the entire thread was lost + didn't make sense to rebuild most of the "formations" as it was all covered under fish and garnered little interest.
{outside a few short and sweet easy shapes people could spot and some made its way into hodoku}
its a type of strong link for chaining : this part i kept alive here an there.
- Code: Select all
/ x /
x * x
/ x /
in essence all the cells for a box are located on 1 row and 1 col. where the box has at least 2 cells, or a max of 5.
the centre cell may or may not be empty of x.
the centre cells is the key hence the * for linking chains around corners via row/col swaps, the * is the exact centre ie "intersection" row & col of the empty rectangle formation hence named ERi
the centre cell is also important to note as if its occupied when using nice-loops this cannot be used as a strong link and can only be used as a weak link because the * is both on and off.
aic's more care is needed for elimination rules as the centre cell is both on and off.
if the star is off then it can be used easily in nice loops as both strong or a weak node.
http://forum.enjoysudoku.com/strong-links-weaklinks-t34655.html?hilit=strong%20link
interesting, i ' ll have to double check as i didn't see the condensed spaces being used when i was looking at hidden sets. or fishOh then my solver does store these spaces actually
- Code: Select all
var pos = strategyManager.RowPositionsAt(row, number);
if (pos.Count != 2) continue;
if (lines.TryGetValue(pos, out var n))
{
if (ProcessRow(strategyManager, pos, row, number, n)) return;
}
okay i do see, but it seems to be running not how i was expecting
: i was expecting Rn[x,n] & Rn[x,n2] = power-set or bit-set or set where this version tags degenerative cases as its unionizes the numbers ie two singles is still a pair, or a single and a digit with 2 spots is still a pair.
current version looks up a number see if it has 2 spots, and then passes it to an analyzer and repeats passing hits to the analyzer.
apologies: i tend to prattle on, this place moved on from nice-loops in 2010ish and focuses on the shorter easier language of a.i.c way less confusion since for discussions. {Same goes for APE/ATE evolved to Almost locked sets instead of analyzing pair matching of subsets tedious and time consuming }I'm sorry but you totally lost me on the AIC vs Nice Loops part.
yes more then likely, I've been here a time.I guess there is quite an experience gap between the two of us.
I currently do the wiki editing on https://www.reddit.com/r/sudoku/wiki/index/ as well: lots of information all in 1 spot.
. My solver indeed doesn't find these for now but I recently found a post about all the types of UR and I'm gonna implement them when I have time.
