The universal property of the model space C₀(I, R) #
Let R be a seminormed ring, I a discrete index type and M a complete ultrametric normed
R-module with contractive scalar action. A bounded family e : I → M, that is an element of
lp (fun _ : I ↦ M) ∞, defines a continuous linear map
ZeroAtInftyContinuousMap.linearCombination R e : C₀(I, R) →L[R] M, a ↦ ∑' i, a i • e i.
The series converges because its terms tend to zero along the cofinite filter, which suffices in
a complete ultrametric group, and the ultrametric inequality bounds its sum by ‖e‖ * ‖a‖. The
map sends the coordinate vector single i 1 to e i, and over a norm-one ring its operator norm
is the supremum ‖e‖ of the norms of the e i. None of this needs a pseudo-uniformiser.
Conversely, over a ring admitting a multiplicative pseudo-uniformiser
(TauCeti.NormedRing.IsTate R) every continuous linear map u : C₀(I, R) →L[R] M has a uniform
bound, so the family i ↦ u (single i 1) is bounded and u is its linear combination map. Over a
commutative such ring this gives the linear isometric equivalence
linearCombinationEquiv R M : lp (fun _ : I ↦ M) ∞ ≃ₗᵢ[R] (C₀(I, R) →L[R] M)
for the operator norm of the scope TauCeti.RingOpNorm. Over a nontrivially normed field the
norm identity is stated for Mathlib's operator norm (opNorm_linearCombination).
This is the basic way to construct continuous linear maps out of the model spaces C₀(I, R) of
nonarchimedean functional analysis, for instance from bounded lifts of the coordinate vectors.
Main definitions #
ZeroAtInftyContinuousMap.linearCombination: the continuous linear mapC₀(I, R) →L[R] Mattached to a bounded family inM.ZeroAtInftyContinuousMap.linearCombinationEquiv: over a commutative ring with a multiplicative pseudo-uniformiser, bounded families correspond isometrically to continuous linear maps.
Main results #
ZeroAtInftyContinuousMap.hasSum_linearCombination: the unconditional expansion∑ i, a i • e i = linearCombination R e a.ZeroAtInftyContinuousMap.linearCombination_single: the value on a coordinate vector.ZeroAtInftyContinuousMap.eq_linearCombination_iff: a continuous linear map is the linear combination map of a bounded family exactly when it sends each unit coordinate vector to the corresponding family member.ZeroAtInftyContinuousMap.linearCombination_eq_tsum_mul: for scalar-valued families, the pairing∑' i, b i * a i.ZeroAtInftyContinuousMap.ringOpNorm_linearCombination: over a norm-one ring the operator norm is the family supremum.ZeroAtInftyContinuousMap.linearCombination_injectiveandZeroAtInftyContinuousMap.linearCombination_self: bounded families are determined by their linear combination maps, and over Tate scalars every continuous linear map is one.
References #
- P. Schneider, Nonarchimedean Functional Analysis, Springer, 2002, §10.
- J.-P. Serre, Endomorphismes complètement continus des espaces de Banach p-adiques, Publ. Math. IHÉS 12 (1962), 69–85.
The ultrametric inequality bounds ∑' i, a i • e i by ‖e‖ * ‖a‖ for a family a vanishing
at infinity and a bounded family e.
On a discrete index, the terms a i • e i of a family a vanishing at infinity against a
bounded family e tend to zero along the cofinite filter.
In a complete ultrametric module, the series ∑' i, a i • e i is summable for every
a ∈ C₀(I, R) and every bounded family e.
The linear combination map of a bounded family e in a complete ultrametric module: the
continuous linear map C₀(I, R) →L[R] M sending a to the unconditional sum ∑' i, a i • e i.
It sends the coordinate vector single i 1 to e i.
Instances For
The value of the linear combination map is the sum ∑' i, a i • e i.
Unconditional expansion: linearCombination R e a is the unconditional sum of the
a i • e i.
The linear combination map sends the coordinate vector single i r to r • e i.
The linear combination map of a bounded family is the unique continuous linear map
sending each unit coordinate vector single i 1 to e i.
Pairing the coordinate family lp.single ∞ i m with a gives a i • m.
The linear combination map is bounded by the supremum of the family.
The operator norm of the linear combination map is at most the supremum of the family.
The norm of a linear combination map: over a norm-one ring, the operator norm of
linearCombination R e is the supremum ‖e‖ = ⨆ i, ‖e i‖ of the family.
The zero family has the zero linear combination map.
The linear combination map is additive in the family.
A bounded family is determined by its linear combination map.
Over a nontrivially normed field, the operator norm of linearCombination K e for Mathlib's
operator norm is the supremum of the family.
Over Tate scalars, the values of a continuous linear map u out of C₀(I, R) at the unit
coordinate vectors form a bounded family, bounded by the operator norm of u.
Every continuous linear map out of C₀(I, R) is a linear combination map over Tate
scalars: u is the linear combination map of its values at the unit coordinate vectors.
Over Tate scalars, every continuous linear map out of C₀(I, R) is the linear combination map
of a bounded family.
Over a commutative ring, the linear combination map is linear in the family.
The pairing of a bounded family b with a ∈ C₀(I, R) is the convergent sum
∑' i, b i * a i.
The universal property of C₀(I, R). Over a commutative ring admitting a multiplicative
pseudo-uniformiser, bounded families I → M in a complete ultrametric module correspond to
continuous linear maps C₀(I, R) →L[R] M, isometrically for the sup norm and the operator norm.
A family e is sent to its linear combination map a ↦ ∑' i, a i • e i, and a map u to its
values u (single i 1) at the unit coordinate vectors.
Instances For
The universal property sends a bounded family to its linear combination map.
The inverse of the universal property evaluates a map at the unit coordinate vectors.