Some code optimizations are carried out on the intermediate code because

Answer: they enhance the portability of the compiler to other target processors

Explanation: Optimizing on an intermediate representation (IR) allows optimization passes to be written in a machine-independent way. The same IR-based optimizations can be reused when retargeting the compiler to different processors, which improves portability and reduces duplication of effort.

• IR decouples optimizations from target-specific details, so optimizations need not be reimplemented for each backend.

• IR can retain high-level information useful for many optimizations while remaining independent of machine instruction sets.

• After machine-independent optimizations on the IR, backends can perform target-specific optimizations on generated machine code.

Why the other choices are not correct:

• Program analysis is sometimes easier on IR, but saying analysis is always more accurate there is misleading; the main goal is portability.

• Dataflow information can be used for optimization at multiple stages; the advantage of IR is convenience and reusability, not exclusivity.

• Front-end information can be attached to the IR and used for optimization, so it is not true that front-end information "cannot otherwise be used".