I am reading through through a paper called HMF: Simple Type Inference for First-Class Polymorphism by Daan Leijen of Microsoft Research.

In the paper it describes how to calculate the polymorphic weight of an expression:

extract from pdf

I am having trouble understanding why the above example [[forall a. [forall b. a -> b]]] has a weight of 1 rather than two. there are two type abstractions where both type variable are not free in their bodies, should these not be added together?


You asked why

[[forall a. [forall b. a -> b]]] has a weight of 1

It doesn't. It has a polymorphic weight of 1. The polymorphic weight strips the binder before it looks at the weight. For the $\rho$ you chose:

$[\![\forall \overline{\alpha} . \rho]\!] = wt(\rho) = 1 \neq 2 = wt(\forall \overline{\alpha} . \rho)$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.