Search NASA⌕ Search

SEARCH · Search NASA

Results for “Inverse problem”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 739 records · Page 41

The behavior of transverse waves in nonuniform solar flux tubes. II. Implications for coronal loop seismology

The seismology of coronal loops using observations of damped transverse oscillations in combination with results from theoretical models is a tool to indirectly infer physical parameters in the solar atmospheric plasma. Existing seismology schemes based on approximations of the period and damping time of kink oscillations are often used beyond their theoretical range of applicability. These approximations assume that the variation of density across the loop is confined to a nonuniform layer much thinner than the radius of the loop, but the results of the inversion problem often do not satisfy this preliminary hypothesis. Here, we determine the accuracy of the analytic approximations of the period and damping time, and the impact on seismology estimates when largely nonuniform loops are considered. We find that the accuracy of the approximations when used beyond their range of applicability is strongly affected by the form of the density profile across the loop, that is observationally unknown and so must be arbitrarily imposed as part of the theoretical model. The error associated with the analytic approximations can be larger than 50% even for relatively thin nonuniform layers. This error directly affects the accuracy of approximate seismology estimates compared to actual numerical inversions. In addition, assuming different density profiles can produce noncoincident intervals of the seismic variables in inversions of the same event. The ignorance about the true shape of density variation across the loop is an important source of error that may dispute the reliability of parameters seismically inferred assuming an ad hoc density profile.

79 ASTRONOMY AND ASTROPHYSICS↗

Composite Bond Line Measurements Based on a Bayesian Analysis of Flash Thermography Data

For bonded composite materials, an accurate characterization of the adhesive bond line is needed to predict failure modes and fracture toughness. In this paper, bond line thickness was estimated from data obtained using through transmission flash thermography. The forward model that predicts back surface temperature is based on a three layer heat diffusion equation with varying diffusivity and flux boundary conditions. The corresponding inverse problem of estimating bond line thickness from measurement data was solved using a Bayesian approach that assumed Gaussian priors for the bond line thickness and thermal diffusivity of the adherends. Finally, the outputs of the thermography based method were compared to measurements that were collected using a micrometer and ultrasound testing.

Flash thermography↗

Vector tomography for reconstructing electric fields with non-zero divergence in bounded domains

In vector tomography (VT), the aim is to reconstruct an unknown multi-dimensional vector field using line integral data. In the case of a 2-dimensional VT, two types of line integral data are usually required. These data correspond to integration of the parallel and perpendicular projection of the vector field along the integration lines and are called the longitudinal and transverse measurements, respectively. In most cases, however, the transverse measurements cannot be physically acquired. Therefore, the VT methods are typically used to reconstruct divergence-free (or source-free) velocity and flow fields that can be reconstructed solely from the longitudinal measurements. In this paper, we show how vector fields with non-zero divergence in a bounded domain can also be reconstructed from the longitudinal measurements without the need of explicitly evaluating the transverse measurements. To the best of our knowledge, VT has not previously been used for this purpose. In particular, we study low-frequency, time-harmonic electric fields generated by dipole sources in convex bounded domains which arise, for example, in electroencephalography (EEG) source imaging. We explain in detail the theoretical background, the derivation of the electric field inverse problem and the numerical approximation of the line integrals. We show that fields with non-zero divergence can be reconstructed from the longitudinal measurements with the help of two sparsity constraints that are constructed from the transverse measurements and the vector Laplace operator. As a comparison to EEG source imaging, we note that VT does not require mathematical modeling of the sources. By numerical simulations, we show that the pattern of the electric field can be correctly estimated using VT and the location of the source activity can be determined accurately from the reconstructed magnitudes of the field. - Highlights: • Vector tomography is used to reconstruct electric fields generated by dipole sources. • Inverse solutions are based on longitudinal and transverse line integral measurements. • Transverse line integral measurements are used as a sparsity constraint. • Numerical procedure to approximate the line integrals is described in detail. • Patterns of the studied electric fields are correctly estimated.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Feature-adjacent multi-fidelity physics-informed machine learning for partial differential equations

Physics-informed neural networks have emerged as an alternative method for solving partial differential equations. However, for complex problems, the training of such networks can still require high-fidelity data which can be expensive to generate. To reduce or even eliminate the dependency on high-fidelity data, we propose a novel multi-fidelity architecture which is based on a feature space shared by the low- and high-fidelity solutions. In the feature space, the projections of the low-fidelity and high-fidelity solutions are adjacent by constraining their relative distance. The feature space is represented with an encoder and its mapping to the original solution space is effected through a decoder. In conclusion, the proposed multi-fidelity approach is validated on forward and inverse problems for steady and unsteady problems described by partial differential equations.

97 MATHEMATICS AND COMPUTING↗

Bayesian soft X-ray tomography using non-stationary Gaussian Processes

In this study, a Bayesian based non-stationary Gaussian Process (GP) method for the inference of soft X-ray emissivity distribution along with its associated uncertainties has been developed. For the investigation of equilibrium condition and fast magnetohydrodynamic behaviors in nuclear fusion plasmas, it is of importance to infer, especially in the plasma center, spatially resolved soft X-ray profiles from a limited number of noisy line integral measurements. For this ill-posed inversion problem, Bayesian probability theory can provide a posterior probability distribution over all possible solutions under given model assumptions. Specifically, the use of a non-stationary GP to model the emission allows the model to adapt to the varying length scales of the underlying diffusion process. In contrast to other conventional methods, the prior regularization is realized in a probability form which enhances the capability of uncertainty analysis, in consequence, scientists who concern the reliability of their results will benefit from it. Under the assumption of normally distributed noise, the posterior distribution evaluated at a discrete number of points becomes a multivariate normal distribution whose mean and covariance are analytically available, making inversions and calculation of uncertainty fast. Additionally, the hyper-parameters embedded in the model assumption can be optimized through a Bayesian Occam's Razor formalism and thereby automatically adjust the model complexity. This method is shown to produce convincing reconstructions and good agreements with independently calculated results from the Maximum Entropy and Equilibrium-Based Iterative Tomography Algorithm methods.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Adding GPU Support to the Markov Chain Monte Carlo Code Catmip

In geophysics, we are confronted with many under-determined inverse problems. For example, all of our observations of earthquakes are made at the Earth’s surface. So, when we try to infer how slip during an earthquake evolves in space and time, we find that there are many potential slip histories that are consistent with our limited observations and our understanding of earthquake physics. One way to approach these problems is with Bayesian analysis which allows us to infer the ensemble of all potential slip models that satisfy the observations and our prior knowledge of earthquake physics. In Bayesian analysis, our prior knowledge is known as the prior probability density function or prior PDF, the fit to the data is known as the data likelihood, and the target PDF that satisfies both the prior PDF and data likelihood is known as the posterior PDF. However, simulating the posterior PDF typically requires using Markov Chain Monte Carlo (MCMC) to draw tens of billions of random realizations of earthquake slip models, which may not be computationally feasible. To make this and similar geophysical inversions computationally tractable, we developed the Cascading Adaptive Transitional Metropolis In Parallel (CATMIP) algorithm. CATMIP is an efficient parallel Markov Chain Monte Carlo (MCMC) sampler that is used for model fitting and uncertainty quantification in geophysics. Example use cases are earthquake rupture modeling, determining mineral composition on Mars, reconstructing the history of ocean salinity, and historical earthquake relocation. CATMIP employs many parallel instances of the Metropolis algorithm for sampling in a transitioning framework. Transitioning is a process in which a set of random samples at equilibrium with a known probability density function (PDF) are used as seeds for the Markov chains to sample successive target PDFs that incrementally move the distribution from the starting seeds to the final desired PDF that describes the relative plausibility of potential values for the model parameters. The algorithm is implemented as a Master-Worker model employing MPI for communication. The worker processes are loosely coupled with global parameters periodically optimized by the master process. This provides a very high amount of parallelism with little communication between updates. During the presentation we will discuss the history of the algorithm and elaborate the earthquake rupture modeling use case for the CATMIP package. Our first step toward GPU optimization was to optimize the code for the CPU. CPU profiling revealed that most of the compute time is spent in calls to level 2 BLAS routines and calls to GSL random number generators. We revised the algorithm to employ level 3 BLAS routines instead. In our presentation we will describe how this was accomplished. Adding GPU support to CATMIP consisted mostly of replacing the calls to GSL with calls to GPU vendor-provided library routines. A small number of loops were directly implemented in CUDA. In the presentation will provide implementation details. Finally, we will discuss methods for profiling and opportunities for further optimizing GPU execution. By creating a code with the flexibility to run on either a CPU or GPU architecture, CATMIP can be used on systems ranging from large CPU-based HPC environments to single servers with GPU acceleration and everything in between.

HECC↗

Application of Gaussian Process Modeling to Analysis of Functional Unreliability

This paper applies Gaussian Process (GP) modeling to analysis of the functional unreliability of a “passive system.” GPs have been used widely in many ways [1]. The present application uses a GP for emulation of a system simulation code. Such an emulator can be applied in several distinct ways, discussed below. All applications illustrated in this paper have precedents in the literature; the present paper is an application of GP technology to a problem that was originally analyzed [2] using neural networks (NN), and later [3, 4] by a method called “Alternating Conditional Expectations” (ACE). This exercise enables a multifaceted comparison of both the processes and the results. Given knowledge of the range of possible values of key system variables, one could, in principle, quantify functional unreliability by sampling from their joint probability distribution, and performing a system simulation for each sample to determine whether the function succeeded for that particular setting of the variables. Using previously available system simulation codes, such an approach is generally impractical for a plant-scale problem. It has long been recognized, however, that a well-trained code emulator or surrogate could be used in a sampling process to quantify certain performance metrics, even for plant-scale problems. “Response surfaces” were used for this many years ago. But response surfaces are at their best for smoothly varying functions; in regions of parameter space where key system performance metrics may behave in complex ways, or even exhibit discontinuities, response surfaces are not the best available tool. This consideration was one of several that drove the work in [2]. In the present paper, (1) the original quantification of functional unreliability using NN [2], and later ACE [3], is reprised using GP; (2) additional information provided by the GP about uncertainty in the limit surface, generally unavailable in other representations, is discussed; (3) a simple forensic exercise is performed, analogous to the inverse problem of code calibration, but with an accident management spin: given an observation about containment pressure, what can we say about the system variables? References 1. For an introduction to GPs, see (for example) Gaussian Processes for Machine Learning, C. E. Rasmussen and C. K. I. Williams (MIT, 2006). 2. Reliability Quantification of Advanced Reactor Passive Safety Systems, J. J. Vandenkieboom, PhD Thesis (University of Michigan, 1996). 3. Z. Cui, J. C. Lee, J. J. Vandenkieboom, and R. W. Youngblood, “Unreliability Quantification of a Containment Cooling System through ACE and ANN Algorithms,” Trans. Am. Nucl. Soc. 85, 178 (2001). 4. Risk and Safety Analysis of Nuclear Systems, J. C. Lee and N. J. McCormick (Wiley, 2011). See especially §11.2.4.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

A fast marching algorithm for the factored eikonal equation

The eikonal equation is instrumental in many applications in several fields ranging from computer vision to geoscience. This equation can be efficiently solved using the iterative Fast Sweeping (FS) methods and the direct Fast Marching (FM) methods. However, when used for a point source, the original eikonal equation is known to yield inaccurate numerical solutions, because of a singularity at the source. In this case, the factored eikonal equation is often preferred, and is known to yield a more accurate numerical solution. One application that requires the solution of the eikonal equation for point sources is travel time tomography. This inverse problem may be formulated using the eikonal equation as a forward problem. While this problem has been solved using FS in the past, the more recent choice for applying it involves FM methods because of the efficiency in which sensitivities can be obtained using them. However, while several FS methods are available for solving the factored equation, the FM method is available only for the original eikonal equation. In this paper we develop a Fast Marching algorithm for the factored eikonal equation, using both first and second order finite-difference schemes. Our algorithm follows the same lines as the original FM algorithm and requires the same computational effort. In addition, we show how to obtain sensitivities using this FM method and apply travel time tomography, formulated as an inverse factored eikonal equation. Numerical results in two and three dimensions show that our algorithm solves the factored eikonal equation efficiently, and demonstrate the achieved accuracy for computing the travel time. We also demonstrate a recovery of a 2D and 3D heterogeneous medium by travel time tomography using the eikonal equation for forward modeling and inversion by Gauss–Newton.

97 MATHEMATICS AND COMPUTING↗

Frequency-domain full-waveform inversion with non-linear descent directions

Full-waveform inversion (FWI) is a highly non-linear inverse problem, normally solved iteratively, with each iteration involving an update constructed through linear operations on the residuals. Incorporating a flexible degree of non-linearity within each update may have important consequences for convergence rates, determination of low model wavenumbers and discrimination of parameters. In this study, we examine one approach for doing so, wherein higher order scattering terms are included within the sensitivity kernel during the construction of the descent direction, adjusting it away from that of the standard Gauss–Newton approach. These scattering terms are naturally admitted when we construct the sensitivity kernel by varying not the current but the to-be-updated model at each iteration. Linear and/or non-linear inverse scattering methodologies allow these additional sensitivity contributions to be computed from the current data residuals within any given update. We show that in the presence of pre-critical reflection data, the error in a second-order non-linear update to a background of s 0 is, in our scheme, proportional to at most (Δs/s 0 ) 3 in the actual parameter jump Δs causing the reflection. In contrast, the error in a standard Gauss–Newton FWI update is proportional to (Δs/s 0 ) 2 . For numerical implementation of more complex cases, we introduce a non-linear frequency-domain scheme, with an inner and an outer loop. A perturbation is determined from the data residuals within the inner loop, and a descent direction based on the resulting non-linear sensitivity kernel is computed in the outer loop. We examine the response of this non-linear FWI using acoustic single-parameter synthetics derived from the Marmousi model. Lastly, the inverted results vary depending on data frequency ranges and initial models, but we conclude that the non-linear FWI has the capability to generate high-resolution model estimates in both shallow and deep regions, and to converge rapidly, relative to a benchmark FWI approach involving the standard gradient.

58 GEOSCIENCES↗

On Boundary-Value Problems for RANS Equations and Two-Equation Turbulence Models

Currently, in engineering computations for high Reynolds number turbulent flows, turbulence modeling continues to be the most frequently used approach to represent the effects of turbulence. Such models generally rely on solving either one or two transport equations along with the Reynolds-Averaged Navier–Stokes (RANS) equations. The solution of the boundary-value problem of any system of partial differential equations requires the complete delineation of the equations and the boundary conditions, including any special restrictions and conditions. In the literature, such a description is often incomplete, neglecting important details related to the boundary conditions and possible restrictive conditions, such as how to ensure satisfying prescribed values of the dependent variables of the transport equations in the far field of a finite domain. In this article, we discuss the possible influence of boundary values, as well as near-field and far-field behavior, on the solution of the RANS equations coupled with transport equations for turbulence modeling. In so doing, we defne the concept of a welldefined boundary-value problem. Additionally, a three-dimensional, rather than a simpler one-dimensional analysis is performed to analyze the near-wall and far-field behavior of the turbulence model variables. This allows an assessment of the decay rate of these variables required to realize the boundary conditions in the far field. This paper also addresses the impact of various transformations of two-equation models (e.g., the model of Wilcox) to remove the singular behavior of the dissipation rate (ω) at the surface boundary. Finally, the issue of well-posedness regarding the governing equations is considered. A compelling argument (although not a proof) for ill-posedness is made for both direct and inverse problems.

Navier–Stokes equations↗

Robustness of Vegetation Optical Depth Retrievals Based on L-Band Global Radiometry

Microwave vegetation optical depth (VOD) and soil moisture (SM) can be simultaneously retrieved based on L-band radiometry with polarization information. VOD is indicative of the vegetation water content (VWC) because it captures the extinction of land surface emission. If the connectivity of VOD to VWC is robust, the pair of VWC-SM observations can be viable bases for understanding soil–plant–atmosphere water relations, providing new perspectives on ecosystem science. Simultaneous SM–VOD retrievals are feasible by inverting the τ−ω model with two independent datasets in dual-channel algorithms. However, given correlated satellite vertical and horizontal brightness temperatures (TBs; TB v and TB h ), an ill-posed inverse problem arises where TB errors result in high uncertainties of retrievals. In this study, we apply the degrees-of-information (DoI) metric and propose a signal-to-noise ratio (SNR) metric to assess the “retrievability” of VOD given the Soil Moisture Active Passive (SMAP) TB v –TB h linear dependence. The application of these metrics allows determining where the VOD retrievals are robust and reliable. This is a necessary step in supporting the applications of VOD in ecology and hydrology. Results show that regions with mainly nonwoody vegetation have the best potential for VOD retrievals, though regularization is necessary. We then assess VOD time variations from two regularization products that reduce the impact of underdetermined inversions: the L3 dual-channel algorithm (L3-DCA) and the multitemporal dual-channel algorithm (MTDCA), which constrain VOD time dynamics with and without using a priori VOD climatology, respectively. Though they both reduce noise, especially in the VOD retrievals, they result in differences in VOD seasonal amplitude and coupling to SM at high frequencies as we outline here.

Microwave↗

Probabilistic constrained Bayesian inversion for transpiration cooling

Abstract To enable safe operations in applications such as rocket combustion chambers, the materials require cooling to avoid material damage. Here, transpiration cooling is a promising cooling technique. Numerous studies investigate possibilities to simulate and evaluate the complex cooling mechanism. One naturally arising question is the amount of coolant required to ensure a safe operation. To study this, we introduce an approach that determines the posterior probability distribution of the Reynolds number using an inverse problem and constraining the maximum temperature of the system under parameter uncertainties. Mathematically, this chance inequality constraint is dealt with by a generalized polynomial chaos expansion of the system. The posterior distribution will be evaluated by different Markov chain Monte Carlo based methods. A novel method for the constrained case is proposed and tested among others on two‐dimensional transpiration cooling models.

97 MATHEMATICS AND COMPUTING↗

Composite Bond Line Measurements Based on A Bayesian Analysis of Flash Thermography Data

For bonded composite materials, an accurate characterization of the adhesive bond line is needed to predict failure modes and fracture toughness. In this paper, bond line thickness was estimated from data obtained using through transmission flash thermography. The forward model that predicts back surface temperature is based on a three layer heat diffusion equation with varying diffusivity and flux boundary conditions. The corresponding inverse problem of estimating bond line thickness from measurement data was solved using a Bayesian approach that assumed Gaussian priors for the bond line thickness and thermal diffusivity of the adherends. Finally, the outputs of the thermography based method were compared to measurements that were collected using a micrometer and ultrasound testing.

Flash thermography↗

Bayesian Inference and the Effects of Varying Uncertainty Models in Charring Ablator Calibration and Uncertainty Quantification Problems

The Mars Science Laboratory (MSL) vehicle utilized a heat shield constructed from NASA’s Phenolic-Impregnated Carbon Ablator (PICA) material to protect the main structure from the high enthalpy environment encountered during hypersonic atmospheric entry. During the vehicle’s descent through Martian atmosphere, multiple thermocouples embedded within the heat shield captured in-depth material temperature data that allow for studies to be conducted on current material response reconstruction tools. In the present work, material temperature data obtained from thermocouples within the MISP-4 plug (MEDLI (Mars Science Laboratory Entry, Descent, and Landing Instrument) Integrated Sensor Plug) are utilized in the calibration of Theoretical Ablative Composite for Open Testing (TACOT) model parameters in conjunction with NASA’s Porous material Analysis Toolbox (PATO) through Bayesian inference where uncertainty due to parametric, modeling, and experimental sources is simultaneously quantified. Prior to the study, a sensitivity analysis is performed through computation of the robust Sobol indices in an effort to study the relationship between input space and model response and to reduce the dimensionality of the statistical inverse problem. The Bayesian inference methodology necessitates an a-priori choice to be made for the uncertainty model for which numerous possibilities are available. Across most works, however, only basic additive or multiplicative models are utilized with pre-defined magnitudes of uncertainty based on a-priori knowledge or to-be-calibrated multipliers of static covariance matrix structures. The present effort explores the effects of informed uncertainty models, ones with temporal dependence that are simultaneously calibrated through Bayesian inference, on calibrated results for parameters that make up the uncertain input space.

Sensitivity Analysis↗

Tuning a RANS k-e model for jet-in-crossflow simulations.

We develop a novel calibration approach to address the problem of predictive k e RANS simulations of jet-incrossflow. Our approach is based on the hypothesis that predictive k e parameters can be obtained by estimating them from a strongly vortical flow, specifically, flow over a square cylinder. In this study, we estimate three k e parameters, C%CE%BC, Ce2 and Ce1 by fitting 2D RANS simulations to experimental data. We use polynomial surrogates of 2D RANS for this purpose. We conduct an ensemble of 2D RANS runs using samples of (C%CE%BC;Ce2;Ce1) and regress Reynolds stresses to the samples using a simple polynomial. We then use this surrogate of the 2D RANS model to infer a joint distribution for the k e parameters by solving a Bayesian inverse problem, conditioned on the experimental data. The calibrated (C%CE%BC;Ce2;Ce1) distribution is used to seed an ensemble of 3D jet-in-crossflow simulations. We compare the ensemble's predictions of the flowfield, at two planes, to PIV measurements and estimate the predictive skill of the calibrated 3D RANS model. We also compare it against 3D RANS predictions using the nominal (uncalibrated) values of (C%CE%BC;Ce2;Ce1), and find that calibration delivers a significant improvement to the predictive skill of the 3D RANS model. We repeat the calibration using surrogate models based on kriging and find that the calibration, based on these more accurate models, is not much better that those obtained with simple polynomial surrogates. We discuss the reasons for this rather surprising outcome.

Lefantzi, Sophia↗

Simultaneous time-optimal control of the inversion of two spin-(1/2) particles

We analyze the simultaneous time-optimal control of two-spin systems. The two noncoupled spins, which differ in the value of their chemical offsets, are controlled by the same magnetic fields. Using an appropriate rotating frame, we restrict the study to the case of opposite shifts. We then show that the optimal solution of the inversion problem in a rotating frame is composed of a pulse sequence of maximum intensity and is similar to the optimal solution for inverting only one spin by using a nonresonant control field in the laboratory frame. An example is implemented experimentally using nuclear magnetic resonance techniques.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

A persistent adjoint method with dynamic time-scaling and an application to mass action kinetics

In this article, we consider an optimization problem where the objective function is evaluated at the fixed-point of a contraction mapping parameterized by a control variable, and optimization takes place over this control variable. Since the derivative of the fixed-point with respect to the parameter can usually not be evaluated exactly, an adjoint dynamical system can be used to estimate gradients. Using this estimation procedure, the optimization algorithm alternates between derivative estimation and an approximate gradient descent step. We analyze a variant of this approach involving dynamic time-scaling, where after each parameter update the adjoint system is iterated until a convergence threshold is passed. Here, we prove that, under certain conditions, the algorithm can find approximate stationary points of the objective function. We demonstrate the approach in the settings of an inverse problem in chemical kinetics, and learning in attractor networks.

97 MATHEMATICS AND COMPUTING↗

2022 AI Testbed Expeditions Report

By exploiting the coherent properties of a light source, coherent diffraction imaging (CDI) is able to obtain the sample image at a nanoscale resolution using the measured diffraction pattern. Bragg Coherent Diffraction Imaging (BCDI) has become valuable for recovering the displacement and strain field of crystals, providing a valuable tool in material science and solid-state physics. X-ray ptychography is another emerging CDI technique that can produce a high-resolution image of the extended sample and has become popular in many research areas (e.g., materials science, biology, electronics, and optics characterization). CDI including BCDI and ptychography has become an established technique in Synchrotron Facilities including the Advanced Photon Source (APS) and will greatly benefit from the 100x coherent flux increase of the upcoming APS Upgrade (APSU). The current image formation process in CDI employs iterative phase retrieval algorithms, which is a time-consuming and computationally expensive process. Especially after APSU, the traditional iterative methods will not be able to match the experimental data acquisition speed. We employ deep learning (DL) approach to replace the iterative approaches, therefore allowing hundreds of times faster recovery of the object. We developed AutoPhaseNN, a DL-based approach which learns to solve the inverse problem without labeled data. Taking 3D BCDI as a representative technique, AutoPhaseNN has been demonstrated to be one hundred times faster than traditional iterative phase retrieval methods while providing comparable image quality. The current network is trained with 64 x 64 x 64 data size, to achieve higher resolution imaging, we will need to scale the network to input and train/infer 3D arrays of size 256 x 256 x 256 (today) and of size 2560x2560x2560 (APSU). However, the scalability of the network is restricted due to the memory-intensive training process. To perform the training for a 256 x 256 x 256 data size, the required memory exceeds the capacity of the current machine. In this project, we explore using Sambanova system to train the network for the direct data inversion for CDI.

36 MATERIALS SCIENCE↗