edge template puzzle

I am trying to learn hex. I have been working through various edge templates that I saw on the internet at Hex Templates [drking.org.uk] . The first one that gave me some trouble was this one. Black to play and win!

link to trmph [www.trmph.com]

This (modulo any copying errors on my part)  is template a4-15 from Dr King's site, with one white move played. Black should hence still be able to win if it's black's move. I would imagine that this is easy for experts but I found this very hard. Those of you that can do this one and want more could also consider what black should do if white had played e11 instead of f9 ( link to trmph).

for e11

for f9

Well done Arek! For f9 your move is the only move on the right that works (you can make the ladder on the left first, but eventually you have to play the move on the right that you suggested). That is the sort of move that I'm just not finding at the minute; this is the reason I'm working through these templates, to learn this sort of stuff.

According to the very bottom of this page on hexwiki, the move I was missing is perhaps called "Tom's move".

Yeah, it is. This edge template is also strongly related.

Yes, Tom's move is a good trick to know. I was able to solve these puzzles in my head, because I was already familiar with the trick.

By the way, you might enjoy this puzzle: black to move and win. It is based on an interesting variant of Tom's move.

Also, the hexwiki page you linked to contains a mistake in the "third row intrusion" section. The suggested move for black doesn't work. A good exercise is to find the counter for white. Another good exercise is to find the correct move for black (the original template is valid).

shalev -- you're right! I verified by computer that the suggested move loses (and that the template is valid and minimal). See also

Intrusion_on_the_3rd_row [hexwiki]

which probably needs to be completely rewritten (the correct counter does not even seem to be covered, and there are also errors with colours of stones in some places e.g. in the second diagram). Currently it is suggesting that the suggested response to a certain third row intrusion is another play on the third row, but maybe the wiki will be edited and fixed up at some point.

I replaced the losing move on hexwiki with one of the three winning moves for black and flagged that the page devoted to the 3rd row intrusion was incorrect.

If you're verifying these using a computer, is it possible to verify my proposed 7th row template? Note that this template is missing 4 columns on the left, because trmph doesn't go to boards larger than 19x19 (and this needs 23). The template is supposed to be symmetric, though.

I don't know :-( I had seen your 7th row template but the solver I have access to, with its default settings, will only play on boards up to 19x19 so I'd have to delve deeper into what it's doing to see if this is possible, and this is hard because the documentation is almost non-existent, so I am not optimistic. I should say also that I did not write any of this code; I am just running it. I am also not optimistic about the time it will take -- it might take a very long time to verify such a template, and getting the code to run on a machine is hard work and I've only done it on a local machine at home rather than a big fancy compute cluster. It might take months to verify as it stands. There will be something like 20-30 moves in the "mustplay" zone and if people were serious about this then for each one a human could suggest the correct reply and the machine could start there, but if each of those runs takes a couple of days then again we're talking about a couple of months, and this is assuming that someone can analyse on a board bigger than 19x19 and I don't know if this is possible.

I read a lot about high row templates recently; I re-read the 6th row template thread that I started in 2008 or so for example (sixth row template thread) and tried to verify some of the assertions in there. Not everything is correct. For example Art Duval made an assertion about a one stone 4th row template on the short diagonal and it seems to me to be (just) false -- Tom's move has a line involving one stone on the 5th row. I now suspect that there might be infinitely many 4th row templates, each involving just one free hex on the 5th row (but anywhere you like, basically). Does that sound right to you? I didn't check it carefully yet.

javerberg and philip both actually *wrote* hex code in the past, I believe; however I think they both seem to have left this site (javerberg also let the links to pictures on his website rot, which makes the 6x6 thread much harder to read). I don't know who javerberg is IRL but Philip has written papers about solving hex and you could probably find and get in touch with him (he was at Alberta a few years ago working in the computer science department).