SEARCH · Search NASA

Results for “high accelerating gradient”

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.

SourceResearch connectionFrom yearThrough year

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

At least 289 records · Page 16

All sources NASA NTRS DOE OSTI

DOE OSTI2024

RESONANT MATCHING SECTION FOR CEBAF ENERGY UPGRADE

Thomas Jefferson National Accelerator Facility (Jeffer- son Lab) currently studies the feasibility of upgrading its energy to 22 GeV. It considers addition of six more linac passes. The highest energy passes will share two new arcs designed using the Fixed-Field Alternating Gradient (FFA) technology. The FFA arcs are built using permanent combined-function magnets. They will be connected to the linacs through transition sections that will match the optics of all six passes to the linacs. With the high number of con- straints and the limited space available, we are investigating a parametric resonance technique to match the optics quasi- independently at each energy. A resonance is excited at each individual energy to selectively control its optics. The reso- nant dipole and quadrupole kick harmonics are imposed for all energies simultaneously using Panofsky corrector mag- nets placed throughout the FFA arcs. This paper presents the current progress on that transition section design

Deitrick, K.↗

DOE OSTI2023

AutoReP: Automatic ReLU Replacement for Fast Private Network Inference

The proliferation of the Machine-Learning-As-A-Service (MLaaS) market has brought to light a number of clients’ data privacy and security concerns. One promising solution is private inference (PI) techniques using cryptographic primitives. These techniques often come with high computation and communication overhead associated with the non-linear operator such as ReLU. Several approaches have been developed in reducing the number of ReLU operations, however, they either require a heuristic threshold selection or introduce significant accuracy drop. This work presents AutoReP, a gradient-based framework for non-linear operators reduction that aims to mitigate these concerns from a systematic perspective. AutoReP automates the process of discrete selection of ReLU and polynomial functions on neurons to accelerate PI applications. We also introduce distribution-aware polynomial approximation (DaPa) to accurately approximate ReLUs under given distribution, preserving model expressivity. Our experimental results demonstrate significant accuracy improvements of 6.12% (94.31%, 12.9K ReLU budget, CIFAR-10), 8.39% (74.92%, 12.9K ReLU budget, CIFAR-100), and 9.45% (63.69%, 55K ReLU budget, Tiny-ImageNet) over current state-of-the-art methods, e.g., SNL. Morever, AutoReP is applied to EfficientNet-B2 on ImageNet dataset, and achieved 75.55% accuracy with 176.1 × ReLU budget reduction.

Peng, Hongwu↗

DOE OSTI2020

Collaborative Research: Natural Organic Matter and Microbial Controls on Mobilization/Immobilization of I and Pu in Soils and Waters Affected by Radionuclide Releases in USA and Japan

In this project, the relationship between natural organic matter (NOM) and two radioactive elements that are relevant to nuclear waste disposal were studied: Plutonium (Pu) and Iodine. The human and environmental risks associated with Pu stem mainly from the very long half-lives of several of its isotopes ( 238 Pu, 88 yr; 239 Pu, 24,100 yr; 240 Pu, 6560 yr) and its radiotoxicity. Understanding Pu biogeochemical behavior in both near-field (>10-11M) and far-field scales (<10-11M) is imperative to the development of approaches for reprocessing Pu, remediation of Pu contamination and accurate assessment of risks posed by disposal practices for Pu-bearing wastes. The environmental mobility of Pu can be affected by redox potential, pH, adsorption, precipitation, complexation, colloid formation, and microbial activity, of which the first characteristic has the most profound influence. Numerous studies have shown high affinity of Pu towards NOM, as well as to mineral phases. NOM is ubiquitous in the environment, e.g., both fulvic and humic acids are able to reduce Pu(V,VI) to Pu(IV) and the redox potential of NOM is positively related to the abundance of phenolic/acidic OH groups. NOM can either facilitate or limit actinide migration, depending on specific biogeochemical conditions including pH, mineral and organic matter characteristics, etc. The other radionuclide of interest is radioiodine ( 129 I). 129I is a major by-product of nuclear fission and of serious concern to the Department of Energy (DOE) as it is among the top risk drivers at existing and potential radiowaste-contaminated sites. The risk of 129 I stems largely from its high bioconcentration factor (90% of the body’s iodine is accumulated in the thyroid), a high inventory at source terms, a very long-half life (16M years), and rapid mobility in the subsurface environment. As a consequence, 129 I has the lowest drinking water standard (1 pCi/L) among all radionuclides in the Federal Register. With a novel and sensitive gas chromatography-mass spectrometry (GC-MS) method developed in our lab, it is possible to quickly and simultaneously determine the distribution of 129 I and stable 127 I forms in environments, as low as 2 pCi/L for 129 I. This method was subsequently validated using accelerator mass spectroscopy, AMS. IO 3 - and organo-I were determined as major species in the groundwater of SRS and the Hanford Site, contrary to thermodynamic predictions that I- should be the dominant species at these sites. Mobility of 129 I was also demonstrated to depend greatly on the I species and its concentration, sediment pH, and redox state, with times to achieve equilibrium taking up to 12 weeks. Along the groundwater pathway in the F-Area of SRS, 129 I- supplied from the seepage basins was transformed to 129 IO 3 - and organo- 129 I with increasing iodine sediment sorption, causing the lower total 127 I and 129 I concentrations along the gradient transect of the waste plume. By contrast, groundwater 129 I concentrations in the wetlands (as high as 1617.3 pCi/L) were greatly elevated with respect to the source term (159.3 pCi/L). While the NOM promoted the uptake of 129 I to the wetland sediment, it also promoted the formation of soluble organic fraction. A small fraction of NOM that is bound to iodine can behave as a mobile organo-I source. Iodide was enzymatically incorporated into NOM, whereas both iodide and iodate were abiotically bound to NOM, under certain conditions. Iodate removal from the mobile aqueous phase can also occur through incorporation into carbonate (e.g., at the Hanford Site, USA). Thus immobilization and re-mobilization of iodine species were influenced by pH, Eh and the presence of NOM and metal oxides, which adds to the complexity of site remediation action. A ground-breaking result was to elucidate the products (i.e. organo-iodine moieties formed via enzymatic and non-enzymatic processes) at the molecular level by nuclear magnetic resonance (NMR) and electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry (ESI-FTICRMS). We found that iodine-NOM interactions may be influenced by NOM hydrophobic aliphatic moieties. From the perspective of ESI-FTICRMS, organo-iodine formulas were ascribed to the groups of unsaturated hydrocarbons, lignins and proteins. Iodate is likely abiotically reduced to reactive iodine species by lignin- and tannin-like compounds or carboxylic-rich alicyclic molecules (CRAM). We also investigated microbial mechanisms in iodine incorporation into NOM. We established that soil bacteria isolated from F-Area of SRS did not accumulate significant amounts of I- (0.2-2%). Intracellular uptake of I- decreases with increasing pH when pH ranged from 4 to 6. In contrast, 44 out of 84 strains isolated from the F-Area of SRS can transform I- to IO 3 - and organo-iodine. In some cases, oxidation was facilitated in the presence of H 2 O 2 . Microbes can also excrete organic acids that enhance I- oxidation by lowering the ambient pH and reacting with H 2 O 2 to form peroxy carboxylic acids. At lower pH values (≤5), H 2 O 2 hydrolysis was the driving force for iodide-oxidation; whereas, at pH ≥ 6, spontaneous decomposition of peroxy carboxylic acids, originating from H 2 O 2 and organic acids were the primary cause of iodide oxidation. Lastly, it was determined that microbial processes involved in Mn (II) are capable of directly oxidizing I- via enzymatic catalysis (i.e., multicopper oxidases), or indirectly through the formation of reactive oxygen species (ROS) and/or biogenic manganese oxides. ROS-mediated oxidation of I- was found to predominate at pH >5, whereas the enzymatic and Mn oxide pathways were more active at pH < 5. Together, this project has resulted in 9 publications in high-impact journals, and the training of 1 Ph.D and 4 undergraduate students.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2024

Quantitative Insight to Fission Gas Pores Distribution in Irradiated Annular U-10Zr Metallic Fuel Using Machine Learning

Metallic fuels, particularly U-10Zr and its performance in reactor irradiation conditions, have been thoroughly investigated and are a promising candidate for next-generation sodium-cooled fast spectrum nuclear reactors. Irradiation in reactors can lead to the formation of fission gas and increased pore formation which can significantly impact fuel performance. Due to the large number of pores and various phases formed in metallic fuel during irradiation, a quantitative description of fission gas pores as a function of irradiation conditions is not yet available, undermining the fidelity of fuel performance modeling to support fuel qualification. It has been difficult to clearly detect pore boundaries and distinguish matrix phases from fission gas pores using optical microscopy by using simple threshold methods working with low magnification images. The pre-trained deep learning model for fission gas pore detection was applied to ~10,260 high magnification scanning electron microscopy images. The model increased the accuracy of fission gas pore segmentation to obtain statistical features, which cannot be processed manually. A pre-trained decision tree model was used to classify pores as isolated or connected pores, providing new insight into the correlation between the movement of lanthanides, solid fission products, and the radial temperature gradient developed in fuel irradiation conditions. This paper emphasizes the potential that artificial intelligence-based machine learning models have to accelerate qualification and support nuclear fuel development.

36 MATERIALS SCIENCE↗

DOE OSTI2021

Elemental segregation to lattice defects in the CrMnFeCoNi high-entropy alloy during high temperature exposures

The influence of small plastic pre-strains on the elevated-temperature stability and microstructure of the equiatomic CrMnFeCoNi FCC solid solution is investigated. Particular attention is given to whether any of the alloy elements segregate to individual dislocations. To that end, CrMnFeCoNi samples were first deformed in tension at room temperature to plastic strains of 0.2 and 2.3%, and subsequently annealed at 973 K for 800 hours. The pre-strains activated planar slip of 1/2<110>-type dislocations on {111}-type glide planes. Interactions of this planar slip with special Σ3 grain boundaries formed a large number of dislocation segments with a <110>-type crystallographic orientation suitable for a credible end-on analysis of dislocation cores in HR-STEM. The cores of the 1/2<110> dislocations pushed up against the investigated grain boundaries were found to be close to the compact configuration. Within the sensitivity of the Super-X EDS mapping, no concentration gradient was detected near dislocations that would indicate enrichment at dislocation cores of any of the elemental constituents of the alloy after the pre-deformation and annealing. However, a Cr-rich tetragonal sigma phase nucleated and grew at grain boundary triple junctions during this anneal, processes that were not accelerated by the enhanced dislocation density present after pre-strain. A clear chromium gradient was observed in the Cr-depleted zones near grain boundaries suggesting that Cr transport occurred by relatively slow diffusion from the bulk to the grain boundaries and then by relatively fast diffusion along the grain boundaries to the precipitates. Accompanying the Cr depletion near grain boundaries is a simultaneous Ni and Mn enrichment, which promotes formation of the L1 0 NiMn phase that is observed on the grain boundaries after prolonged annealing.

36 MATERIALS SCIENCE↗

DOE OSTI2023

CFD-based design optimization of ducted hydrokinetic turbines

Abstract Hydrokinetic turbines extract kinetic energy from moving water to generate renewable electricity, thus contributing to sustainable energy production and reducing reliance on fossil fuels. It has been hypothesized that a duct can accelerate and condition the fluid flow passing the turbine blades, improving the overall energy extraction efficiency. However, no substantial evidence has been provided so far for hydrokinetic turbines. To investigate this problem, we perform a CFD-based optimization study with a blade-resolved Reynolds-averaged Navier–Stokes (RANS) solver to explore the design of a ducted hydrokinetic turbine that maximizes the efficiency of energy extraction. A gradient-based optimization approach is utilized to effectively deal with the high-dimensional design space of the blade and duct geometry, with gradients being calculated through the adjoint method. The final design is re-evaluated through higher-fidelity unsteady RANS (URANS) simulations. Our optimized ducted turbine achieves an efficiency of about 54% over a range of operating conditions, higher than the typical 46% efficiency of unducted turbines.

13 HYDRO ENERGY↗

DOE OSTI2024

Reduced-dimension Bayesian optimization for model calibration of transient vapor compression cycles

Development and calibration of first-principles dynamic models of vapor compression cycles (VCCs) is of critical importance for applications that include control design and fault detection and diagnostics. Nevertheless, the inherent complexity of models that are represented by large systems of differential–algebraic equations leads to significant challenges for model calibration processes that utilize classical gradient-based methods. Bayesian optimization (BO) is a sample-efficient and gradient-free approach using a probabilistic surrogate model and optimal search over a feasible parameter space. Despite the benefits of BO in reducing computational costs, challenges remain in dealing with a high-dimensional calibration task resulting from a large set of parameters that have significant impacts on system behavior and need to be calibrated simultaneously. This paper presents a reduced-dimension BO framework for calibrating transient VCCs models where the calibration space is projected to a low-dimensional subspace for accelerating convergence of the solution algorithm and consequently reducing the number of transient simulations. The proposed approach was demonstrated via two case studies associated with different VCC applications where 10 parameters were calibrated in each case using laboratory measurements. The reduced-dimension BO framework only required 1 / 8 th of the iterations associated with a standard BO method that deals with high-dimensional calibration parameters for converged solutions and yielded comparable accuracy. Furthermore, both calibrated models revealed significant accuracy improvements compared to uncalibrated models.

Ma, Jiacheng↗

DOE OSTI2020

Steady detonation propagation in thin channels with strong confinement

We examine asymptotically the dynamics of two-dimensional, steady detonation wave propagation and failure for a strongly confined high explosive (HE), in which the width of the explosive is small relative to the reaction zone length. An energy balance equation is derived, which shows how the longitudinal acceleration of subsonic flow behind the detonation shock is influenced both by chemical reaction and by the effects of HE boundary streamline deflection, specifically via the induced rate of change of mass flux through the detonation wave. The latter serves to either counteract or reinforce the acceleration of longitudinal flow, depending on the sign of the gradient of the boundary streamline deflection at the detonation shock. The analysis is valid for general equations of state and chemical reaction rates in the HE. The asymptotically derived form of the energy equation represents an eigenvalue problem for the determination of the steady detonation propagation speed, solved via a shooting method. We explore specific results for ideal and stiffened equations of state, along with a pressure-dependent reaction rate for which changes in the pressure exponent and reaction order are also studied. We consider the influences of both straight and curved HE boundary streamline shapes. The asymptotic analysis reveals significant physical insights into how detonation propagation and failure are affected by strong confinement.

42 ENGINEERING↗

DOE OSTI2024

Excellent performance of 650 MHz single-cell niobium cavity after electropolishing

Electropolishing process and cathodes have undergone modification and optimization for both low- and high-beta 650 MHz five-cell niobium cavities for PIP-II. Cavities treated with these modified electropolishing conditions exhibited smooth surfaces and good performance in baseline tests. Nonetheless, due to administrative constraints on project cavities, maximum gradient performance testing was not conducted. This paper presents a study conducted on a single-cell 650 MHz cavity utilizing the optimized electropolishing conditions, highlighting the maximum performance attained for this specific cavity. The cavity tested at 2 K in a vertical cryostat reached a superior accelerating field gradient of 53.3 MV/m at Q0 of 1.6x1010, which is the highest gradient attained for this type of large-sized cavities.

43 PARTICLE ACCELERATORS↗

DOE OSTI2023

Synchrotron Polarization Signatures of Surface Waves in Supermassive Black Hole Jets

Abstract Supermassive black holes in active galactic nuclei are known to launch relativistic jets, which are observed across the entire electromagnetic spectrum and thought to be efficient particle accelerators. Their primary radiation mechanism for radio emission is polarized synchrotron emission produced by a population of nonthermal electrons. In this Letter, we present a global general relativistic magnetohydrodynamical (GRMHD) simulation of a magnetically arrested disk (MAD). After the simulation reaches the MAD state, we show that waves are continuously launched from the vicinity of the black hole and propagate along the interface between the jet and the wind. At this interface, a steep gradient in velocity is present between the mildly relativistic wind and the highly relativistic jet. The interface is, therefore, a shear layer, and due to the shear, the waves generate roll-ups that alter the magnetic field configuration and the shear layer geometry. We then perform polarized radiation transfer calculations of our GRMHD simulation and find signatures of the waves in both total intensity and linear polarization, effectively lowering the fully resolved polarization fraction. The telltale polarization signatures of the waves could be observable by future very long baseline interferometric observations, e.g., the next-generation Event Horizon Telescope.

Astronomy & Astrophysics↗

DOE OSTI2024

Progress and Challenges in Traveling-Wave SRF cavity development at Fermilab

Traveling-wave (TW) technology can push the accelerator field gradient of niobium SRF cavity to 70MV/m or higher beyond the limit of 50~60MV/m in Standing-wave (SW) technology. The early stages of TW SRF cavity developments had been funded by several SBIR grants to Euclid Techlabs and completed in collaboration with Fermilab through a 1-cell prototype and a proof-of-principle 3-cell TW cavity. The TW resonance excitation in the 3-cell TW cavity at 2K was demonstrated through the low power RF test in early 2024. A high-power test of the 3-cell in TW mode being prepared. To advance a design and technology to fabricate a novel high gradient TW SRF cavity, FNAL proposed a half-meter TW RF design and R&Ds to realize that are in progress. Here we will report the recent progress in the 3-cell TW cavity and the challenges towards a half-meter scale TW cavity.

Furuta, F.↗

DOE OSTI2022

Fed-DeepONet: Stochastic Gradient-Based Federated Training of Deep Operator Networks

The Deep Operator Network (DeepONet) framework is a different class of neural network architecture that one trains to learn nonlinear operators, i.e., mappings between infinite-dimensional spaces. Traditionally, DeepONets are trained using a centralized strategy that requires transferring the training data to a centralized location. Such a strategy, however, limits our ability to secure data privacy or use high-performance distributed/parallel computing platforms. To alleviate such limitations, in this paper, we study the federated training of DeepONets for the first time. That is, we develop a framework, which we refer to as Fed-DeepONet, that allows multiple clients to train DeepONets collaboratively under the coordination of a centralized server. To achieve Fed-DeepONets, we propose an efficient stochastic gradient-based algorithm that enables the distributed optimization of the DeepONet parameters by averaging first-order estimates of the DeepONet loss gradient. Then, to accelerate the training convergence of Fed-DeepONets, we propose a moment-enhanced (i.e., adaptive) stochastic gradient-based strategy. Finally, we verify the performance of Fed-DeepONet by learning, for different configurations of the number of clients and fractions of available clients, (i) the solution operator of a gravity pendulum and (ii) the dynamic response of a parametric library of pendulums.

Moya, Christian↗

DOE OSTI2024

Progress and Challenges in Traveling-Wave SRF cavity development at Fermilab

Furuta, F.↗

DOE OSTI2023

Ultrafast Wavefront Shaping via Space-Time Refraction

A myriad of metasurfaces have been demonstrated that manipulate light by spatially structuring thin optical layers. Manipulation of the optical properties of such layers in both space and time can unlock new physical phenomena and enable new optical devices. Examples include photon acceleration and frequency conversion, which modifies Snell’s relation to a more general, nonreciprocal form. Here, we combine theory and experiment to realize wavefront shaping and frequency conversion on subpicosecond time-scales by inducing space-time refractive index gradients in epsilon-near-zero (ENZ) films with femtosecond light pulses. Furthermore, we experimentally tune wavefront steering by controlling the incident angle of the beams and the pump–probe delay without the need for nanostructure fabrication. As a demonstration of this approach, we leverage the ultrafast, high-bandwidth optical response of transparent oxides in their ENZ wavelength range to create large refractive index gradients and new types of nonreciprocal, ultrafast two-dimensional (2D) optics, including an ultrathin transient lens.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Application of Machine Learning and Data Augmentation Algorithms in the Discovery of Metal Hydrides for Hydrogen Storage

The development of efficient and sustainable hydrogen storage materials is a key challenge for realizing hydrogen as a clean and flexible energy carrier. Among various options, metal hydrides offer high volumetric storage density and operational safety, yet their application is limited by thermodynamic, kinetic, and compositional constraints. In this work, we investigate the potential of machine learning (ML) to predict key thermodynamic properties—equilibrium plateau pressure, enthalpy, and entropy of hydride formation—based solely on alloy composition using Magpie-generated descriptors. We significantly expand an existing experimental dataset from ~400 to 806 entries and assess the impact of dataset size and data augmentation, using the PADRE algorithm, on model performance. Models including Support Vector Machines and Gradient Boosted Random Forests were trained and optimized via grid search and cross-validation. Results show a marked improvement in predictive accuracy with increased dataset size, while data augmentation benefits are limited to smaller datasets and do not improve accuracy in underrepresented pressure regimes. Furthermore, clustering and cross-validation analyses highlight the limited generalizability of models across different material classes, though high accuracy is achieved when training and testing within a single hydride family (e.g., AB2). The study demonstrates the viability and limitations of ML for accelerating hydride discovery, emphasizing the importance of dataset diversity and representation for robust property prediction.

augmentation↗

DOE OSTI2025

Digital Twin Framework for PIP-II Linac: AI-Driven Multi-Scale Modeling from Ion Source to 800 MeV

The PIP-II linac will enable >1.2 MW beam power for DUNE, requiring unprecedented operational reliability across its warm front-end (RFQ, MEBT) and five distinct SRF sections operating at 162.5/325/650 MHz. We present a comprehensive digital twin framework uniquely combining a fully differentiable fast beam transport code with neural network surrogates trained on high-fidelity PIC simulations, capturing space charge and nonlinear dynamics beyond traditional envelope codes while achieving 10⁴ speedup at <1% accuracy. End-to-end differentiability enables gradient-based optimization across 500+ parameters simultaneously previously impossible with conventional tools while the model incorporates static/dynamic errors and serves as a virtual commissioning platform for diverse hardware integration. The framework facilitates reinforcement learning for pulsed/CW mode transitions, predictive maintenance through anomaly detection, and autonomous tuning algorithm development with real-time execution capability. Validation against physics simulations shows excellent agreement for the front-end, with initial results demonstrating potential for 30% commissioning time reduction and proactive fault mitigation, providing a scalable blueprint for operating next-generation high-intensity accelerators.

Pathak, Abhishek [Fermilab] (ORCID:000000021704208↗

DOE OSTI2025

Digital Twin Framework for PIP-II Linac: AI-Driven Multi-Scale Modeling from Ion Source to 800 MeV

The PIP-II linac will enable >1.2 MW beam power for DUNE, requiring unprecedented operational reliability across its warm front-end (RFQ, MEBT) and five distinct SRF sections operating at 162.5/325/650 MHz. We present a comprehensive digital twin framework uniquely combining a fully differentiable fast beam transport code with neural network surrogates trained on high-fidelity PIC simulations, capturing space charge and nonlinear dynamics beyond traditional envelope codes while achieving 10⁴× speedup at <1% accuracy. End-to-end differentiability enables gradient-based optimization across 500+ parameters simultaneously—previously impossible with conventional tools—while the model incorporates static/dynamic errors and serves as a virtual commissioning platform for diverse hardware integration. The framework facilitates reinforcement learning for pulsed/CW mode transitions, predictive maintenance through anomaly detection, and autonomous tuning algorithm development with real-time execution capability. Validation against physics simulations shows excellent agreement for the front-end, with initial results demonstrating potential for 30% commissioning time reduction and proactive fault mitigation, providing a scalable blueprint for operating next-generation high-intensity accelerators.

Pathak, Abhishek [Fermilab] (ORCID:000000021704208↗

DOE OSTI2023

An autoencoder compression approach for accelerating large-scale inverse problems

Abstract Partial differential equation (PDE)-constrained inverse problems are some of the most challenging and computationally demanding problems in computational science today. Fine meshes required to accurately compute the PDE solution introduce an enormous number of parameters and require large-scale computing resources such as more processors and more memory to solve such systems in a reasonable time. For inverse problems constrained by time-dependent PDEs, the adjoint method often employed to compute gradients and higher order derivatives efficiently requires solving a time-reversed, so-called adjoint PDE that depends on the forward PDE solution at each timestep. This necessitates the storage of a high-dimensional forward solution vector at every timestep. Such a procedure quickly exhausts the available memory resources. Several approaches that trade additional computation for reduced memory footprint have been proposed to mitigate the memory bottleneck, including checkpointing and compression strategies. In this work, we propose a close-to-ideal scalable compression approach using autoencoders to eliminate the need for checkpointing and substantial memory storage, thereby reducing the time-to-solution and memory requirements. We compare our approach with checkpointing and an off-the-shelf compression approach on an earth-scale ill-posed seismic inverse problem. The results verify the expected close-to-ideal speedup for the gradient and Hessian-vector product using the proposed autoencoder compression approach. To highlight the usefulness of the proposed approach, we combine the autoencoder compression with the data-informed active subspace (DIAS) prior showing how the DIAS method can be affordably extended to large-scale problems without the need for checkpointing and large memory.

Mathematics↗