Seeking clarification of definition of sashimi

MCondron

I'm working on some code for finned/sashimi configurations, and I've found a couple of patterns I'm not sure if I should call "finned" or "sashimi". The term "sashimi" seems mainly to be used in a situation where a fin is keeping a pattern classified as something larger than it would be without the fin, but I have two questions:

(1) How far must a pattern "collapse" in order for "sashimi" to be used?
(2) What if the fin is stabilizing the pattern but there is still a vertex in the box with the fin?

Example for question (1): Swordfish collapses to X-wing -- is this sashimi?

rkral

MCondron · Posted: Tue Jan 09, 2007 4:31 am Post subject:

Hmm... So you're saying that a finless Franken interpretation of a configuration is favored over seeing it as a sashimi? It seems that fins are less desirable than the more complex "franken/mutant" arrangements of fish.

I do plan to implement franken fish, including with fins (and indeed have spent a lot of time on this, although it's not quite ready yet) but I want to make sure that I get right how to describe the patterns. It's a bit surprising to see a franken swordfish designation favored over calling it a sashimi swordfish, but I can see it either way.

Mainly, the question I'm facing is what my sashimi detection code should avoid. Any vertex cell in the same box seems to violate the idea of sashimi, and if that leads to labeling a configuration as a franken-something, that's OK with me.

And regarding the collapse to x-wing, I think I see now that this can be detected without too much grief. Thanks for the hint.

Mike

MCondron · Posted: Tue Jan 09, 2007 4:36 am Post subject:

I had identified the skyscraper dichotomy a while ago and just ignored it since the x-wings will lead to the same eliminations. However, since it seems to be a popular pattern with a name and all, I've added explicit code to find skyscrapers, and I've placed it to run before any of the finned fish code.

rkral · Posted: Tue Jan 09, 2007 5:19 am Post subject:

tarek · Posted: Tue Jan 09, 2007 6:50 pm Post subject:

There are several prolonged discussions over on the players forums regarding the sashimi/franken overlap...........

This started with the special form of sashimi swordfish, where 3 sashimi swordfish (the headless/0 true vertices line variation) were needed to achieve what a franken swordfish would do in 1 step.......

In these cases I do think that the franken fish would be favoured.

tarek

MCondron · Posted: Sat Jan 13, 2007 4:34 pm Post subject:

Thanks for all the help understanding this. Mainly what I'm trying to do is make sure my code achieves two goals:

1) Code should not miss anything that should be categorized as being a legitimate example of a particular configuration

2) Code should not waste time looking for patterns that are better classified as something else.

It is becoming apparent that these two goals sometimes cannot be achieved at the same time. I suppose the best thing to do is design the code for each solution method to be as broad as possible, then allow the order in which the methods are called determine which is "favored".

MCondron · Posted: Sat Jan 13, 2007 10:06 pm Post subject:

Or, perhaps another way to select the "favored" classification of a configuration would be the one that eliminates the most candidates in a single "move". The problem with that is that you have to test each method and keep track of how many eliminations each method generates, then pick the one with the most. The classification depends on the status of the "victims" as much as it does on the base/root set.

BTW -- could someone point me to a specific spot where the sashimi/franken issue is discussed on the Player's Fourm? I do spend a fair amount of time there, but the volume of posts is so large it is like trying to sip from the ocean sometimes.

MCondron · Posted: Sat Jan 13, 2007 10:17 pm Post subject:

OR...(yet another thought)...maybe the "favored" description/classification of a configuration should be the one that gives the most potential eliminations in a single step, regardless of how many of those potential eliminations are actually realized in a particular board configuration.

rkral · Posted: Sun Jan 14, 2007 2:04 am Post subject:

MCondron · Posted: Sun Jan 14, 2007 4:10 am Post subject:

Thanks, rkral. I am still in the process of digesting the material on those threads. Perhaps in another couple of weeks I will be in a position to write a summary. There's a lot of good material there, but there is also a lot of false starts, dead ends, etc. I've spent a lot of time in research in my professional life, so I understand the nature of the process, but there does come a time when a review is in order. Myth Jellies' recent post on fins is also on my reading list, but also doesn't quite address this issue.

rkral · Posted: Sun Jan 14, 2007 11:29 am Post subject:

MCondron · Posted: Sun Jan 14, 2007 3:13 pm Post subject: