To b]denis_berthier, Jason and blue, Thank you very much for your post.
denis_berthier and
Jasondenis_berthier wrote:I could see the consecutive spaces when I edited your post, but not in my browser (if you type consecutive spaces in html, it can "see" only one of them).
In html, you should use " " instead of spaces, but I don't know if the software of this forum accepts this.
If you type no-break spaces (alt-space) instead of mere spaces, I think it will work. (It works on my Mac.)
Jason, if you're reading this, can you give any advice ?
JasonLion wrote:Using the code tag is the best way I know of to show things where the spacing needs to be controlled. The UTF version of non-breaking space (alt-space) does work, but with a variable width font it doesn't always work out well in practice (depending on what you are trying to do). It adds extra space just fine, but you won't be able to line things up except at the start of the line.
Thank you very much for your advise.
For Japanese (and Chinese and Korean), most simple way is
to use CJK space character " " like this " ".
In any way, I had to check by "Preview" before "Submit".
denis_berthierdenis_berthier wrote:I wouldn't be so affirmative. Even if there is some confluence problem, we can't be sure it is the cause of what you see in this particular example.
Without knowing your chain rules precisely, I can't say much. Many sets of chain rules have the confluence property: reversible chains (xy-chains, AICs,...) and non-reversible ones (braids, g-braids, ...). Other sets don't: whips, g-whips,... (but, as I said before, effective occurrences of problems are very rare). Also, you spoke of "vectors"; I never heard of such a technique.
Please wait for few days. I will report after checking rating log.
I try to describe vector-based rules r14 & r15, in English.
It is based on "Off Off Vector" between two candidates.
"Off Off Vector" from Ca to Cb indicates "If Ca is not the big number, then Cb cannot be the big number".
The process of r14 is as follows.
[1]Get "Off Off Vectors"s by following processes
(1-1)If there are the strong link between Ca and Cx, and the strong/weak link between Cx and Cb,
"Off Off Vector" is defined from Ca to Cb.
(1-2)After Ca is eliminated, r7(Locked Candidates), r8(Naked Pair/Triple/Quad) and r9(Hidden Pair/Triple/Quad) is applied separately. If Cb is eliminated by any oh these application, "Off Off Vectors" is defined from Ca to Cb.
[2]Make network(s) by all "Off Off Vectors"s
For each connected component of the network(s),
for each node (candidate: denoted as Cｆ), test [3}
[3} CS(Cｆ) is defined as a set of all candidates Cr where exists a chained vectors from Cf to Cr.
If one of following contradiction condition is concluded, Cf is placed as the big number.
(A)all candidates(N) in a unit are included in CS(Cf), where N a certain number.
(B)all candidates in a cell are included in CS(Cf)
(My intention is a focused candidate(Cf) and a reached candidate(Cr).)
r15 uses "On On Vector"s, which are gotten by reversing "Off Off Vector"s .
Same procedure as r15 is proceeded.
If one of following contradiction condition is concluded, Cf is eliminated.
(A)2(or more) candidates(N) in a unit are included in CS(Cf), where N a certain number.
(B)2(or more) candidates in a cell are included in CS(Cf)
Followings are WEB page for explanation. Although explanation is in Japanese language, picture may help your understanding.In both picture, red lines are strong links.Green vectors are "Off Off Vector"s in r14, "On On Vector"s in r15.
r14:
http://www.tknottet.sakura.ne.jp/Test/G8.htmlr15:
http://www.tknottet.sakura.ne.jp/Test/G9.htmldenis_berthierdenis_berthier wrote:One way you can have an idea whether your set of rules has a problem with confluence is checking them individually for a stronger (but easier to check) property (stability for confluence): do you have a rule that can be applied in some case to produce an elimination of some candidate x, and such that no rule in your set could still eliminate x if another candidate y had been eliminated (or asserted) before? Needless to check this with the basic rules (Singles+locked-candidates+Subsets) we mentioned before (we know that set has the confluence property).
The only alternative approach I can imagine is, try to spot precisely where the divergence in your example occurs and to see what happens at that point in each case: which rule is applied or not, why it is applied in one case and not in the other, ... anything like that.
As I mentioned in above, I'm investigating what happens on two versions of Imam_bayildi.
After that I'll study on the confluence program of my T.
I suppose there are 2 problems of rule property. One is behavior on isomorphs, The other is behavior when candidate y was eliminated (as you mentioned). denis_berthier, would you kindly tell me the words for these? Are they confluence and stability ?
blueblue wrote:I've coded your rating method, and tried it with some simpler versions of "T", on the puzzles that are currently listed on your web page.
The results and puzzles are below.
There are some interesting things to notice.
1) For some of the puzzles, the rating goes up as more techniques are added.
2) Look at the ratings for the "Discrepancy" puzzle ! (8th line)
Thank you very much for your very fast action.
Would you kindly rate following 3 puzzles?
coloin wrote: (Tue Feb 24, 2015 5:08 am)
I don't think you should forget about the Easter Monster puzzle where the sk-loop was first found or the Fata_Morgana or the puzzles with high Suexratt
- Code: Select all
1.......2.9.4...5...6...7...5.9.3.......7.......85..4.7.....6...3...9.8...2.....1#Easter Monster
- Code: Select all
........3..1..56...9..4..7......9.5.7.......8.5.4.2....8..2..9...35..1..6........#Fata_Morgana
- Code: Select all
..3......4...8..36..8...1...4..6..73...9..........2..5..4.7..686........7..6..5..#Suexrat9-10364
These may or may not score highly in your program - i am suspecting the suexrat9-10364 will
I found " the rating goes up as more techniques are added" around 17Hints35042.
But is it logically possible?
