Matrix coefficients of continuous representations #
This file defines the matrix coefficient
g ↦ ⟪π g v, w⟫ of a representation whose operator-valued action is continuous, and develops its
algebra: sesquilinearity in the defining vectors, the matrix-multiplication identity at a product
of group elements, the behaviour under left and right translation of the group argument, the
involution coming from inversion, and the uniform bound for unitary representations.
Mathlib's ContRepresentation bundles continuous linear action operators but does not require
continuity of the map from the acting topological monoid to those operators. The continuity
hypothesis is therefore explicit in matrixCoeff and its API; since it is a Prop argument, two
matrix coefficients built from different continuity proofs are equal by proof irrelevance.
Main definitions #
TauCeti.ContRepresentation.matrixCoeff: the matrix coefficientg ↦ ⟪π g v, w⟫as an element ofC(G, 𝕜).TauCeti.ContRepresentation.matrixCoeffₛₗ: the matrix coefficients of a fixed representation bundled as a sesquilinear mapV →ₗ⋆[𝕜] V →ₗ[𝕜] C(G, 𝕜), in Mathlib'sinnerₛₗconvention.
Main statements #
TauCeti.ContRepresentation.matrixCoeff_apply_mul_eq_sum: in an orthonormal basis, matrix coefficients multiply like matrices,π_{ik}(g * h) = ∑ j, π_{ij}(g) · π_{jk}(h).TauCeti.ContRepresentation.matrixCoeff_comp_mulRightandTauCeti.ContRepresentation.matrixCoeff_comp_mulLeft: right and left translates of a matrix coefficient ofπare again matrix coefficients ofπ. This is theG × G-action that structures the span of the matrix coefficients.TauCeti.ContRepresentation.star_matrixCoeff: the conjugate of a matrix coefficient of a unitary representation is a matrix coefficient composed with inversion.TauCeti.ContRepresentation.matrixCoeff_trivial: the matrix coefficients of the trivial representation are constant.TauCeti.ContRepresentation.norm_matrixCoeff_le: the uniform bound‖v‖ * ‖w‖for a unitary representation.
The algebra of matrix coefficients is the Layer 3 milestone of the
compact-groups roadmap; it is what makes the
span of the matrix coefficients of a fixed representation a translation-stable subspace of C(G),
stable under conjugation followed by inversion of the argument, and it feeds the Schur
orthogonality relations of Layer 4. Conjugation alone leaves that span in general: it produces a
matrix coefficient of the contragredient representation, which is not built here. The mathematical
development follows Daniel Bump, Lie Groups, second edition, Chapters 2–4.
The matrix coefficient g ↦ ⟪π g v, w⟫ of a representation with continuous
operator-valued action.
Equations
- TauCeti.ContRepresentation.matrixCoeff π hπ v w = { toFun := fun (g : G) => inner 𝕜 ((π g) v) w, continuous_toFun := ⋯ }
Instances For
Evaluation of a matrix coefficient.
A matrix coefficient at the identity is the inner product of its defining vectors.
Sesquilinearity in the defining vectors #
Following Mathlib's inner-product convention, a matrix coefficient is conjugate-linear in its first vector and linear in its second.
The matrix coefficients of a fixed representation, bundled as a sesquilinear map: conjugate
linear in the first vector, linear in the second, exactly as Mathlib's innerₛₗ.
Bundling records the sesquilinearity in the form the span of the matrix coefficients is built from,
and supplies the remaining additive identities (map_neg, map_sub, map_sum) through the
LinearMap API.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Translating the group argument #
Translating the argument of a matrix coefficient on the right moves the action onto the first
vector. The pointwise statement needs no continuity of the multiplication of G.
Matrix coefficients multiply like matrices: expanding the intermediate vector π h v in an
orthonormal basis writes the coefficient at a product as a sum of products of coefficients. In the
basis notation π_{ij}(g) = ⟪π g eⱼ, eᵢ⟫ this is π_{ik}(g * h) = ∑ j, π_{ij}(g) · π_{jk}(h).
The trivial representation, and restriction along a homomorphism #
A matrix coefficient of the trivial representation is the constant function at the inner
product of its defining vectors; the trivial action is constant, so continuous_const is its
continuity witness. Which constants arise depends on V: they are the values of the inner product,
so all of 𝕜 when some ⟪v, w⟫ ≠ 0, and only 0 when V is trivial.
A matrix coefficient of an invariant submodule is the matrix coefficient of the ambient
representation at the underlying vectors, so passing to a subrepresentation creates no new
matrix coefficients; continuous_subrepresentation hπ is the continuity witness of the restricted
action.
Restricting a representation along a continuous homomorphism precomposes its matrix
coefficients; the restricted action is π ∘ φ, so hπ.comp hφ is its continuity witness.
The right translate of a matrix coefficient of π is again a matrix coefficient of π: the
translation is absorbed into the first vector.
Translating the argument of a matrix coefficient of a unitary representation on the left moves the action of the inverse onto the second vector.
Evaluating a matrix coefficient of a unitary representation at an inverse conjugates it and
swaps its defining vectors. This is the involution that the span of the matrix coefficients carries;
see star_matrixCoeff for the packaged form.
The left translate of a matrix coefficient of a unitary representation is again a matrix coefficient: the translation is absorbed into the second vector.
The conjugate of a matrix coefficient of a unitary representation is the matrix coefficient
with its vectors swapped, precomposed with inversion. So the span of the matrix coefficients of a
unitary representation is stable under the star operation of C(G, 𝕜) followed by inversion of the
argument.
The uniform norm of a matrix coefficient of a unitary representation is at most the product of the norms of its defining vectors.