Fpu_renameSourceAs described in the Fpu module documentation, there are problems when mixing some C-lib or ocaml native functions with interval programming on 64 bits machine.
The standard floating point functions results will always lie in the low; high interval computed by the Fpu module, but they are slightly different on 32 and 64 bits machines.
Using Open Fpu.Rename at the beginning of your program guarantees that floating computation will give the same results on 32 and 64 bits machines for all transcendantal functions but not for ordinary arithmetic functions.
NB: while most transcendantal function are almost as fast, and sometimes faster than their "standard" ocaml counterparts, +., -., *. and /. are much slower (from 50% to 100% depending on the processor). If you want to rename also +., -., *. and /. then use the Fpu.Rename_all module.
Computes x mod y
square root function
log function
exp function
Computes x^y
Computes cos(x) for x in [-2^63, 2^63]
Computes sin(x) for x in [-2^63, 2^63]
Computes tan(x) for x in [-2^63, 2^63]
arc-sinus function
arc-cosine function
atan2 function
arc-tan function
cosh function
sinh function
tanh function