Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
Jan 14, 2011, 12:35 AM
Q13 and P13 apparently lose to 5. ...L10 (according to mmai12). The AI picks a low block for move 6 as best and a reply of 6. ...L11. A high block does appear to be a losing proposition as well (I checked it with the ai). Also, keep in mind that with the VCT setting, the computer isn't limited to searching 6 moves out- provided it can find a series of threats by the 6th move, it will continue analysis beyond the 6th move (which is why it takes so much longer). 5. N13 is what I'm looking at currently, since the computer plays 6. ...J12 in response- often a sign that it doesn't have a clear win in sight. I haven't completed the AI analysis of that one though, and don't know whether it would choose a different 5th if given the choice.
Message was edited by: watsu at Jan 13, 2011 6:36 PM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
Jan 14, 2011, 2:53 PM
Hey Zoey, I just noticed your previous comment:
"thanks for the links. i read it a bit. i just have a issue with the R3U2 kinda coordinates. and was too lazy to translate 200 of these into viewable form. i think it would be very nice to one day translate all that text into clear text, and to make all the moves into viewable boards. it would really come to life if we did that. to take it a step further would be to make a video of it narrating the story of it, along with pictures.. "
I agree, translating coordinate systems is a bit of a chore. Fortunately, for a bulk of these old games this has already been done!
At Mark Mammel's 5-in-a-row site, where you can download his WPente program which includes the mm_ai that is used here, you can ALSO download a large series of completed games from the '83 world championship:
there is a link that says: "Download 131 Pente games..."
This is a direct link to the zip file which contains the save files for these games. Since these files can be immediately opened within Mark's program, the text file associated with it is already translated into modern coordinates. When I was learning the game I would read the analysis in the newsletter and scroll through the moves within the WPente program to see what was going on -- extremely valuable resource while learning the game! It's even interesting to see when certain blunders were made under time pressure, etc...
I remember that it looked like a few of the games are missing and a couple of them might have some mistakes -- either they were originally notated incorrectly during the tournament or they have been incorrectly translated to the new coordinates -- but nearly all of them are correct.
I believe these saved games correspond with the large set of links at the bottom, called cpente01 and so on. A few of the other smaller samples of newsletters on that site, such as the ones containing the analysis of the Hammer opening, are not included in those save files unfortunately. It would be cool if someone would do the work to get those made and then perhaps get in touch with Mark so that he can link to those as well so that they are all downloadable from one location at the same place that you can download the WPente program to view the files...
In addition, I remember reading on these forums a couple of different times that some players (and members here) still possess hard copies of some additional newsletters from the 80s that were not made available on that renju site (and so are not available for download anywhere that I know of). It would be EXTREMELY excellent for the community if someone would do a little work to scan these and post them online and give us a link to them so that these important documents will not be lost forever.
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
Jan 14, 2011, 4:44 PM
Ok, I might have been too broad when I was speaking of all the possible moves to respond to the hammer opening. Most of those choices would be poor and I think we should be able to use some simple human analysis to eliminate those. So, I'll take a crack at narrowing it down:
First, as I've explained earlier in the thread, the primary threat for black after the 4th move is 5 ... L10, often continuing with 6 ... L11 but other options for continuation exist also, such as 6 ... O7.
So, assuming that white completely wastes the 5th move (say, by playing 5. A1), here are a couple of examples of how black will win:
Or:
Or:
So, when looking at each 5th move for white, we should first quickly look to see if it has any chance to stop this threat. If not, then black wins with 5 ... L10 for that option.
Using this quick logic, I'll try to eliminate at least half of the 5th moves here shortly...
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
Jan 14, 2011, 4:48 PM
Ok, after a very quick look through the 5th moves on the 13 line, I could only eliminate 4 of them!
Eliminated: G13,H13,P13,Q13
This is based on having no impact on the example line I just posted using 5 ... L10, which wins as shown.
All of the other moves cause at least some complication with that example line and would cause at least some minor deviation, so I just skipped over those for now. I'm betting that L10 probably beats all of those moves, but the line would be different.
Next up, a first pass through the 12 line...
-------------------------
UPDATE:
Amazingly, in only one case could the example line be played straight through against moves on the 12 line:
ELIMINATED: Q12
-------------------------
UPDATE:
Several of the moves on the 11 line were worthless:
ELIMINATED: G11,H11,O11,P11,Q11
-------------------------
UPDATE:
Next is the 10 line:
ELIMINATED: G10,H10,P10,Q10
-------------------------
UPDATE:
Next!
ELIMINATED: G9,H9,J9,Q9
--------------------------
UPDATE:
The 8 line. I'm including at least one example where white plays directly into a capture so the capture is made and then the line plays out exactly.
ELIMINATED: G8,H8,J8,P8,Q8
--------------------------
UPDATE:
7 line. Again, I'm including cases of playing directly into a capture, taking the capture, and then following the line as shown.
ELIMINATED: G7,H7,J7,L7,N7,Q7
--------------------------
UPDATE:
Ok, the 6 line. Moves down here are all generally pretty bad I think.
ELIMINATED: G6,H6,J6,L6,M6,N6,P6,Q6
---------------------------
UPDATE:
Well, last (and least) is the 5 line. Terrible moves. Only O5 causes the slightest of variations from the sample line, so it gets left off the list for now.
ELIMINATED: G5,H5,J5,K5,L5,M5,N5,P5,Q5
----------------------------
Ok, that wraps up my first pass at all of the moves to eliminate the most obvious losing moves first. Note that K7 and K6 are sort of special cases so they should be looked at later.
I've eliminated these moves based on the fact that they have absolutely no impact in steering P2 away from its primary winning line that I showed above -- in a few cases white played directly into a capture, which was taken, and then the rest of the line played out as shown. These confirm a winning response for black of 5 ... L10 for each of these choices for white's 5th move -- and they do so as straightforward as possible.
The next step would be to make a second pass through the rest of the remaining move choices and look at L10 again as a possible winning move. In these cases, the line shown cannot be played out exactly, but in many cases there is a simple alternative that presents itself based upon the placement of white's 5th stone. Once these minor deviations are explored, these moves too can be eliminated.
It will be interesting to see how these sorts of findings match up with any computer or ai analysis performed by other players.
So, to recap, more than half of the moves are eliminated at first glance since they lose easily to 5 ... L10 with the line shown.
There are 36 moves remaining to be looked at further. I would predict that at least half of these still lose to 5 ... L10 without TOO much complication.
Message was edited by: up2ng at Jan 14, 2011 11:26 AM
Message was edited by: up2ng at Jan 14, 2011 1:17 PM
Message was edited by: up2ng at Jan 14, 2011 1:34 PM
Message was edited by: up2ng at Jan 14, 2011 1:42 PM
Message was edited by: up2ng at Jan 14, 2011 2:30 PM
Message was edited by: up2ng at Jan 14, 2011 2:47 PM
Message was edited by: up2ng at Jan 14, 2011 3:04 PM
Message was edited by: up2ng at Jan 14, 2011 3:18 PM
Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
Jan 14, 2011, 8:12 PM
Thanks for the elimination assistance. As a note- for 5.N13 the mmai P2 response is P9. I haven't completed the running of that line yet, but if human analysis can add it as another postential P2 winning threat, it might help eliminate a few more moves from consideration.
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
Jan 16, 2011, 8:47 PM
Here's what appears to be a secondary winning line threat based on the mmai play against 5. N13. I'll submit it in the more generalised form of the wasted 5. A1, as up2ng has done. I'm not sure yet how many (if any) this line might eliminate from up2ng's list, but it does seem like a strong alternate winning threat line to me. There's also the alternate winning threat of 10. ...N8. Also, the three lines up2ng mentioned upthread were exactly what I was seeing from the mmai against the Q13 and P13 5ths.
ETA: it only seems to help out on the 12 and 13 lines, from what I can tell at first look, but it does appear to win against O13, N13, L12, K12, J12, H12, and G12 without complications.
ETA: this fifth/sixth combo for P2 is also what the AI uses against both M13 and L13. There are some variations in the lines, particularly L13, but the combination appears to lead to a win. That's as far as I've made it so far.
ETA: the line below appears to work for M13, P12 and Q12 (and probably for other moves as well, though I haven't looked yet)
10. P7 looks odd, but P2 has a strong winning threat with 10. ... P7
ETA: Minor correction to add the number 13 to the sentence "it only seems to help out on the 12 and 13 lines". That sense was already present in the list of moves, but I omitted the number 13 when I originally typed it.
Message was edited by: watsu at Jan 18, 2011 9:10 PM
Message was edited by: watsu at Jun 2, 2020, 8:20 AM
Message was edited by: watsu at Jun 6, 2022, 10:58 AM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Solve the Hammer Opening for P1, & become a Pente Legend.
Posted:
May 31, 2020, 7:54 AM
Earlier, I got asked whether or not the Hammer opening had been exhaustively shown to be a P1 loss. Not too my knowledge, but here's where the community had gotten to as of a bit less than a decade ago. Maybe someone else will pick up the torch? ETA: I've added another board above which appears to work for a couple of new positions on the 12 line. Combined with moves which are already covered in the database (only the top 20 move results show up, but if you actually input the positions you'll find more games; be sure to check the db for K10 K9 etc. as well) this probably will wrap it up for the hammer, but be sure to try something new, undiscovered and magical not covered here on one or more of the top players, just in case...
Message was edited by: watsu at Jun 2, 2020, 3:00 PM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat