The New Sudoku Players' Forum

[Mauriès Robert]

Hi Denis,
I believe I have already asked you this question, but I need to understand it better.
How do you determine that a candidate is a target even before building the chain (biv-chain, whip, braid etc.) that will effectively eliminate that target?
Cordialy
Robert

[denis_berthier]

Hi Denis,
I believe I have already asked you this question, but I need to understand it better.
How do you determine that a candidate is a target even before building the chain (biv-chain, whip, braid etc.) that will effectively eliminate that target?

How do you determine that 3 cells in a row will be part of a Triplet before you've found a full Triplet?

[Mauriès Robert]

Hi Denis,

Hi Denis,
I believe I have already asked you this question, but I need to understand it better.
How do you determine that a candidate is a target even before building the chain (biv-chain, whip, braid etc.) that will effectively eliminate that target?

How do you determine that 3 cells in a row will be part of a Triplet before you've found a full Triplet?

This type of answer doesn't help me much.
Or I deduce that for an arbitrarily chosen candidate you try a channel, if it leads to contradiction it was a target, if it doesn't lead to anything you always arbitrarily choose another candidate, etc...
Robert

[denis_berthier]

Hi Denis,

Hi Denis,
I believe I have already asked you this question, but I need to understand it better.
How do you determine that a candidate is a target even before building the chain (biv-chain, whip, braid etc.) that will effectively eliminate that target?

How do you determine that 3 cells in a row will be part of a Triplet before you've found a full Triplet?

This type of answer doesn't help me much.
Or I deduce that for an arbitrarily chosen candidate you try a channel, if it leads to contradiction it was a target, if it doesn't lead to anything you always arbitrarily choose another candidate, etc...
Robert

Randomly or by intuition if you're a human player.
Exactly as for Triplets.

[Mauriès Robert]

Hi Denis,
I believe I have already asked you this question, but I need to understand it better.
How do you determine that a candidate is a target even before building the chain (biv-chain, whip, braid etc.) that will effectively eliminate that target?

Randomly or by intuition if you're a human player.
Exactly as for Triplets.

It seems to me that a Triplet, like an X-wing or a swordfich, are patterns visible to the naked eye. I doubt that this is the case for a braid [13] for example.
How does the programmed machine (solver) choose a target then?
It tries all the candidates of the puzzle and for each candidate it looks if a pattern (biv-chain, whip, braid, etc...) is possible?
If this is the case, you can well imagine that this is impossible for a resolution by hand without resorting to chance.
It seems to me then that it is more productive to focus on pairs (candidates or a group of closely related candidates) to look for a target.

PS : I answered your questions in Robert's puzzles 2020-10-20 feed
Robert

[denis_berthier]

Hi Denis,
I believe I have already asked you this question, but I need to understand it better.
How do you determine that a candidate is a target even before building the chain (biv-chain, whip, braid etc.) that will effectively eliminate that target?

Randomly or by intuition if you're a human player.
Exactly as for Triplets.

It seems to me that a Triplet, like an X-wing or a swordfich, are patterns visible to the naked eye. I doubt that this is the case for a braid [13] for example.

I'm sure you know the comparison is absurd. A Triplet should be compared to a whip[3] or braid[3].
And most puzzles don't need longer whips.

[Mauriès Robert]

I'm sure you know the comparison is absurd. A Triplet should be compared to a whip[3] or braid[3].
And most puzzles don't need longer whips.

So can you answer my question, which I will repeat:
How does the programmed machine (solver) choose a target then?
It tries all the candidates of the puzzle and for each candidate it looks if a pattern (biv-chain, whip, braid, etc...) is possible?
If this is the case, you can well imagine that this is impossible for a resolution by hand without resorting to chance.
It seems to me then that it is more productive to focus on pairs (candidates or a group of closely related candidates) to look for a target.
Thank you in advance.
Robert

[denis_berthier]

So can you answer my question, which I will repeat:
How does the programmed machine (solver) choose a target then?
It tries all the candidates of the puzzle and for each candidate it looks if a pattern (biv-chain, whip, braid, etc...) is possible?
If this is the case, you can well imagine that this is impossible for a resolution by hand without resorting to chance.
It seems to me then that it is more productive to focus on pairs (candidates or a group of closely related candidates) to look for a target.

Targets are not tried randomly: partial-whips[1] are as obvious to find as whips[1], the simplest pattern after Singles and as obvious to find as bivalue cells.
Partial-whips[1] are the starting point for longer partial-whips, some of which may produce full whips. It's as simple as that.
And it's much easier to follow a single stream of reasoning than two or more in your approach.

[Mauriès Robert]

Targets are not tried randomly: partial-whips[1] are as obvious to find as whips[1], the simplest pattern after Singles and as obvious to find as bivalue cells.
Partial-whips[1] are the starting point for longer partial-whips, some of which may produce full whips. It's as simple as that.
And it's much easier to follow a single stream of reasoning than two or more in your approach.

Yes, the whips [1] are obvious - they correspond to alignments. I'll admit that whip [2], whip [3], see whip [4] are easy to see, but in your somewhat difficult puzzles resolutions you go beyond that and I find it hard to believe that these long whips can be found without making unsuccessful attempts.
Excuse me, but one question calls for another, I discover that there are partial whips. What are partial whips?
Robert

[denis_berthier]

Targets are not tried randomly: partial-whips[1] are as obvious to find as whips[1], the simplest pattern after Singles and as obvious to find as bivalue cells.
Partial-whips[1] are the starting point for longer partial-whips, some of which may produce full whips. It's as simple as that.
And it's much easier to follow a single stream of reasoning than two or more in your approach.

Yes, the whips [1] are obvious - they correspond to alignments. I'll admit that whip [2], whip [3], see whip [4] are easy to see, but in your somewhat difficult puzzles resolutions you go beyond that and I find it hard to believe that these long whips can be found without making unsuccessful attempts.

94+ % of the puzzles can be solved in W4.
99,8+ % in W7
(unbiased stats - see my paper on this topic)
99% of the puzzles proposed in the newspapers can be solved in W1.
So, what are we talking about? An infinitesimal % of puzzles, interesting for 2 or 3 handfuls of people in the world.

A partial-whip is an unfinished whip, with a pending right-linking candidate. A whip is finished when there is no rlc for its last csp-var. You should have a look at the user manual, where clear graphics are given (also for t-whips). It can be downloaded from GitHub or Researchgate