Feedback and Questions

Very interesting report, thank you. Been while since I wrote that but I suspect I was rotating the puzzle to position the X-Wing in a certain direction for illustrative purposes but added the original textual strings into the page. It does, however, show that the strategies used (at least the early ones) are independent of the rotation of the puzzle. For the advanced strategies there are often many choices and it often picks the first example on a top-left to bottom-right 1 to 9 search order. I've put the correct strings in now.

Thanks for pointing it out and have a great Christmas

Great to hear from you again, fantastic you are enjoying Sudoku and the site. Have a wonderful Christmas!

You are very welcome, many thanks for the kind words. Best of luck solving!

The 9 in E3 is a 'hidden single' which means that it is the only 9 in one direction. There are other 9s in the box so we can't use the box to fix 9. There are many 9s in row E so again 9 is not fixable with the row. But if you look in the column you will see 9 appears only once. Therefore we must put 9 in that sole remaining spot. Same happens with the 3 - last remaining cell in the column.

Have a look on this page.

Yes, do you see the line above the big board which starts "candidates can be [] edited or [] highlighted" ? Tick "Edited" and then click on cells. You will be able to add or delete candidates. Press enter to get out of the cell and "Take Step" to continue.

Multi-colouring is strictly about how to get eliminations between two separate groups of chain sets. By separate I mean there must be no pairs connecting them at any point. I haven't worked out the implications of using multi-coloring that combined two 3D-medusa nets - it would be horribly complicated and you'd have to prove separate-ness for all numbers. It's theoretically possible and an interesting idea.

I always think I've exhausted the possibilities only for someone to point out a new variant or leap of logic that makes something else possible, so I hesitate to say I don't think you can conclude anything from 'candidate OFF' alone. But here is why I think it's not necessary to consider it. OFF will only have a knock on effect if it is one side of a pair. OFF will force ON in bi-value and bi-location cases, eg when there are two candidates left in one cell. If there are 3 or more OFF doesn't say which of the others is ON. But if it is bi-value then why not start with the ON candidate - starting with the other OFF just adds one link to the chain and you're back on the previously mastered Digit/Nishio Forcing Chains.

I'd like to extend those features as well but there is a limit on what I can store in the browser and using javascript. The apps we are building have much more capacity in these senses including full undo/redo. I'm also creating a membership/account system. Storing personal puzzle definitions would be a nice feature of that so I'll add that to the wish list.

Some puzzles are relatively easy but also possess one bottleneck that requires a hard strategy to get through. My generator marks these sort of puzzles as undesirable to publish and I filter them out. The scoring is based on a) the number of rounds which is a measure of the number of opportunities to solve anything at any one time and b) the scores of each strategy used (plus some heuristic modifiers). So the grade not about which one hard strategy is used - in this case I think the X-Wing bumped it over from easy to moderate - but the overall accumulation of score parts.
More here if you are interested:
Sudoku Creation and Grading