## January 3, 2015

Post puzzles for others to solve here.

### Re: January 3, 2015

blue wrote:
DonM wrote:2. The over/inappropriate of computer solvers has affected the advance of good Eureka notation. A number of the computer solvers were programmed years ago and will spit out contructs and/or various terms that were defined/used in the first fews years of sudoku, but which were eventually no longer used as it was determined that either they could be incorporated into Eureka notation or they should be totally discarded as being too assumptive. I find it startling to periodically see some of these be resurrected on this forum as if we are back in 2006-7.

Hi Don,

You've used the term "assumptive", and mentioned the idea that some things can be "too assumptive", and (perhaps) used the word in other ways, recently ... "least assumptive", I think (?) ... but I've never been able to grasp exactly what you mean.
Can I prevail upon you to do your best to clarify the meaning(s) ?
E.g. "more or less assumptive than something else" ... what might that mean ?
I'm guessing that techniques like singles, locked sets and unfinned fish, you'ld classify as being "non-assumptive" (?).

Also, would you shed some light on a threshold (or threshold area) where you perceive that things have become "too assumptive" ?
Couch it in terms of "IMHO", if you like. In fact, please do.

(Okay, assume all of this as being 'IMHO")
To start with, I've never been totally happy with the term 'assumptive'. One of the many definitions of it that I would find relative would be 'assuming something without proof'. But in the end, here we're really talking about the degree of guessing.

Keeping in mind that we are talking about a puzzle game, my challenge in solving is to keep the element of guessing to a minimum and that means using as many proven patterns as possible. Just for reference, at the extreme of greatest assumptiveness or most guessing would be the use of the forcing chain method where you randomly start with one digit in a bivalue cell, see where it goes and then start with the other digit and do likewise. This requires the use of minimal logic- it's essentially a mindless exercise. At the other extreme of least assumptiveness (least guessing) would be finding something like an x-wing or a naked pair. The finding of these patterns is the real fun of solving sudoku- you find the pattern and the resulting exclusions are reliable and proven.

As the puzzles get more difficult, the less you will be able to use the simpler patterns to solve them. The increasing complexity of Eureka-notated chains to solve increasingly difficult puzzles is an exercise in using what is really an exercise in finding patterns that are more obscure than an x-wing, but are patterns nonetheless. For instance, after one has become more experienced in the pattern solving process, doesn't one get to the point that you can see a pattern of strong links that you've learned will often lead to an exclusion? If one can quickly see the exclusions that result from a simple naked pair pattern, isn't it a short jump to see the exclusions possible from a simple almost naked pair (ALS) pattern?

When you get to the really difficult puzzles, it seems self-evident that the element of assumptiveness is going to increase. When one uses one Kraken cell chain (essentially trifurcating a cell) to solve a puzzle, isn't one getting farther away from pattern-solving than if one used 2 chains each with perhaps an ALS or somesuch? I still consider that there are elements of a pattern in Kraken solutions, but just less so than in simpler chains.

What is the threshold for too much assumptiveness? It's a difficult question because some of it is subjective and some of it depends on what the objective is. On this forum, the objective (that seemed to develop almost unspoken some time ago) has been mostly to solve these daily puzzles with one-line solutions. That sometimes requires using constructs that wouldn't be considered as appropriate if the objective was to use the simplest chains possible even if more than one was required. Still, even with the one-line objective, I would find anything bordering on the forcing chain method as being mindless guessing which doesn't require any skill, doesn't improve one's skillset and isn't any fun (remember this is all IMHO ).

Digressing: While I understand the fun of doing these daily puzzles with one-line puzzles, my concern about there not being at least the occasional more difficult puzzles is that new solvers will think that bifurcating methods, just writing 'if this then that, if that then this' or using various coloring methods is all you need to solve any puzzle, no matter the difficulty. Plus, one's solving skills aren't likely to increase beyond a certain point.

From what I remember about the Eureka forum, the conversations could become a little heated. In fact, it was so bad (in that way) that after just a few visits, I tended to avoid it like the plague.

If one looked closely, there really were only a few individuals who routinely stirred things up. The view I chose to see were the incredible minds that took the time to make sudoku solving an incredibly rewarding pastime. Even so, the most devisiveness occurred circa 2006-8, but during latter 2008 and on there was a period of very productive manual solving of the 'extremes' where most everyone had a positive demeanor and was cooperative in keeping to reasonable standardization. BTW, anyone remember The Eureka Challenge?
This is a side issue, but: I wonder if you think that there may have been advances in "Eureka notation" over the past (almost a full) decade, and that it might be worthwhile to view it as an "evolving standard" ?

It is, but all I would ask is that any changes be made based on sound Boolean logic, that they be as intuitive as possible (ie. the expression of the pattern be as easy to discern as possible) and that the changes not be made to make life easier for the writer at the expense of understanding by the reader.

This post has become somewhat long-winded, but I responded with the same seriousness that seemed intended by your question.

Regards,
Don
DonM
2013 Supporter

Posts: 475
Joined: 13 January 2008

### Re: January 3, 2015

To wipe the baby's bottom on a couple of issues:

1) Aran, I must admit I was winging it a bit when I responded to you yesterday about an Almost XYWIng as I couldn't quickly locate <this post>

Whether the proof of a pattern is considered assumptive or not is irrelevant as players are expected to know both how to recognise it and the inference(s) it derives. For any readers that don't there should be a source reference available.

2) Some methods are definitely more 'assumptive' than others. My gauge of how assumptive one is to count how the maximum number of truth states are 'open' at any time in the solution path. For an unbranched chain the start node is always open and is being compared with the current end node as it is being traversed, so the count is two. As a chain branches the count increases. Another more laborious way would be to count the ANDs that would be required if the path was formally notated using Booleans. Any internal branching in a pattern is ignored as the pattern as whole is either true or false.
David P Bird
2010 Supporter

Posts: 1009
Joined: 16 September 2008
Location: Middle England

### Re: January 3, 2015

aran wrote:Eleven
Mine would be
(v=wx)r56c2 - (w|x=yz)r89c2 - (yz=u)r7c1 - (u=r)r7c8 - (r=v)r4c8 => r4c123 <> v

wx are excluded from r89c2 every bit as much as yz from r7c1
therefore there is no need to adopt a theoretical wIx which I think confuses rather than enlightens
but if you do wish for wIx then do you not need yIz ?

This is the last time, that i will explain anything about AIC. I already regretted, that i noted, that Steve's link was wrong.

A strong link term l=r in AIC is defined by "if l is false, then r is true" or (not l) => r. This logically implies "if r is false, then l is true" or (not r) => l.
The weak link l-r is defined by "if l is true, then r is false" or l => (not r). This logically implies "if r is true, then l is false" or r => (not l)
So it is absolutely symmetric and can be written/read from left to right as well as from right to left.

Remember the definitions i wrote:
l = (w|x)r89c2 means w or x in the 2 cells
r = yzr89c2 means both y and z in the 2 cells (more commonly used in this sense)

("w or x" here of course means w or x or both)

Then (w|x=yz)r89c2 means:
If not (w or x) (<=> not w and not x) are in the cells r89c2, then (y and z) (a pair) are in r89c2
or from right to left
If not (y and z) (<=> not y or not z) in r89c2, then (w or x) in r89c2

Please mind the and's and or's.

(wx=yz)r89c2 would mean:
If not (w and x) (<=> not w or not x) are in the cells r89c2, then (y and z) (a pair) are in r89c2
This is wrong. If only one of w or x is missing in the cells, there will be no pair yz there.

If you are the meaning that (wx=yz)r89c2 implicitely means (w|x=yz)r89c2, it is wrong, if you write it from right to left:
You would have (u=yz)r7c1 - (y|z=wx)r89c2, and that is wrong (if y or z in r7c1, then not both are excluded from r89c2).

When you talk about "the reversibilty may require branching" you seem to mix this with reversing a contradiction chain.

PS: Do you remember [Edit:] Bill Richter, who was omnipresent at Eureka ?
Smile to find posts like this
eleven

Posts: 1799
Joined: 10 February 2008

### Re: January 3, 2015

Hi DonM,

Sorry it has taken me so long to reply.
Thank you for your detailed response.
I think I understand now.

From what I remember about the Eureka forum, the conversations could become a little heated. In fact, it was so bad (in that way) that after just a few visits, I tended to avoid it like the plague.

If one looked closely, there really were only a few individuals who routinely stirred things up. The view I chose to see were the incredible minds that took the time to make sudoku solving an incredibly rewarding pastime. Even so, the most devisiveness occurred circa 2006-8, but during latter 2008 and on there was a period of very productive manual solving of the 'extremes' where most everyone had a positive demeanor and was cooperative in keeping to reasonable standardization.

I'm regretting that I missed it all

Best Regards,
Blue.
blue

Posts: 611
Joined: 11 March 2013

### Re: January 3, 2015

eleven wrote:PS: Do you remember [Edit:] Bill Richter, who was omnipresent at Eureka ?
Smile to find posts like this

That was a nice trip to yesteryear. The subject matter was particularly interesting because while Nice Loop notation was still being used at the time of this post on the 'old' version of this forum, it would almost disappear from that point on as Eureka notation took over. IMO, this occurred because it was easier to notate constructs in more complex chains using AICs with Eureka notation. Incidentally, when Myth Jellies introduced AICs in 2006, he used his own notation, but dropped it in favor of Eureka notation soon after.

Ah-h, Bill Richter! He certainly added an element of eccentricity to the Eureka forum. He had both an advanced education in mathematics and an aptitude for the subject. But, unfortunately, he created a barrier to communicating his ideas to the rest of us by creating his own notation, Or Chains, and presenting all his solutions using it. Maybe he was ahead of his time, but no one picked up the use of Or Chain notation or, for the most part, tried to follow his solutions.
DonM
2013 Supporter

Posts: 475
Joined: 13 January 2008

Previous