A study in Chain Forging: Page 4

Welcome back!

This proof is multi-page, multi-faceted and entirely too long for one sudoku meal. Hopefully, in spite of my verbosity, you have managed to enjoy the previous pages:

  1. First Page
  2. Second Page
  3. Third Page

This page is a bit unusual, as the entire page is dedicated to just one step. The idea that I will try to present in great detail is very simple. The potential application of this idea is wide. In general, the idea is one of a Super Cell. Super cells are nothing but a way to reduce the puzzle to smaller more managable chunks of information.


The Foundation of a native super cell


Locked 9s

Box b2 and row 2 have bilocation 9s at ac2. No matter what happens, the finished puzzle will have a pair 9x at ac2. Thus, for the purposes of chain interactions in both row 2 and box b2, one can view the puzzle as if there exists one less cell.

Thus, the foundation of a native super cell can be as simple as one candidate limited to two locations. That is simple enough, I hope!


A representation of the Native super cell


Super Cell graphic

Again, only for the purposes of chain reactions in row 2 and box b2 does this Super Cell configuration exist. Note that one can delete the 9s, and that one must include each other candidate that exists in both cells as if they existed in one cell.


What can be done with such a Super Cell? Without any loss of generality, one can treat this super cell for all chains pieces that interact only in row 2 or in box b2 as if this cell were just a normal cell. The graphic below illustrates some possible chain reactions emanating from this super cell only in box b2.


Chain pieces using the Super Cell


Super Cell Chain graphic

Study this graphic. It is nothing but some easy interactions within box b2 that are the direct result of the strong 9s at ac2. However, for the purposes of these interactions, the 9s are nothing but a reduction carrier that allows the following chain piece:


  • b23=1 == ac2=1 -- {super cell by strong 9s ac2}=2 == c3=2 -- c3=6 == b3=6
Thus we have now have another type of super cell caused by the end links of the chain above:
  • b3=6 == b23=1
  • =>Super cell 16, strong only, in column b
If there exists something in column b that interacts with this relationship, we will have some sort of chain. Of course, I would not have introduced this discussion if this relationship did not exist.


Continuous, or wrap around, chain using super cell plus chain pieces plus cell b9


Super Cell Chained to a bivalue cell

If one practices with the concept of Super Cells, then immediately upon inspection one can see the equivalent of a strong inference cell 16 in column b in box b2. Thus, one has the equivalent of pair 16 in row b at b9, b23 ish. Thus:

  • b23=1 == b3=6 -- b9=6 == b9=1
  • Note b9=1 -- b23=1, thus a wrap around chain
  • Thus b23=1 -- b3=6
  • Thus we also have a wrap around chain inside box b2 using the Super Cell!
The entire configuration is one big wrap around chain, allowing mulitple strong set deductions, thus providing one with mutliple possible eliminations.


As a complete chain, without writing down the super cell idea, and bringing back the 9s, one can write many possible chains. One way to write it follows:

  • {Hidden pair 29 at ac2} == c3=2 -- c3=6 == b3=6 -- b9=6 == b9=1 -- b23=1 == {Hidden pair 19 at ac2} thus:
    1. ac2: {Hidden pair 19} == {Hidden pair 29} forces only 129 at ac2
    2. c3=6 == c3=2 forces only 26 at c3
    3. b3=6 == b9=6 does nothing, as we already had that
    4. b9=1 == b23=1 => b145678 ≠1 => b47≠1

There is a Sue De Cox available here that eliminates exactly the same items, I think. I have found in my experience that a hidden set cross naked set representation as shown above usually co-exists with a Sue De Cox. For me, what I have detailed here is much easier to find. Also, the logic seems less strained, as less simultaneous conditions are required. Furthermore, the concept of such Super Cells is not limited to only continuous, or wrap around, chains. It can be used for many more types of reductions. Thus, understanding this concept gets much more Bang for the trouble.


The actual eliminations detailed


Eliminations detailed for this step

Bringing back the temporary disappearing candidate 9, and placing all the candidates back in their proper locations, one can now make the actual puzzle eliminations deduced by this technique. Note that we have knocked off some 4's and 3s without actually considering any 4's or 3's in our chains. This is very typical with this type of continuous chain. One can also view this deduction as a multi-hub at ac2,c3 using 29 as a bridge connected to the rim cell b9 by two strong spokes, the 6s and ones emanting from the multi-hub. Wow, that is too wordy!


Alternate view

Hopefully the following view makes sense.

  • Let the 2's and 9's form a super cell at ac2,c3. Note this super cell exists only with repect to box b2.
  • The 1's and 6's come out of this super cell within box b2 and are aligned in row b.
  • Row b already has a cell limited to only 16, so one clearly has a wrap around chain
  • The only problem with this view point is one needs to be more careful determining the precise reductions possible
Here is a possible way to write this alternate chain idea:
  • {Hidden triple 269 at ac2,c3} == b3=6 -- b9=6 == b9=1 -- b23=1 == {Hidden pair 19 at ac2}
The problem that I have with this valid represetation is that it is not clear what is happening at cell c3. However, by consistently using a larger super cell, not only does the idea become more simple, but also the valid eliminations are quite clear.


Same pattern, using a super cell of size 3


Size 3 super cell

Using the Hidden triple concept derived from the size 3 super cell, derives the following presentation of the same chain:
  • {Hidden triple 269 at ac2,c3} == b3=6 -- b9=6 == b9=1 -- b23=1 == {Hidden triple 129 at ac2,c3} proving clearly:
    • ac2,c3 are reduced to only 1269
    • 16 cannot exist outside of b239 in row b
Hopefully, one can see that the idea of the super cell easily expands to multiple cells.

In other words, the super cell is formed by candidates 29 at ac2, c3. So, temporarily erase 29. Place candidates 1346 into the super cell. See the easy wrap around chain:

  • Super Cell=6 == b3=6 -- b9=6 == b9=1 -- b23=1 == Super Cell=1
Now, make all the eliminations. Add back 29 to the super cell after one is finished. Quite Easily Done!

As an interesting aside, note that the template for the larger pattern, after using the size three super cell, is my favorite type of Y Wing Style.


Super Cell Summary

To this point, I have introduced two general types of Super Cells:

  1. Reduce a group of cells to one cell using strength in location. Allows one to temporarily erase some candidates from the group of cells. Is only valid in houses that contain all the cells that form the super cell.
  2. Form the equivalent of a strong-only cell by the use of a chain. This idea serves as a visualized link point for chain continuation.
For me, the main value in thinking with the concept of Super Cells is precisely that chain linkage is made much easier. Nevertheless, this concept has been treated earlier. The general form of this concept is Useful Groups. The concept of useful groups stems entirely from the concept of using Booleans in chains. Thus, once again, the true foundation for this concept is quite simple:

This concludes the fourth page of this proof. More possibly unchartered territories still exist. If you are adventuresome, you may enjoy visiting the next page.

3 Comments
Indicate which comments you would like to be able to see

michael  From paris, france
Steve, IMHO : your Super Cell looks brilliant and if it is not a new concept, which it may well be, then it is surely an original and illuminating presentation of some existing concept!
I am just going to put in my own words the idea as I figure it to myself.
If n candidates are locked into n+1 cells in a container, then from the perspective of that container, you can ignore (temporarily) those n candidates and consider that all remaining candidates for those n+1 cells are lodged in a single super cell (SC) which is free to link in the usual way to any cell within the container. Cells other than SC are of course free to link in the usual way outside the container.
To take your example of an SC size 3 (where size refers to the number of cells 'covered' by SC and not to the number of candidates in SC):
we know that 2 and 9 are limited to a2 c2 and c3. Thus they definitely lock two of those cells even though we don't know (or care !) which two. Consequently one and one only of the remaining candidates for those three cells (1 3 4 6) will fill the remaining cell, and again we don't know (or care) which one. As long as we remain within the container, we can regard this bunch of candidates as being in one cell, the super cell SC. Accordingly there is a valid strong link on the 6s from SC to B3 and on the 1s from SC to B2 B3, and a valid weak link between 1 and 6 in SC.
On the other hand no valid link exists between the 6s in SC and c9 which lies outside the container 'over which SC has jurisdiction'.
In passing to be noted that the notion of n candidates locked into n+1 cells is entirely different from an Almost Locked Set where there are n+1 candidates for n cells.
10/May/07 10:18 PM
David PB  From UK
What voltage are these new super-cells? Will they fit my torch?
11/May/07 6:42 AM
Steve  From Ohio    Supporting Member
Check out my page
Quite funny David!

Michael - it is quite good to present the idea formally. Of course it can look much like nothing but the execution of Almost Hidden Sets. Then it is clearly nothing new, but rather perhaps nothing but a better way to visualize relationships quickly.
However, the concept is not quite limited to only Almost Hidden Sets. In fact, I believe the idea has applications in a wide variety of situations. Hopefully, I will someday get around to more of them.
11/May/07 1:36 PM
Please Log in to post a comment.

Not a member? Joining is quick and free. As a member you get heaps of benefits.

Join Now Login