It is usually easy enough to create one or a few instances of a graph with a certain property. E.g., you can make an educated guess about which model already in igraph will produce a certain property, then vary its parameters. Or smoothly interpolate between the adjacency matrices of two graphs having different values of a property. E.g. a star graph will have betweenness centralization of nearly 1, a cycle graph nearly 0.
Will the graphs you obtain this way be representative of most graphs with the same 'centralization' (or other property) value? Probably not. That will also be the case if you choose more "random looking" graphs as starting points, such as one from a preferential attachment model and a random regular graph. Therein lies the problem with using any arbitrary graph with a property instead of sampling them uniformly. On 3 April 2018 at 02:28, John Erwin Banez <jsba...@up.edu.ph> wrote: > I am trying to test robustness of a method I am trying to develop. Having > igraphs with varied centrality helps. > > > > On Mon, Apr 2, 2018 at 5:48 PM, Szabolcs Horvát <szhor...@gmail.com> > wrote: > >> If you want uniform sampling, then this is a research-level problem (not >> a technical one). If you don't, I wonder what conclusion you could draw >> from the results. >> >> On 2 April 2018 at 08:10, John Erwin Banez <jsba...@up.edu.ph> wrote: >> >>> How do I generate several igraphs with varying graph level >>> centralization? I am familiar with sample_(...), that variations of >>> 'sample_' allow me to generate graphs GIVEN a MODEL....but not given graph >>> level centralization. >>> >>> I would like to have for example: >>> >>> graph_1 <- centralization=.50 >>> graph_2 <- centralization=.30 >>> >>> . >>> >>> . >>> >>> . >>> >>> graph_n <- centralization=.00 >>> >>> >>> -- >>> John Erwin Bañez >>> Assistant Professor >>> College of Social Work and Community Development >>> University of the Philippines, Diliman >>> >>> _______________________________________________ >>> igraph-help mailing list >>> igraph-help@nongnu.org >>> https://lists.nongnu.org/mailman/listinfo/igraph-help >>> >>> >> >> _______________________________________________ >> igraph-help mailing list >> igraph-help@nongnu.org >> https://lists.nongnu.org/mailman/listinfo/igraph-help >> >> > > > -- > John Erwin Bañez > Assistant Professor > College of Social Work and Community Development > University of the Philippines, Diliman > > _______________________________________________ > igraph-help mailing list > igraph-help@nongnu.org > https://lists.nongnu.org/mailman/listinfo/igraph-help > >
_______________________________________________ igraph-help mailing list igraph-help@nongnu.org https://lists.nongnu.org/mailman/listinfo/igraph-help
