§6.10, Theorem 1 — Minimum-norm solution through the adjoint
ProvedVectorSpaceOpt.minimum_norm_solution_via_adjointLet be a continuous linear map between real Hilbert spaces with closed range. Fix and . Then solves with minimum norm among all exact solutions if and only if there is such that
The explicit membership y ∈ range A records the solvability implicit in the source's phrase “solution of Ax = y.” This equivalence characterizes the canonical exact solution as lying in the adjoint range and prepares the pointwise minimum-norm choice used to define the pseudoinverse operator.
import Mathlib open scoped InnerProduct
namespace VectorSpaceOpt
theorem minimum_norm_solution_via_adjoint
{G H : Type*} [NormedAddCommGroup G] [InnerProductSpace ℝ G] [CompleteSpace G]
[NormedAddCommGroup H] [InnerProductSpace ℝ H] [CompleteSpace H]
(A : G →L[ℝ] H) (hA : IsClosed (A.range : Set H))
(y : H) (hy : y ∈ A.range) (x : G) :
(A x = y ∧ ∀ u : G, A u = y → ‖x‖ ≤ ‖u‖) ↔
∃ z : H, (A ∘L (A†)) z = y ∧ x = (A†) z := by sorry
end VectorSpaceOptRead-back
What the Lean code literally says, in plain math · gpt-5
Let and be complete real inner-product spaces, let be continuous and real-linear with closed range, let be explicitly assumed to belong to that range, and let be arbitrary. The conjunction that and for every exact solution with holds if and only if there exists some such that and . The witness is only existential and is not asserted to be unique; the statement also covers and trivial spaces.
Confirmed by the mission captain (proposal self-audit).