3 ms·
> The programming language of a CAS should have a type system suitable for doing mathematics. I find it hard to understand what type theorists are actually eve
by astrange 5y ago
> The programming language of a CAS should have a type system suitable for doing mathematics.
I find it hard to understand what type theorists are actually ever talking about because their terminology is too hard to understand (like "intuitionistic type theory" "calculus of constructions" etc).
But looking at practical PL type systems, it's always seemed like they aren't that useful for math.
- for basic engineering math you'd want units like inch vs. centimeter that just add properly, or inch vs square inch that don't add together, but it doesn't seem like those are easy to model with types.
- more abstractly, how do you use types to prove that a matrix multiplication is correct? What even is the type of an intermediate term in a matrix multiplication?
- what’s the type backing a number? Is it a bignum? But now addition is O(n) not O(1) which actually adds a lot of security issues.