Pith. sign in

REVIEW 3 major objections 5 minor 51 references

On the sample complexity of Fourier compressed sensing: wavelets versus shearlets

T0 review · 3 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Shearlets' superior image sparsity does not translate into fewer Fourier samples for compressed sensing.

desk verdict The negative result on shearlet Fourier sample complexity is likely right but its theoretical core leans on a heuristic optimality claim; the numerics are the stronger evidence. read the letter →

arxiv 2607.20020 v1 pith:TCVNLSKE submitted 2026-07-22 math.FA cs.ITmath.ITmath.OC

Giovanni S. Alberti , Alessandro Felisi , Iş{i}l Güleken This is my paper · ORCID
classification math.FAcs.ITmath.ITmath.OC MSC 42C1594A2065T60
keywords compressedsensingshearletswaveletssamplecomplexitylocalcoherenceFouriermeasurementsframetheorycartoon-likeimages
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper asks whether the known advantage of shearlets in sparsely representing images—an N-term approximation error of order N^{-2} log^3 N versus N^{-1} for wavelets—also buys a proportional reduction in the number of Fourier measurements needed for compressed sensing. The answer it argues is no. The local coherence between shearlets and Fourier exponentials decays only as (1+||n||)^{-3/4}, slower than the wavelet decay (1+||n||)^{-1}; this slower decay puts a quadratic-in-sparsity term in the sample-complexity bound even under the most optimistic assumptions about frame redundancy. Numerically, phase diagrams on piecewise-smooth test images confirm that shearlets need only a minor (roughly factor-two) measurement advantage over wavelets, far below the sparsity gain. A sympathetic reader should care because it delimits when directional representation systems can actually accelerate Fourier-sampling modalities such as MRI.

What carries the argument

The two quantities that carry the argument are the local coherence between the sparsifying frame and the Fourier measurement basis, and the localization factor η_{s,D} that measures the redundancy of the frame. For cone-adapted shearlets, the paper proves a decay bound |ψhat_i(n)| ≤ C ||n||^{-3/4} for the Fourier transform on integer grids, and argues this rate is essentially optimal via a standard Littlewood-Paley heuristic; for wavelets the corresponding decay is ||n||^{-1}. The difference in decay exponents is what turns the ideal linear sample complexity m ~ s into m ~ s^2 for shearlets. The localization factor, which would require lower frame bounds for arbitrary finite shearlet subfram

What would settle it

Exhibit a cone-adapted shearlet frame whose elements satisfy |ψhat_i(n)| ≤ C (1+||n||)^{-1} on integer grids while retaining the frame and approximation properties; the paper's comparison then gives m_shear ~ s_shear, overturning the quadratic conclusion. Alternatively, produce a phase-diagram study on cartoon-like images in which shearlets require an order of magnitude fewer Fourier samples than wavelets at matched recovery quality.

Watch

Extended reading notes

Core claim

The central claim is that, for subsampled Fourier compressed sensing, shearlets do not deliver the sample-complexity reduction their approximation rates suggest. For cartoon-like images, shearlets have effective sparsity s_shear while wavelets have s_wave approximately s_shear^2; however the general frame recovery bound m ≳ η^2 ||ω||_2^2 s, combined with the essentially optimal local-coherence bound |ψhat_i(n)| ≲ ||n||^{-3/4} for cone-adapted shearlets, yields m_shear ≳ s_shear^2 even when the localization factor η is taken to be of order one. Comparing this with the wavelet bound m_wave ≳ s_wave yields only a logarithmic (resolution-dependent) improvement for shearlets; the paper's phase-di

Load-bearing premise

The negative conclusion depends on the claim that the decay |ψhat_i(n)| ≲ ||n||^{-3/4} is essentially optimal for cone-adapted shearlets; that optimality is argued from a Littlewood-Paley heuristic, and if some shearlet generator achieved the wavelet-like decay ||n||^{-1}, the sample-complexity comparison would flip.

Editorial extensions

If this is right

  • If the paper's bounds are tight, standard shearlet–Fourier compressed sensing will require m growing as s_shear^2, not as s_shear, so the theoretical regime where shearlets beat wavelets in sample count is at best logarithmic in resolution.
  • The practical message for imaging: switching from a wavelet to a shearlet sparsifying dictionary in an ℓ1-analysis recovery pipeline is unlikely to let one drop the sampling rate proportionally to the reduction in coefficients; observed gains are of order a factor of two.
  • The bottleneck is incoherence, not sparsity: the slower Fourier decay of shearlets offsets their better N-term approximation, and the unresolved localization factor only adds uncertainty on the pessimistic side.
  • Future approaches seeking to make directional systems pay off in Fourier sampling must either alter the sampling distribution, use multilevel sparsity-aware schemes, or find shearlet generators with faster Fourier decay while retaining frame bounds.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • One testable extension: if a cone-adapted shearlet family can be engineered so that its generators decay like wavelets (exponent -1) at integer frequencies, the paper's own comparison would predict m_shear ~ s_shear, restoring a proportional advantage; the remaining obstacle would then be the localization factor.
  • The paper's analysis suggests a broader design principle: for structured Fourier measurements, the sampling distribution should be matched to the coherence profile of the dictionary, and dictionaries with excellent approximation rates but slow Fourier decay will be throttled by the coherence term. Other directional systems may face a similar trade-off.
  • The roughly factor-two numerical advantage might partly reflect the redundancy and finite-scale implementation of digital shearlets rather than the asymptotic coherence bound; experiments at higher resolutions and with different wavelet families would separate finite-size effects from the asymptotic story.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper asks whether the superior sparsity of shearlets for cartoon-like images translates into fewer Fourier measurements in compressed sensing. The theoretical part proves an upper bound for the local coherence of cone-adapted shearlets with Fourier measurements (Proposition 2), yielding decay O(||n||_∞^{-3/4}), which is slower than the wavelet decay O(||n||^{-1}) recalled in Remark 3; it discusses the localization factor and shows that controlling it requires lower frame bounds that may be exponentially small; and, assuming the coherence bound is optimal and ignoring localization/balancing, it derives m_shear ≈ 2^{j0} versus m_wave ≈ j0 2^{j0}, i.e. only a logarithmic improvement. The numerical part compares stationary-wavelet and ShearLab shearlet sparsification on synthetic disk images; it confirms better shearlet N-term approximation but phase diagrams show roughly comparable sample complexity.

Significance. A fully established negative result would be valuable: it would show that asymptotic approximation-rate advantages do not automatically translate into sample-complexity gains under coherence-based compressed sensing with subsampled Fourier data, and it identifies concrete technical bottlenecks (localization factor, lower frame bounds). The paper is unusually transparent about its optimistic assumptions. However, the central theoretical scaling rests on the unproven optimality lower bound in Remark 4; without it, the comparison in Section 3.4 is only an upper-bound heuristic. The numerical experiments are suggestive but lack confidence intervals and a noise-level specification.

major comments (3)
  1. [§3.4 and Remark 4] The conclusion m_shear ≈ s_shear^2 rests on treating Proposition 2's upper bound μloc_k ≲ (1+||n_k||_∞)^{-3/4} as tight. Remark 4 is only a heuristic: the Littlewood-Paley lower bound on Σ_i |ψhat_i(n)|^2 does not imply a single shearlet at the dominant scale satisfies |ψhat_i0(n)| ≳ 2^{-3j/4}; several scales/shears could contribute to the sum while each individual Fourier transform decays faster than 2^{-3j/4}. If the true worst-case decay were O(||n||_∞^{-1}) (compatible with Proposition 2), then ||ω_shear||_2^2 would be O(j0) and the Section 3.4 estimate would give m_shear ≈ j0 2^{j0/2}, a substantial improvement over m_wave ≈ j0 2^{j0}. A matching lower bound for the specific shearlet frames covered by Proposition 2, or for the bandlimited Parseval frame of Theorem 4, is needed before the quadratic-in-sparsity claim is stated.
  2. [§4.2–4.3.2] The numerical experiments are not fully reproducible because the noise level σ is never specified. Section 4.2 defines Gaussian noise with variance σ² and sets ε_Ω = σ√|Ω|, but no value of σ is reported for the phase diagrams. Since the success/failure thresholds ε_a=25 and ε_r=0.15 are fixed, the phase transition locations depend on σ. In addition, T=6 trials per grid point makes each displayed probability a multiple of 1/6; no confidence intervals or error bars are given, so the 'roughly factor 2' advantage is not quantitatively supported. Reporting σ, the number of trials, and per-point standard errors (or at least a few bootstrap intervals) is necessary to substantiate the empirical claim.
  3. [§3.4] The comparison assumes η_s,D≈1 and that the balancing property/localization factor do not degrade the shearlet bound. These are explicitly acknowledged as optimistic, but the paper states the result as 'even under optimal theoretical conditions.' As Section 3.3 shows, no frame satisfying all the hypotheses of Theorem 1 (Parseval, good NLA rate, and η_s,D=O(1)) is identified. The statement should be phrased as conditional on these unverified assumptions, and the abstract's 'even under optimal theoretical conditions' should be qualified accordingly, unless such a frame is exhibited or a matching lower bound on η_s,D is provided.
minor comments (5)
  1. [Remark 7] The text says 'bounded from below by an expression of the form C j0' but Theorem 8 gives an exponential lower bound of the form C^{j0}; the notation should be corrected.
  2. [§4.3] Figure 2's y-axis label contains a typo ('mininum'). Figures 2 and 3 would benefit from error bars or confidence bands, especially because the phase diagrams use only T=6 trials per grid point.
  3. [§4.1.2] The sentence 'with ρ% different Fourier measurements' is unclear; ρ is only defined later as 100m/K². Please define ρ before using it.
  4. [§4.1.1] The shearlet system uses scales up to 2 while the stationary wavelet transform uses 4 levels. Although the total coefficient counts are matched, the finest scale/resolution differs; a sentence justifying that this does not bias the comparison would be useful.
  5. [§3.4] Equation (15) omits the log factors that appear in Theorem 1; this is acceptable for a scaling comparison but should be stated explicitly near the equation.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the central sample-complexity comparison follows from an in-paper coherence bound plus an external recovery theorem; only non-load-bearing self-citations are present.

full rationale

The derivation chain is not circular. The sample-complexity comparison in Section 3.4 combines Theorem 1, quoted from the external paper Krahmer–Needell–Ward [28], with Proposition 2, proven in the paper from the decay assumptions in Definition 11, and with the standard nonlinear-approximation scalings s_wave ~ 2^{j0} and s_shear ~ 2^{j0/2} cited from external wavelet/shearlet literature. No parameter is fitted to the target conclusion: the constant C in Proposition 2 is computed from the generator parameters, not calibrated. The 'essentially optimal' Remark 4 is the only passage supplying the lower bound that converts the coherence upper bound into the m_shear ~ s_shear^2 scaling, and it is a Littlewood-Paley heuristic rather than a complete theorem. That is a rigor/correctness concern, not circularity, because the lower-bound claim is not defined in terms of, or fitted to, the sample-complexity result. The authors' own works [3,4] are cited only for alternative infinite-dimensional compressed-sensing formulations and an unweighted noise model; the central finite-dimensional comparison, the shearlet coherence estimate, and the numerical phase diagrams do not rely on those self-citations. The numerical experiments are empirical comparisons using a sampling density derived from the coherence bounds, not a prediction obtained from fitted constants. Thus no circular reduction is exhibited; the score is set to 2 only to acknowledge the presence of non-load-bearing self-citations.

Assumptions & free parameters 1 free parameters · 7 assumptions · 0 invented entities

The paper introduces no invented entities. The theoretical claim depends on standard frame-based recovery theorems and external bounds (coherence of wavelets, lower frame bounds for wavelet subframes), all cited; the paper's own contribution is the shearlet-specific coherence bound and the obstacle analysis. Numerical thresholds are arbitrary but not fitted to force the conclusion.

free parameters (1)
  • Numerical phase-diagram calibration parameters (epsilon_a=25, epsilon_r=0.15, T=6) = epsilon_a=25; epsilon_r=0.15; T=6
    Hand-chosen experimental thresholds and number of Monte Carlo trials; the same qualitative conclusion holds for both success criteria, so these are not fitted to produce the negative result.
assumptions (7)
  • domain assumption Feasible cone-adapted shearlet generators (compactly supported [26] and bandlimited Parseval [24]) exist with the decay conditions of Definition 11.
    Proposition 2's coherence bound is conditional on these generators; existence is cited, not proved here (Theorems 3 and 4 in Appendix C).
  • domain assumption Nonlinear approximation rates: ||f-f_N||^2 = O(N^{-1}) for wavelets and O(N^{-2} log^3 N) for shearlets on cartoon-like images.
    Used in §1.2 and §3.4 to derive s_wave ~ s_shear^2; cited to [39, 23, 5, 31, 30].
  • domain assumption Local coherence of Daubechies wavelets with Fourier exponentials decays as (1+||n||_8)^{-1} [25, Theorem 2.3].
    Baseline for the wavelet sample complexity in §3.4; accepted from cited literature.
  • domain assumption Lower frame bounds for finite subframes of wavelets/shearlets can be exponentially small in the number of translations and finest scale [11, Proposition 2.3; 12, Theorems 7-8].
    Used in Remark 7 to argue that the localization factor cannot be bounded usefully.
  • standard math Theorem 1 (recovery of Parseval frames with weighted l1) holds as stated in [28, Corollary 2.9].
    The entire sample-complexity framework (6) is built on this theorem.
  • ad hoc to paper For the theoretical comparison in §3.4, the authors optimistically assume the balancing property and localization factor do not degrade the shearlet bound (localization factor ~1, bandwidth c2^{j0} for both).
    Explicitly stated as optimistic; if false, shearlet complexity would be worse, which only strengthens the negative conclusion. It is a simplifying assumption for the comparison.
  • standard math Parabolic molecules localization bound [22, Theorem 2.9] and L/2-admissibility of the canonical parametrization [22, Lemma 2.8].
    Used in Lemma 3 to bound the sum of square roots of cross-correlations between frame elements.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the sample complexity of Fourier compressed sensing: wavelets versus shearlets." pith.science (2026). https://pith.science/paper/TCVNLSKE

@misc{pith2026260720020,
  author       = {Pith},
  title        = {Pith review of: On the sample complexity of Fourier compressed sensing: wavelets versus shearlets},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TCVNLSKE}},
  note         = {Machine review of arXiv:2607.20020}
}
read the original abstract

This paper explores the measurement requirements for signal recovery in compressed sensing, comparing the performance of shearlet frames with traditional wavelet systems. Directional representation systems such as shearlets are known for their ability to sparsely represent images with anisotropic features, which allows for efficient nonlinear approximations. The central question we address is whether this difference in sparsity allows for a proportional reduction in the number of Fourier measurements needed for successful reconstruction. On the theoretical front, we study the obstacles encountered when trying to apply standard frame-based recovery results to shearlet systems. First, we show that the (optimal) local coherence between Fourier measurements and cone-adapted shearlets decays more slowly than the corresponding local coherence for wavelets. Second, managing the sparsifying system's redundancy relies on evaluating a localization factor, which requires lower frame bound estimates that can become exponentially small. Thus, even under optimal theoretical conditions, the number of samples required for shearlets scales quadratically with sparsity, which offers no substantial theoretical reduction over the standard wavelet benchmark. These theoretical limitations are assessed through a series of numerical experiments on a dataset of piecewise smooth images. While empirical observations confirm that shearlets can accurately represent these images using fewer coefficients than wavelets, phase diagrams indicate that this advantage in sparsity does not yield a proportional reduction in required Fourier samples. Ultimately, we conclude that despite the superior nonlinear approximation rates of shearlets, their practical sample complexity in compressed sensing scenarios with subsampled Fourier measurements remains comparable to that of traditional wavelets.

Figures

Figures reproduced from arXiv: 2607.20020 by the authors.

Figure 1
Figure 1. Examples of synthetic ground-truth images [PITH_FULL_IMAGE:figures/full_fig_p023_1.png] view at source ↗
Figure 2
Figure 2. Minimum values of N needed to achieve σN pg : 256q ď ε 2 with ε “ 25, for test functions g : 256 with support boundary length nL for wavelets and shearlets. available data vary. In compressed sensing, this point of view goes back to the seminal work of Donoho and Tanner [18, 17], who showed that for Gaussian sensing matrices and ℓ 1 -based recovery, one observes a phase transition phenomenon in the diagrams. More sp… view at source ↗
Figure 3
Figure 3. Phase diagrams for the empirical success probability of the analysis [PITH_FULL_IMAGE:figures/full_fig_p028_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 3 linked inside Pith

  1. [1]

    On optimal wavelet reconstructions from Fourier samples: Linearity and universality of the stable sampling rate

    Ben Adcock, Anders C. Hansen, and Clarice Poon. “On optimal wavelet reconstructions from Fourier samples: Linearity and universality of the stable sampling rate”. In: Applied and Computational Harmonic Analysis 36.3 (2014), pp. 387–415

    2014

  2. [2]

    Breaking the coherence barrier: a new theory for compressed sensing

    Ben Adcock et al. “Breaking the coherence barrier: a new theory for compressed sensing”. In: Forum Math. Sigma 5 (2017), Paper No. e4, 84

    2017

  3. [3]

    Infinite dimensional compressed sensing from anisotropic measurements and applications to inverse problems in PDE

    Giovanni S. Alberti and Matteo Santacesaria. “Infinite dimensional compressed sensing from anisotropic measurements and applications to inverse problems in PDE”. In: Applied and Computational Har- monic Analysis 50 (Jan. 2021), pp. 105–146

    2021

  4. [4]

    Alberti et al

    Giovanni S. Alberti et al. Compressed sensing for inverse problems II: applications to deconvolution, source recovery, and MRI. 2025. arXiv: 2501.01929 [math.FA]. url: https://arxiv.org/abs/2501. 01929

    arXiv 2025

  5. [5]

    New tight frames of curvelets and optimal representa- tions of objects with piecewise C 2 singularities

    Emmanuel J. Cand` es and David L. Donoho. “New tight frames of curvelets and optimal representa- tions of objects with piecewise C 2 singularities”. In: Communications on Pure and Applied Mathe- matics 57.2 (2004), pp. 219–266

    2004

  6. [6]

    Sparsity and incoherence in compressive sampling

    Emmanuel J. Cand` es and Justin Romberg. “Sparsity and incoherence in compressive sampling”. In: Inverse Problems 23.3 (2007), pp. 969–985

    2007

  7. [7]

    Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information

    Emmanuel J. Cand` es, Justin Romberg, and Terence Tao. “Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information”. In: IEEE Transactions on Information Theory 52.2 (2006), pp. 489–509

    2006

  8. [8]

    Near-Optimal Signal Recovery From Random Projections: Universal Encoding Strategies?

    Emmanuel J. Cand` es and Terence Tao. “Near-Optimal Signal Recovery From Random Projections: Universal Encoding Strategies?” In:IEEE Transactions on Information Theory52.12 (2006), pp. 5406– 5425

    2006

Show all 51 references
  1. [9]

    Compressed sensing with coherent and redundant dictionaries

    Emmanuel J. Cand` es et al. “Compressed sensing with coherent and redundant dictionaries”. In: Applied and Computational Harmonic Analysis31.1 (Jan. 2011), pp. 59–73

    2011

  2. [10]

    An Introduction to Frames and Riesz Bases

    Ole Christensen. An Introduction to Frames and Riesz Bases. Springer International Publishing, 2016. 36

    2016

  3. [11]

    Frames of exponentials: lower frame bounds for finite subfamilies and approximation of the inverse frame operator

    Ole Christensen and Alexander M. Lindner. “Frames of exponentials: lower frame bounds for finite subfamilies and approximation of the inverse frame operator”. In: Linear Algebra and its Applications 323.1–3 (Jan. 2001), pp. 117–130

    2001

  4. [12]

    Lower Bounds for Finite Wavelet and Gabor Systems

    Ole Christensen and Alexander M. Lindner. “Lower Bounds for Finite Wavelet and Gabor Systems”. In: Approximation Theory and Its Applications17.1 (2001), pp. 18–29

    2001

  5. [13]

    Localization of frames ii

    Elena Cordero and Karlheinz Gr¨ ochenig. “Localization of frames ii.” In: Applied and Computational Harmonic Analysis 17.1 (2004), pp. 29–47

    2004

  6. [14]

    Nonlinear Approximation

    Ronald A. DeVore. “Nonlinear Approximation”. In: Acta Numerica 7 (1998), pp. 51–150

    1998

  7. [15]

    Compressed sensing

    David L. Donoho. “Compressed sensing”. In: IEEE Transactions on Information Theory52.4 (2006), pp. 1289–1306

    2006

  8. [16]

    Wedgelets: Nearly Minimax Estimation of Edges

    David L. Donoho. “Wedgelets: Nearly Minimax Estimation of Edges”. In: The Annals of Statistics 27.3 (1999), pp. 859–897. (Visited on 06/21/2026)

    1999

  9. [17]

    Counting the Faces of Randomly-Projected Hypercubes and Orthants, with Applications

    David L. Donoho and Jared Tanner. “Counting the Faces of Randomly-Projected Hypercubes and Orthants, with Applications”. In: Discrete & Computational Geometry43.3 (2010), pp. 522–541

    2010

  10. [18]

    Observed Universality of Phase Transitions in High-Dimensional Geometry, with Implications for Modern Data Analysis and Signal Processing

    David L. Donoho and Jared Tanner. “Observed Universality of Phase Transitions in High-Dimensional Geometry, with Implications for Modern Data Analysis and Signal Processing”. In: Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences...

    1906

  11. [19]

    Intrinsic localization of frames

    Massimo Fornasier and Karlheinz Gr¨ ochenig. “Intrinsic localization of frames”. In: Constructive Ap- proximation 22.3 (2005), pp. 395–415

    2005

  12. [20]

    A mathematical introduction to compressive sensing

    Simon Foucart and Holger Rauhut. A mathematical introduction to compressive sensing. Applied and Numerical Harmonic Analysis. Birkh¨ auser/Springer, New York, 2013

    2013

  13. [21]

    α-Molecules

    Philipp Grohs and Gitta Kutyniok. “ α-Molecules”. In: Applied and Computational Harmonic Analysis 41.1 (July 2016), pp. 297–336

    2016

  14. [22]

    Parabolic Molecules

    Philipp Grohs and Gitta Kutyniok. “Parabolic Molecules”. In: Foundations of Computational Math- ematics 14.2 (Sept. 2013), pp. 299–337

    2013

  15. [23]

    Optimally Sparse Multidimensional Representation Using Shear- lets

    Kanghui Guo and Demetrio Labate. “Optimally Sparse Multidimensional Representation Using Shear- lets”. In: SIAM Journal on Mathematical Analysis39.1 (Jan. 2007), pp. 298–318

    2007

  16. [24]

    The Construction of Smooth Parseval Frames of Shearlets

    Kanghui Guo and Demetrio Labate. “The Construction of Smooth Parseval Frames of Shearlets”. In: Math. Model. Nat. Phenom.8.1 (2013), pp. 82–105

    2013

  17. [25]

    Analyzing the structure of multidimensional compressed sensing problems through coherence

    Alex Jones, Ben Adcock, and Anders Hansen. Analyzing the structure of multidimensional compressed sensing problems through coherence. 2016. arXiv: 1610.07497 [cs.IT]. url: https://arxiv.org/abs/ 1610.07497

    2016 arXiv

  18. [26]

    Construction of compactly supported shear- let frames

    Pisamai Kittipoom, Gitta Kutyniok, and Wang-Q Lim. “Construction of compactly supported shear- let frames”. In: Constr. Approx. 35.1 (Feb. 2012), pp. 21–72

    2012

  19. [27]

    Playing with Duality: An overview of recent primal- dual approaches for solving large-scale optimization problems

    Nikos Komodakis and Jean-Christophe Pesquet. “Playing with Duality: An overview of recent primal- dual approaches for solving large-scale optimization problems”. In: IEEE Signal Processing Magazine 32.6 (2015), pp. 31–54

    2015

  20. [28]

    Compressive Sensing with Redundant Dictio- naries and Structured Measurements

    Felix Krahmer, Deanna Needell, and Rachel Ward. “Compressive Sensing with Redundant Dictio- naries and Structured Measurements”. In: SIAM Journal on Mathematical Analysis 47.6 (2015), pp. 4606–4629

    2015

  21. [29]

    Stable and robust sampling strategies for compressive imaging

    Felix Krahmer and Rachel Ward. “Stable and robust sampling strategies for compressive imaging”. In: IEEE Transactions on Image Processing23.2 (2014), pp. 612–622. 37

    2014

  22. [30]

    Introduction to Shearlets

    Gitta Kutyniok and Demetrio Labate. “Introduction to Shearlets”. In: Shearlets: Multiscale Analysis for Multivariate Data. Ed. by Gitta Kutyniok and Demetrio Labate. Boston: Birkh¨ auser Boston, 2012, pp. 1–38

    2012

  23. [31]

    Compactly supported shearlets are optimally sparse

    Gitta Kutyniok and Wang-Q Lim. “Compactly supported shearlets are optimally sparse”. In: Journal of Approximation Theory163.11 (Nov. 2011), pp. 1564–1589

    2011

  24. [32]

    Optimal compressive imaging of Fourier data

    Gitta Kutyniok and Wang-Q Lim. “Optimal compressive imaging of Fourier data”. In: SIAM J. Imaging Sci. 11.1 (2018), pp. 507–546

    2018

  25. [33]

    ShearLab 3D: Faithful Digital Shearlet Trans- forms Based on Compactly Supported Shearlets

    Gitta Kutyniok, Wang-Q Lim, and Rafael Reisenhofer. “ShearLab 3D: Faithful Digital Shearlet Trans- forms Based on Compactly Supported Shearlets”. In: ACM Transactions on Mathematical Software 42.1 (2016), 5:1–5:42

    2016

  26. [34]

    Sparse multidimensional representation using shearlets

    Demetrio Labate et al. “Sparse multidimensional representation using shearlets”. In: Wavelets XI. Vol. 5914. International Society for Optics and Photonics. SPIE, 2005, 59140U

    2005

  27. [35]

    Sparse MRI: The application of compressed sensing for rapid MR imaging

    Michael Lustig, David Donoho, and John M. Pauly. “Sparse MRI: The application of compressed sensing for rapid MR imaging”. In: Magnetic Resonance in Medicine58.6 (2007), pp. 1182–1195

    2007

  28. [36]

    Generalized sampling reconstruction from Fourier measurements using compactly sup- ported shearlets

    Jackie Ma. “Generalized sampling reconstruction from Fourier measurements using compactly sup- ported shearlets”. In: Applied and Computational Harmonic Analysis42.2 (2017), pp. 294–318

    2017

  29. [37]

    Shearlet-based compressed sensing for fast 3D cardiac MR imaging using iterative reweighting

    Jackie Ma et al. “Shearlet-based compressed sensing for fast 3D cardiac MR imaging using iterative reweighting”. In: Physics in Medicine & Biology63.23 (2018), p. 235004

    2018

  30. [38]

    Sparse recovery from Fourier measurements using compactly supported shear- lets

    Jackie Ken Yip Ma. “Sparse recovery from Fourier measurements using compactly supported shear- lets”. PhD thesis. Technische Universit¨ at Berlin, 2016

    2016

  31. [39]

    A wavelet tour of signal processing

    St´ ephane Mallat. A wavelet tour of signal processing. Academic Press, 1998

    1998

  32. [40]

    Certain topics in telegraph transmission theory

    Harry Nyquist. “Certain topics in telegraph transmission theory”. In: Transactions of the American Institute of Electrical Engineers47.2 (1928), pp. 617–644

    1928

  33. [41]

    UNLocBoX: A MATLAB convex optimization toolbox for proximal-splitting methods

    Nathanael Perraudin et al. UNLocBoX: A MATLAB convex optimization toolbox for proximal-splitting methods. 2016. arXiv: 1402.0779 [cs.LG]. url: https://arxiv.org/abs/1402.0779

    2016 arXiv

  34. [42]

    Structure dependent sampling in compressed sensing: Theoretical guarantees for tight frames

    Clarice Poon. “Structure dependent sampling in compressed sensing: Theoretical guarantees for tight frames”. In: Applied and Computational Harmonic Analysis42.3 (May 2017), pp. 402–451

    2017

  35. [43]

    Compressed Sensing and Redundant Dictionaries

    Holger Rauhut, Karin Schnass, and Pierre Vandergheynst. “Compressed Sensing and Redundant Dictionaries”. In: IEEE Transactions on Information Theory54.5 (2008), pp. 2210–2219

    2008

  36. [44]

    On sparse reconstruction from Fourier and Gaussian mea- surements

    Mark Rudelson and Roman Vershynin. “On sparse reconstruction from Fourier and Gaussian mea- surements”. In: Communications on Pure and Applied Mathematics61.8 (2008), pp. 1025–1045

    2008

  37. [45]

    Communication in the presence of noise

    Claude E. Shannon. “Communication in the presence of noise”. In: Proceedings of the IRE37.1 (1949), pp. 10–21

    1949

  38. [46]

    ShearLab 3D software

    ShearLab. ShearLab 3D software. https://shearlab.math.lmu.de/software/. 2016

    2016

  39. [47]

    A parametrix construction for wave equations with C 1,1 coefficients

    Hart F. Smith. “A parametrix construction for wave equations with C 1,1 coefficients”. In: Annales de l’Institut Fourier48.3 (1998), pp. 797–835

    1998

  40. [48]

    MATLAB version 24.2 (R2024b)

    The MathWorks, Inc. MATLAB version 24.2 (R2024b). Natick, Massachusetts, United States, 2024. url: https://www.mathworks.com

    2024

  41. [49]

    Wavelet Toolbox version 24.2 (R2024b)

    The MathWorks, Inc. Wavelet Toolbox version 24.2 (R2024b). Natick, Massachusetts, United States,

  42. [50]

    Compressed sensing magnetic resonance imaging based on shearlet sparsity and nonlocal total variation

    Ali Pour Yazdanpanah and Emma E. Regentova. “Compressed sensing magnetic resonance imaging based on shearlet sparsity and nonlocal total variation”. In: Journal of Medical Imaging4.2 (2017), p. 026003. 38

    2017

  43. [2024]

    url: https://www.mathworks.com/products/wavelet.html

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.