For each element in a matrix, find the sum of all of its neighbors

问题

Given a matrix, I want to find the sum of the neighbors for each element (so the result is a matrix). The neighbors are the values above, below and beside the given element ,if they exist (not considering the diagonal elements).

Example:

> z = matrix(1:9, 3, 3, byrow=T)
> z
     [,1] [,2] [,3]
[1,]    1    2    3
[2,]    4    5    6
[3,]    7    8    9

And the expected result is :

> result
     [,1] [,2] [,3]
[1,]    6    9    8
[2,]   13   20   17
[3,]   12   21   14

What is the simplest way I can do this in R without using loops?

回答1:

One way would be to make matrices with the neighbor on each side and add them together.

rbind(z[-1,],0) + rbind(0,z[-nrow(z),]) + cbind(z[,-1],0) + cbind(0,z[,-ncol(z)])
##      [,1] [,2] [,3]
## [1,]    6    9    8
## [2,]   13   20   17
## [3,]   12   21   14

回答2:

Here's one way to do it. I won't claim it's the simplest, but it certainly avoids looping.

m <- matrix(1:16, 4, 4, byrow=TRUE)

result <- array(sapply(seq_along(m), function(i) {
  ind <- which(col(m) == col(m)[i] & abs(row(m)[i] - row(m)) == 1 | 
                 row(m) == row(m)[i] & abs(col(m)[i] - col(m)) == 1)
  sum(m[ind])
}), dim(m))


result

     [,1] [,2] [,3] [,4]
[1,]    7   10   13   11
[2,]   16   24   28   23
[3,]   28   40   44   35
[4,]   23   38   41   27

来源：https://stackoverflow.com/questions/22572901/for-each-element-in-a-matrix-find-the-sum-of-all-of-its-neighbors