Category Theory for Programmers Chapter 2: Types and Functions

  1. Implement memoize 
    const memoize = (f) => {
  const cache = new Map();
  return function() {
    const args = JSON.stringify(arguments);
    if (cache.has(args)) { return cache.get(args); }
    const result = f.apply(null, arguments);
    cache.set(args, result);
    return result;
  }
}
  2. Memoizing Math.random just gives the same number forever. 
    const random = memoize(Math.random)

  3. Math.random doesn’t take a seed, but the above function can give different values with different inputs. Not really sure what the intention here is, but this works.

  4. Only the first function is pure. It gives the same results with no side-effects for the inputs. If memoized, nothing changes. (2) reads input, so can give different results. (3) has side-effects by writing to stdout which can be differnt between calls. (4) has a value that persists between calls so successive calls have different values.

  5. There are 4 functions from Bool to Bool. 
    const id = x => x
const not = x => !x
const alwaysTrue = x => true
const alwaysFalse = x => false

  6. The category with types Void, (), and Bool has the following morphisms:

    • IdVoid: Void → Void
    • absurd: Void → a
    • Id(): () → ()
    • discard: a → ()
    • True(): () → Bool
    • False(): () → Bool
    • IdBool: Bool → Bool
    • NotBool: Bool → Bool
    • TrueBool: Bool → Bool
    • FalseBool: Bool → Bool

    The weird thing here is why there are no morphisms to Void. For this, we fall back to what a morphism is, and here it is a function from a set to a set. This also means, that a function takes every value of the domain to a unique value in the codomain. In the instance of Void, we can define a function with Void as the domain because even though Void is the empty set, we can still define a function for all x in ∅ (even if there are none). However, for a function that does have values, like Bool, for there to be a function to Void, we are saying for all x in {true, false} there exists a unique value in Void such that the relationship holds. However, there is no value in Void so we cannot have a function.

    This differs with () which is a singleton set, eg {👾}. For this we can define functions from () that send 👾 to some other value, and similar, send all values from another domain to just one value.