ROMNet: a hybrid reduced order modeling and machine learning approach to waveform inversion

Liliana Borcea; Alexander Mamonov; Kui Ren; Haizhao Yang; Chugang Yi

math.NAAug 25, 2026

ROMNet: a hybrid reduced order modeling and machine learning approach to waveform inversion

Authors: Liliana Borcea, Alexander Mamonov, Kui Ren, Haizhao Yang, Chugang Yi

Organizations: Applied Physics and Applied Mathematics, Columbia University, New York, NY, 10027. · Department of Mathematics, University of Houston, Houston, TX 77204-3008. · Department of Mathematics, University of Maryland, College Park, MD 20742.

Abstract

Waveform inversion seeks to estimate the wave speed of a heterogeneous, inaccessible medium, from time-resolved measurements of the waves at user controlled sensors. We consider this inverse problem for acoustic waves and an active array of source/receiver sensors that emit probing signals and measure the generated pressure waves. The forward map, from the wave speed to the measurements, is nonlinear and oscillatory. The oscillations cause cycle skipping, the main impediment to using the standard, nonlinear least-squares data fitting formulation, known as full waveform inversion (FWI). A recently introduced alternative waveform inversion approach computes from the measurements an algebraic surrogate of the wave operator, a reduced order model (ROM) matrix, which is then used to estimate the wave speed. The mapping from the measurements to the ROM is nonlinear, but well understood. It is computed efficiently, in a non-iterative manner. The nonlinear mapping from the ROM to the wave speed is less understood, and its approximation involves time-consuming optimization. Our goal in this paper is to use a neural network to map the ROM matrix to a nearby one, that has a simpler and explicit dependence on the wave speed. This simplifies and reduces the computational cost of the ROM-based waveform inversion. We introduce the methodology, called ROMNet, and test it with numerical simulations, using two training data sets: The first set consists of random media with variations of the wave speed modeled by a superposition of Gaussians with random amplitudes and standard deviations. The second is the publicly available GeoFWI dataset introduced for benchmarking FWI using deep learning. We compare the performance of ROMNet with the direct ROM-based inversion and with two representative deep learning approaches to FWI: Fourier-DeepONet" and InversionNet".

Figures & tables

Figure 1: One-dimensional simulations in a homogeneous and layered medium.The domain is $[0,2.5]$ km, with homogeneous Dirichlet boundary conditions, and the source is at $x=0.06$ km. The homogeneous medium has a wave speed $3$ km/s. The layered medium has the same wave speed, except in the interval $[0.35,2.1]$ km, which is divided into four equal layers, with wave speeds $3.5,2.5,3.5,2.5$ km/s, from top to bottom. The snapshots in the homogeneous medium (a) and the heterogeneous medium (b). Orthogonalized snapshots in the homogeneous medium (c) and the heterogeneous medium (d). The abscissa is the discrete time $t_{j}=j\tau$ , for $j=0,\ldots,n-1$ . The ordinate is the travel time $T$ . For each $t_{j}$ , we use colors to plot the amplitudes of $u_{j}(x(T))$ and $v_{j}(x(T))$ , respectively.

Figure 2: Two dimensional simulations. The reference medium is homogeneous, with wave speed $c_{o}({{\itbf x}})\equiv\bar{c}=3$ km/s. The true speed $c({{\itbf x}})$ differs from it only inside two thin horizontal bars outlined by the dashed rectangles, where the wave speed is $4$ km/s. The top two rows show the snapshots in the reference and heterogeneous media, respectively, at three time instants. The bottom two rows show the corresponding orthonormal basis vectors. The array is linear and lies at the top of the plotted domain, where we have a Dirichlet boundary condition. The subscript $o$ in the titles refers to $c_{o}$ , and the yellow triangles are the sensors in the array. The contour of the reflectors is outlined in all the pictures, including those corresponding to the reference medium, to aid in the interpretation.

Figure 3: ROMNet inversion (top branch) and the ROM inversion (bottom branch). Both start from the data-driven

\mathbfcal{A}^{{\scalebox{0.5}[0.4]{ROM}}}

. The network

\mathcal{G}_{\bm{\theta}}

is trained offline (dashed).

Figure 4: (a) True model $c({{\itbf x}})$ and the linear array located near the top boundary. The abscissa and ordinate are in km and the colorbar is in km/s. (b)–(c) Decimal logarithms of the FWI and ROM objective functions $\mathcal{O}^{{\scalebox{0.5}[0.4]{FWI}}}$ , $\mathcal{O}^{{\scalebox{0.5}[0.4]{ROM}}}$ . (d)–(e) plot the decimal logarithms of the objective functions $\mathcal{O}^{{\scalebox{0.5}[0.4]{ML}}}$ , for $\gamma=5$ : In (d) we use the learned approximation of $\mathcal{F}^{{\scalebox{0.5}[0.4]{ML}}}$ while in (e) we use the exact $\mathcal{F}^{{\scalebox{0.5}[0.4]{ML}}}$ . The axes in (b)–(e) are: the contrast of the wave speed in the abscissa and the interface position in the ordinate. The global minimum is indicated with ${\color[rgb]{1,0,1}\bigcirc}$ and $\times$ marks the minimum of each objective on the search grid. The true parameters are not included in the search.

method	relative $L^{2}$ error
ROMNet	0.0297
Fourier-DeepONet	0.0332
InversionNet	0.0419
ROM	0.0293
Oracle	0.0232

Table 1: Mean relative

L^{2}

error for all methods, over11 held-out test samples.

method	rel. $L^{2}$	MAE	RMSE	SSIM
ROMNet	0.0297	0.0476	0.0890	0.7443
Fourier-DeepONet	0.0332	0.0517	0.0995	0.7438
InversionNet	0.0419	0.0676	0.1258	0.5902

Table 2: Complementary error metrics over the 11 held-out test samples.

Figure 5: Recovered wave speed on two held-out Random-Gaussian test samples. From left to right, the columns show ground truth $c_{\rm true}$ , ROMNet, Fourier-DeepONet, InversionNet, Oracle, and the ROM inversion results. The $L^{2}$ relative error of each estimate is printed in its panel.

Figure 6: Recovered velocity on three out-of-distribution probes. From left to right: true $c({{\itbf x}})$ , ROMNet, Fourier-DeepONet, InversionNet, Oracle, and the ROM inversion results. The $L^{2}$ relative error of each estimate is printed in its panel.

	$\gamma=1$	$\gamma=2$	$\gamma=3$	$\gamma=4$	$\gamma=5$
ROMNet	0.0304	0.0299	0.0298	0.0297	0.0297
Oracle	0.0241	0.0233	0.0231	0.0231	0.0232
ROM	0.0301	0.0293	0.0292	0.0291	0.0289

Table 3: Effect of the windowing parameter

\gamma

on the mean

L^{2}

relative error of the reconstruction. The ROMNet and Oracle rows are averaged over the same

11

held-out test media as Table 1 , and the ROM row over

3

of them, which is why its value at

\gamma=5

differs slightly from the one in Table 1 .

Figure 7: From left to right: true $c({{\itbf x}})$ , the hat-basis inversion with ROMNet and the Oracle; the Gaussian-dictionary inversion with ROMNet and the Oracle.

category	sample size	ROMNet	Fourier- DeepONet	InversionNet	ROM	Oracle
fold	15	0.0264	0.0284	0.0528	0.0235	0.0180
fault	15	0.0382	0.0459	0.0799	0.0252	0.0186
salt	15	0.0428	0.0523	0.0967	0.0530	0.0241
overall	45	0.0358	0.0422	0.0765	0.0339	0.0202

Table 4: Mean

L^{2}

relative error on

15

random samples per structural category of the GeoFWI test set (

45

in total). Best learning-based method in each row is in bold.

Figure 8: Three velocity models with folded layering. From left to right: True $c({{\itbf x}})$ , ROMNet, Fourier-DeepONet, InversionNet, Oracle and ROM inversion.

Figure 9: Three velocity models with faulted layering. Columns as in Figure 8 .

Figure 10: Three velocity models containing a salt body. Columns as in Figure 8 .

Figure 11: GeoFWI: the initial wave speed of the ROM inversion (smoothed true $c({{\itbf x}})$ , middle column) and the reference $c_{o}({{\itbf x}})$ of ROMNet (right column), for a medium with a salt body (top) and a folded medium (bottom). The true wave speeds are in the left column. Both initial wave speeds depend only on depth; the ROM inversion uses a different one for each medium, whereas $c_{o}({{\itbf x}})$ is the same for all media.

Figure 12: Two crops of the BP-2004 model. From left to right: True $c({{\itbf x}})$ , ROMNet, Fourier-DeepONet, InversionNet, Oracle, ROM.

Figure 13: The evolution of the estimated wave speed with the ROM-based inversion, when starting with the initial guess given by ROMNet. From left to right: True $c({{\itbf x}})$ , the ROMNet estimate (the same as in Fig. 12 ) and the estimates after $5$ , $10$ , $15$ and $20$ Gauss-Newton iterations.

	per iter.	total ( $K\!=\!40$ )	error
ROM	$19.4$ s	$14$ min	$0.0295$
ROMNet	$\mathbf{0.26}$ s	$\mathbf{10.5}$ s	$0.0296$

Table 5: Wall time and

L^{2}

relative error for ROM and ROMNet. Per-iteration times exclude the one-time operator build and compilation, and the totals are measured end-to-end sums, averaged over the three test samples on an idle GPU. The errors are averaged over the same three media, which is why they differ slightly from those of Table 1 , and the ROM inversion and ROMNet are both run for

K=40

iterations. The two methods use the same configuration described in appendix A.2 .

Figure 14: Random Gaussians dataset: $L^{2}$ relative error of the reconstruction per Gauss–Newton iteration, for the same configuration as in Table 5 .

Figure 15: GeoFWI dataset: ROMNet estimates for 3 test media, from noiseless data without and with the regularization ( 4.3 )–( 4.4 ), and from data with $1\%$ additive noise, with the regularization.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Random Gaussians
Domain, grid	$\Omega=[0,2.0]\times[0,2.5]$ km, uniform grid of $100\times 125$ points
Probing pulse	$\omega_{o}/(2\pi)=6$ Hz, $B/(2\pi)=4$ Hz
$\bar{c}$ and $\lambda_{\rm min}$	$\bar{c}=3$ km/s, $\lambda_{\rm min}=2\pi\bar{c}/(\omega_{o}+B)=0.3$ km
Array	$m=10$ sensors at depth $0.14$ km, at horizontal positions $0.30,0.456,\ldots,1.70$ km (spacing $0.156$ km, aperture $a=1.4$ km)
Time sampling	$\tau=0.040$ s, $n=16$ snapshots
Reference speed	$c_{o}({{\itbf x}})\equiv\bar{c}=3$ km/s

Appendix

Table 6: Setup for the Random Gaussians data set.

GeoFWI [ 40 ]
Domain, grid	$\Omega=[0,2.05]\times[0,3.0]$ km, uniform grid of $123\times 180$ points. GeoFWI models, given on $100\times 100$ grids, are resampled to this grid.
Probing pulse	$\omega_{o}/(2\pi)=6$ Hz, $B/(2\pi)=4$ Hz
$\bar{c}$ and $\lambda_{\rm min}$	$\bar{c}=1.5$ km/s (the smallest wave speed in the data set, in water), $\lambda_{\rm min}=2\pi\bar{c}/(\omega_{o}+B)=0.15$ km
Array	$m=10$ sensors at depth $0.15$ km, at horizontal positions $0.30,0.461,\ldots,1.75$ km (spacing $0.161$ km, aperture $a=1.45$ km)
Time sampling	$\tau=0.0435$ s, $n=40$ snapshots for ROMNet; the ROM inversion uses $n=56$ snapshots, with the domain extended below the imaging region, so that the bottom boundary has no effect.
Reference $c_{o}({{\itbf x}})$	the per-depth median of the training wave speeds; it equals $1.69$ km/s at the depth of the array and $3.83$ km/s at the bottom

Appendix

Table 7: Setup for the GeoFWI data set.

	Random Gaussians	GeoFWI
Number of functions	$28\times 28+1=N_{c}=785$	$20\times 32+1=N_{c}=641$
Spacing of the centers	$0.072$ km $\times$ $0.090$ km	$0.105$ km $\times$ $0.095$ km
$\rho_{x},\rho_{y}$ ( $0.7\times$ spacing)	$0.050$ km, $0.063$ km	$0.074$ km, $0.067$ km
Windowing $\gamma$	$5$	$5$

Appendix

Table 8: Setup of the Stage 2 (Gauss–Newton) inversion for the two data sets.

Inverse Problems

Inverse problems infer hidden causes, parameters, or signals from indirect, noisy measurements using physical models, priors, and statistical methods.

See all 246 papers

Oct 11, 2026

Full Waveform Inversion with a Learned Preconditioner

Full-waveform inversion (FWI) is a powerful approach for reconstructing subsurface velocity models, but its practical performance is limited by ill-conditioning, sensitivity to the update direction, and the high cost of curvature-based methods. In this project, we investigate whether a neural operator trained on reverse-time-migration (RTM) images can provide a useful learned preconditioner for FWI. We compare projected FWI against neural-preconditioned FWI using either a standard Fourier neural operator (FNO) or a Jacobian-informed FNO (JI-FNO). The neural preconditioner applies the normal operator of the learned RTM map to modify the true Devito FWI gradient. Our experiments show that the JI-FNO preconditioner achieves the lowest model error while maintaining data-residual decay comparable to baseline projected FWI. In contrast, the standard FNO preconditioner is unstable and gives worse model recovery, despite using the same neural-preconditioning framework. We also identify a failure mode of a nonlinear neural Gauss Newton-type inner solve: directly matching the neural RTM output to the FWI gradient creates an objective-space mismatch and does not yield a reliable descent direction. These results suggest that forward RTM-image prediction alone is insufficient for learned preconditioning; accurate learned derivatives and objective-consistent use of the neural operator are critical for producing useful inversion updates.

Jeongjin Park

Seismic Imaging Inverse Problems Learning to Optimize Image Inverse Problems Fourier Neural Operator

Ask PDF

Apr 8, 2026

SPAMoE: Spectrum-Aware Hybrid Operator Framework for Full-Waveform Inversion

Full-waveform inversion (FWI) is pivotal for reconstructing high-resolution subsurface velocity models but remains computationally intensive and ill-posed. While deep learning approaches promise efficiency, existing Convolutional Neural Networks (CNNs) and single-paradigm Neural Operators (NOs) struggle with one fundamental issue: frequency entanglement of multi-scale geological features. To address this challenge, we propose Spectral-Preserving Adaptive MoE (SPAMoE), a novel spectrum-aware framework for solving inverse problems with complex multi-scale structures. Our approach introduces a Spectral-Preserving DINO Encoder that enforces a lower bound on the high-to-low frequency energy ratio of the encoded representation, mitigating high-frequency collapse and stabilizing subsequent frequency-domain modeling. Furthermore, we design a novel Spectral Decomposition and Routing mechanism that dynamically assigns frequency bands to a Mixture-of-Experts (MoE) ensemble comprising FNO, MNO, and LNO. On the ten OpenFWI sub-datasets, experiments show that SPAMoE reduces the average MAE by 44.4% relative to the best officially reported OpenFWI baseline, thereby establishing a new architectural framework for learning-based full-waveform inversion. Our code and data are available at https://github.com/zhenyuwang12366/SPAMoE

Zhenyu Wang, Peiyuan Li, Yongxiang Shi +3

China University of Mining and Technology - Beijing · City University of Hong Kong (Dongguan) · The Hong Kong University of Science and Technology (Guangzhou)

Inverse Problems PDE Inverse Problems Frequency-Domain Feature Learning Mixture of Experts Fourier Neural Operator

Ask PDF

Mar 23, 2026

Unveiling the Mechanism of Continuous Representation Full-Waveform Inversion: A Wave Based Neural Tangent Kernel Framework

Full-waveform inversion (FWI) estimates physical parameters in the wave equation from limited measurements and has been widely applied in geophysical exploration, medical imaging, and non-destructive testing. Conventional FWI methods are limited by their notorious sensitivity to the accuracy of the initial models. Recent progress in continuous representation FWI (CR-FWI) demonstrates that representing parameter models with a coordinate-based neural network, such as implicit neural representation (INR), can mitigate the dependence on initial models. However, its underlying mechanism remains unclear, and INR-based FWI shows slower high-frequency convergence. In this work, we investigate the general CR-FWI framework and develop a unified theoretical understanding by extending the neural tangent kernel (NTK) for FWI to establish a wave-based NTK framework. Unlike standard NTK, our analysis reveals that wave-based NTK is not constant, both at initialization and during training, due to the inherent nonlinearity of FWI. We further show that the eigenvalue decay behavior of the wave-based NTK can explain why CR-FWI alleviates the dependency on initial models and shows slower high-frequency convergence. Building on these insights, we propose several CR-FWI methods with tailored eigenvalue decay properties for FWI, including a novel hybrid representation combining INR and multi-resolution grid (termed IG-FWI) that achieves a more balanced trade-off between robustness and high-frequency convergence rate. Applications in geophysical exploration on Marmousi, 2D SEG/EAGE Salt and Overthrust, 2004 BP model, and the more realistic 2014 Chevron models show the superior performance of our proposed methods compared to conventional FWI and existing INR-based FWI methods.

Ruihua Chen, Yisi Luo, Bangyu Wu +1

School of Mathematics and Statistics, Xi’an Jiaotong University, Xi’an, Shaanxi, China · Macao Institute of Systems Engineering, Macau University of Science and Technology, Taipa, Macao

Inverse Problems Neural Tangent Kernel Implicit Neural Representations

Ask PDF

Abstract

Inverse Problems

Explore similar work

Full Waveform Inversion with a Learned Preconditioner

SPAMoE: Spectrum-Aware Hybrid Operator Framework for Full-Waveform Inversion

Unveiling the Mechanism of Continuous Representation Full-Waveform Inversion: A Wave Based Neural Tangent Kernel Framework