Nulling at short wavelengths: theoretical performance constraints and a demonstration of faint companion detection inside the diffraction limit with a rotating-baseline interferometer

Bertrand Mennesson; Eugene Serabyn; Jonas K\"uhn; Kurt Liewer; Stefan Martin

REVIEW 2 major objections 4 minor 1 cited by

Nulling at short wavelengths: theoretical performance constraints and a demonstration of faint companion detection inside the diffraction limit with a rotating-baseline interferometer

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A rotating fiber nuller detected a 1.08% companion at 30 mas—inside the 88 mas diffraction limit—by fitting the second harmonic of the null-depth rotation curve.

desk verdict First real on-sky demonstration of companion detection by rotating-baseline nulling inside the diffraction limit, with a clean one-parameter fit and a credible but not fully characterized null-floor calibration. read the letter →

arxiv 1908.05977 v1 pith:LTFGEGFN submitted 2019-08-16 astro-ph.IM

Eugene Serabyn , Bertrand Mennesson , Stefan Martin , Kurt Liewer , Jonas Kühn This is my paper · ORCID

classification astro-ph.IM

keywords nullinginterferometryrotatingbaselineinnerworkinganglecompaniondetectionsingle-modefiberself-calibrationstellardiameternear-infrared

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

The paper establishes that high-accuracy nulling interferometry can be pushed into the near-infrared and into the diffraction-limited core of a bright star's image, provided phase noise is handled statistically rather than by brute-force hardware stabilization. Its demonstration is companion detection by baseline rotation: on the spectroscopic binary eta Peg, the Ks-band null depth measured while the baseline rotates traces a $\cos^2$ curve whose amplitude yields a secondary-to-primary flux ratio of $(1.08 \pm 0.06) \times 10^{-2}$, and whose non-zero minimum yields a primary-diameter leakage of $(4.8 \pm 1.6) \times 10^{-4}$. The companion separation was 30 mas, about one third of the 88 mas diffraction-limited beam, well inside both the telescope's diffraction core and typical coronagraphic inner working angles. If these results hold, rotating-baseline nullers on large telescopes would open a small-angle, high-contrast regime—inner hot dust, brown-dwarf companions, young planets, and eventually hot Jupiters—that coronagraphs cannot reach.

What carries the argument

The load-bearing mechanism is the rotating-baseline cross-aperture nuller: two sub-apertures within one telescope pupil are combined on a single-mode fiber with an achromatic $\pi$ phase shift, so the anti-symmetric stellar field cannot propagate in the fiber while off-axis light passes through the sinusoidal transmission fringe. Rotating an image of the pupil rotates the baseline on the sky, and for a companion at small angle $\theta_p$ the transmitted intensity goes as $\cos^2\alpha_p$, placing the companion signal at the second harmonic of the rotation angle, with amplitude proportional to $(k b \theta_p)^2 I_p$. The other load-bearing element is the nulling self-calibration (NSC) algorithm, which fits the probability density of the fluctuating null-depth stream and recovers the astrophysical null even when phase fluctuations are large; this relaxes the phase-stabilization requirement from about $\lambda/300$ to about $\lambda/4$.

What would settle it

Re-observe eta Peg or a comparable binary with an independently known companion flux ratio, interleaving at least two calibrator stars at matched airmass on multiple nights; if the recovered flux ratio shifts by more than the formal $0.06 \times 10^{-2}$ error, or the recovered primary diameter moves outside $3.23 \pm 0.07$ mas, the single measured null floor does not transfer between targets.

Watch

Extended reading notes

Core claim

The central discovery is that the null-depth rotation curve of a single-baseline nuller cleanly separates two astrophysical signals that both live inside the diffraction limit. For the binary eta Peg, with known separation 30 mas and position angle 170°, the calibrated null depths follow the expected $\cos^2$ pattern of a companion well inside the first constructive fringe; with the binary geometry fixed, the best fit gives a Ks-band secondary-to-primary flux ratio of $(1.08 \pm 0.06) \times 10^{-2}$, corresponding to a Ks magnitude difference of $4.92 \pm 0.06$ mag. The two minima of the curve, where the null fringe extinguishes both stars, do not reach zero: their average, $(4.8 \pm 1.6) \times 10^{-4}$, is the leakage from the primary's finite disk and translates to a stellar diameter of $3.66^{+0.56}_{-0.68}$ mas, consistent with the independent long-baseline value of $3.23 \pm 0.07$ mas. Because the secondary lay inside the diffraction-limited beam and inside typical coronagraphic inner working angles, this is a direct realization of the rotating-nuller concept proposed for space-based exoplanet detection.

Load-bearing premise

The instrumental null-depth floor, measured as an average of 34 null depths on eight other stars ($6.3 \pm 0.7 \times 10^{-4}$), is assumed to be identical for eta Peg and is subtracted from every calibrated point; if that floor is target-dependent—through residual dispersion, pointing drift, or a faint unresolved source—the recovered flux ratio and primary diameter would be systematically biased.

Editorial extensions

If this is right

Companions and stellar disks can be measured at separations below the classical diffraction limit, down to about $0.2$–$0.3\,\lambda/D$, with null-depth accuracies of several $10^{-4}$.
The baseline length needed for a given stellar-diameter accuracy shrinks as the square root of the null-depth accuracy; a $10^{-4}$ nuller needs roughly ten times shorter baselines than a $10^{-2}$ visibility measurement for the same diameter accuracy.
Because the nulling self-calibration algorithm reconstructs the astrophysical null from the statistics of the fluctuations, the hardware requirement relaxes from holding a $\sim\lambda/300$ fringe to simply avoiding fringe hops of more than about $\lambda/4$.
On 30–40 m telescopes the same technique would provide H- and K-band inner working angles of a few milliarcseconds and contrasts of $10^{-4}$ to $10^{-5}$, reaching the innermost hot Jupiters, inner exozodiacal dust, and long-term-trend radial-velocity companion candidates.
Nulling and coronagraphic dark-speckle techniques can be viewed as opposite ends of a single family of measurement methods, differing mainly in the number of sub-apertures contributing to the dark field.

Reading between the lines

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

A natural extension is to use two simultaneous baselines or two wavelengths to break the brightness–separation degeneracy in the product $I_p\theta_p^2$, which remains when neither the companion flux nor its separation is known independently.
The assumption of a star-independent null floor could be tested by interleaving calibrators at matched airmass on the same night; if the floor varies, the single $6.3 \times 10^{-4}$ correction would need to be replaced by a per-target model.
Because the second-harmonic signature is achromatic in form, the same rotation-curve analysis could be applied at shorter wavelengths, where the diffraction limit and inner working angle scale down but the phase-stability requirement tightens.
Paving a large telescope pupil with many small nulling sub-apertures would let a single observation sample several baselines and orientations, replacing physical rotation with instantaneous rotation-curve coverage for snapshot surveys.

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.

Desk Editor's Note

read the letter

Colleague,

The thing to know about 1908.05977 is that it actually delivers the long-promised demonstration: a companion detected by rotating-baseline nulling inside the diffraction limit. The target is the known binary eta Peg, so the paper is not claiming a new discovery; the new thing is the technique working on sky, validating the Bracewell idea with a single-aperture pupil rotator. The Ks-band null-depth rotation curve is fit with one free parameter (the flux ratio), giving 1.08 +/- 0.06 x 10^-2 with reduced chi2 0.49, and the two minimum points give a stellar diameter consistent with the LBI value. That is a real result.

The theoretical sections are also sound. The null-depth fluctuation derivation (Eqns 2-13) is correct and useful: it shows how the astrophysical null can be recovered from the mean and variance of the measured nulls, and the Appendix on why nulling beats visibility at the fringe minimum is a clear statement of the noise advantage. The PFN description is thorough and the limitations are honestly stated, including the one rejected scan and the scalar instrumental floor subtraction.

The soft spot is exactly the one the stress test flags. The instrumental floor is characterized as an average over 34 measurements on 8 stars, 6.3 +/- 0.7 x 10^-4, and that scalar is subtracted from all eta Peg points. No unresolved calibrator was observed through the same K-mirror rotation sequence, and the floor is not shown as a function of baseline angle. If the floor has a coherent cos(2a) component, it would leak into the companion amplitude and the inferred flux ratio. The rms scatter 0.7 x 10^-4 limits random errors, not a coherent floor harmonic. However, the size of the effect is bounded by the data: the companion modulation amplitude is about 2.3 x 10^-3, and the two minimum points, where a floor harmonic would also distort the stellar-diameter measurement, sit near the expected diameter leakage of 3.7 x 10^-4. A large angle-dependent floor at those orientations would force the inferred diameter away from the independent LBI value. So the central detection is probably secure, but the calibration argument is weaker than it should be for a technique demonstration. A calibrator through the same rotation sequence, or at least a plot of the 34 floor measurements versus baseline angle, would tighten it.

One editorial slip: the text in Section 4 says the ~10^-2 null depth occurs with the baseline perpendicular to the binary separation, but the equations and the data (maximum at PA 170 degrees, same as the binary PA) imply it occurs when the baseline is parallel. Worth fixing.

I would send this to review. The demonstration is novel, the analysis is honest, and the one-parameter fit is a strong check. The missing angle-resolved floor calibration is a legitimate referee question, not a fatal flaw. For anyone working on high-contrast small-angle techniques, this is a useful citation.

Referee Report

2 major / 4 minor

Summary. The paper describes the Palomar Fiber Nuller (PFN), a rotating-baseline nulling interferometer operating in the near-infrared on the Palomar 5 m telescope. It derives theoretical performance constraints for high-accuracy NIR nulling, including a statistical treatment of phase fluctuations (Sections 2–3), and reports the first demonstration of faint-companion detection via active baseline rotation, using the known spectroscopic binary η Peg. The measured Ks-band null-depth rotation curve is fit with one free parameter (companion flux ratio), yielding a secondary-to-primary flux ratio of 1.08 ± 0.06 × 10^-2, and the two null-depth minima provide a stellar diameter of 3.66 (+0.56/−0.68) mas, consistent with the long-baseline interferometric value of 3.23 ± 0.07 mas. The paper also discusses the nulling self-calibration (NSC) algorithm and prospects for applying cross-aperture nulling on larger telescopes.

Significance. If the results hold, this is a significant instrument demonstration: it is the first implementation of the rotating-baseline nulling concept originally proposed for space-based exoplanet detection, and it extends nulling to the near-infrared (Ks band) with null-depth accuracies of a few × 10^-4. The theoretical derivation of the null-depth estimate, especially the use of the measured null-depth variance to correct the mean null (Eqs. 2–13), is clearly presented and appears correct. The η Peg fit is good, with reduced chi2 = 0.49, and the companion separation and position angle are taken from an external catalog, avoiding circularity in the detection claim. The stellar diameter derived from two null-depth points is consistent with the independent LBI measurement. The paper also provides a detailed error budget and a thoughtful discussion of future applications on 30 m class telescopes.

major comments (2)

[Section 4] The instrumental null-depth floor is subtracted as a scalar constant (6.3 ± 0.7 × 10^-4) from all η Peg measurements, and the 0.7 × 10^-4 rms is added to the per-point error bars; however, the companion signal itself is a cos(2α) modulation of the null depth versus baseline angle, so a baseline-angle-dependent component of the floor at that frequency is degenerate with the companion flux ratio and would not be removed by the scalar subtraction. The paper does not report the floor as a function of baseline position angle nor demonstrate that the 34 survey measurements sample a range of angles, and the rms scatter does not bound a coherent floor harmonic. I request that the authors either present the floor measurements versus baseline angle, provide an argument that the survey geometry covers the full angle range, or add an explicit systematic uncertainty to the derived companion flux ratio and stellar diameter that accounts for a plausible floor harmonic amplitude.
[Section 4 and Fig. 3] The text is internally inconsistent about the baseline orientation that nulls the companion. The text states that null depths range from ~10^-2 with the baseline perpendicular to the binary separation vector to 3.7 × 10^-4 with the baseline parallel, and later identifies the two lowest null-depth points as being for baseline orientations parallel to the separation; this contradicts Eq. 15, in which the companion transmission vanishes when the baseline is perpendicular to the separation (cos α = 0). The figure caption, labeling 170° as the null-depth maximum and 260° as the minimum, is consistent with Eq. 15. The numerical analysis may be unaffected, but the text must be corrected so that the description of which data points correspond to the companion null is unambiguous.

minor comments (4)

[Section 4] Please specify whether the model fit uses the exact sinusoidal response (Eq. 15) or the small-angle approximation (Eq. 17); at η Peg's separation k b θ / 2 ≈ 0.72 rad, the two differ by up to about 20% near the peak, and if Eq. 17 is used the resulting systematic bias on the derived flux ratio should be quantified.
[Eq. (14a)] The word 'were' in the enumeration of the variance terms should be 'where'.
[Section 4] With 7 data points and one fitted parameter, the reduced chi2 of 0.49 corresponds to chi2 ≈ 2.9 for 6 degrees of freedom; the statement that the error bars 'may be slightly overestimated' is acceptable but could be made more quantitative by citing the corresponding chi2 probability.
[Acknowledgements] The word 'Tedchnology' should be 'Technology'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: companion detection and diameter estimate are calibrated against external data and first-principles model.

full rationale

The paper's derivation chain is self-contained rather than circular. The eta Peg companion separation and position angle are taken from the external 6th Catalog of visual binary stars (Hartkopf, Mason & Rafferty 2008), and the Ks-band flux ratio is the single free parameter fitted to the null-depth rotation curve; it is not an input to the model. The stellar-diameter contribution is estimated from the two lowest null-depth points and then compared with the independent LBI diameter of Nordgren et al. (2001), so the external benchmark is not used to force the result. The companion-response and diameter equations (Eqs. 15-22) are derived in the text from basic fringe and geometric arguments, not assumed from the eta Peg data. The instrumental null floor is a scalar calibration derived from 34 null-depth measurements on eight other stars; subtracting it is a standard calibration step, and any possible baseline-angle dependence of that floor is a systematic-error concern, not circularity. The NSC algorithm is cited to prior same-author work, but the present paper itself derives the relevant null-depth statistics (Eqs. 2-13), and NSC functions as a data-extraction tool rather than as an input that defines the astrophysical signal. No fitted parameter is renamed as a prediction, and no equation reduces to its own input by construction.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claims rest on standard small-angle approximations, Gaussian phase statistics, a uniform-disk model for the primary, and the assumption that the instrumental null floor measured on other stars applies to eta Peg. No new physical entities are introduced; the only fitted parameters are the companion flux ratio, the instrumental null floor, and a secondary stellar-diameter estimate.

free parameters (3)

Companion flux ratio of eta Peg B = 1.08 ± 0.06 x 10^-2
Free parameter in the rotation-curve fit to eta Peg; determines the companion flux ratio (Section 4).
Instrumental null depth floor = 6.3 ± 0.7 x 10^-4
Average null depth measured on 34 observations of 8 stars; used as a target-independent calibration that is subtracted from the eta Peg nulls (Section 4).
Primary stellar diameter leakage = 4.8 ± 1.6 x 10^-4 (3.66 +0.56 -0.68 mas)
Derived from the two null-depth points with the baseline parallel to the binary separation; used as a consistency check against the LBI diameter (Section 4).

assumptions (6)

domain assumption Phase fluctuations are Gaussian
Eqn. 10 uses the Gaussian fourth-moment relation (Delta_phi)^4 = 3 sigma_phi^4 (Section 2.3).
domain assumption Null-depth error terms are small and independent
Eqn. 14a sums variance terms assuming small, independent errors (Section 2.4).
domain assumption Stellar disk is uniform
Eqn. 18 for null depth from a resolved star assumes a uniform disk (Section 3b).
standard math Small-angle point-source response
Eqn. 16 is a small-angle expansion of the fringe response for a point companion (Section 3a).
domain assumption Catalog astrometry for eta Peg is accurate
Separation (30 mas) and position angle (170 degrees) on 2015 June 26 are taken from the 6th Catalog of Visual Binary Stars (Section 4).
domain assumption Instrumental null floor is stable across targets
The floor from eight other stars is assumed to apply to eta Peg and is subtracted (Section 4).

how reviews work

0 comments

Cite this review

Pith. "Pith review of Nulling at short wavelengths: theoretical performance constraints and a demonstration of faint companion detection inside the diffraction limit with a rotating-baseline interferometer." pith.science (2026). https://pith.science/paper/LTFGEGFN

@misc{pith2026190805977,
  author       = {Pith},
  title        = {Pith review of: Nulling at short wavelengths: theoretical performance constraints and a demonstration of faint companion detection inside the diffraction limit with a rotating-baseline interferometer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LTFGEGFN}},
  note         = {Machine review of arXiv:1908.05977}
}

read the original abstract

The Palomar Fiber Nuller (PFN) is a rotating-baseline nulling interferometer that enables high-accuracy near-infrared (NIR) nulling observations with full azimuth coverage. To achieve NIR null-depth accuracies of several x 10-4, the PFN uses a common-mode optical system to provide a high degree of symmetry, single-mode-fiber beam combination to reduce sensitivity to pointing and wavefront errors, extreme adaptive optics to stabilize the fiber coupling and the cross-aperture fringe phase, rapid signal calibration and camera readout to minimize temporal effects, and a statistical null-depth fluctuation analysis to relax the phase stabilization requirement. Here we describe the PFN final design and performance, and provide a demonstration of faint-companion detection by means of nulling-baseline rotation, as originally envisioned for space-based nulling interferometry. Specifically, the Ks-band null-depth rotation curve measured on the spectroscopic binary eta Peg reflects both a secondary star 1.08 +/- 0.06 x 10-2 as bright as the primary, and a null-depth contribution of 4.8 +/- 1.6 x 10-4 due to the size of the primary star. With a 30 mas separation at the time, eta Peg B was well inside both the telescope diffraction-limited beam diameter (88 mas) and typical coronagraphic inner working angles. Finally, we discuss potential improvements that can enable a number of small-angle nulling observations on larger telescopes.

Figures

Figures reproduced from arXiv: 1908.05977 by the authors.

**Figure 1.** Time fraction (probability) that instantaneous nulls depths are below a given value. The dotted curve is representative of first generation AO wavefront-error correction levels (Strehl ratio = 0.6), and the solid curve is representative of ExAO wavefront-error correction levels (Strehl ratio = 0.95) [PITH_FULL_IMAGE:figures/full_fig_p024_1.png] view at source ↗

**Figure 4.** Block diagram of the Palomar Fiber Nuller’s final configuration. The large red arrows trace the flow of the Ks-band light, while the thin blue arrows follow the other wavelengths as they split off to other cameras. (The J-band starlight is not used, being filtered out prior to reaching the pointing camera.) Insets along the sides show: Top left: The subaperture mask, with our favored elliptical apertures seen at 3 o… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Super-resolution wavefront reconstruction in adaptive-optics with pyramid sensors
astro-ph.IM 2025-12 conditional novelty 6.0 of 10

Offset sampling grids in a pyramid wavefront sensor double the number of reconstructable wavefront modes and, with full-rank intensity processing, allow simultaneous phase and amplitude sensing.

Reference graph

Works this paper leans on

2 extracted references · 1 canonical work pages · cited by 1 Pith paper

[1]

Direct Imaging of Exoplanets: Science and Techniques,

Absil, O., Karlsson, A. & Kaltenegger, L., 2003, Proc. SPIE, 4852, 431 Absil, O. et al., 2011, A&A, 535, A68 Absil, O. et al., 2013, A&A, 555, A104 Angel, J. R. P. 1990, in The Next Generation Space Telescope, ed. P. Bely & C. J. Burrows (Baltimore : Space Telescope Science Institute), p. 81 Angel, J. R. P & Wolf, N. J. 1997, ApJ, 475, 373 Atkinson, D., H...

2003
[2]

Single beam signal from zeta Aql (K_mag = 2.9) on 2015 June 28, showing the stability of the stellar signal injected into our SM fiber (under 1.4” seeing conditions ). This data was taken during a normal observing sequence, alternating between the two indiv idual beams, the dark and the combined beam, but the other three states have been removed here. Eac...

work page 2015

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Direct Imaging of Exoplanets: Science and Techniques,

Absil, O., Karlsson, A. & Kaltenegger, L., 2003, Proc. SPIE, 4852, 431 Absil, O. et al., 2011, A&A, 535, A68 Absil, O. et al., 2013, A&A, 555, A104 Angel, J. R. P. 1990, in The Next Generation Space Telescope, ed. P. Bely & C. J. Burrows (Baltimore : Space Telescope Science Institute), p. 81 Angel, J. R. P & Wolf, N. J. 1997, ApJ, 475, 373 Atkinson, D., H...

2003

[2] [2]

Single beam signal from zeta Aql (K_mag = 2.9) on 2015 June 28, showing the stability of the stellar signal injected into our SM fiber (under 1.4” seeing conditions ). This data was taken during a normal observing sequence, alternating between the two indiv idual beams, the dark and the combined beam, but the other three states have been removed here. Eac...

work page 2015