network visualization linux

[Elijah Wright]

We've used R's sna package locally to generate lots (and lots, and lots, 
and lots...) of graph plots.  It works well, particularly for work with 
data that may not conform to people's initial expectations of what 
relations may be...

However... it is sometimes a bit memory-hungry.  :)  You are correct in 
asserting that you will need quite a bit of memory to plot a large network 
- I've used numerous 16GB and 32GB machines running Sun's Solaris to plot 
absurdly-sized networks with what I would call a good degree of success.

All of the standard tricks for working with large matrices work quite well 
with R-sna.  [In effect, there's a real payoff for properly trimming or 
partitioning your data as early in the analysis process as possible...] 
Getting to know the other R packages that deal with matrix-oriented data 
is also highly recommended, for those who want to make serious use of 
'sna'.  Several of the more useful tools are not actually *in* the sna 
package, but found inside other toolkits that are packaged for R.


A few of Katy Borner's students have simply run Pajek under Wine; that 
seems to work well enough.  I don't think I've seen anyone run UCINET 
under Wine, but I suppose that should also be possible.

--elijah


On Tue, 14 Nov 2006, Thomas Krichel wrote:

> Date: Tue, 14 Nov 2006 13:36:27 -0600
> From: Thomas Krichel <krichel at OPENLIB.ORG>
> Reply-To: ASIS&T Special Interest Group on Metrics
>     <SIGMETRICS at LISTSERV.UTK.EDU>
> To: SIGMETRICS at LISTSERV.UTK.EDU
> Subject: Re: [SIGMETRICS] network visualization linux
> 
> Adminstrative info for SIGMETRICS (for example unsubscribe):
> http://web.utk.edu/~gwhitney/sigmetrics.html
>
>  Frank Havemann writes
>
>> I recommend the sna-package of R:
>> http://www.r-project.org/
>
>  I have had problems with sna and had to abandon it to
>  calculate my 4k+ author network paths. I kept running into
>  memory problems even on my HP Proliant 8 processor box
>  with close to 4G of memory. I ended up having to write
>  my own software to do the calculations. I have not used its
>  vizualization features.
>
>  But that was last year. Maybe this is fixed now.
>
>
>
>  Cheers,
>
>  Thomas Krichel                             mailto:krichel at openlib.org
>                                        http://openlib.org/home/krichel
>                                    RePEc:per:1965-06-05:thomas_krichel
>                                                skype id: thomaskrichel
>