Ernie Esser

A General Framework for a Class of First Order Primal-Dual Algorithms for Convex Optimization in Imaging Science

We generalize the primal-dual hybrid gradient (PDHG) algorithm proposed by Zhu and Chan in [An Efficient Primal-Dual Hybrid Gradient Algorithm for Total Variation Image Restoration, CAM Report 08-34, UCLA, Los Angeles, CA, 2008] to a broader class of convex optimization problems. In addition, we survey several closely related methods and explain the connections to PDHG. We point out convergence results for a modified version of PDHG that has a similarly good empirical convergence rate for total variation (TV) minimization problems. We also prove a convergence result for PDHG applied to TV denoising with some restrictions on the PDHG step size parameters. We show how to interpret this special case as a projected averaged gradient method applied to the dual functional. We discuss the range of parameters for which these methods can be shown to converge. We also present some numerical comparisons of these algorithms applied to TV denoising, TV deblurring, and constrained

Fast "Online" Migration with Compressive Sensing

l_1

Total Variation Regularization Strategies in Full-Waveform Inversion

minimization problems.

Matrix Completion on Unstructured Grids - 2-D Seismic Data Regularization and Interpolation

We present a novel adaptation of a recently developed relatively simple iterative algorithm to solve large-scale sparsity-promoting optimization problems. Our algorithm is particularly suitable to large-scale geophysical inversion problems, such as sparse least-squares reverse-time migration or Kirchoff migration since it allows for a tradeoff between parallel computations, memory allocation, and turnaround times, by working on subsets of the data with different sizes. Comparison of the proposed method for sparse least-squares imaging shows a performance that rivals and even exceeds the performance of state-of-the art one-norm solvers that are able to carry out least-squares migration at the cost of a single migration with all data.

SVD-free Low-rank Matrix Factorization - Wavefield Reconstruction Via Jittered Subsampling and Reciprocity

We propose an extended full-waveform inversion formulation that includes general convex constraints on the model. Though the full problem is highly nonconvex, the overarching optimization scheme arrives at geologically plausible results by solving a sequence of relaxed and warm-started constrained convex subproblems. The combination of box, total variation, and successively relaxed asymmetric total variation constraints allows us to steer free from parasitic local minima while keeping the estimated physical parameters laterally continuous and in a physically realistic range. For accurate starting models, numerical experiments carried out on the challenging 2004 BP velocity benchmark demonstrate that bound and total variation constraints improve the inversion result significantly by removing inversion artifacts, related to source encoding, and by clearly improved delineation of top, bottom, and flanks of a high-velocity high-contrast salt inclusion. The experiments also show that for poor starting models t...

Using Common Information in Compressive Time-lapse Full-waveform Inversion

Seismic data interpolation via rank-minimization techniques has been recently introduced in the seismic community. All the existing rank-minimization techniques assume the recording locations to be on a regular grid, e.g. sampled periodically, but seismic data are typically irregularly sampled along spatial axes. Other than the irregularity of the sampled grid, we often have missing data. In this paper, we study the effect of grid irregularity to conduct matrix completion on a regular grid for unstructured data. We propose an improvement of existing rank-minimization techniques to do regularization. We also demonstrate that we can perform seismic data regularization and interpolation simultaneously. We illustrate the advantages of the modification using a real seismic line from the Gulf of Suez to obtain high quality results for regularization and interpolation, a key application in exploration geophysics.

A Lifted l1/l2 Constraint for Sparse Blind Deconvolution

Recently computationally efficient rank optimization techniques have been studied extensively to develop a new mathematical tool for the seismic data interpolation. So far, matrix completion problems have been discussed where sources are subsample according to a discrete uniform distribution. In this paper, we studied the effect of two different subsampling techniques on seismic data interpolation using rank-regularized formulations, namely jittered subsampling over uniform random subsampling. The other objective of this paper is to combine the fact of source-receiver reciprocity with the rank-minimization techniques to enhance the accuracy of missing-trace interpolation. We illustrate the advantages of jittered subsampling and reciprocity using a seismic line from Gulf of Suez to obtain high quality results for interpolation, a key application in exploration geophysics.

Resolving scaling ambiguities with the ℓ1/ℓ2 norm in a blind deconvolution problem with feedback

The use of time-lapse seismic data to monitor changes in the subsurface has become standard practice in industry. In addition, full-waveform inversion has also been extended to time-lapse seismic to obtain useful time-lapse information. The computational cost of this method are becoming more pronounced as the volume of data increases. Therefore, it is necessary to develop fast inversion algorithms that can also give improved time-lapse results. Rather than following existing joint inversion algorithms, we are motivated by a joint recovery model which exploits the common information among the baseline and monitor data. We propose a joint inversion framework, leveraging ideas from distributed compressive sensing and the modified Gauss-Newton method for full-waveform inversion, by using the shared information in the time-lapse data. Our results on a realistic synthetic example highlight the benefits of our joint inversion approach over a parallel inversion method that does not exploit the shared information. Preliminary results also indicate that our formulation can address time-lapse data with inconsistent acquisition geometries.

Application of a Convex Phase Retrieval Method to Blind Seismic Deconvolution

We propose a modification to a sparsity constraint based on the ratio of l1 and l2 norms for solving blind seismic deconvolution problems in which the data consist of linear convolutions of different sparse reflectivities with the same source wavelet. We also extend the approach to the Estimation of Primaries by Sparse Inversion (EPSI) model, which includes surface related multiples. Minimizing the ratio of l1 and l2 norms has been previously shown to promote sparsity in a variety of applications including blind deconvolution. Most existing implementations are heuristic or require smoothing the l1/l2 penalty. Lifted versions of l1/l2 constraints have also been proposed but are challenging to implement. Inspired by the lifting approach, we propose to split the sparse signals into positive and negative components and apply an l1/l2 constraint to the difference, thereby obtaining a constraint that is easy to implement without smoothing the l1 or l2 norms. We show that a method of multipliers implementation of the resulting model can recover source wavelets that are not necessarily minimum phase and approximately reconstruct the sparse reflectivities. Numerical experiments demonstrate robustness to the initialization as well as to noise in the data.

Constrained waveform inversion for automatic salt flooding

Compared to more mundane blind deconvolution problems, blind deconvolution in seismic applications involves a feedback mechanism related to the free surface. The presence of this feedback mechanism gives us an unique opportunity to remove ambiguities that have plagued blind deconvolution for a long time. While beneficial, this feedback by itself is insufficient to remove the ambiguities even with ℓ1 constraints. However, when paired with an ℓ1/ℓ2 constraint the feedback allows us to resolve the scaling ambiguity under relatively mild assumptions. Inspired by lifting approaches, we propose to split the sparse signal into positive and negative components and apply an ℓ1/ℓ2 constraint to the difference, thereby obtaining a constraint that is easy to implement. Numerical experiments demonstrate robustness to the initialization as well as to noise in the data.