Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§6.11, Proposition 1 — Structural identities of the pseudoinverse

Proved
VectorSpaceOpt.pseudoinverse_identities

by wenxinzhang · Aug 25, 2026 · Mathlib c5ea003 (Lean v4.30.0)

adjointfunctional-analysishilbert-spaceoperator-identitiespseudoinverse

Let A ⁣:G→HA \colon G \to HA:G→H be continuous linear between real Hilbert spaces with closed range, and let B ⁣:H→GB \colon H \to GB:H→G satisfy IsPseudoinverse A B. Then AAA is the pseudoinverse of BBB, B†B^\daggerB† is the pseudoinverse of A†A^\daggerA†, and

BAB=B,ABA=A,(BA)†=BA.BAB=B,\qquad ABA=A,\qquad (BA)^\dagger=BA.BAB=B,ABA=A,(BA)†=BA.

Moreover, there are pseudoinverses CCC of A†AA^\dagger AA†A and DDD of AA†AA^\daggerAA† with B=CA†=A†DB=CA^\dagger=A^\dagger DB=CA†=A†D. Since BBB is already a continuous linear map, its linearity and boundedness encode items 1–2 of the source proposition; the displayed clauses encode items 3–9. The bundle is the operator-calculus capstone and a reusable API for generalized-inverse reasoning.

Preamble
import Definitions.Def_VectorSpaceOpt_pseudoinverse_predicate

open scoped InnerProduct
Formal statement
namespace VectorSpaceOpt

theorem pseudoinverse_identities
    {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))
    (B : H →L[ℝ] G) (hB : IsPseudoinverse A B) :
    IsPseudoinverse B A ∧
    IsPseudoinverse (A†) (B†) ∧
    (B ∘L A) ∘L B = B ∧
    (A ∘L B) ∘L A = A ∧
    (B ∘L A)† = B ∘L A ∧
    (∃ C : G →L[ℝ] G, IsPseudoinverse ((A†) ∘L A) C ∧ B = C ∘L (A†)) ∧
    (∃ D : H →L[ℝ] H, IsPseudoinverse (A ∘L (A†)) D ∧ B = (A†) ∘L D) := by sorry

end VectorSpaceOpt
Source
David G. Luenberger, Optimization by Vector Space Methods, §6.11, Proposition 1, book p. 165 (PDF physical p. 185), https://sites.science.oregonstate.edu/~show/old/142_Luenberger.pdf
Read-back

What the Lean code literally says, in plain math · gpt-5

Let GGG and HHH be complete real inner-product spaces, let A:G→HA : G → HA:G→H be continuous and real-linear with closed range, and let B:H→GB : H → GB:H→G be continuous and real-linear. For continuous real-linear maps T:U→VT : U → VT:U→V and S:V→US : V → US:V→U between real normed spaces, write P(T,S)P(T,S)P(T,S) for the literal property that, for every v∈Vv ∈ Vv∈V, (i) ‖T(S(v))−v‖≤‖T(u)−v‖‖T(S(v))-v‖ ≤ ‖T(u)-v‖‖T(S(v))−v‖≤‖T(u)−v‖ for every u∈Uu ∈ Uu∈U, and (ii) every u∈Uu ∈ Uu∈U satisfying ‖T(u)−v‖≤‖T(w)−v‖‖T(u)-v‖ ≤ ‖T(w)-v‖‖T(u)−v‖≤‖T(w)−v‖ for every w∈Uw ∈ Uw∈U also satisfies ‖S(v)‖≤‖u‖‖S(v)‖ ≤ ‖u‖‖S(v)‖≤‖u‖; this is the custom predicate IsPseudoinverse⁡(T,S)\operatorname{IsPseudoinverse}(T,S)IsPseudoinverse(T,S) and does not require uniqueness. Assuming P(A,B)P(A,B)P(A,B), all of the following hold simultaneously: P(B,A)P(B,A)P(B,A); P(A†,B†)P(A†,B†)P(A†,B†); BAB=BB A B=BBAB=B; ABA=AA B A=AABA=A; (BA)†=BA(B A)†=B A(BA)†=BA; there exists a continuous real-linear C:G→GC : G → GC:G→G such that P(A†A,C)P(A†A,C)P(A†A,C) and B=CA†B=C A†B=CA†; and there exists a continuous real-linear D:H→HD : H → HD:H→H such that P(AA†,D)P(A A†,D)P(AA†,D) and B=A†DB=A†DB=A†D. Every juxtaposed product here denotes composition with the rightmost map applied first. The witnesses CCC and DDD are not asserted to be unique, and no finite-dimensionality or nontriviality assumption is present.

Human review
  • Endorsed by Shuze Chen · Aug 25, 2026

  • Endorsed by wenxinzhang · Aug 25, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me