R中邻接矩阵的传递归约

如何解决R中邻接矩阵的传递归约

我有一个成对矩阵，我可以将其视为图的邻接矩阵。我希望应用传递归约算法来找到边最少但保留原始图的连通性的图 - 见下图。

我的矩阵的头部看起来像这样：

               EN_DavaW      EN_DrumW   CN_ShainW  CN_Glasdrum        19-CCP
EN_DavaW     0.0000000000  2.286985e-03 0.014775598  0.013954988 -0.0149552822
EN_DrumW    -0.0022869851  0.000000e+00 0.013133681  0.011270755 -0.0166146429
CN_ShainW   -0.0147755985 -1.313368e-02 0.000000000 -0.001550990 -0.0244997421
CN_Glasdrum -0.0139549879 -1.127075e-02 0.001550990  0.000000000 -0.0328348644
19-CCP       0.0149552822  1.661464e-02 0.024499742  0.032834864  0.0000000000

在这个矩阵中，正整数可以用从 Pop 1 到 Pop 2 的箭头来表示。而负值则表示从 Pop 2 到 Pop1。

我正在努力寻找可用于 R version 4.02 的软件包，以便在我的矩阵上执行此操作。 我查看了包 nem，更具体地说是函数 nem::transitive.reduction see here，但它不适用于上述版本。即使通过 bioconductor

安装

是否有其他包或者我可以创建自己的函数来对成对矩阵进行传递归约？

解决方法

我认为您可以尝试像下面这样的 igraph + relations 组合

library(igraph)
library(relations)

g <- graph_from_adjacency_matrix(m,mode = "directed",weighted = TRUE)
df <- get.data.frame(g)
r <- endorelation(
  domain = as.list(unique(unlist(df[c("from","to")]))),graph = df[c("from","to")]
)
mat <- relation_incidence(transitive_reduction(r))
mattr <- m[row.names(mat),colnames(mat)] * mat
gtr <- graph_from_adjacency_matrix(mattr,weighted = TRUE)

• 原点图

• 传递归约图

---

数据

> dput(m)
structure(c(0,-0.0022869851,-0.0147755985,-0.0139549879,0.0149552822,0.002286985,-0.01313368,-0.01127075,0.01661464,0.014775598,0.013133681,0.00155099,0.024499742,0.013954988,0.011270755,-0.00155099,0.032834864,-0.0149552822,-0.0166146429,-0.0244997421,-0.0328348644,0),.Dim = c(5L,5L),.Dimnames = list(c("EN_DavaW","EN_DrumW","CN_ShainW","CN_Glasdrum","19-CCP"),c("EN_DavaW","19-CCP")))