Print

Print


*****  To join INSNA, visit http://www.insna.org  *****

Phillip,

If I remember correctly, calculation of mean/variance should be done by
Monte Carlo approximation,
and at least in the case of mean NetMiner does it. I don't remember whether
they provide variance,
but in my vague memory I think they at least give you p-values, which is
more meaningful than variance
(since those triad census statistics are not necessarily (and almost surely
not) normally distributed).

I'm sorry I can't give you details since I can't run the program now,
but I'm pretty sure you have the solution (in any form) there.

Unfortunately NetMiner is a commercial program, but they provide trial
versions for restricted
node size of the graph. Since you can't run triad census on large graphs
anyway,
(since you also need to get mean/variance of those statistics, the
computational cost blows up
so I guess it's only doable for graphs with only 10~100 nodes)
this would not be a big problem.

(You can get it in: http://www.netminer.com )

If you want a scalable implementation of this you may try using nauty,
a software for graph isomorphism testing.
It could be that someone has already implemented triad census based on
nauty,
but I haven't seen it personally.

You can get nauty in:
http://cs.anu.edu.au/people/bdm/nauty/

I think SOCNET has been receiving multiple question about triad census
lately.
I wonder what is the reason? Can you elaborate on in what context you're
interested in triad census?



Thanks,
Hyokun Yun
Ph.D Candidate
Department of Statistics
Purdue University

<http://cs.anu.edu.au/people/bdm/nauty/>
On Fri, Jan 28, 2011 at 2:49 PM, Philip A Sinclair <[log in to unmask]
> wrote:

> ***** To join INSNA, visit http://www.insna.org *****
> Dear List,
>  Please is there an available program that calculates the mean and
> variance of the triad census vector for U|MAN random graph models?
>
> Regards,
> Philip Sinclair
> _____________________________________________________________________
> SOCNET is a service of INSNA, the professional association for social
> network researchers (http://www.insna.org). To unsubscribe, send an email
> message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET
> in the body of the message.

_____________________________________________________________________
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.