Re: [igraph] to find connected components in the network

[Gábor Csárdi]

Hi,

On Mon, Jul 5, 2010 at 3:25 PM, Rachita Halehalli <address@hidden> wrote:
> Hi,
> Thanks for your inputs. from past one week internet was down here.
> I have 11 networks each one having nearly 11000 nodes and over 50000edges.
> When i was trying to find shortest paths it wasn't fitting into memory.

well, how could we help this? If you really need to calculate all
shortest paths at the same time, then just buy more memory. For your
graph you need 50000x50000x8 bytes, about 19GB. If you use Python or
R, then the whole thing is copied once, so you need twice as much.
Another option is not calculating all shortest paths at the same time.

> Clusters actually gives one large connected component with over 10000nodes
> and about 400 peripheral connections with 2-6nodes.
> I need to find highly connected subgragphs in the large component which r
> connected by articulation point. Even after removing node i get same results
> for clusters function.
> It is same with decompose graph.

Then most probably you are doing something wrong. Or we did something
wrong. Please show us an example that goes wrong, or at least your
code.

>  What should i do to get local subcomponents which are connected by
> articulation points?

Show us your code, and we can tell you what you are doing wrong. Or we
can find the bug in our code, if there is one.

Thanks,
Gabor

> I want to find out what are these components/ subgraphs and their local
> network properties and function.
>
> Thanks and regards
>
> Rachita
>
> On 30 June 2010 20:59, Gábor Csárdi <address@hidden> wrote:
>>
>> Hi,
>>
>> what exactly is tedious? If I understand you correctly, all you need
>> to do is removing the articulation points one by one from the network,
>> and then calling clusters() or decompose.graph() on the remainder
>> graphs.
>>
>> If your graph is not too big and everything fits into the memory, then
>> you can do all this with
>>
>> rem <- lapply(A, delete.vertices, graph=g)
>> comps <- lapply(rem, decompose.graph)
>>
>> where 'A' is the number vector with the articulation points and 'g' is
>> your graph. If you just need the vertex ids and not the actual graphs,
>> then use 'clusters' instead of 'decompose.graph', 'clusters' is much
>> faster.
>>
>> Is this what you want?
>>
>> Best,
>> Gabor
>>
>> On Wed, Jun 30, 2010 at 4:26 PM, Rachita Halehalli <address@hidden>
>> wrote:
>> >
>> > Hi,
>> > Can any one help me out to find components in the network.
>> >
>> > I have list of articulation points for a network.
>> > Now i want to find out the different components that each articulation
>> > point
>> > connects..
>> >
>> > Degree of articulation points in the list  varies from 2 to 100. It
>> > becomes
>> > very tedious to use fucntion 'subcomponent'.
>> >
>> > Thanks and warm regards
>> >
>> > -Rachita Halehalli
>> > Graduate student
>> > CDFD
>> > Hyderabad
>> >
>> > _______________________________________________
>> > igraph-help mailing list
>> > address@hidden
>> > http://lists.nongnu.org/mailman/listinfo/igraph-help
>> >
>> >
>>
>>
>>
>> --
>> Gabor Csardi <address@hidden>     UNIL DGM
>
>
>
> --
> Thanks and warm regards,
>
> Rachita H R
> Graduate Student
> CDFD,
> Hyderabad
> India.
>



-- 
Gabor Csardi <address@hidden>     UNIL DGM