rxFun: Add/Create C functions for use in rxode2

In nlmixr2/rxode2: Facilities for Simulating from ODE-Based Models

Add/Create C functions for use in rxode2

Description

Add/Create C functions for use in rxode2

Usage

rxFun(name, args, cCode)

rxRmFun(name)

Arguments

name	This can either give the name of the user function or be a simple R function that you wish to convert to C. If you have rxode2 convert the R function to C, the name of the function will match the function name provided and the number of arguments will match the R function provided. Hence, if you are providing an R function for conversion to C, the rest of the arguments are implied.
args	This gives the arguments of the user function
cCode	This is the C-code for the new function

Examples

# Right now rxode2 is not aware of the function fun
# Therefore it cannot translate it to symengine or
# Compile a model with it.

try(rxode2("a=fun(a,b,c)"))

# Note for this approach to work, it cannot interfere with C
# function names or reserved rxode2 special terms.  Therefore
# f(x) would not work since f is an alias for bioavailability.

fun <- "
double fun(double a, double b, double c) {
  return a*a+b*a+c;
}
" # C-code for function

rxFun("fun", c("a", "b", "c"), fun) ## Added function

# Now rxode2 knows how to translate this function to symengine

rxToSE("fun(a,b,c)")

# And will take a central difference when calculating derivatives

rxFromSE("Derivative(fun(a,b,c),a)")

## Of course, you could specify the derivative table manually
rxD("fun", list(
  function(a, b, c) {
    paste0("2*", a, "+", b)
  },
  function(a, b, c) {
    return(a)
  },
  function(a, b, c) {
    return("0.0")
  }
))

rxFromSE("Derivative(fun(a,b,c),a)")

# You can also remove the functions by `rxRmFun`

rxRmFun("fun")

# you can also use R functions directly in rxode2


gg <- function(x, y) {
  x + y
}

f <- rxode2({
 z = gg(x, y)
})


e <- et(1:10) |> as.data.frame()

e$x <- 1:10
e$y <- 21:30

rxSolve(f, e)

# Note that since it touches R, it can only run single-threaded.
# There are also requirements for the function:
#
# 1. It accepts one value per argument (numeric)
#
# 2. It returns one numeric value

# If it is a simple function (like gg) you can also convert it to C
# using rxFun and load it into rxode2

rxFun(gg)

rxSolve(f, e)

# to stop the recompile simply reassign the function
f <- rxode2(f)

rxSolve(f, e)

rxRmFun("gg")
rm(gg)
rm(f)


# You can also automatically convert a R function to R code (and
# calculate first derivatives)

fun <- function(a, b, c) {
  a^2+b*a+c
}

rxFun(fun)

# You can see the R code if you want with rxC

message(rxC("fun"))

# you can also remove both the function and the
# derivatives with rxRmFun("fun")

rxRmFun("fun")

nlmixr2/rxode2 documentation built on Jan. 11, 2025, 8:48 a.m.