Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Merged by Bors] - feat(analysis/normed_space/lp_equiv): equivalences between lp and pi_Lp, bounded_continuous_function #15872

Closed
wants to merge 7 commits into from
Closed
Show file tree
Hide file tree
Changes from 5 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
172 changes: 172 additions & 0 deletions src/analysis/normed_space/lp_equiv.lean
Original file line number Diff line number Diff line change
@@ -0,0 +1,172 @@
/-
Copyright (c) 2022 Jireh Loreaux. All rights reserved.
Released under Apache 2.0 license as described in the file LICENSE.
Authors: Jireh Loreaux
-/
import analysis.normed_space.lp_space
import analysis.normed_space.pi_Lp
import topology.continuous_function.bounded

/-!
# Equivalences among $$L^p$$ spaces

In this file we collect a variety of equivalences among various $$L^p$$ spaces. In particular,
when `α` is a `fintype`, given `E : α → Type u` and `p : ℝ≥0∞`, there is a natural linear isometric
equivalence `lp_pi_Lpₗᵢ : lp E p ≃ₗᵢ pi_Lp p E`. In addition, when `α` is a discrete topological
space, the bounded continuous functions `α →ᵇ β` correspond exactly to `lp (λ _, β) ∞`. Here there
can be more structure, including ring and algebra structures, and we implement these equivalences
accordingly as well.

We keep this as a separate file so that the various $$L^p$$ space files don't import the others.

j-loreaux marked this conversation as resolved.
Show resolved Hide resolved
Recall that `pi_Lp` is just a type synonym for `Π i, E i` but given a different metric and norm
structure, although the topological, uniform and bornological structures coincide definitionally.
These structures are only defined on `pi_Lp` for `fintype α`, so there are no issues of convergence
to consider.

While `lp` is also a type synonym for `Π i, E i`, it allows for infinite index types. There is a
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved
predicate `mem_ℓp` which says that the relevant `p`-norm is finite and `lp` is the subtype of those
elements with finite `lp` norm.

## TODO

* Equivalence between `lp` and `measure_theory.Lp`, for `f : α → E` (i.e., functions rather than
pi-types) and the counting measure on `α`

-/

open_locale ennreal

section lp_pi_Lp

variables {α : Type*} {E : α → Type*} [Π i, normed_add_comm_group (E i)] {p : ℝ≥0∞}

/-- When `α` is `finite`, every `f : pre_lp E p` satisfies `mem_ℓp f p`. -/
lemma mem_ℓp.all [finite α] (f : Π i, E i) : mem_ℓp f p :=
begin
rcases p.trichotomy with (rfl | rfl | h),
{ exact mem_ℓp_zero_iff.mpr {i : α | f i ≠ 0}.to_finite, },
{ exact mem_ℓp_infty_iff.mpr (set.finite.bdd_above (set.range (λ (i : α), ∥f i∥)).to_finite) },
{ casesI nonempty_fintype α, exact mem_ℓp_gen ⟨finset.univ.sum _, has_sum_fintype _⟩ }
end

variables [fintype α]

/-- The canonical `equiv` between `lp E p ≃ pi_Lp p E` when `E : α → Type u` with `[fintype α]`. -/
def equiv.lp_pi_Lp : lp E p ≃ pi_Lp p E :=
{ to_fun := λ f, f,
inv_fun := λ f, ⟨f, mem_ℓp.all f⟩,
left_inv := λ f, lp.ext $ funext $ λ x, rfl,
right_inv := λ f, funext $ λ x, rfl }

lemma coe_equiv_lp_pi_Lp (f : lp E p) : equiv.lp_pi_Lp f = f := rfl
lemma coe_equiv_lp_pi_Lp_symm (f : pi_Lp p E) : (equiv.lp_pi_Lp.symm f : Π i, E i) = f := rfl

lemma equiv_lp_pi_Lp_norm (f : lp E p) : ∥equiv.lp_pi_Lp f∥ = ∥f∥ :=
begin
unfreezingI { rcases p.trichotomy with (rfl | rfl | h) },
{ rw [pi_Lp.norm_eq_card, lp.norm_eq_card_dsupport], refl },
{ rw [pi_Lp.norm_eq_csupr, lp.norm_eq_csupr], refl },
{ rw [pi_Lp.norm_eq_sum h, lp.norm_eq_tsum_rpow h, tsum_fintype], refl },
end

/-- The canonical `add_equiv` between `lp E p` and `pi_Lp p E` when `E : α → Type u` with
`[fintype α]` and `[fact (1 ≤ p)]`. -/
def add_equiv.lp_pi_Lp [fact (1 ≤ p)] : lp E p ≃+ pi_Lp p E :=
{ map_add' := λ f g, rfl,
.. equiv.lp_pi_Lp }

@[simp] lemma coe_add_equiv_lp_pi_Lp [fact (1 ≤ p)] (f : lp E p) :
add_equiv.lp_pi_Lp f = f := rfl
@[simp] lemma coe_add_equiv_lp_pi_Lp_symm [fact (1 ≤ p)] (f : pi_Lp p E) :
(add_equiv.lp_pi_Lp.symm f : Π i, E i) = f := rfl
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved

section equivₗᵢ
variables (𝕜 : Type*) [nontrivially_normed_field 𝕜] [Π i, normed_space 𝕜 (E i)]

/-- The canonical `linear_isometry_equiv` between `lp E p` and `pi_Lp p E` when `E : α → Type u`
with `[fintype α]` and `[fact (1 ≤ p)]`. -/
noncomputable def lp_pi_Lpₗᵢ [fact (1 ≤ p)] : lp E p ≃ₗᵢ[𝕜] pi_Lp p E :=
{ map_smul' := λ k f, rfl,
norm_map' := equiv_lp_pi_Lp_norm,
.. add_equiv.lp_pi_Lp }

variables {𝕜}

@[simp] lemma coe_lp_pi_Lpₗᵢ [fact (1 ≤ p)] (f : lp E p) :
lp_pi_Lpₗᵢ 𝕜 f = f := rfl
@[simp] lemma coe_lp_pi_Lpₗᵢ_symm [fact (1 ≤ p)] (f : pi_Lp p E) :
((lp_pi_Lpₗᵢ 𝕜).symm f : Π i, E i) = f := rfl
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved

end equivₗᵢ

end lp_pi_Lp

section lp_bcf

open_locale bounded_continuous_function
open bounded_continuous_function

-- note: `R` and `A` are explicit because otherwise Lean has elaboration problems
variables {α E : Type*} (R A 𝕜 : Type*) [topological_space α] [discrete_topology α]
variables [normed_ring A] [norm_one_class A] [nontrivially_normed_field 𝕜] [normed_algebra 𝕜 A]
variables [normed_add_comm_group E] [normed_space 𝕜 E] [non_unital_normed_ring R]


section normed_add_comm_group

/-- The canonical map between `lp (λ (_ : α), E) ∞` and `α →ᵇ E` as an `add_equiv`. -/
noncomputable def add_equiv.lp_bcf :
lp (λ (_ : α), E) ∞ ≃+ (α →ᵇ E) :=
{ to_fun := λ f, of_normed_add_comm_group_discrete f (∥f∥) $ le_csupr (mem_ℓp_infty_iff.mp f.prop),
inv_fun := λ f, ⟨f, f.bdd_above_range_norm_comp⟩,
left_inv := λ f, lp.ext rfl,
right_inv := λ f, ext $ λ x, rfl,
map_add' := λ f g, ext $ λ x, rfl }

@[simp] lemma coe_add_equiv_lp_bcf (f : lp (λ (_ : α), E) ∞) :
(add_equiv.lp_bcf f : α → E) = f := rfl
@[simp] lemma coe_add_equiv_lp_bcf_symm (f : α →ᵇ E) : (add_equiv.lp_bcf.symm f : α → E) = f := rfl
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved

/-- The canonical map between `lp (λ (_ : α), E) ∞` and `α →ᵇ E` as a `linear_isometry_equiv`. -/
noncomputable def lp_bcfₗᵢ : lp (λ (_ : α), E) ∞ ≃ₗᵢ[𝕜] (α →ᵇ E) :=
{ map_smul' := λ k f, rfl,
norm_map' := λ f, by { simp only [norm_eq_supr_norm, lp.norm_eq_csupr], refl },
.. add_equiv.lp_bcf }

variables {𝕜}

@[simp] lemma coe_lp_bcfₗᵢ (f : lp (λ (_ : α), E) ∞) : (lp_bcfₗᵢ 𝕜 f : α → E) = f := rfl
@[simp] lemma coe_lp_bcfₗᵢ_symm (f : α →ᵇ E) : ((lp_bcfₗᵢ 𝕜).symm f : α → E) = f := rfl
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved

end normed_add_comm_group

section ring_algebra

/-- The canonical map between `lp (λ (_ : α), R) ∞` and `α →ᵇ R` as a `ring_equiv`. -/
noncomputable def ring_equiv.lp_bcf : lp (λ (_ : α), R) ∞ ≃+* (α →ᵇ R) :=
{ map_mul' := λ f g, ext $ λ x, rfl, .. @add_equiv.lp_bcf _ R _ _ _ }

variables {R}
@[simp] lemma coe_ring_equiv_lp_bcf (f : lp (λ (_ : α), R) ∞) :
(ring_equiv.lp_bcf R f : α → R) = f := rfl
@[simp] lemma coe_ring_equiv_lp_bcf_symm (f : α →ᵇ R) :
((ring_equiv.lp_bcf R).symm f : α → R) = f := rfl
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved

variables (α) -- even `α` needs to be explicit here for elaboration

-- the `norm_one_class A` shouldn't really be necessary, but currently it is for
-- `one_mem_ℓp_infty` to get the `ring` instance on `lp`.
/-- The canonical map between `lp (λ (_ : α), A) ∞` and `α →ᵇ A` as an `alg_equiv`. -/
noncomputable def alg_equiv.lp_bcf : lp (λ (_ : α), A) ∞ ≃ₐ[𝕜] (α →ᵇ A) :=
{ commutes' := λ k, rfl, .. ring_equiv.lp_bcf A }

variables {α A 𝕜}
@[simp] lemma coe_alg_equiv_lp_bcf (f : lp (λ (_ : α), A) ∞) :
(alg_equiv.lp_bcf α A 𝕜 f : α → A) = f := rfl
@[simp] lemma coe_alg_equiv_lp_bcf_symm (f : α →ᵇ A) :
((alg_equiv.lp_bcf α A 𝕜).symm f : α → A) = f := rfl
j-loreaux marked this conversation as resolved.
Show resolved Hide resolved

end ring_algebra

end lp_bcf
5 changes: 0 additions & 5 deletions src/analysis/normed_space/lp_space.lean
Original file line number Diff line number Diff line change
Expand Up @@ -51,11 +51,6 @@ say that `∥-f∥ = ∥f∥`, instead of the non-working `f.norm_neg`.
* More versions of Hölder's inequality (for example: the case `p = 1`, `q = ∞`; a version for normed
rings which has `∥∑' i, f i * g i∥` rather than `∑' i, ∥f i∥ * g i∥` on the RHS; a version for
three exponents satisfying `1 / r = 1 / p + 1 / q`)
* Equivalence with `pi_Lp`, for `α` finite
* Equivalence with `measure_theory.Lp`, for `f : α → E` (i.e., functions rather than pi-types) and
the counting measure on `α`
* Equivalence with `bounded_continuous_function`, for `f : α → E` (i.e., functions rather than
pi-types) and `p = ∞`, and the discrete topology on `α`

-/

Expand Down