{"id":"19d0cc4a-7483-4f0a-a1ae-ebd99fc70068","arxiv_id":"2512.10889","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For two dipolar emitters in high-NA microscopy, polarization-filtered image-inversion interferometry nearly saturates the quantum Fisher-information bound on estimating their separation, for arbitrary orientations.","lead":"This paper computes the best possible precision limits for measuring the distance between two glowing dipole molecules in a high-numerical-aperture microscope, now including polarization effects. It shows that the standard image-inversion interferometer only reaches the quantum limit for special orientations, and that splitting the light by radial and azimuthal polarization restores near-optimal resolution.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Zernike n=8 truncation with no convergence check may understate the QCRB, so the claimed near-saturation of the polarization-filtered III is not yet established quantitatively.","rationale":"The paper's intended contribution is quantitative: it claims the polarized III comes close to the QCRB for arbitrary dipole orientations (within a factor ~2 across orientation space, and 'nearly all' at fixed orientations). That claim depends on the QCRB benchmark being computed accurately. The reader's identified weakest assumption (equal brightness) is real but self-acknowledged in Sec. V and does not affect the internal validity of the calculations for the stated model. The Zernike truncation is an unacknowledged numerical approximation that directly controls the benchmark. Because QFI is monotone under the truncation map, the reported QCRB is only an upper bound; if convergence is slow—plausible given the (1-r^2)^(-1/4) prefactor in Eq. (5)—the gap between the polarized III and the true quantum limit could be larger than plotted. This is checkable without new physics and is the single most load-bearing issue for the quantitative claims. I therefore keep the conditional verdict: the paper's qualitative conclusions (finite QFI, polarization filtering helps) are well supported, but the numerical benchmarks should be verified before quantitative statements are relied upon.","tokens_in":16881,"tokens_out":10429,"duration_ms":113219,"concrete_test":"Recompute the QFI for representative cases (e.g., Θ=π/3, Φ=π/3 and the isotropic mixture, for l = 1, 10, 50, 100 nm) with Zernike truncations n=10, 12, and 16, and estimate the n→∞ limit by Richardson extrapolation or by an iterative eigensolver on the full 2049×2049 sampled operator. If the QCRB values plotted in Figs. 6, 10, and 11 shift by more than ~5% at any plotted separation, the quantitative saturation claims are not converged and require revision; if they are stable, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is the unverified truncation in the QFI computation. Section III computes K(l;Θ,Φ) by expanding |ψ±> in Zernike modes truncated at n=8, giving a 90×90 density matrix. This is a compression of the true single-photon Hilbert space; the QFI of the compressed state is a lower bound on the full QFI (quantum Fisher information is monotone under CPTP maps), so the plotted QCRB (1/K) is an upper bound on the true quantum limit. The central quantitative claim—that the radial/azimuthal filtered III approaches the QCRB (Figs. 6, 10, 11 and the \"within a factor of ~2\" statement)—therefore depends on the convergence of this truncation. The pupil field in Eq. (5) contains the edge-enhancement factor (1-r^2)^(-1/4), so Zernike coefficients decay only algebraically; n=8 is asserted with no convergence check. If the true QFI is larger than the truncated value, the reported saturation gaps widen and the \"nearly saturates\" claim is overstated. The equal-brightness assumption, by contrast, is explicitly acknowledged and scoped in Sec. V; it narrows the claim but does not threaten the internal numerics. A secondary indicator that the numerics need scrutiny is the typo in Eq. (23), where J^(1) appears twice instead of J^(1)+J^(2).","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies fundamental quantum and classical precision limits for estimating the separation of two closely spaced, mutually incoherent dipole emitters in high-NA optical microscopy. Using a vectorial Green's-tensor model of the back-focal-plane field, the authors construct the single-photon state, compute the quantum Fisher information (QFI) via the symmetric logarithmic derivative, and compare the corresponding quantum Cramér-Rao bound (QCRB) with classical Fisher-information bounds for direct imaging and image-inversion interferometry (III). They consider two source models: fixed, equal, known dipole orientations, and isotropic (freely tumbling) emitters. The central claim is that unpolarized III saturates the QCRB only for special orientations (dipoles parallel or perpendicular to the optical axis), while for arbitrary orientations and for isotropic emitters, resolving the collected light into radial and azimuthal polarization components before the III restores near-optimal super-resolving performance, with the combined r/φ-polarized III coming within a factor of about 2 of the QCRB across orientation space.","tokens_in":17201,"tokens_out":15234,"duration_ms":147315,"significance":"If the numerical results are reliable, the paper provides a valuable extension of quantum-inspired superresolution from scalar wave models to the fully vectorial, high-NA regime relevant to single-molecule fluorescence microscopy. The work is clearly framed in standard quantum estimation theory, the physical model is stated explicitly, and the comparison of several concrete measurement schemes (direct imaging, unpolarized III, r-polarized III, φ-polarized III, and their combination) is useful and well organized. The authors are also commendably explicit about the scope of their assumptions, particularly the equal-brightness and known-orientation limitations. The main weakness is that the quantitative conclusions—especially the 'within a factor of ~2' saturation claim—rest on a numerically truncated QFI computation whose convergence is not established.","major_comments":[{"comment":"The QCRB curves in Figs. 3, 6, 11 and the ratios in Figs. 8-10 are computed from a Zernike expansion truncated at n=8 (90x90 density matrix), with no convergence test. The pupil field in Eq. (5) contains (1-r^2)^(-1/4), so Zernike coefficients decay only algebraically; truncation error at n=8 may be significant, especially at the small separations used in the polar plots. Since the claim that the polarization-filtered III is 'within a factor of ~2' of the QCRB compares against these truncated QCRBs, this is load-bearing. Note that QFI monotonicity under CPTP maps does not straightforwardly bound the renormalized truncated state (postselection can increase per-photon QFI), so the error direction is unknown. Please add convergence checks (n=10,12,16) or compute the QFI directly from the 2x2 Gram matrix of |ψ±> over the full pupil, avoiding truncation entirely.","section":"Section III, Numerical Methods"}],"minor_comments":[{"comment":"The total III Fisher information is written as J^(1)_III + J^(1)_III; the second term should be J^(2)_III. As written, channel 1 is double-counted.","section":"Eq. (23)"},{"comment":"The text says normalization yields the states |ψ±> before the Zernike expansion, but it is not stated whether the states are renormalized after truncation at n=8. This ambiguity affects the interpretation of the subsequent QFI calculation and should be clarified.","section":"Section III"},{"comment":"The text refers to 'gold dashed' curves while the figure legends use 'yellow dashed'. Please unify the color terminology for the φ-polarized III curves.","section":"Section IV"},{"comment":"The statement that unequal known orientations 'should only make the resolution problem easier' is presented as an intuitive argument. Since the paper does not analyze unequal orientations, please qualify this claim or provide a supporting argument/reference.","section":"Section V"},{"comment":"The statement 'Code and data are available upon request' is less reproducible than a permanent repository. Consider depositing the numerical code used for the QFI and CRB computations.","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well written and honest about its limitations. The main technical risk is the numerical convergence of the QFI computation, which is load-bearing for the quantitative saturation claim. If the authors provide convergence checks or an alternative truncation-free calculation, the conclusions are likely to hold qualitatively and probably quantitatively. The equal-brightness assumption is explicitly scoped and is not a fundamental flaw. Note also that the paper's own Eq. (23) typo and the missing renormalization statement are easily fixable. I would support acceptance after a revision that addresses the numerical convergence issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — this paper does something genuinely new: it takes the quantum-inspired super-resolution program from scalar point sources to the vectorial dipole emitters that actually appear in single-molecule microscopy, and it quantifies both the QFI and the classical bounds for several measurement schemes under high-NA collection. The central practical message is clear: the unpolarized image inversion interferometer saturates the QCRB only for special dipole orientations (parallel or perpendicular to the optical axis), and once you resolve the collected light into radial and azimuthal polarization components, the III gets within about a factor of two of the quantum limit across orientation space. That claim is interesting and, if right, useful for people building passive super-resolution setups for known two-locus scenes.\n\nThe formalism is standard: equal-brightness mixture of two single-photon states built from the vectorial Green's tensor, SLD for the QFI, and the usual Fisher information for direct imaging and the interferometric variants. The isotropic-emitter extension is a reasonable limiting case. The authors are honest about the model's idealizations — equal brightness, known orientations or full rotational averaging, no FRET — and they explicitly scope the claims to those assumptions. The citation pattern is appropriate; the reliance on Refs. 70/71 for the polarization-filtering rationale is natural since those are prior experimental/theoretical papers on that method.\n\nThe soft spot is the numerics, and it is the same one your reader flagged. The QFI is computed by expanding the one-photon states in a Zernike basis truncated at n=8, giving a 90x90 density matrix, with no convergence check. Because the pupil field has the edge-enhancement factor (1-r^2)^(-1/4), the Zernike coefficients decay algebraically, and it is not obvious that n=8 is enough. This matters: the QFI of the truncated state is a lower bound on the true QFI, so the reported QCRB is an upper bound. If the true QFI is larger, the gaps between the classical CRBs and the quantum bound are wider than plotted, and the 'within a factor of ~2' statement could be optimistic. This is a quantitative weakness, not a fatal one; it affects the degree of near-saturation, not the existence of the effect. The typo in Eq. (23) — J^(1) twice instead of J^(1)+J^(2) — is minor and probably just a typo, but it should be fixed. Code is 'available upon request' rather than released; for a numerical paper of this type, a released script would make the convergence question easy to check.\n\nMy recommendation: this deserves a serious referee. The vectorial treatment of the two-source problem is a real step beyond the scalar approximation, and the polarization-filtered III is a concrete, experimentally feasible prescription. Have a referee demand the convergence check and a small table of QFI values at n=6,8,10; that would settle the main concern. I'd bring it to reading group, and I'd cite it if I worked in this area.","headline":"Extends quantum-inspired super-resolution to vectorial dipole emitters with a practical polarization-filtering fix; the numerics need a convergence check before the near-saturation claim is trusted.","tokens_in":17691,"tokens_out":1989,"would_cite":true,"duration_ms":19892,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Filtering light by polarization restores quantum-limited separation estimates for two dipole emitters at any orientation, whereas plain parity sorting only works for special orientations.","keywords":["quantum Fisher information","super-resolution microscopy","dipole emitters","image inversion interferometry","polarization filtering","Cramér–Rao bound","Rayleigh's curse","high numerical aperture"],"falsifier":"Measure the separation-estimation variance of an azimuthally polarized image inversion interferometer on a pair of fixed-orientation dipoles with known unequal brightness at a subdiffraction separation; if the observed variance departs from the predicted Cramér–Rao bound by more than photon-counting statistics, the equal-brightness assumption is doing the work.","tokens_in":16778,"feed_emoji":"🔬","tokens_out":10677,"duration_ms":89937,"temperature":0.7,"pith_summary":"This paper asks whether the recently discovered quantum advantage in resolving two closely spaced optical sources survives when the sources are real electric dipoles focused by a high-numerical-aperture lens, where the usual scalar-wave approximation breaks down. The authors compute quantum and classical Fisher information bounds for estimating the separation of two incoherent dipole emitters, in two regimes: fixed, equal orientation and completely isotropic orientation. They find that the quantum Cramér–Rao bound stays far below the direct-imaging bound at small separations, so super-resolution is possible for dipoles too. The practical message is that an image inversion interferometer, which sorts light by parity, only saturates the quantum bound for dipoles oriented parallel or perpendicular to the optical axis; for all other orientations, splitting the light into radially and azimuthally polarized components before parity sorting restores nearly all of the promised advantage. This matters because it identifies an implementable optical modification that extends quantum-inspired passive super-resolution to realistic single-molecule emitters.","feed_headline":"Polarization filter restores quantum super-resolution for dipoles","feed_subtitle":"For any dipole orientation, azimuthal-polarization sorting nearly hits the quantum limit for subdiffraction separation.","key_machinery":"Two tools carry the argument. The quantum Fisher information, via the symmetric logarithmic derivative computed numerically from the one-photon density matrix expanded in Zernike modes, sets the ultimate precision bound. The image inversion interferometer sorts the field by parity: for a definite-parity field, destructive interference nulls one output port as the separation vanishes, and the Fisher information is read out from the brightening of that dark port. The crucial identity is that, for any dipole orientation, the azimuthally polarized component of the collected field has definite parity under inversion whereas the radially polarized component mixes parities; a vortex half-wave plate","core_discovery":"For equal, mutually incoherent dipole emitters, the quantum Fisher information for their separation remains finite as the separation vanishes, despite fully vectorial emission, while direct imaging obeys Rayleigh's curse. An unmodified image inversion interferometer (parity sorting) saturates the quantum bound only for dipoles exactly parallel or perpendicular to the optical axis. For arbitrary orientations and for isotropic emitters, resolving the field into azimuthal and radial components before parity sorting restores near-quantum-limited performance; the azimuthal component alone carries most of the separation information except near the optical axis.","pith_inferences":["Editorial inference: an adaptive protocol that first estimates the dipole orientation could extend near-quantum-limited separation estimation to unknown, unequal orientations, a case the paper leaves open.","Editorial inference: because the azimuthal channel carries nearly all separation information, the radial channel could be dedicated to centroid or orientation estimation in a multi-parameter scheme without sacrificing separation precision.","Editorial inference: the parity argument that underpins the fix should transfer to other parity-sorting demultiplexing implementations, so polarization filtering may generalize beyond the specific image inversion interferometer."],"forward_implications":["For dipoles fixed parallel or perpendicular to the optical axis, the unmodified image inversion interferometer already saturates the quantum Cramér–Rao bound, so no polarization optics are needed in those limiting cases.","For arbitrary fixed orientations, azimuthally polarized image inversion interferometry recovers nearly all of the quantum advantage at small separations; discarding the radially polarized light costs little except for dipoles nearly parallel to the optical axis.","Measuring the radial and azimuthal components in separate interferometers closes most of the remaining gap at moderate subdiffraction separations and preserves saturation in the limiting orientations.","A pair of isotropic emitters behaves like an intermediate orientation case: the azimuthally polarized III measurement captures most of the super-resolving information, consistent with a recent experimental realization cited in the paper.","The result provides a passive, non-switching route to resolving two simultaneously emitting molecules at separations well below the diffraction limit, provided the scene is known to contain exactly two sources."],"fun_headline_variants":["Azimuthal filtering restores quantum-limited separation for dipoles","Parity sorting fails for tilted dipoles; azimuthal basis fixes it","For arbitrary dipole orientations, azimuthal splitting hits the quantum bound","Vectorial emission tamed: azimuthal filter restores super-resolution"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the two emitters are equally bright and mutually incoherent, so every collected photon is equally likely to have come from either source; if their brightnesses differ or their phases are partially locked, the Fisher information curves change and the near-saturation of the quantum bound may fail.","fun_headline_variants_meta":{"raw":{"variants":["Azimuthal filtering restores quantum-limited separation for dipoles","Parity sorting fails for tilted dipoles; azimuthal basis fixes it","For arbitrary dipole orientations, azimuthal splitting hits the quantum bound","Vectorial emission tamed: azimuthal filter restores super-resolution"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1155,"prompt_tokens":704,"completion_tokens":451,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":448,"completion_tokens_details":{"reasoning_tokens":377}},"tokens_in":448,"tokens_out":451,"duration_ms":5520,"temperature":1.0,"reasoning_tokens":377,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T16:58:44.442898+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the separation-estimation variance of an azimuthally polarized image inversion interferometer on a pair of fixed-orientation dipoles with known unequal brightness at a subdiffraction separation; if the observed variance departs from the predicted Cramér–Rao bound by more than photon-counting statistics, the equal-brightness assumption is doing the work.","supporting_citations":[],"review_version":1}