betaFeedback

WorksMathematics

Norm-optimal model order reduction in reproducing kernel Hilbert spaces

Abstract

We consider the problem of model order reduction (MOR) in reproducing kernel Hilbert spaces (RKHSs). We prove generalizations of well-known necessary interpolatory conditions for norm-optimal reduced-order models, originally stated for rational functions in H2H_2 Hardy-Hilbert spaces. We construct an alternative to the iterative rational Krylov (IRKA) algorithm for RKHSs with differentiable kernels to obtain norm-optimal reduced-order models. Then, we introduce generalized Malmquist-Takenaka (GMT) functions that can be used to define orthonormal bases in an arbitrary RKHS. Using GMT functions, we propose unitary transformations between separable RKHSs. This allows us to extend our optimality conditions and subsequent algorithms to MOR problems posed in RKHSs with non-differentiable kernels. The potential and versatility of the proposed MOR framework is demonstrated in several numerical experiments. We perform norm-optimal MOR on various machine learning models such as support vector machines and deep ReLU networks. In addition, we show how the proposed MOR framework can be used to solve seemingly unrelated problems. In particular, we provide an example for fitting high degree polynomials to data and also obtain a FEM-like method for solving partial differential equations (PDEs) with an optimal kernel basis.

Mathematics Subject Classification 2020: 41A65 (primary); 41A05, 42B35, 46E22, 65D15 (secondary).

Keywords: reproducing kernel Hilbert spaces, reproducing kernels, Malmquist–Takenaka systems, model order reduction, support vector machines, neural networks.

Introduction

The study of model order reduction (MOR) deals with the simplification of mathematical models referred to as full–order models (FOMs). In MOR, FOMs and their simplified approximants usually have prescribed structure. The obtained approximant is called the reduced–order model (ROM). Traditionally, MOR techniques have been developed for reducing the complexity of dynamical systems [5, 11, 30, 31, 42]. Since many control engineering applications must operate in real time under limited hardware capacity, control schemes based on ROMs have proven particularly useful [49]. Well–known techniques exist for the reduction of models that describe linear time–invariant (LTI) systems [30, 31, 42, 58, 60].

In order to address the MOR of LTI systems, several well–known methods have been developed that preserve the structure of the full order model, but provide theoretical guarantees on the performance of the ROM. One such well–known method is balanced truncation (BT) [30], which operates on the state–space representation of LTI systems. It relies on a balanced realization of the system, which can be obtained by solving Lyapunov equations. Then, the least controllable and observable states in the balanced system are disregarded to obtain the ROM. One important benefit of the BT method is the theoretical guarantee it provides on the H∞H_{\infty} error of the ROM.

Another popular approach to the reduction of LTI models is the iterative rational Krylov algorithm (iterative rational Krylov algorithm (IRKA)) introduced in [31]. IRKA is an iterative method that computes the ROM and guarantees that the result satisfies necessary interpolatory (or, equivalently, orthogonality–based) conditions for optimal approximation in the H2H_2 norm. In recent works [11, 12], the orthogonality–based optimality conditions proposed in [31] for H2(C+)H_2(\mathbb{C}^{+}) have been adapted to H2H_2 spaces corresponding to more general domains. For example, in [11, 12], interpolatory necessary conditions and an IRKA analog have been developed for H2H_2 spaces defined over simply connected sets. We also mention [8], where H2H_2–norm–optimal ROMs are obtained for bilinear systems. Again, the construction uses a variation of IRKA and satisfies an appropriate analog of the interpolatory conditions necessary for optimal approximation in the H2H_2 norm.

In this study, our objective is to further develop MOR methods along several directions of generalization. First, we exploit the fact that the H2(C+)H_2(\mathbb{C}^{+}) Hardy–Hilbert space is an reproducing kernel Hilbert space (RKHS) containing functions C+→C\mathbb{C}^{+} \to\mathbb{C}, where C+\mathbb{C}^{+} denotes the open right complex half–plane. We show that the necessary interpolatory optimality conditions proposed in [31, 42] for H2H_2–norm–optimal MOR can be generalized to an RKHS setting. The proposed optimality conditions are then used to construct iterative schemes to obtain norm–optimal ROMs. We note that our optimality conditions and subsequent algorithms can only be applied directly in RKHSs whose kernel is Wirtinger differentiable with respect to both variables. We refer to RKHSs adhering to this condition as admissible. Several relevant large models, such as deep ReLU neural networks [14, 54, 59], cannot be represented in admissible RKHSs.

To overcome this issue, we introduce the RKHS variants of Malmquist–Takenaka (MT) functions [44, 60] and refer to them as generalized Malmquist–Takenaka (GMT) functions. The classical MT functions form an orthonormal system in the Hardy–Hilbert space H2(D)H_2(\mathbb{D}), where D:={z∈C:∣z∣<1}\mathbb{D} := \{z \in\mathbb{C} : |z| < 1\} denotes the interior of the complex unit disk. We prove that the proposed RKHS formulation of the MT system retains some important properties of its rational counterpart. We mention that drawing an analogy between special complex analytic functions and general RKHSs is not without precedent. In [26], for example, an RKHS formulation of finite Blaschke products (special inner functions of H2(D)H_2(\mathbb{D})) is proposed. Recently, in [55], the rational MT system has been defined for a family of RKHSs known as boundary weighted Hardy spaces. We note that our definition of GMT functions contains these and other previous generalizations such as [47] as special cases. Using the introduced GMT bases and the Hilbert space isomorphism theorem, we define unitary transformations between any two separable RKHSs. These bijections allow us to apply the proposed optimality conditions and MOR algorithms by transforming the problem to an admissible RKHS such as H2(D)H_2(\mathbb{D}).

Finally, we present a collection of novel applications of norm–optimal MOR. The purpose of these examples is to demonstrate the versatility of the proposed RKHS–MOR framework. To this end, we include examples that perform RKHS–norm–optimal MOR on modern machine learning models. In particular, we consider the MOR of support vector machines and deep ReLU neural networks. In addition, we consider two somewhat surprising applications. These illustrate that, under some circumstances, our framework can be used to obtain norm–optimal approximations of FOMs that are not fully known. In particular, we first provide an academic example of fitting high–degree polynomials to data in an RKHS–norm–optimal manner. Then, we demonstrate how our RKHS approximation framework can be used to obtain kernel–basis weak solutions to certain partial differential equations (PDEs). These four applications range from an illustrative academic example to fully developed case studies, with the PDE application in particular presented as a proof of concept for a genuinely new use of the framework, rather than a completed method.

Below we list the key novelties of our study.

  1. Optimality conditions in general RKHSs. We prove necessary conditions for RKHS norm optimality of the reduced–order model in admissible RKHSs. These results, see § 4, unify and generalize conditions known for H2(C+)H_2(\mathbb{C}^+) approximation [31, 42].

  1. Iterative RKHS MOR algorithms. We develop iterative schemes that, in admissible RKHSs, produce ROMs satisfying the introduced optimality conditions.

  1. Generalized Malmquist–Takenaka (GMT) functions. We introduce GMT functions for RKHSs, showing that existing variants, including the classical H2(D)H_2(\mathbb{D}) function system and previous generalizations [47, 55], arise as special cases. We show interpolation properties, zero locations, factorizations, and several other important properties of GMT functions.

(4) The GMT–transformation. We introduce the GMT transformation: a unitary map between two arbitrary separable RKHSs. We use the GMT transformation to solve MOR problems posed in non–admissible RKHSs.

(5) Diverse applications. We provide a diverse collection of example applications. We perform RKHS–norm–optimal MOR on support vector machines (SVMs) and neural networks. We develop an adaptive PDE discretization scheme based on the proposed approximation framework.

The paper is organized as follows. In § 2, we revisit the definition of RKHSs and provide a formal problem statement. In § 3, we review related mathematical concepts and prior results, focusing on properties of RKHSs and H2H_2–optimal MOR. Our novel optimality conditions are introduced in § 4. GMT functions, their properties and the GMT transformation are introduced in § 6. § 7 contains our proposed MOR algorithms, while in § 8, we demonstrate their effectiveness through four numerical examples. Finally, in § 10, we summarize our results and discuss future research directions.

Problem statement

In this work, we are interested in the MOR of FOMs belonging to RKHSs. In order to provide a formal problem statement, we recall the definition of RKHSs and reproducing kernels.

Definition 2.1 (Reproducing kernel Hilbert space). Let Ω⊆C\Omega\subseteq\mathbb{C} and H\mathcal{H} be a Hilbert space of functions Ω→C\Omega\to\mathbb{C}. If the point evaluation functionals

Tλ:H→C,Tλ(F):=F(λ)(λ∈Ω)\mathcal{T}_{\lambda}: \mathcal{H} \to\mathbb{C}, \qquad\mathcal{T}_{\lambda}(F) := F(\lambda) \quad(\lambda\in\Omega)

are bounded (continuous) for all λ∈Ω\lambda\in\Omega, then H\mathcal{H} is called a reproducing kernel Hilbert space.

By Riesz’s representation theorem (see, e.g., [13]), any bounded linear operator K\mathcal{K} acting on a Hilbert space H\mathcal{H} can be expressed as

K(F)=⟨F,hK⟩H(F∈H)\mathcal{K}(F) = \langle F, h_{\mathcal{K}}\rangle_{\mathcal{H}} \quad(F \in\mathcal{H})

for some unique hK∈Hh_{\mathcal{K}} \in\mathcal{H}. Hence, if H\mathcal{H} is an RKHS, for all λ∈Ω\lambda\in\Omega, there is a unique ξλ∈H\xi_{\lambda} \in\mathcal{H} such that

Tλ(F)=F(λ)=⟨F,ξλ⟩H(F∈H).\mathcal{T}_{\lambda}(F) = F(\lambda) = \langle F, \xi_{\lambda}\rangle_{\mathcal{H}} \quad(F \in\mathcal{H}).

This gives rise to the concept of a reproducing kernel.

Definition 2.2 (Reproducing kernel, kernel slice). Let H\mathcal{H} be an RKHS of functions C⊇Ω→C\mathbb{C} \supseteq\Omega\to\mathbb{C}. The map

ξ:Ω×Ω→C,\xi: \Omega\times\Omega\to\mathbb{C},

that satisfies

Tλ(F)=F(λ)=⟨F,ξ(⋅,λ)⟩H=⟨F,ξλ⟩H(F∈H)T_{\lambda}(F) = F(\lambda) = \langle F, \xi(\cdot,\lambda)\rangle_{\mathcal{H}} = \langle F,\xi_{\lambda}\rangle_{\mathcal{H}} \qquad(F \in\mathcal{H})

is called the reproducing kernel associated with H\mathcal{H}. For some fixed λ∈Ω\lambda\in\Omega, the map

ξλ:=ξ(⋅,λ)∈H\xi_{\lambda} := \xi(\cdot,\lambda) \in\mathcal{H}

is called a kernel slice.

It can verified that a reproducing kernel is Hermitian (i.e., ξ(x,λ)=ξ(λ,x)‾\xi(x,\lambda) = \overline{\xi(\lambda,x)} for all x,λ∈Ωx,\lambda\in\Omega) and positive semi–definite (i.e., ∑i,j=0n−1cicj‾ξ(xi,xj)≥0\sum_{i,j=0}^{n-1}c_i\overline{c_j}\xi(x_i,x_j) \ge0 for all x0,…,xn−1x_0,\ldots,x_{n-1}, c0,…,cn−1∈Cc_0,\ldots,c_{n-1} \in\mathbb{C} and n∈Nn \in\mathbb{N}). By the Moore–Aronszajn theorem [2], an RKHS and its associated reproducing kernel are in one–to–one correspondence. RKHSs and kernels have many useful properties, which have been exploited in various data–driven modeling applications such as machine learning (see, e.g., [20, 33, 65]). In § 3.1, we briefly review some of these properties necessary for the methods proposed in this study. For a thorough review of RKHSs, we recommend [52].

We are now ready to formally state the problem of MOR in an RKHS. For this, we first define generalized model spaces.

Definition 2.3 (Generalized model space, dimension). Let Ω⊆C\Omega\subseteq\mathbb{C} and H\mathcal{H} be an RKHS of functions Ω→C\Omega\to\mathbb{C} with reproducing kernel ξ\xi. The generalized generalized model space in H\mathcal{H} of dimension n∈Nn \in\mathbb{N} is defined by

Gξn:={G∈span⁡{ξλ0,…,ξλn−1}:(λ0,…,λn−1)∈Ω#n}.\mathcal{G}_{\xi}^{n} := \left\{G \in\operatorname{span}\{\xi_{\lambda_0},\ldots,\xi_{\lambda_{n-1}}\} : (\lambda_0,\ldots,\lambda_{n-1}) \in\Omega_{\#}^{n}\right\}.

where we define

Ω#n:={(λ0,…,λn−1)∈Ωn:λj≠λk for j≠k}.\Omega_{\#}^{n} := \left\{(\lambda_0,\ldots,\lambda_{n-1}) \in\Omega^{n} : \lambda_j \ne\lambda_k \text{ for } j \ne k\right\}.

We refer to the functions in generalized model spaces as models throughout the paper. Our objective in this study is to approximate models with functions that adhere to a special structure, e.g., simpler models. To obtain a meaningful MOR problem, we need to define the order, which serves as a measure of model complexity.

Definition 2.4 (Model order). Let n∈Nn \in\mathbb{N} be the order of the model GG, if G∈Gξn⊆HG \in\mathcal{G}_{\xi}^{n} \subseteq\mathcal{H}, but G∉Gξr⊆HG \notin\mathcal{G}_{\xi}^{r} \subseteq\mathcal{H} for any r<nr < n.

In § 6 and § 8, we encounter problems where we need to transform the FOM into a companion admissible RKHS. This will be done via a unitary transformation F\mathcal{F} acting between the original and companion RKHSs. For such cases, the ROM belongs to a weaker notion of a model space (in the original RKHS), specified by the following definition.

Definition 2.5 (Weak model space, dimension, weak model, order). Let Ω⊆C\Omega\subseteq\mathbb{C} and H\mathcal{H} be an RKHS of functions Ω→C\Omega\to\mathbb{C} with reproducing kernel ξ\xi. Let furthermore Ω^⊆C\widehat{\Omega} \subseteq\mathbb{C} and H^\widehat{\mathcal{H}}

be an RKHS of functions Ω^→C\widehat{\Omega} \to\mathbb{C} with kernel ξ^\widehat{\xi}. The weak model space in H\mathcal{H} (with respect to H^\widehat{\mathcal{H}}) of dimension n∈Nn \in\mathbb{N} is

Wξn:={G=F−1[G^]:G^∈Gξn⊆H^, F:H→H^ is unitary}.\mathbb{W}_{\xi}^{n} := \left\{ \mathsf{G} = \mathcal{F}^{-1}[\widehat{\mathsf{G}}] : \widehat{\mathsf{G}} \in\mathbb{G}_{\xi}^{n} \subseteq\widehat{\mathcal{H}},\ \mathcal{F} : \mathcal{H} \to\widehat{\mathcal{H}}\text{ is unitary} \right\}.

If G∈Wξn\mathsf{G} \in\mathbb{W}_{\xi}^{n}, but G∉Wξr\mathsf{G} \notin\mathbb{W}_{\xi}^{r} for any r<nr < n, we say that G\mathsf{G} is a H\mathcal{H}–weak model of order nn.

Remark 2.6. (a) We have G∈Wξn\mathsf{G} \in\mathbb{W}_{\xi}^{n} if it can be written as a linear combination of kernel slice preimages under a given unitary transformation F:H→H^\mathcal{F} : \mathcal{H} \to\widehat{\mathcal{H}}.

(b) For an H\mathcal{H}–weak model G∈Wξn\mathsf{G} \in\mathbb{W}_{\xi}^{n}, there is no guarantee that there exist nn points (λ0,…,λn−1)∈Ω^n(\lambda_0,\ldots,\lambda_{n-1}) \in\widehat{\Omega}^{n} and r0,…,rn−1∈C\mathfrak{r}_0,\ldots,\mathfrak{r}_{n-1} \in\mathbb{C} such that

G=∑k=0n−1rkξλk.\mathsf{G} = \sum_{k=0}^{n-1} \mathfrak{r}_k \xi_{\lambda_k}.

where ξ\xi is the reproducing kernel of H\mathcal{H}. On the other hand, by Definition 2.5, G∈Wξn\mathsf{G} \in\mathbb{W}_{\xi}^{n} belongs to an nn–dimensional subspace of H\mathcal{H} denoted by Wn\mathcal{W}_n. Here, Wn\mathcal{W}_n is an nn–dimensional RKHS in its own right (see Theorem 3.6 in § 3). Consequently, G\mathsf{G} can be written as

G=∑k=0n−1wkξλkWn.\mathsf{G} = \sum_{k=0}^{n-1} \mathfrak{w}_k \xi_{\lambda_k}^{\mathcal{W}_n}.

for some (λ0,…,λn−1)∈Ωn(\lambda_0,\ldots,\lambda_{n-1}) \in\Omega^n and w0,…,wn−1∈C\mathfrak{w}_0,\ldots,\mathfrak{w}_{n-1} \in\mathbb{C}, where ξWn\xi^{\mathcal{W}_n} denotes the kernel of the nn–dimensional RKHS Wn⊆H\mathcal{W}_n \subseteq\mathcal{H}.

We define norm–optimal MOR in an RKHS H\mathcal{H} as follows.

References

  1. [1]A. Angino, T. Dózsa, and M. Voigt, H₂-optimal model order reduction using hyperbolic geometry. arXiv preprint, 2026, arXiv:2610.16249arxiv.org/abs/2610.16249
  2. [2]N. Aronszajn, Theory of reproducing kernels. Trans. Amer. Math. Soc. 68 (1950), no. 3, 337–404 Zbl 0061.26204 MR 0051437
  3. [3]K. E. Atkinson, An Introduction to Numerical Analysis. 2nd edn., John Wiley & Sons, New York, NY, 1989 Zbl 0402.65001 MR 1007135
  4. [4]M. Barrault, Y. Maday, N. C. Nguyen, and A. T. Patera, An ‘empirical interpolation’ method: application to efficient reduced–basis discretization of partial differential equations. C. R. Math. Acad. Sci. Paris 339 (2004), no. 9, 667–672 Zbl 1061.65118 MR 2103208DOI
  5. [5]U. Baur, P. Benner, and L. Feng, Model order reduction for linear and nonlinear systems: a system–theoretic perspective. Arch. Comput. Methods Eng. 21 (2014), no. 4, 331–358 Zbl 1348.93075 MR 3274477DOI
  6. [6]C. Beattie and S. Gugercin, Model reduction by rational interpolation. In Model Reduction and Approximation: Theory and Algorithms, edited by P. Benner, M. Ohlberger, A. Cohen, and K. Willcox, chap. 7, pp. 297–334, Comput. Sci. Eng. 15, SIAM, Philadelphia, PA, 2017 MR 3672151
  7. [7]R. Becker and R. Rannacher, An optimal control approach to a posteriori error estimation in finite element methods. Acta Numer. 10 (2001), 1–102 Zbl 1105.65349 MR 2009692
  8. [8]P. Benner and T. Breiten, Interpolation–based H₂-model reduction of bilinear control systems. SIAM J. Matrix Anal. Appl. 33 (2012), no. 3, 859–885 Zbl 1256.93027 MR 3023455DOI
  9. [9]J.-P. Berrut and L. N. Trefethen, Barycentric Lagrange interpolation. SIAM Rev. 46 (2004), no. 3, 501–517 Zbl 1061.65006 MR 2115059DOI
  10. [10]D. Blalock, J. J. Gonzalez Ortiz, J. Frankle, and J. Guttag, What is the state of neural network pruning? Proc. Mach. Learn. Syst. 2 (2020), 129–146arxiv.org/abs/2003.03033
  11. [11]A. Borghi and T. Breiten, H₂-optimal rational approximation on general domains. Adv. Comput. Math. 50 (2024), no. 3, article no. 28 Zbl 1541.41011 MR 4734558arxiv.org/abs/2305.01511
  12. [12]A. Borghi and T. Breiten, Data–driven optimal approximation on Hardy spaces in simply connected domains. Adv. Comput. Math. 51 (2025), no. 6, article no. 59 Zbl 8146323 MR 5000002arxiv.org/abs/2507.15837
  13. [13]J. B. Conway, A Course in Functional Analysis. 2nd edn., Grad. Texts in Math. 96, Springer, New York, NY, 2007 Zbl 0706.46003 MR 1070713
  14. [14]I. Daubechies, R. DeVore, S. Foucart, B. Hanin, and G. Petrova, Nonlinear approximation and (deep) ReLU networks. Constr. Approx. 55 (2022), no. 1, 127–172 Zbl 1501.41003 MR 4376561
  15. [15]C. de Boor, Divided differences. Surv. Approx. Theory 1 (2005), 46–69 Zbl 1071.65027 MR 2221566
  16. [16]R. De Figueiredo, A generalized Fock space framework for nonlinear system and signal analysis. IEEE Trans. Circuits Systems 30 (1983), no. 9, 637–647 Zbl 0539.93041 MR 0718397DOI
  17. [17]P. Duren and A. Schuster, Bergman Spaces. Math. Surveys Monogr. 100, American Mathematical Society, Providence, RI, 2004 Zbl 1059.30001 MR 2033762
  18. [18]P. L. Duren, Theory of Hᵖ Spaces. Pure Appl. Math. 38, Academic Press, New York, NY, 1970 Zbl 0215.20203 MR 0268655DOI
  19. [19]T. Dózsa, A. Angino, Z. Szabó, J. Bokor, and M. Voigt, Adaptive kernel methods. arXiv preprint, 2026, arXiv:2601.21707arxiv.org/abs/2601.21707
  20. [20]T. Dózsa, F. Deuschle, B. Cornelis, and P. Kovács, Variable projection support vector machines and some applications using adaptive Hermite expansions. Int. J. Neural Syst. 34 (2024), no. 1, article no. 2450004DOI
  21. [21]T. Dózsa, A. J. R. Pelling, and M. Voigt, Impulse response estimation via Laguerre–Fourier expansion. arXiv preprint, 2026, arXiv:2608.14769arxiv.org/abs/2608.14769
  22. [22]T. Dózsa, F. Schipp, and A. Soumelidis, On Bernoulli’s method. SIAM J. Numer. Anal. 62 (2024), no. 3, 1259–1277 Zbl 1545.65077 MR 4750273DOI
  23. [23]T. Dózsa, M. Szabari, A. Soumelidis, and P. Kovács, Pole identification using discrete Laguerre expansion and variable projection. IFAC-PapersOnLine 56 (2023), no. 2, 3994–3999DOI
  24. [24]T. Dózsa, P. Őri, M. Szabari, E. Simonyi, A. Soumelidis, and I. Lakatos, Brake disc deformation detection using intuitive feature extraction and machine learning. Machines 12 (2024), no. 4, article no. 214DOI
  25. [25]T. Dózsa, P. Őri, G. Ungvári, Z. Szabó, A. Soumelidis, I. Lakatos, and P. Kovács, Detection of brake disc deformation with adaptive wavelet neural networks. IEEE Trans. Instrum. Meas. 75 (2026), article no. 2510713DOI
  26. [26]C. Felder and T. Le, Analogues of finite Blaschke products as inner functions. Bull. Lond. Math. Soc. 54 (2022), no. 4, 1197–1219 Zbl 1529.30058 MR 4468315DOI
  27. [27]G. Flagg, C. Beattie, and S. Gugercin, Convergence of the iterative rational Krylov algorithm. Systems Control Lett. 61 (2012), no. 6, 688–691 Zbl 1401.93051 MR 2924212DOI
  28. [28]S. R. Garcia, J. Mashreghi, and W. T. Ross, Finite Blaschke Products and Their Connections. Springer, Cham, 2018 Zbl 1398.30002 MR 3793610DOI
  29. [29]I. Gözse and A. Soumelidis, Realizing system poles identification on the unit disc based on the Fourier transform of Laguerre coefficients. In 2015 23rd Mediterranean Conference on Control and Automation (MED), pp. 821–826, Torremolinos, Spain, 2015DOI
  30. [30]S. Gugercin and A. C. Antoulas, A survey of model reduction by balanced truncation and some new results. Control. 77 (2004), no. 8, 748–766 Zbl 1061.93022 MR 2072207
  31. [31]S. Gugercin, A. C. Antoulas, and C. Beattie, H₂ model reduction for large-scale linear dynamical systems. SIAM J. Matrix Anal. Appl. 30 (2008), no. 2, 609–638 Zbl 1159.93318 MR 2421462
  32. [32]B. Hanin and D. Rolnick, Complexity of linear regions in deep networks. In Proceedings of the 36th International Conference on Machine Learning, edited by K. Chaudhuri and R. Salakhutdinov, pp. 2596–2604, Proc. Mach. Learn. Res. 97, Long Beach, CA, 2019arxiv.org/abs/1901.09021
  33. [33]M. A. Hearst, S. T. Dumais, E. Osuna, J. Platt, and B. Schölkopf, Support vector machines. IEEE Intell. Syst. Appl. 13 (1998), no. 4, 18–28
  34. [34]P. S. C. Heuberger, P. M. J. Van Den Hof, and B. Wahlberg (eds.), Modelling and Identification with Rational Orthogonal Basis Functions. Springer, London, 2005
  35. [35]K. Hornik, Approximation capabilities of multilayer feedforward networks. Neural Netw. 4 (1991), no. 2, 251–257DOI
  36. [36]C. Horowitz, Zeros of functions in the Bergman spaces. Duke Math. J. 41 (1974), no. 4, 693–710 Zbl 0293.30035 MR 0357747DOI
  37. [37]Y. Jang and L. Grigori, Randomized orthogonalization process with reorthogonalization. Numer. Linear Algebra Appl. 32 (2025), no. 4, article no. e70029 Zbl 8127837 MR 4939755
  38. [38]P. E. T. Jorgensen and J. Tian, New duality in choices of feature spaces via kernel analysis. Sampl. Theory Signal Process. Data Anal. 23 (2025), no. 1, article no. 5 Zbl 1575.46021 MR 4867336arxiv.org/abs/2501.11206
  39. [39]D. P. Kingma and J. Ba, Adam: a method for stochastic optimization. arXiv preprint, 2014, presented at the 3rd International Conference on Learning Representations (ICLR), San Diego, CA, USA, 2015, arXiv:1412.6980arxiv.org/abs/1412.6980
  40. [40]P. Kovács, G. Bognár, C. Huber, and M. Huemer, VPNet: variable projection networks. Int. J. Neural Syst. 32 (2022), no. 1, article no. 2150054
  41. [41]P. Kovács, S. Fridli, and F. Schipp, Generalized rational variable projection with application in ECG compression. IEEE Trans. Signal Process. 68 (2020), 478–492 Zbl 1543.94230DOI
  42. [42]L. Meier and D. Luenberger, Approximation of linear constant systems. IEEE Trans. Automat. Control 12 (1967), no. 5, 585–588
  43. [43]J. Mercer, Functions of positive and negative type, and their connection with the theory of integral equations. Philos. Trans. Roy. Soc. London Ser. A 209 (1909), 415–446 Zbl 40.0408.02DOI
  44. [44]W. Mi, T. Qian, and F. Wan, A fast adaptive model reduction method based on Takenaka-Malmquist systems. Systems Control Lett. 61 (2012), no. 1, 223–230 Zbl 1256.93029 MR 2878709DOI
  45. [45]P. Mlinarić, C. A. Beattie, Z. Drmač, and S. Gugercin, IRKA is a Riemannian gradient descent method. IEEE Trans. Automat. Control 70 (2025), no. 5, 2979–2991 Zbl 8062336 MR 4898622arxiv.org/abs/2311.02031
  46. [46]G. Montúfar, R. Pascanu, K. Cho, and Y. Bengio, On the number of linear regions of deep neural networks. In Advances in Neural Information Processing Systems 27, edited by Z. Ghahramani, M. Welling, C. Cortes, N. D. Lawrence, and K. Q. Weinberger, pp. 2924–2932, Montreal, QC, Canada, 2014arxiv.org/abs/1402.1869
  47. [47]S. Müller and R. Schaback, A Newton basis for kernel spaces. J. Approx. Theory 161 (2009), no. 2, 645–655 Zbl 1185.41004 MR 2563073
  48. [48]Z. Nagy-Csiha, M. Pap, and F. Weisz, Construction of rational interpolations using Malmquist-Takenaka systems. Constr. Math. Anal. 6 (2023), no. 1, 55–76 Zbl 1524.42052 MR 4566456DOI
  49. [49]G. Obinata and B. D. O. Anderson, Model Reduction for Control System Design. Comm. Control Engrg. Ser., Springer, London, 2001
  50. [50]M. Pap, Hyperbolic wavelets and multiresolution in H²(T). J. Fourier Anal. Appl. 17 (2011), no. 5, 755–776 Zbl 1228.24042 MR 2838106
  51. [51]M. Pap, Multiresolution in the Bergman space. Ann. Univ. Sci. Budapest. Sect. Comput. 39 (2013), 333–353 Zbl 1289.30234arxiv.org/abs/1105.5729
  52. [52]V. I. Paulsen and M. Raghupathi, An Introduction to the Theory of Reproducing Kernel Hilbert Spaces. Cambridge Stud. Adv. Math. 152, Cambridge University Press, Cambridge, 2016 Zbl 1364.46004 MR 3526117DOI
  53. [53]P. P. Petrushev and V. A. Popov, Rational Approximation of Real Functions. Encyclopedia Math. Appl. 28, Cambridge University Press, Cambridge, 1987 Zbl 0644.41010 MR 0940242DOI
  54. [54]G. Plonka, Y. Riebe, and Y. Kolomoitsev, Spline representation and redundancies of one-dimensional ReLU neural network models. Anal. Appl. (Singap.) 21 (2023), no. 1, 127–163 Zbl 1507.41005 MR 4539213arxiv.org/abs/2207.14609
  55. [55]T. Qian, Kernel Takenaka-Malmquist systems and adaptive approximation in boundary weighted Hardy spaces. arXiv preprint, 2026, arXiv:2607.24473arxiv.org/abs/2607.24473
  56. [56]J. N. Reddy, An Introduction to the Finite Element Method. 2nd edn., McGraw-Hill Ser. Mech. Engrg., McGraw-Hill, New York, 1993
  57. [57]F. Schipp and J. Bokor, Rational bases generated by Blaschke product systems. IFAC Proc. Vol. 36 (2003), no. 16, 1309–1314DOI
  58. [58]C. Schröder and M. Voigt, Balanced truncation model reduction with a priori error bounds for LTI systems with nonzero initial value. J. Comput. Appl. Math. 420 (2023), article no. 114708 Zbl 1504.93049 MR 4497919arxiv.org/abs/2006.02495
  59. [59]A. Shilton, S. Gupta, S. Rana, and S. Venkatesh, Reproducing kernel Banach space models for neural networks with application to Rademacher complexity analysis. In Advances in Neural Information Processing Systems 38, pp. 77765–77809, Neural Information Processing Systems Foundation, San Diego, CA, USA and Mexico City, Mexico, 2025DOI
  60. [60]A. Soumelidis, J. Bokor, F. Schipp, and Z. Szabó, Hyperbolic geometrical approach to model reduction. IFAC-PapersOnLine 50 (2017), no. 1, 12905–12910
  61. [61]I. Stakgold and M. J. Holst, Green’s Functions and Boundary Value Problems. 3rd edn., Pure Appl. Math. (Hoboken) 99, John Wiley & Sons, Hoboken, NJ, 2011 Zbl 1221.35001 MR 2789179
  62. [62]G. Stepaniants, Learning partial differential equations in reproducing kernel Hilbert spaces. J. Mach. Learn. Res. 24 (2023), article no. 86 Zbl 1576.68175 MR 4582508
  63. [63]Z. Szabó, J. Bokor, and F. Schipp, Identification of rational approximate models in H∞ using generalized orthonormal basis. IEEE Trans. Automat. Control 44 (1999), no. 1, 153–158 Zbl 1056.93528 MR 1666925
  64. [64]T. Terao, K. Ozaki, and T. Ogita, LU-Cholesky QR algorithms for thin QR decomposition. Parallel Comput. 92 (2020), article no. 102571 ML 4057043
  65. [65]V. Vapnik and R. Izmailov, Complete statistical theory of learning: Learning using statistical invariants. In Proceedings of the Ninth Symposium on Conformal and Probabilistic Prediction and Applications, edited by A. Gammerman, V. Vovk, Z. Luo, E. Smirnov, and G. Cherubin, pp. 4–40. Proc. Mach. Learn. Res. 128, 2020
  66. [66]A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, Attention is all you need. In Advances in Neural Information Processing Systems 30, edited by I. Guyon, U. von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. V. N. Vishwanathan, and R. Garnett, pp. 5998–6008, Long Beach, CA, USA, 2017arxiv.org/abs/1706.03762
  67. [67]P. Vincent and Y. Bengio, Kernel matching pursuit. Mach. Learn. 48 (2002), no. 1-3, 165–187 Zbl 0998.68120
  68. [68]G. Wahba, Spline Models for Observational Data. CBMS-NSF Regional Conf. Ser. in Appl. Math. 59, SIAM, Philadelphia, PA, 1990 Zbl 0813.62001 MR 1045442
  69. [69]B. Wahlberg, System identification using Laguerre models. IEEE Trans. Automat. Control 36 (1991), no. 5, 551–562 Zbl 0738.93078 MR 1101710
  70. [70]X. Wei, C. Shen, J. Yang, and H. V. Poor, Random orthogonalization for federated learning in massive MIMO systems. IEEE Trans. Wireless Commun. 23 (2024), no. 3, 2469–2485
  71. [71]S.-S. Xie, S. Heo, S. Kim, G. Woo, and S. Yi, Numerical solution of one-dimensional Burgers’ equation using reproducing kernel function. J. Comput. Appl. Math. 214 (2008), no. 2, 417–434 Zbl 1140.65069 MR 2398343DOI
  72. [72]D.-X. Zhou, Derivative reproducing properties for kernel methods in learning theory. J. Comput. Appl. Math. 220 (2008), no. 1–2, 456–463 Zbl 1152.68049 MR 2444183DOI
  73. [73]K. Zhu, Analysis on Fock Spaces. Grad. Texts in Math. 263, Springer, New York, NY, 2012 Zbl 1262.30003 MR 2934601DOI

Paper details

vlnc:W_zp74dvcv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents