Here's toy language for 32-bit float computations in OCaml. For example, if you want to compute the n'th harmonic number, you could encode this as: let harmonic : compiled_fn = compile { name ="harmonic"; inputs = ["n"]; body = Sum("i", Num 1.0, Var "n", Div(Num 1.0, Var "i")); } Your function gets compiled to native code using LLVM. You can then call 'harmonic' like this: run harmonic [10.0] Don't use this compiler in a space vehicle. Do use it to learn about LLVM. Cheers, Alex