I couldn't solve Brainbashers July the 8th 'very hard' sudoku without resorting to 'super hard' techniques, and that, 3 or 4 times.
I suppose I overlooked something, but can anybody remember it being especially hard, or even harder than usual

On 14/07/2022 02:51, gerson wrote:
> I couldn't solve Brainbashers July the 8th 'very hard' sudoku without resorting to 'super hard'
> techniques, and that, 3 or 4 times.
> I suppose I overlooked something, but can anybody remember it being especially hard, or even harder
> than usual

I just had a go at it. I was using the auto-pencil marks and highlighting features and it took me 7
minutes. So for me that elevates it out of the "came out easily" category (under 3 minutes), but it
didn't seem especially hard. Normally I have to follow "hooks(?)" around the grid to gradually
eliminate numbers, which I'm guessing is an advanced technique, but I didn't need that at all for
this puzzle. (But I'm not really clear on what constitutes a super-hard technique.)

Most days I just do the "very hard" puzzle, because it takes me about the amount of time I have to
spend on it (5-10 mins), and if it takes <3 mins I'm dissatisfied so I do the super-hard one.

Mike.

"Mike Terry" wrote in message news:PYGdnes3fZABp03_nZ2dnUU7-aPNnZ2d@brightview.co.uk...

On 14/07/2022 02:51, gerson wrote:
> I couldn't solve Brainbashers July the 8th 'very hard' sudoku without resorting to 'super hard'
> techniques, and that, 3 or 4 times.
> I suppose I overlooked something, but can anybody remember it being especially hard, or even harder
> than usual

I just had a go at it. I was using the auto-pencil marks and highlighting features and it took me 7
minutes. So for me that elevates it out of the "came out easily" category (under 3 minutes), but it
didn't seem especially hard. Normally I have to follow "hooks(?)" around the grid to gradually
eliminate numbers, which I'm guessing is an advanced technique, but I didn't need that at all for
this puzzle. (But I'm not really clear on what constitutes a super-hard technique.)

Most days I just do the "very hard" puzzle, because it takes me about the amount of time I have to
spend on it (5-10 mins), and if it takes <3 mins I'm dissatisfied so I do the super-hard one.

Mike.

Thanks
By 'super hard techniques' I was referring to the last four things on the list you get when you
click on his 'Sudoku help page' in the Notes.
Anyhow, I'd like to know what "hooks(?)" "round the grid' is. Is it what I call chaining, such as,
- if this one's a 3 then that one's an 8 which makes that other one a 2 and so on - ?

On 17/07/2022 10:13, gerson wrote:
>
>
> "Mike Terry"  wrote in message news:PYGdnes3fZABp03_nZ2dnUU7-aPNnZ2d@brightview.co.uk...
> On 14/07/2022 02:51, gerson wrote:
>> I couldn't solve Brainbashers July the 8th 'very hard' sudoku without resorting to 'super hard'
>> techniques, and that, 3 or 4 times.
>> I suppose I overlooked something, but can anybody remember it being especially hard, or even
>> harder than usual
>
> I just had a go at it.  I was using the auto-pencil marks and highlighting features and it took me 7
> minutes.  So for me that elevates it out of the "came out easily" category (under 3 minutes), but it
> didn't seem especially hard.  Normally I have to follow "hooks(?)" around the grid to gradually
> eliminate numbers, which I'm guessing is an advanced technique, but I didn't need that at all for
> this puzzle.  (But I'm not really clear on what constitutes a super-hard technique.)
>
> Most days I just do the "very hard" puzzle, because it takes me about the amount of time I have to
> spend on it (5-10 mins), and if it takes <3 mins I'm dissatisfied so I do the super-hard one.
>
> Mike.
>
> Thanks
> By 'super hard techniques' I was referring to the last four things on the list you get when you
> click on his 'Sudoku help page' in the Notes. Anyhow, I'd like to know what "hooks(?)" "round the
> grid' is. Is it what I call chaining, such as,
> - if this one's a 3 then that one's an 8 which makes that other one a 2 and so on - ?

I don't instantly recognise my "hooks" in the list of techniques on the help page. It's rather like
the "remote pairs", except it's much more general than that. ("remote pairs" can be considered a
simple case of my technique.) If you take the remote pairs examples and just focus on ONE of the
pairs of digits, that's more like it.

Basically, the approach is to highlight occurences of a chosen digit, and identify two of them such
that one or other of those two must be correct. Then any common buddies of the two cannot be that
digit. In practice we start with one of the digits, and say "if this square ISN'T the chosen digit,
then [any valid deductions working around the grid until we identify some other square that MUST be
the chosen digit]". Then we eliminate common buddies. (So there is a strong element of focussed
"chaining" in your sense, but there is a specific goal we are working towards - not just random "I
wonder what happens if this square is a 6".

Anyway, looking at the list of techniques, a number of them are just special cases of my technique.
E.g. X-Wing, Remote Pairs, XY-Wing, XYZ-Wing are all variations of my technique, and I'd spot all of
them in a Brainbashers sudoku through that mechanism if not more directly. (X-Wing is just a
cut-down Swordfish, which in turn is a cut-down generalised pattern applying to any number of
columns, not just 2 or 3! I look for the general pattern which covers X-Wing, but I've never
thought to learn "XYZ-Wing" as a pattern as I'd (probably) find it with the hook(?) pattern anyway,
so it wouldn't be any extra help...)

Hmmm, I searched for "hook" and sudoku etc. and found a link which does explain the technique I mean
but only in a simplified form.

<https://www.sudokudragon.com/guidehook.htm>

[click the "stop" button on the page to turn off the annoying auto-play, then use left and right
buttons to step through the demo]

My explanation (referring to the linked web page diagrams) would be:
- We are focussing on 6's (On BrainBashers I would highlight all 6's)
- [step through demo to step 6 of 22]
- the highlighted square Dc is our starting point
- Dc can be 6, but IF IT'S NOT, then...
- it's the 8, so Ac is 7, and Ca is 6. Bingo! [###]
- Put another way - EITHER Dc or Ca is 6. (This is the important point.)
- So any common buddies of those two squares can't be 6
In the given scenario, that means Da and Ea can't be 6.

The demo says "we must have 3 squares..." but that's not really required at all - what we must have
is the starting square Dc which may be the 6, and the [###] chain of deductions above (any length)
leading to another square also having to 6 (if the starting square is assumed not to be 6). Then we
can apply the common buddy conclusion for those two cells. So it's all quite flexible, but also
needs some experience/practice to work out /fruitful/ applications of the rule!

I used the rule a couple of times in todays "very hard" puzzle, so if you like I could try to
reproduce the positions where I used it to explain further. (But being able to highlight the chosen
number occurences is quite important in practice as it helps direct our goal, hence which deduction
links might be more fruitful. Highlighting will be missing in any diagram I manage to post!)

Regards,
Mike.

On 17/07/2022 16:38, Mike Terry wrote:
> On 17/07/2022 10:13, gerson wrote:
>>
>>
>> "Mike Terry"  wrote in message news:PYGdnes3fZABp03_nZ2dnUU7-aPNnZ2d@brightview.co.uk...
>> On 14/07/2022 02:51, gerson wrote:
>>> I couldn't solve Brainbashers July the 8th 'very hard' sudoku without resorting to 'super hard'
>>> techniques, and that, 3 or 4 times.
>>> I suppose I overlooked something, but can anybody remember it being especially hard, or even
>>> harder than usual
>>
>> I just had a go at it.  I was using the auto-pencil marks and highlighting features and it took me
>> 7 minutes.  So for me that elevates it out of the "came out easily" category (under 3 minutes),
>> but it didn't seem especially hard.  Normally I have to follow "hooks(?)" around the grid to
>> gradually eliminate numbers, which I'm guessing is an advanced technique, but I didn't need that
>> at all for this puzzle.  (But I'm not really clear on what constitutes a super-hard technique.)
>>
>> Most days I just do the "very hard" puzzle, because it takes me about the amount of time I have to
>> spend on it (5-10 mins), and if it takes <3 mins I'm dissatisfied so I do the super-hard one.
>>
>> Mike.
>>
>> Thanks
>> By 'super hard techniques' I was referring to the last four things on the list you get when you
>> click on his 'Sudoku help page' in the Notes. Anyhow, I'd like to know what "hooks(?)" "round the
>> grid' is. Is it what I call chaining, such as,
>> - if this one's a 3 then that one's an 8 which makes that other one a 2 and so on - ?
>
> I don't instantly recognise my "hooks" in the list of techniques on the help page.  It's rather like
> the "remote pairs", except it's much more general than that.  ("remote pairs" can be considered a
> simple case of my technique.)  If you take the remote pairs examples and just focus on ONE of the
> pairs of digits, that's more like it.

Correction - I meant "...and just focus on ONE of the DIGITS in the pairs..." I.e. if all the pairs
are (6,8), we would only be focusing on the 6's. (The 8's could be relevent as part of the chain of
deductions, as in "remote pair" examples, but other deductions typically occur as well...)

>
> Basically, the approach is to highlight occurences of a chosen digit, and identify two of them such
> that one or other of those two must be correct.  Then any common buddies of the two cannot be that
> digit.  In practice we start with one of the digits, and say "if this square ISN'T the chosen digit,
> then [any valid deductions working around the grid until we identify some other square that MUST be
> the chosen digit]".  Then we eliminate common buddies.  (So there is a strong element of focussed
> "chaining" in your sense, but there is a specific goal we are working towards - not just random "I
> wonder what happens if this square is a 6".
>
> Anyway, looking at the list of techniques, a number of them are just special cases of my technique.
> E.g. X-Wing, Remote Pairs, XY-Wing, XYZ-Wing are all variations of my technique, and I'd spot all of
> them in a Brainbashers sudoku through that mechanism if not more directly.  (X-Wing is just a
> cut-down Swordfish, which in turn is a cut-down generalised pattern applying to any number of
> columns, not just 2 or 3!  I look for the general pattern which covers X-Wing, but I've never
> thought to learn "XYZ-Wing" as a pattern as I'd (probably) find it with the hook(?) pattern anyway,
> so it wouldn't be any extra help...)
>
> Hmmm, I searched for "hook" and sudoku etc. and found a link which does explain the technique I mean
> but only in a simplified form.
>
>   <https://www.sudokudragon.com/guidehook.htm>
>
> [click the "stop" button on the page to turn off the annoying auto-play, then use left and right
> buttons to step through the demo]
>
> My explanation (referring to the linked web page diagrams) would be:
> -  We are focussing on 6's  (On BrainBashers I would highlight all 6's)
> -  [step through demo to step 6 of 22]
> -  the highlighted square Dc is our starting point
> -  Dc can be 6, but IF IT'S NOT, then...
> -  it's the 8, so Ac is 7, and Ca is 6.  Bingo!     [###]
> -  Put another way - EITHER Dc or Ca is 6.  (This is the important point.)
> -  So any common buddies of those two squares can't be 6
>    In the given scenario, that means Da and Ea can't be 6.
>
> The demo says "we must have 3 squares..." but that's not really required at all - what we must have
> is the starting square Dc which may be the 6, and the [###] chain of deductions above (any length)
> leading to another square also having to 6 (if the starting square is assumed not to be 6).  Then we
> can apply the common buddy conclusion for those two cells.  So it's all quite flexible, but also
> needs some experience/practice to work out /fruitful/ applications of the rule!
>
> I used the rule a couple of times in todays "very hard" puzzle, so if you like I could try to
> reproduce the positions where I used it to explain further.  (But being able to highlight the chosen
> number occurences is quite important in practice as it helps direct our goal, hence which deduction
> links might be more fruitful.  Highlighting will be missing in any diagram I manage to post!)
>
> Regards,
> Mike.
>

On 14/07/2022 2:51 am, gerson wrote:
> I couldn't solve Brainbashers July the 8th 'very hard' sudoku
> without resorting to 'super hard' techniques, and that, 3 or 4
> times.
> I suppose I overlooked something, but can anybody remember it
> being especially hard, or even harder than usual

Damn you, I've now tried it six times. No joy.

--
Richard Heathfield
Email: rjh at cpax dot org dot uk
"Usenet is a strange place" - dmr 29 July 1999
Sig line 4 vacant - apply within

On 17/07/2022 17:16, Richard Heathfield wrote:
> On 14/07/2022 2:51 am, gerson wrote:
>> I couldn't solve Brainbashers July the 8th 'very hard' sudoku without resorting to 'super hard'
>> techniques, and that, 3 or 4 times.
>> I suppose I overlooked something, but can anybody remember it being especially hard, or even
>> harder than usual
>
> Damn you, I've now tried it six times. No joy.
>
>

The BrainBashers sudokus are machine-generated somehow, and I can't imagine that the generation
processes doesn't perform a solve-check using just the allowed solution techniques, which will all
be programmed into the generator as runtime selectable options.

So I'd be very surprised if the puzzle /wasn't/ solvable using only the listed "very hard" techniques.

Well, it's not impossible but that would indicate a basic software bug in the generator - and
wouldn't that have been spotted before now, e.g. in unit testing?

So probably something has just been overlooked as G. suspects. What to do...?

G. could post the puzzle here, at the point where he gets stuck and has to use a more advanced
technique, then others might just spot what was overlooked? Or the puzzle at the "stuck" point
could be fed into one of the many online step-by-step solvers that allow you to specify the list of
techniques to use. (Sounds like an obvious way forward, but these online solvers have their own
names for the techniques allowed - some names are commonly recognised everywhere, but there's no
standard registry for Sudoko techniques! So all the Brainbasher techniques will need to be mapped
to the right online-solver techniques, which might be tricky in places...)

BTW, I've been looking at my times for recent puzzles, and it seems I took a fair bit longer on 14th
and 17th July (a little under 20 mins). I don't know if that means those puzzles were genuinely
harder, because in the end it can just come down to the luck of the order you try different things...

Mike.

On 17/07/2022 5:57 pm, Mike Terry wrote:
> On 17/07/2022 17:16, Richard Heathfield wrote:
>> On 14/07/2022 2:51 am, gerson wrote:
>>> I couldn't solve Brainbashers July the 8th 'very hard' sudoku
>>> without resorting to 'super hard' techniques, and that, 3 or 4
>>> times.
>>> I suppose I overlooked something, but can anybody remember it
>>> being especially hard, or even harder than usual
>>
>> Damn you, I've now tried it six times. No joy.
>>
>>
>
> The BrainBashers sudokus are machine-generated somehow, and I
> can't imagine that the generation processes doesn't perform a
> solve-check using just the allowed solution techniques, which
> will all be programmed into the generator as runtime selectable
> options.

Be that as it may, I'm going to bloody well do this.

Solution follows, from which you will learn NOTHING. :-(
Solution follows, from which you will learn NOTHING. :-
Solution follows, from which you will learn NOTHING. :
Solution follows, from which you will learn NOTHING.
Solution follows, from which you will learn NOTHING.
Solution follows, from which you will learn NOTHING
Solution follows, from which you will learn NOTHIN
Solution follows, from which you will learn NOTHI
Solution follows, from which you will learn NOTH
Solution follows, from which you will learn NOT
Solution follows, from which you will learn NO
Solution follows, from which you will learn N
Solution follows, from which you will learn
Solution follows, from which you will learn
Solution follows, from which you will lear
Solution follows, from which you will lea
Solution follows, from which you will le
Solution follows, from which you will l
Solution follows, from which you will
Solution follows, from which you will
Solution follows, from which you wil
Solution follows, from which you wi
Solution follows, from which you w
Solution follows, from which you
Solution follows, from which you
Solution follows, from which yo
Solution follows, from which y
Solution follows, from which
Solution follows, from which
Solution follows, from whic
Solution follows, from whi
Solution follows, from wh
Solution follows, from w
Solution follows, from
Solution follows, from
Solution follows, fro
Solution follows, fr
Solution follows, f
Solution follows,
Solution follows,
Solution follows
Solution follow
Solution follo
Solution foll
Solution fol
Solution fo
Solution f
Solution
Solution
Solutio
Soluti
Solut
Solu
Sol
So
S

Block 1 top left is 2, 6, or 9.

2 leads to a contradiction about a month down the line.
So does 6. So top left is a 9.

The next cell along is 2, 4, or 6.

2 again leads to a contradiction. Put a 4 there, everything
works, and elegance be damned.

--
Richard Heathfield
Email: rjh at cpax dot org dot uk
"Usenet is a strange place" - dmr 29 July 1999
Sig line 4 vacant - apply within