如何在 R 中删除 igraph 中的节点？

Question

我正在尝试编写一个脚本，根据节点的连接数使用 igraph 在 R 中删除 Barabasi-Albert 网络中的节点（我正在尝试从“复杂的错误和攻击容限”中重新创建一些基本结果网络”由 Albert、Jeong 和 Barabasi 撰写的论文）。我首先尝试删除五个随机节点，这些节点的连接数少于网络中节点的平均连接数。但是，当我在尝试删除节点后可视化网络时，它看起来确实有所不同，但看起来节点并没有被删除。所以我不相信脚本有效。

nnodes=50 #number of nodes
test.graph<-barabasi.game(nnodes,.5) #create B-A network
test.graph2=test.graph #create a second B-A network for removing nodes
bar_mat=matrix(0,nrow=nnodes,ncol=1) #create empty matrix
for (i in 1:nnodes){
bar_mat[i,]=sum(test.graph[,i]) #sums up the number of connections of each node
}
###Visualizing the network before removing nodes
barabasi.community<-walktrap.community(test.graph) #this is supposed to visualize the most
#connected nodes in the network
members<-membership(barabasi.community)
plot.igraph(test.graph,
        layout=layout.fruchterman.reingold,
        vertex.size=10,
        vertex.label.cex=.5,
        edge.arrow.size=.5,
        mark.groups=list(members),
        mark.col="green"
)
f=c()

for (k in 1:5){ #checking five random nodes
 a=sample(1:nrow(bar_mat),1) #select random node
 if(bar_mat[a,]<=mean(bar_mat)){
  test.graph2 <- delete.vertices(test.graph2,a) # this is supposed to delete 
  #the node based on if it has lower than the average amount of connections
  i=i+1 #counting how many nodes of the five are actually removed
 }
f[k]=a #putting the nodes tested into a vector
a=0 #resetting a
}
###Visualizing network after node removal 
barabasi.community2<-walktrap.community(test.graph2)
members2<-membership(barabasi.community2)
plot.igraph(test.graph2,
        layout=layout.fruchterman.reingold,
        vertex.size=10,
        vertex.label.cex=.5,
        edge.arrow.size=.5,
        mark.groups=list(members2),
        mark.col="pink"
) 

脚本在节点数量较少（如大约 50 个）但节点数量较多（大约 100 个）时运行，我收到以下错误：

Error in delete.vertices(test.graph2, a) : 
At iterators.c:759 : Cannot create iterator, invalid vertex id, Invalid vertex id

我认为这与节点的命名约定有关，但我不确定。我是网络科学的新手，我不是最好的程序员，所以我非常感谢任何帮助。谢谢！

Answer 1

由于每次都是从nrow(bar_mat) 中随机抽样，所以会出现这个错误，你可以选择一个已经被删除的节点。例如，循环第一次运行并选择第 24 行。它将工作并删除节点 24。它再次运行并选择第 12 行。它将工作并删除节点 12。它再次运行并选择第 24 行。现在它可以'不要运行，因为图形不再有节点 24！否则，您的代码很好，并且确实在删除节点。

您可以通过多种方式解决此问题，例如一次对 5 个随机节点进行采样（这里我做了 20 个以清楚地表明节点已被删除）：

library(igraph)

nnodes=100 #number of nodes
test.graph<-barabasi.game(nnodes,.5) #create B-A network
test.graph2=test.graph #create a second B-A network for removing nodes
bar_mat=matrix(0,nrow=nnodes,ncol=1) #create empty matrix
for (i in 1:nnodes){
  bar_mat[i,]=sum(test.graph[,i]) #sums up the number of connections of each node
}

vcount(test.graph)
#> [1] 100

# Sample 5 random nodes
set.seed(1491)
a=sample(1:nrow(bar_mat), 20)

for (i in a) {
  if(bar_mat[i,]<=mean(bar_mat)){a
    test.graph2 <- delete.vertices(test.graph2,i)
    }
  }
#> Error in delete.vertices(test.graph2, i): At iterators.c:759 : Cannot create iterator, invalid vertex id, Invalid vertex id

vcount(test.graph)
#> [1] 100
vcount(test.graph2)
#> [1] 88

^{由reprex package (v0.3.0) 于 2020-04-07 创建}