I want to process data frame as follows, where I want to get the sum of 2 vectors and append it to a data frame as a row vector. 2 vectors are row vector of considering row and column vector which start just below the considering row with a fixed length.

```
data
A b1 b2 b3
1 2 2 2
2 3 3 3
3 4 4 4
4 5 5 5
5 6 6 6
output (expected)
A b1 b2 b3
1 4 5 6
2 6 7 8
3 8 9 -
4 10 - -
5 - - -
```

In the example if 1st row is considered, two vectors are

```
row vector r- [2 2 2]
column vector c - [2,3,4]
```

After getting the transpose of column vector I can add tow vectors and append it to a new data frame. This process must be done to all the rows.

Easiest way to do this is looping, but in R loops are not efficient, instead `apply`

function can be used. However in this scenario, to do that need to know what is the current row number.

Is there a way to do this efficiently in R

r answered 4 months ago G. Grothendieck #1

**1) rollapply** We can use `rollapply`

to form the matrix of subvectors of `A`

and then add that together with an initial column of zero to `m`

. Note that we pad `A`

with NA values so that the result of `rollapply`

is the appropriate shape.

```
library(zoo)
m <- cbind(A = 1:5, b1 = 2:6, b2 = 2:6, b3 = 2:6) # input matrix
nc1 <- ncol(m) - 1
A <- c(m[, 1], rep(NA, nc1))
cbind(0, rollapply(A[-1], nc1, c)) + m
```

giving:

```
A b1 b2 b3
[1,] 1 4 5 6
[2,] 2 6 7 8
[3,] 3 8 9 NA
[4,] 4 10 NA NA
[5,] 5 NA NA NA
```

**2) base** This solution is similar but does not use any packages. The first two lines are the same as in (1).

```
nc1 <- ncol(m) - 1
A <- c(m[, 1], rep(NA, nc1))
cbind(0, embed(A[-1], nc1)[, seq(nc1, 1)]) + m
```

giving:

```
A b1 b2 b3
[1,] 1 4 5 6
[2,] 2 6 7 8
[3,] 3 8 9 NA
[4,] 4 10 NA NA
[5,] 5 NA NA NA
```