Birman–Schwinger Principle¶

Turn a perturbed-operator eigenvalue question into a fixed-threshold eigenvalue question for a resolvent-sandwiched auxiliary operator, preserving eigenspace and counting information under stated hypotheses.

Version: v2 · 2026-09-06 · History
Domain-specific #: 1385
Origin domain: spectral theory
Subdomain: operator theory
Aliases: Birman-Schwinger principle

Core Idea¶

The Birman–Schwinger principle converts the spectral question “is E an eigenvalue of the perturbed operator?” into “does an auxiliary bounded operator have a prescribed eigenvalue?” The conversion factors the perturbation and sandwiches the unperturbed resolvent between those factors. In its canonical attractive self-adjoint form, let

\[ H=H_0-V,\qquad V\geq 0, \]

where H_0 is self-adjoint and bounded below. For a real spectral parameter E below the spectrum of H_0, define

\[ K(E)=V^{1/2}(H_0-E)^{-1}V^{1/2}. \]

When the products are well defined and the required compactness or discreteness hypotheses hold,

\[ \dim\ker(H-E)=\dim\ker(K(E)-I). \]

Thus E is an eigenvalue of H exactly when 1 is an eigenvalue of K(E), with the same geometric multiplicity. In the standard attractive Schrödinger setting, the number of eigenvalues of H not exceeding E also equals the number of eigenvalues of K(E) at least 1, counted with multiplicity.^[1]

The map is concrete. If (H-E)psi=0, then

\[ (H_0-E)\psi=V\psi, \qquad \phi=V^{1/2}\psi, \qquad K(E)\phi=\phi. \]

Conversely, a threshold eigenvector phi recovers an original eigenvector by psi=(H_0-E)^(-1)V^(1/2)phi. The resolvent condition prevents the map from collapsing a nonzero eigenvector into zero. The principle therefore preserves the spectral verdict while changing the operator class and the tools that can be used on it.

The sign +1 belongs to the convention H=H_0-V. If the perturbation is written H=H_0+V and factored as V=V_2^*V_1, the standard auxiliary criterion is commonly -1 for V_1(H_0-z)^(-1)V_2^*. Modern generalized treatments make this convention dependence explicit and provide the additional analytic machinery needed for non-self-adjoint algebraic multiplicities and Jordan chains.^[2]

Structural Signature¶

Sig role-phrases:

the operator pair — the baseline H_0, whose resolvent is known, and the perturbed operator H_0-V in the locked attractive convention
the factored perturbation — V^(1/2) on each side, or a declared two-factor generalization
the parameterized resolvent bridge — an admissible E in the resolvent set of H_0, below its spectrum in the elementary count form, together with (H_0-E)^(-1)
the auxiliary operator family — K(E)=V^(1/2)(H_0-E)^(-1)V^(1/2)
the fixed spectral threshold — eigenvalue 1 under the attractive convention
the eigenspace transport — psi -> V^(1/2)psi and its resolvent-based inverse
the counting transfer — original eigenvalues at or below E become auxiliary eigenvalues at or above 1

The defining invariant is not that the two operators have the same entire spectrum. They generally do not. It is the parameter-linked equivalence

\[ E\in\sigma_p(H) \quad\Longleftrightarrow\quad 1\in\sigma_p(K(E)), \]

together with the corresponding kernel dimension. In the positive compact self-adjoint case, monotonicity of (H_0-E)^(-1) as E rises toward the free spectrum turns individual threshold crossings into an eigenvalue-counting identity. Compactness is a hypothesis to verify, not a consequence of merely writing the sandwich.

What It Is Not¶

Not an assertion that H and K(E) are isospectral. One eigenvalue of one parameter-dependent auxiliary operator encodes one spectral parameter of the original problem.
Not perturbation theory in a small parameter. The equivalence is exact under its hypotheses; no power series or weak-coupling truncation defines it.
Not just the resolvent identity. Resolvent algebra helps construct and analyze the relation, but the principle additionally factors the perturbation and fixes the eigenvalue threshold.
Not automatic compactness. Boundedness, relative compactness, Schatten membership, and integral-kernel estimates require conditions on the baseline operator and perturbation.
Not an unrestricted threshold theorem. At the bottom of essential spectrum the free resolvent may cease to be bounded, and resonances or embedded eigenvalues require limiting or generalized versions.
Not a universal same-multiplicity statement for non-self-adjoint operators. Geometric multiplicity, algebraic multiplicity, and the multiplicity of a zero of an analytic operator-valued function must be distinguished.^[2]
Not the Birman–Schwinger operator itself. K(E) is a constituent; the principle is the equivalence and the deductions it licenses.
Not the Birman–Schwinger bound. A bound is a downstream inequality obtained after the counting function is estimated by trace-ideal or kernel information.

Scope of Application¶

Negative spectrum of Schrödinger operators. For H=-Delta-V with V>=0, E<0, the free resolvent is well defined and the Birman–Schwinger operator is an integral operator. In three dimensions its kernel uses

\[ (-\Delta-E)^{-1}(x,y) =\frac{e^{-\sqrt{|E|}|x-y|}}{4\pi|x-y|}. \]

Potential factors localize this Green kernel. Schatten or Hilbert–Schmidt estimates on the resulting operator yield upper bounds on the number of bound states.^[1] This was the setting of the independent 1961 work of Birman and Schwinger.^[3]^[4]

Eigenvalue inequalities. Once negative eigenvalues are represented as threshold exceedances of a positive compact operator, singular-value and trace-ideal inequalities become available. The technique underlies classical Birman–Schwinger estimates and upper bounds on bound-state counts through trace-ideal or Hilbert–Schmidt control.^[1]

Spectral gaps. Generalized forms count eigenvalues introduced into gaps of an essential spectrum. The elementary one-sided monotonicity formula may be replaced by a difference of counting functions or a spectral-flow statement. A gap application is therefore in scope, but it must carry its own sign and endpoint conventions.

Singular and lower-dimensional interactions. Point interactions, surface-supported delta interactions, and distributional potentials can admit Birman–Schwinger formulations through quadratic forms, trace maps, or self-adjoint extensions. The factor is then not necessarily multiplication by an ordinary function. The one-center delta interaction is a standard exactly solvable case.^[5]

Non-self-adjoint spectral problems. Factorizations V=V_2^*V_1 and analytic operator families extend the technique to complex spectral parameters and non-normal operators. This scope is genuine, but algebraic multiplicity is attached to a zero of an analytic operator-valued function and its Jordan chains, not naively to the frozen compact operator alone.^[2]

Finite-dimensional models. Matrices satisfy the same algebra and are useful for testing signs, eigenspace maps, and threshold counts. They do not exhibit the main analytic gain—turning an unbounded differential operator into a compact integral operator—but they remain literal instances.

Clarity¶

Three declarations make an application unambiguous.

First, state the perturbation sign. This entry locks H=H_0-V, V>=0, so the auxiliary threshold is +1. Changing the sign of the perturbation changes the threshold convention. A derivation that silently borrows K(E) from one convention and the threshold from another is wrong.

Second, state where E lies. The simple version requires E in rho(H_0); the standard Schrödinger count formula takes E<0 because the free Laplacian has spectrum [0,infinity). Setting E=0 by substitution is not licensed when the resolvent becomes singular. One must take a justified limit or use a threshold theorem.

Third, distinguish the three strengths of claim:

E is an eigenvalue of H iff 1 is an eigenvalue of K(E);
the corresponding eigenspaces have the same dimension;
an entire eigenvalue count equals a threshold count.

The third needs positivity, ordering, compactness/discreteness, and the appropriate location of E; it is not merely the first statement repeated.

The most economical recognition test is therefore: Can one identify a baseline resolvent, a factored perturbation, a same-parameter auxiliary operator, a fixed threshold, and a two-way eigenvector map? If any of those roles is absent, the argument may be related resolvent analysis but is not yet the Birman–Schwinger principle.

Manages Complexity¶

The original spectral problem may combine an unbounded differential operator, a difficult domain, and a perturbing potential. Directly finding its negative eigenvalues requires solving an equation in which differentiation, boundary conditions, and the potential act simultaneously. The Birman–Schwinger move isolates the baseline difficulty inside a resolvent whose kernel or mapping properties may already be known, then exposes the perturbation as factors on the two sides.

That re-representation changes the available proof technology. Compact operators have discrete nonzero spectra; positive compact operators admit ordered eigenvalues; Schatten norms and traces control how many eigenvalues can exceed one; integral kernels expose dimension, decay, localization, and singularity. A differential-equation question becomes a threshold-counting question on an operator with matrix-like spectral behavior.

The principle also separates two failure layers. If the eigenvalue-one equivalence is invalid, inspect the factorization, domains, and resolvent set. If the equivalence is valid but an estimate is weak, inspect the chosen norm or kernel bound. This prevents a poor downstream inequality from being misdiagnosed as a failure of the spectral transformation itself.

The auxiliary operator depends on E, so one can study eigenvalue motion rather than solve the original problem afresh at every energy. In the positive self-adjoint setting, its eigenvalues move monotonically as E approaches the free spectral threshold. Bound states appear exactly when an auxiliary eigenvalue crosses one.

Abstract Reasoning¶

Use the following protocol in the locked self-adjoint setting:

Write the perturbed operator as H=H_0-V with V>=0, including the form-domain interpretation if V is not a bounded multiplication operator.
Choose E in the resolvent set of H_0; for the elementary counting theorem, place it below the free spectrum.
Verify that V^(1/2)(H_0-E)^(-1)V^(1/2) is well defined and bounded.
Verify compactness or the weaker spectral condition required by the intended conclusion.
Transport a proposed original eigenvector by phi=V^(1/2)psi and check K(E)phi=phi.
Transport back by psi=(H_0-E)^(-1)V^(1/2)phi; use E in rho(H_0) to prove the maps are inverse on the kernels.
Only after the kernel equivalence is secure, invoke positivity and min–max or coupling monotonicity to transfer an eigenvalue count.
Estimate the auxiliary threshold count with the least lossy available instrument: direct eigenvalues, operator norm, trace, Schatten norm, or kernel inequality.

Several deductions follow. If ||K(E)||<1, then E cannot be an eigenvalue of H; in a monotone region, an appropriate uniform sub-threshold bound can exclude a whole spectral interval. If K(E) is positive compact and has m eigenvalues at least one, then H has m eigenvalues at or below E, under the count theorem's hypotheses. If one auxiliary eigenvalue crosses one as a coupling parameter varies, the corresponding original bound state is born at that crossing. If compactness fails, the eigenvector algebra may remain meaningful while the finite threshold-count conclusion fails—so the lost conclusion identifies the missing hypothesis.

Knowledge Transfer¶

Literal transfer occurs across operator-theoretic spectral settings that preserve every role. Schrödinger, Dirac, waveguide, graph, point-interaction, and spectral-gap problems may change the baseline resolvent and factor space, yet the same sequence remains: choose a spectral parameter, factor the perturbation, sandwich the resolvent, test a fixed auxiliary eigenvalue, and translate multiplicity or counts under the relevant hypotheses.

The most valuable transferable knowledge is diagnostic. A sign mismatch points to the chosen decomposition of H. A bad threshold limit points to the resolvent set. A failure of compactness points to missing decay, localization, or embedding. A non-self-adjoint multiplicity dispute points to the need for analytic operator-family zeros and Jordan-chain machinery. These diagnostics move intact among subfields because they arise from the same operator roles rather than from a metaphor.

Beyond mathematics and mathematical physics, “replace a hard problem by an equivalent easier representation” is only the portable skeleton. It belongs to the live Problem Representation prime. Calling a business reformulation or data transformation “Birman–Schwinger” would be analogy by renaming: it has no resolvent, factored operator perturbation, spectral parameter, or eigenvalue-one test. The named identity stays in spectral theory.

Examples¶

Canonical: a rank-one matrix perturbation¶

Let

\[ H_0=\begin{pmatrix}1&0\\0&4\end{pmatrix},\qquad v=\frac1{\sqrt2}\binom11,\qquad V=4|v\rangle\langle v|. \]

Then

\[ H=H_0-V=\begin{pmatrix}-1&-2\\-2&2\end{pmatrix}. \]

Its characteristic polynomial is x^2-x-6, so its eigenvalues are -2 and 3. Since V has rank one, K(E) has one nonzero eigenvalue:

\[ \mu(E) =4\langle v,(H_0-E)^{-1}v\rangle =2\left(\frac1{1-E}+\frac1{4-E}\right). \]

At E=-2,

\[ \mu(-2)=2\left(\frac13+\frac16\right)=1, \]

exactly detecting the eigenvalue -2 of H. The vector (2,1)^T is an H-eigenvector at -2; applying V^(1/2) maps it to the v direction, on which K(-2) acts as the identity. The counting statement also works: mu(-1)=7/5>1, matching one eigenvalue of H at or below -1, whereas mu(-3)=11/14<1, matching none at or below -3.

Mapped back: the baseline operator is the diagonal H_0; the factored perturbation is the positive rank-one V; the spectral parameter is E; the resolvent is diagonal with entries (1-E)^(-1) and (4-E)^(-1); the auxiliary family has the single active eigenvalue mu(E); the threshold is one; the eigenvector map carries (2,1)^T to span{v}; and the inequalities at -1 and -3 verify the count transfer on both sides of the crossing.

Applied / In Practice: an attractive one-dimensional point interaction¶

Consider the rigorously realized Hamiltonian formally written

\[ H=-\frac{d^2}{dx^2}-g\delta_0,\qquad g>0. \]

The rigorous object is defined through a quadratic form or self-adjoint extension; one does not literally multiply by sqrt(delta). Write a bound energy as E=-kappa^2, kappa>0. The free resolvent kernel on the real line is

\[ G_\kappa(x,y)=\frac{e^{-\kappa|x-y|}}{2\kappa}. \]

Because the interaction is supported at one point, the Birman–Schwinger operator acts on a one-dimensional auxiliary space by the scalar

\[ gG_\kappa(0,0)=\frac{g}{2\kappa}. \]

The threshold equation is therefore

\[ \frac{g}{2\kappa}=1, \qquad \kappa=\frac g2, \qquad E=-\frac{g^2}{4}. \]

The recovered bound-state wave function is proportional to exp(-g|x|/2), the unique decaying solution with the point-interaction jump condition. This is an exactly solvable member of the point-interaction class treated systematically by Albeverio and collaborators.^[5]

Mapped back: the baseline is the free one-dimensional Laplacian; the perturbation is the attractive point interaction of strength g; the spectral parameter is -kappa^2; the resolvent bridge is its Green function; the factor maps a wave function to its value at the interaction point; the auxiliary operator is the scalar g/(2kappa); the fixed threshold is one; and solving the auxiliary scalar equation recovers both the bound energy and the decaying eigenfunction. The example also exposes the form-domain boundary that a naive sqrt(delta) notation would hide.

Structural Tensions¶

T1: Exact equivalence versus estimate quality. The eigenvalue-one correspondence can be exact while the trace, norm, or kernel estimate applied afterward is loose. A weak bound does not refute the principle; it shows that information was lost in the chosen estimate of K(E). Conversely, a sharp operator inequality cannot rescue an invalid factorization. Diagnostic: Has error entered in constructing the equivalence, or only in bounding the auxiliary operator after the equivalence is established?

T2: Unbounded original operator versus compact auxiliary operator. The principal gain comes from changing operator class, but compactness does not come for free. Potentials without sufficient localization, threshold energies, or bad embeddings can leave the auxiliary operator noncompact. The same transformation that promises matrix-like behavior therefore makes its analytic hypotheses load-bearing. Diagnostic: Which specific decay, relative-compactness, or embedding result establishes compactness here?

T3: Fixed sign convention versus portable notation. The attractive form H_0-V yields threshold +1; the additive form H_0+V typically yields -1 after factorization. Both conventions are legitimate, but combining parts of them creates a false spectral criterion. Diagnostic: Can every sign in the auxiliary equation be derived from the declared decomposition of the perturbed operator?

T4: Pointwise spectral equivalence versus global eigenvalue counting. A two-way kernel map needs fewer assumptions than a finite count identity. Positivity, ordering, compactness, and spectral location enter when one moves from “this E is an eigenvalue” to “this many eigenvalues lie below E.” The stronger conclusion is more useful and more fragile. Diagnostic: Is the argument claiming one eigenspace correspondence or an ordered count, and have the additional hypotheses for the latter been verified?

T5: Interior resolvent parameter versus threshold limit. Keeping E in rho(H_0) makes the inverse bounded and the eigenvector maps direct. Approaching essential spectrum is often where the most interesting bound states and resonances occur, but it can destroy that bounded inverse. The method gains physical relevance as its elementary hypotheses weaken. Diagnostic: Is the threshold operator an honest bounded operator at the stated energy, a norm limit, a weak limit, or only formal notation?

T6: Self-adjoint transparency versus non-self-adjoint reach. Positivity and min–max theory make the self-adjoint principle easy to interpret. Non-self-adjoint extensions reach resonances and complex potentials, but geometric and algebraic multiplicities diverge and Jordan chains become operator-family data. Diagnostic: Which multiplicity is being preserved, and is it attached to a frozen operator or to a zero of an analytic family?

T7: Domain autonomy versus structural reduction. Problem Representation captures the portable move from a hard formulation to a more tractable equivalent one, and Eigenvalue/Eigenvector captures the spectral target. Neither supplies the resolvent sandwich, sign-sensitive threshold, or eigenspace transport. Cross-domain reach belongs to the primes; precise spectral diagnosis belongs to this node. Diagnostic: Does the proposed use still require a factored perturbation, baseline resolvent, spectral parameter, and threshold eigenvalue? If not, only the parent skeleton has transferred.

Structural–Framed Character¶

Birman–Schwinger Principle is structural-leaning on the structural–framed spectrum. Its core is a formal equivalence rather than an institutional rule or value-laden judgment, yet its identity cannot escape operator-theoretic spectral vocabulary and remain the named principle.

Evaluative weight points structural. The principle neither approves nor condemns a spectrum. It states when two eigenvalue assertions correspond and what can be counted. Whether a bound state is desirable is external to the mechanism.

Human-practice-bound points structural. Once operators, domains, and potentials are fixed, the spectral relation does not depend on an observer, profession, or convention of social interpretation. Mathematical agents discover and prove the relation; they do not constitute it by adopting it.

Institutional origin points structural. Birman and Schwinger independently developed the technique in 1961, but that historical origin is not a role in the theorem. Removing the research institutions or naming tradition does not change the operator relation.^[3]^[4]

Vocabulary-travels points framed/domain-bound. Resolvent, spectral parameter, perturbation factor, compact operator, eigenspace, and multiplicity remain operative rather than decorative. Stripping them leaves only a generic reformulation pattern and destroys the recognition test.

Import-versus-recognize splits by range. Transfer among Schrödinger, Dirac, point-interaction, gap, and non-self-adjoint spectral settings is recognition of the same operator mechanism under altered hypotheses. Transfer to unrelated problem solving would be import by analogy and belongs to Problem Representation, not to this named principle.

The portable skeleton is one exact re-representation: preserve a target verdict while moving it into a representation with better operations. That skeleton is inherited from Problem Representation. The fixed spectral target is supplied by Eigenvalue/Eigenvector. The resolvent-factorization package is what keeps the child inside spectral theory.

Its character: a strongly structural mathematical mechanism whose recognition remains domain-framed by operator spectra, resolvents, and factorized perturbations.

Structural Core vs. Domain Accent¶

This section decides why Birman–Schwinger Principle is a domain-specific abstraction rather than a prime.

What is skeletal (could lift toward a cross-domain prime). A difficult question is changed into an equivalent question in a representation with better-behaved operations, while a declared verdict is preserved. The input problem, transform, invariant verdict, new representation, and improved tool set form a portable problem-representation skeleton. In this case the original representation is an eigenvalue problem for H; the new one is a threshold problem for K(E); the invariant verdict is existence and kernel dimension; and the improved tools are compact-operator spectra, traces, singular values, and kernels. Problem Representation already owns that cross-domain architecture.

What is domain-bound. Every discriminating role is operator-theoretic: H_0 and its resolvent set, an additive or form perturbation, a factorization, the sandwiched resolvent, a spectral parameter, point and discrete spectrum, eigenspaces, multiplicities, compactness, Schatten ideals, and a sign-sensitive fixed eigenvalue. Remove the resolvent and one has no Birman–Schwinger operator. Remove the factorization and the eigenspace transport disappears. Remove the same-parameter threshold equation and one has only a generic operator comparison. Even “one” is not a generic threshold here: its value is forced by the normalization and perturbation sign.

Why this does not clear the prime bar. A prime's operative vocabulary and recognition test must survive materially unrelated substrates. The thin lesson “re-represent a problem so its invariant becomes easier to inspect” does survive, but that lesson is not distinctively Birman–Schwinger and is already cataloged as Problem Representation. The full identity transfers literally only within spectral and operator theory. Elsewhere one would have to rename ordinary constraints as spectra, ordinary transformations as resolvents, and ordinary pass/fail cutoffs as eigenvalue one. That is analogy, not recognition. The node is autonomous because its in-domain deductions— eigenspace transport, bound-state exclusion, multiplicity preservation, and threshold counts—cannot be reconstructed from the parent primes alone.

Problem Representation. The Birman–Schwinger principle is a strict domain instance: it replaces the original perturbed-operator eigenproblem with a resolvent-sandwiched threshold problem whose operator class exposes different proof operations. It adds a fixed spectral mechanism far beyond the parent's generic encoding choice.
Eigenvalue and Eigenvector. Strictly presupposed rather than a taxonomic genus. The principle transports eigenvectors and multiplicities and tests a fixed auxiliary eigenvalue, but it is a relation between spectral problems rather than an eigenpair itself.
Transformation. A valid broad interpretive lens, declined as a direct structured parent because Problem Representation more precisely captures why this particular transformation is performed and what tractability it changes.
Factorization. A constitutive internal move, declined as a direct parent. Factoring V alone does not entail a resolvent, a parameter-linked operator family, or threshold correspondence.
Threshold. Related but declined. The value one is a spectral test produced by normalization; it is not a generic response-regime threshold.
Perturbation Theory. A neighboring technique, not instantiated. The live prime requires a small parameter and series approximation, neither of which defines the exact Birman–Schwinger relation.

Relationships to Other Abstractions¶

Current abstraction Birman–Schwinger Principle Domain-specific

Parents (2) — more general patterns this builds on

Birman–Schwinger Principle is a kind of Problem Representation Prime

Problem Representation. The Birman–Schwinger principle is a strict domain instance: it replaces the original perturbed-operator eigenproblem with a resolvent-sandwiched threshold problem whose operator class exposes different proof.
Birman–Schwinger Principle presupposes Eigenvalue And Eigenvector Prime

Eigenvalue and Eigenvector. Strictly presupposed rather than a taxonomic genus.

Hierarchy paths (3) — routes to 3 parentless roots

Birman–Schwinger Principle → Problem Representation → Representation → Abstraction

Show alternative paths (2)

Neighborhood in Abstraction Space¶

Birman–Schwinger Principle sits in a sparse region of the domain-specific corpus (76^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Fredholm Kernel — 0.84
Locally Optimal Block Preconditioned Conjugate Gradient — 0.83
Jacobi Method — 0.83
Gelfand–Naimark–Segal construction — 0.82
Quantum Operation — 0.82

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

Birman–Schwinger operator or kernel. This is the auxiliary object K(E), not the equivalence connecting its threshold spectrum to H. Tell: Is the phrase naming one operator family, or the two-way spectral inference built from it?
Birman–Schwinger bound or estimate. This is a downstream inequality for a bound-state count, usually obtained from a trace ideal or integral kernel. Tell: Is equality of spectral verdicts being stated, or only an upper/lower numerical estimate after that equality is used?
Lippmann–Schwinger equation. This is an integral equation for scattering states or wave functions using a Green operator. It can look algebraically similar but does not by itself impose the Birman–Schwinger factor sandwich and fixed auxiliary eigenvalue. Tell: Is the equation solving a state with an incident term, or testing a homogeneous threshold eigenproblem?
Feshbach–Schur map. This eliminates one block of an operator relative to a projection and produces an energy-dependent effective operator on a retained subspace. Tell: Is the construction based on block/projection elimination, or on perturbation factors sandwiching the baseline resolvent?
Fredholm alternative. This distinguishes invertibility from existence of nontrivial kernels for compact perturbations of the identity. It often supports Birman–Schwinger analysis but does not create the original-to- auxiliary spectral map. Tell: Has a perturbed eigenvalue E been tied to a resolvent-sandwiched operator at threshold one?
Resolvent identity. This algebraically relates resolvents of different operators or parameters. Tell: Is there also a perturbation factorization and a two-way eigenvector map, or only an identity among inverses?
Perturbation Theory. This approximates quantities through a series in a small coupling or correction. Tell: Would the method still be exact at finite or strong coupling under the stated operator hypotheses?
Min–max principle. This variationally orders eigenvalues of a self-adjoint operator. It helps prove counting consequences but is not the resolvent transformation. Tell: Is the main object a Rayleigh quotient, or the threshold spectrum of K(E)?
Problem Representation. This is the broader catalog prime for an encoding that changes available solution operations. Tell: Do the resolvent, factored perturbation, spectral parameter, eigenspace map, and fixed eigenvalue remain indispensable?
Correspondence Principle. This relates an older and newer theory in an appropriate limiting regime. Tell: Is the relation between theories across a limit, or between two operator eigenproblems at the same spectral parameter?
Equivalence Principle. This is the gravitational identification of inertial and gravitational behavior. Tell: Are gravity and local frames involved, or a factorized operator perturbation and resolvent?
Generalized Birman–Schwinger principle. This is not a competing concept but a qualified extension with broader factorizations, non-self-adjoint operators, analytic operator-valued zeros, or singular potentials. Tell: Does the claimed multiplicity require Jordan-chain or operator-family machinery beyond the elementary self-adjoint theorem?

References¶

[1] Aleksey Kostenko. Trace Ideals with Applications, section 4.1, “Bound state problems.” Lecture notes. Lemma 4.1.1 and Proposition 4.1.1 state the eigenvalue-one, multiplicity, and counting forms and apply trace/Hilbert–Schmidt estimates. registry ↩a ↩b ↩c

[2] Jussi Behrndt, A. F. M. ter Elst, and Fritz Gesztesy. “The Generalized Birman–Schwinger Principle.” Transactions of the American Mathematical Society 375, no. 2 (2022), 799–845. Gives abstract factorized and non-self-adjoint forms and treats geometric and algebraic multiplicities, analytic operator-valued zeros, and Jordan chains. registry ↩a ↩b ↩c

[3] M. Sh. Birman. “On the spectrum of singular boundary-value problems.” Matematicheskii Sbornik 55(97), no. 2 (1961), 125–174. Original independent source; English translation in American Mathematical Society Translations, Series 2, volume 53 (1966), 23–80. registry ↩a ↩b

[4] Julian Schwinger. “On the Bound States of a Given Potential.” Proceedings of the National Academy of Sciences 47, no. 1 (1961), 122–129. Independent original bound-state treatment. registry ↩a ↩b

[5] Sergio Albeverio, Friedrich Gesztesy, Raphael Høegh-Krohn, and Helge Holden. Solvable Models in Quantum Mechanics. Springer, 1988, “The One-Center δ-Interaction in One Dimension,” pp. 75–90. Authoritative treatment of the point-interaction realization used in the applied example. registry ↩a ↩b