creint wrote:denis_berthier wrote:creint wrote:If 7r4c1 is true then it leads to no 4 possible in c4, so it can be removed.
Starting at Blue point (false) and follow the manual added blue line.
Note that blue point was group 7r56c2 = 7r3c2 .., but gets reduced to a single in the end.
Note that 7r4c4 was already excluded at start of chain.
Note that . refers to 4r57c4
Another name could be something like "unit contradiction in net".
The graphic is good (though it doesn't display the continuity of the sequence of candidates), but there is no "group".
"." doesn't refer to anything but to the absence of a candidate that is neither t- nor z-. There's no need for any exotic name for it.
When you are reading the logic from left to right you don't see why 7r5c2 is not used.
I think most readers try to read step by step. A bit more details would improve the readability for most of readers, they probably don't have an user interface that can display the whips.
-For some readers it should include both candidates if they are both used in a Strong link: -7r56c2 -> 7r3c2, on its own -7r6c2 -> 7r3c2 is invalid (7r5c2 could also be true).
-For some readers include 7r4c1 -> -7r4c4, because it is an extra step the user has to do when manually drawing. It could be read as invalid because -7r7c4 -> 7r5c4 by itself is invalid (7r4c4 could also be true).
If 7r56c2 are both are excluded by rest of chain you would have given ".".
In my graphic, the link between the target and the first llc is drawn.
For the rest, the fundamental point you're still missing is,
a whip is NOT a chain of inferences. It is a static pattern, a continuous sequence of candidates, visible in a grid without any logic reasoning. The z- and t- candidates are irrelevant: they could disappear (by another rule being applied before this whip), the whip pattern would remain UNCHANGED.The whip pattern can be USED as a support for inferences and only then does one need to take the z- and t-candidates into account MOMENTARILY, for a single step along the chain. For each of these candidates, it is straightforward to check that it is linked to the target or to a previous rlc. (Such volatile links could optionally be drawn in thin dotted lines, but certainly not in any way that would imply confusion with the real whip.)
I think the main problem for some people here is getting rid of the view of a chain as a chain of inferences. I have already talked about some unwanted consequences of this old view, such as the inability to understand that a chain is a pattern like any other pattern (Subset...) or such as counting the inference steps as a measure of complexity.
creint wrote:Why are other false points shown? With more abstraction you only need true points in the right order and you can construct the logic. For example: 7r4c1 - c2n7{r3} - c2n2{r2} - c2n5{r8}.
Absurd. This would hide
one of the main properties of the chain, continuity.
This would not be more abstraction, but only destruction of the right level of abstraction. This is what Robert has done in his downgraded version of whips/braids/S-braids under the name of (anti-)tracks.
Beauty is in the eye of the beholder. Show Notre-Dame to some people, they will see a marvel of architecture. Show it to other people (AnotherName), they will only see tons of limestone that can be assembled in any way. Show it to still other people (Robert), they will want to eliminate the roof and install a pool on it instead.
What I'm asking is simple: respect my definitions; they are not arbitrary; every part of them has its reasons. If someone doesn't understand them or doesn't like them, it's not a problem for me: don't use them. But if someone hasn't even read the definitions, the only thing he may honestly do is keep his mouth shut.
(I'm not talking about you, creint; I think you understand them, in spite of some remaining confusion).
creint wrote:*Links on this forum do work but it probably should not contain [] characters.
I had used links before and I wondered why this didn't work. I have renamed it and it works. Thanks for the tip.
Also note that for including an image from the web, the tags should be [img] and not [url]. Using this would avoid having to follow links to unreliable sites.
