Passive discrete-time systems with a Pontryagin state space

Passive discrete-time systems with Hilbert spaces as an incoming and outgoing space and a Pontryagin space as a state space are investigated. A geometric characterization when the index of the transfer function coincides with the negative index of the state space is given. In this case, an isometric (co-isometric) system has a product representation corresponding to the left (right) Krein-Langer factorization of the transfer function. A new criterion, based on the inclusion of reproducing kernel spaces, when a product of two isometric (co-isometric) systems preserves controllability (observability), is obtained. The concept of the defect function is expanded for generalized Schur functions, and realizations of generalized Schur functions with zero defect functions are studied.


Introduction
Let U and Y be separable Hilbert spaces.The generalized Schur class S κ (U, Y) consists of L(U, Y)-valued functions S(z) which are meromorphic in the unit disc D and holomorphic in a neighbourhood Ω of the origin such that the Schur kernel (1.1) K S (w, z) = 1 − S(z)S * (w) 1 − z w , w, z ∈ Ω, has κ negative squares (κ = 0, 1, 2, . ..).This means that for any finite set of points w 1 , . . ., w n in the domain of holomorphy ρ(S) ⊂ D of S and vectors f 1 , . . ., f n ⊂ Y, the Hermitian matrix has at most κ negative eigenvalues, and there exists at least one such matrix that has exactly κ negative eigenvalues.It is known from the reproducing kernel theory [1], [4], [23], [27], [30] that the kernel (1.1) generates the reproducing kernel Pontryagin space H(S) with negative index κ.The spaces H(S) are called generalized de Branges-Rovnyak spaces, and the elements in H(S) are functions defined on ρ(S) with values in Y.The notation S * (z) means (S(z)) * , a function S # (z) is defined to be S * (z) and S # ∈ S κ (Y, U) whenever S ∈ S κ (U, Y) [1, Theorem 2.5.2].
The class S 0 (U, Y) is written as S(U, Y) and it coincides with the Schur class, that is, functions holomorphic and bounded by one in D. The results first obtained by Kreȋn and Langer [26], see also [1, §4.2] and [21] are inverse Blaschke products, and they have unitary values everywhere on the unit circle T. It follows from these factorizations that many properties of the functions in the Schur class S(U, Y) hold also for the generalized Schur functions.
The properties of the generalized Schur functions can be studied by using operator colligations and transfer function realizations.An operator colligation Σ = (T Σ ; X , U, Y; κ) consists of a Pontryagin space X with the negative index κ (state space), Hilbert spaces U (incoming space), and Y (outgoing space) and a system operator T Σ ∈ L(X ⊕ U, X ⊕ Y).The operator T Σ can be written in the block form (1.3) where A ∈ L(X ) (main operator), B ∈ L(U, X ) (control operator), C ∈ L(X , Y) (observation operator), and D ∈ L(U, Y) (feedthrough operator).Sometimes the colligation is written as Σ = (A, B, C, D; X , U, Y; κ).It is possible to allow all spaces to be Pontryagin or even Kreȋn spaces, but colligations with only the state space X allowed to be a Pontryagin space will be considered in this paper.The colligation generated by (1.3) is also called a system since it can be seen as a linear discrete-time system of the form where {h k } ⊂ X , {ξ k } ⊂ U and {σ k } ⊂ Y.In what follows, "system" always refers to (1.3), since other kind of systems are not considered.
When the system operator T Σ in (1.3) is a contraction, the corresponding system is called passive.If T Σ is isometric (co-isometric, unitary), then the corresponding system is called isometric (co-isometric, conservative).The transfer function of the system (1.3) is defined by (1.4) θ Σ (z) := D + zC(I − zA) −1 B, whenever I − zA is invertible.Especially, θ is defined and holomorphic in a neighbourhood of the origin.The values θ Σ (z) are bounded operators from U to Y.The adjoint or dual system is Σ * = (T * Σ ; X , Y, U; κ) and one has θ Σ * (z) = θ Σ # (z).Since contractions between Pontryagin spaces with the same negative indices are bi-contractions, Σ * is passive whenever Σ is.If θ is an L(U, Y)-valued function and θ Σ (z) = θ(z) in a neighbourhood of the origin, then the system Σ is called a realization of θ.A realization problem for the function θ ∈ S κ (U, Y) is to find a system Σ with a certain minimality property (controllable, observable, simple, minimal); for details, see Theorem 2.4, such that Σ is a realization of θ.
If κ = 0, the system reduces to the standard Hilbert space setting of the passive systems studied, for instance, by de Branges and Rovnyak [18,19], Ando [2], Sz.-Nagy and Foias [32], Helton [24], Brodskiȋ [20], Arov [5,6] and Arov et al [7,8,9,10,13].The theory has been extended to Pontryagin state space case by Dijksma et al [21,22], Saprikin [28], Saprikin and Arov [12] and Saprikin et al [11].Especially, in [28], Arov's well-known results of minimal and optimal minimal systems are generalized to the Pontryagin state space settings.Part of those results are used in [11], where transfer functions, Kreȋn-Langer factorizations, and the corresponding product representation of system are studied and, moreover, the connection between bi-inner transfer functions and systems with bi-stable main operators are generalized to the Pontryagin state space settings.In this paper those results will be further expanded and improved.
The case when all the spaces are indefinite, the theory of isometric, co-isometric and conservative systems is considered, for instance, in [1], see also [23].The indefinite reproducing kernel spaces were first studied by Schwartz in [29] and Sorjonen in [30].
The paper is organized as follows.In Section 2, basic notations and definitions about the indefinite spaces and their operators are given.Also, the left and right Kreȋn-Langer factorizations are formulated, and the boundary value properties of generalized Schur functions are introduced.After that, basic properties of linear discrete time systems, or operator colligations, especially in Pontryagin state space, are recalled without proofs.However, the extension of Arov's result about the weak similarity between two minimal passive realizations of the same transfer function, is given with a proof.
Section 3 deals mainly with the dilations, embeddings and products of two systems.The transfer function θ Σ of the passive system Σ = (T Σ ; X , U, Y; κ) is a generalized Schur function with negative index no larger than the negative index of the state space X , but the theory of passive systems will often be meaningful only if the indices are equal.A simple geometric criterion for these indices to coincides is given in Lemma 3.2.Main results in this section contain criteria when the product of two co-isometric (isometric) systems preserves observability (controllability).These results are obtained in Theorems 3.6 and 3.7).The criteria involve the reproducing kernel spaces induced by the generalized Schur functions.Moreover, Theorem 3.9 expands the results of [11] about the realizations of generalized Schur functions and their product representations corresponding to the Kreȋn-Langer factorizations.In the end of Section 3, it is obtained that if A is the main operator of Σ = (T Σ ; X , U, Y; κ) such that θ Σ ∈ S κ (U, Y), then there exist unique fundamental decompositions Section 4 expands and generalizes the results of [6] and [11] about the realizations of bi-inner functions.It will be shown that the notions of stability and co-stability can be generalized to the Pontryagin state space settings in a similar manner as bi-stability is generalized in [11].Moreover, the results of [3] about the realizations of ordinary Schur functions with zero defect functions will be generalized.This yields a class of generalized Schur functions with boundary value properties very close to those of inner functions in a certain sense.

Pontryagin spaces, Kreȋn-Langer factorizations and linear systems
Let X be a complex vector space with a Hermitian indefinite inner product •, • X .The anti-space of X is the space −X that coinsides with X as a vector space but its inner product is − •, • X .Notions of orthogonality and orthogonal direct sum are defined as in the case of Hilbert spaces, and X ⊕ Y is often denoted by X Y ⊺ .Space X is said to be a Kreȋn space if it admits a decomposition X = X + ⊕ X − where (X ± , ± •, • X ) are Hilbert spaces.Such a decomposition is called a fundamental decomposition.In general, it is not unique.However, a fundamental decomposition determines the Hilbert space |X | = X + ⊕(−X − ) with the strong topology which does not depend on the choice of the fundamental decomposition.The dimensions of X + and X − , which are also independent of the choice of the fundamental decomposition, are called the positive and negative indices ind ± X = dim X ± of X .In what follows, all notions of continuity and convergence are understood to be with respect to the strong topology.All spaces are assumed to be separable.A linear manifold N ⊂ X is a regular subspace, if it is itself a Kreȋn space with the inherited inner product of •, • X .A Hilbert subspace is a regular subspace such that its negative index is zero, and a uniformly negative subspace is a regular subspace with positive index zero, i.e., an anti-Hilbert space.
If N ⊂ X is a regular subspace, then X = N ⊕ N ⊥ , where ⊥ refers to orthogonality w.r.t.indefinite inner product •, • X .Observe that N is regular precisely when N ⊥ is regular.Denote by L(X , Y) the space of all continuous linear operators from the Kreȋn space X to the Kreȋn space Y.Moreover, L(X ) stands for L(X , X ).Domain of a linear operator T is denoted by D(T ), kernel by ker T and T ↾ N is a restriction of T to the linear manifold N .The adjoint of A ∈ L(X , Y) is an operator A * ∈ L(Y, X ) such that Ax, y Y = x, A * y X for all x ∈ X and y ∈ Y. Classes of invertible, self-adjoint, isometric, co-isometric and unitary operators are defined as for Hilbert spaces, but with respect to the indefinite inner product.For self-adjoint operators A, B ∈ L(X , Y), the inequality A ≤ B means that Ax, x ≤ Bx, x for all x ∈ X .A self-adjoint operator P ∈ L(X ) is an •, • -orthogonal projection if P 2 = P .The unique orthogonal projection onto a regular subspace N of X exists and is denoted by P N .A Pontryagin space is a Kreȋn space X such that ind − X < ∞.A linear operator A ∈ L(X , Y) is a contraction if Ax, Ax ≤ x, x for all x ∈ X .If X and Y are Pontryagin spaces with the same negative index, then the adjoint of a contraction A ∈ L(X , Y) is still a contraction, i.e., A is a bi-contraction.The identity operator of the space X is denoted by I X or just by I when the corresponding space is clear from the context.For further information about the indefinite spaces and their operators, we refer to [14], [17] and [23].
For ordinary Schur class S(U, Y), it is well known [32] that S ∈ S(U, Y) has non-tangential strong limit values almost everywhere (a.e.) on the unit circle T. It follows that S ∈ S(U, Y) can be extended to L ∞ (U, Y) function, that is, the class of weakly measurable a.e.defined and essentially bounded L(U, Y)-valued functions on T.Moreover, S(ζ) is contractive a.e. on T. If S ∈ S(U, Y) has isometric (co-isometric, unitary) boundary values a.e. on T, then S is said to be inner (co-inner, bi-inner).
If U = Y, then the notations S(U) and S κ (U) are often used instead of S(U, U) and S κ (U, U).Suppose that P ∈ L(U) is an orthogonal projection from the Hilbert space U to an arbitrary one dimensional subspace.Then a function defined by of simple Blaschke-Potapov factors is called Blaschke product of degree n, and it is also bi-inner and invertible on D \ {α 1 , . . ., α n }.The following factorization theorem was first obtained by Kreȋn and Langer [26], see also [1, §4.2] and [21].The following result [32, Theorem V.4.2], which involves the notion of an outer function (for the definition, see [32]), will be utilized.

on T, then there exist a Hilbert space K and an outer function
a.e. on T.Moreover, ϕ is unique up to a left constant unitary factor.
Since Blaschke products are unitary on T, it follows that Theorem 2.3 guarantees that there exists an outer function ϕ S with properties introduced in Theorem 2.3 for N S .An easy modification of Theorem 2.3 shows that there exists a Schur function ψ S such that ψ # S is an outer function, . Moreover,it follows from the identies (2.5) and (2.6) that ϕ S = ϕ S l and ψ S = ψ Sr . (2.7) The function ϕ S is called the right defect function and ψ S is the left defect function.Let Σ = (A, B, C, D; X , U, Y; κ) be a passive system.The following subspaces are called respectively controllable, observable and simple subspaces.The system Σ is said to be controllable (observable, simple) if X c = X (X o = X , X s = X ) and minimal if it is both controllable and observable.When Ω ∋ 0 is some symmetric neighbourhood of the origin, that is, z ∈ Ω whenever z ∈ Ω, then also are also bi-contractions.Moreover, the operators B and C * are contractions but not bicontractions unless κ = 0.
The following realization theorem is known, and the parts (i)-(iii) can be found e.g. in [1, Chapter 2] and the part (iv) in [ Conversely, if the system Σ has some of the properties (i)-(iv), then θ Σ ∈ S κ (U, Y), where κ is the negative index of the state space of Σ.
It is also true that the transfer function of passive system is a generalized Schur function, but its index may be smaller than the negative index of the state space [28, Theorem 2.2].For a conservative system Σ it is known from [1, Theorem 2.1.2(3)] that the index of the transfer function θ Σ of Σ co-insides with the negative index of the state space X of Σ if and only if the space (X s ) ⊥ is a Hilbert subspace.This result holds also in more general settings when Σ is passive, as it will be proved in Lemma 3.2, after introducing some machinery.
and there exists a unitary operator U : X 1 → X 2 such that Moreover, the realizations Σ 1 and Σ 2 are said to be weakly similar if D 1 = D 2 and there exists an injective closed densely defined possibly unbounded linear operator Z : X 1 → X 2 with the dense range such that Unitary similarity preserves dynamical properties of the system and also the spectral properties of the main operator.If two realizations of θ ∈ S κ (U, Y) both have the same property (i), (ii) or (iii) of Theorem 2.4, then they are unitarily similar [1, Theorem 2.1.3].In Hilbert state space case, results of Helton [24] and Arov [5] state that two minimal passive realizations of θ ∈ S(U, Y) are weakly similar.However, weak similarity preserves neither the dynamical properties of the system nor the spectral properties of its main operator.The following theorem shows that Helton's and Arov's statement holds also in Pontryagin state space settings.Proof is similar to the one given in the Hilbert space settings in [15, Theorem 3.2] and [16,Theorem 7.13].
Then they are weakly similar.
Proof.Decompose the system operators as in (1.3).In a sufficiently small neighbourhood of the origin, the functions θ Σ 1 and θ Σ 2 have the Neumann series which coincide.Hence and let Z be the closure of the graph of R.
Rx n , and the continuity implies Since Σ 2 is observable, it follows from (2.9) that (2.14) and therefore y = 0.This implies that Z is a closed densely defined linear operator.Since Σ 2 is controllable, the range of Z is dense.
To prove the injectivity, let x ∈ D(Z) such that Zx = 0. Then there exists {x n } n∈N ⊂ D(R) such that x n → x and Rx n → 0. By the continuity, Since Z is closed, equations (2.15) and (2.16) show that A 1 x ∈ D(Z) and ZA 1 x = A 2 Zx.Since (2.17) and (2.18) hold also, it has been shown that Z is a weak similarity.
Remark 2.6.It should be noted that Theorem 2.5 holds also when all the spaces are Pontryagin, Kreȋn or, if one defines the observability criterion as ∩ n∈N 0 ker CA n = {0}, even Banach spaces.This result can also be derived from [31, p. 704].

Julia operators, dilations, embeddings and products of systems
The system (1.3) can be expanded to a larger system either without changing the transfer function or without changing the main operator.Both of these can be done by using the Julia operator, see (3.1) below.For a proof of the next theorem and some further details about Julia operators, see [ A dilation of a system (1.3) is any system of the form Σ = ( A, B, C, D; X , U, Y; κ ′ ), where That is, the system operator T Σ of Σ is of the form (3.3) Then the system Σ is called a restriction of Σ, and it has an expression (3.4) Σ = (P X A↾ X , P X B, C↾ X , D; P X X , U, Y; κ).
Dilations and restrictions are denoted by mostly without subscripts when the corresponding state spaces are clear.A calculation show that the transfer functions of the original system and its dilation coincide.
The second way to expand the system (1.3) is called an embedding, which is any system determined by the system operator (3.6) where U ′ and Y ′ are Hilbert spaces.The transfer function of the embedded system is where θ Σ is the transfer function of the original system.For a passive system there always exist a conservative dilation [28, Theorem 2.1] and a conservative embedding [11, p. 7].Both of these can be constructed such that the system operator of the expanded system is the Julia operator of T Σ .Such expanded systems are called Julia dilation and Julia embedding, respectively.If the passive system (1.3) is simple (controllable, observable, minimal), then so is any conservative embedding (3.6) of it.This follows from the fact that BU ⊂ B U and C * Y ⊂ C * Y.A detailed proof of simplicity can be found in [11,Theorem 4.3].The same argument works also in the rest of the cases.However, it can happen that a simple passive system has no simple conservative dilation, even in the case when the original system is minimal, see the example on page 15 in [11].Lemma 3.2.Let θ Σ be the transfer function of a passive system Σ = (T Σ ; X , U, Y; κ).If θ Σ ∈ S κ (U, Y), then the spaces (X c ) ⊥ , (X o ) ⊥ and (X s ) ⊥ are Hilbert subspaces of X .Moreover, if one of the spaces (X c ) ⊥ , (X s ) ⊥ and (X s ) ⊥ is a Hilbert subspace, then so are the others and Lemma 2.5] that (X c ) ⊥ and (X o ) ⊥ are Hilbert spaces.It easily follows from (2.8) and (2.9) that ⊥ is also a Hilbert space, and the first claim is proved.Suppose next that (X s ) ⊥ is a Hilbert space.Consider a conservative embedding Σ of Σ, and represent the system operator T Σ as in (3.6).The first identity in (3.7) shows that the transfer function of any embedding of Σ has the same number of poles (counting multiplicities) as θ Σ , and therefore it follows from Lemma 2.2 that the indices of θ Σ and θ Σ coincides.Denote the simple subspace of the embedded system as X s .Since X s ⊂ X s , it holds ( X s ) ⊥ ⊂ (X s ) ⊥ , and therefore ( X s ) ⊥ is also a Hilbert space.It follows from [1, Theorem 2.1.2(3)] that the transfer function θ Σ of Σ belongs to S κ ( U, Y), which implies now θ Σ ∈ S κ (U, Y).Then the first claim proved above implies that (X c ) ⊥ and (X o ) ⊥ are Hilbert subspaces.
If one assumes that (X c ) ⊥ or (X o ) ⊥ is a Hilbert space, the identity (3.8) shows that (X s ) ⊥ is a Hilbert space as well.Then the argument above can be applied, and the second claim is proved.
The product or cascade connection of two systems Σ Written in the form (1.3), one has X = X 1 X 2 and (3.10) Note that A 2 = A↾ X 2 and (3.11) The product Σ 2 • Σ 1 is defined when the incoming space of Σ 2 is the outgoing space of Σ 1 .
Again, direct computations show that θ Σ 2 •Σ 1 = θ Σ 2 θ Σ 1 whenever both functions are defined.For the dual system one has ( It follows from the identity (3.11) that the product Σ 2 • Σ 1 is conservative (isometric, co-isometric, passive) whenever Σ 1 and Σ 2 are.Also, if the product is isometric (co-isometric, conservative) and one factor of the product is conservative, then the other factor must be isometric (co-isometric, conservative).
It is known (c.f.e.g.[1, Theorem 1.2.1]) that if Σ 2 • Σ 1 is controllable (observable, simple, minimal), then so are Σ 1 and Σ 2 .The converse statement is not true.The following lemma gives necessary and sufficient conditions when the product is observable, controllable or simple.The simple case is handled in [11,Lemma 7.4].
Let Ω = Ω be a symmetric neighbourhood of the origin such that the transfer function where x 1 ∈ X 1 and x 2 ∈ X 2 .Then Σ is observable if and only if (3.12) has only the trivial solution, and Σ is controllable if and only if (3.13) has only the trivial solution.Moreover, Σ is simple if and only if the pair of equations consisting of (3.12) and (3.13) has only the trivial solution.
Proof.Write the system operator T Σ 2 •Σ 1 in (3.9) in the form (1.3).It follows from (2.11)-(2.13) that x ∈ (X o ) ⊥ ⇐⇒ C(I − zA) −1 x = 0 for all z ∈ Ω; (3.14) x ∈ (X c ) ⊥ ⇐⇒ B * (I − zA * ) −1 x = 0 for all z ∈ Ω; (3.15) x ∈ (X s ) ⊥ ⇐⇒ B * (I − zA * ) −1 x = 0 and C(I − A) −1 x = 0 for all z ∈ Ω. (3.16) Decompose x = x 1 ⊕x 2 , where x 1 ∈ X 1 and x 2 ∈ X 2 .With respect to the this decomposition, the definition of the main operator A from (3.10) yields From this relation and (3.10), it follows that the right hand side of (3.14) is equivalent to = 0 for all z ∈ Ω. (3.17) Similar calculations show that the right hand side of (3.15) is equivalent to Expanding the identity (3.17) and using the definition of the transfer function one gets that (3.17) is equivalent to That is, the identity (3.17) is equivalent to (3.12).Similar calculations and the identity shows that the identity (3.18) is equivalent to (3.13).The results follow now by observing that if the system Σ is observable, controllable or simple, then, respectively, ( Part (iii) of the theorem below with an additional condition that all the realizations are conservative, is proved in [11,Theorem 7.3,7.6].Similar techniques will be used to expand this result as follows.
Theorem 3.4.Let θ ∈ S κ (U, Y) and let θ = θ r B −1 r = B −1 l θ l be its Kreȋn-Langer factorizations.Suppose that Proof.Suppose first that Σ B −1 r is a simple passive system and Σ θr is a passive system.The results from [11,Theorems 9.4 and 10.2] show that all the simple passive realizations of B −1 r are conservative and minimal.Thus, the assumptions guarantees that Σ B −1 r is conservative and minimal.Represent the system operators T Σ B −1 r and T Σ θr as Let Ω = Ω be a symmetric neighbourhood of the origin such that B −1 The space X − r is κ-dimensional anti-Hilbert space, and all the poles of B −1 r are also poles of r is a Hilbert space, the operator A 2 is a Hilbert space contraction, and (I − zA 2 ) −1 exists for all z ∈ D. That is, the right hand side of (3.20) is holomorphic in D, and then so is the left hand side also.Since θ r and B r have no common zeros in the sense of Theorem 2.1 and the zeros of B r are the poles of B −1 r , the factor θ r (z) cannot cancel out the poles of C 1 (I − zA 1 ) −1 x 1 (For a more detailed argument, see the proof of [11,Theorem 7.3]).That is, θ r (z)C 1 (I − zA 1 ) −1 x 1 is holomorphic in D only if C 1 (I − zA 1 ) −1 x 1 ≡ 0. Then also C 2 (I − zA 2 ) −1 x 2 ≡ 0, and it follows from (2.12) that x 1 ∈ (X − r o ) ⊥ and x 2 ∈ (X + r o ) ⊥ .Since the system Σ B −1 r is minimal, x 1 = 0.If the system Σ θr is observable, then x 2 = 0, and it follows from Lemma 3.3 that Σ θr • Σ B −1 r is observable and passive, and part (i) is proven.Next suppose that x 1 and x 2 satisfy (3.20) and ) ⊥ , and therefore If the system Σ θr is simple, then x 2 = 0, and it follows from Lemma 3.3 that Σ θr • Σ B −1 r is simple and passive, and the first claim of the part (iii) is proven.The other claim in part (iii) and also part (ii) follow now by considering the dual systems.
The product of the form Σ B −1 l • Σ θ l does not necessarily preserve observability as is shown in Example 3.8 below.A counter-example is constructed with the help of the following realization result.For the proof and more details, see [1, Theorem 2.2.1].Lemma 3.5.Let S ∈ S κ (U, Y) and let H(S) be the Pontryagin space induced by the reproducing kernel (1.1).Then the system Σ = (A, B, C, D, H(S), U, Y; κ) where is co-isometric and observable realization of S.Moreover, C(I −zA) −1 h = h(z) for h ∈ H(S).
The system Σ in Lemma 3.5 is called a canonical co-isometric realization of S.
If the systems Σ 1 and Σ 2 in Lemma 3.3 have additional properties, a criterion for observability that does not explicitly depend on a system operator can be obtained.

and only if the following two conditions hold:
(i) Proof.Since all co-isometric observable realizations of S 1 and S 2 are unitarily similar, it can be assumed that Σ 1 and Σ 2 are realized as in Lemma 3.5.Let Ω be a neighbourhood of the origin such that S 1 and S 2 are analytic in Ω.By combining Lemma 3.5 and the condition (3.12) in Lemma 3.3, it follows that Σ is observable if and only if holds for every z ∈ Ω only when h 1 ≡ 0 and h 2 ≡ 0. Assume the conditions (i) and (ii).Then S 2 (z)h 1 (z) = −h 2 (z) can hold only if h 2 ≡ 0. Since the mapping h 1 → S 2 h 1 is an isometry, it has only the trivial kernel.Therefore h 1 ≡ 0, and sufficiency is proven.
Conversely, assume that Σ is co-isometric and observable.The condition (3.23) shows that the mapping h 1 → S 2 h 1 has only the trivial kernel, and It now follows from [1, Theorem 4.1.1]that H(S 1 ) and S 2 H(S 1 ) are contained contractively in H(S), and h 1 → S 2 h 1 is a partial isometry.Since it has only the trivial kernel, it is an isometry, and (ii) holds.Since (3.24) holds and H(S 1 ) and S 2 H(S 1 ) are contained contractively in H(S), a result from [1,Theorem 1.5.3]shows that H(S 1 ) and S 2 H(S 1 ) are actually contained isometrically in H(S).Therefore H(S 1 ) ⊥ = S 2 H(S 1 ) so the condition (i) holds and the necessity is proven.
The dual version can be obtained by using the canonical isometric realizations from [1, Theorem 2.2.2] or taking adjoint systems in Theorem 3.6.
be isometric and controllable realizations of the functions S 1 ∈ S(U, Y 1 ) and S 2 ∈ S(Y 1 , Y), respectively.Then Σ = Σ 2 • Σ 1 is isometric and controllable realization of S = S 2 S 1 if and only if the following two conditions hold: Proof.The theorem will be proved in two steps.In the first step, it is assumed that Σ 1 is simple, Σ 2 is observable and Σ 3 is controllable.In the second step, the general case will be proved by using the results from the first step.
Step 1. (i) This is stated essentially in [11,Theorem 7.2] but without proof.According to [21,Theorem 4.4], Σ 1 = (T Σ ; X , U, Y; κ) can be represented as the products of the form where X − r and X − l are κ-dimensional anti-Hilbert spaces.Subscripts refer "right" and "left", because it will be proved that the factorizations of the transfer function θ of Σ 1 corresponding to the product representations above are actually Kreȋn-Langer factorizations.Since all the realizations in (3.27) are simple and conservative, it follows from Lemma 3.2 that θ Σ r2 , θ Σ l1 ∈ S(U, Y), θ Σ r1 S κ (U), θ Σ l2 ∈ S κ (Y), and the spaces are Hilbert spaces.But since the state spaces X − r and X − l are anti-Hilbert spaces, all the spaces in (3.28) must be the zero spaces.Thus Σ r1 and Σ l2 are minimal.By using the unitary similarity introduced on page 7 it can be deduced now that all co-isometric observable realizations of θ r2 and θ l1 are conservative and minimal, and then it follows from [1, Theorem A3] that θ r2 and θ l1 are inverse Blaschke products, which gives the result.
(ii) It is known (cf.e.g.[1, Theorem 2.4.1]) that the co-isometric and observable realization Σ 2 = (A, B, C, D; X , U, Y; κ) of the function θ has a simple and conservative dilation Σ 2 = ( A, B, C, D; X , U, Y; κ) such that (3.29) where X 0 is a Hilbert space.By [11,Theorem 7.7], there exist unique fundamental decompositions X = X + ⊕ X − and X = X + ⊕ X − such that AX + ⊂ X + and A X + ⊂ X + .Then (X 0 ⊕ X + ) ⊕ X − is a fundamental decomposition of X , and for x 0 ∈ X 0 and x + ∈ X + (3.30) This yields X + = X 0 ⊕ X + and X − = X − 2 .Part (i) shows that Σ 2 can be represented as r .The transfer functions of the components are θ r and B −1 r , respectively, and Σ θr is simple and conservative and Σ B −1 r is conservative and minimal.It follows from [11,Theorem 7.7] that the state spaces of Σ θr and Σ B −1 r are X + and X − , respectively.Thus By using the representation above and (3.2)-(3.5), it follows that and res X + →X + Σ θr := Σ θr .Since Σ 2 is co-isometric and observable and Σ B −1 r is minimal and conservative, Σ θr must be co-isometric and observable.That is, Σ is the desired representation.(iii) This can be done by using [1, Theorem 2.4.3] and then proceeding along the lines of the proof of (ii) .
Step 2. (i) Denote Σ 1 = (A, B, C, D; X ; U, Y; κ).Since the index of the transfer function θ coincides with the negative index of X , Lemma 3.2 shows that (X s ) ⊥ is a Hilbert space.It easily follows from (2.10) that C(X s ) ⊥ = {0}, B * (X s ) ⊥ = {0}, AX s ⊂ X s and A(X s ) ⊥ ⊂ (X s ) ⊥ .This implies that the system operator has the representation (3.31) Easy calculations show that the restriction  (ii) Denote Σ 2 = (A, B, C, D; X , U, Y; κ).Lemma 3.2 show that (X o ) ⊥ is a Hilbert space.From the identity (2.9) it follows easily that A(X o ) ⊥ ⊂ (X o ) ⊥ and C(X o ) ⊥ = {0}.This implies that the system operator can be represented as Moreover, the restriction res have the properties introduced in Part 1 (ii).The final statement is obtained by proceeding as in the proof of (i).
(iii) The proof is similar to the proofs of (i) and (ii) and hence the details are omitted.
Proposition 3.10.Suppose that A ∈ L(U, Y) is the main operator of a passive system Σ = (T Σ ; X , U, Y; κ) such that the index of the transfer function of Σ is κ.Then there exist unique fundamental decompositions Proof.Embed the system Σ in a conservative system Σ = (T Σ , X , U, Y, κ) without changing the main operator and the state space.Now the first identity in (3.7) shows that the transfer function θ Σ of Σ has the same amount of poles (counting multiplicities) as the transfer function of the original system.Hence it follows from Lemma 2.2 that the index of θ Σ is κ.The representations in Theorem 3.9 (i) combined with the decomposition of the main operator A in (3.7) give the claimed fundamental decompositions.The decomposition X + 1 ⊕ X − 1 corresponds to the one induced by the product representation Σ = Σ θr • Σ B −1 r , where θ = θ r B −1 r is the right Kreȋn-Langer factorization of θ.Similarly, the decomposition X + 2 ⊕ X − 2 corresponds to the one induced by the product representation Σ = Σ B −1 where θ = B −1 l θ l is the left Kreȋn-Langer factorization of θ.To prove the uniqueness, the fact that A has no negative eigenvector with corresponding eigenvalue modulus one is needed.To this end, assume that Ax = λx for some x ∈ X and λ ∈ T. Consider again a conservative embedding Σ of Σ, and represent Σ as in (3.6) .Then, Since Σ is conservative, the system operator T Σ of Σ is unitary.Therefore x, x X = λx, λx X + Cx, Cx Y , and since Y is a Hilbert space and |λ| = 1, it must be Cx = 0.Then, CA n x = λ n Cx = 0 for any n ∈ N 0 .That is, x ∈ ( X o ) ⊥ , where X o is the observable subspace of the system Σ.Since the index of θ is κ, the subspace ( X o ) ⊥ is a Hilbert space by Lemma 3.2, and x must be non-negative.
The following theorem in the Hilbert state space case was obtained in [3, Theorem 1.1].The proof therein was based on the block parametrization of the system operator.The proof given here for the general case is based on the existence of minimal passive realizations.It also uses some techniques appearing in the proof of [6, Theorem 1] and, in addition, implements the product representations provided in Theorem 3.9.T ,1 (I − zA) −1 B = 0 for every z in some neighbourhood of the origin.Since Σ is controllable, it follows from (2.11) that D * T ,1 = 0 and then D T = 0, which means that Σ is isometric.
If Σ is chosen to be minimal passive, the previous argument shows that Σ is an isometric and minimal realization of θ.Since the controllable isometric realizations of θ are unitarily similar, they all are now also minimal.This proves the first statement in (i).
If θ has isometric boundary values a.e., then θ l in the left Kreȋn-Langer factorization of θ is inner.Consider the product Σ = Σ B −1 • Σ θ l as in the Theorem 3.9.Let X + 1 ⊕ X − 1 and X + 2 ⊕ X − 2 be the unique fundamental decompositions of X of, given by Proposition 3.10, such that AX + 1 ⊂ X + 1 and A * X + 2 ⊂ X + 2 .The case κ = 0 from [3,Theorem 1.1] shows that the main operator of Σ θ l belongs to C 0 • , and then the main operator of Σ * θ l , which is A * ↾ X + 2 ,

1 rand B − 1 l
, show that S ∈ S κ (U, Y) has Kreȋn-Langer factorizations of the form S = S r B −1 r = B −1 l S l , where S r , S l ∈ S 0 (U, Y).The functions B − is a simple Blaschke-Potapov factor.Easy calculations show that b is holomorphic in the closed unit disc D, it has unitary values everywhere on T and b(z) is invertible whenever z ∈ D \ {α}.In particular, b ∈ S 0 (U) is bi-inner.

Theorem 2 . 1 .
Suppose S ∈ S κ (U, Y).Then(2.3)S(z) = S r (z)B −1 r (z) where S r ∈ S(U, Y) and B r is a Blaschke product of degree κ with values in L(U) such that B r (w)f = 0 and S r (w)f = 0 for some w ∈ D only if f = 0. Moreover,(2.4)S(z) = B −1 l (z)S l (z) where S l ∈ S(U, Y) and B l is a Blaschke product of degree κ with values in L(Y) such that B * l (w)g = 0 and S * l (w)g = 0 for some w ∈ D only if g = 0. Conversely, any function of the form (2.3) or (2.4) belongs to S κ ′ for some κ ′ ≤ κ, and κ ′ = κ exactly when the functions have no common zeros in sense as described above.Both factorizations are unique up to unitary constant factors.The factorization (2.3) is called the right Kreȋn-Langer factorization and (2.4) is the left Kreȋn-Langer factorization.It follows that S ∈ S κ (U, Y) has κ poles (counting multiplicities) in D, contractive strong limit values exist a.e. on T and S can also be extended to L ∞ (U, Y)-function.Actually, these properties also characterize the generalized Schur functions.This result will be stated for reference purposes.For the proof of the sufficiency, see[21, Proposition 7.11].Lemma 2.2.Let S be an L(U, Y)-valued function holomorphic at the origin.Then S ∈ S κ (U, Y) if and only if S is meromorphic on D with finite pole multiplicity κ and lim r→1 sup |z|=r S(z) ≤ 1 holds.A function S ∈ S κ (U, Y) and the factors S r and S l in (2.3) and (2.4) have simultaneously isometric (co-isometric, unitary) boundary values since the factors B −1 l and B −1 r have unitary values everywhere on T.

23 ,Theorem 3 . 1 .
Lecture 2].Suppose that X 1 and X 2 are Pontryagin spaces with the same negative index, and A : X 1 → X 2 is a contraction.Then there exist Hilbert spaces D A and D A * , linear operators D A : D A → X 1 , D A * : D A * → X 2 with zero kernels and a linear operator L : D A → D A * such that(3.1)

are the realizations of θ r , θ l , B − 1 r
and B −1 l , respectively.Then: (i) If Σ θr and Σ B −1 r are observable and passive, then so is Σ θr • Σ B −1 r ; (ii) If Σ θ l and Σ B −1 l are controllable and passive, then so is Σ B −1 l • Σ θ l ; (iii) If all the realizations described above are simple passive, then so are Σ θr • Σ B −1 r and Σ B −1 l • Σ θ l .

1 rand Σ B − 1 l
dimensional anti-Hilbert spaces, Σ θr and Σ θ l are conservative and simple and Σ B −are conservative and minimal.It can be now deduced that X has the fundamental decompositions (