MemoFunc - A Function Memoization Package for R

A simple way to memoise function results to improve performance by eliminating unnecessary computation or data retrieval activities.

Functions can be memoized with a simple call to memo.

> # a simple example function
> simple.function <- function (value) {
+   print("Executing!")
+   value
+ }

> # call memo function to memoise a function
> simple.function.memo <- memo(simple.function)

> # or like this
> simple.function %<>% memo()

> # or like this
> simple.function2 <- (function (value) value) %>% memo()

Calling a memo is exactly like calling a normal function, in fact it is a normal function! The memo has all the same arguments and defaults as the origional function so it can be used in legacy code without the need for any risky refactoring.

> # the first time we call the memo the function will execute
> simple.function(10)
[1] "Executing!"
[1] 10

> # if we call the function again with the same parameter values then
> # the cached value will be returned
> simple.function(10)
[1] 10

> # calling the memo with a different set of parameter values will
> # cause the function to execute
> simple.function(20)
[1] "Executing!"
[1] 20

Memoing a function can significantly improve the performance of a system by limiting how often expensive call are made. Functions that return a NULL value can be memoed by using the allow.null argument.

> # consider a slow function which is memoised, note that we have used the allow.null argument
> # so that NULL is cached when returned from a function, the default is FALSE
> slow.function <- (function (value) Sys.sleep(value)) %>% memo(allow.null = TRUE)

> # the first time we call the slow function it takes some time
> system.time(slow.function(3))
   user  system elapsed 
   0.00    0.00    3.01 

> # subsequent calls make use of the cache and are much faster
> system.time(slow.function(3))
   user  system elapsed 
   0.01    0.00    0.02 

Installation

devtools::install_github("rwetherall/memofunc")