Quasi-Adam: Accelerating Adam using quasi-Newton approximations
Explore the source record for details and available documents.
Mathematics and Computing↗
SEARCH · Engineering Papers
Results for “ADAM”
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.
Improving ADAM through an implicit-explicit (IMEX) time-stepping approach
The ADAM optimizer, often used in machine learning for neural network training, corresponds to an underlying ordinary differential equation (ODE) in the limit of very small learning rates. Here, this work shows that the classical ADAM algorithm is a first-order implicit-explicit (IMEX) Euler discretization of the underlying ODE. Employing the time discretization point of view, we propose new extensions of the ADAM scheme obtained by using higher-order IMEX methods to solve the ODE. Based on this approach, we derive a new optimization algorithm for neural network training that performs better than classical ADAM on several regression and classification problems.
97 MATHEMATICS AND COMPUTING↗
Advancing Agrivoltaic Modeling With ADAM
Designing an agrivoltaic system presents a complex set of tradeoffs around PV system configuration, resulting performance, agricultural needs, and system economics. Developing tools for agrivoltaic analysis can assist system designers when making decisions related to these tradeoffs. We are developing the Agrivoltaics Design and Analysis Model (ADAM) as a free, publicly available web tool for agrivoltaics economics analysis. Features of ADAM include automated calculations for available agrivoltaic cropland, user-friendly configuration changes, inter-row irradiance calculations, and integrated economic modeling of energy and non-energy revenues. We will discuss the iterative process of agrivoltaic tool development including tradeoffs between accuracy and uncertainty, feature prioritization, and ease of use driven by our multi-disciplinary stakeholder design process. Finally, we will share modeling results from existing agrivoltaics systems as a preliminary case study.
14 SOLAR ENERGY↗
Iridium-cobalt oxide synthesized via surfactant-assisted Adams fusion for efficient oxygen evolution in acidic and alkaline media
A series of iridium-cobalt (Ir-Co) oxide catalysts were synthesized using a modified surfactant-assisted Adams fusion method and evaluated for the oxygen evolution reaction (OER) in both acidic and alkaline media. The effect of varying the Ir/Co ratio was systematically studied and compared with commercial Ir black and pure IrO 2 samples. The actual elemental ratios were quantified using X-ray photoelectron spectroscopy (XPS), energy-dispersive X-ray spectroscopy (EDS), and inductively coupled plasma-mass spectrometry (ICP-MS). Among the synthesized samples, the Ir 6 Co 4 catalyst exhibited the best performance in acidic media, achieving an iR-corrected overpotential of 292 mV. In alkaline conditions, it demonstrated comparable activity to Ir black, with an iR-corrected overpotential of 263 mV. XPS and electron energy loss spectroscopy analyses revealed that increasing Co content led to a higher fraction of metallic Ir (Ir 0 ) in the catalyst. In addition, the role of Ir 3+ in enhancing OER activity was explored. A strong correlation was observed between higher Ir 3+ content and improved OER performance in acidic conditions.
Adams fusion↗
Modeling viscosity of commercial glass melts: Adam–Gibbs equation and Gaussian process regression
Not Available
36 MATERIALS SCIENCE↗
Digitization of Bashforth and Adams Table II for Axisymmetric Capillary Surface Profiles
Explore the source record for details and available documents.
Post Guillen, Donna↗
Challenges in Training PINNs: A Loss Landscape Perspective
This paper explores challenges in training Physics Informed Neural Networks (PINNs), emphasizing the role of the loss landscape in the training process. We examine difficulties in minimizing the PINN loss function, particularly due to ill conditioning caused by differential operators in the residual term. We compare gradient-based optimizers Adam, L-BFGS, and their combination Adam+L-FGS, showing the superiority of Adam+L-BFGS, and introduce a novel secondorder optimizer, NysNewton-CG (NNCG), which significantly improves PINN performance. Theoretically, our work elucidates the connection between ill-conditioned differential operators and ill-conditioning in the PINN loss and shows the benefits of combining first- and second-order optimization methods. Our work presents valuable insights and more powerful optimization strategies for training PINNs, which could improve the utility of PINNs for solving difficult partial differential equations.
Rathore, Pratik↗
DESI 2024: Constraints on physics-focused aspects of dark energy using DESI DR1 BAO data
Baryon acoustic oscillation data from the first year of the Dark Energy Spectroscopic Instrument (DESI) provide near percent-level precision of cosmic distances in seven bins over the redshift range z=0.1–4.2. Here, this paper is the follow-up to the original DESI BAO cosmology paper [A. G. Adame et al. (DESI Collaboration), arXiv:2404.03002], which considered the conventional w 0 w a cold dark matter (CDM) model. We use the novel DESI data, together with other cosmic probes, to constrain the background expansion history using some well-motivated physical classes of dark energy. In particular, we explore three physics-focused behaviors of dark energy from the equation of state and energy density perspectives: the thawing class (matching many simple quintessence potentials), emergent class (where dark energy comes into being recently, as in phase transition models), and mirage class [where phenomenologically the distance to cosmic microwave background (CMB) last scattering is close to that from a cosmological constant Λ despite dark energy dynamics]. All three classes fit the data at least as well as Λ CDM, and indeed can improve on it by Δχ 2 ≈ –5 to –17 for the combination of DESI BAO with CMB and supernova data while having one more parameter. The mirage class does essentially as well as w 0 w a CDM and exhibits moderate to strong Bayesian evidence preference with respect to Λ CDM. These classes of dynamical behaviors highlight worthwhile avenues for further exploration into the nature of dark energy.
79 ASTRONOMY AND ASTROPHYSICS↗
2021–2022 Can Do Colorado E-Bike Full-Scale Pilot Program Study
In 2021–2022, the Colorado Energy Office conducted a full-scale pilot program study on e-bike usage as part of the Can Do Colorado initiative, providing e-bikes to low-income participants across the state. A [2020 mini pilot program study](https://www.nlr.gov/transportation/secure-transportation-data/tsdc-2020-can-do-colorado-e-bike-pilot-program.html) informed the full-scale study. Both studies used pedal-assist e-bikes, which feature an electric motor and battery to help power the bike. The motor amplifies the power behind each pedal stroke, augmenting the energy you put into the bike. #### Data Collection Agency The Colorado Energy Office conducted the study in partnership with local organizations in Adams and Broomfield counties (Smart Commute Metro North), Boulder (Community Cycles), Durango (Four Corners Office for Resource Efficiency), Fort Collins (City of Fort Collins), Pueblo (Pueblo County), and Vail (Town of Vail). #### Survey Methodology Program participants received an e-bike and accessories at no cost and manually submitted travel data and feedback via the CanBikeCO smartphone app. Developed in partnership with NLR, the app used a customized version of the open-source [NLR OpenPATH platform](https://www.nlr.gov/transportation/openpath.html). #### Survey Records and Data Survey records include 170 participants. The six datasets contain up to 18 months of partially automated travel diaries, combining sensed and surveyed travel behavior data—patterns of multimodal, end-to-end, individual human mobility—as well as demographic information from participants. The number of e-bike trips and e-bike miles traveled per location are 1,560 and 4,179 for Adams and Broomfield counties; 8,481 and 27,000 for Boulder; 2,815 and 6,307 for Durango; 3,483 and 7,080 for Fort Collins; 4,022 and 14,887 for Pueblo, and 1,206 and 3,3361 for Vail.
32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
Crossing The Gap Using Variational Quantum Eigensolver: A Comparative Study
Within the evolving domain of quantum computational chemistry, the Variational Quantum Eigensolver (VQE) has been developed to explore not only the ground state but also the excited states of molecules. In this study, we compare the performance of Variational Quantum Deflation (VQD) and Subspace-Search Variational Quantum Eigensolver (SSVQE) methods in determining the low-lying excited states of $LiH$. Our investigation reveals that while VQD exhibits a slight advantage in accuracy, SSVQE stands out for its efficiency, allowing the determination of all low-lying excited states through a single parameter optimization procedure. We further evaluate the effectiveness of optimizers, including Gradient Descent (GD), Quantum Natural Gradient (QNG), and Adam optimizer, in obtaining $LiH$'s first excited state, with the Adam optimizer demonstrating superior efficiency in requiring the fewest iterations. Moreover, we propose a novel approach combining Folded Spectrum VQE (FS-VQE) with either VQD or SSVQE, enabling the exploration of highly excited states. We test the new approaches for finding all three $H_4$'s excited states. Folded Spectrum SSVQE (FS-SSVQE) can find all three highly excited states near $-1.0$ Ha with only one optimizing procedure, but the procedure converges slowly. In contrast, although Folded spectrum VQD (FS-VQD) gets highly excited states with individual optimizing procedures, the optimizing procedure converges faster.
Chen, I-Chi↗
Exploring the Structure–Function Relationship in Iridium–Cobalt Oxide Catalyst for Oxygen Evolution Reaction across Different Electrolyte Media
Renewable hydrogen generation from water electrolysis offers a viable path to decarbonization if the costs can be reduced. The iridium-based anode catalyst is one of the most expensive components in electrolyzers. We propose reducing iridium usage by substituting Ir with Co, a more affordable metal, in the mixed oxide phase to enhance the catalytic activity while minimizing Ir consumption. A modified surfactant-assisted Adams fusion synthesis technique was developed as a scalable method for producing IrCo oxide nanoparticles. The synthesized material outperforms the commercial baseline, iridium oxide with carbon (IrOx_C), in both acidic and alkaline media. Acid etching (IrCo_ae) further enhances activity by selectively removing Co to expose more active sites. IrCo_ae achieved a significantly lower overpotential at 10 mA/cm 2 compared to IrOx_C, with reductions of approximately 18% under acidic conditions and 14% under alkaline conditions. This work demonstrates that the proposed synthesis method enables efficient Ir utilization and can be adapted to enhance catalyst stability for renewable hydrogen production.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Data Associated with "Identifying Suitable Front Contacts for High-Efficiency Cd(Se,Te) Solar Cells on Space-Qualified Cover Glass"
This is data associated with the publication " Identifying Suitable Front Contacts for High-Efficiency Cd(Se,Te) Solar Cells on Space-Qualified Cover Glass " by Aesha P. Patel, Ryan Muzzio, Matthew R. Young, Robert Morrissey, Suresh Chaulagain, B. Edward Sartor, Prabodika N. Kaluarachchi, Christian Velez, Joshua A. Brown, Joel N. Duenow, Stephen Glynn, Michael J. Heben, Zhaoning Song, Nikolas J. Podraza, Adam B. Phillips, Randy J. Ellingson, Matthew O. Reese. All data associated with each figure in the manuscript and supplementary should be available in this dataset. A readme file is also included to provide some guidance. Abstract: Deployment of photovoltaics in space requires devices that combine high-efficiency, low areal mass, and resilience to harsh environments. Historically, high-efficiency multijunction III–V materials have dominated space power systems; however, their high cost and limited manufacturing throughput motivate the exploration of scalable alternatives. While CdTe-based thin-film photovoltaics offer an attractive option, their performance on non-conventional substrates can suffer from front-contact instability under higher-temperature processing. Here, the role of front-contact chemistry in limiting cell performance is investigated using CdTe-based devices fabricated on 150 μm thick Ceria-doped space-qualified 0214 Corning glass. A matrix of four transparent conducting oxides (TCOs: CTO, AZO, ITO, IZO) combined with two n-type emitters (MZO, IGO) reveals chemical stability at the front interface—rather than absorber composition alone—governs recombination losses, voltage deficits, and device reproducibility. Chemically stable front-contact combinations suppress elemental diffusion and interfacial degradation, resulting in significantly improved carrier lifetimes and junction quality. These insights are validated through record-certified Cd(Se,Te) cell efficiencies of 18.4% under AM1.5G and 16.2% under AM0 illumination on ultra-thin glass. Beyond CdTe, this work provides a general framework for the rational selection of TCO/emitter interfaces in superstrate thin-film photovoltaics, including emerging technologies like metal halide perovskites, while enabling high-efficiency, lightweight photovoltaics for space applications.
14 SOLAR ENERGY↗
Optimizing the optimizer for physics-informed neural networks and Kolmogorov-Arnold networks
Physics-Informed Neural Networks (PINNs) have revolutionized the computation of PDE solutions by integrating partial differential equations (PDEs) into the neural network’s training process as soft constraints, becoming an important component of the scientific machine learning (SciML) ecosystem. More recently, physics-informed Kolmogorv-Arnold networks (PIKANs) have also shown to be effective and comparable in accuracy with PINNs. In their current implementation, both PINNs and PIKANs are mainly optimized using first-order methods like Adam, as well as quasi-Newton methods such as BFGS and its low-memory variant, L-BFGS. However, these optimizers often struggle with highly nonlinear and non-convex loss landscapes, leading to challenges such as slow convergence, local minima entrapment, and (non)degenerate saddle points. In this study, we investigate the performance of Self- Scaled BFGS (SSBFGS), Self-Scaled Broyden (SSBroyden) methods and other advanced quasi-Newton schemes, including BFGS and L-BFGS with different line search strategies. These methods dynamically rescale updates based on historical gradient information, thus enhancing training efficiency and accuracy. We systematically compare these optimizers – using both PINNs and PIKANs – on key challenging PDEs, including the Burgers, Allen-Cahn, Kuramoto-Sivashinsky, Ginzburg-Landau, and Stokes equations. Additionally, we evaluate the performance of SSBFGS and SSBroyden for Deep Operator Network (DeepONet) architectures, demonstrating their effectiveness for data-driven operator learning. Our findings provide state-of-the-art results with orders-of-magnitude accuracy improvements without the use of adaptive weights or any other enhancements typically employed in PINNs. More broadly, our work reveal insights into the effectiveness of quasi-Newton optimization strategies in significantly improving the convergence and accurate generalization of PINNs and PIKANs.
97 MATHEMATICS AND COMPUTING↗
Fast-RF-Shimming: Accelerate RF shimming in 7T MRI using deep learning
Ultrahigh field (UHF) Magnetic Resonance Imaging (MRI) offers an elevated signal-to-noise ratio (SNR), enabling exceptionally high spatial resolution that benefits both clinical diagnostics and advanced research. However, the jump to higher fields introduces complications, particularly transmit radiofrequency (RF) field ($B^{+}_{1}$) inhomogeneities, manifesting as uneven flip angles and image intensity irregularities. These artifacts can degrade image quality and impede broader clinical adoption. Traditional RF shimming methods, such as Magnitude Least Squares (MLS) optimization, effectively mitigate $B^{+}_{1}$ inhomogeneity, but remain time-consuming. Recent machine learning approaches, including RF Shim Prediction by Iteratively Projected Ridge Regression and other deep learning architectures, suggest alternative pathways. Although these approaches show promise, challenges such as extensive training periods, limited network complexity, and practical data requirements persist. In this paper, we introduce a holistic learning-based framework called Fast-RF-Shimming, which achieves a 5000 × speed-up compared to the traditional MLS method. In the initial phase, we employ random-initialized Adaptive Moment Estimation (Adam) to derive the desired reference shimming weights from multi-channel $B^{+}_{1}$ fields. Next, we train a Residual Network (ResNet) to map $B^{+}_{1}$ fields directly to the ultimate RF shimming outputs, incorporating the confidence parameter into its loss function. Finally, we design Non-uniformity Field Detector (NFD), an optional post-processing step, to ensure the extreme non-uniform outcomes are identified. Comparative evaluations with standard MLS optimization underscore notable gains in both processing speed and predictive accuracy, which indicates that our technique shows a promising solution for addressing persistent inhomogeneity challenges.
Deep learning↗
Quantum solver for single-impurity Anderson models with particle-hole symmetry
Quantum embedding methods, such as dynamical mean-field theory (DMFT), provide a powerful framework for investigating strongly correlated materials. A central computational bottleneck in DMFT is in solving the Anderson impurity model (AIM), whose exact solution is classically intractable for large bath sizes. In this work, we benchmark a quantum-classical hybrid solver tailored for particle-hole symmetric AIMs, using the variational quantum eigensolver to prepare the ground state of the model with shallow quantum circuits. The solver uses shallow quantum ansätze and one set of variational parameters to prepare the ground state and its particle and hole excitations, enabling the construction of the impurity Green’s function through a continued-fraction expansion. We evaluate the performance of this approach across a few bath sizes and interaction strengths under noisy, shot-limited conditions. We compare three optimization routines (COBYLA, Adam, and L-BFGS-B) in terms of convergence and fidelity, assess the benefits of estimating a quantum-computed moment correction to the variational energies, and benchmark the approach by comparing the density of states computed from the impurity Green’s function against that obtained using a classical pipeline. Our results demonstrate the feasibility of Green’s function construction on near-term devices and establish practical benchmarks for quantum impurity solvers embedded within self-consistent DMFT loops.
Karabin, Mariia [ORNL]↗
Optimizers for stabilizing likelihood-free inference
A growing number of applications in particle physics and beyond use neural networks as unbinned likelihood ratio estimators applied to real or simulated data. Precision requirements on the inference tasks demand a high-level of stability from these networks, which are affected by the stochastic nature of training. We show how physics concepts can be used to stabilize network training through a physics-inspired optimizer. In particular, the energy conserving descent (ECD) optimization framework uses classical Hamiltonian dynamics on the space of network parameters to reduce the dependence on the initial conditions while also stabilizing the result near the minimum of the loss function. We develop a version of this optimizer known as , which has few free hyperparameters with limited ranges guided by physical reasoning. We apply to representative likelihood-ratio estimation tasks in particle physics and find on average that it out-performs the widely used Adam optimizer. We expect that ECD will be a useful tool for wide array of data-limited problems, where it is computationally expensive to exhaustively optimize hyperparameters and mitigate fluctuations with ensembling.
Monte Carlo methods↗
Directed Flow and the Chiral Magnetic Effect in Ultrarelativistic Collisions with CMS at the LHC (Final Technical Report for DOE Grant # DE-SC0021080)
This Final Technical Report covers the years of activity for the Grant DE-SC0021080. It reflects the efforts of Alice Mignerey, the PI, Timothy Koeth, the co-PI, graduate students Eric Adams and Samuel Lascio, and undergraduates Alyssa Marinelli and Max Bernard. Results from three separate aspects of the project are presented: 1) understanding the radiation damage of the positive and negative sides of the Spectator Reaction Plane detector (SRPD) for data from the 2018 PbPb run at the CERN LHC. 2) Results of the directed flow in PbPb collisions at $\sqrt{SNN}$ = 5.02 TeV. 3) Activation studies of the sectors of the positive side of the SRPD and radiation damage studies on fused quartz using the Maryland 4-MeV electron accelerator. The most significant result is that the reaction plane derived directed flow for both participant and spectator reaction planes is many times larger than that obtained by ALICE.
73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
Directed Flow and the Chiral Magnetic Effect in Ultrarelativistic Collisions with CMS at the LHC (Final Technical Report for DOE Grant # DE-SC0021080)
This Final Technical Report covers the years of activity for the Grant DE-SC0021080. It reflects the efforts of Alice Mignerey, the PI, Timothy Koeth, the co-PI, graduate students Eric Adams and Samuel Lascio, and undergraduates Alyssa Marinelli and Max Bernard. Results from three separate aspects of the project are presented: 1) understanding the radiation damage of the positive and negative sides of the Spectator Reaction Plane Detector (SRPD) for data from the 2018 PbPb run at the CERN LHC. 2) Results of the directed flow in PbPb collisions at √sN N = 5.02 TeV. 3) Activation studies of the sectors of the positive side of the SRPD and radiation damage studies on fused quartz using the Maryland 4-MeV electron accelerator. The most significant result is that the reaction plane derived directed flow for both participant and spectator reaction planes is many times larger than that obtained by ALICE.
73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗