Given a column, 1) take the negative numbers, 2) make them positive, 3) sqrt() it, 4) then put it back in as negative

[pathos]

I have a small puzzle.

Given a column of a continuous variable, I would like to:

1. square root only the positive numbers
2. take out the negative numbers, make them positive, square root them, then put them back into the column
3. ignore the 0s and don't take the square root, for computational efficiency sake

I thought about using ifelse, but this sounds computationally heavy. Would there be a better way?

[jrmuirhead]

set.seed(123)

(mydf<- data.frame(x = rnorm(10)))
#>              x
#> 1  -0.56047565
#> 2  -0.23017749
#> 3   1.55870831
#> 4   0.07050839
#> 5   0.12928774
#> 6   1.71506499
#> 7   0.46091621
#> 8  -1.26506123
#> 9  -0.68685285
#> 10 -0.44566197
mydf[5, "x"] <- 0 # Set one value to zero deliberately

(mydf <- within(mydf, {y = sign(x) * sqrt(abs(x))}))
#>              x          y
#> 1  -0.56047565 -0.7486492
#> 2  -0.23017749 -0.4797682
#> 3   1.55870831  1.2484824
#> 4   0.07050839  0.2655342
#> 5   0.00000000  0.0000000
#> 6   1.71506499  1.3096049
#> 7   0.46091621  0.6789081
#> 8  -1.26506123 -1.1247494
#> 9  -0.68685285 -0.8287659
#> 10 -0.44566197 -0.6675792

^{Created on 2022-03-29 by the reprex package (v2.0.1)}

[pathos]

Ah sorry, I should have clarified from the beginning for a dplyr or tidytable equivalent

[nirgrahamuk]

@jrmuirhead

(mydf <- within(mydf, {y = sign(x) * sqrt(abs(x))}))

@pathos

for a dplyr or tidytable equivalent

(mydf <- mydf%>% mutate( y = sign(x) * sqrt(abs(x))))

Note : This might be simplified by removing sign(x) * , as when you list out what to do with them (2) you dont say to make them negative as a subsequent step.
Also, the solution doesnt skip 0's but I don't see a motivated reason to skip them either.