Re: [igraph] preserving multiple graphs

[Gabor Csardi]

Uri,

On Sun, Feb 10, 2008 at 07:45:19PM -0500, Uri Shwed wrote:
> Hi Gabor and all,
> I'm trying to analyze the unfolding of a network over time, using  
> cohesive blocks. Since towards the end of my time frame the network  
> becomes quite huge, I thought it would be useful to first store all  
> the graphs into a list, and then cohesive.blocks the periods  
> separately.  This has taught me that lists don't store graphs - is  
> that true? How do one stores graphs then? I wouldn't mind even saving  
> them as separate files, but I do need a way to give those files  
> serial names automatically.

hmmm, lists are perfect for storing graphs, in fact, they are perfect 
for storing anything. E.g.

l <- list( ba.game(1000, m=3, dir=FALSE), 
           erdos.renyi.game(1000, 3000, type="gnm") )
sapply(l, transitivity)

works fine.

Before checking your code. There is a bug in the 0.4.5 release in the 
cohesive.blocks function, sometimes the cohesive blocks are not
calculated correctly. We will release a corrected version, igraph 0.5
soon. If you can't wait, see this message for preliminary packages:
http://lists.gnu.org/archive/html/igraph-help/2008-02/msg00011.html

> I hope I'm making some sense...I'm just doing my first baby steps in  
> R. Here's the code I have:
> 
> 
> ##this function takes data of 3 columns - sender(citing), receiver  
> (cited) and year(citey), and uses a moving window to build many  
> networks, and store them by the first year in the window .
> 
> window.matgrf<-function(data, start=1950, end=1990, lag=8) {
>         t<-start
>         lag<-lag
>         end<-end
>         last<-end-lag
>         years<-start:end
>         lis<<-list(year=NA,mtrx=NA,grf=NA)

Do not use <<- here, it is not needed. Actually <<- is very rarely 
needed and it will almost always a source of great confusion, whenever 
you use it. 

>       lstname<<-list(year=NA,mtrx=NA,grf=NA,blks=NA,blkgrf=NA) //This is  
> for future use with faster machines - maybe I could get the blocks on  
> this run.
>         for (t in start:last) {
>                 mtx<-matrix(c(data$citing[data$citey>=t & data$citey  
> <=t+lag], data$cited [data$citey >=t & data$citey
>                 <=t+lag]), ncol=2)
>                 print(t)
>                 grfy<-graph.edgelist(mtx)
>               print(is.igraph(grfy))
>               lis<-list(year=t, matrx=mtx,grf=grfy)
>                 lstname[t]<<-list(lis)
>                 t<-t+1
> 
>                 }
>       lstname
>          }
> ##so to build a list called Set:
> 
>  Set<-window.matgrf(FullData, end=1994)
> ## and then to get the cohesive blocks for 1977 -  1985:
> cohesive.blocks(Set[[1975]][2])
> 

It would be easier if you sent a reproducible example, with the
code only i would need to figure out how your input data looks like,
made up some toy input data, etc. From the code it seems to me that
you don't have a list of graphs here, but a list of lists, and the
third ('grf') elements of the "small" lists are the graphs. 
If this is true then you need something like

cohesive.blocks(Set[[1975]]$grf)

or if you're sure that the graph is always the third element, then

cohesive.blocks(Set[[1975]][[3]])

Gabor

ps. it is ok if you're not on the list, but then please note thar 
your emails will be delayed, as a i need to acknowledge them "by hand".

-- 
Csardi Gabor <address@hidden>    UNIL DGM