An introduction to t-Go and a javascript viewer for t-Go SGF data from Little Golem Go forum

26 replies. Last post: 2017-12-30

Reply to this topic Return to forum

An introduction to t-Go and a javascript viewer for t-Go SGF data from Little Golem
  • Malcolm Schonfield ★ at 2017-06-15

    I would welcome feedback on this new article I’ve written: An introduction to toroidal Go.
    Also, I’ve made a tool for reviewing t-Go games from little Golem. It’s alpha software and provided on an “as is” basis: link.


  • ypercube ★ at 2017-06-16

    Thank you for the great article.

    Regarding sizes, what would you think of a rectangular torus board? Examples: 9x17, 5x17, 7x15, 11x19, 7x19, 5x19.

  • ypercube ★ at 2017-06-16

    I created a user tournament: Torus Torus, Torus 2017

    Two games per opponent, starts in June 20.

  • gamesorry at 2017-06-17

    I’m getting the following error when trying to register for the tourney:

    HTTP ERROR 404

    Problem accessing /ng/a/index.jsp. Reason:


    anyone seeing the same?

  • Malcolm Schonfield ★ at 2017-06-18

    Hi Gamesorry, I was able to register with no problems using Chrome on an android phone. Maybe you could try using a different browser? Also the one time I had a technical problem on little golem (it involved a triple ko!) I raised the ticket at the following URL and it got sorted out very quickly:

  • gamesorry at 2017-06-18

    Oh I just looked it up in the User Tournament page instead of the clicking on the link above, and successfully registered.

    It might have something to do with the url:



  • gamesorry at 2017-06-18

    hmm actually both work for me now. Not sure what happened

  • Sighris at 2017-06-19

    Malcolm Schonfield, very cool!  Thanks for sharing! 

    ypercube ★, Interesting idea / question. 

    Another T-Go possibility is to add a shift/twist (literally) into the mapping of the surface such that on the 5x5 T-Go game with 25 intersections named: 

    • E1 E2 E3 E4 E5
    • D1 D2 D3 D4 D5
    • C1 C2 C3 C4 C5
    • B1 B2 B3 B4 B5
    • A1 A2 A3 A4 A5

    “A1” would connect to “B5” (instead of A5); “B1”  would connect to “C5”; “C1” would connect to “D5”; and "E1would connect to “A5” (instead of E5).... 

    The above is a one unit shift/twist; but a 2 or more unit shift could also be done.  

  • ypercube ★ at 2017-06-20

    @Sighris yes, that’s another way to wrap a square in a torus.

    I think there are other way as well, to wrap a torus. Not necessarily with a rectangular shape.

    Example, with shifts both vertical and horizontal and an a shape of a 3x3 square with a square removed:

    A3 B3 C3

    A2 B2 B3

    A1 B1

    The tiling:

    C2 A1 B1 A3 B3 C3 ...

    A3 B3 C3 A2 B2 C2 ...

    A2 B2 C2 A1 B1 ...

    A1 B1 A3 B3 C3 ...

  • ypercube ★ at 2017-06-20

    Correction for the shape (tiling is correct):

    A3 B3 C3

    A2 B2 C2

    A1 B1

  • Malcolm Schonfield ★ at 2017-06-25

    @ypercube, about your question "what would you think of a rectangular torus board? Examples: 9x17, 5x17, 7x15, 11x19, 7x19, 5x19"

    Yes, these would probably be nice spaces to play Go on. I don’t think it necessary has to be odd numbers. Still, I would expect the game to be have less strategic potential with narrow boards. It could be nice to compare 7x19 (=133 points) with 11*11 (121 points).

    @Sighris and ypercube, I hadn’t thought of that. I would call these variants “twisted toroidal Go”. These also would be nice spaces to play Go on.

    Thanks both of you for your kind words about my article.

  • Sighris at 2017-07-01

    ypercube ★, Interesting.... IIRC, two squares, one smaller than the other, can also be used to map a torus and doing so requires a “twisting” {meaning ALL of BOTH the X and Y coordinates/lines pass through the center/hole of the torus (and wrap around the “outside”.  If I can find some spare/entertainment/math(geometry)-fun time I will explain that more clearly later. 
    Malcolm Schonfield, yes, it would be interesting to "compare 7x19 (=133 points) T-Go with 11*11 (121 points) T-Go! - but my intuition (which might be wrong) says that a square mapping (11x11) would be more interesting, or at least easier to comprehend for Go players use to playing on a square board, than the rectangular 7x19 (=133 points) mapping... what if we took the rectangular idea to the extreme, what would it be like to play on a 3x50 (150 points) T-GoBan?  Super-weird? It would look like... no, it would be that every stone played was a wall of stones with a 2-space hole in the wall on each side of the single stones; like this: OOXOOXOOXOOXOOXO... even though all of the stone(s)/X would be the same stone/X!  And with one more move/play (just two moves) a player could make (two) wall(s) like OXXOXXOXXOXXOXXO... spaced 149 intersections apart. That would get very weird very quickly IMO. 

  • Malcolm Schonfield ★ at 2017-12-03

    The tournament has just ended as I resigned my last game. Congratulations to gamesorry for winning! Some very interesting games were played.

    On a different topic: at the moment I do feel that the komi is bit too small. I reckon 7.5 would be better than 4.5. How do other players feel?

  • gamesorry at 2017-12-04


    Thanks! It was a close competition w. After I lost to William Fraser I thought there was no chance I could win the tournament (I think I was behind at certain points in both of our games). Every game was tough and interesting – T-Go definitely introduced some new skill set into go.

    Regarding the komi, I don’t have a strong idea about how much it should be. In this tournament my winning rates for black and white seem to be the same – but that’s too small number of samples and maybe we could summarize the overall winning rate of taking black with the current komi and come up with an idea :)

  • Malcolm Schonfield ★ at 2017-12-21

    I’ve improved the viewer, and also open sourced the code.

    link to the viewer

    code on github

    A sample of it in action:

    A review of the fighting game against Jos Dekker with more extensive comments; browsable via the SGF viewer

    Hope you like it! Comments are welcome, as always.

  • Sighris at 2017-12-25

    I like your viewer and your blog. See your other thread (here on LG Go forum) for a question I have about your game #1873384 with Jos Dekker. I would like your viewer better if it would allow us / me to try out variations AND display them correctly with the moves being “reflected / repeated” in the duplicated boarder zones... Also it would be nice if the viewer showed the coordinates letters and numbers to make it easier to discuss the moves. - but other than that, good job!!! :) 

  • Malcolm Schonfield ★ at 2017-12-25

    Thanks for your comments. For the interactive version where moves/variations can be tried out: I have some ideas about how to extend the viewer like you say; we’ll see... :)

    As for coordinates, that should be do-able too.


  • Sighris at 2017-12-26

    I see you added an option (under the 3 bars button) to add the coordinates, thanks! Now I have a question, yesterday when I looked at your review I saw a game tree with comments on the variations... I’m not seeing it now and I can’t find it... What am I doing wrong? 

  • Sighris at 2017-12-26

    ...oh wait.... I think I was looking at the wrong web-page... I just (re)found the webpage with the move variations tree < >... but now I’m getting ready to log-off the 'net (it is 1:40am here in LA)... so I’ll write more tomorrow. 

  • Sighris at 2017-12-26

    ...oh, BTW, the coordinates don’t match up with the LittleGolem coordinates because the “A” and the “1” start in the reflected area instead of the... what should I call it... the center-board? The inner square? Just look at your coordinates and the coordinates on the LittleGolem webpages/games. 

  • Malcolm Schonfield ★ at 2017-12-26

    Just thought I’d mention this SL page I created a couple of months ago, and updated just now:

    Also, I just put this piece on Edgeless Go online:

    (For those not too scared of mathematical notation/terminology.) The main content is:

    This paper on locally grid graphs (referenced in the link above).

    @hypercube: I think your example from 2017-06-20 isn’t locally grid-like – it’s still interesting though :)

  • Sighris at 2017-12-28

    oh cool!  The first image < Figure:1 > shows a version of the “twisted torus” I described (above, 2017-06-19)... BTW, I think {with 90% certanty} the twist can be done in both the X and Y axis at the same time, so that in my above example, not only could “A” connect to “B” etc. (see above) but in addition, A1 could connect to E2, A2 to E3, A3 to E4, etc. 

  • Malcolm Schonfield ★ at 2017-12-28


    I don’t see how that could work , adding a twist both vertically and horizontally. Could you add a 3 lines of wraparound to  demonstrate? 

  • Malcolm Schonfield ★ at 2017-12-29


    I think that in order to get a one-point twist in both directions we have to break a symmetry so the four corner points only have 3 neighbours, not four. 

  • Sighris at 2017-12-30

    @Malcolm, it is possible to shift in both directions (X and Y axis) and maintain having a grid mapping in which all intersections have 4 neighbors / liberties; but if a shift is made in both (the X and Y) directions a hole is created which needs to be filled.  So if you shift by 2 in both directions a 2x2 square patch is needed to fill the hole. See the image I just shared (to demonstrate) in our FaceBook Go group {I tagged you, it is here: at the bottom (at least at this time) of this thread: }.  If we were to shift (twist) the X axis 3 and the Y axis 5 it would create a 3x5 hole to fill, thus adding 15 intersections to the grid map (of the surface of the toroid)... This is what I meant by 2017-07-01 (above) "IIRC, two squares, one smaller than the other, can also be used to map a torus and doing so requires a “twisting”..." 

  • Sighris at 2017-12-30

    @Malcolm, When making a shift in both the X and Y directions (on a Toroidal Go Game surface mapping) we have at least 3 choices; 1) fill in the hole with a patch, 2) leave the hole in the surface, thus creating 4 edges around the hole where the intersections have only three liberties {just like on the edge of a normal/flat GoBan}, or 3) “stretch”/extend the connection/liberty across the “hole” but this can result in the weirdness of having a position/intersection seem to have two states {for example both empty AND occupied by White; or occupied by BOTH White and Black} but in the mathematical modeling there would really be no weirdness, but it would probably be confusing to many (most?) Go Players. ... For images of these boards / playing-surfaces see the FaceBook Go Group discussion I mentioned in my previous post (15 hours ago / above). 

Return to forum

Reply to this topic

Include game board: [game;id:123456] or [game;id:123456;move:20] or [game;id:123456;move:20;title:some text]