Efficient computation of minimum-variance wave-front reconstructors with sparse matrix techniques

Enhanced-resolution Shack-Hartmann wavefront sensing for extended objects

Adaptive optics systems for large-aperture solar telescopes, especially multiconjugate adaptive optics systems, suffer from a fundamental trade-off between wavefront sampling rate and sub-aperture resolution. We introduce an enhanced-resolution Shack-Hartmann wavefront sensing method that decouples sub-aperture resolution from the desired wavefront sampling rate. We experimentally verified the validity of this method. Results show that by synthesizing multiple low-spatial samplings, this method is capable to sense higher-frequency aberrations beyond any low-spatial sampling involved in the synthesis, and it allows higher sub-aperture resolution and higher operating bandwidths, which can better fulfill the needs of solar adaptive optics.

Constrained deconvolution from wavefront sensing using the frozen flow hypothesis and complex wavelet regularization

Deconvolution from wavefront sensing (DWFS) is a high-performance image restoration technique designed to compensate for atmospheric turbulence-induced wavefront distortions. It uses simultaneously recorded short-exposure images of the object and high cadence wavefront sensor (WFS) data to estimate both the wavefronts and the object. Conventional DWFS takes no account of the temporal correlations in WFS data, which limits the reconstruction of high-spatial frequency components of wavefront distortion and then the recovery of the object. This paper takes the frozen flow hypothesis (FFH) to model the temporal evolution of atmospheric turbulence. Under this assumption, a joint estimation is performed in a Bayesian framework to simultaneously estimate the object and the turbulence phases with strict constraints imposed by WFS data and the FFH. It takes into account the temporal correlations in WFS data as well as the available a priori knowledge about the object and turbulence phases. Taking advantage of the sparse analysis prior of the object in the wavelet domain, a sparse regularization of the object based on the 2D dual-tree complex wavelet transform is proposed. Numerical experiments show that the proposed method is robust and effective for high-resolution image restoration in different seeing conditions.

Sparse data-driven wavefront prediction for large-scale adaptive optics

This paper presents a computationally efficient wavefront aberration prediction framework for data-driven control in large-scale adaptive optics systems. Our novel prediction algorithm splits prediction into two stages: a high-resolution and a low-resolution stage. For the former, we exploit sparsity structures in the system matrices in a data-driven Kalman filtering algorithm and constrain the identified gain to be likewise sparse; for the latter, we identify a dense Kalman gain and perform corrections to the suboptimal predictions of the former on a smaller grid. This novel prediction framework is able to retain the robustness to measurement noise of the standard Kalman filter in a much more computationally efficient manner, in both its offline and online aspects, while minimally sacrificing performance; its data-driven nature further compensates for modeling errors. As an intermediate result, we present a sparsity-exploiting data-driven Kalman filtering algorithm able to quickly estimate an approximate Kalman gain without solving the Riccati equation.

Phase retrieval of large-scale time-varying aberrations using a non-linear Kalman filtering framework

This paper presents a computationally efficient framework in which a single focal-plane image is used to obtain a high-resolution reconstruction of dynamic aberrations. Assuming small-phase aberrations, a non-linear Kalman filter implementation is developed whose computational complexity scales close to linearly with the number of pixels of the focal-plane camera. The performance of the method is tested in a simulation of an adaptive optics system, where the small-phase assumption is enforced by considering a closed-loop system that uses a low-resolution wavefront sensor to control a deformable mirror. The results confirm the computational efficiency of the algorithm and show a large robustness against noise and model uncertainties.

Nonlinear wavefront reconstruction methods for pyramid sensors using Landweber and Landweber-Kaczmarz iterations

Accurate and robust wavefront reconstruction methods for pyramid wavefront sensors are in high demand, as these sensors are planned to be part of many instruments currently under development for ground-based telescopes. The pyramid sensor relates the incoming wavefront and its measurements in a nonlinear way. Nevertheless, almost all existing reconstruction algorithms are based on a linearization of the model. The assumption of a linear pyramid sensor response is justifiable in closed-loop adaptive optics (AO) when the measured phase information is small, but, depending on the system, may not be feasible due to unpreventable errors such as non-common path aberrations. In order to solve the nonlinear inverse problem of wavefront reconstruction from pyramid sensor data, we introduce two new methods based on the nonlinear Landweber and Landweber-Kaczmarz iterations. Using these algorithms, we experience high-quality wavefront estimation, especially for the non-modulated sensor, by still keeping the numerical effort feasible for large-scale AO systems.

Tensor-based predictive control for extremely large-scale single conjugate adaptive optics

In this paper we propose a data-driven predictive control algorithm for large-scale single conjugate adaptive optics systems. At each time sample, the Shack-Hartmann wavefront sensor signal sampled on a spatial grid of size N×N is reshuffled into a d-dimensional tensor. Its spatial-temporal dynamics are modeled with a d-dimensional autoregressive model of temporal order p, where each tensor storing past data undergoes a multilinear transformation by factor matrices of small sizes. Equivalently, the vector form of this autoregressive model features coefficient matrices parametrized with a sum of Kronecker products between d-factor matrices. We propose an Alternating Least Squares algorithm for identifying the factor matrices from open-loop sensor data. When modeling each coefficient matrix with a sum of r terms, the computational complexity for updating the sensor prediction online reduces from O(pN⁴) in the unstructured matrix case to O(prd N^2(d+1)d). Most importantly, this model structure breaks away from assuming any prior spatial-temporal coupling as it is discovered from the data. The algorithm is validated on a laboratory testbed that demonstrates the ability to accurately decompose the coefficient matrices of large-scale autoregressive models with a tensor-based representation, hence achieving high data compression rates and reducing the temporal error especially for a large Greenwood per sample frequency ratio.

Fast iterative tomographic wavefront estimation with recursive Toeplitz reconstructor structure for large-scale systems

Tomographic wavefront reconstruction is the main computational bottleneck to realize real-time correction for turbulence-induced wavefront aberrations in future laser-assisted tomographic adaptive-optics (AO) systems for ground-based giant segmented mirror telescopes because of its unprecedented number of degrees of freedom, N, i.e., the number of measurements from wavefront sensors. In this paper, we provide an efficient implementation of the minimum-mean-square error (MMSE) tomographic wavefront reconstruction, which is mainly useful for some classes of AO systems not requiring multi-conjugation, such as laser-tomographic AO, multi-object AO, and ground-layer AO systems, but is also applicable to multi-conjugate AO systems. This work expands that by Conan [Proc. SPIE9148, 91480R (2014)PSISDG0277-786X10.1117/12.2054472] to the multi-wavefront tomographic case using natural and laser guide stars. The new implementation exploits the Toeplitz structure of covariance matrices used in an MMSE reconstructor, which leads to an overall O(N log N) real-time complexity compared with O(N²) of the original implementation using straight vector-matrix multiplication. We show that the Toeplitz-based algorithm leads to 60 nm rms wavefront error improvement for the European Extremely Large Telescope laser-tomography AO system over a well-known sparse-based tomographic reconstruction; however, the number of iterations required for suitable performance is still beyond what a real-time system can accommodate to keep up with the time-varying turbulence.

GPU implementation for spline-based wavefront reconstruction

This paper presents an adaptation of the distributed-spline-based aberration reconstruction method for Shack-Hartmann (SH) slope measurements to extremely large-scale adaptive optics systems and the execution on graphics processing units (GPUs). The introduction of a hierarchical multi-level scheme for the elimination of piston offsets between the locally computed wavefront (WF) estimates solves the piston error propagation observed for a large number of partitions with the original version. To obtain a fully distributed method for WF correction, the projection of the phase estimates is locally approximated and applied in a distributed fashion, providing stable results for low and medium actuator coupling. An implementation of the method with the parallel computing platform CUDA exploits the inherently distributed nature of the algorithm. With a standard off-the-shelf GPU, the computation of the adaptive optics correction updates is accomplished in under 1 ms for the benchmark case of a 200×200 SH array.

Framework to trade optimality for local processing in large-scale wavefront reconstruction problems

We show that the minimum variance wavefront estimation problems permit localized approximate solutions, in the sense that the wavefront value at a point (excluding unobservable modes, such as the piston mode) can be approximated by a linear combination of the wavefront slope measurements in the point's neighborhood. This enables us to efficiently compute a wavefront estimate by performing a single sparse matrix-vector multiplication. Moreover, our results open the possibility for the development of wavefront estimators that can be easily implemented in a decentralized/distributed manner, and in which the estimate optimality can be easily traded for computational efficiency. We numerically validate our approach on Hudgin wavefront sensor geometries, and the results can be easily generalized to Fried geometries.

Multi time-step wavefront reconstruction for tomographic adaptive-optics systems

In tomographic adaptive-optics (AO) systems, errors due to tomographic wavefront reconstruction limit the performance and angular size of the scientific field of view (FoV), where AO correction is effective. We propose a multi time-step tomographic wavefront reconstruction method to reduce the tomographic error by using measurements from both the current and previous time steps simultaneously. We further outline the method to feed the reconstructor with both wind speed and direction of each turbulence layer. An end-to-end numerical simulation, assuming a multi-object AO (MOAO) system on a 30 m aperture telescope, shows that the multi time-step reconstruction increases the Strehl ratio (SR) over a scientific FoV of 10 arc min in diameter by a factor of 1.5-1.8 when compared to the classical tomographic reconstructor, depending on the guide star asterism and with perfect knowledge of wind speeds and directions. We also evaluate the multi time-step reconstruction method and the wind estimation method on the RAVEN demonstrator under laboratory setting conditions. The wind speeds and directions at multiple atmospheric layers are measured successfully in the laboratory experiment by our wind estimation method with errors below 2  ms^-1. With these wind estimates, the multi time-step reconstructor increases the SR value by a factor of 1.2-1.5, which is consistent with a prediction from the end-to-end numerical simulation.

Performance of the Gemini Planet Imager's adaptive optics system

The Gemini Planet Imager's adaptive optics (AO) subsystem was designed specifically to facilitate high-contrast imaging. A definitive description of the system's algorithms and technologies as built is given. 564 AO telemetry measurements from the Gemini Planet Imager Exoplanet Survey campaign are analyzed. The modal gain optimizer tracks changes in atmospheric conditions. Science observations show that image quality can be improved with the use of both the spatially filtered wavefront sensor and linear-quadratic-Gaussian control of vibration. The error budget indicates that for all targets and atmospheric conditions AO bandwidth error is the largest term.

Spatio-angular minimum-variance tomographic controller for multi-object adaptive-optics systems

Multi-object astronomical adaptive optics (MOAO) is now a mature wide-field observation mode to enlarge the adaptive-optics-corrected field in a few specific locations over tens of arcminutes. The work-scope provided by open-loop tomography and pupil conjugation is amenable to a spatio-angular linear-quadratic-Gaussian (SA-LQG) formulation aiming to provide enhanced correction across the field with improved performance over static reconstruction methods and less stringent computational complexity scaling laws. Starting from our previous work [J. Opt. Soc. Am. A31, 101 (2014)10.1364/JOSAA.31.000101JOAOD61084-7529], we use stochastic time-progression models coupled to approximate sparse measurement operators to outline a suitable SA-LQG formulation capable of delivering near optimal correction. Under the spatio-angular framework the wavefronts are never explicitly estimated in the volume, providing considerable computational savings on 10-m-class telescopes and beyond. We find that for Raven, a 10-m-class MOAO system with two science channels, the SA-LQG improves the limiting magnitude by two stellar magnitudes when both the Strehl ratio and the ensquared energy are used as figures of merit. The sky coverage is therefore improved by a factor of ~5.

Anti-aliasing Wiener filtering for wave-front reconstruction in the spatial-frequency domain for high-order astronomical adaptive-optics systems

Computationally efficient wave-front reconstruction techniques for astronomical adaptive-optics (AO) systems have seen great development in the past decade. Algorithms developed in the spatial-frequency (Fourier) domain have gathered much attention, especially for high-contrast imaging systems. In this paper we present the Wiener filter (resulting in the maximization of the Strehl ratio) and further develop formulae for the anti-aliasing (AA) Wiener filter that optimally takes into account high-order wave-front terms folded in-band during the sensing (i.e., discrete sampling) process. We employ a continuous spatial-frequency representation for the forward measurement operators and derive the Wiener filter when aliasing is explicitly taken into account. We further investigate and compare to classical estimates using least-squares filters the reconstructed wave-front, measurement noise, and aliasing propagation coefficients as a function of the system order. Regarding high-contrast systems, we provide achievable performance results as a function of an ensemble of forward models for the Shack-Hartmann wave-front sensor (using sparse and nonsparse representations) and compute point-spread-function raw intensities. We find that for a 32×32 single-conjugated AOs system the aliasing propagation coefficient is roughly 60% of the least-squares filters, whereas the noise propagation is around 80%. Contrast improvements of factors of up to 2 are achievable across the field in the H band. For current and next-generation high-contrast imagers, despite better aliasing mitigation, AA Wiener filtering cannot be used as a standalone method and must therefore be used in combination with optical spatial filters deployed before image formation actually takes place.

Sub-nanometer flattening of 45 cm long, 45 actuator x-ray deformable mirror

We have built a 45 cm long x-ray deformable mirror (XDM) of super-polished single-crystal silicon that has 45 actuators along the tangential axis. After assembly, the surface height error was 19 nm rms. With use of high-precision visible-light metrology and precise control algorithms, we have actuated the XDM and flattened its entire surface to 0.7 nm rms controllable figure error. This is, to our knowledge, the first sub-nanometer active flattening of a substrate longer than 15 cm.

Finite element-wavelet hybrid algorithm for atmospheric tomography

Reconstruction of the refractive index fluctuations in the atmosphere, or atmospheric tomography, is an underlying problem of many next generation adaptive optics (AO) systems, such as the multiconjugate adaptive optics or multiobject adaptive optics (MOAO). The dimension of the problem for the extremely large telescopes, such as the European Extremely Large Telescope (E-ELT), suggests the use of iterative schemes as an alternative to the matrix-vector multiply (MVM) methods. Recently, an algorithm based on the wavelet representation of the turbulence has been introduced in [Inverse Probl.29, 085003 (2013)] by the authors to solve the atmospheric tomography using the conjugate gradient iteration. The authors also developed an efficient frequency-dependent preconditioner for the wavelet method in a later work. In this paper we study the computational aspects of the wavelet algorithm. We introduce three new techniques, the dual domain discretization strategy, a scale-dependent preconditioner, and a ground layer multiscale method, to derive a method that is globally O(n), parallelizable, and compact with respect to memory. We present the computational cost estimates and compare the theoretical numerical performance of the resulting finite element-wavelet hybrid algorithm with the MVM. The quality of the method is evaluated in terms of an MOAO simulation for the E-ELT on the European Southern Observatory (ESO) end-to-end simulation system OCTOPUS. The method is compared to the ESO version of the Fractal Iterative Method [Proc. SPIE7736, 77360X (2010)] in terms of quality.

Static and predictive tomographic reconstruction for wide-field multi-object adaptive optics systems

Multi-object adaptive optics (MOAO) systems are still in their infancy: their complex optical designs for tomographic, wide-field wavefront sensing, coupled with open-loop (OL) correction, make their calibration a challenge. The correction of a discrete number of specific directions in the field allows for streamlined application of a general class of spatio-angular algorithms, initially proposed in Whiteley et al. [J. Opt. Soc. Am. A15, 2097 (1998)], which is compatible with partial on-line calibration. The recent Learn & Apply algorithm from Vidal et al. [J. Opt. Soc. Am. A27, A253 (2010)] can then be reinterpreted in a broader framework of tomographic algorithms and is shown to be a special case that exploits the particulars of OL and aperture-plane phase conjugation. An extension to embed a temporal prediction step to tackle sky-coverage limitations is discussed. The trade-off between lengthening the camera integration period, therefore increasing system lag error, and the resulting improvement in SNR can be shifted to higher guide-star magnitudes by introducing temporal prediction. The derivation of the optimal predictor and a comparison to suboptimal autoregressive models is provided using temporal structure functions. It is shown using end-to-end simulations of Raven, the MOAO science, and technology demonstrator for the 8 m Subaru telescope that prediction allows by itself the use of 1-magnitude-fainter guide stars.

Improving wavefront reconstruction accuracy by using integration equations with higher-order truncation errors in the Southwell geometry

Least-squares (LS)-based integration computes the function values by solving a set of integration equations (IEs) in LS sense, and is widely used in wavefront reconstruction and other fields where the measured data forms a slope. It is considered that the applications of IEs with smaller truncation errors (TEs) will improve the reconstruction accuracy. This paper proposes a general method based on the Taylor theorem to derive all kinds of IEs, and finds that an IE with a smaller TE has a higher-order TE. Three specific IEs with higher-order TEs in the Southwell geometry are deduced using this method, and three LS-based integration algorithms corresponding to these three IEs are formulated. A series of simulations demonstrate the validity of applying IEs with higher-order TEs in improving reconstruction accuracy. In addition, the IEs with higher-order TEs in the Hudgin and Fried geometries are also deduced using the proposed method, and the performances of these IEs in wavefront reconstruction are presented.

Distributed Kalman filtering compared to Fourier domain preconditioned conjugate gradient for laser guide star tomography on extremely large telescopes

This paper discusses the performance and cost of two computationally efficient Fourier-based tomographic wavefront reconstruction algorithms for wide-field laser guide star (LGS) adaptive optics (AO). The first algorithm is the iterative Fourier domain preconditioned conjugate gradient (FDPCG) algorithm developed by Yang et al. [Appl. Opt.45, 5281 (2006)], combined with pseudo-open-loop control (POLC). FDPCG's computational cost is proportional to N log(N), where N denotes the dimensionality of the tomography problem. The second algorithm is the distributed Kalman filter (DKF) developed by Massioni et al. [J. Opt. Soc. Am. A28, 2298 (2011)], which is a noniterative spatially invariant controller. When implemented in the Fourier domain, DKF's cost is also proportional to N log(N). Both algorithms are capable of estimating spatial frequency components of the residual phase beyond the wavefront sensor (WFS) cutoff frequency thanks to regularization, thereby reducing WFS spatial aliasing at the expense of more computations. We present performance and cost analyses for the LGS multiconjugate AO system under design for the Thirty Meter Telescope, as well as DKF's sensitivity to uncertainties in wind profile prior information. We found that, provided the wind profile is known to better than 10% wind speed accuracy and 20 deg wind direction accuracy, DKF, despite its spatial invariance assumptions, delivers a significantly reduced wavefront error compared to the static FDPCG minimum variance estimator combined with POLC. Due to its nonsequential nature and high degree of parallelism, DKF is particularly well suited for real-time implementation on inexpensive off-the-shelf graphics processing units.

Wavefront reconstruction in adaptive optics systems using nonlinear multivariate splines

This paper presents a new method for zonal wavefront reconstruction (WFR) with application to adaptive optics systems. This new method, indicated as Spline based ABerration REconstruction (SABRE), uses bivariate simplex B-spline basis functions to reconstruct the wavefront using local wavefront slope measurements. The SABRE enables WFR on nonrectangular and partly obscured sensor grids and is not subject to the waffle mode. The performance of SABRE is compared to that of the finite difference (FD) method in numerical experiments using data from a simulated Shack-Hartmann lenslet array. The results show that SABRE offers superior reconstruction accuracy and noise rejection capabilities compared to the FD method.

Sky coverage modeling for the whole sky for laser guide star multiconjugate adaptive optics

The scientific productivity of laser guide star adaptive optics systems strongly depends on the sky coverage, which describes the probability of finding natural guide stars for the tip/tilt wavefront sensor(s) to achieve a certain performance. Knowledge of the sky coverage is also important for astronomers planning their observations. In this paper, we present an efficient method to compute the sky coverage for the laser guide star multiconjugate adaptive optics system, the Narrow Field Infrared Adaptive Optics System (NFIRAOS), being designed for the Thirty Meter Telescope project. We show that NFIRAOS can achieve more than 70% sky coverage over most of the accessible sky with the requirement of 191 nm total rms wavefront.

Computational performance comparison of wavefront reconstruction algorithms for the European Extremely Large Telescope on multi-CPU architecture

The European Southern Observatory (ESO) is studying the next generation giant telescope, called the European Extremely Large Telescope (E-ELT). With a 42 m diameter primary mirror, it is a significant step from currently existing telescopes. Therefore, the E-ELT with its instruments poses new challenges in terms of cost and computational complexity for the control system, including its adaptive optics (AO). Since the conventional matrix-vector multiplication (MVM) method successfully used so far for AO wavefront reconstruction cannot be efficiently scaled to the size of the AO systems on the E-ELT, faster algorithms are needed. Among those recently developed wavefront reconstruction algorithms, three are studied in this paper from the point of view of design, implementation, and absolute speed on three multicore multi-CPU platforms. We focus on a single-conjugate AO system for the E-ELT. The algorithms are the MVM, the Fourier transform reconstructor (FTR), and the fractal iterative method (FRiM). This study enhances the scaling of these algorithms with an increasing number of CPUs involved in the computation. We discuss implementation strategies, depending on various CPU architecture constraints, and we present the first quantitative execution times so far at the E-ELT scale. MVM suffers from a large computational burden, making the current computing platform undersized to reach timings short enough for AO wavefront reconstruction. In our study, the FTR provides currently the fastest reconstruction. FRiM is a recently developed algorithm, and several strategies are investigated and presented here in order to implement it for real-time AO wavefront reconstruction, and to optimize its execution time. The difficulty to parallelize the algorithm in such architecture is enhanced. We also show that FRiM can provide interesting scalability using a sparse matrix approach.

Cumulative Reconstructor: fast wavefront reconstruction algorithm for Extremely Large Telescopes

The Cumulative Reconstructor (CuRe) is a new direct reconstructor for an optical wavefront from Shack-Hartmann wavefront sensor measurements. In this paper, the algorithm is adapted to realistic telescope geometries and the transition from modified Hudgin to Fried geometry is discussed. After a discussion of the noise propagation, we analyze the complexity of the algorithm. Our numerical tests confirm that the algorithm is very fast and accurate and can therefore be used for adaptive optics systems of Extremely Large Telescopes.

Analysis of the improvement in sky coverage for multiconjugate adaptive optics systems obtained using minimum variance split tomography

The scientific utility of laser-guide-star-based multiconjugate adaptive optics systems depends upon high sky coverage. Previously we reported a high-fidelity sky coverage analysis of an ad hoc split tomography control algorithm and a postprocessing simulation technique. In this paper, we present the performance of a newer minimum variance split tomography algorithm, and we show that it brings a median improvement at zenith of 21 nm rms optical path difference error over the ad hoc split tomography control algorithm for our system, the Narrow Field Infrared Adaptive Optics System for the Thirty Meter Telescope. In order to make the comparison, we also validated our previously developed sky coverage postprocessing software using an integrated simulation of both high- (laser guide star) and low-order (natural guide star) loops. A new term in the noise model is also identified that improves the performance of both algorithms by more properly regularizing the reconstructor.

System design considerations to improve isoplanatism for adaptive optics retinal imaging

Adaptive optics (AO) retinal images are limited by anisoplanatism; wavefront shape varies across the field of view such that only a limited area can achieve diffraction-limited image quality at one time. We explored three alternative AO modalities designed to reduce this effect, drawn from work in astronomy. Optical design analysis and computer modeling was undertaken to predict the benefit of each modality for various schematic eyes and various complexities of the imaging system. Off-axis performance was found to be limited by system parameters and not by the eye itself, due to the inherent off-axis characteristics of the eye's gradient index lens. This rendered the alternative AO modalities ineffectual compared with conventional AO but did suggest several methods by which anisoplanatism may be reduced by altering the design of conventional AO systems. Several of these design possibilities were explored with further modeling. The best-performing method involved the replacement of system lenses with gradient index versions inspired by the human eye lens. Mirror-based relay optics also demonstrated good off-axis performance, but their advantage was lost in regions of the system suffering from uncorrected higher-order aberration. Incorporating "off-the-plane" beam deviations ameliorated this loss substantially. In this work we also show, to our knowledge for the first time, that the ideal location of a single AO corrector need not lie in the pupil plane.

Closed-loop control of a woofer-tweeter adaptive optics system using wavelet-based phase reconstruction

A novel closed-loop control technique for adaptive optics (AO) systems based on a wavelet-based phase reconstruction technique and a woofer-tweeter controller is presented. The wavelet-based reconstruction technique is based on obtaining a Haar decomposition of the phase screen directly from gradient measurements and has been extended here with the use of a Poisson solver to improve performance. This method is O(N) (i.e., a linear computation cost as number of actuators increases) and is the fastest of the known O(N) reconstruction techniques. The controller configuration is based on the woofer-tweeter controller to control low- and high-spatial-frequency aberrations, respectively. The separation of the woofer and tweeter signals is done using a computationally efficient method that is based on the availability of a low-spatial-resolution reconstruction during the wavelet synthesis process. The performance of the proposed technique is evaluated using a simulated AO system and phase screens generated to reflect atmospheric turbulence with various dynamic characteristics. Results indicate that the combination of the wavelet-based phase reconstruction and woofer-tweeter controller leads to very good results with respect to speed and accuracy.

Fast reconstruction and prediction of frozen flow turbulence based on structured Kalman filtering

Efficient and optimal prediction of frozen flow turbulence using the complete observation history of the wavefront sensor is an important issue in adaptive optics for large ground-based telescopes. At least for the sake of error budgeting and algorithm performance, the evaluation of an accurate estimate of the optimal performance of a particular adaptive optics configuration is important. However, due to the large number of grid points, high sampling rates, and the non-rationality of the turbulence power spectral density, the computational complexity of the optimal predictor is huge. This paper shows how a structure in the frozen flow propagation can be exploited to obtain a state-space innovation model with a particular sparsity structure. This sparsity structure enables one to efficiently compute a structured Kalman filter. By simulation it is shown that the performance can be improved and the computational complexity can be reduced in comparison with auto-regressive predictors of low order.

Intersubject variability of foveal cone photoreceptor density in relation to eye length

PURPOSE

Adaptive optics scanning laser ophthalmoscopy (AOSLO) under optimized wavefront correction allows for routine imaging of foveal cone photoreceptors. The intersubject variability of foveal cone density was measured and its relation to eye length evaluated.

METHODS

AOSLO was used to image 18 healthy eyes with axial lengths from 22.86 to 28.31 mm. Ocular biometry and an eye model were used to estimate the retinal magnification factor. Individual cones in the AOSLO images were labeled, and the locations were used to generate topographic maps representing the spatial distribution of density. Representative retinal (cones/mm(2)) and angular (cones/deg(2)) cone densities at specific eccentricities were calculated from these maps.

RESULTS

The entire foveal cone mosaic was resolved in four eyes, whereas the cones within 0.03 mm eccentricity remained unresolved in most eyes. The preferred retinal locus deviated significantly (P < 0.001) from the point of peak cone density for all except one individual. A significant decrease in retinal density (P < 0.05) with increasing axial length was observed at 0.30 mm eccentricity but not closer. Longer, more myopic eyes generally had higher angular density near the foveal center than the shorter eyes, but by 1°, this difference was nullified by retinal expansion, and so angular densities across all eyes were similar.

CONCLUSIONS

The AOSLO can resolve the smallest foveal cones in certain eyes. Although myopia causes retinal stretching in the fovea, its effect within the foveola is confounded by factors other than cone density that have high levels of intersubject variability.

Fast minimum variance wavefront reconstruction for extremely large telescopes

We present what we believe to be a new algorithm, FRactal Iterative Method (FRiM), aiming at the reconstruction of the optical wavefront from measurements provided by a wavefront sensor. As our application is adaptive optics on extremely large telescopes, our algorithm was designed with speed and best quality in mind. The latter is achieved thanks to a regularization that enforces prior statistics. To solve the regularized problem, we use the conjugate gradient method, which takes advantage of the sparsity of the wavefront sensor model matrix and avoids the storage and inversion of a huge matrix. The prior covariance matrix is, however, non-sparse, and we derive a fractal approximation to the Karhunen-Loève basis thanks to which the regularization by Kolmogorov statistics can be computed in O(N) operations, with N being the number of phase samples to estimate. Finally, we propose an effective preconditioning that also scales as O(N) and yields the solution in five to ten conjugate gradient iterations for any N. The resulting algorithm is therefore O(N). As an example, for a 128 x 128 Shack-Hartmann wavefront sensor, the FRiM appears to be more than 100 times faster than the classical vector-matrix multiplication method.

High fidelity sky coverage analysis via time domain adaptive optics simulations

We describe a high fidelity simulation method for estimating the sky coverage of multiconjugate adaptive optics systems; this method is based upon the split tomography control architecture, and employs an AO simulation postprocessing technique to evaluate system performance with hundreds of randomly generated natural guide star (NGS) asterisms. A novel technique to model the impact of quadratic wavefront aberrations upon the NGS point spread functions is described; this is used to model the variations in system performance with different asterisms, and is crucial for obtaining accurate results with the postprocessing technique. Several design and algorithm improvements help to reduce the residual wavefront error in the tip/tilt and plate scale modes that are controlled using the NGS asterism. These improvements include choosing the right wavefront sensor (WFS) pixel size, optimal pixel weights, and type II control of the plate scale modes.

Wavefront reconstruction over a circular aperture using gradient data extrapolated via the mirror equations

Methods for extrapolating gradient data outside a circular aperture from measurements obtained within a circular aperture are presented. The proposed methods are required to be computationally efficient and to avoid the excitation of additional waffle modes in Fried alignment. It is shown that, using an octagon as an intermediate step from the circle to the square in the extrapolation process, the computations or residual reconstruction error can be reduced. The resulting computational cost is as low as O(N(1/2)), where N is the number of measurement points. The performances of the extrapolation methods are studied in connection with a recently developed O(N) wavefront reconstruction algorithm based on wavelet filter banks [IEEE J. Sel. Top. Signal Process. 2, 781 (2008)] Experiments indicate that, as expected, there is a significant reconstruction error if no extrapolation is used. Further, the proposed extrapolation techniques lead to a reconstruction with data that are marginally different from a pupil masked reconstruction using data from a square aperture.

Comparison of minimum-norm maximum likelihood and maximum a posteriori wavefront reconstructions for large adaptive optics systems

The performances of various estimators for wavefront sensing applications such as adaptive optics (AO) are compared. Analytical expressions for the bias and variance terms in the mean squared error (MSE) are derived for the minimum-norm maximum likelihood (MNML) and the maximum a posteriori (MAP) reconstructors. The MAP estimator is analytically demonstrated to yield an optimal trade-off that reduces the MSE, hence leading to a better Strehl ratio. The implications for AO applications are quantified thanks to simulations on 8-m- and 42-m-class telescopes. We show that the MAP estimator can achieve twice as low MSE as MNML methods do. Large AO systems can thus benefit from the high quality of MAP reconstruction in O(n) operations, thanks to the fast fractal iterative method (FrIM) algorithm (Thiébaut and Tallon, submitted to J. Opt. Soc. Am. A).

Comparison of Control Algorithms for a MEMS-based Adaptive Optics Scanning Laser Ophthalmoscope

We compared four algorithms for controlling a MEMS deformable mirror of an adaptive optics (AO) scanning laser ophthalmoscope. Interferometer measurements of the static nonlinear response of the deformable mirror were used to form an equivalent linear model of the AO system so that the classic integrator plus wavefront reconstructor type controller can be implemented. The algorithms differ only in the design of the wavefront reconstructor. The comparisons were made for two eyes (two individuals) via a series of imaging sessions. All four controllers performed similarly according to estimated residual wavefront error not reflecting the actual image quality observed. A metric based on mean image intensity did consistently reflect the qualitative observations of retinal image quality. Based on this metric, the controller most effective for suppressing the least significant modes of the deformable mirror performed the best.

Split atmospheric tomography using laser and natural guide stars

Laser guide star (LGS) atmospheric tomography is described in the literature as integrated minimum-variance tomographic wavefront reconstruction from a concatenated wavefront-sensor measurement vector consisting of many high-order, tip/tilt (TT)-removed LGS measurements, supplemented by a few low-order natural guide star (NGS) components essential to estimating the TT and tilt anisoplanatism (TA) modes undetectable by the TT-removed LGS wavefront sensors (WFSs). The practical integration of these NGS WFS measurements into the tomography problem is the main subject of this paper. A split control architecture implementing two separate control loops driven independently by closed-loop LGS and NGS measurements is proposed in this context. Its performance is evaluated in extensive wave optics Monte Carlo simulations for the Thirty Meter Telescope (TMT) LGS multiconjugate adaptive optics (MCAO) system, against the delivered performance of the integrated control architecture. Three iterative algorithms are analyzed for atmospheric tomography in both cases: a previously proposed Fourier domain preconditioned conjugate gradient (FDPCG) algorithm, a simple conjugate gradient (CG) algorithm without preconditioning, and a novel layer-oriented block Gauss-Seidel conjugate gradient algorithm (BGS-CG). Provided that enough iterations are performed, all three algorithms yield essentially identical closed-loop residual RMS wavefront errors for both control architectures, with the caveat that a somewhat smaller number of iterations are required by the CG and BGS-CG algorithms for the split approach. These results demonstrate that the split control approach benefits from (i) a simpler formulation of minimum-variance atmospheric tomography allowing for algorithms with reduced computational complexity and cost (processing requirements), (ii) a simpler, more flexible control of the NGS-controlled modes, and (iii) a reduced coupling between the LGS- and NGS-controlled modes. Computation and memory requirements for all three algorithms are also given for the split control approach for the TMT LGS AO system and appear feasible in relation to the performance specifications of current hardware technology.

Warm-started wavefront reconstruction for adaptive optics

Future extreme adaptive optics (ExAO) systems have been suggested with up to 10(5) sensors and actuators. We analyze the computational speed of iterative reconstruction algorithms for such large systems. We compare a total of 15 different scalable methods, including multigrid, preconditioned conjugate-gradient, and several new variants of these. Simulations on a 128x128 square sensor/actuator geometry using Taylor frozen-flow dynamics are carried out using both open-loop and closed-loop measurements, and algorithms are compared on a basis of the mean squared error and floating-point multiplications required. We also investigate the use of warm starting, where the most recent estimate is used to initialize the iterative scheme. In open-loop estimation or pseudo-open-loop control, warm starting provides a significant computational speedup; almost every algorithm tested converges in one iteration. In a standard closed-loop implementation, using a single iteration per time step, most algorithms give the minimum error even in cold start, and every algorithm gives the minimum error if warm started. The best algorithm is therefore the one with the smallest computational cost per iteration, not necessarily the one with the best quasi-static performance.

Performance study of Kalman filter controller for multiconjugate adaptive optics

We compare the performance of the Kalman filter (KF)-based and the minimum variance (MV) control algorithms for a zonal adaptive optics with a phase temporal prediction step included for effective compensation of the errors attributable to latencies in the system. The main goal is to evaluate the performance achievable by the computationally more expensive KF approach, which explicitly accounts for the atmospheric turbulence temporal behavior through a first-order autoregressive evolution model, and the simpler MV algorithm, with and without temporal prediction. For a representative example, the Gemini-South 8 m telescope multiconjugate adaptive optics system performance of the KF and the MV controllers has been compared with respect to their turbulence compensation capability. We show that the KF algorithm, as expected, shows superior performance to that of the MV algorithm, especially for extremely low sampling rates and large control latencies. We also show that for moderate control latencies the MV algorithm with a temporal prediction step added to it approaches the performance of the KF technique.

Correctability limitations imposed by plane-wave scintillation in multiconjugate adaptive optics

Plane-wave scintillation is shown to impose multiconjugate adaptive optics (MCAO) correctability limitations that are independent of wavefront sensing and reconstruction. Residual phase and log-amplitude variances induced by scintillation in weak turbulence are derived using linear (diffraction-based) diffractive MCAO spatial filters or (diffraction-ignorant) geometric MCAO proportional gains as open-loop control parameters. In the case of Kolmogorov turbulence, expressions involving the Rytov variance and/or weighted C(2)(n) integrals apply. Differences in performance between diffractive MCAO and geometric MCAO resemble chromatic errors. Optimal corrections based on least squares imply irreducible performance limits that are validated by wave-optic simulations.

Fourier domain preconditioned conjugate gradient algorithm for atmospheric tomography

By 'atmospheric tomography' we mean the estimation of a layered atmospheric turbulence profile from measurements of the pupil-plane phase (or phase gradients) corresponding to several different guide star directions. We introduce what we believe to be a new Fourier domain preconditioned conjugate gradient (FD-PCG) algorithm for atmospheric tomography, and we compare its performance against an existing multigrid preconditioned conjugate gradient (MG-PCG) approach. Numerical results indicate that on conventional serial computers, FD-PCG is as accurate and robust as MG-PCG, but it is from one to two orders of magnitude faster for atmospheric tomography on 30 m class telescopes. Simulations are carried out for both natural guide stars and for a combination of finite-altitude laser guide stars and natural guide stars to resolve tip-tilt uncertainty.

Wavefront reconstruction from its gradients

Wavefronts reconstructed from measured gradients are composed of a straightforward integration of the measured data, plus a correction term that disappears when there are no measurement errors. For regions of any shape, this term is a solution of Poisson's equation with Dirichlet conditions (V = 0 on the boundaries). We show that for rectangular regions, the correct solution is not a periodic one, but one expressed with Fourier cosine series. The correct solution has a lower variance than the periodic Fourier transform solution. Similar formulas exist for a circular region with obscuration. We present a near-optimal solution that is much faster than fast-Fourier-transform methods. By use of diagonal multigrid methods, a single iteration brings the correction term to within a standard deviation of 0.08, two iterations, to within 0.0064, etc.

Multigrid algorithm for least-squares wavefront reconstruction

Multigrid (MG) methods are presented for fast, efficient, flexible, and robust least-squares wavefront reconstruction in extremely high-resolution conventional adaptive optics, or ExAO. We demonstrate that MG can robustly handle a variety of sensor-actuator geometries, and it can accommodate deformable mirror influence function models that are more realistic than the common piecewise bilinear model. With MG one can also easily incorporate additional penalty, or regularization, terms to damp out the waffle mode in Fried geometry and to damp out instabilities due to actuators near the pupil boundary with poorly sensed influence. We present closed-loop simulation results that suggest that only one or two MG iterations per time step are needed to control an ExAO system.

Optimal modal fourier-transform wavefront control

Optimal modal Fourier-transform wavefront control combines the speed of Fourier-transform reconstruction (FTR) with real-time optimization of modal gains to form a fast, adaptive wavefront control scheme. Our modal basis is the real Fourier basis, which allows direct control of specific regions of the point-spread function. We formulate FTR as modal control and show how to measure custom filters. Because the Fourier basis is a tight frame, we can use it on a circular aperture for modal control even though it is not an orthonormal basis. The modal coefficients are available during reconstruction, greatly reducing computational overhead for gain optimization. Simulation results show significant improvements in performance in low-signal-to-noise-ratio situations compared with nonadaptive control. This scheme is computationally efficient enough to be implemented with off-the-shelf technology for a 2.5 kHz, 64 x 64 adaptive optics system.

Large-scale wave-front reconstruction for adaptive optics systems by use of a recursive filtering algorithm

We propose a new recursive filtering algorithm for wave-front reconstruction in a large-scale adaptive optics system. An embedding step is used in this recursive filtering algorithm to permit fast methods to be used for wave-front reconstruction on an annular aperture. This embedding step can be used alone with a direct residual error updating procedure or used with the preconditioned conjugate-gradient method as a preconditioning step. We derive the Hudgin and Fried filters for spectral-domain filtering, using the eigenvalue decomposition method. Using Monte Carlo simulations, we compare the performance of discrete Fourier transform domain filtering, discrete cosine transform domain filtering, multigrid, and alternative-direction-implicit methods in the embedding step of the recursive filtering algorithm. We also simulate the performance of this recursive filtering in a closed-loop adaptive optics system.

Closed-loop stability and performance analysis of least-squares and minimum-variance control algorithms for multiconjugate adaptive optics

Recent progress has been made to compute efficiently the open-loop minimum-variance reconstructor (MVR) for multiconjugate adaptive optics systems by a combination of sparse matrix and iterative techniques. Using spectral analysis, I show that a closed-loop laser guide star multiconjugate adaptive optics control algorithm consisting of MVR cascaded with an integrator control law is unstable. Tosolve this problem, a computationally efficient pseudo-open-loop control (POLC) method was recently proposed. I give a theoretical proof of the stability of this method and demonstrate its superior performance and robustness against misregistration errors compared with conventional least-squares control. This can be accounted for by the fact that POLC incorporates turbulence statistics through its regularization term that can be interpreted as spatial filtering, yielding increased robustness to misregistration. For the Gemini-South 8-m telescope multiconjugate system and for median Cerro Pachon seeing, the performance of POLC in terms of rms wave-front error averaged over a 1-arc min field of view is approximately three times superior to that of a least-squares reconstructor. Performance degradation due to 30% translational misregistration on all three mirrors is approximately a 30% increased rms wave-front error, whereas a least-squares reconstructor is unstable at such a misregistration level.

Robustness study of the pseudo open-loop controller for multiconjugate adaptive optics

Robustness of the recently proposed "pseudo open-loop control" algorithm against various system errors has been investigated for the representative example of the Gemini-South 8-m telescope multiconjugate adaptive-optics system. The existing model to represent the adaptive-optics system with pseudo open-loop control has been modified to account for misalignments, noise and calibration errors in deformable mirrors, and wave-front sensors. Comparison with the conventional least-squares control model has been done. We show with the aid of both transfer-function pole-placement analysis and Monte Carlo simulations that POLC remains remarkably stable and robust against very large levels of system errors and outperforms in this respect least-squares control. Approximate stability margins as well as performance metrics such as Strehl ratios and rms wave-front residuals averaged over a 1-arc min field of view have been computed for different types and levels of system errors to quantify the expected performance degradation.

Order-N sparse minimum-variance open-loop reconstructor for extreme adaptive optics

A scalable sparse minimum-variance open-loop wave-front reconstructor for extreme adaptive optics (ExAO) systems is presented. The reconstructor is based on Ellerbroek's sparse approximation of the wave-front inverse covariance matrix [J. Opt. Soc. Am. A 19, 1803 (2002)]. The baseline of the numerical approach is an iterative conjugate gradient (CG) algorithm for reconstructing a spatially sampled wave front at N grid points on a computational domain of size equal to the telescope's primary mirror's diameter D that uses a multigrid (MG) accelerator to speed up convergence efficiently and enhance its robustness. The combined MGCG scheme is order N and requires only two CG iterations to converge to the asymptotic average Strehl ratio (SR) and root-mean-square reconstruction error. The SR and reconstruction squared error are within standard deviation with figures obtained from a previously proposed MGCG fast-Fourier-transform based minimum-variance reconstructor that incorporates the exact wave-front inverse covariance matrix on a computational domain of size equal to 2D. A cost comparison between the present sparse MGCG algorithm and a Cholesky factorization based algorithm that uses a reordering scheme to preserve sparsity indicates that the latter method is still competitive for real-time ExAO wave-front reconstruction for systems with up to N approximately equal to 10(4) degrees of freedom because the update rate of the Cholesky factor is typically several orders of magnitude lower than the temporal sampling rate.

Preconditioned conjugate gradient wave-front reconstructors for multiconjugate adaptive optics

Multiconjugate adaptive optics (MCAO) systems with 10(4)-10(5) degrees of freedom have been proposed for future giant telescopes. Using standard matrix methods to compute, optimize, and implement wavefront control algorithms for these systems is impractical, since the number of calculations required to compute and apply the reconstruction matrix scales respectively with the cube and the square of the number of adaptive optics degrees of freedom. We develop scalable open-loop iterative sparse matrix implementations of minimum variance wave-front reconstruction for telescope diameters up to 32 m with more than 10(4) actuators. The basic approach is the preconditioned conjugate gradient method with an efficient preconditioner, whose block structure is defined by the atmospheric turbulent layers very much like the layer-oriented MCAO algorithms of current interest. Two cost-effective preconditioners are investigated: a multigrid solver and a simpler block symmetric Gauss-Seidel (BSGS) sweep. Both options require off-line sparse Cholesky factorizations of the diagonal blocks of the matrix system. The cost to precompute these factors scales approximately as the three-halves power of the number of estimated phase grid points per atmospheric layer, and their average update rate is typically of the order of 10(-2) Hz, i.e., 4-5 orders of magnitude lower than the typical 10(3) Hz temporal sampling rate. All other computations scale almost linearly with the total number of estimated phase grid points. We present numerical simulation results to illustrate algorithm convergence. Convergence rates of both preconditioners are similar, regardless of measurement noise level, indicating that the layer-oriented BSGS sweep is as effective as the more elaborated multiresolution preconditioner.

Numerical simulations of multiconjugate adaptive optics wave-front reconstruction on giant telescopes

We present sample Monte Carlo simulation results to illustrate the trends in multiconjugate adaptive optics (MCAO) performance as the telescope aperture diameter increases from 8 to 32 m with all other first-order system parameters held constant. The MCAO system considered includes three deformable mirrors, a 1-arc min square field of view, and five wave-front-sensing references consisting of either natural guide stars or laser guide stars at a range of either 30 or 90 km. The rms residual wave-front error decreases slowly with increasing aperture diameter with natural guide stars, whereas performance degrades significantly with increasing aperture diameter for laser guide stars at 30 km if the number of guide stars is held fixed. Performance with laser guide stars at 90 km is a weak function of telescope aperture diameter in the range from 8 to 32 m, with rms wave-front errors no more than 20% greater than the corresponding natural guide-star case for the same level of wave-front sensor's measurement noise.

Local, hierarchic, and iterative reconstructors for adaptive optics

Adaptive optics systems for future large optical telescopes may require thousands of sensors and actuators. Optimal reconstruction of phase errors using relative measurements requires feedback from every sensor to each actuator, resulting in computational scaling for n actuators of n2. The optimum local reconstructor is investigated, wherein each actuator command depends only on sensor information in a neighboring region. The resulting performance degradation "global" modes is quantified analytically, and two approaches are considered for recovering global performance. Combining local and global estimators in a two-layer hierarchic architecture yields computations scaling with n(4/3); extending this approach to multiple layers yields linear scaling. An alternative approach that maintains a local structure is to allow actuator commands to depend on both local sensors and prior local estimates. This iterative approach is equivalent to a temporal low-pass filter on global information and gives a scaling of n(3/2). The algorithms are simulated by using data from the Palomar Observatory adaptive optics system. The analysis is general enough to also be applicable to active optics or other systems with many sensors and actuators.

Multigrid preconditioned conjugate-gradient method for large-scale wave-front reconstruction

We introduce a multigrid preconditioned conjugate-gradient (MGCG) iterative scheme for computing open-loop wave-front reconstructors for extreme adaptive optics systems. We present numerical simulations for a 17-m class telescope with n = 48756 sensor measurement grid points within the aperture, which indicate that our MGCG method has a rapid convergence rate for a wide range of subaperture average slope measurement signal-to-noise ratios. The total computational cost is of order n log n. Hence our scheme provides for fast wave-front simulation and control in large-scale adaptive optics systems.