useful idea in general). This means there is a 4 in R5C5, and the rest is fairly easy.
19. This has a few harder deductions. Using diagonal adjacencies, you find that R1C3, R4C6 and R3C7 all contain numbers. Now there are two important logical steps: the first is to conclude that R1C3 must be a 3 (if it was a 1, the 3 clue in R2C2 wouldn't work) and the second is that if R5C6 and R4C7 both contain numbers, there would be adjacent 4's in R4C6 and R4C7. The next hard step is a small lookahead to conclude that R2C3 cannot contain a number. The rest is simple, bar a small lookahead to conclude that R6C4 cannot have a number in (would cut off bottom left region).
20. This is a tough cookie. Shade R1C8, conclude R24C2 contain numbers, and conclude that R8C7 and R7C8 must both be 3. Tough lookahead to conclude that R5C1 cannot contain a number (if it is a 3, then R6C1 must be 2 or 3, and if it is a 1, then R4C2 is a 3 and so R3C2 must also be a 3). Then conclude that R2C5 must be shaded, and later on R5C6 in a similar fashion. The rest is easy.
Obviously, there is no need to solve only using logical methods, but do not make mad guesses! These will often be unhelpful: just make small bifurcations in constrained areas.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
I made a tower defence puzzle for practice since it is a completely new puzzle type. Hope it helps!
Last edited by Puzzle_Maestro on Tue 10 Mar, 2020 4:17 pm, edited 2 times in total.
Re: WPF Puzzle GP 2020
Freddie, you star! Thanks so much for taking the time to write that up.
EDIT: So it turns out that these puzzles are called Sukoro in other places... https://www.janko.at/Raetsel/Sukoro/index.htm
EDIT: So it turns out that these puzzles are called Sukoro in other places... https://www.janko.at/Raetsel/Sukoro/index.htm
Re: WPF Puzzle GP 2020
The IB has some points in it now: https://gp.worldpuzzle.org/sites/defaul ... IB_v01.pdf
Re: WPF Puzzle GP 2020
Are you sure? Looks good to me...Puzzle_Maestro wrote: ↑Sun 08 Mar, 2020 1:57 pmEDIT: this puzzle is wrong, because I misread the instructions for the puzzle. I'll try to correct it as quickly as possible.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
The puzzle is mostly fine, there is just one small problem. I missed the 'at least' in 'A cell with a circled number does not attack, but must be attacked by at least as many cells as its number'. I would probably only need to add a few more corner clues to disambiguate.Feadoor wrote: ↑Tue 10 Mar, 2020 12:48 pmAre you sure? Looks good to me...Puzzle_Maestro wrote: ↑Sun 08 Mar, 2020 1:57 pmEDIT: this puzzle is wrong, because I misread the instructions for the puzzle. I'll try to correct it as quickly as possible.
Re: WPF Puzzle GP 2020
I managed to solve it and got a unique solution, though. Anyway, here's one that I made. I think it's reasonably difficult for its size, and has some interesting logic.

 Posts: 45
 Joined: Mon 11 Mar, 2019 5:23 pm
Re: WPF Puzzle GP 2020
Very nice puzzle Sam! Amusingly enough, I'd experimented with the same setup as you, so I found the breakin very quickly, but it still took me ages to solve... Finish was cute.
Also, you're absolutely right about my puzzle being unique: I must have missed something blindingly obvious. Oops!
Also, you're absolutely right about my puzzle being unique: I must have missed something blindingly obvious. Oops!