CODEWITHSUNDEEP
r
Adrian
Adrian

Reputation: 843

Multiplying elements of a matrix depending on numbers and strings of the row names and column names

I have a very large matrix that has column names and row names. These column and row names are identical and contain a three letter string with a number.

The three letter string repeats itself and only the number changes. After several repetitions, the string changes and the number starts from 1 again.

To give a small example, I have a matrix a:

a <- matrix(c(1:36), nrow = 6, byrow = TRUE)

names <- paste(rep(c("aaa" , "bbb", "ccc"), each = 2) , rep(c(1:2) , times = 3))

rownames(a) <- names
colnames(a) <- names

that gives:

      aaa 1 aaa 2 bbb 1 bbb 2 ccc 1 ccc 2
aaa 1     1     2     3     4     5     6
aaa 2     7     8     9    10    11    12
bbb 1    13    14    15    16    17    18
bbb 2    19    20    21    22    23    24
ccc 1    25    26    27    28    29    30
ccc 2    31    32    33    34    35    36

For each element of this matrix I would like to do a multiplication.

So basically for every element where the string of the row / column names are different, I want to match the number of the row name / column name and multiply with the other string.

If "aaa" is matched with "bbb" then:

matrix[aaa (number n), bbb (number m)] * matrix[bbb (number n), aaa (number m)]

if "aaa" is equal to "aaa" then

matrix[aaa (number n), aaa (number m)] * matrix[aaa (number n), aaa (number m)]

or basically the element squared.

It looks confusing so I give a few examples:

In the end, the matrix should give:

      aaa 1 aaa 2 bbb 1 bbb 2 ccc 1 ccc 2
aaa 1     1     4    39    56   125   156
aaa 2    49    64   171   200   341   384
bbb 1    39    56   225   256   459   504
bbb 2   171   200   441   484   759   816
ccc 1   125   156   459   504   841   900
ccc 2   341   384   759   816  1225  1296

which I get using terrible code:

b <- a^2

b[1,3] <- a[1,3] * a[3,1]
b[1,4] <- a[1,4] * a[3,2]
b[1,5] <- a[1,5] * a[5,1]
b[1,6] <- a[1,6] * a[5,2]
b[2,3] <- a[2,3] * a[4,1]
b[2,4] <- a[2,4] * a[4,2]
b[2,5] <- a[2,5] * a[6,1]
b[2,6] <- a[2,6] * a[6,2]

b[3,1] <- a[3,1] * a[1,3]
b[3,2] <- a[3,2] * a[1,4]
b[3,5] <- a[3,5] * a[5,3]
b[3,6] <- a[3,6] * a[5,4]
b[4,1] <- a[4,1] * a[2,3] 
b[4,2] <- a[4,2] * a[2,4]
b[4,5] <- a[4,5] * a[6,3]
b[4,6] <- a[4,6] * a[6,4]

b[5,1] <- a[5,1] * a[1,5]
b[5,2] <- a[5,2] * a[1,6]
b[5,3] <- a[5,3] * a[3,5]
b[5,4] <- a[5,4] * a[3,6]
b[6,1] <- a[6,1] * a[2,5]
b[6,2] <- a[6,2] * a[2,6]
b[6,3] <- a[6,3] * a[4,5]
b[6,4] <- a[6,4] * a[4,6]

Is there a code that lets me do this type of matching? Maybe also when there are more strings and numbers?

Upvotes: 0

Views: 87

Answers (2)

Onyambu
Onyambu

Reputation: 79208

An easy way that is vectorized Using base R you can do:

b = expand.grid(dimnames(a))[2:1]
d = with(read.table(text=paste(b[,1],b[,2])),cbind(paste(V3,V2),paste(V1,V4)))
t(array(a[as.matrix(b)]*a[d],dim(a)))
     [,1] [,2] [,3] [,4] [,5] [,6]
[1,]    1    4   39   56  125  156
[2,]   49   64  171  200  341  384
[3,]   39   56  225  256  459  504
[4,]  171  200  441  484  759  816
[5,]  125  156  459  504  841  900
[6,]  341  384  759  816 1225 1296

Upvotes: 0

Roland
Roland

Reputation: 132696

You can use a data.table join:

library(data.table)
#melt to long format
DT <- setDT(melt(a))

#split ID columns
DT[, c("Var1str", "Var1num", "Var2str", "Var2num") := cbind(tstrsplit(Var1, " "), 
                                                            tstrsplit(Var2, " "))]

#simple join on split IDs, flipping the strings
DT[DT, res := value * i.value, on = c(Var1num = "Var1num", Var2num = "Var2num", 
                                      Var1str = "Var2str", Var2str = "Var1str")]

#create wide-format matrix
res <- dcast(DT, Var1 ~ Var2, value.var = "res")
rn <- res[["Var1"]]
res <- as.matrix(res[, -1])
rownames(res) <- rn
res
#      aaa 1 aaa 2 bbb 1 bbb 2 ccc 1 ccc 2
#aaa 1     1     4    39    56   125   156
#aaa 2    49    64   171   200   341   384
#bbb 1    39    56   225   256   459   504
#bbb 2   171   200   441   484   759   816
#ccc 1   125   156   459   504   841   900
#ccc 2   341   384   759   816  1225  1296

Upvotes: 3

Related Questions