pith. sign in

arxiv: 2510.09806 · v2 · submitted 2025-10-10 · 🌌 astro-ph.IM

AMIGO: a Data-Driven Calibration of the JWST Interferometer

Louis Desdoigts , Benjamin Pope , Max Charles , Peter Tuthill , Dori Blakely , Doug Johnstone , Shrishmoy Ray , Anand Sivaramakrishnan
show 3 more authors
Jens Kammerer Deepashri Thatte Rachel Cooper
This is my paper

Pith reviewed 2026-05-18 07:19 UTC · model grok-4.3

classification 🌌 astro-ph.IM
keywords JWSTAperture Masking Interferometrydata-driven calibrationH2RG detectorhigh-contrast imagingkernel phaseastrometrydifferentiable modeling
0
0 comments X

The pith

Amigo calibrates JWST aperture masking interferometer by forward-modeling optics and detector physics to detect faint companions at 100 mas.

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

The paper introduces Amigo, a data-driven calibration framework for the JWST Near Infrared Imager and Slitless Spectrograph's aperture masking interferometer. It uses an end-to-end differentiable model built in Jax with dLux to simulate optics, detector physics, and readout, with a neural module handling non-linear charge effects in the H2RG sensor. This approach enables better mitigation of systematics like the brighter-fatter effect and extraction of kernel phases from up-the-ramp data. Demonstrations on commissioning observations recover the ABDor AC binary with precise astrometry and detect both the outer and inner companions to HD206893 at contrasts near 10 magnitudes and separations of 100 mas, outperforming prior pipelines. If correct, this re-establishes AMI as a strong option for diffraction-limited high-contrast imaging.

Core claim

Amigo is a full-system forward model of the JWST AMI implemented as a differentiable pipeline that includes an embedded neural submodule to capture non-linear charge redistribution in the detector. When applied to commissioning data it recovers the ABDor AC binary with high-precision astrometry and detects both HD206893B and the inner substellar companion HD206893c at separations of only 100 mas and contrasts approaching 10 magnitudes, results that exceed those from all published pipelines.

What carries the argument

The end-to-end differentiable architecture using Jax and dLux that forward-models the generation of up-the-ramp detector reads with an embedded neural sub-module for non-linear charge redistribution effects.

If this is right

  • Optimal extraction of robust observables such as kernel amplitudes and phases becomes possible while mitigating the brighter-fatter effect.
  • High-precision astrometry on binary stars like ABDor AC is achieved from existing commissioning data.
  • Detection of inner substellar companions at small separations approaching 100 mas and contrasts near 10 magnitudes is enabled.
  • AMI is re-established as a viable competitor for high-contrast imaging at the diffraction limit.

Where Pith is reading between the lines

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

  • Similar differentiable modeling of detector effects could improve analysis pipelines for other JWST instruments facing charge migration issues.
  • The open-source release permits re-analysis of existing AMI observations to search for additional faint companions.
  • Integration with kernel-phase techniques might extend the contrast limits for future high-resolution JWST programs.

Load-bearing premise

The neural sub-module accurately captures non-linear charge redistribution effects in the H2RG detector without introducing new systematics or overfitting to the commissioning data.

What would settle it

Re-processing the HD206893 commissioning data with Amigo fails to recover the inner companion at the claimed separation and contrast, or independent astrometry on ABDor AC differs significantly from the reported values.

Figures

Figures reproduced from arXiv: 2510.09806 by Anand Sivaramakrishnan, Benjamin Pope, Deepashri Thatte, Dori Blakely, Doug Johnstone, Jens Kammerer, Louis Desdoigts, Max Charles, Peter Tuthill, Rachel Cooper, Shrishmoy Ray.

Figure 1
Figure 1. Figure 1: Left panel: Schematic diagram of the 7-hole NRM projected over the primary mirror. Middle panel: The resulting PSF (i.e. interferogram) from the non-redundant mask, visualised on a square-root scale to highlight low-power features. Right panel: The power-spectrum of the PSF featuring baseline-specific regions of fringe power known in the literature as splodges that can be conveniently found by Fourier tran… view at source ↗
Figure 2
Figure 2. Figure 2: High level flow diagram of the AMIGO model and pipeline, showing the input and output product and shapes passed between each modular component. nλ is the number of wavelengths modelled by the optics, ng is the number of groups in the data, and nints is the number of integrations. Each of these model and pipeline components are discussed in detail in their own section. In order to ensure model generalisatio… view at source ↗
Figure 3
Figure 3. Figure 3: Residuals from second-order polynomial fits to ramp data, shown before (top row) and after (bottom row) applying a sine-wave-based correction for ADC integral non-linearity. The left panels plot residuals as a function of the ramp value, while the right panels show the same residuals folded over a modulo 1024 pattern, revealing periodic structure. Prior to correction, the residuals exhibit a strong sinusoi… view at source ↗
Figure 4
Figure 4. Figure 4: Left panel: Residual between the calibrated aperture mask and its idealised undistorted counterpart. Right panel: PSF residuals of the four primary optical effects on the PSF. Top left: Instrumental jitter, applied through a convolution with a Gaussian kernel. Top right: Primary mirror aberrations, modelled using Zernike polynomials on the primary. Bottom left: Aperture mask distortions, modelled by applyi… view at source ↗
Figure 5
Figure 5. Figure 5: Flow chart of the injection of visibility signals to forwards-modelled PSFs. This demonstrates how high-resolution visibility signals can be directly injected into any PSF model provided the appropriate set of visibility basis vectors. An example binary-star signal is injected as a demonstrator. 3.2.1 Latent Visibility Model This simple visibility modelling method satisfies the require￾ments for use in a f… view at source ↗
Figure 6
Figure 6. Figure 6: Demonstration of the produced latent visibility basis used for model-fitting. Top panel: The normalised eigenvalues for each visibility basis vector, ordered by their impact on the PSF in the image plane. Bottom: Representative log amplitude and phase basis vectors over a range of indexes. We can see that higher basis indices have increasing spatial resolution over the OTF, with low order ones picking out … view at source ↗
Figure 7
Figure 7. Figure 7: Example Delay-Insensitive Subspace of Calibrated Observables (DISCO) basis vectors found from the GO 1843 observation, discussed in Section 7.3. Top panel: Normalised log amplitude and phases basis vector variances. Bottom panel: Selection of representative DISCO basis vectors. Matching with the latent visibility basis vectors, low index vectors are better constrained and have lower spatial fidelity. 4. Th… view at source ↗
Figure 8
Figure 8. Figure 8: Schematic diagram of the EDM architecture, showing the three major components and the embedded neural network used to capture non-linear charge migration. 4.2.1 Neural Network Implementation & Architecture The non-linear ramp employs a recurrent architecture to mir￾ror the time evolution of the BFE. Provided with the over￾sampled illuminance I and a charge distribution qt , it seeks to predict the charge d… view at source ↗
Figure 9
Figure 9. Figure 9: Demonstration of the EDM BFE model and recovered detector parameters. Starting with the normalised input charge distribution (top left), the CNN predicts a series of distortion coefficients that are applied to a 3× oversampled set of coordinates for each pixel. The distorted output pixel positions are visualised in the top middle panel. These output positions have their overlap fractions with each neighbou… view at source ↗
Figure 10
Figure 10. Figure 10: Recovered OPD maps from the calibration data for each filter. The top row shows the full OPD maps recovered for the F380M, F430M, and F480M filters, revealing large-scale piston, tip, and tilt terms across the segmented aperture. These low-order aberrations are expected given the off-axis subarray placement used in AMI mode. The bottom row displays the same OPD maps with piston, tip, and tilt removed from… view at source ↗
Figure 11
Figure 11. Figure 11: Summary of AMIGO model fit diagnostics across all five dithers for the F430M calibration data. Each column corresponds to a single dither position. The top row shows the per-pixel log-likelihoods from the final fit, highlighting the location of the target PSF. The middle row displays the average residual z-score per pixel, computed by averaging the uncertainty-normalised residuals across all groups in the… view at source ↗
Figure 12
Figure 12. Figure 12: Schematic diagram of the basic workflow of extracting the DISCOs from data. visibilities. This operation is subtraction, not a division, as AMIGO recovers the log complex visibilities (see Section 3.2.1). The same projection is performed on the parameter covariance matrices to propagate uncertainties correctly. The final step in this procedure is the projection to the statistically-independent DISCO space… view at source ↗
Figure 13
Figure 13. Figure 13: Summary of the fits to the AB Dor C companion. The top panels shows the log-likelihood detection maps over companion position for each filter (F380M, F430M, F480M) from a grid search over the recovered DISCOs. Each panel shows the marginalised log-likelihood surface as a function of companion offset in ∆RA and ∆DEC, revealing a strong and consistent peak in all filters. The greyed central region denotes t… view at source ↗
Figure 14
Figure 14. Figure 14: Joint posterior distributions from an MCMC fit to the AB Dor C companion. Two fit types are shown: Joint — simultaneously modelling astrometry and photometry across all three filters — and per-filter fits. The parameters shown include the separation (mas), position angle (degrees), and contrasts (∆mag) in the F380M, F430M, and F480M bands. The joint-fit samples are shown in black. One- and two-σ credible … view at source ↗
Figure 15
Figure 15. Figure 15: Sensitivity limits derived using the Ruffio et al. (2018) method for calculating 3σ contrast upper limits as a function of angular separation, applied to the F380M, F430M, and F480M filters. The shaded regions denote the ±1σ variation across azimuthal annuli. The dashed black line indicates the approximate contrast limit as calculated from Equation 11, confirming that model performance is consistent with … view at source ↗
Figure 16
Figure 16. Figure 16: Summary of the fits to the HD 206893 companions. The top two panels shows the marginalised log-likelihood surface as a function of companion offset in ∆RA and ∆DEC, revealing a strong and consistent peak in all filters. The top row shows the detection maps for the full data, with the GRAVITY prediction for the B companion shown as a white circle. The middle row shows the detection map after the best-fit B… view at source ↗
Figure 17
Figure 17. Figure 17: Joint posterior distributions from an MCMC fit to the HD 206893 B companion. Two fit types are shown: Joint — simultaneously modelling astrometry and photometry across all three filters — and per-filter fits. While both companions are fit simultaneously, only the B companions samples are shown here for clarity. The parameters shown include the separation (mas), position angle (degrees), and contrasts (∆ma… view at source ↗
Figure 18
Figure 18. Figure 18: Joint posterior distributions from an MCMC fit to the HD 206893 c companion. Two fit types are shown: Joint — simultaneously modelling astrometry and photometry across all three filters — and per-filter fits. While both companions are fit simultaneously, only the B companions samples are shown here for clarity. The parameters shown include the separation (mas), position angle (degrees), and contrasts (∆ma… view at source ↗
Figure 19
Figure 19. Figure 19: Sensitivity limits for HD 206893 derived using the Ruffio et al. (2018) method for calculating 3σ contrast upper limits as a function of angular separation, applied to the F380M, F430M, and F480M filters. The shaded regions denote the ±1σ variation across azimuthal annuli. The dashed black line indicates the approximate contrast limit as calculated from Equation 11, confirming that model performance is co… view at source ↗
Figure 20
Figure 20. Figure 20: Summary of AMIGO model fit diagnostics across all five dithers for the F380M calibration data. Each column corresponds to a single dither position. The top row shows the per-pixel log-likelihoods from the final fit, highlighting the location of the target PSF. The middle row displays the average residual z-score per pixel, computed by averaging the uncertainty-normalised residuals across all groups in the… view at source ↗
Figure 21
Figure 21. Figure 21: Summary of AMIGO model fit diagnostics across all five dithers for the F480M calibration data. Each column corresponds to a single dither position. The top row shows the per-pixel log-likelihoods from the final fit, highlighting the location of the target PSF. The middle row displays the average residual z-score per pixel, computed by averaging the uncertainty-normalised residuals across all groups in the… view at source ↗
read the original abstract

The James Webb Space Telescope (JWST) hosts a non-redundant Aperture Masking Interferometer (AMI) in its Near Infrared Imager and Slitless Spectrograph (NIRISS) instrument, providing the only dedicated interferometric facility aboard - magnitudes more precise than any interferometric experiment previously flown. However, the performance of AMI (and other high resolution approaches such as kernel phase) in recovery of structure at high contrasts has not met design expectations. A major contributing factor has been the presence of uncorrected detector systematics, notably charge migration effects in the H2RG sensor, and insufficiently accurate mask metrology. Here we present Amigo, a data-driven calibration framework and analysis pipeline that forward-models the full JWST AMI system - including its optics, detector physics, and readout electronics - using an end-to-end differentiable architecture implemented in the Jax framework and in particular exploiting the dLux optical modelling package. Amigo directly models the generation of up-the-ramp detector reads, using an embedded neural sub-module to capture non-linear charge redistribution effects, enabling the optimal extraction of robust observables, for example kernel amplitudes and phases, while mitigating systematics such as the brighter-fatter effect. We demonstrate Amigo's capabilities by recovering the ABDor AC binary from commissioning data with high-precision astrometry, and detecting both HD206893B and the inner substellar companion HD206893c: a benchmark requiring contrasts approaching 10 magnitudes at separations of only 100 mas. These results exceed outcomes from all published pipelines, and re-establish AMI as a viable competitor for imaging at high contrast at the diffraction limit. Amigo is publicly available as open-source software community resource.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces AMIGO, a data-driven calibration framework and analysis pipeline for the JWST NIRISS Aperture Masking Interferometer (AMI). It forward-models the full system—including optics, mask metrology, and H2RG detector up-the-ramp reads—via an end-to-end differentiable architecture in JAX/dLux. An embedded neural sub-module captures non-linear charge redistribution effects to mitigate systematics such as the brighter-fatter effect. The pipeline is demonstrated on commissioning data by recovering high-precision astrometry for the ABDor AC binary and detecting both HD206893B and the inner companion HD206893c at contrasts approaching 10 magnitudes at ~100 mas separations, outperforming all published pipelines and re-establishing AMI as a high-contrast competitor at the diffraction limit. The software is released as open-source.

Significance. If the neural module proves generalizable, AMIGO could meaningfully advance high-contrast interferometric imaging with JWST by addressing detector systematics that have limited AMI performance to date. The differentiable end-to-end approach and public code release are clear strengths that support reproducibility. However, the headline claims rest on demonstration rather than exhaustive validation, so the significance is currently provisional pending clearer evidence that gains arise from improved physics modeling rather than dataset-specific fitting.

major comments (3)
  1. [Abstract / Results] Abstract and Results (demonstration on HD206893): The claim of detecting HD206893c at ~10 mag contrast / 100 mas and exceeding all published pipelines is presented without quantitative error bars, SNR values, false-alarm probabilities, or injected-signal recovery tests. This is load-bearing for the central claim that AMIGO re-establishes AMI as competitive.
  2. [Methods (neural sub-module)] Methods (neural sub-module training): The manuscript does not specify whether the commissioning datasets used for the ABDor AC and HD206893 demonstrations were held out from neural-module training, nor does it report cross-validation, regularization details, or ablation tests that isolate the neural module's contribution versus the physical forward model alone. This directly affects whether performance gains are generalizable or risk overfitting to the same data shown in the results.
  3. [Validation / Results] §4 (or equivalent validation section): No independent test set or synthetic-data benchmarks are described to confirm that the neural module captures non-linear charge redistribution without introducing new systematics, which is required to substantiate the comparison to prior pipelines.
minor comments (2)
  1. [Methods] Notation: The distinction between kernel amplitudes/phases extracted before and after neural correction could be clarified with an explicit equation or diagram in the methods.
  2. [Figures] Figure clarity: Residual maps in the demonstration figures would benefit from explicit scale bars and a statement of the rms level achieved relative to prior pipelines.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed review of our manuscript on AMIGO. We have addressed each major comment point by point below. Revisions have been made to incorporate quantitative metrics, clarify training procedures, and add validation benchmarks, thereby strengthening the evidence for the framework's performance and generalizability.

read point-by-point responses

  1. Referee: [Abstract / Results] Abstract and Results (demonstration on HD206893): The claim of detecting HD206893c at ~10 mag contrast / 100 mas and exceeding all published pipelines is presented without quantitative error bars, SNR values, false-alarm probabilities, or injected-signal recovery tests. This is load-bearing for the central claim that AMIGO re-establishes AMI as competitive.

    Authors: We agree that explicit quantitative support strengthens the central claims. In the revised manuscript we have updated both the abstract and the Results section to report error bars on the recovered astrometry for HD206893c, SNR values >5 for both companions, and false-alarm probabilities <10^{-4} derived from the kernel-phase likelihood ratio. We have also added injected-companion recovery tests performed on the same commissioning datasets at contrasts and separations matching the real detections; these tests show reliable recovery with AMIGO while prior pipelines yield higher false-negative rates. These additions directly substantiate the performance comparison. revision: yes

  2. Referee: [Methods (neural sub-module)] Methods (neural sub-module training): The manuscript does not specify whether the commissioning datasets used for the ABDor AC and HD206893 demonstrations were held out from neural-module training, nor does it report cross-validation, regularization details, or ablation tests that isolate the neural module's contribution versus the physical forward model alone. This directly affects whether performance gains are generalizable or risk overfitting to the same data shown in the results.

    Authors: We thank the referee for identifying this omission. The revised Methods section now explicitly states that the ABDor AC and HD206893 commissioning observations were held out from neural-module training; the submodule was trained solely on a separate set of calibration frames acquired during the same commissioning period. We report 5-fold cross-validation results, L2 regularization with coefficient 0.005, and ablation experiments comparing the full differentiable model against the physical forward model alone. On the held-out validation data the neural component reduces residual systematics by approximately 25 percent, confirming that the gains arise from improved modeling rather than overfitting. revision: yes

  3. Referee: [Validation / Results] §4 (or equivalent validation section): No independent test set or synthetic-data benchmarks are described to confirm that the neural module captures non-linear charge redistribution without introducing new systematics, which is required to substantiate the comparison to prior pipelines.

    Authors: We acknowledge the value of explicit validation. The revised §4 now contains a dedicated validation subsection that describes an independent test set drawn from additional JWST NIRISS AMI observations never used in training or the primary science demonstrations. We also present synthetic-data benchmarks in which known non-linear charge-redistribution effects are injected into simulated up-the-ramp reads; the neural module recovers these effects with residuals consistent with photon noise and without introducing new systematic patterns, as quantified by reduced-chi-squared values near unity on the test data. These results support that the performance improvements are attributable to better physics modeling. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper presents an end-to-end differentiable forward model in Jax/dLux that incorporates physical optics, detector readout, and an embedded neural sub-module for non-linear charge redistribution in H2RG sensors. The central results consist of applying this calibrated pipeline to commissioning data to recover astrometry for ABDor AC and detect companions in HD206893 at high contrast, with explicit comparison to outcomes from all published pipelines. No equations or steps are described that reduce the reported astrometric recoveries or detections by construction to the neural-module training on the identical datasets; the model is framed as a physical simulation augmented by learned corrections rather than a self-referential fit. The demonstration is externally benchmarked against independent prior pipelines, satisfying the criteria for a self-contained derivation without load-bearing self-citation chains or fitted inputs renamed as predictions.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Limited details available from abstract; the model relies on assumptions about detector physics and the completeness of the forward simulation.

free parameters (1)
  • neural sub-module weights
    Parameters of the embedded neural network trained to model charge migration effects.
axioms (1)
  • domain assumption The end-to-end model accurately represents JWST optics, mask metrology, and detector readout physics.
    Invoked as the basis for forward-modeling up-the-ramp reads.

pith-pipeline@v0.9.0 · 5873 in / 1335 out tokens · 28163 ms · 2026-05-18T07:19:43.467237+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

123 extracted references · 123 canonical work pages · 7 internal anchors

  1. [1]

    2025, AJ, 169, 254

    Adam, A., Stone, C., Bottrell, C., et al. 2025, AJ, 169, 254

    work page 2025

  2. [2]

    2023, Computer Methods in Applied Mechanics and Engineering, 406, 115902

    Akhare, D., Luo, T., & Wang, J.-X. 2023, Computer Methods in Applied Mechanics and Engineering, 406, 115902

    work page 2023

  3. [3]

    2014, Journal of Instrumentation, 9, C03048–C03048

    Antilogus, P., Astier, P., Doherty, P., Guyonnet, A., & Regnault, N. 2014, Journal of Instrumentation, 9, C03048–C03048

    work page 2014

  4. [4]

    H., et al

    Argyriou, I., Lage, C., Rieke, G. H., et al. 2023, A&A, 680, A96

    work page 2023

  5. [5]

    P., Bendek, E., Monacelli, B., et al

    Bailey, V. P., Bendek, E., Monacelli, B., et al. 2023, in Society of Photo- Optical Instrumentation Engineers (SPIE) Conference Series, V ol. 12680, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, 126800T

    work page 2023

  6. [6]

    E., Haniff, C

    Baldwin, J. E., Haniff, C. A., Mackay, C. D., & Warner, P. J. 1986, Nature, 320, 595

    work page 1986

  7. [7]

    O., Kammerer, J., Pueyo, L., et al

    Balmer, W. O., Kammerer, J., Pueyo, L., et al. 2025, AJ, 169, 209

    work page 2025

  8. [8]

    2015, SIAM Review, 57, 483

    Benner, P., Gugercin, S., & Willcox, K. 2015, SIAM Review, 57, 483

    work page 2015

  9. [9]

    A Conceptual Introduction to Hamiltonian Monte Carlo

    Betancourt, M. 2017, arXiv e-prints, arXiv:1701.02434

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  10. [10]

    P., Jankowiak, M., et al

    Bingham, E., Chen, J. P., Jankowiak, M., et al. 2019, J. Mach. Learn. Res., 20, 28:1

    work page 2019

  11. [11]

    2025, AJ, 169, 152

    Blakely, D., Johnstone, D., Stolker, T., et al. 2025, AJ, 169, 152

    work page 2025

  12. [12]

    2022, ApJ, 931, 3

    Blakely, D., Francis, L., Johnstone, D., et al. 2022, ApJ, 931, 3

    work page 2022

  13. [13]

    2024, The James Webb Interfer- ometer: Space-based interferometric detections of PDS 70 b and c at 4.8 µm, arXiv:2404.13032

    Blakely, D., Johnstone, D., Cugno, G., et al. 2024, The James Webb Interfer- ometer: Space-based interferometric detections of PDS 70 b and c at 4.8 µm, arXiv:2404.13032

    work page arXiv 2024

  14. [14]

    M., & Mourard, D

    Bonneau, D., Delfosse, X., Clausse, J. M., & Mourard, D. 2012, SearchCal: The JMMC Evolutive Search Calibrator Tool, ASCL, record ascl:1210.012

    work page 2012

  15. [15]

    B., et al

    Born, M., W olf, E., Bhatia, A. B., et al. 1999, Principles of Optics: Electro- magnetic Theory of Propagation, Interference and Diffraction of Light, 7th edn. (Cambridge University Press)

    work page 1999

  16. [16]

    Dynamic Filter Networks

    Brabandere, B. D., Jia, X., Tuytelaars, T., & Gool, L. V. 2016, Dynamic Filter Networks, arXiv:1605.09673

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  17. [17]

    2018, JAX: composable transfor- mations of Python+NumPy programs

    Bradbury, J., Frostig, R., Hawkins, P., et al. 2018, JAX: composable transfor- mations of Python+NumPy programs

    work page 2018

  18. [18]

    Broyden, C. G. 1970, IMA Journal of Applied Mathematics, 6, 76

    work page 1970

  19. [19]

    2023, ApJL, 947, L30

    Calissendorff, P., De Furio, M., Meyer, M., et al. 2023, ApJL, 947, L30

    work page 2023

  20. [20]

    2023, OJA, 6, 15

    Campagne, J.-E., Lanusse, F., Zuntz, J., et al. 2023, OJA, 6, 15

    work page 2023

  21. [21]

    2023, JA TIS, 9, 028004

    Chaushev, A., Sallum, S., Lozi, J., et al. 2023, JA TIS, 9, 028004

    work page 2023

  22. [22]

    C., Tuthill, P

    Cheetham, A. C., Tuthill, P. G., Sivaramakrishnan, A., & Lloyd, J. P. 2012, Optics Express, 20, 29457

    work page 2012

  23. [23]

    2025, arXiv e-prints, arXiv:2502.00100

    Czekala, I., Jennings, J., Zawadzki, B., et al. 2025, arXiv e-prints, arXiv:2502.00100 De Furio, M., Lew, B., Beichman, C., et al. 2023, 948, 92

    work page arXiv 2025

  24. [24]

    2024, arXiv e-prints, arXiv:2406.08704

    Desdoigts, L., Pope, B., Gully-Santiago, M., & Tuthill, P. 2024, Differen- tiable Optics with dLux II: Optical Design Maximising Fisher Information, arXiv:2406.08704

    work page arXiv 2024

  25. [25]

    Desdoigts, L., Pope, B. J. S., Dennis, J., & Tuthill, P. G. 2023, JA TIS, 9, doi:10.1117/1.jatis.9.2.028007

    work page doi:10.1117/1.jatis.9.2.028007 2023

  26. [26]

    Dholakia, S., Dholakia, S., & Pope, B. J. S. 2024, arXiv e-prints, arXiv:2410.03449

    work page arXiv 2024

  27. [27]

    J., Adam, A., et al

    Dia, N., Yantovski-Barth, M. J., Adam, A., et al. 2025, arXiv e-prints, arXiv:2501.02473

    work page arXiv 2025

  28. [28]

    B., Beaulieu, M., et al

    Doyon, R., Hutchings, J. B., Beaulieu, M., et al. 2012, in Space Telescopes and Instrumentation 2012: Optical, Infrared, and Millimeter Wave, ed. M. C. Clampin, G. G. Fazio, H. A. MacEwen, & J. M. O. Jr., V ol. 8442, International Society for Optics and Photonics (SPIE), 84422R

    work page 2012

  29. [29]

    J., Hutchings, J

    Doyon, R., Willott, C. J., Hutchings, J. B., et al. 2023, arXiv:2306.03277

    work page arXiv 2023

  30. [30]

    2019, Journal of Open Source Software, 4, 1352

    Dube, B. 2019, Journal of Open Source Software, 4, 1352

    work page 2019

  31. [31]

    T., Smith, J., Rubinstein, M., et al

    Feng, B. T., Smith, J., Rubinstein, M., et al. 2023, arXiv e-prints, arXiv:2304.11751

    work page arXiv 2023

  32. [32]

    1970, The Computer Journal, 13, 317

    Fletcher, R. 1970, The Computer Journal, 13, 317

    work page 1970

  33. [33]

    Follette, K. B. 2023, An Introduction to High Contrast Differential Imaging of Exoplanets and Disks, arXiv:2308.01354

    work page arXiv 2023

  34. [34]

    2023, dfm/tinygp: The tiniest of Gaussian Process libraries, doi:10.5281/ZENODO.7646759

    Foreman-Mackey, D. 2023, dfm/tinygp: The tiniest of Gaussian Process libraries, doi:10.5281/ZENODO.7646759

    work page doi:10.5281/zenodo.7646759 2023

  35. [35]

    2015, A&A, 579, A68

    Gallenne, A., Mérand, A., Kervella, P., et al. 2015, A&A, 579, A68

    work page 2015

  36. [36]

    P., Mather, J

    Gardner, J. P., Mather, J. C., Clampin, M., et al. 2006, Space Science Reviews, 123, 485

    work page 2006

  37. [37]

    Chandler-Mather et al. supplementary material

    Gardner, J. P., Mather, J. C., Abbott, R., et al. 2023, PASP, 135, 068001

    work page 2023

  38. [38]

    1970, Mathematics of Computation, 24, 23

    Goldfarb, D. 1970, Mathematics of Computation, 24, 23

    work page 1970

  39. [39]

    Goodman, J. W. 2005, Introduction to Fourier Optics, 3rd edn. (Roberts and Company Publishers)

    work page 2005

  40. [40]

    2024, PASP, 136, 014503

    Goudfrooij, P., Grumm, D., V olk, K., & Bushouse, H. 2024, PASP, 136, 014503

    work page 2024

  41. [41]

    Z., Pueyo, L., Sivaramakrishnan, A., & Lacour, S

    Greenbaum, A. Z., Pueyo, L., Sivaramakrishnan, A., & Lacour, S. 2015, ApJ, 798, 68

    work page 2015

  42. [42]

    Z., Cheetham, A., Sivaramakrishnan, A., et al

    Greenbaum, A. Z., Cheetham, A., Sivaramakrishnan, A., et al. 2019, AJ, 157, 249

    work page 2019

  43. [43]

    Griewank, A., & Walther, A. 2008, Evaluating Derivatives: Principles and Techniques of Algorithmic Differentiation, Second Edition, Other Titles in Applied Mathematics (Society for Industrial and Applied Mathematics (SIAM, 3600 Market Street, Floor 6, Philadelphia, PA 19104))

    work page 2008

  44. [44]

    Gully-Santiago, M., & Morley, C. V. 2022, ApJ, 941, 200

    work page 2022

  45. [45]

    A., Kuchner, M

    Guyon, O., Pluzhnik, E. A., Kuchner, M. J., Collins, B., & Ridgway, S. T. 2006, ApJS, 167, 81–99

    work page 2006

  46. [46]

    2015, A&A, 575, A41

    Guyonnet, A., Astier, P., Antilogus, P., Regnault, N., & Doherty, P. 2015, A&A, 575, A41

    work page 2015

  47. [47]

    R., Millman, K

    Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357

    work page 2020

  48. [48]

    2024, exoplanet-dev/jaxoplanet: Astronomical time series analysis with JAX, doi:10.5281/zenodo.10736936

    Hattori, S., Garcia, L., Murray, C., et al. 2024, exoplanet-dev/jaxoplanet: Astronomical time series analysis with JAX, doi:10.5281/zenodo.10736936

    work page doi:10.5281/zenodo.10736936 2024

  49. [49]

    2002, Optics, 4th edn

    Hecht, E. 2002, Optics, 4th edn. (San Francisco: Addison-Wesley)

    work page 2002

  50. [50]

    2020, Optax: composable gradient transformation and optimisation, in JAX!

    Hessel, M., Budden, D., Viola, F., et al. 2020, Optax: composable gradient transformation and optimisation, in JAX!

    work page 2020

  51. [51]

    D., et al

    Hinkley, S., Lacour, S., Marleau, G. D., et al. 2023, A&A, 671, L5

    work page 2023

  52. [52]

    Hinton, S. R. 2016, JOSS, 1, 00045

    work page 2016

  53. [53]

    Data analysis recipes: Fitting a model to data

    Hogg, D. W., Bovy, J., & Lang, D. 2010, arXiv e-prints, arXiv:1008.4686

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  54. [54]

    W., & Villar, S

    Hogg, D. W., & Villar, S. 2024, Is machine learning good or bad for the natural sciences?, arXiv:2405.18095

    work page arXiv 2024

  55. [55]

    M., Hogg, D

    Horta, D., Price-Whelan, A. M., Hogg, D. W., Ness, M. K., & Casey, A. R. 2025, AJ, 169, 314

    work page 2025

  56. [56]

    Hunter, J. D. 2007, Computing In Science & Engineering, 9, 90

    work page 2007

  57. [57]

    Ireland, M. J. 2013, MNRAS, 433, 1718

    work page 2013

  58. [58]

    H., Benedict, G

    Jefferys, W. H., Benedict, G. F., Hemenway, P. D., Shelus, P. J., & Duncombe, R. L. 1985, Celestial Mechanics, 37, 299

    work page 1985

  59. [59]

    Jennison, R. C. 1958, MNRAS, 118, 276

    work page 1958

  60. [60]

    A., Vandal, T., et al

    Kammerer, J., Cooper, R. A., Vandal, T., et al. 2023, PASP, 135, 014502

    work page 2023

  61. [61]

    E., Kevrekidis, I

    Karniadakis, G. E., Kevrekidis, I. G., Lu, L., et al. 2021, Nature Reviews Physics, 3, 422

    work page 2021

  62. [62]

    1991, Laplace’s method in Bayesian analysis, doi:10.1090/conm/115/07

    Kass, R., Tierney, L., & Kadane, J. 1991, Laplace’s method in Bayesian analysis, doi:10.1090/conm/115/07

    work page doi:10.1090/conm/115/07 1991

  63. [63]

    On neural differential equations

    Kidger, P. 2022, On Neural Differential Equations, arXiv:2202.02435

    work page arXiv 2022

  64. [64]

    Patrick Kidger and Cristian Garcia

    Kidger, P., & Garcia, C. 2021, Equinox: neural networks in JAX via callable PyTrees and filtered transformations, arXiv:2111.00254

    work page arXiv 2021

  65. [65]

    Adam: A Method for Stochastic Optimization

    Kingma, D. P., & Ba, J. 2014, arXiv e-prints, arXiv:1412.6980

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  66. [66]

    2024, Nature, 632, 1060–1066

    Kochkov, D., Yuval, J., Langmore, I., et al. 2024, Nature, 632, 1060–1066

    work page 2024

  67. [67]

    E., Hook, R

    Krist, J. E., Hook, R. N., & Stoehr, F. 2011, in Optical Modeling and Perfor- mance Predictions V, ed. M. A. Kahan, V ol. 8127, International Society for Optics and Photonics (SPIE), 81270J

    work page 2011

  68. [68]

    Lajoie, C.-P., Lallo, M., Meléndez, M., et al. 2023, OTE Science Performance Memo 2 - A Y ear of Wavefront Sensing with JWST in Flight: Cycle 1 Telescope Monitoring & Maintenance Summary, Technical Report JWST- STScI-008497, arXiv:2307.11179

    work page arXiv 2023

  69. [69]

    2011, JModOptic, 58, 1678

    Lakshminarayanan, V., & Fleck, A. 2011, JModOptic, 58, 1678

    work page 2011

  70. [70]

    M., Hankins, M

    Lau, R. M., Hankins, M. J., Sanchez-Bermudez, J., et al. 2023, arXiv:2311.15948

    work page arXiv 2023

  71. [71]

    2015, Nature, 521, 436

    LeCun, Y., Bengio, Y., & Hinton, G. 2015, Nature, 521, 436

    work page 2015

  72. [72]

    2023, Inverse Problems, 39, 035008

    Liaudat, T., Starck, J.-L., Kilbinger, M., & Frugier, P.-A. 2023, Inverse Problems, 39, 035008

    work page 2023

  73. [73]

    2024, PASP, 136, 114504 34 Louis Desdoigtset al

    Lucas, M., Norris, B., Guyon, O., et al. 2024, PASP, 136, 114504 34 Louis Desdoigtset al

    work page 2024

  74. [74]

    MacKay, D. J. C. 2002, Information Theory, Inference & Learning Algorithms (USA: Cambridge University Press)

    work page 2002

  75. [75]

    Margossian, C. C. 2019, WIREs Data Mining and Knowledge Discovery, 9, doi:10.1002/widm.1305

    work page doi:10.1002/widm.1305 2019

  76. [76]

    2020, New insights and perspectives on the natural gradient method, arXiv:1412.1193

    Martens, J. 2020, New insights and perspectives on the natural gradient method, arXiv:1412.1193

    work page arXiv 2020

  77. [77]

    2010, ApJ, 724, 464

    Martinache, F. 2010, ApJ, 724, 464

    work page 2010

  78. [78]

    2013, PASP, 125, 422

    Martinache, F. 2013, PASP, 125, 422

    work page 2013

  79. [79]

    2016, in SPIE Conference Series, V ol

    Martinache, F. 2016, in SPIE Conference Series, V ol. 9907, Optical and In- frared Interferometry and Imaging V, ed. F. Malbet, M. J. Creech-Eakman, & P. G. Tuthill, 990712

    work page 2016

  80. [80]

    2020, A&A, 636, A72

    Martinache, F., Ceau, A., Laugier, R., et al. 2020, A&A, 636, A72

    work page 2020

Showing first 80 references.