I have started a thread on the Sudoku Discussions forum about the use of logic in Sudoku. The discussion started off pretty good, but it now seems to be stalling at logic notation. Good links were placed and I’m trying to catch up on lots of the material. The question remains: how do we use logic in sudoku?

Here is an example:

Take a bivalue cell C. It has 2 candidates, C1 and C2. One of them must be true, the other one false. We can write this as: (C1 v C2), ~(C1 & C2). This works fine with 2 candidates, but how about 3, or 9?

These puzzles have been posted by MikeJapan and h3lix on DJ-Ape’s forum.

They are an interesting variation which combine a zero-killer with crosswords in the empty regions. I adapted SumoCue to handle these puzzles. In “letter” mode, the program shows the placed digits and candidates as letters in the “zero” regions. In stead of cage combinations, there is a new tool that shows all remaining valid words for each region. This tool can be set to “horizontal” and “vertical” mode, because a single cell can belong to 2 words.

I have already implemented a new solving technique that eliminates candidates that do not occur in any valid word. There is a background interface with the Internet Anagram Server at Wordsmith to retrieve all valid words for the given letters. The SumoCue V2 save format can save the letter table.

I’m not sure what techniques to implement next. It’s easy to build the word lists into DLX and verify a unique solution, but it is nicer to have more human executable solving techniques. The next target could be invalid word combinations, where the combination violates the sudoku rule by placing repeating digits in a box. This could lead to words being eliminated from the list, because they cannot be used in any combination, which in turn could cause letter eliminations.

It looks like a pretty far-fetched technique, but it is still logical.

Another technique that I’ve been thinking about is a variation on the killer pair/triple/quad. When a word, when placed, would eliminate all candidates from a cell, or all combinations from a cage, the word cannot be placed. All cells of the cage must “see” the word, so this is not a technique that can be used on a large scale.

I’ll see what comes next.

These are 3 variations, which I have implemented in the latest version of SudoCue. I still need to build these extra contraints into the DLX solver, so the program can also generate these variations. The human style solver supports and solves each of these variations. A nice opportinity to publish 3 new series of daily puzzles. I’ve already generated a collection of Windokus that would easily fit into the Nightmare category.

This is a one-off puzzle, which I posted on both DJ Ape’s forum and on the forum supporting Mike Mepham’s site. A novel concept. Convert a list of names into cage sums.

I know there are very few people who read this blog, so it is safe to add the full list of names here, including the first names that I refused to post with the original puzzle.

A = Ravi Shankar
B = Lucille Ball
C = Yoko Ono
D = Oskar Schindler
E = John F Kennedy

F = Gene Krupa
G = John Pritchard

H = Lionel Hampton
I = Nelson Mandela
J = John Cage
K = Errol Flynn

L = Jim Lovell
M = Ella Fitzgerald
N = George Patton

O = Joe Shuster
P = Karel Appel
Q = Orson Welles
R = Bruno Kreisky

S = Alec Guinness
T = Niels Bohr
U = Edith Piaf

V = Dizzy Gillespie
W = Maria Callas
X = Edward Teller
Y = Joseph Barbera

If you managed to find this list, good for you!

This concept can be worked out, replacing the names with triva questions that each have the cage sum as the answer. A whole new line of puzzles can be created this way.

A stunning achievement. Mike Mepham, who has compiled a 4 page puzzle supplement for the Daily Telegraph and the Sunday Telegraph, has included a debut HaniDoku puzzle. Visits to my www.hanidoku.com website have grown to 3 figures, from the usual 10 to 15.

Interesting to see how the feedback will be. This variant is so very promising.

I just calculated the POM space for the variant with interesting results. Only 244 templates exist, but with the optional digits, there can be upto 8 positions in a template that may or may not contain a digit. I cannot yet oversee the full implications of the optional digits on the templates. It could be that there are “incomplete” templates that cannot be expanded to one of the 244 full templates, but which are nevertheless valid for a digit with optional positions. In-depth testing must be done to gain some insight.

I have been involved in so many development issues that blogging was sort of forgotten.

The release of version 1.2.0 of SudoCue went like a train. No single bug report has come in while I noticed several hundreds of people have already downloaded this version. Without any initiative from my side, the program is now listed at a dozen or more freeware download sites. Seems that sudokumakers are not the only people stealing stuff from the competition hehehe…

Clueless 3 has been released. These puzzles are immensely popular. Site visits have almost doubled since I started publishing them. This #3 is a naughty one. There are optional clues. Another first in sudokuworld.

New batches have arrived from SudokuGo, my little maker. About 20 of the 60,000 generated puzzles are Nightmare-worthy. The bar is very high.

Had an interesting discussion on both the programmers and the players forum about finned fish. Devised an algorithm that can find them all, regardless of the presence of vertices. My fellow programmers are a bit reluctant to follow suit and need more proof. I’ll give it to them soon. But then my solver is the first to catch any fish in a sudoku.

Lots of forum discussions going on. Especially on the player’s forum it is hard to focus on productive threads. You gotta read a lot to pick up the nice pieces.

I noticed some very strange behaviour in the Sudoku Top 50, the list that we all like to be high in with our Sudoku websites. The site, with a free vote for sudocue.net, is linked with http://www.inertiasoftware.com/top50/index.php?id=157.

They used to reset the counters once a day, around 12:00 GMT. Now the reset at 17:00 GMT, and, strangly enough, a second reset somewhere in the middle of the night, but never at the same time.

This gives the list 2 counting periods, from 17:00 to 03:00 (or something) and from 03:00 to 17:00.

Now here’s what happens. The site www.su-doku.net (the home of Sudoku on the Internet, yeah..) never makes a high score in the 17:00 to 03:00 timeframe, but gradualy finds its way to the top of the list in the 03:00 to 17:00 timeframe. This site happens to be Inertia’s own website, where they demonstrate their online version of Sudoku Quest.

So, what is happening here? How easy would it be for Inertia Software to give their own site a few extra votes from time to time? For me, it would be no problem at all, I can assure you that.

I already disliked the company because they packaged a searchbar with the download of Sudoku Quest, which was installed without asking confirmation (I can get very angry at such behaviour!!!) and I do not trust them and their Top 50.

My own Top 9×9 is almost ready, and I will promote it furiously as the “honest” Sudoku site ranking list.

Darn, I’m writing too much here….

Uploaded 2 more beautiful nightmares, that will help my addicted players through the weekend.