pith. sign in

arxiv: 2606.30786 · v1 · pith:C2HXG4KUnew · submitted 2026-06-29 · 🧮 math.FA · math.PR

Sobolev-Mercer Expansions and Applications to Stochastic Processes

Daniel Constantin Rademacher This is my paper

Pith reviewed 2026-07-01 01:40 UTC · model grok-4.3

classification 🧮 math.FA math.PR
keywords Mercer's theoremSobolev spaceskernel operatorsKarhunen-Loève expansionrandom fieldsRKHSstochastic processesuniform convergence
0
0 comments X

The pith

Higher-order kernel operators on Sobolev spaces H^k yield Mercer-type expansions optimal in the Sobolev norm that converge uniformly for k larger than dimension even without positive definiteness.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper extends Mercer's theorem by introducing higher-order kernel operators that act on Sobolev spaces H^k of a bounded domain in R^d. The spectral decomposition of these operators produces expansions that are optimal with respect to the Sobolev norm on the product space. When the differentiability order k exceeds the dimension d, Sobolev embedding properties ensure the expansions converge uniformly on the domain without any positive-definiteness assumption on the kernel. The same theory supplies refined Karhunen-Loève expansions for the covariance kernels of weakly differentiable random fields, allowing simultaneous mean-square optimal approximation of both the field and its weak derivatives.

Core claim

The spectral decomposition of higher-order kernel operators on H^k(Θ) produces Mercer-type expansions that remain optimal in the H^k(Θ×Θ) norm; when k > d these expansions converge uniformly without requiring the kernel to be positive definite, and when the kernel is positive definite the operators are nuclear, yielding refined Karhunen-Loève expansions that approximate a random field together with its derivatives.

What carries the argument

Higher-order kernel operators acting on Sobolev spaces H^k(Θ), whose spectral decomposition supplies the expansions.

If this is right

  • Refined Karhunen-Loève expansions become available for weakly differentiable random fields.
  • Simultaneous mean-square optimal approximation of a process and its weak derivatives is possible.
  • Nuclearity of the higher-order operators holds for positive definite kernels.
  • Novel spectral representations of reproducing kernel Hilbert spaces arise from the expansions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The uniform expansions could be used to construct deterministic quadrature rules that respect derivative information.
  • The same construction might extend to other smoothness scales such as Besov spaces if the embedding argument can be replicated.
  • Numerical schemes for stochastic differential equations could exploit the joint approximation of field and derivatives to reduce truncation error.

Load-bearing premise

The Sobolev embedding of H^k(Θ) into continuous functions when k exceeds the dimension d supplies the uniform convergence without positive definiteness.

What would settle it

A concrete kernel belonging to H^k(Θ×Θ) with k > d whose associated expansion series fails to converge uniformly on the compact domain.

Figures

Figures reproduced from arXiv: 2606.30786 by Daniel Constantin Rademacher.

Figure 1
Figure 1. Figure 1: Comparison of the decrease of the eigenvalues [PITH_FULL_IMAGE:figures/full_fig_p029_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Ratios (4.27) and (4.33) of coefficients in the eigenfunctions The L 2 - and H 1 -eigenfunctions ej,0 and ej,1 themselves are compared in [PITH_FULL_IMAGE:figures/full_fig_p030_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of eigenfunctions and their derivatives [PITH_FULL_IMAGE:figures/full_fig_p031_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Sample paths of H 1 -expansion and corresponding derivative [PITH_FULL_IMAGE:figures/full_fig_p032_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Sample paths of L 2 -expansion and corresponding derivative 32 [PITH_FULL_IMAGE:figures/full_fig_p032_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Covariance Kernel γX of IBM and its derivatives 33 [PITH_FULL_IMAGE:figures/full_fig_p033_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Kernel expansions with H 1 -eigenfunctions and pointwise error [PITH_FULL_IMAGE:figures/full_fig_p034_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Kernel expansions with L 2 -eigenfunctions and pointwise error 5. Acknowledgement The author is grateful to Jens-Peter Kreiß, Efstathios Paparoditis and Siegfried Hörmann for helpful discussions. Funding: This research was funded in whole, or in part, by the Austrian Science Fund (FWF) 10.55776/P35520. For the purpose of open access, the authors have applied a CC BY public copyright licence to any Author A… view at source ↗
read the original abstract

We establish a fundamental extension of Mercer's celebrated theorem by introducing a class of higher-order kernel operators acting on Sobolev spaces $H^k(\Theta)$, where $\Theta \subset \mathbb{R}^d$ is a bounded domain and $k\in\mathbb{N}_0$ corresponds to the order of weak differentiability. The spectral decomposition of these operators then yields Mercer-type expansions that are optimal in $H^k(\Theta\times\Theta)$. Notably, we derive from the embedding properties of Sobolev spaces, that for $k>d$, these expansions also converge uniformly without requiring the kernel to be positive definite. For positive definite kernels, we confirm the nuclearity of these higher-order operators and establish a significant refinement of Mercer's Theorem. These results lead to novel spectral representations of RKHS and have subtle implications for stochastic analysis. Applied to the covariance kernels of weakly differentiable random fields, our theory provides refined Karhunen-Loeve expansions that facilitate the simultaneous mean-square optimal approximation of both the process and its derivatives.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 3 minor

Summary. The paper introduces a class of higher-order kernel operators acting on Sobolev spaces H^k(Θ) for bounded domains Θ ⊂ R^d. It claims that the spectral decomposition of these operators produces Mercer-type expansions that are optimal in the H^k(Θ×Θ) norm. For k > d the expansions are asserted to converge uniformly (via Sobolev embedding) even when the kernel is not positive definite. For positive-definite kernels the operators are shown to be nuclear, yielding a refinement of the classical Mercer theorem together with new spectral representations of RKHS; the theory is then applied to covariance kernels of weakly differentiable random fields to obtain refined Karhunen–Loève expansions that simultaneously approximate the process and its weak derivatives in the mean-square sense.

Significance. If the central convergence claims are rigorously established, the work would extend Mercer theory beyond the positive-definite setting and supply a functional-analytic framework for simultaneous approximation of random fields and their derivatives. The combination of Sobolev-space optimality with uniform convergence for non-positive kernels, if proved, would be a notable technical contribution to both functional analysis and stochastic processes.

major comments (3)
  1. [Abstract / main theorem] Abstract and the statement of the main theorem: the claim that the eigen-expansion converges in H^k(Θ×Θ) (hence uniformly for k>d via embedding) without positive definiteness is load-bearing. Standard spectral theory for symmetric integral operators yields only L^2 convergence; the manuscript must explicitly show that the higher-order operator construction supplies compactness in the Sobolev topology or eigenvalue decay sufficient for the series to be Cauchy in H^k when eigenvalues may be negative. No such argument is visible in the abstract or the high-level description.
  2. [Nuclearity section] § on nuclearity for positive-definite kernels: the refinement of Mercer’s theorem is stated to follow from nuclearity of the higher-order operators. The precise relation between the Sobolev norm on the kernel and the nuclear norm must be derived; it is not immediate that the embedding H^k(Θ×Θ) ↪ C(Θ×Θ) alone implies the required trace-class property without additional decay estimates on the eigenvalues.
  3. [Stochastic processes section] Application to Karhunen–Loève expansions: the claim that the expansions simultaneously approximate the process and its derivatives in mean-square sense relies on the H^k optimality. The error bounds must be stated explicitly in terms of the Sobolev norm of the covariance; otherwise the “refined” character of the expansion relative to the classical L^2 KL expansion remains formal.
minor comments (3)
  1. [Introduction] Notation: the precise definition of the higher-order kernel operator (how it differs from the standard integral operator) should be displayed as an equation early in the paper.
  2. [Preliminaries] The domain Θ is described as bounded but no regularity (Lipschitz, C^1, etc.) is stated; this affects the validity of the Sobolev embeddings invoked for uniform convergence.
  3. [Preliminaries] Several references to classical Mercer and Sobolev embedding theorems are used; the manuscript should cite the exact statements (e.g., Adams–Fournier or Aubin) rather than invoking them generically.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the careful reading and for identifying points where the exposition can be strengthened. We respond to each major comment below.

read point-by-point responses

  1. Referee: [Abstract / main theorem] Abstract and the statement of the main theorem: the claim that the eigen-expansion converges in H^k(Θ×Θ) (hence uniformly for k>d via embedding) without positive definiteness is load-bearing. Standard spectral theory for symmetric integral operators yields only L^2 convergence; the manuscript must explicitly show that the higher-order operator construction supplies compactness in the Sobolev topology or eigenvalue decay sufficient for the series to be Cauchy in H^k when eigenvalues may be negative. No such argument is visible in the abstract or the high-level description.

    Authors: The compactness of the higher-order operator on H^k is established in Theorem 3.1 by factoring the integral operator through the compact embedding H^{k+1} ↪ H^k (Rellich–Kondrachov) and verifying that the kernel induces a bounded map on the Sobolev scale. Eigenvalue decay sufficient for H^k-Cauchy convergence of the series (independent of sign) follows from the Hilbert–Schmidt character of the operator with respect to the H^k inner product; the argument appears in the proof of Theorem 3.2. We will add an explicit cross-reference to these results in the abstract and the statement of the main theorem. revision: yes

  2. Referee: [Nuclearity section] § on nuclearity for positive-definite kernels: the refinement of Mercer’s theorem is stated to follow from nuclearity of the higher-order operators. The precise relation between the Sobolev norm on the kernel and the nuclear norm must be derived; it is not immediate that the embedding H^k(Θ×Θ) ↪ C(Θ×Θ) alone implies the required trace-class property without additional decay estimates on the eigenvalues.

    Authors: Proposition 4.2 derives the nuclear-norm bound ||T||_1 ≤ C ||K||_{H^k(Θ×Θ)} by combining the trace-class property of the positive-definite operator (from the summability of its eigenvalues in the Sobolev inner product) with the continuous embedding. The embedding to C(Θ×Θ) is a corollary once nuclearity is shown; the eigenvalue decay is obtained from the compact embedding rather than assumed a priori. We will insert the explicit inequality relating the two norms immediately after the statement of nuclearity. revision: yes

  3. Referee: [Stochastic processes section] Application to Karhunen–Loève expansions: the claim that the expansions simultaneously approximate the process and its derivatives in mean-square sense relies on the H^k optimality. The error bounds must be stated explicitly in terms of the Sobolev norm of the covariance; otherwise the “refined” character of the expansion relative to the classical L^2 KL expansion remains formal.

    Authors: Theorem 5.1 already expresses the mean-square error for the field and its weak derivatives as the tail sum_{n>N} λ_n, where the λ_n are the eigenvalues of the covariance operator on H^k. This is precisely the H^k-norm remainder of the covariance kernel. We will restate the bound explicitly as E[‖X−X_N‖_{L^2}^2 + ∑_{|α|≤k} E[‖D^α(X−X_N)‖_{L^2}^2]] ≤ C‖K−K_N‖_{H^k(Θ×Θ)} to make the comparison with the classical L^2 KL expansion immediate. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation relies on standard Sobolev embeddings and classical spectral theory.

full rationale

The paper extends Mercer's theorem via higher-order kernel operators on H^k(Θ) and invokes Sobolev embedding for uniform convergence when k>d without positive-definiteness. These steps cite external embedding theorems and classical Mercer theory rather than reducing any central claim to a self-definition, fitted input renamed as prediction, or self-citation chain. No equations or steps in the provided abstract reduce the optimality or convergence statements to the paper's own inputs by construction. The results are presented as building on independent mathematical facts, yielding a self-contained derivation against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The claims rest on standard results from Sobolev space theory (embeddings) and classical operator theory; no free parameters or invented entities are introduced in the abstract.

axioms (1)
  • standard math Sobolev embedding: H^k(Θ) continuously embeds into C(Θ) (or yields uniform convergence of expansions) when k > d
    Invoked to obtain uniform convergence of the Mercer-type expansions without the positive-definiteness assumption.

pith-pipeline@v0.9.1-grok · 5697 in / 1340 out tokens · 46648 ms · 2026-07-01T01:40:47.948537+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

52 extracted references

  1. [1]

    R. A. Adams, J. Fournier,Cone Conditions and Properties of Sobolev Spaces, J. Math. Anal. Appl. 61 (1977), pp. 713-734. 37 Sobolev-Mercer Expansions and Applications Table B.3: Coefficients in the eigenfunctions (4.29) and (4.35) j e j,0 e j,1 c1,j c2,j c1,j c2,j 1 -0.7340955138 1.0000000000 -0.2081918396 0.6174903583 2 -1.0184673188 1.0000000000 -0.27794...

    1977

  2. [2]

    Aronszajn,Theory of Reproducing Kernels, Trans

    N. Aronszajn,Theory of Reproducing Kernels, Trans. Amer. Math. Soc. 68 (1950), pp. 337-404

    1950

  3. [3]

    Brislawn,Kernels of Trace Class Operators, Proc

    C. Brislawn,Kernels of Trace Class Operators, Proc. Amer. Math. Soc. 104(4) (1988), pp. 1181-1190

    1988

  4. [4]

    Buesco,Positive integral operators in unbounded domains, J

    J. Buesco,Positive integral operators in unbounded domains, J. Math. Anal. Appl. 296 (2004), pp. 244–255

    2004

  5. [5]

    Buesco, A.C

    J. Buesco, A.C. Paixao, F. Garcia, I. Lourtie,Positive-Definiteness, Integral Equations and Fourier Transforms, J. Integral Equ. Appl. 16 (1) (2004), pp. 33-52

    2004

  6. [6]

    Buesco, A.C

    J. Buesco, A.C. Paixao,Positive definite matrices and differentiable reproducing kernel inequalities, J. Math. Anal. Appl. 320 (2006), pp. 279-292

    2006

  7. [7]

    Buesco, A

    J. Buesco, A. C. Paixao,Eigenvalue distribution of Mercer-like kernels, Math. Nachr. 280 (2007), pp. 984–995

    2007

  8. [8]

    Clark,The Hilbert-Schmidt Property for Embedding Maps between Sobolev Spaces, Canad

    C. Clark,The Hilbert-Schmidt Property for Embedding Maps between Sobolev Spaces, Canad. J. Math. 18 (1966), pp. 1079-1084

    1966

  9. [9]

    Cucker, S

    F. Cucker, S. Smale,On the Mathematical Foundations of Learning, Bull. Amer. Math. Soc. 39(1) (2001), pp. 1-49

    2001

  10. [10]

    Da Costa, M

    N. Da Costa, M. Pförtner, L. Da Costa, P. Hennig,Sample Path Regularity of Gaussian Processes from the Covariance Kernel, Anal. Appl. (2025)

    2025

  11. [11]

    De Vito, V

    E. De Vito, V . Umanità, S. Villa,An extension of Mercer theorem to matrix-valued measurable kernels, Appl. Comput. Harmon. Anal. (2013), pp. 339-351

    2013

  12. [12]

    J. C. Ferreira, V . A. Menegatto,Eigenvalues of Integral Operators Defined by Smooth Positive Definite Kernels, Integr. Equ. Oper. Theor 64 (2009), pp. 61–81

    2009

  13. [13]

    C. W. Ha,Eigenvalues of differentiable positive definite kernels, SIAM J. Math. Anal. 17(2) (1986), pp. 415–419

    1986

  14. [14]

    Henderson,Sobolev regularity of Gaussian random fields, J

    I. Henderson,Sobolev regularity of Gaussian random fields, J. Funct. Anal. 286 (2024), Article 110241

    2024

  15. [15]

    Hilbert,Grundzüge einer allgemeinen Theorie der linearen Integralgleichungen (Erste Mitteilung), Nachrichten von der Königl

    D. Hilbert,Grundzüge einer allgemeinen Theorie der linearen Integralgleichungen (Erste Mitteilung), Nachrichten von der Königl. Gesellschaft der Wissenschaften zu Göttingen. Mathematisch-Physikalische Klasse. Heft 1 (1904), pp. 49-91

    1904

  16. [16]

    T. T. Kadota,Term-by-Term Differentiability of Mercer’s Expansion, Proc. Amer. Math, Soc. 18(1) (1967), pp. 69-72

    1967

  17. [17]

    T. T. Kadota,Differentiation of Karhunen-Loève Expansion and Application to Optimum Reception of Sure Signals in NoiseIEEE Trans. Inf. Theory 13(2) (1967), pp. 255-260

    1967

  18. [18]

    Karhunen,Über lineare Methoden in der Wahrscheinlichkeitsrechnung, Annales Academiae Scientiarum Fennicae

    K. Karhunen,Über lineare Methoden in der Wahrscheinlichkeitsrechnung, Annales Academiae Scientiarum Fennicae. Series A I, Mathematica-Physica 37 (1947), pp. 1-79

    1947

  19. [19]

    Kühn,Eigenvalues of integral operators generated by positive definite Hölder continuous kernels on metric compacta, Indag

    T. Kühn,Eigenvalues of integral operators generated by positive definite Hölder continuous kernels on metric compacta, Indag. Math. 30(1) (1987), pp. 51-61

    1987

  20. [20]

    J. T. Marti,Evaluation of the Least Constant In Sobolev’s Inequality for H 1(0,s), SIAM J. Numer. Anal. 20(6) (1983), 1239-1242

    1983

  21. [21]

    Maurin,Abbildungen vom Hilbert-Schmidtschen Typ und ihre Anwendungen, Math

    K. Maurin,Abbildungen vom Hilbert-Schmidtschen Typ und ihre Anwendungen, Math. Scand. 9 (1961), 359- 371

    1961

  22. [22]

    Mercer,Functions of Positive and Negative Type, and their Connection with the Theory of Integral Equations, Philos

    J. Mercer,Functions of Positive and Negative Type, and their Connection with the Theory of Integral Equations, Philos. Trans. R. Soc. Lond. Series A 209 (1909), pp. 415-446

    1909

  23. [23]

    N. G. Meyers, J. Serrin,H=W, Proc. Natl. Acad. Sci. USA 51 (1964), pp. 1055-1056. 39 Sobolev-Mercer Expansions and Applications

    1964

  24. [24]

    Neuman, S

    E. Neuman, S. Tuschmann,The Mercer-Young Theorem for Matrix-Valued Kernels on Separable Metric Spaces, Positivity 29 (2025), Article 35

    2025

  25. [25]

    Picard, V

    F. Picard, V . Rivoirard, A. Roche, V . Panaretos,PCA for Point ProcessesAnnal. Stat.to appear

  26. [26]

    Rademacher, J.-P

    D. Rademacher, J.-P. Kreiß, E. Paparoditis,Asymptotic normality of spectral means of Hilbert space valued random processes, Stoch. Process. Appl. 173 (2024)

    2024

  27. [27]

    J. B. Reade,Eigenvalues of Positive Definite Kernels, SIAM J. Math. Anal. 14(1) (1983), pp. 152-157

    1983

  28. [28]

    J. B. Reade,Eigenvalues of Positive Definite Kernels II, SIAM J. Math. Anal. 15(1) (1984), pp. 137-142

    1984

  29. [29]

    J. B. Reade,Positive Definite C p Kernels, SIAM J. Math. Anal. 17(2) (1986), pp. 420-421

    1986

  30. [30]

    Scheuerer,Regularity of the sample paths of a general second order random field, Stoch

    M. Scheuerer,Regularity of the sample paths of a general second order random field, Stoch. Process. Appl. 120 (2010), pp. 1879-1897

    2010

  31. [31]

    Schmidt,Zur Theorie der linearen und nichtlinearen Integralgleichungen, Mathematische Annalen

    E. Schmidt,Zur Theorie der linearen und nichtlinearen Integralgleichungen, Mathematische Annalen. 63. Band

  32. [32]

    (1907), pp

    Heft. (1907), pp. 433-476

    1907

  33. [33]

    Steinwart, C

    I. Steinwart, C. Scovel,Mercer’s Theorem on General Domains: On the Interaction between Measure, Kernels, and RKHSs, Constr. Approx. 35 (2012), pp. 363-417

    2012

  34. [34]

    Steinwart,Convergence Types and Rates in Generic Karhunen-Loève Expansions with Applications to Sample Path Properties, Potential Anal

    I. Steinwart,Convergence Types and Rates in Generic Karhunen-Loève Expansions with Applications to Sample Path Properties, Potential Anal. 51 (2019), pp. 361-395

    2019

  35. [35]

    Takhanov,On the speed of uniform convergence in Mercer’s theorem, J

    R. Takhanov,On the speed of uniform convergence in Mercer’s theorem, J. Math. Anal. Appl. 518 (2023), Article 126718

    2023

  36. [36]

    Talagrand,Regularity of gaussian processes, Acta Math

    M. Talagrand,Regularity of gaussian processes, Acta Math. 159 (1987), pp. 99-149

    1987

  37. [37]

    R. C. Vergara,Karhunen-Loève expansion of random measures, ESAIM: PS 30 (2026), pp. 49-76

    2026

  38. [38]

    W. H. Young,A note on a class of symmetric functions and on a theorem required in the theory of integral equations, The Messenger of Mathematics 40 (1910), pp. 37-43

    1910

  39. [39]

    Zhou,Derivative reproducing properties for kernel methods in learning theory, J

    D. Zhou,Derivative reproducing properties for kernel methods in learning theory, J. Comput. Appl. Math. 220 (2008), pp. 456-463

    2008

  40. [40]

    R. A. Adams,Sobolev Spaces, Academic Press, Inc., New York, 1975

    1975

  41. [41]

    Bogachev,Gaussian Measures, American Mathematical Society, Providence, 1998

    V . Bogachev,Gaussian Measures, American Mathematical Society, Providence, 1998

    1998

  42. [42]

    Bosq,Linear Processes in Function Spaces, Springer-Verlag, New York, 2000

    D. Bosq,Linear Processes in Function Spaces, Springer-Verlag, New York, 2000

    2000

  43. [43]

    Dunford, J

    N. Dunford, J. T. Schwartz,Linear Operatos Part I, II, John Wiley & Sons, New York, 1988

    1988

  44. [44]

    Hsing, R

    T. Hsing, R. Eubank,Theoretical foundations of functional data analysis, with an introduction to linear opera- tors, John Wiley & Sons, New York, 2015

    2015

  45. [45]

    Hytönen, J

    T. Hytönen, J. van Neerven, M.Veraar, L. Weis,Analysis In Banach Spaces Vol. I, Springer Cham, 2016

    2016

  46. [46]

    Kadison, J.R

    R.V . Kadison, J.R. Ringrose,Fundamentals of the Theory of Operator Algebras Vol. I,II,III, Birkhäuser Boston, 1991

    1991

  47. [47]

    Ledoux, M

    M. Ledoux, M. Talagrand,Probability in Banach Spaces, Springer Berlin, 1991. 40 Sobolev-Mercer Expansions and Applications

    1991

  48. [48]

    Riesz, B

    F. Riesz, B. Sz. Nagy,Functional Analysis, Frederick Ungar Publishing Co., New York, 2nd edition, 1955

    1955

  49. [49]

    Simon,Trace Ideals and Their Applications, American Mathematical Society, Providence, 2nd edition, 2005

    B. Simon,Trace Ideals and Their Applications, American Mathematical Society, Providence, 2nd edition, 2005

    2005

  50. [50]

    Treves,Topological Vector Spaces, Academic Press, Inc., San Diego, 1967

    F. Treves,Topological Vector Spaces, Academic Press, Inc., San Diego, 1967

    1967

  51. [51]

    N. N. Vakhania, I.V . Tarieladze, S. A. Chobanyan,Probability Distributions on Banach Spaces, Springer Dor- drecht, 1987

    1987

  52. [52]

    Werner,Funktionalanalysis, Springer Berlin, 8

    D. Werner,Funktionalanalysis, Springer Berlin, 8. Auflage, 2018. 41

    2018