r - Cumulatively apply function in dplyr -

consider following data:

df <- data.frame(names = sample(letters[1:5], 20, replace = t), numbers = 1:20) 

for have function cumsum, such each row in group, computes cumulative sum of numbers row

library(dplyr) df %>%   group_by(names) %>%   mutate(cumsum_numbers = cumsum(numbers)) 

i wish apply general function my_fn cumulatively in same manner cumsum. my_fn has general form:

my_fn <- function(vector){   # stuff  vector   return(x) # x numeric scalar  } 

that is, takes vector of previous values row, , returns scalar.

the following code not work:

df %>%   group_by(names) %>%   mutate(cumsum_numbers = my_fn(numbers)) # apply my_fn                              # each group in numbers, returning                              # same value each grouping level 

so guess want like:

df %>%   group_by(names) %>%   mutate(cumsum_numbers = cum_my_fn(numbers)) 

note example function mean calculating cumulative mean. interestingly dplyr has implemented cummean, don't know internal workings of can't work out how implement behaviour general function.