Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 6.4 — Kato–Rellich (self-adjoint case, with the lower bound (6.3))

Proved
TeschlQM.KatoRellich.kato_rellich

by mikedeng1 · Sep 28, 2026 · Mathlib 0df444a (Lean v4.33.1)

functional-analysiskato-rellichp2o-batch-books5p2o-gran-per-chapterp2o-plan-bookp2o-v1perturbation-theoryself-adjoint-operators

Let AAA be a self-adjoint operator and BBB a symmetric operator in a complex Hilbert space H\mathfrak{H}H, and suppose the AAA-bound of BBB is less than one. Then A+BA + BA+B, with domain D(A+B)=D(A)\mathfrak{D}(A + B) = \mathfrak{D}(A)D(A+B)=D(A), is self-adjoint.

If moreover AAA is bounded from below by γ\gammaγ, and a∈[0,1)a \in [0,1)a∈[0,1), b≥0b \ge 0b≥0 are constants for which (6.1) holds, then A+BA + BA+B is bounded from below by

γ−max⁡(a∣γ∣+b,  b1−a).\gamma - \max\Big(a|\gamma| + b,\; \frac{b}{1 - a}\Big).γ−max(a∣γ∣+b,1−ab​).

This is the basic perturbation theorem for self-adjoint operators: it gives self-adjointness of Schrödinger operators −Δ+V-\Delta + V−Δ+V for potentials that are small relative to the kinetic energy.

Formalization Note. The book prints the lower bound (6.3) as γ−max⁡(a∣γ∣+b,b/(a−1))\gamma - \max(a|\gamma| + b, b/(a-1))γ−max(a∣γ∣+b,b/(a−1)). Because b/(a−1)≤0b/(a-1) \le 0b/(a−1)≤0, that expression equals γ−a∣γ∣−b\gamma - a|\gamma| - bγ−a∣γ∣−b, and this is false: in C2\mathbb{C}^2C2 with A=diag⁡(0,6.57)A = \operatorname{diag}(0, 6.57)A=diag(0,6.57) (γ=0\gamma = 0γ=0), BBB the real symmetric matrix with entries −4.00,3.32;3.32,−9.10-4.00, 3.32; 3.32, -9.10−4.00,3.32;3.32,−9.10, and a=0.931a = 0.931a=0.931, the best bbb is ≈5.97\approx 5.97≈5.97 while the lowest eigenvalue of A+BA + BA+B is ≈−6.66<−b\approx -6.66 < -b≈−6.66<−b. The statement uses b/(1−a)b/(1 - a)b/(1−a), which is Kato's constant (Perturbation Theory for Linear Operators, Thm V.4.11). "AAA-bound less than one" is relativeBound A B < 1 in [0,∞][0,\infty][0,∞], which includes relative boundedness. The sum A+BA + BA+B is Mathlib's sum of LinearPMaps, defined on D(A)∩D(B)\mathfrak{D}(A) \cap \mathfrak{D}(B)D(A)∩D(B); the first conjunct asserts that this is D(A)\mathfrak{D}(A)D(A).


Retired 2026-10-27: a Lean-vs-paper fidelity audit found this statement weaker than, or narrower than, the cited result. A corrected statement will be resubmitted.

Preamble
import Mathlib
import Definitions.Def_TeschlQM_Shared_IsSymmetric
import Definitions.Def_TeschlQM_KatoRellich_IsRelativelyBounded
import Definitions.Def_TeschlQM_Shared_IsBoundedBelowBy
Formal statement
namespace TeschlQM.KatoRellich

open scoped ENNReal

/-- Teschl, Theorem 6.4 (Kato–Rellich), p. 135, self-adjoint case with the lower bound (6.3).
If `A` is self-adjoint and `B` is symmetric with `A`-bound less than one, then `A + B` with
`𝔇(A + B) = 𝔇(A)` is self-adjoint. If moreover `A ≥ γ` and (6.1) holds with constants `a < 1`, `b`,
then `A + B ≥ γ - max(a|γ| + b, b/(1 - a))`. The book prints `b/(a - 1)`, which is false
(see the moderation notes); `b/(1 - a)` is the corrected constant (Kato, Thm V.4.11). -/
theorem kato_rellich {H : Type*} [NormedAddCommGroup H] [InnerProductSpace ℂ H] [CompleteSpace H]
    (A B : H →ₗ.[ℂ] H) (hA : IsSelfAdjoint A) (hB : TeschlQM.Shared.IsSymmetric B)
    (hbound : relativeBound A B < 1) :
    (A + B).domain = A.domain ∧ IsSelfAdjoint (A + B) ∧
      ∀ γ a b : ℝ, a < 1 → IsRelativelyBoundedWith A B a b → TeschlQM.Shared.IsBoundedBelowBy A γ →
        TeschlQM.Shared.IsBoundedBelowBy (A + B) (γ - max (a * |γ| + b) (b / (1 - a))) := by sorry

end TeschlQM.KatoRellich
Source
Teschl, Mathematical Methods in Quantum Mechanics, AMS GSM 99, 2009, p. 135, Theorem 6.4, Eq. (6.3)
Read-back

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

Let HHH be a complex Hilbert space: a complex inner-product space that is complete in its norm. HHH may be any such space, including the zero space {0}\{0\}{0}. Let AAA and BBB be partially defined linear operators on HHH. Each has a domain, which is a complex linear subspace D(A)\mathcal D(A)D(A) or D(B)\mathcal D(B)D(B) of HHH, and is a C\mathbb CC-linear map from that domain into HHH. The statement does not assume that either domain is dense or closed.

The sum A+BA+BA+B is the operator with domain

D(A+B)=D(A)∩D(B),(A+B)x=Ax+Bx.\mathcal D(A+B) = \mathcal D(A)\cap\mathcal D(B), \qquad (A+B)x = Ax + Bx .D(A+B)=D(A)∩D(B),(A+B)x=Ax+Bx.

Hypotheses:

  1. AAA is self-adjoint: A∗=AA^{*}=AA∗=A as partially defined operators, so the domains and the values are equal. Here A∗A^{*}A∗ is the library's adjoint of a partially defined operator. Its domain is the set of yyy for which x↦⟨y,Ax⟩x\mapsto\langle y, Ax\ranglex↦⟨y,Ax⟩ is continuous on D(A)\mathcal D(A)D(A). On that domain it acts as the usual adjoint when D(A)\mathcal D(A)D(A) is dense, and as the zero map when D(A)\mathcal D(A)D(A) is not dense.
  2. BBB is symmetric: BBB satisfies the imported predicate TeschlQM.Shared.IsSymmetric. Its definition is not shown in the code given here, so its exact content cannot be stated from this code.
  3. Relative bound below 1: relbd⁡(A,B)<1\operatorname{relbd}(A,B) < 1relbd(A,B)<1, where relbd⁡(A,B)\operatorname{relbd}(A,B)relbd(A,B) is the imported quantity relativeBound A B. Its definition and value type are not shown in the code; the extended non-negative reals [0,∞][0,\infty][0,∞] are brought into scope, so it may take values there.

Conclusion. The statement asserts all three of the following:

  • (i) Domain: D(A+B)=D(A)\mathcal D(A+B)=\mathcal D(A)D(A+B)=D(A), which is the same as saying D(A)⊆D(B)\mathcal D(A)\subseteq\mathcal D(B)D(A)⊆D(B).

  • (ii) Self-adjointness: A+BA+BA+B is self-adjoint in the same sense as hypothesis 1, that is, (A+B)∗=A+B(A+B)^{*}=A+B(A+B)∗=A+B.

  • (iii) Lower bound: for every triple of real numbers γ,a,b\gamma, a, bγ,a,b with

    • a<1a<1a<1,
    • AAA and BBB satisfying the imported predicate IsRelativelyBoundedWith A B a b (whose definition is not shown), and
    • AAA satisfying the imported predicate IsBoundedBelowBy A γ (whose definition is not shown),

    the operator A+BA+BA+B satisfies IsBoundedBelowBy with the constant

γ  −  max⁡ ⁣(a ∣γ∣+b,    b1−a).\gamma \;-\; \max\!\Big(a\,|\gamma| + b,\;\; \frac{b}{1-a}\Big).γ−max(a∣γ∣+b,1−ab​).

Nothing in part (iii) requires a≥0a\ge 0a≥0, b≥0b\ge 0b≥0 or γ≥0\gamma\ge 0γ≥0, and aaa may be any real number below 111. Part (iii) contains no reference to relbd⁡(A,B)\operatorname{relbd}(A,B)relbd(A,B). Its only link to the global hypotheses is that it sits inside the same conclusion.

Degenerate cases:

  • Division: because a<1a<1a<1, the denominator 1−a1-a1−a is strictly positive, so the division never falls back on a default value.
  • Negative constants: if negative aaa or bbb satisfy the relative-boundedness predicate, the maximum can be negative. The asserted lower bound for A+BA+BA+B would then be strictly larger than γ\gammaγ. Whether such aaa and bbb can occur depends on the predicate's definition, which is not shown.
  • Zero space: if H={0}H=\{0\}H={0}, every domain is {0}\{0\}{0} and the operators are zero. Conclusions (i) and (ii) then hold trivially, and (iii) reduces to whatever the imported predicates say about the zero operator.
  • Non-dense domain: if D(A)\mathcal D(A)D(A) is not dense, hypothesis 1 compares AAA with the library's zero-valued default adjoint.
  • Vacuous case: if the imported predicates can never hold together for a given (γ,a,b)(\gamma,a,b)(γ,a,b), part (iii) is vacuous for that triple.
Human review
  • Endorsed by Shuze Chen · Oct 2, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 2, 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