# Divisibility of natural numbers ```agda module elementary-number-theory.divisibility-natural-numbers where ``` <details><summary>Imports</summary> ```agda open import elementary-number-theory.addition-natural-numbers open import elementary-number-theory.distance-natural-numbers open import elementary-number-theory.inequality-natural-numbers open import elementary-number-theory.multiplication-natural-numbers open import elementary-number-theory.natural-numbers open import elementary-number-theory.strict-inequality-natural-numbers open import foundation.action-on-identifications-functions open import foundation.binary-relations open import foundation.dependent-pair-types open import foundation.empty-types open import foundation.identity-types open import foundation.logical-equivalences open import foundation.negated-equality open import foundation.negation open import foundation.propositional-maps open import foundation.propositions open import foundation.transport-along-identifications open import foundation.type-arithmetic-empty-type open import foundation.unit-type open import foundation.universe-levels ``` </details> ## Idea A natural number `m` is said to **divide** a natural number `n` if there exists a natural number `k` equipped with an identification `km = n`. Using the Curry-Howard interpretation of logic into type theory, we express divisibility as follows: ```text div-ℕ m n := Σ (k : ℕ), k *ℕ m = n. ``` If `n` is a nonzero natural number, then `div-ℕ m n` is always a proposition in the sense that the type `div-ℕ m n` contains at most one element. ## Definitions ```agda div-ℕ : ℕ → ℕ → UU lzero div-ℕ m n = Σ ℕ (λ k → k *ℕ m = n) quotient-div-ℕ : (x y : ℕ) → div-ℕ x y → ℕ quotient-div-ℕ x y H = pr1 H eq-quotient-div-ℕ : (x y : ℕ) (H : div-ℕ x y) → (quotient-div-ℕ x y H) *ℕ x = y eq-quotient-div-ℕ x y H = pr2 H eq-quotient-div-ℕ' : (x y : ℕ) (H : div-ℕ x y) → x *ℕ (quotient-div-ℕ x y H) = y eq-quotient-div-ℕ' x y H = commutative-mul-ℕ x (quotient-div-ℕ x y H) ∙ eq-quotient-div-ℕ x y H div-quotient-div-ℕ : (d x : ℕ) (H : div-ℕ d x) → div-ℕ (quotient-div-ℕ d x H) x pr1 (div-quotient-div-ℕ d x (u , p)) = d pr2 (div-quotient-div-ℕ d x (u , p)) = commutative-mul-ℕ d u ∙ p ``` ### Concatenating equality and divisibility ```agda concatenate-eq-div-ℕ : {x y z : ℕ} → x = y → div-ℕ y z → div-ℕ x z concatenate-eq-div-ℕ refl p = p concatenate-div-eq-ℕ : {x y z : ℕ} → div-ℕ x y → y = z → div-ℕ x z concatenate-div-eq-ℕ p refl = p concatenate-eq-div-eq-ℕ : {x y z w : ℕ} → x = y → div-ℕ y z → z = w → div-ℕ x w concatenate-eq-div-eq-ℕ refl p refl = p ``` ## Properties ### The quotients of a natural number `n` by two natural numbers `p` and `q` are equal if `p` and `q` are equal ```agda eq-quotient-div-eq-div-ℕ : (x y z : ℕ) → is-nonzero-ℕ x → x = y → (H : div-ℕ x z) → (I : div-ℕ y z) → quotient-div-ℕ x z H = quotient-div-ℕ y z I eq-quotient-div-eq-div-ℕ x y z n e H I = is-injective-left-mul-ℕ ( x) ( n) ( tr ( λ p → x *ℕ (quotient-div-ℕ x z H) = p *ℕ (quotient-div-ℕ y z I)) ( inv e) ( commutative-mul-ℕ x (quotient-div-ℕ x z H) ∙ ( eq-quotient-div-ℕ x z H ∙ ( inv (eq-quotient-div-ℕ y z I) ∙ commutative-mul-ℕ (quotient-div-ℕ y z I) y)))) ``` ### Divisibility by a nonzero natural number is a property ```agda is-prop-div-ℕ : (k x : ℕ) → is-nonzero-ℕ k → is-prop (div-ℕ k x) is-prop-div-ℕ k x f = is-prop-map-is-emb (is-emb-right-mul-ℕ k f) x ``` ### The divisibility relation is a partial order on the natural numbers ```agda refl-div-ℕ : is-reflexive div-ℕ pr1 (refl-div-ℕ x) = 1 pr2 (refl-div-ℕ x) = left-unit-law-mul-ℕ x div-eq-ℕ : (x y : ℕ) → x = y → div-ℕ x y div-eq-ℕ x .x refl = refl-div-ℕ x antisymmetric-div-ℕ : is-antisymmetric div-ℕ antisymmetric-div-ℕ zero-ℕ zero-ℕ H K = refl antisymmetric-div-ℕ zero-ℕ (succ-ℕ y) (pair k p) K = inv (right-zero-law-mul-ℕ k) ∙ p antisymmetric-div-ℕ (succ-ℕ x) zero-ℕ H (pair l q) = inv q ∙ right-zero-law-mul-ℕ l antisymmetric-div-ℕ (succ-ℕ x) (succ-ℕ y) (pair k p) (pair l q) = ( inv (left-unit-law-mul-ℕ (succ-ℕ x))) ∙ ( ( ap ( _*ℕ (succ-ℕ x)) ( inv ( is-one-right-is-one-mul-ℕ l k ( is-one-is-left-unit-mul-ℕ (l *ℕ k) x ( ( associative-mul-ℕ l k (succ-ℕ x)) ∙ ( ap (l *ℕ_) p ∙ q)))))) ∙ ( p)) transitive-div-ℕ : is-transitive div-ℕ pr1 (transitive-div-ℕ x y z (pair l q) (pair k p)) = l *ℕ k pr2 (transitive-div-ℕ x y z (pair l q) (pair k p)) = associative-mul-ℕ l k x ∙ (ap (l *ℕ_) p ∙ q) ``` ### If `x` is nonzero and `d | x`, then `d ≤ x` ```agda leq-div-succ-ℕ : (d x : ℕ) → div-ℕ d (succ-ℕ x) → leq-ℕ d (succ-ℕ x) leq-div-succ-ℕ d x (pair (succ-ℕ k) p) = concatenate-leq-eq-ℕ d (leq-mul-ℕ' k d) p leq-div-ℕ : (d x : ℕ) → is-nonzero-ℕ x → div-ℕ d x → leq-ℕ d x leq-div-ℕ d x f H with is-successor-is-nonzero-ℕ f ... | (pair y refl) = leq-div-succ-ℕ d y H leq-quotient-div-ℕ : (d x : ℕ) → is-nonzero-ℕ x → (H : div-ℕ d x) → leq-ℕ (quotient-div-ℕ d x H) x leq-quotient-div-ℕ d x f H = leq-div-ℕ (quotient-div-ℕ d x H) x f (div-quotient-div-ℕ d x H) leq-quotient-div-ℕ' : (d x : ℕ) → is-nonzero-ℕ d → (H : div-ℕ d x) → leq-ℕ (quotient-div-ℕ d x H) x leq-quotient-div-ℕ' d zero-ℕ f (zero-ℕ , p) = star leq-quotient-div-ℕ' d zero-ℕ f (succ-ℕ n , p) = f (is-zero-right-is-zero-add-ℕ _ d p) leq-quotient-div-ℕ' d (succ-ℕ x) f H = leq-quotient-div-ℕ d (succ-ℕ x) (is-nonzero-succ-ℕ x) H ``` ### If `x` is nonzero, if `d | x` and `d ≠ x`, then `d < x` ```agda le-div-succ-ℕ : (d x : ℕ) → div-ℕ d (succ-ℕ x) → d ≠ succ-ℕ x → le-ℕ d (succ-ℕ x) le-div-succ-ℕ d x H f = le-leq-neq-ℕ (leq-div-succ-ℕ d x H) f le-div-ℕ : (d x : ℕ) → is-nonzero-ℕ x → div-ℕ d x → d ≠ x → le-ℕ d x le-div-ℕ d x H K f = le-leq-neq-ℕ (leq-div-ℕ d x H K) f ``` ### `1` divides any number ```agda div-one-ℕ : (x : ℕ) → div-ℕ 1 x pr1 (div-one-ℕ x) = x pr2 (div-one-ℕ x) = right-unit-law-mul-ℕ x div-is-one-ℕ : (k x : ℕ) → is-one-ℕ k → div-ℕ k x div-is-one-ℕ .1 x refl = div-one-ℕ x ``` ### `x | 1` implies `x = 1` ```agda is-one-div-one-ℕ : (x : ℕ) → div-ℕ x 1 → is-one-ℕ x is-one-div-one-ℕ x H = antisymmetric-div-ℕ x 1 H (div-one-ℕ x) ``` ### Any number divides `0` ```agda div-zero-ℕ : (k : ℕ) → div-ℕ k 0 pr1 (div-zero-ℕ k) = 0 pr2 (div-zero-ℕ k) = left-zero-law-mul-ℕ k div-is-zero-ℕ : (k x : ℕ) → is-zero-ℕ x → div-ℕ k x div-is-zero-ℕ k .zero-ℕ refl = div-zero-ℕ k ``` ### `0 | x` implies `x = 0` and `x | 1` implies `x = 1` ```agda is-zero-div-zero-ℕ : (x : ℕ) → div-ℕ zero-ℕ x → is-zero-ℕ x is-zero-div-zero-ℕ x H = antisymmetric-div-ℕ x zero-ℕ (div-zero-ℕ x) H is-zero-is-zero-div-ℕ : (x y : ℕ) → div-ℕ x y → is-zero-ℕ x → is-zero-ℕ y is-zero-is-zero-div-ℕ .zero-ℕ y d refl = is-zero-div-zero-ℕ y d ``` ### Any divisor of a nonzero number is nonzero ```agda is-nonzero-div-ℕ : (d x : ℕ) → div-ℕ d x → is-nonzero-ℕ x → is-nonzero-ℕ d is-nonzero-div-ℕ .zero-ℕ x H K refl = K (is-zero-div-zero-ℕ x H) ``` ### Any divisor of a number at least `1` is at least `1` ```agda leq-one-div-ℕ : (d x : ℕ) → div-ℕ d x → leq-ℕ 1 x → leq-ℕ 1 d leq-one-div-ℕ d x H L = leq-one-is-nonzero-ℕ d (is-nonzero-div-ℕ d x H (is-nonzero-leq-one-ℕ x L)) ``` ### If `x < d` and `d | x`, then `x` must be `0` ```agda is-zero-div-ℕ : (d x : ℕ) → le-ℕ x d → div-ℕ d x → is-zero-ℕ x is-zero-div-ℕ d zero-ℕ H D = refl is-zero-div-ℕ d (succ-ℕ x) H (pair (succ-ℕ k) p) = ex-falso ( contradiction-le-ℕ ( succ-ℕ x) d H ( concatenate-leq-eq-ℕ d ( leq-add-ℕ' d (k *ℕ d)) p)) ``` ### If `x` divides `y` then `x` divides any multiple of `y` ```agda div-mul-ℕ : (k x y : ℕ) → div-ℕ x y → div-ℕ x (k *ℕ y) div-mul-ℕ k x y H = transitive-div-ℕ x y (k *ℕ y) (pair k refl) H div-mul-ℕ' : (k x y : ℕ) → div-ℕ x y → div-ℕ x (y *ℕ k) div-mul-ℕ' k x y H = tr (div-ℕ x) (commutative-mul-ℕ k y) (div-mul-ℕ k x y H) ``` ### A 3-for-2 property of division with respect to addition ```agda div-add-ℕ : (d x y : ℕ) → div-ℕ d x → div-ℕ d y → div-ℕ d (x +ℕ y) pr1 (div-add-ℕ d x y (pair n p) (pair m q)) = n +ℕ m pr2 (div-add-ℕ d x y (pair n p) (pair m q)) = ( right-distributive-mul-add-ℕ n m d) ∙ ( ap-add-ℕ p q) div-left-summand-ℕ : (d x y : ℕ) → div-ℕ d y → div-ℕ d (x +ℕ y) → div-ℕ d x div-left-summand-ℕ zero-ℕ x y (pair m q) (pair n p) = pair zero-ℕ ( ( inv (right-zero-law-mul-ℕ n)) ∙ ( p ∙ (ap (x +ℕ_) ((inv q) ∙ (right-zero-law-mul-ℕ m))))) pr1 (div-left-summand-ℕ (succ-ℕ d) x y (pair m q) (pair n p)) = dist-ℕ m n pr2 (div-left-summand-ℕ (succ-ℕ d) x y (pair m q) (pair n p)) = is-injective-right-add-ℕ (m *ℕ (succ-ℕ d)) ( ( inv ( ( right-distributive-mul-add-ℕ m (dist-ℕ m n) (succ-ℕ d)) ∙ ( commutative-add-ℕ ( m *ℕ (succ-ℕ d)) ( (dist-ℕ m n) *ℕ (succ-ℕ d))))) ∙ ( ( ap ( _*ℕ (succ-ℕ d)) ( is-additive-right-inverse-dist-ℕ m n ( reflects-order-mul-ℕ d m n ( concatenate-eq-leq-eq-ℕ q ( leq-add-ℕ' y x) ( inv p))))) ∙ ( p ∙ (ap (x +ℕ_) (inv q))))) div-right-summand-ℕ : (d x y : ℕ) → div-ℕ d x → div-ℕ d (x +ℕ y) → div-ℕ d y div-right-summand-ℕ d x y H1 H2 = div-left-summand-ℕ d y x H1 (concatenate-div-eq-ℕ H2 (commutative-add-ℕ x y)) ``` ### If `d` divides both `x` and `x + 1`, then `d = 1` ```agda is-one-div-ℕ : (x y : ℕ) → div-ℕ x y → div-ℕ x (succ-ℕ y) → is-one-ℕ x is-one-div-ℕ x y H K = is-one-div-one-ℕ x (div-right-summand-ℕ x y 1 H K) ``` ### Multiplication preserves divisibility ```agda preserves-div-mul-ℕ : (k x y : ℕ) → div-ℕ x y → div-ℕ (k *ℕ x) (k *ℕ y) pr1 (preserves-div-mul-ℕ k x y (pair q p)) = q pr2 (preserves-div-mul-ℕ k x y (pair q p)) = ( inv (associative-mul-ℕ q k x)) ∙ ( ( ap (_*ℕ x) (commutative-mul-ℕ q k)) ∙ ( ( associative-mul-ℕ k q x) ∙ ( ap (k *ℕ_) p))) ``` ### Multiplication by a nonzero number reflects divisibility ```agda reflects-div-mul-ℕ : (k x y : ℕ) → is-nonzero-ℕ k → div-ℕ (k *ℕ x) (k *ℕ y) → div-ℕ x y pr1 (reflects-div-mul-ℕ k x y H (pair q p)) = q pr2 (reflects-div-mul-ℕ k x y H (pair q p)) = is-injective-left-mul-ℕ k H ( ( inv (associative-mul-ℕ k q x)) ∙ ( ( ap (_*ℕ x) (commutative-mul-ℕ k q)) ∙ ( ( associative-mul-ℕ q k x) ∙ ( p)))) ``` ### If a nonzero number `d` divides `y`, then `dx` divides `y` if and only if `x` divides the quotient `y/d` ```agda div-quotient-div-div-ℕ : (x y d : ℕ) (H : div-ℕ d y) → is-nonzero-ℕ d → div-ℕ (d *ℕ x) y → div-ℕ x (quotient-div-ℕ d y H) div-quotient-div-div-ℕ x y d H f K = reflects-div-mul-ℕ d x ( quotient-div-ℕ d y H) ( f) ( tr (div-ℕ (d *ℕ x)) (inv (eq-quotient-div-ℕ' d y H)) K) div-div-quotient-div-ℕ : (x y d : ℕ) (H : div-ℕ d y) → div-ℕ x (quotient-div-ℕ d y H) → div-ℕ (d *ℕ x) y div-div-quotient-div-ℕ x y d H K = tr ( div-ℕ (d *ℕ x)) ( eq-quotient-div-ℕ' d y H) ( preserves-div-mul-ℕ d x (quotient-div-ℕ d y H) K) ``` ### If `d` divides a nonzero number `x`, then the quotient `x/d` is also nonzero ```agda is-nonzero-quotient-div-ℕ : {d x : ℕ} (H : div-ℕ d x) → is-nonzero-ℕ x → is-nonzero-ℕ (quotient-div-ℕ d x H) is-nonzero-quotient-div-ℕ {d} {.(k *ℕ d)} (pair k refl) = is-nonzero-left-factor-mul-ℕ k d ``` ### If `d` divides a number `1 ≤ x`, then `1 ≤ x/d` ```agda leq-one-quotient-div-ℕ : (d x : ℕ) (H : div-ℕ d x) → leq-ℕ 1 x → leq-ℕ 1 (quotient-div-ℕ d x H) leq-one-quotient-div-ℕ d x H K = leq-one-div-ℕ ( quotient-div-ℕ d x H) ( x) ( div-quotient-div-ℕ d x H) ( K) ``` ### `a/a = 1` ```agda is-idempotent-quotient-div-ℕ : (a : ℕ) → is-nonzero-ℕ a → (H : div-ℕ a a) → is-one-ℕ (quotient-div-ℕ a a H) is-idempotent-quotient-div-ℕ zero-ℕ nz (u , p) = ex-falso (nz refl) is-idempotent-quotient-div-ℕ (succ-ℕ a) nz (u , p) = is-one-is-left-unit-mul-ℕ u a p ``` ### If `b` divides `a` and `c` divides `b` and `c` is nonzero, then `a/b · b/c = a/c` ```agda simplify-mul-quotient-div-ℕ : {a b c : ℕ} → is-nonzero-ℕ c → (H : div-ℕ b a) (K : div-ℕ c b) (L : div-ℕ c a) → ( (quotient-div-ℕ b a H) *ℕ (quotient-div-ℕ c b K)) = ( quotient-div-ℕ c a L) simplify-mul-quotient-div-ℕ {a} {b} {c} nz H K L = is-injective-right-mul-ℕ c nz ( equational-reasoning (a/b *ℕ b/c) *ℕ c = a/b *ℕ (b/c *ℕ c) by associative-mul-ℕ a/b b/c c = a/b *ℕ b by ap (a/b *ℕ_) (eq-quotient-div-ℕ c b K) = a by eq-quotient-div-ℕ b a H = a/c *ℕ c by inv (eq-quotient-div-ℕ c a L)) where a/b : ℕ a/b = quotient-div-ℕ b a H b/c : ℕ b/c = quotient-div-ℕ c b K a/c : ℕ a/c = quotient-div-ℕ c a L ``` ### If `d | a` and `d` is nonzero, then `x | a/d` if and only if `xd | a` ```agda simplify-div-quotient-div-ℕ : {a d x : ℕ} → is-nonzero-ℕ d → (H : div-ℕ d a) → (div-ℕ x (quotient-div-ℕ d a H)) ↔ (div-ℕ (x *ℕ d) a) pr1 (pr1 (simplify-div-quotient-div-ℕ nz H) (u , p)) = u pr2 (pr1 (simplify-div-quotient-div-ℕ {a} {d} {x} nz H) (u , p)) = equational-reasoning u *ℕ (x *ℕ d) = (u *ℕ x) *ℕ d by inv (associative-mul-ℕ u x d) = (quotient-div-ℕ d a H) *ℕ d by ap (_*ℕ d) p = a by eq-quotient-div-ℕ d a H pr1 (pr2 (simplify-div-quotient-div-ℕ nz H) (u , p)) = u pr2 (pr2 (simplify-div-quotient-div-ℕ {a} {d} {x} nz H) (u , p)) = is-injective-right-mul-ℕ d nz ( equational-reasoning (u *ℕ x) *ℕ d = u *ℕ (x *ℕ d) by associative-mul-ℕ u x d = a by p = (quotient-div-ℕ d a H) *ℕ d by inv (eq-quotient-div-ℕ d a H)) ``` ### Suppose `H : b | a` and `K : c | b`, where `c` is nonzero. If `d` divides `b/c` then `d` divides `a/c` ```agda div-quotient-div-div-quotient-div-ℕ : {a b c d : ℕ} → is-nonzero-ℕ c → (H : div-ℕ b a) (K : div-ℕ c b) (L : div-ℕ c a) → div-ℕ d (quotient-div-ℕ c b K) → div-ℕ d (quotient-div-ℕ c a L) div-quotient-div-div-quotient-div-ℕ {a} {b} {c} {d} nz H K L M = tr ( div-ℕ d) ( simplify-mul-quotient-div-ℕ nz H K L) ( div-mul-ℕ ( quotient-div-ℕ b a H) ( d) ( quotient-div-ℕ c b K) ( M)) ``` ### If `x` is nonzero and `d | x`, then `x/d = 1` if and only if `d = x` ```agda is-one-quotient-div-ℕ : (d x : ℕ) → is-nonzero-ℕ x → (H : div-ℕ d x) → (d = x) → is-one-ℕ (quotient-div-ℕ d x H) is-one-quotient-div-ℕ d .d f H refl = is-idempotent-quotient-div-ℕ d f H eq-is-one-quotient-div-ℕ : (d x : ℕ) → (H : div-ℕ d x) → is-one-ℕ (quotient-div-ℕ d x H) → d = x eq-is-one-quotient-div-ℕ d x (.1 , q) refl = inv (left-unit-law-mul-ℕ d) ∙ q ``` ### If `x` is nonzero and `d | x`, then `x/d = x` if and only if `d = 1` ```agda compute-quotient-div-is-one-ℕ : (d x : ℕ) → (H : div-ℕ d x) → is-one-ℕ d → quotient-div-ℕ d x H = x compute-quotient-div-is-one-ℕ .1 x (u , q) refl = inv (right-unit-law-mul-ℕ u) ∙ q is-one-divisor-ℕ : (d x : ℕ) → is-nonzero-ℕ x → (H : div-ℕ d x) → quotient-div-ℕ d x H = x → is-one-ℕ d is-one-divisor-ℕ d x f (.x , q) refl = is-injective-left-mul-ℕ x f (q ∙ inv (right-unit-law-mul-ℕ x)) ``` ### If `x` is nonzero and `d | x` is a nontrivial divisor, then `x/d < x` ```agda le-quotient-div-ℕ : (d x : ℕ) → is-nonzero-ℕ x → (H : div-ℕ d x) → ¬ (is-one-ℕ d) → le-ℕ (quotient-div-ℕ d x H) x le-quotient-div-ℕ d x f H g = map-left-unit-law-coproduct-is-empty ( quotient-div-ℕ d x H = x) ( le-ℕ (quotient-div-ℕ d x H) x) ( map-neg (is-one-divisor-ℕ d x f H) g) ( eq-or-le-leq-ℕ ( quotient-div-ℕ d x H) ( x) ( leq-quotient-div-ℕ d x f H)) ```