Kreĭn–Saakyan L-Resolvent Matrix Theory

• The topic defines the Kreĭn–Saakyan theory as a framework that characterizes all generalized resolvents using gauge subspaces and analytic L-resolvent matrices.
• It employs a rigged space structure and a linear-fractional transformation over Nevanlinna families to unify self-adjoint extension theory and boundary triple formalism.
• The approach leads to explicit constructions of spectral and pseudo-spectral functions, with broad applications to differential operators and canonical systems.

The Kreĭn–Saakyan theory of $\sL$-resolvent matrices characterizes and parametrizes all generalized resolvents of symmetric operators and linear relations in Hilbert and Pontryagin spaces, including the construction of explicit analytic matrix-valued functions governing generalized spectral theory. Central to the theory is the concept of a gauge subspace $\sL$ (proper or improper) and the associated $\sL$-resolvent matrix $W(z)$, which encodes all generalized resolvents via a linear-fractional transformation over Nevanlinna parameter families. This approach unifies self-adjoint extension theory, Nevanlinna-type holomorphic operator functions, rigged Hilbert (or Pontryagin) space constructions, and boundary triple formalism, enabling a fully explicit description of spectral and pseudo-spectral functions for symmetric relations, with key applications to canonical systems and differential operators (Derkach, 4 Dec 2025, Derkach, 20 May 2025, Mogilevskii, 2014).

1. Rigged Space Framework and Gauge Subspaces

A closed symmetric linear relation $A$ in a separable Hilbert space $\sH$ with finite and equal deficiency indices $p<\infty$ is naturally situated in a triplet (rigged chain):

$\sH_+ \subset \sH \subset \sH_-$

where $\sH_+ = \dom(A^*)$ is equipped with the graph norm, and $\sH_-$ is its dual. For $A$ in a Pontryagin space, analogous structures apply, with special attention to the indefinite inner product and finite negative index.

A subspace $\sL\subset \sH_-$ of dimension $p$ is called a gauge of $A$ if, for some $z$ in the complex plane, the dual space decomposes as

$\sH_- = \operatorname{ran}(A-zI) \dotplus \sL$

If $\sL\subset \sH$ this is a proper gauge; if $\sL \cap (\sH_- \setminus \sH) \neq \emptyset$ it is an improper gauge. The set of such $z$ is denoted $\rho(A, \sL)$, and the projection onto $\sL$ parallel to $\operatorname{ran}(A-zI)$ is denoted $\Pi_\sL^z$ (Derkach, 4 Dec 2025, Derkach, 20 May 2025).

This rigged space structure underlies the extension of the classical Kreĭn–Saakyan formula from operators to linear relations and from proper to improper gauges, especially in the context of boundary value problems and the spectral theory of differential systems.

2. $\sL$-Resolvent Matrices and Linear-Fractional Parametrization

Given a self-adjoint extension $\tilde{A}$ of $A$ in some (possibly larger) Hilbert or Pontryagin space, and the canonical resolvent $\widetilde{R}_z = (\widetilde{A} - zI)^{-1}$, the (generalized) compressed resolvent defined on $\sH$ is

$R_z = P_\sH \widetilde{R}_z |_\sH$

where $P_\sH$ is the orthogonal projection. One regularizes at $z=i$ by defining $R^0 = \frac{1}{2}(R_i + R_{-i})$, and then

$\widehat{R}_z = R_z - R^0$

Given an invertible $L : \mathbb{C}^p \to \sL \subset \sH_-$, the operator-valued function

$r(z) := L^* \widehat{R}_z L$

is called the $\sL$-resolvent of $A$. For $z$ in $\rho(A, \sL)$, $r(z)$ is holomorphic and belongs to the Nevanlinna class $\mathcal{R}^p$, characterized by symmetry $r(z^*) = r(z)^*$ and kernel nonnegativity $$K_r(z, w) := \frac{r(z) - r(w)^*}{z - \bar{w}} \geq 0$$ for $z, w \in \mathbb{C}_+$.

All $\sL$-resolvents arise from a fixed analytic $2p \times 2p$ $\sL$-resolvent matrix $W(z)$ through the linear-fractional transformation

$$r(z) = T_{W(z)}[\tau(z)] = (w_{11}(z)\, \tau(z) + w_{12}(z))\,(w_{21}(z)\, \tau(z) + w_{22}(z))^{-1}$$

where $\tau(z)$ ranges over all operator-valued Nevanlinna families of size $p\times p$. The matrix $W(z)$ satisfies a kernel $J_p$-positivity condition, with $$J_p = \begin{pmatrix} 0 & -i I_p \ i I_p & 0 \end{pmatrix}$$ and

$J_p - W(z) J_p W(w)^* \geq 0$

for $z, w \in \mathbb{C}_+$. This parametrization gives a bijection between generalized resolvents and boundary parameter families (Derkach, 4 Dec 2025, Derkach, 20 May 2025).

Concept	Operator/Linear Relation Formulation	Gauge Type
$\sL$-resolvent	$r(z) = L^* \widehat{R}_z L$	Proper/Improper
Parameter Set	$\tau(z)$ in Nevanlinna or generalized Nevanlinna class	Depends on negative index
$\sL$-resolvent matrix	$W(z)\in \mathcal{W}(J_p)$ class	Same structure

3. Boundary Triple Formalism and Weyl Function Machinery

A boundary triple for $A^*$ is a triple $(\mathbb{C}^p, \Gamma_0, \Gamma_1)$ with $\Gamma_0, \Gamma_1 : \dom(A^*) \to \mathbb{C}^p$ surjective and satisfying the abstract Green's identity:

$$(f', g) - (f, g') = (\Gamma_1 f,\, \Gamma_0 g) - (\Gamma_0 f,\, \Gamma_1 g)$$

This setup allows the definition of the Weyl function $M(z)$ and the $\gamma$-field by

$$M(z) \Gamma_0 f_z = \Gamma_1 f_z, \quad \gamma(z) = (\Gamma_0|_{N_z})^{-1},\quad N_z = \ker(A^* - zI)$$

These objects enable one to express all generalized resolvents via the Kreĭn formula:

$$R_z = R_z^0 - \gamma(z)[M(z) + \tau(z)]^{-1} \gamma(z)^*$$

This formula produces a bijection between generalized (including all $\sL$-) resolvents and Nevanlinna boundary families $\tau$, establishing the analytic functional model underlying the resolvent theory. In the case of linear relations and improper gauges, these constructions pass to the anti-dual or distributional boundary mappings (Derkach, 4 Dec 2025, Derkach, 20 May 2025, Mogilevskii, 2014).

4. Construction and Properties of the $\sL$-Resolvent Matrix

The preresolvent data are assembled in

$$\mathfrak{A}(z) = \begin{pmatrix} M(z) & \gamma(z)^* L \ L^* \gamma(z) & L^* R_z^0 L \end{pmatrix}$$

Under appropriate domain conditions, the $(1,2)$ and $(2,1)$ blocks remain invertible, allowing the explicit construction

$$W(z) = \begin{pmatrix} 0 & a_{12} \ -I_p & a_{22} \end{pmatrix}^{-1} \begin{pmatrix} I_p & a_{11} \ 0 & a_{21} \end{pmatrix}$$

Alternatively, $W(z)$ can be written in terms of boundary operators:

$W(z) = (\Gamma \cdot P(z)^*)^*, \qquad P(z) = L^{-1} \Pi_\sL^z$

The class of functions $W(z)$ thus obtained—$\mathcal{W}(J_p)$ in the Hilbert space case, or the corresponding Kreĭn–Saakyan class $\mathcal{W}_\kappa(J_p)$ in the Pontryagin space case (having $\kappa$ negative squares)—characterizes all possible $\sL$-resolvents. J-positivity/nonnegativity of the kernel associated to $W(z)$ characterizes positive definiteness properties of the corresponding resolvent kernels (Derkach, 4 Dec 2025, Derkach, 20 May 2025).

5. Spectral and Pseudo-Spectral Functions

With $\mathbb{R} \subset \rho(A, \sL)$, there exists a vector-valued directing mapping $f \mapsto (z)f = L^{-1} \Pi_\sL^z f$ into $\mathbb{C}^p$. Every nondecreasing left-continuous $p\times p$ matrix-function $\sigma(\lambda)$ on $\mathbb{R}$ is called an LT-spectral function if

$\int ((\lambda)f)^*\, d\sigma(\lambda)\, (\lambda)f \leq \langle f, f \rangle, \quad \text{with equality for } f\in\dom A$

Spectral functions $\sigma$ correspond precisely to those for which the induced generalized Fourier transform is an isometry. Every such $\sigma$ is linked to an $\sL$-resolvent via the Stieltjes-type integral

$$\int \left[(\lambda-z)^{-1} - \frac{\lambda}{1+\lambda^2}\right] d\sigma(\lambda) = r(z) + K$$

where $K$ is a boundary constant. Pseudo-spectral functions are characterized by partial isometries with kernel $\text{mul}\,A$; true spectral functions exist exactly when $A$ is single-valued (operator, no multi-valued part) (Derkach, 4 Dec 2025).

6. Specialization to Canonical Systems

For canonical first-order systems $f' + F(t) f = z H(t) f$ on an interval $(a, b)$ with $H\geq0$ and $\operatorname{tr} H = 1$, the $\sL$-resolvent theory provides a comprehensive parametrization of all self-adjoint extensions and generalized resolvents for the minimal relation $S_\text{min}$ in weighted $L^2_H$ spaces. A natural boundary triple is given by

$$\Gamma_0(f,g) = \frac{1}{\sqrt{2}}(f(a) + f(b)),\quad \Gamma_1(f,g) = -\frac{1}{\sqrt{2}}(f(a) - f(b))$$

The associated Weyl function and $\gamma$-field are computable via the system's monodromy $U(b, z)$:

$$M(z) = -[I - U(b, z)][I + U(b, z)]^{-1},\quad \gamma(z) = \sqrt{2}\, U(\cdot, z)[I + U(b, z)]^{-1}$$

For gauges $\sL$ in the dual space (typically supported at endpoints), explicit formulas for $P(z)$ and $Q(z)$ lead to closed-form rational expressions for $W(z)$ in terms of $U(b, z)$. The general linear-fractional formula recovers all possible boundary conditions and spectral measures (Derkach, 4 Dec 2025, Derkach, 20 May 2025, Mogilevskii, 2014).

7. Connections and Extensions

The Kreĭn–Saakyan theory extends classical self-adjoint extension analysis and Weyl–Titchmarsh theory by fully encoding the dependence of generalized resolvents and spectral measures on abstract boundary data, without restriction to differential or canonical systems. It subsumes both the operator and linear relation settings, accommodating multi-valued parts, improper gauges, and allows for indefinite signatures in Pontryagin and Kreĭn space frameworks. This provides a unified matrix-functional toolkit describing all admissible resolvents, associated pseudo-spectral and spectral families, and furnishes concrete computational machinery for applications in direct and inverse spectral theory (Derkach, 4 Dec 2025, Derkach, 20 May 2025, Mogilevskii, 2014).