From 9b4b5adcfe74b9701bfdf3a53bbc21274026eb33 Mon Sep 17 00:00:00 2001 From: Ralf Jung <jung@mpi-sws.org> Date: Thu, 3 Aug 2023 17:14:37 +0000 Subject: [PATCH] Improvements by Ralf. --- docs/equalities_and_entailments.md | 26 +++++++++++++++++--------- 1 file changed, 17 insertions(+), 9 deletions(-) diff --git a/docs/equalities_and_entailments.md b/docs/equalities_and_entailments.md index 9a81890b3..7ccae535d 100644 --- a/docs/equalities_and_entailments.md +++ b/docs/equalities_and_entailments.md @@ -53,24 +53,32 @@ Here, stdpp adds the following facilities: `n` arguments of the function `f`. This significantly reduces backtracking during `Proper` search and thus improves performance/avoids diverging failing instance searches. These first arguments typically include type variables - (`A : Type` or `B : A → Type`); type class parameters (`C A`); Leibniz - arguments (`i : nat` or `i : Z`). + (`A : Type` or `B : A → Type`), type class parameters (`C A`), and Leibniz + arguments (`i : nat` or `i : Z`), so they cannot be rewritten or don't need + setoid rewriting. Examples: - + For `cons : ∀ A, A → list A → list A` we have `Params (@cons) 1`. - + For `replicate : ∀ A, nat → A → list A` we have `Params (@replicate) 2`. + + For `cons : ∀ A, A → list A → list A` we have `Params (@cons) 1`, + indicating that the type argument named `A` is not up to rewriting. + + For `replicate : ∀ A, nat → A → list A` we have `Params (@replicate) 2` + indicating that the type argument `A` is not up to rewriting and that the + `nat`-typed argument also doesn't show up as rewriteable in the `Proper` + instance (because rewriting with `=` doesn't need such an instance). + For `lookup : ∀ {Lookup K A M}, K → M → option A` we have `Params (@lookup) 5`: there are 3 Type parameters, 1 type class, and a key (which is Leibniz for all instances). - Consequenently, `Proper .. f` instances are always written in such a way that `f` is partially applied with the first `n` arguments from `Params f n`. + Note that implicit arguments count here. This means that `Proper` instances never start with `(=) ==>`. Examples: + `Proper ((≡@{A}) ==> (≡@{list A}) ==> (≡@{list A})) cons` + (where `cons` is `@cons A`, matching the 1 in `Params`) + `Proper ((≡@{A}) ==> (≡@{list A})) (replicate n)` + (where `replicate n` is `@replicate A n`) + `Proper ((≡@{M}) ==> (≡@{option A})) (lookup k)` -- If the function `f` is not a definition, but a parameter (of a higher order - function), then there is no `Params` instance. However, `Proper` premises - are still written using the convention above. Example: + (where `lookup k` is `@lookup K A M _ k`, so 5 parameters are fixed, matching the `Param`) +- Lemmas about higher-order functions often need `Params` premises. + These are also written using the convention above. Example: ``` Lemma set_fold_ind `{Set A C} {B} (P : B → C → Prop) (f : A → B → B) (b : B) : @@ -78,10 +86,10 @@ Lemma set_fold_ind `{Set A C} {B} (P : B → C → Prop) (f : A → B → B) (b ``` - For premises involving predicates (such as `P` in `set_fold_ind` above), we - always write the weakest proper: that is, use `impl` instead of `iff` (and + always write the weakest `Proper`: that is, use `impl` instead of `iff` (and in Iris, write `(⊢)` instead of `(⊣⊢)`). For "simple" `P`s, there should be instances to solve both `impl` and `iff` using `solve_proper`, and for more - complicated cases where `solve_proper` fails, a `impl` is much easier to + complicated cases where `solve_proper` fails, an `impl` is much easier to prove by hand than an `iff`. ## Equivalences on OFEs -- GitLab