Projection methods and applications for seismic nonlinear inverse problems with multiple constraints
ABSTRACT
Nonlinear inverse problems are often hampered by local minima because of missing low frequencies and far offsets in the data, lack of access to good starting models, noise, and modeling errors. A well-known approach to counter these deficiencies is to include prior information on the unknown model, which regularizes the inverse problem. Although conventional regularization methods have resulted in enormous progress in ill-posed (geophysical) inverse problems, challenges remain when the prior information consists of multiple pieces. To handle this situation, we have developed an optimization framework that allows us to add multiple pieces of prior information in the form of constraints. The proposed framework is more suitable for full-waveform inversion (FWI) because it offers assurances that multiple constraints are imposed uniquely at each iteration, irrespective of the order in which they are invoked. To project onto the intersection of multiple sets uniquely, we use Dykstra’s algorithm that does not rely on trade-off parameters. In that sense, our approach differs substantially from approaches, such as Tikhonov/penalty regularization and gradient filtering. None of these offer assurances, which makes them less suitable to FWI, where unrealistic intermediate results effectively derail the inversion. By working with intersections of sets, we avoid trade-off parameters and keep objective calculations separate from projections that are often much faster to compute than objectives/gradients in 3D. These features allow for easy integration into existing code bases. Working with constraints also allows for heuristics, where we built up the complexity of the model by a gradual relaxation of the constraints. This strategy helps to avoid convergence to local minima that represent unrealistic models. Using multiple constraints, we obtain better FWI results compared with a quadratic penalty method, whereas all definitions of the constraints are in terms of physical units and follow from the prior knowledge directly.
REFERENCES
- , 2002, Parallel multiscale Gauss-Newton-Krylov methods for inverse wave propagation:
Proceedings of the Supercomputing, ACM/IEEE 2002 Conference ,41 . - , 2014, Full waveform inversion using simultaneous encoded sources based on first-and second-order optimization methods: Ph.D. thesis, University of Alberta.
- , 2017, Edge-preserving smoothing for simultaneous-source FWI model updates in high-contrast velocity models: Geophysics, 83, no. 2,
A33–A37 , doi:10.1190/geo2017-0563.1 .GPYSA7 0016-8033 - , 2012, Estimating nuisance parameters in inverse problems: Inverse Problems, 28,
115016 , doi:10.1088/0266-5611/28/11/115016 .INPEEY 0266-5611 - , 2013, Regularized seismic full waveform inversion with prior model information: Geophysics, 78, no. 2,
R25–R36 , doi:10.1190/geo2012-0104.1 .GPYSA7 0016-8033 - , 1988, Comparing hard and soft prior bounds in geophysical inverse problems: Geophysical Journal International, 94,
249–261 , doi:10.1111/j.1365-246X.1988.tb05899.x .GJINEA 0956-540X - , 1988, Two-point step size gradient methods: IMA Journal of Numerical Analysis, 8,
141–148 , doi:10.1093/imanum/8.1.141 .IJNADH 0272-4979 - , 2013, POCS-based geophysical constraints in multi-parameter full wavefield inversion:
75th Annual International Conference and Exhibition, EAGE , Extended Abstracts, doi:10.3997/2214-4609.20130409 . - , 2011, Convex analysis and monotone operator theory in Hilbert spaces, 1st ed.: Springer Publishing Company Inc.
- , 2015, Projection methods: Swiss army knives for solving feasibility and best approximation problems with halfspaces: Contemporary Mathematics, 636,
1–40 , doi:10.1090/conm/636 .CTMAEH 1098-3627 - , 2014, Introduction to nonlinear optimization: SIAM.
- , 2015, On the convergence of alternating minimization for convex programming with applications to iteratively reweighted least squares and decomposition schemes: SIAM Journal on Optimization, 25,
185–209 , doi:10.1137/13094829X .SJOPE8 1095-7189 - , 2015, General optimization framework for robust and regularized 3D FWI:
77th Annual International Conference and Exhibition, EAGE , Extended Abstracts, doi:10.3997/2214-4609.201412589 . - , 2007, Convex constrained optimization for the seismic reflection tomography problem: Journal of Applied Geophysics, 62,
158–166 , doi:10.1016/j.jappgeo.2006.10.004 .JAGPEA 0926-9851 - , 2015, Convex optimization algorithms: Athena Scientific.
- , 1999, Nonmonotone spectral projected gradient methods on convex sets: SIAM Journal on Optimization, 10,
1196–1211 , doi:10.1137/S1052623497330963 .SJOPE8 1095-7189 - , 2003, Inexact spectral projected gradient methods on convex sets: IMA Journal of Numerical Analysis, 23,
539–559 , doi:10.1093/imanum/23.4.539 .IJNADH 0272-4979 - , 2005, Robust stopping criteria for Dykstra’s algorithm: SIAM Journal on Scientific Computing, 26,
1405–1414 , doi:10.1137/03060062X .SJOCE3 1064-8275 - , 2011, Distributed optimization and statistical learning via the alternating direction method of multipliers: Now Publishers Inc.,
Foundations and Trends in Machine Learning Series ,1–122 . - , 2004, Convex optimization: Cambridge University Press.
- , 1986, A method for finding projections onto the intersection of convex sets in Hilbert spaces: Springer,
28–47 . - , 2007, Full waveform tomography for lithospheric imaging: Results from a blind test in a realistic crustal model: Geophysical Journal International, 168,
133–151 , doi:10.1111/j.1365-246X.2006.03156.x .GJINEA 0956-540X - , 1995, Multiscale seismic waveform inversion: Geophysics, 60,
1457–1473 , doi:10.1190/1.1443880 .GPYSA7 0016-8033 - , 1990, Basic convergence theory: Technical report, University of Washington.
- , 2006, Computational acceleration of projection algorithms for the linear best approximation problem: Linear Algebra and its Applications, 416,
111–123 , doi:10.1016/j.laa.2005.10.006 .LAAPAW 0024-3795 - , 2011,
Proximal splitting methods in signal processing , in H. H. BauschkeR. S. BurachikP. L. CombettesV. ElserD. R. LukeH. Wolkowicz, eds., Fixed-point algorithms for inverse problems in science and engineering: Springer,Springer optimization and its applications ,185–212 . - , 2002, R-linear convergence of the Barzilai and Borwein gradient method: IMA Journal of Numerical Analysis, 22,
1–10 , doi:10.1093/imanum/22.1.1 .IJNADH 0272-4979 - , 2017, A unified 2D/3D large scale software environment for nonlinear inverse problems: ArXiv
e–prints . - , 2010, Convex optimization & Euclidean distance geometry: Meboo Publishing.
- , 1983, An algorithm for restricted least squares regression: Journal of the American Statistical Association, 78,
837–842 , doi:10.1080/01621459.1983.10477029 . - , 2011, Alternating projection methods: SIAM.
- , 2016, Constrained waveform inversion for automatic salt flooding: The Leading Edge, 35,
235–239 , doi:10.1190/tle35030235.1 . - , 2015, Automatic salt delineation — Wavefield reconstruction inversion with convex constraints:
85th Annual International Meeting, SEG , Expanded Abstracts,1337–1343 , doi:10.1190/segam2015-5877995.1 . - , 2018, Total variation regularization strategies in full-waveform inversion: SIAM Journal on Imaging Sciences, 11,
376–406 , doi:10.1137/17M111328X . - , 1998, Non-linear inversion using general measures of data misfit and model structure: Geophysical Journal International, 134,
213–227 , doi:10.1046/j.1365-246x.1998.00555.x .GJINEA 0956-540X - , 2004, A comparison of automatic techniques for estimating the regularization parameter in non-linear inverse problems: Geophysical Journal International, 156,
411–425 , doi:10.1111/j.1365-246X.2004.02190.x .GJINEA 0956-540X - , 2002, Nonmonotone globalization techniques for the Barzilai-Borwein gradient method: Computational Optimization and Applications, 23,
143–169 , doi:10.1023/A:1020587701058 .CPPPEF 0926-6003 - , 2012, Constrained full-waveform inversion by model reparameterization: Geophysics, 77, no. 2,
R117–R127 , doi:10.1190/geo2011-0196.1 .GPYSA7 0016-8033 - , 2012, Attenuating crosstalk noise with simultaneous source full waveform inversion: Geophysical Prospecting, 60,
759–768 , doi:10.1111/j.1365-2478.2011.01023.x .GPPRAR 0016-8025 - , 2000, Alternating direction method with self-adaptive penalty parameters for monotone variational inequalities: Journal of Optimization Theory and Applications, 106,
337–356 , doi:10.1023/A:1004603514434 .JOTABN 0022-3239 - , 2011, A modified, sparsity-promoting, Gauss-Newton algorithm for seismic waveform inversion:
SPIE Optical Engineering + Applications, International Society for Optics and Photonics ,81380V–81380V . - , 1996, Prestack migration velocity estimation using nonlinear methods: Geophysics, 61,
138–150 , doi:10.1190/1.1443934 .GPYSA7 0016-8033 - , 1988, Subspace methods for large-scale nonlinear inversion: Springer,
139–154 . - , 1992, A modified gradient method for two-dimensional problems in tomography: Journal of Computational and Applied Mathematics, 42,
17–35 , doi:10.1016/0377-0427(92)90160-Y .JCAMDI 0377-0427 - , 2014, Proximal Newton-type methods for minimizing composite functions: SIAM Journal on Optimization, 24,
1420–1443 , doi:10.1137/130921428 .SJOPE8 1095-7189 - , 2009, A comprehensive study of including structural orientation information in geophysical inversions: Geophysical Journal International, 178,
623–637 , doi:10.1111/j.1365-246X.2009.04188.x .GJINEA 0956-540X - , 2013, A contrast source inversion method in the wavelet domain: Inverse Problems, 29,
025015 , doi:10.1088/0266-5611/29/2/025015 .INPEEY 0266-5611 - , 2012, Fast randomized full-waveform inversion with compressive sensing: Geophysics, 77, no. 3,
A13–A17 , doi:10.1190/geo2011-0410.1 .GPYSA7 0016-8033 - , 2016, Modified Gauss-Newton full-waveform inversion explained — Why sparsity-promoting updates do matter: Geophysics, 81, no. 3,
R125–R138 , doi:10.1190/geo2015-0266.1 .GPYSA7 0016-8033 - , 2015, Acoustic- and elastic-waveform inversion using a modified total-variation regularization scheme: Geophysical Journal International, 200,
489–502 , doi:10.1093/gji/ggu393 .GJINEA 0956-540X - , 2017, Full-waveform inversion. Part 1: Forward modeling: The Leading Edge, 36,
1033–1036 , doi:10.1190/tle36121033.1 . - , 2016, The seiscope optimization toolbox: A large-scale nonlinear optimization library based on reverse communication: Geophysics, 81, no. 2,
F1–F15 , doi:10.1190/geo2015-0031.1 .GPYSA7 0016-8033 - , 2012, Linear and nonlinear inverse problems with practical applications: SIAM.
- , 2015, A general analysis of the convergence of ADMM:
Proceedings of the International Conference on Machine Learning ,343–352 . - , 2000, Numerical optimization: Springer.
- , 1993, Generalized subspace methods for large-scale inverse problems: Geophysical Journal International, 114,
12–20 , doi:10.1111/j.1365-246X.1993.tb01462.x .GJINEA 0956-540X - , 1982, LSQR: An algorithm for sparse linear equations and sparse least squares: ACM Transactions on Mathematical Software, 8,
43–71 , doi:10.1145/355984.355989 .ACMSCU 0098-3500 - , 2014, Proximal algorithms: Foundations and trends® in optimization: Mike Casey,
127–239 . - , 2017, Constraints versus penalties for edge-preserving full-waveform inversion: The Leading Edge, 36,
94–100 , doi:10.1190/tle36010094.1 . - , 1998, Slope constrained topology optimization: International Journal for Numerical Methods in Engineering, 41,
1417–1434 , doi:10.1002/(SICI)1097-0207(19980430)41:8<1417::AID-NME344>3.0.CO;2-N .IJNMBH 0029-5981 - , 1999, Seismic waveform inversion in the frequency domain. Part 1: Theory and verification in a physical scale model: Geophysics, 64,
888–901 , doi:10.1190/1.1444597 .GPYSA7 0016-8033 - , 2016, Full-waveform inversion with steerable variation regularization:
86th Annual International Meeting, SEG , Expanded Abstracts,1174–1178 , doi:10.1190/segam2016-13872436.1 . - , 1993, On the Barzilai and Borwein choice of steplength for the gradient method: IMA Journal of Numerical Analysis, 13,
321–326 , doi:10.1093/imanum/13.3.321 .IJNADH 0272-4979 - , 2016, Primer on monotone operator methods: Applied and Computational Mathematics, 15,
3–43 . - , 1997, To Bayes or not to Bayes?: Geophysics, 62,
1045–1046 , doi:10.1190/1.6241045.1 .GPYSA7 0016-8033 - , 2012, Projected Newton-type methods in machine learning, in optimization for machine learning: MIT Press,
305–327 . - , 2010, Convex structure learning in log-linear models: Beyond pairwise potentials:
Proceedings of the 13th International Conference on Artificial Intelligence and Statistics, PMLR ,709–716 . - , 2009, Optimizing costly functions with simple constraints: A limited-memory projected quasi-Newton algorithm:
Proceedings of the Conference on Artificial Intelligence and Statistics ,456–463 . - , 2003, Computation of differential seismograms and iteration adaptive regularization in prestack waveform inversion: Geophysics, 68,
2026–2039 , doi:10.1190/1.1635056 .GPYSA7 0016-8033 - , 2008, Automatic velocity analysis via shot profile migration: Geophysics, 73, no. 5,
VE49–VE59 , doi:10.1190/1.2972021 .GPYSA7 0016-8033 - , 2005, Differential semblance velocity analysis by wave-equation migration:
75th Annual International Meeting, SEG , Expanded Abstracts,2132–2135 , doi:10.1190/1.1817759 . - , 2015, Constrained waveform inversion of colocated VSP and surface seismic data:
77th Annual International Conference and Exhibition, EAGE , Extended Abstracts, doi:10.3997/2214-4609.201412906 . - , 2015, Constraints versus priors: SIAM/ASA Journal on Uncertainty Quantification, 3,
586–598 , doi:10.1137/130920721 . - , 2002, Computational methods for inverse problems: SIAM.
- , 2016, An empirical study of ADMM for nonconvex problems: Presented at the
NIPS Workshop on Nonconvex Optimization . - , 2017, Adaptive ADMM with spectral penalty parameter selection:
Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, PMLR ,718–727 . - , 2015, Full waveform inversion with sparsity constraint in seislet domain:
85th Annual International Meeting, SEG , Expanded Abstracts,1382–1387 , doi:10.1190/segam2015-5932019.1 . - , 2006, Non-monotone spectral projected gradient method applied to full waveform inversion: Geophysical Prospecting, 54,
525–534 , doi:10.1111/j.1365-2478.2006.00554.x .GPPRAR 0016-8025 - , 2002, Geophysical inverse theory and regularization problems: Elsevier,
36 . - , 2017, Sparse-promoting full-waveform inversion based on online orthonormal dictionary learning: Geophysics, 82, no. 2,
R87–R107 , doi:10.1190/geo2015-0632.1 .GPYSA7 0016-8033
