More compact notation to work with subgraphs

[gaborcsardi]

When a function is called on a subgraph of a graph, one would not need to explicitly create the subgraph, just use the more compact notation. From #38.

[gaborcsardi]

It is relatively hard to do this with the R operators that we are allowed to redefine, in a way that is readable. The most logical would be probably to define & for graphs, and also add additional functionality to the V() and E() functions. Then we could write things like

karate & V(1:10)              # Induced subgraph including vertices 1-10
karate & V(color=="red")      # Induced subgraph including the red vertices
karate & V(degree > 0)        # Induced subgraph of the non-isolate vertices

and similarly for edges, here all vertices are kept:

karate & E(1:10)              # Keep only edges 1-10
karate & E(inc(1:5,10))       # Keep edges incident to vertices 1-5 and 10
karate & E(weight >= 1)       # Keep edges with large enough weight

They can be combined like this:

karate & E(weight >= 1) & V(degree > 0)

This will remove edges with small weight, and then remove the isolate vertices.

[jovo]

for reference, i understand that people are quite fond of the syntax used in dplyr: https://github.com/hadley/dplyr specifically:

• select(): focus on a subset of variables
• filter(): focus on a subset of rows
• mutate(): add new columns
• summarise(): reduce each group to a smaller number of summary statistics
• arrange(): re-order the rows

obviously, we'd want something modified more for graphs rather than data frames. i'm not sure how easy it is to use such a syntax, but seemed like potentially worth mentioning....

On Thu, May 8, 2014 at 9:57 PM, Gabor Csardi notifications@github.comwrote:

It is relatively hard to do this with the R operators that we are allowed to redefine, in a way that is readable. The most logical would be probably to define & for graphs, and also add additional functionality to the V() and E() functions. Then we could write things like

karate & V(1:10) # Induced subgraph including vertices 1-10 karate & V(color=="red") # Induced subgraph including the red vertices karate & V(degree > 0) # Induced subgraph of the non-isolate vertices

and similarly for edges, here all vertices are kept:

karate & E(1:10) # Keep only edges 1-10 karate & E(inc(1:5,10)) # Keep edges incident to vertices 1-5 and 10 karate & E(weight >= 1) # Keep edges with large enough weight

They can be combined like this:

karate & E(weight >= 1) & V(degree > 0)

This will remove edges with small weight, and then remove the isolate vertices.

— Reply to this email directly or view it on GitHubhttps://github.com/igraph/xdata-igraph/issues/39#issuecomment-42629175 .

perhaps consider allowing the quest for eudaimonia to guide you openconnecto.me, jovo.me

[gaborcsardi]

dplyr is great, and you are free to use its chaining operator for igraph:

library(igraph)
library(dplyr)
G <- graph.ring(10)
G %.% induced.subgraph(vids=1:5) %.% degree()
# [1] 1 2 2 2 1

If this is what you mean by the syntax used in dplyr.

If you want to have new igraph functions, please open separate issues for them.

[dpmcsuss]

The syntax as is seems pretty good to me now that I know how works

[jovo]

i'm happy if daniel is happy :) (i'm happy anyway, actually :-)

On Fri, May 9, 2014 at 8:47 AM, Daniel Sussman notifications@github.comwrote:

The syntax as is seems pretty good to me now that I know how works

— Reply to this email directly or view it on GitHubhttps://github.com/igraph/xdata-igraph/issues/39#issuecomment-42667256 .

perhaps consider allowing the quest for eudaimonia to guide you openconnecto.me, jovo.me