Can anyone point me to an online database of Steiner triple systems?

My Google-fu is only getting me to descriptions of the few smallest ones, mostly Google book scans (which are rather useless to process using a computer...)


The first answer is not a database of the Steiner triple systems, but rather how many there are up to isomorphism. It is known that they there are 2 of order 13, 80 of order 15, and 11084874829 of order 19. The last number was computed by Kaski and Ostergard, and I suppose that the best approach is to ask them for their data. It's more or less the end of the story, because it is easy to make larger Steiner triple systems, but impossible to compile them into a complete database.

Actually this paper describes a compressed 39-gigabyte file with the Steiner triple systems of order 19, and says that it is available by e-mail request from three of the authors (including Kaski and Ostergard).

It looks like a number of people have the Steiner triple systems of order 15, but I didn't find a paper that simply lists them.

Update 1: It seems that everyone works from the paper "Small Steiner triple systems and their properties", by Mathon, Phelps, and Rosa. This paper is basically an encyclopedia of the 80 Steiner triple systems of order 15 and many of their properties. It also introduces a somewhat standard numbering. The thing to do at this point would be to transcribe the data in this widely cited paper into a file. Google seems to indicate that no such file has been posted to the web.

Update 2: It was done! See file data/steiner.tbl in this GAP package by Nagy and Vojtechovsky. For some reason, a Steiner triple system of order $n$ is also called a Steiner loop of order $n+1$, and that is the terminology that they use. They copied the data from Colbourn and Rosa, Triple Systems, which presumably is the same as in Mathon, Phelps, and Rosa.

How I found it: I Googled one of the hexadecimal strings used to describe one of the STS(15)s. None of the Google's heuristics worked for me, so instead I used the old-fashioned trick of searching for a very specific keyword. It also shows up in a LaTeX PHD thesis, but the GAP file, which has almost the same syntax as Python, is better.

  • 2
    $\begingroup$ If $S$ is a STS, then one can add an element to get $S'=S\cup\{\infty\}$ and define an operation $\star$ on $S'$ so that $\infty$ is the identity element, $x\star x=\infty$ for all $x$ in $S$, and $x\star y=z$ if $\{x,y,z\}$ is a block of~$S$. Then $(S,\star)$ is a loop, and its order is one more than $S$'s. That's why they index things funnily. $\endgroup$ – Mariano Suárez-Álvarez May 20 '10 at 4:36
  • 1
    $\begingroup$ You're very welcome. As usual when I respond to a question in MO, I learned something. $\endgroup$ – Greg Kuperberg May 20 '10 at 4:50
  • $\begingroup$ Sadly, the GAP file has gone AWOL :-/ $\endgroup$ – Mariano Suárez-Álvarez Apr 24 '13 at 7:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.