Modelling Fourier filtering wavefront sensors for PSD-based methods: the AOPERA tool
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.
Vera: I'm Vera.
Jocelyn: And I'm Jocelyn, here with Subrahmanyan.
Vera: We're looking at the paper: "Modelling Fourier filtering wavefront sensors for PSD-based methods".
Title and authors: Vera: So we're looking at this paper called "Modelling Fourier filtering wavefront sensors for PSD-based methods: the AOPERA tool". It’s about creating a way to accurately describe and manage how these Fourier filtering wavefront sensors behave, especially when things aren't linear.
Jocelyn: Right, so it seems like they're building a method using a convolutive formalism to handle that non-linear behavior in the wavefront sensor. This is super important for understanding and predicting how adaptive optics systems will perform when we design them or when we actually run them.
Subrahmanyan: It sounds like they're tackling the core issue of getting an accurate description of these AO systems during their design and operation phases, which is a critical technique for figuring out what to expect.
Vera: The method they developed is based on splitting the residual phase equation into terms related to spatio-temporal error, noise, and aliasing based on a certain reconstruction domain F where D(f)R(f)M⊓(f) equals gamma one two.
Jocelyn: That means they are trying to break down the residual phase into three distinct parts, which is spatio-temporal error, noise error, and aliasing error. That sounds like a systematic way to look at the problem.
Subrahmanyan: And this splitting allows them to compute the AO system response while also including that non-linear behavior of the wavefront sensor within PSD based numerical tools.
Vera: They do this by describing how the wavefront sensor is sensitive combined with managing that non-linearity, which greatly improves the accuracy of describing AO systems through numerical simulations.
Jocelyn: How they do this involves splitting the residual phase equation into those three pieces, but they also compute the spatial PSD of the phase, noted W(f), by integrating over temporal frequencies.
Subrahmanyan: They assume those three terms in equation eight are statistically uncorrelated to obtain the total residual PSD, which is Wres of f equals Wres spatio-temporal plus Wres noise plus Wres aliasing.
Vera: And they make an assumption about the measurement noise being white in the temporal direction for both photon noise and readout noise.
Jocelyn: Then they bring in what they call the frozen flow hypothesis, which assumes that the atmosphere is just a collection of independent turbulent layers sliding with a given wind speed and direction.
Subrahmanyan: Because of this hypothesis, the spatio-temporal PSD simplifies so it only depends on the atmospheric PSD and on the response of the AO loop evaluated at a specific temporal frequency nu equals V times f.
Vera: They then talk about managing non-linearity and gains, saying that sensitivity is non-linear if it changes with the phase being measured, which leads to optical gains defined as OG equals M sub calib divided by M calib.
Jocelyn: The method includes evaluating the current sensitivity through the point spread function seen by the wavefront sensor, solving that using an iterative method involving initialization, residual PSD computation, PSF computation and updating the estimation of sensitivity.
Title and authors: Subrahmanyan: Managing those optical gains can be achieved by updating the reconstructor during AO loop operations to compensate for them and eventually restore gamma one.
Vera: This whole approach is then validated against end-to-end simulations using the OOPAO tool, showing similarity in fitting error, temporal error and noise error with high fidelity.
Jocelyn: And this model and its numerical implementation have been applied to the PROVIDENCE two meter telescope to be built in France by the end of twenty twenty eight.
Subrahmanyan: It also suggests it could be used for ELT instruments which include a pyramid AO system, which is a big deal for future observations.
Vera: They can compute an AO corrected point spread function by separating it into a diffraction static component and a phase residuals component, allowing the calculation of hres equals e to the negative sigma squared res F to the negative n e F to the negative.
Jocelyn: That formula lets you get the science point spread function, which is what we really want to measure after all that filtering is done.
Subrahmanyan: Furthermore, this system can be used for optimal modulation abacus tuning by computing the optimum of an AO system on the fly during on-sky observations following the evolution of turbulence during the night.
Vera: The validation against tools like OOPAO confirms that it provides a high fidelity description of WFS sensitivity and non-linearity, which is what we were aiming for.
Jocelyn: So, to wrap up this paper "Modelling Fourier filtering wavefront sensors for PSD-based methods: the AOPERA tool", it gives us a solid framework for handling non-linear behavior in these wavefront sensors using a convolutive formalism.
Subrahmanyan: It provides a way to split the error budget into spatio-temporal, noise, and aliasing terms and offers tools to manage optical gains during loop operations.
Vera: The implication is that we can get much more accurate performance estimations for AO systems during design or operation phases because this method has been applied to the PROVIDENCE two meter telescope.
Jocelyn: And it opens the door for using these ideas on much larger instruments, like ELT systems with pyramid AO, and even allows us to tune modulation abacuses optimally while observing.
Subrahmanyan: It’s a way to get a better picture of how the atmospheric turbulence affects the signal processing at different scales, which is key for cosmology as well.
Vera: So we’ve seen how they use this AOPERA tool to model those non-linear effects and validate it against other established simulation tools.
Jocelyn: It’s a lot of heavy math, but the result is a tool that helps us understand the real-world performance of AO systems better when they are operating in complex conditions.
Subrahmanyan: The whole point is getting this accurate description for understanding and predicting the behavior of AO systems during design or operation phases.
The paper's summary: Vera: So, this paper is about building something called the AOPERA tool to model these Fourier filtering wavefront sensors better when we look at them using power spectral density methods.
Jocelyn: Right, so it’s essentially creating a way to describe that tricky non-linear behavior in the sensor using a convolution formalism. It’s super important for figuring out how adaptive optics systems will actually perform during design or when they are running on the telescope.
Subrahmanyan: So, we're talking about getting an accurate description of these AO systems during their design and operation phases. That’s a critical technique for understanding what to expect before we even point the telescope at the sky.
Vera: They developed this method by splitting the residual phase equation into spatio-temporal error, noise, and aliasing terms based on some reconstruction domain F where D of f R of f M sub F equals gamma one two.
Jocelyn: That means they're trying to break down the leftover phase into three separate parts—the spatio-temporal error, the noise error, and any aliasing—which is a systematic way to look at the problem.
Subrahmanyan: And this splitting lets them compute the AO system response while also including that non-linear behavior of the wavefront sensor inside their power spectral density based tools.
Vera: They do this by describing how sensitive the wavefront sensor is combined with managing that non-linearity, which greatly improves how accurately we can describe these systems through numerical simulations.
Jocelyn: How they handle it involves splitting that phase equation, but they also compute the spatial power spectral density of the phase, which they call W of f, by integrating over temporal frequencies.
Subrahmanyan: They assume those three terms in equation eight are statistically uncorrelated to get the total residual power spectral density, which is W res of f equals W res spatio-temporal plus W res noise plus W res aliasing.
Vera: And they make an assumption that the measurement noise is white in the temporal direction for both photon noise and readout noise.
Jocelyn: Then they bring in what’s called the frozen flow hypothesis, which assumes that the atmosphere is just a collection of independent turbulent layers sliding with a specific wind speed and direction.
Subrahmanyan: Because of that hypothesis, the spatio-temporal power spectral density simplifies so it only depends on the atmospheric power spectral density and on how the adaptive optics loop responds at a specific temporal frequency nu equals V times f.
Vera: Then they discuss managing non-linearity and gains, saying sensitivity is non-linear if it changes depending on the phase being measured, which leads to optical gains defined as M sub calib divided by M calib.
Jocelyn: The method includes checking the current sensitivity through the point spread function that the wavefront sensor sees, solving that using an iterative process involving initialization, residual power spectral density computation, and updating the estimate of sensitivity.
The paper's summary: Subrahmanyan: Managing those optical gains can be done by updating the reconstructor during AO loop operations to compensate for them and eventually bring gamma one back to one.
Vera: This whole approach is then validated against end-to-end simulations using the OOPAO tool, showing a high fidelity match in how it fits error, temporal error, and noise error.
Jocelyn: And this model has been applied to the PROVIDENCE two meter telescope being built in France by two thousand twenty-eight.
Subrahmanyan: It also suggests it could be used for future very large telescopes with pyramid AO systems.
Vera: They can compute an AO corrected point spread function by separating it into a diffraction static component and a phase residuals component, which lets you calculate the science point spread function using that formula involving e to the negative sigma squared res F to the negative n e F to the negative.
Jocelyn: That formula gives you exactly what we want—the science point spread function after all that filtering is done.
Subrahmanyan: Furthermore, this system can be used for tuning modulation abacuses optimally by computing the best setting for an AO system on the fly during observations as turbulence changes at night.
Vera: The validation against tools like OOPAO confirms it provides a high fidelity description of wavefront sensor sensitivity and non-linearity.
Jocelyn: So, to wrap up this paper, it gives us a solid framework for handling those non-linear effects in these sensors using a convolution formalism.
Subrahmanyan: It provides a way to split the error budget into spatio-temporal, noise, and aliasing terms and offers tools to manage optical gains during loop operations.
Vera: The implication is that we can get much more accurate performance estimations for AO systems during design or operation phases because this method has been applied to the PROVIDENCE two meter telescope.
Jocelyn: And it opens the door for using these ideas on much larger instruments, like ELT systems with pyramid AO, and even allows us to tune modulation abacuses optimally while observing.
Subrahmanyan: It’s a way to get a better picture of how atmospheric turbulence affects the signal processing at different scales, which is key for cosmology as well.
Vera: So we’ve seen how they use this AOPERA tool to model those non-linear effects and validate it against other established simulation tools.
Jocelyn: It’s a lot of heavy math, but the result is a tool that helps us understand the real-world performance of AO systems better when they are operating in complex conditions.
Subrahmanyan: The whole point is getting this accurate description for understanding and predicting the behavior of AO systems during design or operation phases.
Vera: So, it’s about using PSD methods to get a better handle on those tricky non-linear effects in wavefront sensing.
The paper's improvements: Tom: So, we're looking at how they suggest improving this AOPERA tool for modelling Fourier filtering wavefront sensors using power spectral density methods.
Vera: It seems like they’re really focusing on making the estimation of sensitivity more precise by including that non-linear behavior more directly in the math.
Jocelyn: Right, so it's about being able to "manage optical gains" better, meaning they can figure out how the sensor behaves even when things aren't perfectly linear.
Subrahmanyan: So this means any value of gamma is possible with their method because they are accounting for these gains during the AO loop operations.
Vera: They also suggest a new way to compute an AO corrected point spread function by splitting it into a static diffraction part and a phase residuals part.
Jocelyn: That allows them to calculate the science point spread function using that formula involving e to the negative sigma squared res F to the negative n e F to the negative, which is really useful for measurement.
Subrahmanyan: That gives us a concrete way to get that science PSF after all that filtering is done, which feeds directly into our cosmological models.
Vera: And they propose using this tool for optimal modulation abacus tuning by computing the best setting for an AO system on the fly during observations as turbulence evolves at night.
Jocelyn: That means we could potentially tune things in real-time while we're actually looking at the sky, following the changing conditions.
Subrahmanyan: That’s a huge practical implication because it links theoretical models to immediate observational strategies for structure formation studies.
Vera: And they validated this improved tool against tools like OOPAO, showing that it gives a high fidelity description of sensitivity and non-linearity.
Jocelyn: So, even with the added features, the core validation still shows that this model is accurate in describing what we see in simulations.
Subrahmanyan: This suggests that if we can get this level of accuracy for one telescope, it could apply to future very large telescopes with pyramid AO systems.
Vera: The implication is that this approach could be used to design and predict the performance of these systems much more accurately, even on bigger instruments.
Jocelyn: So, the focus shifts from just describing the error budget to actively managing those gains during real-world operation.
Conclusion: Vera: So we’re wrapping up this discussion on the paper "Modelling Fourier filtering wavefront sensors for PSD-based methods: the AOPERA tool". Essentially, they've shown a framework to model those tricky non-linear behaviors in Fourier filtering wavefront sensors using power spectral density methods.
Jocelyn: Right, it boils down to creating a method that accurately describes how the sensor responds when things are messy and not linear, which is key for predicting performance during design or operation.
Subrahmanyan: From a cosmological standpoint, this level of precision in modeling the wavefront sensor response could refine our understanding of how atmospheric turbulence affects signal processing across different scales.
Vera: They showed how splitting the residual phase equation into spatio-temporal error, noise, and aliasing helps them compute that total residual power spectral density.
Jocelyn: And they used the frozen flow hypothesis to simplify that calculation, assuming the atmosphere is just a collection of independent turbulent layers moving at specific wind speeds.
Subrahmanyan: That simplification means the resulting power spectral density only depends on the atmospheric conditions and how much your adaptive optics loop reacts at a certain temporal frequency.
Vera: They also developed ways to manage those optical gains, which are basically non-linear effects where the sensor's sensitivity changes depending on what phase it’s seeing.
Jocelyn: And they showed that by updating the reconstructor during operation, you can compensate for those gains and try to get that gamma one back to one.
Subrahmanyan: That brings us back to the bigger picture, how we predict the behavior of these AO systems when they are actually running on large telescopes.
Vera: They validated this whole thing against tools like OOPAO, which confirms it gives a high fidelity description of sensitivity and non-linearity in simulations.
Jocelyn: It means that for future instruments, like those massive ELT systems with pyramid AO, we have a much better way to estimate how well the system will actually work.
Subrahmanyan: The ability to model these complex interactions is important because it helps us connect observational data back to the fundamental physics of structure formation.
Vera: So, the AOPERA tool gives us a robust numerical implementation for handling those non-linear effects in Fourier filtering wavefront sensors.
Jocelyn: It’s a powerful way to get a better picture of real-world performance when things are complex and turbulent.
Subrahmanyan: Overall, this work provides the tools needed to predict the behavior of AO systems during both their design phase and their actual operation on instruments like PROVIDENCE two meter telescope.
Vera: It’s a solid framework for getting that accurate description we need for those demanding observing conditions.
Jocelyn: We'll keep an eye on how these PSD-based methods evolve as we look at the next set of papers in this field.
Subrahmanyan: Next time, we’ll be looking at how these modeling techniques apply to large-scale structure observations where atmospheric effects are just one part of the puzzle.
Romain JL Fétick, Vincent Chambouleyron, Lisa-Marie Mazzolo, Christophe Vérinaud, Arnaud Striffling, Ƨédric Taissir Héritier-Salama, Benoit Neichel, Jean-Fran߃ois Sauvage, Thierry Fusco
DOTA · ONERA · Aix Marseille University, CNRS, CNES · European Southern Observatory (ESO) · Ƨix Marseille Univ, CNRS, Pytheas, Observatoire de Haute-Provence
astro-ph.IM
Submitted: 2026-09-09
Updated: 2026-10-14
Journal ref: A&A 715, A295 (2026)
DOI: 10.1051/0004-6361/202661016
Code: https://github.com/astro-tiptop/TIPTOP
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 91/100
The gist: Modelling Fourier filtering wavefront sensors for PSD-based methods: the AOPERA tool develops a method to accurately describe and manage non-linear behavior in Fourier filtering wavefront sensors…
Key concepts
- Convolutive Formalism
- This mathematical framework is used to compute the overall response of an adaptive optics system. It combines the sensitivity of the wavefront sensor with non-linearity management, allowing for a more accurate description of how the system reacts to incoming light and atmospheric turbulence.
- Power Spectral Density (PSD) Computation
- The paper computes the spatial PSD by integrating over temporal frequencies. This helps separate different sources of residual phase error, such as spatio-temporal errors, measurement noise, and aliasing, providing a comprehensive view of the system's performance.
- Optical Gains Management
- Wavefront sensor sensitivity can be non-linear if it changes based on the measured phase. The method calculates optical gains (OG) to quantify this non-linearity and uses iterative updates during operation to compensate for these gains, aiming to restore a desired state where gamma equals 1.
Terminology
Summary
Modelling Fourier filtering wavefront sensors for PSD-based methods: the AOPERA tool develops a method to accurately describe and manage non-linear behavior in Fourier filtering wavefront sensors (FF-WFS) using a convolutive formalism, which is critical for understanding and predicting the performance of adaptive optics systems during design or operation The development of a PSD-based method accurately describing Fourier filtering WFS and justification of its validity is a critical technique for understanding the behaviour of AO systems during their design or operation phases
Methodological Framework
The paper develops a method based on the convolutive formalism to compute the AO system response, including electromagnetic phase power spectral density and point spread function, while also allowing for the inclusion of non-linear behaviour of the wavefront sensor within PSD-based numerical tools The mathematical formalism describes FF-WFS sensitivity combined with non-linearity management to greatly improve the accuracy of description of AO systems through numerical simulations The method involves splitting the residual phase equation into terms related to spatio-temporal error, noise, and aliasing based on a piecewise constant reconstruction domain F where D(f)R(f)M⊓(f) = γ (12)
Hypotheses for PSD Computation
The computation of the spatial PSD of the phase, noted W(f), is done by integrating over temporal frequencies The three terms of equation 6 are considered statistically uncorrelated to obtain the total residual PSD: Wres(f) = Wres,spatio-temporal(f) + Wres,noise(f) + Wres,aliasing(f) (8) The measurement noise is considered white in the temporal direction for photon noise and readout noise The frozen flow hypothesis assumes the atmosphere is a sum of independent turbulent layers, each sliding with a given wind speed and direction Thanks to this hypothesis, the spatiotemporal PSD simplifies to depend only on the atmospheric PSD and on the response of the AO loop evaluated at a specific temporal frequency ν = V · f (11)
Managing Non-Linearity and Gains
The sensitivity of a wavefront sensor is non-linear if its sensitivity evolves with the phase to be measured, leading to optical gains defined as OG = M⊓/Mcalib⊓ The method includes the current sensitivity evaluation through the PSF seen by the WFS, which is solved via an iterative method involving initialization, residual PSD computation, PSF computation, and updating the estimation of sensitivity The management of optical gains can be achieved by updating the reconstructor during AO loop operations to compensate for optical gains and eventually restore γ = 1
Validation and Application
The method is validated against end-to-end simulations using the OOPAO tool, showing similarity in fitting error, temporal error, and noise error with high fidelity The model and its numerical implementation have been applied to the PROVIDENCE 2.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Development of a PSD-based method accurately describing Fourier filtering WFS allows for precise estimation of sensitivity by including non-linear behavior, which can be used to
manage optical gains, and thus any value of gamma is the possibility offered by the present work.
-
The improved system can perform accurate AO performance estimations during design or operation phases because the model
has been applied to the PROVIDENCE 2.5 m telescope to be built in France by the end of 2028, and could be used for ELT instruments which include a pyramid AO system.
-
The tool can compute an
AO corrected PSF
by separating it into a diffraction static component and a phase residuals component, allowing for the calculation ofhres = e(-sigma 2 res F-1 n e F-1
to get the science PSF. -
The system can be used for optimal modulation abacus tuning by computing
the optimum of an AO system on the fly, during on-sky observations, following the evolution of the turbulence during the night.
-
The improved tool can validate against end-to-end simulations by comparing its results with tools like OOPAO, ensuring that it provides a
high fidelity
description of WFS sensitivity and non-linearity.
Abstract
Compensation of the atmospheric turbulence thanks to adaptive optics (AO) has now become commonly used for VLT or ELT class telescopes in order to retrieve a resolution close to their diffraction limit. Following the increasing trend of AO system, there is also a stronger necessity for simulations in order to understand and predict their performance facing different observing conditions, that are the evolution of the atmospheric turbulence or the diversity of AO guide source. The so-called PSD based methods are well adapted to the demand thanks to their simplicity and speed. Moreover, they provide a comprehensive breakdown error budget and impact on focal plane, that is of high interest especially in the case of extreme adaptive optics. Their drawback is the challenge to describe non-linear wavefront sensors (WFS) such as the pyramid WFS or the Zernike WFS. There is a necessity to develop fast yet accurate methods to describe the behaviour of AO systems including sensitive WFS. We thus develop a method to compute the AO system response (electromagnetic phase power spectral density, and point spread function) including the non-linear behaviour of the wavefront sensor within PSD-based numerical tools. Mathematical formalism to describe the FF-WFS sensitivity combined to non-linearity management greatly improve the accuracy of description of AO systems through numerical simulations. After a mathematical description of the method, its numerical implementation is compared with end-to-end simulations using the OOPAO tool. Indeed end-to-end simulations are reproducing the response of AO systems with high fidelity, especially regarding WFS sensitivity and non-linearity. Similarity of results for the fitting error, temporal error and noise error proves the validity of our method in managing Fourier filtering WFS sensitivity and non-linearity.
Related papers
- On-sky demonstration of a vector Zernike wavefront sensor in a cascaded adaptive optics system
- Euclid Data Release 1 (DR1): Calibration strategy and observations
- The General Antiparticle Spectrometer (GAPS) Antarctic Balloon Payload
- Detection and excision of radio-frequency interference in radio arrays
- Unified matched-filter and excess-power pipeline for gravitational-wave detector characterization
- Ushering in sub-arcsecond angular resolution observatories in the far-infrared: The case for a double-Fourier interferometer