Pith. sign in

REVIEW 2 major objections 4 minor 124 references

One lifted Green's function settles when optics should focus and when it should mix.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-02 06:02 UTC pith:Q4JUJF4U

load-bearing objection Solid, genuinely useful framework; the abstract's "uniquely maximizes" overstates a result that is conditional on an unproven flattening realizability condition. the 2 major comments →

arxiv 2607.13153 v1 pith:Q4JUJF4U submitted 2026-07-14 physics.optics

Optically Incoherent Photonic Mutual Information

Francis J. Chen , Alessio Amaolo , Pengning Chao , Sean Molesky , Zin Lin , Alejandro W. Rodriguez This is my paper
classification physics.optics
keywords mutual informationincoherent imagingphase retrievalHadamard productGreen's functionphotonic inverse designoptical coherenceMIMO channels
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper attempts to settle, from wave physics, the question of whether an optical front end should focus each scene point onto its own detector or mix light from many points and reconstruct computationally. Its answer is that the choice is fixed by the coherence and statistics of the source ensemble and by the detector law, not by imaging convention. The vehicle is a single linear channel: the Maxwell current-to-field Green's function, lifted to propagate second-order correlations (the mutual intensity), from which coherent communication, phase retrieval, and incoherent imaging follow as different projections. For spatially incoherent sources with isotropic covariance and matching source and detector counts, the paper proves point focusing uniquely maximizes mutual information at any signal-to-noise ratio among channels with fixed total gain, whenever structuring can flatten the channel's singular-value spectrum without changing that gain; the optimum has a flat spectrum, and a nonnegative matrix with a flat spectrum is necessarily a scaled permutation. For coherent sources read by square-law detectors, it shows the reverse — once detectors outnumber sources, interferometric mixing makes relative phases information-bearing and beats focusing — and it derives closed-form upper bounds on any incoherent imager's mutual information from the coherent singular values alone.

Core claim

The paper's central positive result is that, for spatially incoherent sources, the intensity channel is exactly the Hadamard (entrywise) square of the Green's function, F = G*⊙G, a linear map from source intensities to detector intensities. For isotropic Gaussian intensity fluctuations with equal source and detector counts, the mutual information depends on F only through its singular values. Since log(1 + x) is concave, the mutual information at any fixed squared Frobenius norm is bounded by the value of the flat-spectrum channel, and equality forces F to have all singular values equal. An entrywise nonnegative matrix with a flat positive spectrum must be a monomial matrix — a scaled permut

What carries the argument

The load-bearing object is the lifted coherence propagator G⊗G*, the dimension-lifted tensor product of the current-to-field Green's function with its conjugate. It sends the source coherence matrix — the outer product of source amplitudes — to the receiver coherence matrix, so that coherent communication, phase retrieval, and incoherent imaging are all projections of one linear map. Source incoherence selects the diagonal of the source coherence matrix; square-law detection selects the diagonal of the receiver coherence matrix; applying both selections reduces the lift exactly to the Hadamard square F = G*⊙G, the point-spread-function operator of incoherent imaging. Two devices carry the ar

Load-bearing premise

The theorem's unconditional statement assumes that, for the geometry in question, some passive structure can flatten the singular-value spectrum of the intensity channel F = G*⊙G to equal values without reducing its squared Frobenius norm; this is proven analytically only in the paraxial far-field limit and demonstrated numerically in one near-field 4×4 example.

What would settle it

Compute, for a strongly subwavelength source–detector geometry with four sources and four detectors, the largest ratio of smallest to largest singular value of F = G*⊙G at fixed squared Frobenius norm achievable by any passive structure; if this supremum is strictly below one, the flat-spectrum point-focusing optimum is physically unattainable in that geometry and the unconditional uniqueness claim fails.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • In isotropic incoherent imaging with equal source and detector counts, point focusing is not a convention but an information-theoretic optimum: no other passive front end with the same total gain achieves higher mutual information at any noise level.
  • For correlated source ensembles, the point-focusing guarantee is lost, and optimized front ends deliberately spread each source across several detectors, so imaging hardware should be matched to the correlation structure of the scene.
  • In phase retrieval under square-law detection, once detectors outnumber sources, interferometric mixing outperforms focusing and recovers relative-source-phase information that amplitude-only readout discards, approaching the (2M_S − 1)/(2M_S) pre-log ceiling.
  • Every incoherent imager's mutual information, for any source covariance and noise level, is bounded from above by a closed form depending only on the coherent singular values of the Green's function, so structure-agnostic electromagnetic limits transfer to intensity detection.
  • For non-Gaussian source ensembles with the same covariance, the Gaussian-derived mutual information is an upper bound, so the paper's benchmark remains a valid target even when real scene statistics are heavier-tailed or sparse.

Where Pith is reading between the lines

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

  • If the flattening condition fails in some near-field geometry, the true optimum will be a non-focusing compromise; measuring the gap between the best achievable mutual information and the scaled-permutation ideal would quantify the information cost of Maxwell constraints in that geometry.
  • Since the optimal front end is set by the source covariance, an adaptive or reconfigurable imager that estimates scene correlations could switch between focusing and mixing across measurements; the paper's static optima are the natural building blocks of such a policy.
  • The lifted-coherence construction extends naturally to partially coherent sources whose coherence matrix is structured but not diagonal; a testable prediction is that the focusing optimum degrades continuously as the source coherence length grows from zero to full coherence.
  • The focusing theorem bounds channel capacity under a Gaussian isotropic model, not reconstruction performance; for sparse or structured scenes a mixing front end might still yield better estimates even when it carries less of that model's mutual information.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The paper develops a unified end-to-end framework that connects Maxwell-equation propagation to Shannon mutual information by lifting the Green's function to propagate second-order field correlations (mutual intensity). It derives three channel laws from the same underlying operator: coherent-to-coherent communication (existing framework), phase retrieval under square-law detection, and incoherent intensity imaging. For phase retrieval, it derives a high-SNR mode-sorted mutual-information law (Eq. (12)) and numerically demonstrates a transition from point-focusing to interferometric mixing when detectors outnumber sources. For incoherent sources with equal source/detector counts and isotropic source covariance, it shows via a Jensen/majorization argument that mutual information is maximized by a flat singular spectrum of the Hadamard operator F = G* ⊙ G, and that a square nonnegative matrix with flat positive spectrum is necessarily a scaled permutation, i.e., generalized point focusing. This result is explicitly conditional on the realizability of spectrum flattening at fixed Frobenius norm. The paper also derives a closed-form upper bound (Eq. (26)) on incoherent mutual information in terms of the coherent singular values of the Green's function, and shows numerically that correlated source covariances move the optimum away from focusing.

Significance. If the main theorem holds, the paper makes a valuable conceptual contribution: it places focusing-versus-mixing under a common information-theoretic framework, shows that the optimal front end depends on source statistics and detector law rather than on lens-design convention, and provides a structure-agnostic bound (Eq. (26)) connecting incoherent imaging limits to prior coherent capacity bounds [37,55]. The analytic derivations in Secs. III-IV and the appendices are mostly careful: the mode-sorted high-SNR law, the Jensen/majorization bounds, and the monomial-matrix characterization of flat nonnegative spectra are internally consistent. The paper also provides reproducible code and data on GitHub, and the body text is commendably explicit about many limitations, including the conditional nature of the focusing optimality and the Gaussian-fluctuation model. The central weakness is that the unconditional phrasing of the abstract and Fig. 1 goes beyond what is proved: the realizability of the flat-spectrum condition is established only in an idealized far-field limit and one near-field numerical example.

major comments (2)
  1. [Abstract; Sec. IV (before Eq. (22)); Conclusions; Appendix C1] The headline claim that point-focusing uniquely maximizes incoherent mutual information at fixed Frobenius norm is stated unconditionally in the abstract and Fig. 1, but the theorem in Sec. IV is explicitly conditional: it holds only 'whenever structuring can flatten the singular-value spectrum of F while preserving its Frobenius norm.' This realizability hypothesis is not proven for general Maxwell-constrained operators. The only analytic support is the paraxial far-field/prolate-spheroidal construction in Appendix C1, which itself assumes an ideal unitary lens, diffraction-lattice source cells, and point-sampled detectors; the appendix notes that finite-extent detectors integrate positive sinc^2 tails, so exact lattice diagonality holds only for point samples. The only near-field evidence is one 4x4 optimization in Fig. 3. This is load-bearing because it is the paper's central claimed
  2. [Eq. (26) and surrounding text] The derivation of the universal upper bound is sound, but the presentation should be tightened regarding what 'structure-agnostic' means. Eq. (26) is a bound for a given structure's coherent singular values; it becomes a bound over all front ends only after substituting the coherent operator bounds of Refs. [37,55]. The paper states this correctly, but the abstract's phrase 'governed entirely by the coherent singular values' could be read as claiming that Eq. (26) itself is a fundamental limit independent of the structure. In fact, the right-hand side depends on σ_{G,1} and Σσ_{G,j}^2, which are structure-dependent; the structure-agnostic content comes from the cited external bounds. This is a presentation issue, but it affects the interpretation of a headline result.
minor comments (4)
  1. [Abstract] Please add the qualifier 'whenever structuring can flatten the singular-value spectrum at fixed Frobenius norm' to the uniqueness claim, and note that the result is derived under the Gaussian intensity-fluctuation model of Sec. IV. This would align the abstract with the body text and avoid overclaiming.
  2. [Fig. 1] The label 'Focusing maximizes MI' at the incoherent-imaging point should be annotated with an asterisk indicating the flattening condition and the equal source/detector count assumption. As drawn, the marker implies an unconditional theorem.
  3. [Sec. II.D] The notation M'_S ≈ rank_R(Q_x) and M'_R ≈ rank_R(H) uses 'rank_R' without definition. If it means rank over the real field, please define it explicitly, since the subsequent 'effective dimension' discussion is central to the paper's phase-space diagram.
  4. [Appendix C1] The sentence preceding Eq. (C13) says the readout is idealized as point sampling and that exact lattice diagonality holds only for point samples. This limitation should be stated in the main text where the far-field result is invoked, so readers do not over-generalize the exactness of the scaled-permutation channel.

Circularity Check

0 steps flagged

No significant circularity: focusing optimality is derived from Jensen and the monomial structure of flat-spectrum nonnegative matrices, not assumed; the flattening realizability caveat is a correctness limitation rather than a circular step.

full rationale

The central derivation chain is self-contained. For isotropic source covariance and equal source/detector counts, the paper reduces the incoherent mutual information to a function of the singular values of F = G*⊙G (Eq. (20)), applies Jensen's inequality with strict concavity (Eq. (22)), obtains equality only for a flat singular spectrum, and then uses nonnegativity of F to show that a flat positive spectrum forces F to be a scaled permutation (monomial) matrix. This is a mathematical implication from the channel model and standard matrix facts; it does not take 'point focusing' as an input. The paper is explicit that the optimality is conditional: 'point focusing is MI-optimal whenever structuring can flatten the singular value spectrum of F_t,RS without reduction of the sum of its squared singular values' (Sec. IV). The abstract's unconditional phrasing overstates this conditional result, but overstatement of a conditional theorem is a correctness risk, not circularity. The only analytic construction of the flattening premise is the paraxial far-field/prolate-spheroidal argument of Appendix C1, which itself notes that exact lattice diagonality 'holds for point samples alone' and that finite-extent pixels integrate positive sinc^2 tails. That gap is an unproven realizability condition, not an assumption of the target result. The numerical 4×4 near-field example in Fig. 3 is offered as an exhibit, not as the proof. Self-citations to Refs. [37,55] are used to convert Eq. (26) into structure-agnostic bounds on the coherent singular values; the paper states that 'the evaluation of these bounds for specific source–detector geometries is left to future work,' and these cited singular-value bounds are not used to prove the focusing optimum. Thus the self-citations are ancillary, not load-bearing, and no step in the derivation reduces to its own input by definition, fitting, or citation.

Axiom & Free-Parameter Ledger

5 free parameters · 6 axioms · 0 invented entities

The framework introduces no new physical entity; it reinterprets classical mutual-intensity propagation through selection matrices. The main hidden burden is the flattening realizability assumption for the focusing theorem, plus the Gaussian intensity model whose validity is only margin-controlled. Most free parameters are simulation-setup choices rather than fitted constants; none are fitted to the reported MI values, but their absence of sweeps limits the breadth of the numerical claims.

free parameters (5)
  • Simulation geometry and material constants = χ=10+0.01i (Fig. 2), χ=10+0.05i (Fig. 3); design regions 1.5λ×5λ and 6λ×3.5λ; source pitch 1λ; detector segment 3λ
    Chosen by hand; no sweeps over geometry or material are reported, so the claimed transition and the flattening optimum are demonstrated for these settings only.
  • Noise calibration = N = Tr[G_t,RS Q G†_t,RS]/(20 M_R)
    Holds average pre-detection SNR at 20 per run; a modeling choice that changes per structure and enters all MI values in Fig. 2.
  • Source covariance Q = Q = I_M (phase retrieval); Q_B = I and Q_ρ with ρ=0.95 (incoherent runs)
    Held fixed rather than optimized; the Gaussian lower bound and the correlated-mixing results depend on this choice.
  • Nonnegativity margin κ = κ = √(2 ln(M_S/δ)); δ not specified in numerics
    Eq. (21e) enforces the Gaussian intensity model's nonnegativity; without specifying δ, the exact margin used in Fig. 3 is not reproducible.
  • Correlation coefficient ρ = 0.95
    Chosen to produce condition number 39; no sweep over ρ, so the onset of mixing under correlations is not characterized.
axioms (6)
  • domain assumption Maxwell equations define the channel: e_R = G_t,RS j_i with total-field Green's function satisfying (∇×∇×−ε(r)ω²)G_t = ω² I δ.
    Section II B; all channel laws are projections of this operator, so any modeling error in the Green's function propagates everywhere.
  • domain assumption Temporal-average source incoherence: ⟨j_i j_i†⟩ = diag(B), and square-law detection keeps only the diagonal of the receiver mutual intensity.
    Sections II C and IV; this collapses the lifted operator to the Hadamard square F = G*⊙G.
  • domain assumption Gaussian intensity model B ∼ N(µ_B, Q_B) with real AWGN added after detection, plus κ-margin nonnegativity.
    Eqs. (19)-(21e); the log-det MI formula and all bounds are exact only under this model, which the paper acknowledges is an approximation for natural images.
  • ad hoc to paper For the focusing optimality theorem, structuring can flatten the singular spectrum of F at fixed Frobenius norm.
    Sec. IV 'Conditions for MI-optimality of Point Focusing'; proven only in a paraxial far-field limit and shown numerically in one geometry, not a theorem for general geometries.
  • domain assumption High-SNR regime with diagonal GQG† for the mode-sorted channel.
    Appendix B1; Eq. (12) is an asymptotic law with o(1) remainder, and equality requires the decoupled/mode-sorted front end.
  • standard math Known inequalities: Jensen, Hadamard determinant inequality, singular-value majorization for Hadamard products, subadditivity.
    Sec. IV and Appendix A2; used in proofs of Eqs. (22)-(26).

pith-pipeline@v1.3.0-alltime-deepseek · 42550 in / 14360 out tokens · 133865 ms · 2026-08-02T06:02:27.897550+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Optically Incoherent Photonic Mutual Information." pith.science (2026). https://pith.science/paper/Q4JUJF4U

@misc{pith2026260713153,
  author       = {Pith},
  title        = {Pith review of: Optically Incoherent Photonic Mutual Information},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q4JUJF4U}},
  note         = {Machine review of arXiv:2607.13153}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

While traditional evaluations of optical information transfer rely on disjointed abstractions to bridge electromagnetic propagation, coherence, and communication theory, we introduce an end-to-end framework that directly connects rigorous subwavelength wave physics to Shannon mutual information. By lifting the Maxwell current-to-field Green's function to propagate second-order field correlations (the mutual intensity), we establish a unified linear channel model that encapsulates coherent communication, phase retrieval, and incoherent imaging. Applying this framework, we demonstrate that the mutual-information-optimized photonic front end is dictated jointly by available spatial degrees of freedom, source statistics, and detection laws. For coherent sources measured by square-law detectors, we identify a structural transition: when detectors outnumber sources, topology-optimized front ends shift from point-focusing to interferometric mixing. This mixing leverages interference cross terms to make relative source phases information-bearing, yielding mutual information that surpasses the point-focusing amplitude-only baseline. Conversely, for spatially incoherent sources, the channel reduces to the Hadamard square of the Green's function. In this regime, under an isotropic source covariance, we prove that point-focusing uniquely maximizes the mutual information at fixed Frobenius norm. Under source correlations, the optimized front ends instead favor optical mixing. Finally, we derive closed-form upper bounds on achievable incoherent mutual information, governed entirely by the coherent singular values of the underlying electromagnetic operator. Potential applications include near-field microscopy, direct-detection optical datalinks, reference-free phase retrieval, fluorescence and thermal imaging, and structure-agnostic benchmarks for end-to-end-designed computational imagers.

Figures

Figures reproduced from arXiv: 2607.13153 by Alejandro W. Rodriguez, Alessio Amaolo, Francis J. Chen, Pengning Chao, Sean Molesky, Zin Lin.

Figure 1
Figure 1. Figure 1: FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Recovering the phase requires a front end that mixes the source fields, and the fields radiated by each source driven alone in [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

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

124 extracted references · 3 linked inside Pith

  1. [1]

    The main text gives the dimension-lifted propagation of Eq

    Derivation of Hadamard product channel from intensity-detection compression of mutual intensity Square-law detection records the diagonal entries of the receiver mutual intensity matrix, while linear prop- agation is bilinear in the field and carries the en- tire source mutual intensity to the receiver,⟨e Re† R⟩= Gt,RS ⟨jij† i ⟩G † t,RS: even a diagonal (...

  2. [2]

    Over the real field the same expression carries an overall factor of one half

    Modal alignment and the output-entropy subadditivity converse For a linear channelH=UΣV † with isotropic additive noisen∼ CN(0, NI) and a Gaussian input of covariance Q=E[xx †], the mutual information is I(x;y) = log det I+ 1 N HQH† = log det I+ 1 N V †QVΣ 2 , (A3) obtained by invariance of the determinant under unitary similarity, and det(I+AB) = det(I+B...

  3. [3]

    Asymptotic expression for mode-sorted mutual information Channel model.The transmitted signal is a zero- mean proper complex Gaussian current vectorx∼ CN(0,Q),x∈C MS . Through the linear map repre- senting the structured Green’s-function channelG t,RS ∈ CMR×MS (written asGfor notational shorthand), the signal produces a noiseless propagated fielde R =Gx w...

  4. [4]

    (11), has no closed form for generalG t,RS, Q, orN, and the Monte Carlo estimator of Appendix B 3 is costly and noisy to differentiate through

    Derivation of surrogate objective for photonic optimization of phase retrieval The exact mutual information of the phase-retrieval channel, Eq. (11), has no closed form for generalG t,RS, Q, orN, and the Monte Carlo estimator of Appendix B 3 is costly and noisy to differentiate through. For the gradient-based inverse design of the front end we there- fore...

  5. [5]

    We evaluate the mutual information using a nested, variance- reduced Monte Carlo (MC) estimator

    Numerical Evaluation for Phase Retrieval To enable gradient-based photonic inverse design, the mutual information of the phase-retrieval channel must be both numerically tractable and continuously differen- tiable with respect to the structural permittivityε. We evaluate the mutual information using a nested, variance- reduced Monte Carlo (MC) estimator. ...

  6. [6]

    MI-optimality of focusing in the radiative far-field We determine the passive photonic front end that maximizes the mutual information of a spatially inco- herent source radiating across a vacuum gap to an in- tensity detector, and show that, for a source occupy- ing the diffraction-limited lattice of the geometry, in- tensity detection forces the front e...

  7. [7]

    Correlated source covariance The correlated intensity covariance of Fig. 3 correlates the sources in pairs, Qρ =   1ρ0 0 ρ1 0 0 0 0 1ρ 0 0ρ1   , ρ= 0.95, with eigenvalues 1 +ρ,1 +ρ,1−ρ,1−ρ, all strictly pos- itive: sources one and two, and sources three and four, fluctuate together with correlationρ, the two pairs mu- tually uncorrelated, and the ei...

  8. [8]

    Gottesman and E

    Stephen R. Gottesman and E. E. Fenimore. New family of binary arrays for coded aperture imaging.Applied Optics, 28:4344–4352, 1989

    1989

  9. [9]

    Salman Asif, Ali Ayremlou, Aswin C Sankara- narayanan, Ashok Veeraraghavan, and Richard G Bara- niuk

    M. Salman Asif, Ali Ayremlou, Aswin C Sankara- narayanan, Ashok Veeraraghavan, and Richard G Bara- niuk. FlatCam: Thin, lensless cameras using coded aperture and computation.IEEE Transactions on Com- putational Imaging, 3(3):384–397, 2017

    2017

  10. [10]

    Dif- fuserCam: lensless single-exposure 3D imaging.Optica, 5(1):1–9, 2018

    Nick Antipa, Grace Kuo, Reinhard Heckel, Ben Milden- hall, Emrah Bostan, Ren Ng, and Laura Waller. Dif- fuserCam: lensless single-exposure 3D imaging.Optica, 5(1):1–9, 2018

    2018

  11. [11]

    End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging.ACM Transactions on Graphics (TOG), 37(4):1–13, 2018

    Vincent Sitzmann, Steven Diamond, Yifan Peng, Xiong Dun, Stephen Boyd, Wolfgang Heidrich, Felix Heide, and Gordon Wetzstein. End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging.ACM Transactions on Graphics (TOG), 37(4):1–13, 2018

    2018

  12. [12]

    End-to-end nanophotonic inverse design for imag- ing and polarimetry.Nanophotonics, 10(3):1177–1187, 2021

    Zin Lin, Charles Roques-Carmes, Rapha¨ el Pestourie, Marin Soljaˇ ci´ c, Arka Majumdar, and Steven G John- son. End-to-end nanophotonic inverse design for imag- ing and polarimetry.Nanophotonics, 10(3):1177–1187, 2021

    2021

  13. [13]

    End-to-end metasurface inverse de- sign for single-shot multi-channel imaging.Optics Ex- press, 30(16):28358–28370, 2022

    Zin Lin, Rapha¨ el Pestourie, Charles Roques-Carmes, Zhaoyi Li, Federico Capasso, Marin Soljaˇ ci´ c, and Steven G Johnson. End-to-end metasurface inverse de- sign for single-shot multi-channel imaging.Optics Ex- press, 30(16):28358–28370, 2022

    2022

  14. [14]

    Coded aperture imaging with uniformly redundant arrays.Ap- plied Optics, 17(3):337–347, 1978

    Edward E Fenimore and Thomas M Cannon. Coded aperture imaging with uniformly redundant arrays.Ap- plied Optics, 17(3):337–347, 1978

    1978

  15. [15]

    Wiley-Interscience, 2 edition, 2006

    Thomas M Cover and Joy A Thomas.Elements of In- formation Theory. Wiley-Interscience, 2 edition, 2006

    2006

  16. [16]

    A mathematical theory of communication.The Bell System Technical Journal, 27(3):379–423, 1948

    Claude Elwood Shannon. A mathematical theory of communication.The Bell System Technical Journal, 27(3):379–423, 1948. 26

    1948

  17. [17]

    Transmission of information.Bell System Technical Journal, 7(3):535–563, 1928

    Ralph VL Hartley. Transmission of information.Bell System Technical Journal, 7(3):535–563, 1928

    1928

  18. [18]

    On the assessment of optical images.Philosophical Transac- tions of the Royal Society of London

    Peter Berners Fellgett and Edward H Linfoot. On the assessment of optical images.Philosophical Transac- tions of the Royal Society of London. Series A, Mathe- matical and Physical Sciences, 247(931):369–407, 1955

    1955

  19. [19]

    Resolving power and information

    G Toraldo di Francia. Resolving power and information. Journal of the Optical Society of America, 45(7):497– 501, 1955

    1955

  20. [20]

    Light and information

    Denis Gabor. Light and information. InProgress in Optics, volume 1, pages 109–153. Elsevier, 1961

    1961

  21. [21]

    Degrees of freedom in multiple-antenna channels: A sig- nal space approach.IEEE Transactions on Information Theory, 51(2):523–536, 2005

    Ada SY Poon, Robert W Brodersen, and David NC Tse. Degrees of freedom in multiple-antenna channels: A sig- nal space approach.IEEE Transactions on Information Theory, 51(2):523–536, 2005

    2005

  22. [22]

    On the role of the number of degrees of freedom of the field in mimo channels.IEEE Transactions on Antennas and Propagation, 54(2):620– 628, 2006

    Marco Donald Migliore. On the role of the number of degrees of freedom of the field in mimo channels.IEEE Transactions on Antennas and Propagation, 54(2):620– 628, 2006

    2006

  23. [23]

    Space–bandwidth product of optical signals and systems.Journal of the Optical Society of America A, 13(3):470–473, 1996

    Adolf W Lohmann, Rainer G Dorsch, David Mendlovic, Zeev Zalevsky, and Carlos Ferreira. Space–bandwidth product of optical signals and systems.Journal of the Optical Society of America A, 13(3):470–473, 1996

    1996

  24. [24]

    Shadow area and degrees of freedom for free-space communication.IEEE Journal on Selected Areas in Information Theory, 6:325–337, 2025

    Mats Gustafsson. Shadow area and degrees of freedom for free-space communication.IEEE Journal on Selected Areas in Information Theory, 6:325–337, 2025

    2025

  25. [25]

    Degrees of freedom for radiating sys- tems.IEEE Transactions on Antennas and Propaga- tion, 73(2):1028–1038, 2025

    Mats Gustafsson. Degrees of freedom for radiating sys- tems.IEEE Transactions on Antennas and Propaga- tion, 73(2):1028–1038, 2025

    2025

  26. [26]

    Joseph W Goodman.Introduction to Fourier Optics. W. H. Freeman, 4 edition, 2017

    2017

  27. [27]

    Communication in the presence of noise.Proceedings of the IRE, 37(1):10–21, 1949

    Claude E Shannon. Communication in the presence of noise.Proceedings of the IRE, 37(1):10–21, 1949

    1949

  28. [28]

    On limits of wireless communications in a fading environment when using multiple antennas.Wireless Personal Communi- cations, 6(3):311–335, 1998

    Gerard J Foschini and Michael J Gans. On limits of wireless communications in a fading environment when using multiple antennas.Wireless Personal Communi- cations, 6(3):311–335, 1998

    1998

  29. [29]

    Capacity of multi-antenna gaussian chan- nels.European Transactions On Telecommunications, 10(6):585–595, 1999

    Emre Telatar. Capacity of multi-antenna gaussian chan- nels.European Transactions On Telecommunications, 10(6):585–595, 1999

    1999

  30. [30]

    Beitr¨ age zur theorie des mikroskops und der mikroskopischen wahrnehmung.Archiv f¨ ur mikroskopische Anatomie, 9(1):413–468, 1873

    Ernst Abbe. Beitr¨ age zur theorie des mikroskops und der mikroskopischen wahrnehmung.Archiv f¨ ur mikroskopische Anatomie, 9(1):413–468, 1873

  31. [31]

    Sahl, Adam S

    Yoav Shechtman, Steffen J. Sahl, Adam S. Backer, and W. E. Moerner. Optimal point spread function design for 3d imaging.Physical Review Letters, 113:133902, 2014

    2014

  32. [32]

    M. G. L. Gustafsson. Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy.Journal of Microscopy, 198(2):82–87, 2000

    2000

  33. [33]

    Investigations in optics, with special reference to the spectroscope.Philosophical Magazine, 8(49):261–274, 1879

    Lord Rayleigh. Investigations in optics, with special reference to the spectroscope.Philosophical Magazine, 8(49):261–274, 1879

  34. [34]

    Cambridge University Press, 7 edi- tion, 1999

    Max Born and Emil Wolf.Principles of Optics: Elec- tromagnetic Theory of Propagation, Interference and Diffraction of Light. Cambridge University Press, 7 edi- tion, 1999

    1999

  35. [35]

    Near-optimal signal recovery from random projections: Universal en- coding strategies?IEEE Transactions on Information Theory, 52(12):5406–5425, 2006

    Emmanuel J Candes and Terence Tao. Near-optimal signal recovery from random projections: Universal en- coding strategies?IEEE Transactions on Information Theory, 52(12):5406–5425, 2006

    2006

  36. [36]

    Structured com- pressed sensing: From theory to applications.IEEE Transactions on Signal Processing, 59(9):4053–4085, 2011

    Marco F Duarte and Yonina C Eldar. Structured com- pressed sensing: From theory to applications.IEEE Transactions on Signal Processing, 59(9):4053–4085, 2011

    2011

  37. [37]

    Designing lensless imag- ing systems to maximize information capture.Optica, 13(2):227–235, 2026

    Leyla A Kabuli, Henry Pinkard, Eric Markley, Clara S Hung, and Laura Waller. Designing lensless imag- ing systems to maximize information capture.Optica, 13(2):227–235, 2026

    2026

  38. [38]

    Compressed sensing.IEEE Transac- tions on Information Theory, 52(4):1289–1306, 2006

    David L Donoho. Compressed sensing.IEEE Transac- tions on Information Theory, 52(4):1289–1306, 2006

    2006

  39. [39]

    Phase retrieval with application to optical imag- ing: a contemporary overview.IEEE Signal Processing Magazine, 32(3):87–109, 2015

    Yoav Shechtman, Yonina C Eldar, Oren Cohen, Henry Nicholas Chapman, Jianwei Miao, and Mordechai Segev. Phase retrieval with application to optical imag- ing: a contemporary overview.IEEE Signal Processing Magazine, 32(3):87–109, 2015

    2015

  40. [40]

    Saving phase: Injectivity and stability for phase retrieval.Applied and Computational Harmonic Analysis, 37(1):106–125, 2014

    Afonso S Bandeira, Jameson Cahill, Dustin G Mixon, and Aaron A Nelson. Saving phase: Injectivity and stability for phase retrieval.Applied and Computational Harmonic Analysis, 37(1):106–125, 2014

    2014

  41. [41]

    Cambridge University Press, 2 edition, 2012

    Lukas Novotny and Bert Hecht.Principles of Nano- Optics. Cambridge University Press, 2 edition, 2012

    2012

  42. [42]

    Physical limits in electromagnetism.Nature Reviews Physics, 4(8):543–559, 2022

    Pengning Chao, Benjamin Strekha, Rodrick Ku- ate Defo, Sean Molesky, and Alejandro W Rodriguez. Physical limits in electromagnetism.Nature Reviews Physics, 4(8):543–559, 2022

    2022

  43. [43]

    Physical bounds of antennas

    Mats Gustafsson, Doruk Tayli, and Marius Cismasu. Physical bounds of antennas. In Zhi Ning Chen, Duixian Liu, Hisamatsu Nakano, Xianming Qing, and Thomas Zwick, editors,Handbook of Antenna Technolo- gies, pages 197–233. Springer, Singapore, 2016

    2016

  44. [44]

    Maximum shan- non capacity of photonic structures.npj Nanophotonics, 3(1):14, 2026

    Alessio Amaolo, Pengning Chao, Benjamin Strekha, Stefan Clarke, Jewel Mohajan, Francis J Chen, Sean Molesky, and Alejandro W Rodriguez. Maximum shan- non capacity of photonic structures.npj Nanophotonics, 3(1):14, 2026

    2026

  45. [45]

    Efficient inverse design of large- area metasurfaces for incoherent light.ACS Photonics, 10(4):854–860, 2023

    Rapha¨ el Pestourie, Wenjie Yao, Boubacar Kant´ e, and Steven G Johnson. Efficient inverse design of large- area metasurfaces for incoherent light.ACS Photonics, 10(4):854–860, 2023

    2023

  46. [46]

    T- operator limits on optical communication: metaoptics, computation, and input-output transformations.Phys- ical Review Research, 4(1):013020, 2022

    Sean Molesky, Pengning Chao, Jewel Mohajan, Wesley Reinhart, Heng Chi, and Alejandro W Rodriguez. T- operator limits on optical communication: metaoptics, computation, and input-output transformations.Phys- ical Review Research, 4(1):013020, 2022

    2022

  47. [47]

    Bounds on the coupling strengths of communication channels and their information capacities.IEEE Trans- actions on Antennas and Propagation, 73(6):3959–3974, 2025

    Zeyu Kuang, David AB Miller, and Owen D Miller. Bounds on the coupling strengths of communication channels and their information capacities.IEEE Trans- actions on Antennas and Propagation, 73(6):3959–3974, 2025

    2025

  48. [48]

    Information-theoretic design for high-dimensional computational imaging

    Eric Markley, Leyla Kabuli, Tiffany Chien, Henry Pinkard, and Laura Waller. Information-theoretic design for high-dimensional computational imaging. InComputational Optical Sensing and Imaging, page CTh4B.4. Optica Publishing Group, 2024

    2024

  49. [49]

    The singular values of a hadamard product: A basic in- equality.Linear and Multilinear Algebra, 21(4):345–365, 1987

    T Ando, Roger A Horn, and Charles R Johnson. The singular values of a hadamard product: A basic in- equality.Linear and Multilinear Algebra, 21(4):345–365, 1987

    1987

  50. [50]

    Cambridge University Press, 1991

    Roger A Horn and Charles R Johnson.Topics in Matrix Analysis. Cambridge University Press, 1991

    1991

  51. [51]

    Nir Shlezinger, Ron Dabora, and Yonina C. Eldar. Mea- surement matrix design for phase retrieval based on mu- tual information.IEEE Transactions on Signal Process- ing, 66(2):324–339, 2018

    2018

  52. [52]

    Cambridge University Press, 2017

    Massimo Franceschetti.Wave Theory of Information. Cambridge University Press, 2017. 27

    2017

  53. [53]

    Waves, modes, communications, and optics: a tutorial.Advances in Optics and Photonics, 11(3):679–825, 2019

    David AB Miller. Waves, modes, communications, and optics: a tutorial.Advances in Optics and Photonics, 11(3):679–825, 2019

    2019

  54. [54]

    Inverse design in nanophotonics.Nature Photonics, 12(11):659–670, 2018

    Sean Molesky, Zin Lin, Alexander Y Piggott, Weil- iang Jin, Jelena Vuckovi´ c, and Alejandro W Rodriguez. Inverse design in nanophotonics.Nature Photonics, 12(11):659–670, 2018

    2018

  55. [55]

    On signal reconstruction without phase.Applied and Computa- tional Harmonic Analysis, 20(3):345–356, 2006

    Radu Balan, Pete Casazza, and Dan Edidin. On signal reconstruction without phase.Applied and Computa- tional Harmonic Analysis, 20(3):345–356, 2006

    2006

  56. [56]

    Invertibility and robust- ness of phaseless reconstruction.Applied and Computa- tional Harmonic Analysis, 38(3):469–488, 2015

    Radu Balan and Yang Wang. Invertibility and robust- ness of phaseless reconstruction.Applied and Computa- tional Harmonic Analysis, 38(3):469–488, 2015

    2015

  57. [57]

    An algebraic characterization of injectivity in phase retrieval.Applied and Computational Harmonic Analysis, 38(2):346–356, 2015

    Aldo Conca, Dan Edidin, Milena Hering, and Cynthia Vinzant. An algebraic characterization of injectivity in phase retrieval.Applied and Computational Harmonic Analysis, 38(2):346–356, 2015

    2015

  58. [58]

    A small frame and a certificate of its injectivity

    Cynthia Vinzant. A small frame and a certificate of its injectivity. In2015 International Conference on Sam- pling Theory and Applications (SampTA), pages 197–

  59. [59]

    High-snr capac- ity of wireless communication channels in the noncoher- ent setting: A primer.AEU-International Journal of Electronics and Communications, 65(8):707–712, 2011

    Giuseppe Durisi and Helmut B¨ olcskei. High-snr capac- ity of wireless communication channels in the noncoher- ent setting: A primer.AEU-International Journal of Electronics and Communications, 65(8):707–712, 2011

    2011

  60. [60]

    A comparison of the informational capacities of amplitude- and phase-modulation commu- nication systems.Proceedings of the IRE, 41(6):748– 759, 1953

    Nelson M Blachman. A comparison of the informational capacities of amplitude- and phase-modulation commu- nication systems.Proceedings of the IRE, 41(6):748– 759, 1953

    1953

  61. [61]

    Extra dof of near-field holographic mimo communications leveraging evanescent waves

    Ran Ji, Shuo Chen, Chongwen Huang, Jun Yang, Wei EI Sha, Zhaoyang Zhang, Chau Yuen, and M´ erouane Debbah. Extra dof of near-field holographic mimo communications leveraging evanescent waves. IEEE Wireless Communications Letters, 12(4):580–584, 2023

    2023

  62. [62]

    Indexed singular value bounds on scattering operators: How many chan- nels can a photonic device support?arXiv preprint arXiv:2510.01128, 2026

    Paul Virally, Pengning Chao, Alessio Amaolo, Alejan- dro W Rodriguez, and Sean Molesky. Indexed singular value bounds on scattering operators: How many chan- nels can a photonic device support?arXiv preprint arXiv:2510.01128, 2026

    Pith/arXiv arXiv 2026

  63. [63]

    Near-field mimo communications for 6g: Fundamentals, challenges, potentials, and future direc- tions.IEEE Communications Magazine, 61(1):40–46, 2023

    Mingyao Cui, Zidong Wu, Yu Lu, Xiuhong Wei, and Linglong Dai. Near-field mimo communications for 6g: Fundamentals, challenges, potentials, and future direc- tions.IEEE Communications Magazine, 61(1):40–46, 2023

    2023

  64. [64]

    Haiyang Zhang, Nir Shlezinger, Francesco Guidi, Da- vide Dardari, and Yonina C. Eldar. 6g wireless com- munications: From far-field beam steering to near- field beam focusing.IEEE Communications Magazine, 61(4):72–77, 2023

    2023

  65. [65]

    Near-field optics: mi- croscopy, spectroscopy, and surface modification beyond the diffraction limit.Science, 257(5067):189–195, 1992

    Eric Betzig and Jay K Trautman. Near-field optics: mi- croscopy, spectroscopy, and surface modification beyond the diffraction limit.Science, 257(5067):189–195, 1992

    1992

  66. [66]

    Evanescent coupling device design for waveguide- integrated group iv photodetectors.Journal of Light- wave Technology, 28(23):3387–3394, 2010

    Donghwan Ahn, Lionel C Kimerling, and Jurgen Michel. Evanescent coupling device design for waveguide- integrated group iv photodetectors.Journal of Light- wave Technology, 28(23):3387–3394, 2010

    2010

  67. [67]

    On the capacity of mimo optical wire- less channels.IEEE Transactions on Information The- ory, 66(9):5660–5682, 2020

    Longguang Li, Stefan M Moser, Ligong Wang, and Mich` ele Wigger. On the capacity of mimo optical wire- less channels.IEEE Transactions on Information The- ory, 66(9):5660–5682, 2020

    2020

  68. [68]

    A pix- elated mimo wireless optical communication system

    Steve Hranilovic and Frank R Kschischang. A pix- elated mimo wireless optical communication system. IEEE Journal of Selected Topics in Quantum Electron- ics, 12(4):859–874, 2006

    2006

  69. [69]

    Communications-inspired projection design with appli- cation to compressive sensing.SIAM Journal on Imag- ing Sciences, 5(4):1185–1212, 2012

    William R Carson, Minhua Chen, Miguel RD Ro- drigues, Robert Calderbank, and Lawrence Carin. Communications-inspired projection design with appli- cation to compressive sensing.SIAM Journal on Imag- ing Sciences, 5(4):1185–1212, 2012

    2012

  70. [70]

    End-to-end meta-imagers: Information- theoretic objectives and generalized focusing optima

    Lukas Kienesberger, Zeyu Kuang, Yaxi Liu, and Owen D Miller. End-to-end meta-imagers: Information- theoretic objectives and generalized focusing optima. arXiv preprint arXiv:2606.16724, 2026

    Pith/arXiv arXiv 2026

  71. [71]

    Cambridge University Press, 2005

    David Tse and Pramod Viswanath.Fundamentals of Wireless Communication. Cambridge University Press, 2005

    2005

  72. [72]

    Joint tx-rx beamforming design for multicar- rier mimo channels: A unified framework for convex optimization.IEEE Transactions on Signal Processing, 51(9):2381–2401, 2003

    Daniel P´ erez Palomar, John M Cioffi, and Miguel´Angel Lagunas. Joint tx-rx beamforming design for multicar- rier mimo channels: A unified framework for convex optimization.IEEE Transactions on Signal Processing, 51(9):2381–2401, 2003

    2003

  73. [73]

    John Wiley & Sons, 2 edition, 2015

    Joseph W Goodman.Statistical Optics. John Wiley & Sons, 2 edition, 2015

    2015

  74. [74]

    Physical bounds on antennas of arbi- trary shape

    Mats Gustafsson. Physical bounds on antennas of arbi- trary shape. In2011 Loughborough Antennas & Propa- gation Conference, pages 1–5. IEEE, 2011

    2011

  75. [75]

    Fundamental bounds on mimo antennas.IEEE Antennas and Wire- less Propagation Letters, 17(1):21–24, 2018

    Casimir Ehrenborg and Mats Gustafsson. Fundamental bounds on mimo antennas.IEEE Antennas and Wire- less Propagation Letters, 17(1):21–24, 2018

    2018

  76. [76]

    Physi- cal bounds and radiation modes for mimo anten- nas.IEEE Transactions on Antennas and Propagation, 68(6):4302–4311, 2020

    Casimir Ehrenborg and Mats Gustafsson. Physi- cal bounds and radiation modes for mimo anten- nas.IEEE Transactions on Antennas and Propagation, 68(6):4302–4311, 2020

    2020

  77. [77]

    Capacity bounds and degrees of freedom for mimo antennas constrained by q-factor.IEEE Transac- tions on Antennas and Propagation, 69(9):5388–5400, 2021

    Casimir Ehrenborg, Mats Gustafsson, and Miloslav ˇCapek. Capacity bounds and degrees of freedom for mimo antennas constrained by q-factor.IEEE Transac- tions on Antennas and Propagation, 69(9):5388–5400, 2021

    2021

  78. [78]

    Toward a circuit theory of communication.IEEE Transactions on Cir- cuits and Systems I: Regular Papers, 57(7):1663–1683, 2010

    Michel T Ivrlaˇ c and Josef A Nossek. Toward a circuit theory of communication.IEEE Transactions on Cir- cuits and Systems I: Regular Papers, 57(7):1663–1683, 2010

    2010

  79. [79]

    High-snr power offset in multiantenna communica- tion.IEEE Transactions on Information Theory, 51(12):4134–4151, 2005

    Angel Lozano, Antonia M Tulino, and Sergio Verd´ u. High-snr power offset in multiantenna communica- tion.IEEE Transactions on Information Theory, 51(12):4134–4151, 2005

    2005

  80. [80]

    Natural im- age statistics and neural representation.Annual Review of Neuroscience, 24(1):1193–1216, 2001

    Eero P Simoncelli and Bruno A Olshausen. Natural im- age statistics and neural representation.Annual Review of Neuroscience, 24(1):1193–1216, 2001

    2001

Showing first 80 references.