## Harvest from the Assistant Report

This post summarizes what led to the review of Bob Hanson’s web report on his Sudoku Assistant, and what was gained from it. There is also an instructive checkpoint on a T&E method endorsed in the report.

I’ve been on Bob Hanson’s case for 4 ½ months! Sorry, Bob. It started with what I thought would be a quick follow up to my solution of the Easter Monster by Sysudoku human engineered methods. Then I discovered the Hanson and Marans article I had saved for review was about another “Easter Monster”. I found cause to label this one the HM Easter Monster. It was an engaging adventure to review this article and carry out their questionable battle plan via Sysudoku tactics. The H&M account was sketchy.

And in doing so, I was challenged by the H&M’s references to ALS results that I couldn’t duplicate by ALS methods I knew. Coming across a trace of a Sudoku Assistant solution of HM Easter Monster, I could not decipher it without the missing notes. This brought me quite naturally to the study of Hanson’s web report that has occupied me and my patient readers for three months.

Before summarizing the results of this study, let’s tie up a loose end and the basis for your homework assignment. I suggested you try Gail Nelson’s BV Bifurcated Nets on a Andrew Stuart dual cell example analyzed earlier. The goal was to find any of Bob’s three eliminations 1 – 3 before reaching the inevitable contradiction.

The starting bv was specified to be 28r1c9. Leaving the AIC ANL in place, the checkpoint grid uses dashed wink curves and solid slink curves to develop the truth nets blue(8) and green(2) in lock step fashion. It stops after step 5, when it becomes clear (black winks) that the green net is forcing two 5-candidates into r3.

I admit an ulterior motive for the assignment , to make the point that doing bv bifurcated nets systematically is no picnic.

There are no 1 – 3 eliminations. The blue clues bring an immediate collapse. The same result, with much less effort, is to just try 2, or 8 in r1c9. All we have accomplished is to dress up T&E to look like something profound. The logical path to a solution starting with the AIC ANL, remains concealed. Of course, we now have the blue chain of true candidates for our trouble, but in the sysudokie way of thinking, the puzzle won. The only logic we used is that one of the bv partners is true. Duh.

You can say that Gail’s method is OK if used in desperation when all known logical avenues are exhausted. I don’t think so. If standard advanced methods are truly exhausted, or extreme methods for monsters, then the 1 – 3 eliminations will be very long in coming. Also, at that point there will be much easier means to assemble a bipolar set of candidates for trial. We have not defeated the puzzle, even a monster, unless we can show human means of reaching the solution. Finally, when it comes to that, with the aid of ©PowerPoint, the development of the trial trace, with the grid simplifying at each step, is a much more reasonable human task than bv bivalue nets done systematically.

And one more thing. After the AIC ANL, were we close to desperation here in this example? Far from it, with a rich field of bv, and the slink network ripe for coloring.

Trying that, there are good prospects for bridge removals, since

!(green&red) =>

blue + orange.

But chute Sc5 contains

8(3+7)(2 + 5) + 524, and the 524 SdC verification trial brings

blue and orange, and

Sc4 =739 with it. As you might expect, this trial solves the puzzle.

Yes, some might call Sue de Coq verification a guess, but the distinction is clear. Bv trials are the favorite, easy guess of the uninitiated. But without Sue de Coq logic, the above would be an extremely unlikely guess to make. I call it a victory for Sysudoku Sue de Coq technique.

So what lessons did we learn by peering over Bob’s shoulder? I’ll place them in the Sysudoku Order of Battle as we list them as encountered :

1. Take your scratchpad suset sashimi finder with you as you scan the X-panels. Discover the sashimi. Try the available fins. Do the kraken analysis that SA doesn’t.
2. In the bv scan, watch for naked multiples spilling over into an overlapping unit. We identify such a scattered beast as a bent naked n-set, BNS0 or BNS1, and look for victims. The restriction to one or zero common numbers in the bent naked wings eliminates almost all candidates.
3. Enumerating ALS in the extreme phase, look for a second restricted common in your ALS pair. You may have a locked set.
4. From obvious extensions of nice loops and clusters, look for AIC legs forming polarity ANL.
5. In the bv scan, look for unit stripping pairs. Then look for forcing chains between them. Both cannot be true. If one confirms the other, that originating candidate is removed. If they confirm each other, both are removed.

The Sysudoku Order of Battle is amended accordingly.

Next in the blog, we leave advanced and extreme puzzles for a while, and introduce some challenging basic level collections and published series, checkpointing selections from them as a clinic on Sysudoku basic solving.

We start with Nikoli’s X-treme puzzle 180, so if you care to prepare for the basic solving party, try it out. For best checkpoint comparisons, save the grid after the 1: and 7: markings.