Fix -Wconversion warnings in simulation code.
Cast to smaller integer types explicitly.
Generally avoid platform-dependent types (size_t) in simulation code.
Use float versions of math.h functions, not double.