Lost: gsf's "A Ternary Monoid for Sudoku Coloring"

Advanced methods and approaches for solving Sudoku puzzles

Lost: gsf's "A Ternary Monoid for Sudoku Coloring"

Postby udosuk » Wed May 30, 2007 12:50 pm

I don't know why, but it seems the forum database is a little bit corrupted.:!: At least one of the threads is lost from the directory page:

http://forum.enjoysudoku.com/viewtopic.php?t=5378

:?:
Last edited by udosuk on Thu May 31, 2007 9:57 am, edited 2 times in total.
udosuk
 
Posts: 2698
Joined: 17 July 2005

Postby Son of Pappocom » Thu May 31, 2007 11:15 am

Funniness abounds, and not the ha-ha type of funniness.

I've been conversing with the owner of the account "rep'nA" who was a regular poster for well over a year. His account and posts seem to have vanished. I'm at a loss to explain this, as there is no way to actually do that within phpBB, at least not without painstakingly deleting every post (which was certainly not done -- by us at least).

Since rep'nA started that thread, the original post is missing, which would account for it not appearing in the forum topic listing.

The backup and restore system of phpBB is an all-or-nothing kind of thing, so there appears to be no way to restore rep'nA's account. Threads with contributions by him will still be visible (minus his comments, except those quoted in other posts). Threads he started will also be accessible (assuming he was not the sole contributor to them) but only via direct links (i.e. bookmarks or browsing histories).

As I say, I can't account for why or how this happened, which puts the kibosh on figuring out the solution. Any phpBB gurus in the house, I'm open to suggestions, but even on the official phpBB forums the consensus is that what's gone is gone.

Regards
Scott
Son of Pappocom
 
Posts: 217
Joined: 10 October 2005

Postby udosuk » Thu May 31, 2007 11:48 am

Oh my god... What a terrible thing to happen...:(

From my past exchanging with rep'nA I don't think he should be in trouble for what he said or did (that I'm aware of), and he has been providing very valuable ideas and advices, especially in the "Help with particular puzzles" section lately. I'm not into conspiracy theories and don't know what to say now either. I hope at least he's well in the "real life".
udosuk
 
Posts: 2698
Joined: 17 July 2005

Postby gsf » Thu May 31, 2007 12:54 pm

Son of Pappocom wrote:Funniness abounds, and not the ha-ha type of funniness.

I'm inclined to make a new thread to preserve the main note in the index
or should I wait until this gets straightened out?
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby Son of Pappocom » Thu May 31, 2007 1:51 pm

Go right ahead. I doubt there will be any way to retrieve the lost thread in a more official way.
Son of Pappocom
 
Posts: 217
Joined: 10 October 2005

Postby udosuk » Thu May 31, 2007 1:56 pm

I've just edited the main title of this very thread in case you want to preserve the original title when you revive that lost thread.
udosuk
 
Posts: 2698
Joined: 17 July 2005

re: vanished

Postby Pat » Thu May 31, 2007 4:04 pm

Son of Pappocom wrote:Funniness abounds, and not the ha-ha type of funniness.

I've been conversing with the owner of the account "rep'nA" who was a regular poster for well over a year. His account and posts seem to have vanished. I'm at a loss to explain this, as there is no way to actually do that within phpBB, at least not without painstakingly deleting every post (which was certainly not done -- by us at least).

Since rep'nA started that thread, the original post is missing, which would account for it not appearing in the forum topic listing.

The backup and restore system of phpBB is an all-or-nothing kind of thing, so there appears to be no way to restore rep'nA's account. Threads with contributions by him will still be visible (minus his comments, except those quoted in other posts). Threads he started will also be accessible (assuming he was not the sole contributor to them) but only via direct links (i.e. bookmarks or browsing histories).

As I say, I can't account for why or how this happened, which puts the kibosh on figuring out the solution. Any phpBB gurus in the house, I'm open to suggestions, but even on the official phpBB forums the consensus is that what's gone is gone.




no way to do that ??

Pat (2006.Dec.17) wrote:
        -- Nick67 was a valued regular contributor to the forum,
        definitely not a spammer --
Son of Pappocom (2006.Nov.6) wrote:When removing members from the board (for spam, voluntary deletion, or whatever),
the users' posts are removed

User avatar
Pat
 
Posts: 3438
Joined: 18 July 2005

Postby Ocean » Thu May 31, 2007 10:20 pm

First, to gsf: I should say that I found your article very interesting. The reason I have not commented on it, is that it takes time to fully "understand" the concepts (or relate them to my own algorithmic experiments - which i also do not fully understand yet...).

And, in order not to forget repn'A's original post:

rep'nA Fri Apr 13, 2007 4:52 pm wrote:gsf,

Your article is very nicely done. I love the observation that double 3's acts as an identity. I have a few questions if you don't mind.

Firstly, you present a very helpful multiplication table:

Code: Select all
000 0      100 0      200 0      300 0
010 0      110 0      210 0      310 0
020 0      120 0      220 0      320 0
030 0      130 0      230 0      330 0
001 0      101 0      201 0      301 0
011 0      111 0      211 0      311 0
021 0      121 1      221 0      321 1
031 0      131 1      231 0      331 1
002 0      102 0      202 0      302 0
012 0      112 0      212 2      312 2
022 0      122 0      222 0      322 0
032 0      132 0      232 2      332 2
003 0      103 0      203 0      303 0
013 0      113 0      213 2      313 2
023 0      123 1      223 0      323 1
033 0      133 1      233 2      333 3



and write, for example that

gsf wrote:...a segment { 1-edge 3-edge 1-edge } (denoted 131) is equivalent to a single 1-edge.



In the case where the segment is even a tri-cycle, one obtains a 1-edge from the "exceptional" vertex to itself, implying that the vertex is not the induced subgraph color. Is this correct? The reason I doubt it is that this would imply that any of the combinations in your table that produce a 1 would imply that the vertex is not the induced subgraph color. But you only list 4 of the 7 as being able to make a coloring deduction. A similar situation occurs for tri-cycles multiplying to give you 2, where you would be able to deduce instead that the exceptional vertex is the induced subgraph color. Again, there are 6 such instances in the table, but you only point out 3 specific cases as causing deductions. I suppose what makes me most skeptical of my interpretation is the clearly nonsensical situation that occurs when the tri-cycles multiply to 3. For then you should simultaneously be able to deduce that the vertex is and is not the induced subgraph color.

To put my question a different way, when you write:
gsf wrote:This next figure shows the seven tri-cycles that provide coloring information.
Code: Select all

        (0)            (0)            (0)
        / \            / \            / \
       1   1          1   1          3   3
      /     \        /     \        /     \
     *---2---*      *---3---*      *---2---*


        (1)            (1)            (1)            (1)
        / \            / \            / \            / \
       2   2          2   3          3   3          2   2
      /     \        /     \        /     \        /     \
     *---1---*      *---1---*      *---1---*      *---3---*




(0) means that the vertex cannot be the induced subgraph color, and (1) means that the vertex must be the induced subgraph color.



why are these "the" seven tri-cycles...? What doesn't work, for instance, with

Code: Select all

       (0)
       / \ 
      1   3
     /     \
    *---2---*       

:?:


My second question has to do with your example:

gsf wrote:......5.9..7..462.....2..7......6.134.6...8.587.1......2..7.....358..2..1.8......

contains a few y-cycles and y-knots:

y-cycle 4 5 a/9 [96]-[66][65][25]=[98] => [96]-[78][98]-

denotes a Y cycle on color 4 of size 5 (5 vertices, 5 edges). 1-edges are denoted by "-" ([96]-[66]), 2-edges are denoted by "=" ([25]=[98]), and 3-edges are denoted by adjacent vertex labels ([66][65]). This cycle collapses to a tri-cycle with two 1-edges ([96]-[78] and [98]-[96]) and one 3-edge ([78][98]). From the tri-cycle figure above vertex (cell) [96] cannot be the color 4.



What is "color 4"? Looking at my pencilmark grid, it doesn't appear to be the number 4.

Have you already suppressed the links giving the 2-edge between [25] and [98] or is there a brutally obvious reason for them being 2-linked?

Using your multiplication table, I would view the above deduction as (133)21 =>121 implying the exceptional vertex, [96], is not color 4. Where does the tri-cycle [96]-[78][98]- come from? [78] isn't even in the original cycle.

Hmm. Maybe I will stop now and let you respond.

Last edited by rep'nA on Fri Apr 13, 2007 6:16 pm; edited 1 time in total
Ocean
 
Posts: 442
Joined: 29 August 2005

Postby gsf » Thu May 31, 2007 10:47 pm

Ocean wrote:First, to gsf: I should say that I found your article very interesting. The reason I have not commented on it, is that it takes time to fully "understand" the concepts (or relate them to my own algorithmic experiments - which i also do not fully understand yet...).

thanks
the neat part about programming sudoku is that it can be done sans terminology
the hard part comes when independent parties have to differentiate alternate vs. new work
some of this work has crossover with x/xy chain cycle work
I believe the edge equivalence/collapse and tri-cycle observations are the new part here

I created a new thread A Ternary Monoid for Sudoku Coloring just in case the corruption lingers in this thread
we can move further discussion and any lost tidbits over there
Last edited by gsf on Fri Jun 01, 2007 12:52 am, edited 1 time in total.
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby Myth Jellies » Fri Jun 01, 2007 4:26 am

Are there any other accounts deleted--perhaps other account names with an apostrophe in the name, or something like that? A quick search shows that there are no names currently that begin with an apostrophe, though there may never have been one in the first place.
Myth Jellies
 
Posts: 593
Joined: 19 September 2005

Re: re: vanished

Postby Son of Pappocom » Fri Jun 01, 2007 10:46 am

Pat wrote:
Son of Pappocom wrote:I've been conversing with the owner of the account "rep'nA" who was a regular poster for well over a year. His account and posts seem to have vanished. I'm at a loss to explain this, as there is no way to actually do that within phpBB, at least not without painstakingly deleting every post

no way to do that ??

Yup, no way to do that -- "at least not without painstakingly deleting every post." If a spammer slips through we delete the user account and then clean up the posts by hand. It's not something that would mistakenly happen to a genuine member.
Son of Pappocom
 
Posts: 217
Joined: 10 October 2005

Postby Son of Pappocom » Fri Jun 01, 2007 10:47 am

Myth Jellies wrote:Are there any other accounts deleted--perhaps other account names with an apostrophe in the name, or something like that? A quick search shows that there are no names currently that begin with an apostrophe, though there may never have been one in the first place.

We'll know soon enough -- rep'nA's replacement account also has an apostrophe.
Son of Pappocom
 
Posts: 217
Joined: 10 October 2005

Postby ravel » Fri Jun 01, 2007 10:20 pm

I regret very much, that rep'nA's posts are lost. They were one of my favorites. Anyhow, i did read them, and he is still posting. Best wishes to his family, i am sure, they will solace him:)
ravel
 
Posts: 998
Joined: 21 February 2006

re: vanished

Postby Pat » Mon Jun 04, 2007 12:36 pm

Son of Pappocom (2007.Jun.1) wrote:
Pat wrote:
Son of Pappocom wrote:I've been conversing with the owner of the account "rep'nA" who was a regular poster for well over a year. His account and posts seem to have vanished. I'm at a loss to explain this, as there is no way to actually do that within phpBB, at least not without painstakingly deleting every post

no way to do that ??

Yup, no way to do that -- "at least not without painstakingly deleting every post." If a spammer slips through we delete the user account and then clean up the posts by hand. It's not something that would mistakenly happen to a genuine member.

OK, sorry, there goes one theory
User avatar
Pat
 
Posts: 3438
Joined: 18 July 2005

Postby Ruud » Sun Aug 12, 2007 8:49 am

I admire everybody's effort to recover the lost messages, but there might be a better way to do this.

It is clear that the rep'nA user record has been deleted. All the database queries that join to the user table will no longer include data linked to this user profile.

First, the user ID for the old rep'nA account must be recovered. The user ID for the new re'born account is 15727.

Because we know for sure that rep'nA has started thread number 5378, the following query can retrieve his old user ID:

Code: Select all
select `poster_id` from `topics` where `topic_id` = 5378

To double-check, there should not be a record in the `users` table with this `user_id`.

The board administrator can perform these 2 queries to restore all data:

Code: Select all
update `posts` set `poster_id` = 15727 where `poster_id` = (repna_id)
update `topics` set `poster_id` = 15727 where `poster_id` = (repna_id)


This would reassign all posts from the deleted account to the new account.

To correct the post count for the new user account, the following query will do the trick:

Code: Select all
update `users` set `user_posts` = (select count(*) from `posts` where `poster_id` = 15727) where `user_id` = 15727

Not required for recovery, but it would improve Adam's status. He does look a bit like a n00b right now.:D

Ruud
Last edited by Ruud on Mon Aug 13, 2007 12:33 pm, edited 2 times in total.
Ruud
 
Posts: 664
Joined: 28 October 2005

Next

Return to Advanced solving techniques