Enabling numerically exact local solver for waveform inversion—a low-rank approach

Kumar, Rajiv; Willemsen, Bram; Herrmann, Felix J.; Malcolm, Alison

doi:10.1007/s10596-019-09832-9

Cited by 12 publications

(4 citation statements)

References 58 publications

(64 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The memory cost of immersive wave modeling is the bottleneck since most personal laptops are limited by storing the large pre-computed Green's functions in dynamic random access memory (DRAM), which has been discussed in Broggini et al (2017). One potential solution is to apply low-rank approximations to the large pre-computed Green's functions so the occupied memory space can be largely reduced while maintaining good accuracy (Kumar et al, 2019).…”

Section: Appendix 4a Deriving the Elastic Representation Theoremmentioning

confidence: 99%

Elastic immersive wave experimentation

Robertsson

Manen

2022

Geophysical Journal International

View full text Add to dashboard Cite

We describe an elastic wave propagation laboratory that enables a solid object to be artificially immersed within an extended (numerical) environment such that a physical wave propagation experiment carried out in the solid drives the propagation in the extended (numerical) environment and vice-versa. The underlying method of elastic immersive wave experimentation for such a laboratory involves deploying arrays of active multi-component sources at the traction-free surface of the solid (e.g., a cube of granitic rock). These sources are used to accomplish two tasks: (1) cancel outgoing waves, and (2) emit ingoing waves representing the first-order interactions between the physical and extended domains, computed using, e.g., a finite-difference (FD) method. Higher-order interactions can be built by alternately carrying out the processes for canceling the outgoing waves and the FD simulations for generating the ingoing waves. We validate the proposed iterative scheme for realizing elastic immersive wave experimentation using two-dimensional (2D) synthetic wave experiments.

show abstract

Section: Appendix 4a Deriving the Elastic Representation Theoremmentioning

confidence: 99%

Elastic immersive wave experimentation

Robertsson

Manen

2022

Geophysical Journal International

View full text Add to dashboard Cite

show abstract

“…Such IBCs take into account all of the wavefield interactions between altered subdomain 2 and unchanged subdomain 1, but, for each frequency, it requires as many wave-equation solutions as grid points discretizing the edges of the domain of interest for simulating the Green's functions in the entire medium. To keep the computational burden reasonable, Kumar et al (2019) proposed a rank minimization-based framework to compute a low-rank factorized form of the Green's functions, which requires fewer wave-equation solutions. FWI is classically solved based on a reduced approach formulation with local optimization algorithms (Pratt et al, 1998), the only affordable approach to the problem considering the size of the data and model spaces, particularly in 3D.…”

Section: Introductionmentioning

confidence: 99%

Localized Wavefield Inversion (LWI): an Adaptation of Multi-Block ADMM for Localized FWI

Aghamiry¹,

Gholami²,

Operto³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Full-waveform inversion (FWI) is a high-resolution and computationally intensive imaging technique to reconstruct unknown parameters in the computational model in which the waves propagate; however, an accurate model of only part of this medium is required for some applications. To decrease the computational burden of such problems, target-oriented FWI was proposed where the redatumed data on the part of the medium or localized solvers for the wave equation are used. On the other hand, the classical formulation of FWI suffers from non-linearity and ill-posedness, which makes FWI sensitive to the initial model, the low-frequency content of the data, and limited illumination. In this study, we propose a localized version of the alternating direction method of multipliers (ADMM)-based FWI method, which was proposed to solve these problems in classical FWI. In our localized FWI or LWI, the medium is decomposed into a few subdomains, where some of them are updated, and the others are kept fixed based on an adaptation of multi-block ADMM, which is a powerful algorithm for solving inverse problems with decomposition and block separability. Numerical tests on the Marmousi model for a time-lapse application confirm the computational efficiency and robustness against background velocity model errors.

show abstract

“…It requires a large number of Green's functions computations in the full domain to calculate the exact boundary conditions at every time step, and by doing so, this methodology can account for every interaction that the local domain may have with the exterior domain. Lately, IBC has been used in the frequency domain for salt boundary problems by Willemsen et al (2016), and it was also tested using a randomized singular value decomposition to compute the low-rank approximation of the Green's function in two and three dimension (Kumar et al, 2019) .…”

Section: Introductionmentioning

confidence: 99%

Comparing Strategies for Local FWI: FD Injection and Immersive Boundary Conditions

Farias¹,

Pestana²

2020

Preprint

View full text Add to dashboard Cite

Conventional Full Waveform Inversion requires calculating the objective function to be minimized and construction a gradient using the whole property model, when is often the case where geoscientist are only interested in a local region. In this study, we use two strategies to perform local FWI in time domain. One that disregards the interaction of the locally altered model with the exterior (FD injection) and the second that can take these iterations into account (Immersive Boundary Conditions). Numerical tests show the influence of whether or not to set aside these interactions for different accuracy of the exterior model.

show abstract

Enabling numerically exact local solver for waveform inversion—a low-rank approach

Abstract: colm, Enabling numerically exact local solver for waveform-inversion-a low-rank approach, Computational Geosciences

Cited by 12 publications

References 58 publications

Elastic immersive wave experimentation

Elastic immersive wave experimentation

Localized Wavefield Inversion (LWI): an Adaptation of Multi-Block ADMM for Localized FWI

Comparing Strategies for Local FWI: FD Injection and Immersive Boundary Conditions

Contact Info

Product

Resources

About