more progress on proof

This commit is contained in:
Michael Zhang 2024-06-10 22:56:36 -04:00
parent 100db8f8f0
commit be266049f0
10 changed files with 407 additions and 58 deletions

View file

@ -1,79 +1,143 @@
{-# OPTIONS --prop #-}
module Ahmed.Day1 where
open import Data.Empty
open import Data.Unit
open import Data.Bool
open import Data.Nat
open import Data.Fin
open import Data.Product
open import Relation.Nullary using (Dec; yes; no)
import Relation.Binary.PropositionalEquality as Eq
open import Agda.Builtin.Sigma
open import Data.Bool
open import Data.Empty
open import Data.Fin
open import Data.Maybe
open import Data.Nat
open import Data.Product
open import Data.Sum
open import Data.Unit
open import Relation.Nullary
open Eq using (_≡_; refl; trans; sym; cong; cong-app)
open Eq.≡-Reasoning using (begin_; _≡⟨⟩_; step-≡; _∎)
id : {A : Set} A A
id {A} x = x
data type : Set where
bool : type
_-→_ : type type type
data ctx : Set where
nil : ctx zero
_,_ : {n} ctx n type ctx (suc n)
data term : Set where
`_ : term
`true : term
`false : term
`if_then_else_ : term term term term
`λ[_]_ : (τ : type) (e : term) term
_∙_ : term term term
data term : (n : ) Set where
`_ : {n} (m : Fin n) term n
true : {n} term n
false : {n} term n
λ[_]_ : {n} (τ₁ : type) (e : term (suc n)) term n
_∙_ : {n} term n term n term n
data ctx : Set where
nil : ctx
cons : ctx type ctx
postulate
lookup : {n} Fin n ctx n type
-- lookup {n} Γ
lookup : ctx Maybe type
lookup nil _ = nothing
lookup (cons ctx₁ x) zero = just x
lookup (cons ctx₁ x) (suc n) = lookup ctx₁ n
data typing : {n} {Γ : ctx n} term n type Set where
`_ : {n Γ τ} (m : Fin n) lookup m Γ τ typing {n} {Γ} (` m) τ
true : {n Γ} typing {n} {Γ} true bool
false : {n Γ} typing {n} {Γ} false bool
data sub : Set where
nil : sub
cons : sub term sub
data is-value : {n} (e : term n) Set where
true : {n} is-value {n} true
false : {n} is-value {n} false
λ[_]_ : {n} (τ : type) (e : term (suc n)) is-value {n} (λ[ τ ] e)
subst : term term term
subst (` zero) v = v
subst (` suc x) v = ` x
subst `true v = `true
subst `false v = `false
subst (`if x then x₁ else x₂) v = `if (subst x v) then (subst x₁ v) else (subst x₂ v)
subst (`λ[ τ ] x) v = `λ[ τ ] subst x v
subst (x x₁) v = (subst x v) (subst x₁ v)
value = {n} Σ (term n) (is-value)
data value-rel : type term Set where
v-`true : value-rel bool `true
v-`false : value-rel bool `false
v-`λ[_]_ : {τ e} value-rel τ (`λ[ τ ] e)
lift-term : {n} term n term (suc n)
lift-term {n} (` m) = ` (suc m)
lift-term {n} true = true
lift-term {n} false = false
lift-term {n} (λ[ τ₁ ] e) = λ[ τ₁ ] lift-term e
lift-term {n} (e e₁) = lift-term e lift-term e₁
data good-subst : ctx sub Set where
nil : good-subst nil nil
cons : {ctx τ γ v}
good-subst ctx γ
value-rel τ v
good-subst (cons ctx τ) γ
subst : {n} (e : term (suc n)) (v : term n) term n
subst {n} (` zero) v = v
subst {n} (` suc m) v = ` m
subst true v = true
subst false v = false
subst (λ[ τ ] e) v = λ[ τ ] subst e (lift-term v)
subst (e₁ e₂) v = subst e₁ v subst e₂ v
data step : term term Set where
step-if-1 : {e₁ e₂} step (`if `true then e₁ else e₂) e₁
step-if-2 : {e₁ e₂} step (`if `false then e₁ else e₂) e₁
step-`λ : {τ e v} step ((`λ[ τ ] e) v) (subst e v)
data step : {n} term n term n Set where
β-step : {n τ e e₂} step {n} ((λ[ τ ] e) e₂) (subst e e₂)
data steps : term term Set where
zero : {e} steps zero e e
suc : {e e' e''} (n : ) step e e' steps n e' e'' steps (suc n) e e''
-- irreducible : ∀ {n} → (e : term) → ∃[ e' ] mapsto e'
data well-typed : ctx term type Set where
type-`true : {ctx} well-typed ctx `true bool
type-`false : {ctx} well-typed ctx `false bool
type-`ifthenelse : {ctx e e₁ e₂ τ}
well-typed ctx e bool
well-typed ctx e₁ τ
well-typed ctx e₂ τ
well-typed ctx (`if e then e₁ else e₂) τ
type-`x : {ctx x}
(p : Is-just (lookup ctx x))
well-typed ctx (` x) (to-witness p)
type-`λ : {ctx τ τ₂ e}
well-typed (cons ctx τ) e τ₂
well-typed ctx (`λ[ τ ] e) (τ -→ τ₂)
type-∙ : {ctx τ₁ τ₂ e₁ e₂}
well-typed ctx e₁ (τ₁ -→ τ₂)
well-typed ctx e₂ τ₂
well-typed ctx (e₁ e₂) τ₂
-- type-soundness : ∀ {e τ} →
_⊢__ : (Γ : ctx) (e : term) (τ : type) Set
Γ e τ = (well-typed Γ e τ) × ( {n} (e' : term) steps n e e' value-rel τ e' λ e'' step e' e'')
value-relation : {n} (τ : type) (e : term n) Set
expr-relation : {n} (τ : type) (e : term n) Set
irreducible : term Set
irreducible e = ¬ ( λ e' step e e')
value-relation bool true =
value-relation bool false =
value-relation (τ₁ -→ τ₂) ((λ[ τ ] e) v) = expr-relation τ₂ {! !}
value-relation _ _ =
expr-relation {n} τ e = {! !}
data term-relation : type term Set where
e-term : {τ e}
( {n} (e' : term) steps n e e' irreducible e' value-rel τ e')
term-relation τ e
semantic-soundness : {n} (Γ : ctx n) (τ : type) (e : term n) Set
-- semantic-soundness {n} Γ τ e = (γ : substitution n) → expr-relation τ {! !}
_⊨__ : (Γ : ctx) (e : term) (τ : type) Set
_⊨__ Γ e τ = (γ : sub) (good-subst Γ γ) term-relation τ e
fundamental : {Γ e τ} Γ e τ Γ e τ
fundamental {Γ} {`true} {bool} type-sound γ good-sub = e-term λ e' steps irred
[ id , (λ exists ⊥-elim (irred exists)) ] (snd type-sound e' steps)
fundamental {Γ} {`false} {bool} type-sound γ good-sub = e-term λ e' steps irred
[ id , (λ exists ⊥-elim (irred exists)) ] (snd type-sound e' steps)
fundamental {Γ} {`λ[ τ ] e} {τ₁ -→ τ₂} type-sound γ good-sub = e-term f
where
f : {n : } (e' : term) steps n (`λ[ τ ] e) e' irreducible e' value-rel (τ₁ -→ τ₂) e'
f e' steps irred = [ id , (λ exists ⊥-elim (irred exists)) ] (snd type-sound e' steps)
fundamental {Γ} {`if e then e₁ else e₂} {τ} type-sound γ good-sub = e-term f
where
f : {n : } (e' : term) steps n (`if e then e₁ else e₂) e' irreducible e' value-rel τ e'
f .(`if e then e₁ else e₂) zero irred =
let
ts : well-typed Γ (`if e then e₁ else e₂) τ
ts = fst type-sound
ts2 = snd type-sound
in ⊥-elim (irred {! !})
f e' (suc n x steps₁) irred = {! !}
fundamental {Γ} {` x} {τ} type-sound γ good-sub = {! !}
fundamental {Γ} {e e₁} {τ} type-sound γ good-sub = {! !}
-- fundamental {Γ} {`true} {τ -→ τ₁} type-sound γ good-sub = e-term f
-- where
-- el : ∀ {A} → well-typed Γ `true (τ -→ τ₁) → A
-- el ()
-- f : {n : } (e' : term) → steps n `true e' → irreducible e' → value-rel (τ -→ τ₁) e'
-- f e' steps irred = el (fst type-sound)
-- fundamental {Γ} {`false} {τ -→ τ₁} type-sound γ good-sub = e-term f
-- where
-- el : ∀ {A} → well-typed Γ `false (τ -→ τ₁) → A
-- el ()
-- f : {n : } (e' : term) → steps n `false e' → irreducible e' → value-rel (τ -→ τ₁) e'
-- f e' steps irred = el (fst type-sound)

View file

@ -481,4 +481,41 @@ $V_(k, S) db("ref" tau) = { l | l in V_(k-1) db(tau) }$
== Lecture 4
$V_k db(ref tau) = { l | S(L) in V db(tau)}$
$V_k db(ref tau) = { l | S(L) in V db(tau)}$
#pagebreak()
== Lecture 5
Talking about relations between pairs of values.
$V db(tau) = { (v_1, v_2) | ... }$ \
$Epsilon db(tau) = { (e_1, e_2) | ... }$
Proofs of contextual equivalences are difficult to do. FOr example, consider this case:
$lambda x. ... [ ] ...$
Tough to reason about holes inside of a lambda, even with induction on contexts.
Instead, set up a binary logical relation, and prove:
$Gamma tack.r e_1 approx e_2 : tau = forall gamma_1, gamma_2 in G db(tau) ==> (gamma_1 (e_1) , gamma_2 (e_2)) in Epsilon db(tau)$
Not just useful for relations between programs in the same language, but also programs in two different languages.
To do step indexing, you can only count steps on one side. Then in that case, it would only be an approximation relation. To prove equivalence, you would have to show that one side approximates the other and vice versa separately.
Logical relations using step indexing with no types:
$ V_k = &{ "true" , "false" } \
union &{lambda x.e | forall j < k , forall v in V_j => subst(x,v,e) in Epsilon_j} \
$
This is sound because the step index is going down.
https://www.khoury.northeastern.edu/home/amal/papers/next700ccc.pdf
https://www.khoury.northeastern.edu/home/amal/papers/impselfadj.pdf
Matthews-Findler '07: https://users.cs.northwestern.edu/~robby/pubs/papers/popl2007-mf-color.pdf

View file

@ -99,4 +99,12 @@ Compose $M_1$, $M_2$, and $M_3$ into a bigger module.
#tree(
axi[$(m, f("arg"), m') in "enq"_M$],
uni[$m."deq"("arg") --> m'$],
)
)
#pagebreak()
== Lecture 2
- Language Definition
- Compilation to circuits
- Example weak simulation + refinement theorem

4
chong/notes.typ Normal file
View file

@ -0,0 +1,4 @@
#import "../common.typ": *
#import "@preview/prooftrees:0.1.0": *
#show: doc => conf("Language-Based Security", doc)

138
holtzen/notes.typ Normal file
View file

@ -0,0 +1,138 @@
#import "../common.typ": *
#import "@preview/prooftrees:0.1.0": *
#show: doc => conf("Probabilistic Programming", doc)
#import "@preview/simplebnf:0.1.0": *
#let prob = "probability"
#let tt = "tt"
#let ff = "ff"
#let tru = "true"
#let fls = "false"
#let ret = "return"
#let flip = "flip"
#let real = "real"
#let Env = "Env"
#let Bool = "Bool"
#let dbp(x) = $db(#x)_p$
#let dbe(x) = $db(#x)_e$
#let sharpsat = $\#"SAT"$
#let sharpP = $\#"P"$
- Juden Pearl - Probabilistic graphical models
*Definition.* Probabilistic programs are programs that denote #prob distributions.
Example:
```
x <- flip 1/2
x <- flip 1/2
return x if y
```
$x$ is a random variable that comes from a coin flip. Instead of having a value, the output is a _function_ that equals
$
db( #[```
x <- flip 1/2
x <- flip 1/2
return x if y
```]
)
=
cases(tt mapsto 3/4, ff mapsto 1/4) $
tt is "semantic" true and ff is "semantic" false
Sample space $Omega = {tt, ff}$, #prob distribution on $Omega$ -> [0, 1]
Semantic brackets $db(...)$
=== TinyPPL
Syntax
$
#bnf(
Prod( $p$, annot: $sans("Pure program")$, {
Or[$x$][_variable_]
Or[$ifthenelse(p, p, p)$][_conditional_]
Or[$p or p$][_conjunction_]
Or[$p and p$][_disjunction_]
Or[$tru$][_true_]
Or[$fls$][_false_]
Or[$e$ $e$][_application_]
}),
)
$
$
#bnf(
Prod( $e$, annot: $sans("Probabilistic program")$, {
Or[$x arrow.l e, e$][_assignment_]
Or[$ret p$][_return_]
Or[$flip real$][_random_]
}),
)
$
Semantics of pure terms
- $dbe(p) : Env -> Bool$
- $dbe(x) ([x mapsto tt]) = tt$
- $dbe(tru) (rho) = tt$
- $dbe(p_1 and p_2) (rho) = dbe(p_1) (rho) and dbe(p_2) (rho)$
- the second $and$ is a "semantic" $and$
env is a mapping from identifiers to B
Semantics of probabilistic terms
- $dbe(e) : Env -> ({tt, ff} -> [0, 1])$
- $dbe(flip 1/2) (rho) = [tt mapsto 1/2, ff mapsto 1/2]$
- $dbe(ret p) (rho) = v mapsto cases(1 "if" dbp(p) (rho) = v, 0 "else")$
// - $dbe(x <- e_1 \, e_2) (rho) = dbp(e_2) (rho ++ [x mapsto dbp(e_1)])$
- $dbe(x <- e_1 \, e_2) (rho) = v' mapsto sum_(v in {tt, ff}) dbe(e_1) (rho) (v) times db(e_2) (rho [x mapsto v])(v')$
- "monadic semantics" of PPLs
- https://homepage.cs.uiowa.edu/~jgmorrs/eecs762f19/papers/ramsay-pfeffer.pdf
- https://www.sciencedirect.com/science/article/pii/S1571066119300246
Getting back a probability distribution
=== Tractability
What is the complexity class of computing these? #sharpP
- Input: boolean formula $phi$
- Output: number of solutions to $phi$
- $sharpsat(x or y) = 3$
https://en.wikipedia.org/wiki/Toda%27s_theorem
This language is actually incredibly intractable. There is a reduction from TinyPPL to #sharpsat
Reduction:
- Given a formula like $phi = (x or y) and (y or z)$
- Write a program where each variable is assigned a $flip 1/2$:
$x <- flip 1\/2 \
y <- flip 1\/2 \
z <- flip 1\/2 \
ret (x or y) and (y or z)$
How hard is this?
$#sharpsat (phi) = 2^("# vars") times db("encoded program") (emptyset) (tt)$
*Question.* Why do we care about the computational complexity of our denotational semantics?
_Answer._ Gives us a lower bound on our operational semantics.
*Question.* What's the price of adding features like product/sum types?
_Answer._ Any time you add a syntactic construct, it comes at a price.
=== Systems in the wild
- Stan https://en.wikipedia.org/wiki/Stan_(software)
- https://www.tensorflow.org/probability Google
- https://pyro.ai/ Uber

19
stoughton/example1 Normal file
View file

@ -0,0 +1,19 @@
type key. (* encryption keys, key_len bits *)
type text. (* plaintexts, text_len bits *)
type cipher. (* ciphertexts *)
module type ENC = {
proc key_gen(): key
proc enc(k: key, x: text): cipher
proc dec(k: key, c: cipher): text
}
module Cor (Enc: ENC) = {
proc main(x: text): bool = {
var k: key; var c: cipher; var y: text;
k <@ Enc.key_gen();
c <@ Enc.enc(k, x);
y <@ Enc.dec(k, c);
return x = y;
}
}

6
stoughton/notes.typ Normal file
View file

@ -0,0 +1,6 @@
#import "../common.typ": *
#import "@preview/prooftrees:0.1.0": *
#show: doc => conf("The Real/Ideal Paradigm", doc)
https://formosa-crypto.gitlab.io/projects/

30
zdancewic/Day1.agda Normal file
View file

@ -0,0 +1,30 @@
{-# OPTIONS --prop #-}
module Zdancewic.Day1 where
open import Data.Bool
open import Data.Nat
open import Data.String
mem = String ->
infixl 20 _∙_
data aexp : Set where
data bexp : Set where
data com : Set where
`_:=_ : String com
`if_then_else_ : bexp com com com
`skip : com
_∙_ : com com com
beval : mem bexp Bool
denote : com mem mem
denote `skip st = st
denote (`if x then c else c₁) st = if beval st x then denote c st else denote c₁ st
denote (` x := n) st = λ y if x == y then n else st x
denote (c c₁) st = denote c₁ (denote c st)
data ceval : com mem mem Prop where
E_skip : {st} ceval `skip st st

20
zdancewic/Day1.v Normal file
View file

@ -0,0 +1,20 @@
Require Import Coq.Strings.String.
Module Day1.
Inductive aexp : Type :=
.
Inductive bexp : Type :=
.
Inductive com : Type :=
| CSkip
| CAssign (x : string) (a : aexp)
| CSeq (c1 c2 : com)
| CIf (b : bexp) (c1 c2 : com)
| CWhile (b : bexp) (c : com)
.
Fixpoint aeval (m : mem) (a : aexp) : nat :=
.

23
zdancewic/notes.typ Normal file
View file

@ -0,0 +1,23 @@
#import "../common.typ": *
#import "@preview/prooftrees:0.1.0": *
#show: doc => conf("Steve Zdancewic", doc)
- Reasoning about monadic programs in Coq
-
Imp
```
Inductive com : Type :=
| CSkip
| CAssign (x : string) (a : aexp)
| CSeq (c1 c2 : com)
| CIf (b : bexp) (c1 c2 : com)
| CWhile (b : bexp) (c : com)
.
```
"Relationally defined operational semantics"
State monad