2
$\begingroup$

I am looking for software packages for working with exponential random graph models (fitting/generating them and sampling from the graph distributions).

I have only found two packages so far, both using R: ergm/statnet and RSiena, with the former being much more popular.

Are there any other packages available?

I'm particularly interested in methods that are both published/documented and have an available implementation to play with.

$\endgroup$
2
$\begingroup$

Find PNet here: http://sna.unimelb.edu.au/PNet

This is Java based software for fitting exponential random graph models, now including a multilevel version.

Incidentally, RSiena does not fit ERGM models. The old R-independent SIENA software (which is no longer maintained) did, however.

$\endgroup$
3
  • $\begingroup$ Thanks for clarifying that about RSiena. I only looked at the webpage, but haven't played with it yet. (I played with ergm a bit) $\endgroup$
    – Szabolcs
    Apr 20 '13 at 17:01
  • $\begingroup$ So is these two (ergm, PNet) what people typically use in this field? Or do they usually write their own implementations? Also, since you seem to be familiar with this topic, do you know if ergm allows giving graph measures directly to the fitting algorithm instead of a giving an actual graph? (For example, specify the number of vertices, edges and triangles, instead of supplying a graph with exactly that number of vertices, edges and triangles.) $\endgroup$
    – Szabolcs
    Apr 20 '13 at 17:05
  • 1
    $\begingroup$ I'd say most published articles using ERGM models either used the Statnet package, or used PNet. I don't even know of other well developed packages for ERGM (of course, that doesn't imply there aren't any!). If you're estimating parameters, I don't see how you could simply supply a few statistics and expect useful parameters in return. However, you could specify some parameters yourself and simulate graphs which can be used for study. Maybe I misunderstand your circumstances and goal, though. $\endgroup$
    – ndoogan
    Apr 20 '13 at 17:09
0
$\begingroup$

Siena 3.0 is working just fine performing ERGM. It uses the same algorithm as the PNet. It is just Tom (the developer for Siena) is moving on to Siena 4.0 for longitudinal SNA. It does not mean Siena 3.0 is useless though. In fact, my favorite is still Siena 3.0 because it is simple and easy to use. My guess is that PNet is also very easy to handle, I just happen to pick Siena when I run ERGM first.

Many people mentioned R and illustrated ERGM with R. I do not know much about R, although it seems it has deep learning curve.

$\endgroup$

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.