Abstract
For the solution of operator equations, Stevenson introduced a definition of frames, where a Hilbert space and its dual are not identified. This means that the Riesz isomorphism is not used as an identification, which, for example, does not make sense for the Sobolev spaces and . In this article, we are going to revisit the concept of Stevenson frames and introduce it for Banach spaces. This is equivalent to -Banach frames. It is known that, if such a system exists, by defining a new inner product and using the Riesz isomorphism, the Banach space is isomorphic to a Hilbert space. In this article, we deal with the contrasting setting, where and are not identified, and equivalent norms are distinguished, and show that in this setting the investigation of -Banach frames make sense.
1. Introduction
The standard definition of frames found first in the paper by Duffin and Schaefer [Citation1] is the following: (1) (1)
Here, means that there are constants such that .
This concept led to a lot of theoretical work, see e.g., [Citation2–6], but has been used also extensively in signal processing [Citation7], quantum mechanics [Citation8], acoustics [Citation9], and various other fields.
Frames can be used also to represent operators. For the numerical solution of operator equations, the (Petrov-) Galerkin scheme [Citation10] is used, where operators are represented by , called the stiffness or system matrix. The collection consists of the ansatz functions, the collection are the test functions. If and live in the same space, this is called Galerkin scheme, otherwise it is called Petrov–Galerkin scheme.
In finite and boundary element approaches, not only bases were used [Citation11, Citation12], but also frames have been applied, e.g., in [Citation13–17]. Recently, such operator representations got also a more theoretical treatment [Citation18–20].
In numerical applications, it is often advantageous to have self-adjoint matrices, e.g., for Krylov subspace methods, which necessitates to use the same sequence for the discretization at both sides, i.e., investigating . Note that this matrix is self-adjoint if O is, and semi-positive if O is. Positivity is in general not preserved, only if a system without redundancy is used, i.e., a Riesz sequence. Partial differential operators are typically operators of the form , while boundary integral operators might also be smoothing operators which map in accordance with . One possible solution is to work with Gelfand triples i.e., . This is explicitly done for the concept of Gelfand frames [Citation21].
Another possibility is the following, introduced by Stevenson in [Citation17] and used, e.g., in [Citation15]: A collection is called a (Stevenson) frame for , if (2) (2)
Note the difference to the definition (1) by Duffin and Schaefer, which is significant only if the Riesz isomorphism is not employed. Here, the Gelfand triple is only implicitly used and, if the fully general setting is used, the density of the spaces is not required.
Clearly, the definitions (1) and (2) are equivalent by the Riesz isomorphism. On the other hand, if the isomorphism is not considered, but another one is utilized, for example, considering the triple , then the Riesz isomorphism is usually used as an identification on the pivot space , and therefore and cannot be considered to be equal.
In this article, we consider the original definition by Stevenson and re-investigate in full detail all the derivation to ensure that the Riesz identification does not ‘creep in’ again.
On a more theoretical level, let us consider Banach frames [Citation22–24]. Thus, we consider a Banach space X, a sequence space Xd, and a sequence . This is a Xd-frame if
It is called a Banach frame if a reconstruction operator exists, i.e., there exists with for all .
In this setting, -frames were not considered to be interesting as they are isomorphic to Hilbert frames, see e.g., [Citation25, Proposition 3.10]: Let be a -frame for X. Then, X can be equipped with an inner product , becoming a Hilbert space, and is a (Hilbert) frame for X. The proof uses the Riesz isomorphism in the last line. But if a context is considered, where this isomorphism cannot be applied, like for example a Gelfand triple setting, suddenly the concept of -frames might become nontrivial again, and the concept of Stevenson frames is different to a (standard Hilbert space) frame. In this article, we investigate this approach.
The rest of this article is structured as follows. In Section 2, we motivate Gelfand triples by a simple example arising from the variational formulation of second-order elliptic partial differential equations. Section 3 then provides the main ingredients we need, especially it introduces the different notions of frames for solving operator equations. By an illustrative example, we show that Stevenson frames seem to offer the most flexible concept for the discretization of operator equations. Finally, in Section 4, we generalize Stevenson frames to Banach spaces and discuss the consequences.
2. Motivation: Solving operator equations
Let and define the bilinear form by . Assume that a satisfies the following properties:
Let a be bounded, i.e., there is a constant CS, such that
This is equivalent to O being bounded.
Let a be elliptic, i.e., there exists a constant CE such that
Both conditions are equivalent to O being bounded, boundedly invertible, and positive, see e.g., [Citation26, Citation27].
The general goal is to find the solution such that (3) (3)
This is the weak formulation of the operator equation , setting for and .
In numerical approximation schemes, to get an approximate solution, finite dimensional subspaces are considered and the solution such that is calculated. The error between the continuous solution and the approximate solution is orthogonal to the space V, which is known as the Galerkin orthogonality: for all . Note that, in difference to, e.g., a Gelfand triple approach, the norms on V and are the same in the setting above. Instead, the Gelfand triple setting would be with .
We shall illustrate the setting also by a practical example from the theory of partial differential equations. To that end, assume that Ω is a bounded domain in and let be the space of all square-integrable functions . As space we consider the Sobolov space which consists of all functions in whose first-order week derivatives are also square-integrable and which are zero at the boundary . Thus, the variational formulation of the Poisson equation reads (4) (4) compare [Citation26] for example. The bilinear form
is continuous and elliptic due to Friedrichs’ inequality, cf. [Citation26], and the linear form is continuous provided that . Hereby, the inner product in the pivot space is continuously extended onto the duality pairing . Hence, the underlying Gelfant triple is .
3. Main definitions and notations
3.1. Dual pairs
Let X, Y be vector spaces and a(x, y) a bilinear functional on X × Y. Then (X, Y) is called a dual pair [Citation28], if
1.
2.
In short, the notation is used. A classical example is a Banach space X and its dual space . But looking at other dual pairs allows to have an explicit form for the dual elements [Citation29].
Note that often an isomorphism is considered as an identity. For example, by using the Riesz mapping , the dual space is often identified with . If two or more isomorphisms are involved, this identification, of course, can only be considered for one of those isomorphisms. For example, if we consider two Hilbert spaces , the Riesz isomorphism can be considered only for one of them to be an identification, see also Section 3.3.2.
3.2. Gelfand triples
Let X be a Banach space and a Hilbert space. Then, the triple is called a Banach Gelfand triple [Citation30], if , where X is dense in , and is -dense in . The prototype of such a triple is in case of sequence spaces.
Note that, even if we consider the spaces all being Hilbert spaces – such a sequence is also called rigged Hilbert spaces [Citation31] – the Riesz isomorphism, in general, is not just the composition of the inclusion with its adjoint. This depends on the chosen concrete dual pairing.
As another example, consider the triple , which has been presented in the practical example for the Poisson equation in Section 2.
3.3. Frames
A sequence in a separable Hilbert space is a frame for , if there exist positive constants and (called lower and upper frame bound, respectively) that satisfy (5) (5)
An upper (resp. lower) semi-frame is a complete system that only satisfies the upper (resp. lower) frame inequality, see [Citation32, Citation33]. A frame where the two bounds can be chosen to be equal, i.e., , is called tight. We will denote the corresponding sequences in by and in the following, where we consider general discrete index sets . A sequence that is a frame for its closed linear span is called a frame sequence.
By we denote the analysis operator defined by . The adjoint of is the synthesis operator . The frame operator can be written as . It is positive and invertible. Note that those ‘frame-related’ operators can be defined as possibly unbounded operators for any sequence in the Hilbert space [Citation34].
By using the canonical dual frame , i.e., for all k, we get a reconstruction formula:
The Gramian matrix is defined by , also called the mass matrix. This matrix defines an operator on by matrix multiplication, corresponding to . Similarly, we can define the cross-Gramian matrix between two different frames and . Clearly,
If, for the sequence , there exist constants such that the inequalities are fulfilled, is called a Riesz sequence. If is complete, it is called a Riesz basis.
3.3.1. Banach frames
The concept of frames can be extended to Banach spaces [Citation22–24]:
Let X be a Banach space and Xd be a Banach space of scalar sequences. A sequence in the dual is called an Xd-frame for the Banach space X, if there exist constants such that
An Xd-frame is called a Banach frame with respect to a sequence space Xd, if there exists a bounded reconstruction operator , such that for all . In our setting, we use p-frames, that is for , especially, we use .
A family is called a q-Riesz sequence for X, if there exist constants such that (6) (6) for all finite scalar sequence . The family is called a q-Riesz basis if it fulfills (6) and .
Any q-Riesz basis for is a p-frame for X, where , compare [Citation23].
3.3.2. Gelfand frames
A frame for is called a Gelfand frame [Citation21] for the Gelfand triple if there exists a Gelfand triple of sequence spaces , such that the synthesis operator and the analysis operator are bounded. As a result, see [Citation21, Citation35], this means that is a Banach frame for Xd and a Banach frame for .
In many approaches, see e.g. [Citation21], it is assumed for the implementation that there exists an isomorphism . Should Xd be nonreflexive, then it is also assumed that is an isomorphism. If DB is a diagonal operator, i.e., and , then is a Hilbert frame for X and is a Hilbert frame for . This is shown for real weights in [Citation36]. It is easy to see also for complex weights when using a weighted frame viewpoint [Citation37, Citation38]. These cases cover the weighted spaces .
The above setting can be generalized as follows: We define, similar to [Citation25], the sesquilinear form . It is obviously bounded and elliptic, and, in particular, is equivalent to . Therefore, is a Hilbert space which is isomorphic to . Now let , where δl is the standard basis in . This is a Hilbert space frame for X. Similarly, is a Hilbert space frame for . As a consequence X and are Hilbert spaces, but and the inner products and the corresponding norms are changed, albeit equivalent to the original ones.
3.4. Stevenson frames
We consider the duality without using the Riesz isomorphism. In particular, we use the duality with respect to a second Hilbert space .
Definition 3.1
([Citation17]). A sequence is called a (Stevenson) frame for if there exists constants such that (7) (7)
Different to the Gelfand frames setting, we do not assume density.
Typically, we consider Sobolev spaces and the L2-inner product, which we can consider as co-orbit spaces with the sequence spaces varying w. Here, invertible operators between different spaces exist, see Section 3.3.2, and density is also given. In this article, we treat the most general setting.
In [Citation17], the author states ‘We adapted the definition of a frame given in [Citation39, Section 3] by identifying with its dual via the Riesz mapping’. Then, the following results are stated, also in [Citation15], without proofs: The analysis operator , is bounded by (7), as is its adjoint . It can be easily shown that is the synthesis operator with . Especially, one has
Define the frame operator . It is a mapping , which is boundedly invertible. We can show that the sequence is a (Stevenson) -frame with bounds and . Here, and . Furthermore, it holds and, therefore, .
We have the reconstructions (8) (8) and (9) (9) for all and .
The cross-Gramian matrix is the orthogonal projection on and coincides with . Therefore, .
In this article, we are revisiting those statements, make them slightly more general, in order to make sure that not using the Riesz isomorphism is possible.
3.5. An illustrative example
Let be a sufficiently smooth, bounded domain. We consider a multiscale analysis, i.e., a dense, nested sequence of finite dimensional subspaces consisting of piecewise polynomial ansatz functions , such that and
One might think here of a multigrid decomposition of standard Lagrangian finite element spaces or of a sequence of spline spaces originating from dyadic subdivision.
Trial spaces Vj which are used for the Galerkin method satisfy typically a direct or Jackson estimate. This means that (10) (10) holds for all uniformly in j. Here, is the -orthogonal projection onto the trial space Vj and denotes the Sobolev space of order q. The upper bound d > 0 refers in general to the maximum order of the polynomials which can be represented in Vj, while the factor refers to the mesh size of Vj, i.e., the diameter of the finite elements, compare [Citation26] for example.
Besides the Jackson type estimate (10), there also holds the inverse or Bernstein estimate (11) (11) for all , where the upper bound refers to the regularity of the functions in the trial spaces Vj. There holds for trial functions based on cardinal B-splines, since they are globally -smooth, and for standard Lagrangian finite element shape functions, since they are only globally continuous.
A crucial requirement is the uniform frame stability of the systems under consideration, i.e., the existence of constants such that (12) (12) holds uniformly for all j. This stability is satisfied for example by Lagrangian finite element basis functions defined on a multigrid hierarchy resulting from uniform refinement of a given coarse grid, see [Citation26] for example. It is also satisfied by B-splines defined on a dyadic subdivision of the domain under consideration.
Having a multiscale analysis at hand, it can be used for telescoping a given function to account for the fact that Sobolev norms act different on different length scales. Namely, the interplay of (10) and (11) gives rise to the norm equivalence (13) (13) for all , where and denotes the dual to , see [Citation40] for a proof.
In accordance with [Citation15], using (12), we can estimate
The latter sum converges provided that q > 0 and we arrive at
In view of the norm equivalence (13), we have thus proven that there exist constants such that (14) (14) for all . Therefore, in accordance with Definition 3.1, the collection (15) (15) defines a Stevenson frame for , where with duality related to . Notice that this frame underlies the construction of the so-called BPX preconditioner, see e.g., [Citation40–42]. Especially, by removing all basis functions which are associated with boundary nodes, one gets a Stevenson frame for , as required for the Galerkin discretization of elliptic partial differential equations, compare Section 2.
We like to emphasize that the collection (15) does not define a Gelfand frame, since (14) does not hold in , i.e., for q = 0. Hence, the concept of Stevenson frames seems to be more flexible than the concept of Gelfand frames.
3.6. Operator representation in frame coordinates
For orthonormal sequences, it is well known that operators can be uniquely described by a matrix representation [Citation43]. The same can be constructed with frames and their duals, see [Citation18, Citation19].
Let be a frame in with bounds , and let be a frame in with .
1. Let be a bounded, linear operator. Thus, the infinite matrix
defines a bounded operator from to with . As an operator , we have
2. On the other hand, let M be an infinite matrix defining a bounded operator from to . Then, the operator defined by
is a bounded operator from to with
and
Please note that there is a classification of matrices that are bounded operators from to [Citation44].
If we start out with frames, more properties can be proved [Citation18]: Let be a frame in with bounds in with .
1. It holds
Therefore, for all :
2. is injective and is surjective.
3. If , then .
4. Let be any frame in , and and . Then, it holds
Note that, in the Hilbert space of Hilbert–Schmidt operators, the tensor product is a Bessel sequence/frame sequence/Riesz sequence, if the starting sequences and are [Citation45], with being the analysis and being the synthesis operator. This relation is even an equivalence [Citation46].
For the invertibility, it can be shown [Citation20, Citation47]: If and only if O is bijective, then is bijective as operator from onto . In this case, one has
If we have an operator equation , we use which implies for all . Setting , and , we thus have
Note that, for numerical computations, see e.g. [Citation17, Citation21], the system of linear equations is solved. Then, is the solution to , avoiding the numerically expensive calculation of a dual frame [Citation48–50]. If the frame is redundant, then uk can be different to . If a Tychonov regularization is used, we obtain by [Citation51, Prop. 5.1.4].
4. Stevenson frames revisited
As some of the references dealing with Stevenson frames used an unlucky formulation, when stating if or if not the Riesz isomorphism is used, see e.g. [Citation17, Citation21], the authors decided to check everything again, and pay particular attention to the avoidance of the Riesz isomorphism, i.e., to not use .
To not use the Riesz isomorphism in a treatment of Hilbert spaces is mind-boggling, so we decided to use Banach spaces, to be sure to avoid all pitfalls. (Note, however, that the Riesz isomorphism will be used on the sequence space .) In particular, this is a generalization of the original definition. The used spaces are necessarily isomorphic to Hilbert spaces, but not Hilbert spaces per se.
4.1. Stevenson Banach frames
We start out with a generalized definition. (We will show that this is isomorphic, but not identical to the original definition.)
Definition 4.1.
Let be a dual pair of reflexive Banach spaces. Let . It is called a Stevenson Banach frame for X, if there exist bounds such that
The analysis operator is bounded by by definition. (Note that we use here the notation which is more common for Banach spaces [Citation24].) As a consequence of the open mapping theorem, is one-to-one and has closed range.
For with finitely many nonzero entries, i.e., , consider
By using a standard density argument and the reflexivity, it can easily be shown that , where is the synthesis operator with . The bound of is also . The sum converges unconditionally. Indeed, consider . Then, let be a finite set, such that
For another finite index set , we thus find
Hence, by e.g. [Citation28, IV.5.1] and the fact that is a Hilbert space, we deduce (16) (16)
We define the frame operator , which is a mapping . In particular, the operator is self-adjoint. By definition of , it follows that (17) (17)
Hence, is bounded with bound . Furthermore, we have (18) (18) which implies that is one-to-one and positive. By [Citation28, IV.5.1], this also means that has dense range. also has a bounded inverse since
Therefore, it has closed range [Citation52, Theorem XI.2.1]. Consequently, is onto and bijective with Thus, is also self-adjoint, and (19) (19)
Theorem 4.1.
The sequence is a Stevenson Banach frame for with bounds and . The range of its analysis operator coincides with the one of the primal frame, i.e., . The related operators are and . For and , we have the reconstructions
Proof.
It obviously holds . Moreover, we have on the one hand and on the other hand
Hence, is an -frame. By employing the invertibility of for , we get
This implies , where and . Furthermore, , because it holds for all .
Finally, we have the reconstructions for all and for all .
As we have the sharper upper bound. On the other hand, since defines a positive sesquilinear form, the Cauchy–Schwarz inequality implies
Thus, with , there holds
and consequently
So, the sharper bounds and follow. □
The fact that is very different to the Gelfand frame setting, where the ranges even live in different sequence spaces.
Theorem 4.2.
The cross-Gramian matrix is the orthogonal projection on and coincides with .
Proof.
We have that the cross-Gramian matrix of a frame and its dual is a projection:
Next, it holds
In addition, since
we conclude . Thus, is self-adjoint.□
Theorem 4.3.
The collection is a Stevenson Banach frame for X with bounds and if and only if
In particular, for any with and , we have .
Proof.
Given we have the representation
Hence,
By (16) and Theorem 4.1, there follows .□
Consequently, a Stevenson frame is a Riesz basis for X if and only if is one-to-one.
4.2. Is a Hilbert space?
Set . This is, trivially, a symmetric and positive bilinear form by above and, therefore, an inner product on . Hence, is a pre-Hilbert space with this inner product. By (17) and (18), the corresponding norm is equivalent to the original one as . Thus, is a Hilbert space.
Note that, in particular for numerics, it is sometimes not enough to consider equivalent norms. While well-posed problems stay well-posed for equivalent norms, this becomes important for concrete implementations, as things like condition numbers, constants in convergence rates, etc. are considered.
From a frame theory perspective, switching to an equivalent norm can destroy or create tightness, in particular, the switch from one norm to the other changes the frame bound ratio . We refer the reader to, e.g., weighted and controlled frames [Citation37], which are under very mild conditions equivalent to classical Hilbert frames. Nonetheless, they have applications for example in the implementation of wavelets on the sphere [Citation53, Citation54], and nowadays become important for the scaling of frames [Citation55, Citation56]. As a trivial example, look at , where is an orthonormal basis for . Then, is a tight frame with . Looking at the reweighted version , we loose tightness, since this frame has bounds A = 1 and B = 4. Note that there exists an invertible bounded operator that maps the single elements from into , i.e., they are equivalent sequences [Citation57].
Also note that, if it does not make sense to assume that , then cannot be a Hilbert space frame per se. This can only be true for the subsequence of X′ the Stevenson frame , where I is an isomorphism from to X, for example, choosing . In this case, the frame bounds are preserved, but the roles of primal and dual frames interchange.
This especially means that, if the frame bound ratio is important, distinguishing -Banach frames from Hilbert frames is necessary, especially if concrete examples for X and are used, where an identification is not possible, i.e., . As such, Definition 4.1 is, of course, equivalent to the standard frame definition for Hilbert spaces, but the frame bound ratio changes.
We like to remark that, by using the dual frame, one can also conclude that X itself is a Hilbert space.
4.3. Matrix representation
Let us also revisit the statements about the matrix representation of operators [Citation15, Citation17]. To this end, let be Stevenson Banach frame for X.
Let us now consider an operator and define Then, , which implies
(As in Section 3.6, we could consider different sequences, and the arguments would still work, but following the argument in the Introduction and for easy reading we will not.)
For an invertible operator O, we have
(For the analog result in the Hilbert frame case, see [Citation20, Citation47].) Equivalently,
Therefore, as is the orthogonal projection on the operator is boundedly invertible, as Furthermore,
If O is symmetric, then is symmetric. If O is nonnegative, so is . In particular, we have now have settled all statements in [Citation15, Citation17].
Acknowledgment
The authors like to thank Stephan Dahlke, Wolfgang Kreuzer, and Diana Stoeva for fruitful discussions.
Additional information
Funding
References
- Duffin, R. J., Schaeffer, A. C. (1952). A class of nonharmonic Fourier series. Trans. Amer. Math. Soc. 72(2):341–366.
- Aldroubi, A. (1995). Portraits of frames. Proc. Amer. Math. Soc. 123(6):1661–1668.
- Benedetto, J. J., Heller, W. (1990). Irregular sampling and the theory of frames, I. Note Di Matematica. 10(1):103–125.
- Christensen, O. (2016). An Introduction to Frames and Riesz Bases, 2nd ed. Cham: Birkhäuser/Springer.
- Pesenson, I., Mhaskar, H., Mayeli, A., Le Gia, Q. T., Zhou, D.-X. (2017). Frames and Other Bases in Abstract and Function Spaces, Applied and Numerical Harmonic Analysis. Basel: Birkhäuser.
- Pilipović, S., Stoeva, D. T. (2014). Fréchet frames, general definition and expansion. Anal. Appl. 12(02):195–208.
- Bölcskei, H., Hlawatsch, F., Feichtinger, H. G. (1998). Frame-theoretic analysis of oversampled filter banks. IEEE Trans. Signal Process. 46(12):3256–3268.
- Gazeau, J.-P. (2009). Coherent States in Quantum Physics. Weinheim: Wiley.
- Balazs, P., Holighaus, N., Necciari, T., Stoeva, D. T. Frame theory for signal processing in psychoacoustics, excursions in harmonic analysis. In: Radu Balan, John J. Benedetto, Wojciech Czaja, and Kasso Okoudjou, eds. Applied and Numerical Harmonic Analysis, Vol. 5, Basel: Birkhäuser, pp. 225–268.
- Sauter, S., Schwab, C. (2011). Boundary Element Methods, Springer Series in Computational Mathematics. Berlin–Heidelberg: Springer.
- Dahmen, W., Schneider, R. (1999). Composite wavelet basis for operator equations. Math. Comp. 68(228):1533–1567.
- Gaul, L., Kögler, M., Wagner, M. (2003). Boundary Element Methods for Engineers and Scientists. Berlin: Springer.
- Griebel, M. (1994). Multilevelmethoden als Interrelationship über Erzeugendensystemen. Stuttgart: B.G. Teubner.
- Grohs, P., Obermeier, A. (2016). Optimal adaptive ridgelet schemes for linear transport equations. Appl. Comput. Harmon. Anal. 41(3):768–814.
- Harbrecht, H., Schneider, R., Schwab, C. (2008). Multilevel frames for sparse tensor product spaces. Numer. Math. Math. 110(2):199–220.
- Oswald, P. (1994). Multilevel Finite Element Approximation. Stuttgart: B.G. Teubner.
- Stevenson, R. (2003). Adaptive solution of operator equations using wavelet frames. SIAM J. Numer. Anal. 41(3):1074–1100.
- Balazs, P. (2008). Matrix-representation of operators using frames. Sampl. Theory Signal Image Process. 7(1):39–54.
- Balazs, P., Gröchenig, K., (2017). A guide to localized frames and applications to Galerkin-like representations of operators, frames and other bases in abstract and function spaces, In: Pesenson, I., Mhaskar, H., Mayeli, A., Le Gia, Q. T., Zhou, D.-X., eds. Applied and Numerical Harmonic Analysis, Vol. 1, Basel: Birkhäuser, pp. 47–79.
- P., Balazs, G., Rieckh, (2011). Oversampling operators: Frame representation of operators. Analele Universitatii “Eftimie Murgu” 2:107–114.
- Dahlke, S., Fornasier, M., Raasch, T. (2007). Adaptive frame methods for elliptic operator equations. Adv. Comput. Math. 27(1):27–63.
- Casazza, P., Christensen, O., Stoeva, D. T. (2005). Frame expansions in separable Banach spaces. J. Math. Anal. Appl. 307(2):710–723.
- Christensen, O., Stoeva, D. T. (2003). p-frames in separable Banach spaces. Adv. Comput. Math. 18(2/4):117–126.
- Gröchenig, K. (1991). Describing functions: Atomic decompositions versus frames. Monatsh. Math. 112(1):1–41.
- Stoeva, D. T. (2009). Xd-frames in Banach spaces and their duals. Int. J. Pure Appl. Math. 52(1):1–14.
- Braess, D. (2001). Finite Elements. Theory, Fast Solvers, and Applications in Solid Mechanics, 2nd ed. Cambridge: Cambridge University Press.
- Brenner, S. C., Scott, L. R. (1994). The Mathematical Theory of Finite Element Methods. Texts in Applied Mathematics. New York, NY: Springer.
- Werner, D. (1995). Funktionalanalysis. Berlin: Springer.
- Heuser, H. (2006). Funktionalanalysis – Theorie und Anwendungen, 4th ed. Stuttgart: Teubner.
- Cordero, E., Feichtinger, H. G., Luef, F. (2008). Banach Gelfand Triples for Gabor Analysis, Pseudo-Differential Operators. Lecture Notes in Mathematics, Vol. 1949. Berlin: Springer, pp. 1–33.
- Antoine, J.-P. (1998). Quantum Mechanics beyond Hilbert Space. In: Irreversibility and Causality Semigroups and Rigged Hilbert Spaces. Berlin–Heidelberg: Springer, pp. 1–33.
- Antoine, J.-P., Balazs, P. (2011). Frames and semi-frames. J. Phys. A: Math. Theor. 44(20):205201.
- Antoine, J.-P., Balazs, P. (2012). Frames, semi-frames, and Hilbert scales. Numer. Funct. Anal. Optim. 33(7199):736–769.
- Balazs, P., Stoeva, D. T., Antoine, J.-P. (2011). Classification of general sequences by frame-related operators. Sampl. Theory Signal Image Process. 10(2):151–170.
- Kappei, J. (2012). Adaptive Wavelet Frame Methods for Nonlinear Elliptic Problems. Berlin: Logos-Verlag.
- Werner, M. (2009). Adaptive wavelet frame domain decomposition methods for elliptic operator equations. Ph.D. dissertation. Philipps-Universität Marburg, Germany.
- Balazs, P, Antoine, J.-P., Gryboś, A.. (2010). Weighted and controlled frames: Mutual relationship and first numerical properties. Int. J. Wavelets. Multiresolut. Inf. Process. 8(1):109–132.
- Stoeva, D. T., Balazs, P. (2012). Weighted frames and frame multipliers, Annual of the University of Architecture, Civil Engineering and Geodesy XLIII–LIV 2004–2009 (Fasc. II Mathematics Mechanics), pp. 33–42.
- Daubechies, I. (1992). Ten Lectures on Wavelets, CBMS-NSF Regional Conference Series in Applied Mathematics. Philadelphia, PA: SIAM.
- Dahmen, W. (1997). Wavelet and multiscale methods for operator equations. Anu. Numerica. 6:55–228.
- Bramble, J., Pasciak, J., Xu, J. (1990). Parallel multilevel preconditioners. Math. Comp. Comput. 55(191):1–22.
- Oswald, P. (1990). On function spaces related to finite element approximation theory. Z Anal. Anwend. 9(1):43–66.
- Gohberg, I., Goldberg, S., Kaashoek, M. A. (2003). Basic Classes of Linear Operators. Basel: Birkhäuser.
- Crone, L. (1971). A characterization of matrix operator on ℓ2. Math. Z Z. 123(4):315–317.
- Balazs, P. (2008). Hilbert-Schmidt operators and frames – Classification, best approximation by multipliers and algorithms. Int. J. Wavelets. Multiresolut. Inf. Process. 06(02):315–330.
- Bourouihiya, A. (2008). The tensor product of frames. Sampl. Theory Signal Image Process. 7(1):427–438.
- Balazs, P., Rieckh, G. (2016). Redundant Representation of Operators arXiv:1612.06130,
- Balazs, P., Feichtinger, H. G., Hampejs, M., Kracher, G. (2006). Double preconditioning for Gabor frames. IEEE Trans. Signal Process. 54(12):4597–4610.
- Janssen, A. J. E. M., Søndergaard, P. (2007). Iterative algorithms to approximate canonical gabor windows: Computational aspects. J. Fourier Anal. Appl. 13(2):211–241.
- Perraudin, N., Holighaus, N., Søndergaard, P., Balazs, P. (2018). Designing Gabor windows using convex optimization. Appl. Math. Comput. 330:266–287.
- Gröchenig, K. (2001). Foundations of Time-Frequency Analysis. Boston, MA: Birkhäuser.
- Gohberg, I., Goldberg, S., Kaashoek, M. A. (1990). Classes of Linear Operators. Operator Theory: Advances and Applications, Vol. I, Basel: Birkhäuser.
- Bogdanova, I., Vandergheynst, P., Antoine, J.-P., Jacques, L., Morvidone, M. (2005). Stereographic wavelet frames on the sphere. Appl. Comput. Harmon. Anal. 19(2):223–252.
- Jacques, L. (2004), Ondelettes, Repères et Couronne Solaire, Ph.D. thesis, Univ. Cath. Louvain, Louvain-la-Neuve.
- Casazza, P. G., Chen, X. (2017). Frame scalings: A condition number approach. Linear Algebra Appl. 523:152–168.
- Kutyniok, G., Okoudjou, K. A., Philipp, F., Tuley, E. K. (2013). Scalable frames. Linear Algebra Appl. 438(5):2225–2238.
- Casazza, P. (2000). The art of frame theory. Taiwanese J. Math. Math. 4(2):129–202.