Quelle  Normed_Space.thy

(*  Title:      HOL/Hahn_Banach/Normed_Space.thy
  Author: Gertrud Bauer, TU Munich
*)

section ‹Normed vector spaces›

theory Normed_Space
imports Subspace
begin

subsection ‹Quasinorms›

text ‹
  A 🪙‹seminorm› ‹∥⋅∥› is a function on a real vector space into the reals that
  has the following properties: it is positive definite, absolute homogeneous
  and subadditive.
 ›

locale seminorm =
  fixes V :: "'a::{minus, plus, zero, uminus} set"
  fixes norm :: "'a ==> real"    (‹∥_∥›)
  assumes ge_zero [intro?]: "x ∈ V ==> 0 ≤ ∥x∥"
    and abs_homogenous [intro?]: "x ∈ V ==> ∥a ⋅ x∥ = ∣a∣ * ∥x∥"
    and subadditive [intro?]: "x ∈ V ==> y ∈ V ==> ∥x + y∥ ≤ ∥x∥ + ∥y∥"

declare seminorm.intro [intro?]

lemma (in seminorm) diff_subadditive:
  assumes "vectorspace V"
  shows "x ∈ V ==> y ∈ V ==> ∥x - y∥ ≤ ∥x∥ + ∥y∥"
proof -
  interpret vectorspace V by fact
  assume x: "x ∈ V" and y: "y ∈ V"
  then have "x - y = x + - 1 ⋅ y"
    by (simp add: diff_eq2 negate_eq2a)
  also from x y have "∥…∥ ≤ ∥x∥ + ∥- 1 ⋅ y∥"
    by (simp add: subadditive)
  also from y have "∥- 1 ⋅ y∥ = ∣- 1∣ * ∥y∥"
    by (rule abs_homogenous)
  also have "… = ∥y∥" by simp
  finally show ?thesis .
qed

lemma (in seminorm) minus:
  assumes "vectorspace V"
  shows "x ∈ V ==> ∥- x∥ = ∥x∥"
proof -
  interpret vectorspace V by fact
  assume x: "x ∈ V"
  then have "- x = - 1 ⋅ x" by (simp only: negate_eq1)
  also from x have "∥…∥ = ∣- 1∣ * ∥x∥" by (rule abs_homogenous)
  also have "… = ∥x∥" by simp
  finally show ?thesis .
qed


subsection ‹Norms›

text ‹
  A 🪙‹norm› ‹∥⋅∥› is a seminorm that maps only the ‹0› vector to ‹0›.
 ›

locale norm = seminorm +
  assumes zero_iff [iff]: "x ∈ V ==> (∥x∥ = 0) = (x = 0)"


subsection ‹Normed vector spaces›

text ‹
  A vector space together with a norm is called a 🪙‹normed space›.
 ›

locale normed_vectorspace = vectorspace + norm

declare normed_vectorspace.intro [intro?]

lemma (in normed_vectorspace) gt_zero [intro?]:
  assumes x: "x ∈ V" and neq: "x ≠ 0"
  shows "0 < ∥x∥"
proof -
  from x have "0 ≤ ∥x∥" ..
  also have "0 ≠ ∥x∥"
  proof
    assume "0 = ∥x∥"
    with x have "x = 0" by simp
    with neq show False by contradiction
  qed
  finally show ?thesis .
qed

text ‹
  Any subspace of a normed vector space is again a normed vectorspace.
 ›

lemma subspace_normed_vs [intro?]:
  fixes F E norm
  assumes "subspace F E" "normed_vectorspace E norm"
  shows "normed_vectorspace F norm"
proof -
  interpret subspace F E by fact
  interpret normed_vectorspace E norm by fact
  show ?thesis
  proof
    show "vectorspace F"
      by (rule vectorspace) unfold_locales
    have "Normed_Space.norm E norm" ..
    with subset show "Normed_Space.norm F norm"
      by (simp add: norm_def seminorm_def norm_axioms_def)
  qed
qed

end