Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Operational Efficiency”

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 1,513 records · Page 84
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

CO2 Sorption in Aminopolymer-Based Direct Air Capture Composites Through Fluorescent Detection

Aminopolymer-mesoporous oxide composites are becoming widely recognized for their ability to adsorb CO2 directly from the air. However, a better fundamental understanding of the nature of these materials under different environmental operating conditions will be necessary before they become ubiquitous in industrial direct air capture (DAC) systems. For example, gas diffusion in general is coupled with polymer segmental mobility in these composites; diffusion is therefore affected by interactions of the polymer with the pore wall of the mesoporous oxide, by the relative humidity of the air, and by electrostatic crosslinks that develop as a function of CO2 sorption. Disentangling these and additional complicated effects in order to better optimize operating conditions is a major challenge for the field. Here, we present a method for quantifying the adsorption of CO2 and moisture in a polyethylenimine (PEI) - Al2O3 composite using infrared sensing while we simultaneously monitor polymer mobility with a fluorescent probe molecule doped into the composite. Both the fluorescence intensity and shape of the emission spectra are strongly dependent on the mobility of the supporting medium. We monitor the polymer mobility and CO2 adsorption kinetics across a series of different relative humidities (RH) with this technique, and we observed that the relative difference between polymer mobility and sorption kinetics from one RH to the next are quite dramatic. We interpret these results and discuss how they can be used to inform the design of more efficient DAC systems for real-world operating conditions.

carbon capture↗
DOE OSTI2020

Reducing Cost of Chlorinated Volatile Organic Compound Remediation by Transitioning from Active to Passive Soil Vapor Extraction - 20157

Areas of high chlorinated volatile organic compound (cVOC) contamination at the Savannah River Site (SRS) have been undergoing remediation via soil vapor extraction, sometimes coupled with thermal treatments to enhance extraction rates. These active systems are effective in removing large amounts of contaminant mass from the subsurface and mitigating the impacts to groundwater. However, as extraction rates decline, costs must be evaluated with respect to the benefit of continued active operation. A decision framework for identifying conditions when a transition to a more passive remediation is appropriate has been developed with state and federal regulatory agencies. Two remediation areas have recently been transitioned from active soil vapor extraction (ASVE) to passive soil vapor extraction (PSVE) at the SRS. Performance evaluation goals including plume stabilization, overall mass removal trends, environmental sustainability and costs were considered in transitioning from active remediation to passive technologies at both sites. At the Dynamic Underground Stripping (DUS) project at the M-Area Settling Basin, ASVE was combined with steam injection to extract cVOCs during active operations. Steam injection occurred from September 2005 to September 2009. DUS utilized 63 steam injection wells, 34 active vapor extraction wells, and 3 active soil vapor extraction units (SVEUs). Two active SVEUs had 60 horsepower blowers; the third had a 25-horsepower blower. Mass removal was closely tracked during DUS operations; over 181,437 kilograms (400,000 pounds) of cVOCs were removed while active steam injection occurred. After steaming was stopped, ASVE continued. The 34 active wells were evaluated in 2012. The ASVE wells were grouped into categories of high, medium, and low extraction rates. High producing wells remained connected to a single active SVEU. Low producing wells were abandoned, and the medium producing wells were transitioned to PSVE (Microblowers{sup TM}). Microblowers{sup TM} utilize a dedicated blower per well and are solar powered. This passive technology provides energy, maintenance, and operation costs savings while still providing an efficient reduction in cVOC migration to groundwater. In 2018, the remaining ASVE wells were evaluated again. The purpose of this testing was to identify which wells removed the most mass. An optimal well configuration was determined. The criteria to discontinue ASVE was removal of less than 18 kilograms (40 pounds) per week of cVOCs. After 3 months of shutdown (rebound conditions), 2.7 kilograms (5.9 pounds) of cVOCs per week were being removed. Data from the rebound test justified ending ASVE and transitioning wells with higher extraction rates to PSVE. Wells that had depleted the cVOC mass within their zone of influence were abandoned. Performance data from existing PSVE wells justified ending PSVE at wells with depleted extraction rates. Currently the system has 16 PSVE wells operating. Another ASVE system was being used to treat cVOC contaminated soil at the A-Area Miscellaneous Rubble Pile (AMRP) at SRS. System operation began in 2004, with 7 ASVE wells connected to a 60- horsepower blower. Mass removal rates and contaminant concentrations remained consistently low over the ASVE lifespan at AMRP. This indicated that mass removal was diffusion limited. With this data, the 7 ASVE wells were transitioned to PSVE in 2017. Twelve pressure monitoring points were also transitioned to PSVE wells. AMRP currently has 19 PSVE wells operating. Both transitions from active to passive remediation had concurrence from the United States Environmental Protection Agency and the South Carolina Department of Health and Environmental Control. These transitions ensure that only the necessary amount of energy is being exerted to remediate the environment. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2020

Designing hierarchical nanoporous membranes for highly efficient gas adsorption and storage

Nanoporous membranes with two-dimensional materials such as graphene oxide have attracted attention in volatile organic compounds (VOCs) and H 2 adsorption because of their unique molecular sieving properties and operational simplicity. However, agglomeration of graphene sheets and low efficiency remain challenging. Therefore, we designed hierarchical nanoporous membranes (HNMs), a class of nanocomposites combined with a carbon sphere and graphene oxide. Hierarchical carbon spheres, prepared following Murray’s law using chemical activation incorporating microwave heating, act as spacers and adsorbents. Hierarchical carbon spheres preclude the agglomeration of graphene oxide, while graphene oxide sheets physically disperse, ensuring structural stability. The obtained HNMs contain micropores that are dominated by a combination of ultramicropores and mesopores, resulting in high VOCs/H 2 adsorption capacity, up to 235 and 352 mg/g at 200 ppmv and 3.3 weight % (77 K and 1.2 bar), respectively. Our work substantially expands the potential for HNMs applications in the environmental and energy fields.

42 ENGINEERING↗
DOE OSTI2021

Efficient photoconductive terahertz detection through photon trapping in plasmonic nanocavities

We present a photoconductive terahertz detector that employs a plasmonic nanocavity to offer high-sensitivity and broadband operation when used in a terahertz time-domain spectroscopy system even at very low optical pump power levels. By employing a plasmonic nanocavity, all of the photocarriers are generated within a 100 nm distance from the photoconductor contact electrodes, enabling a short transport time for almost all of the photogenerated carriers. As a result, the photoconductive detector maintains high quantum efficiency and ultrafast operation simultaneously, enabling high-sensitivity and broadband operation at very low optical pump power levels. We utilize a photoconductive detector based on a plasmonic nanocavity optimized for operation at a 770 nm optical wavelength in a terahertz time-domain spectroscopy system and demonstrate a 100 dB signal-to-noise ratio and a 0.1–6 THz noise-equivalent bandwidth at a record-low average optical pump power of 0.1 mW, compared to the state-of-the-art photoconductive terahertz detectors. The extremely low optical power budget of the demonstrated photoconductive detector makes this detector attractive for multi-pixel terahertz imaging systems.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Progress Towards Microcalorimeter Gamma Spectroscopy Analysis for Nuclear Fuel Separations

The High Efficiency and Resolution Microcalorimeter Spectrometer, 400 pixels (HERMES-400) is an ultra-high resolution gamma-ray spectrometer at Idaho National Laboratory’s Materials and Fuels Complex. HERMES-400 is intended to enable precise nondestructive analysis of advanced fuel cycle materials. Key upgrades and repairs of the instrument were performed this year to improve performance. First, the fourth and final microcalorimeter detector module was installed, meeting the milestone for its integration into the system. Operation with all four modules enables higher detector efficiency than any previous microcalorimeter gamma-ray system. Second, a thinner magnetic shield was installed. This shields the sensitive microcalorimeter pixels from stray magnetic fields while allowing transmission of lower-energy gamma rays, which are important to isotopic analysis. Finally, the cryostat, which provides cooling of the modules to their operating temperature, was found to have a helium coolant leak, which was repaired. HERMES-400 is in a stable state to perform repeated, long-term gamma-ray measurements with good detector efficiency and energy resolution. In FY25, irradiated TRISO fuel and salts from advanced electrochemical recycling techniques are candidates for microcalorimeter measurements. A final upgrade of the magnetic shielding and vibration damping of the system is also planned to maximize the system performance.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2023

High-Fidelity Energy Deposition Ignition Model Coupled with Flame Propagation Models at Engine-like Flow Conditions

With the heightened pressure on car manufacturers to increase the efficiency and reduce the carbon emissions of their fleets, more challenging engine operation has become a viable option. Highly dilute, boosted, and stratified charge, among others, promise engine efficiency gains and emissions reductions. At such demanding engine conditions, the spark-ignition process is a key factor for the flame initiation propagation and the combustion event. From a computational standpoint, there exist multiple spark-ignition models that perform well under conventional conditions but are not truly predictive under strenuous engine operation modes, where the underlying physics needs to be expanded. In this paper, a hybrid Lagrangian-Eulerian spark-ignition (LESI) model is coupled with different turbulence models, grid sizes, and combustion models. The ignition model, previously developed, relies on coupling Eulerian energy deposition with a Lagrangian particle evolution of the spark channel, at every time-step. The spark channel is attached to the electrodes and allowed to elongate at a speed derived from the flow velocity. The LESI model is used to simulate spark ignition in a nonquiescent crossflow environment at engine-like conditions, using converge commercial computational fluid dynamics (CFD) solver. The results highlight the consistency, robustness, and versatility of the model in a range of engine-like setups, from typical with Reynolds-averaged Navier-Stokes (RANS) and a larger grid size to high fidelity with large-eddy simulation (LES) and a finer grid size. The flame kernel growth is then evaluated against Schlieren images from an optical constant volume ignition chamber with a focus on the performance of flame propagation models, such as G-equation and thickened flame model, versus the baseline well-stirred reactor model. Finally, future development details are discussed.

Advanced ignition modeling↗
DOE OSTI2024

Estimates of lithium mass yields from produced water sourced from the Devonian-aged Marcellus Shale

Abstract Decarbonatization initiatives have rapidly increased the demand for lithium. This study uses public waste compliance reports and Monte Carlo approaches to estimate total lithium mass yields from produced water (PW) sourced from the Marcellus Shale in Pennsylvania (PA). Statewide, Marcellus Shale PW has substantial extractable lithium, however, concentrations, production volumes and extraction efficiencies vary between the northeast and southwest operating zones. Annual estimates suggest statewide lithium mass yields of approximately 1160 (95% CI 1140–1180) metric tons (mt) per year. Production decline curve analysis on PW volumes reveal cumulative volumetric disparities between the northeast (median = 2.89 X 10 7 L/10-year) and southwest (median = 5.56 × 10 7 L/10-year) regions of the state, influencing lithium yield estimates of individual wells in southwest [2.90 (95% CI 2.80–2.99) mt/10-year] and northeast [1.96 (CI 1.86–2.07) mt/10-year] PA. Moreover, Mg/Li mass ratios vary regionally, where NE PA are low Mg/Li fluids, having a median Mg/Li mass ratio of 5.39 (IQR, 2.66–7.26) and SW PA PW is higher with a median Mg/Li mass ratio of 17.8 (IQR, 14.3–20.7). These estimates indicate substantial lithium yields from Marcellus PW, though regional variability in chemistry and production may impact recovery efficiencies.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Automatic Code Generation for High-Performance Graph Algorithms

Graph problems are common across fields of scientific computing and social sciences. However, despite their importance, implementing graph algorithms effectively on modern computing systems is a challenging task that requires significant programming effort and generally results in customized implementations. Current computing and memory hierarchies are not architected for irregular computations resulting in challenges for graph algorithms to achieve high performance on those architectures. In this paper, we present GraphX, a novel compiler framework and DSL designed to simplify the development of efficient graph algorithms and achieve high performance on modern computing systems. GraphX consists of a DSL for efficient implementation of graph algorithms, various optimizations, such as support for sparse linear algebra and workspace transformations, optimized graph primitives, including semiring and masking, and a high-performance code generation engine. Using GraphX, users can implement graph algorithms using a semantically-rich language with graph-oriented operators. GraphX uses these semantics to automatically generate efficient code for target architectures, increasing performance and portability across architectures. The composable nature of GraphX makes it possible to extend the set of optimizations and architectures without modifying the source code. We demonstrate GraphX outperforms state-of-the-art graph libraries, such as LAGraph, up to $3.7 speedup in semiring operations, $2.19 speedup in an important sparse computational kernel, and $9.05 speedup in graph processing algorithms.

compiler, graph algorithms, semiring, masking, wor↗
DOE OSTI2022

Fuel property impacts on gaseous and PM emissions from a multi-mode single-cylinder engine

The U.S. Department of Energy’s Co-Optima initiative has focused on improving fuel economy and vehicle performance while reducing emissions through the simultaneous development of emerging sustainable fuels with beneficial properties and advanced combustion strategies. A major thrust has been the development of advanced compression ignition (ACI) combustion strategies of gasoline range fuels in combination with spark-ignited (SI) combustion in a single engine capable of multi-mode operation to achieve high power density with enhanced part load efficiency. The aim of this study was to further the understanding of how emissions from both ACI and SI strategies operating on the same fuels in the same engine are impacted by different fuel properties. This investigation focused on particulate matter (PM) and gaseous hydrocarbon emissions from 6 different fuels across 3 different combustion modes on the same single-cylinder engine designed for multi-mode operation: SI combustion, partial fuel stratification (PFS), and spark-assisted compression ignition (SACI). In each of these modes, 3 different CA50 phasings were studied such that all 6 fuels could be studied at the same phasings. Three of the six different fuels used were specially formulated in a previous investigation to study the impact of fuel distillation and aromatic content while maintaining the research octane number (RON) and octane sensitivity. Additionally, neat isooctane and two ethanol containing fuels (RD5-87 and Co-Optima E30) were studied. Different fuel and phasing impacts on emissions were observed across the three combustion modes. Fuel properties were found to impact soot PM and particle number more than the CA50 phasing, while the phasing had more impact on NOx emissions. The NOx emissions were reduced in the PFS mode for all fuels compared to SI combustion, but the SACI combustion mode did not reduce NOx emissions. Although PFS produced low soot PM emissions like SI, total PM mass emissions were significantly higher due to large organic carbon (OC) PM mass contribution. Both PFS and SACI had greater particle number emission than SI operation with small nuclei mode particles dominating in PFS compared to large agglomeration particles in SACI.

42 ENGINEERING↗
DOE OSTI2025

Comprehensive Evaluation of the Advanced Moderator Module (AMM)

The U.S. nuclear sector is now focusing on advancing and deploying microreactors that blend efficiency, compactness, and safety. For thermal neutron microreactors designed for high temperature operations, employing a metal hydride-based moderator presents a valuable opportunity to enhance neutron efficiency. Argonne National Laboratory (ANL) has pioneered an Advanced Moderator Module (AMM) featuring an innovative containment strategy.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2024

Probing scrambling and operator size distributions using random mixed states and local measurements

The dynamical spreading of quantum information through a many-body system, typically called scrambling, is a complex process that has proven to be essential to describe many properties of out-of-equilibrium quantum systems. Scrambling can, in principle, be fully characterized via the use of out-of-time-ordered correlation functions, which are notoriously hard to access experimentally. In this work, we put forward an alternative toolbox of measurement protocols to experimentally probe scrambling by accessing properties of the operator size probability distribution, which tracks the size of the support of observables in a many-body system over time. Our measurement protocols require the preparation of separable mixed states together with local operations and measurements, and combine the tools of randomized operations, a modern development of near-term quantum algorithms, with the use of mixed states, a standard tool in NMR experiments. We demonstrate how to efficiently probe the probability-generating function of the operator distribution and discuss the challenges associated with obtaining the moments of the operator distribution. We further show that manipulating the initial state of the protocol allows us to directly obtain the individual elements of the distribution for small system sizes. Published by the American Physical Society 2024

Blocher, Philip Daniel (ORCID:0000000163027567)↗
DOE OSTI2020

Hybrid Analytics Solution for Power Plant Operations

Expert Microsystems, Inc., MapEx Software Inc. and XMPLR Energy LLC have teamed to submit a proposal to the US Department of Energy’s Office of Fossil Energy to improve the efficiency, reliability and flexibility of existing coal-based power plants. The project, titled Hybrid Analytics Solution to Improve Coal Power Plant Operations (the “Hybrid Analytics” project), will develop, demonstrate and commercialize a novel approach that integrates two proven real-time monitoring techniques. The Hybrid Analytics approach integrates into a single, integrated solution: a) an established, advanced data-driven analytics solution that includes artificial intelligence, advanced pattern recognition and machine-learning techniques and b) a well-proven, first principle thermal heat balance model solution. The future role of coal plants will depend on the ability to cycle and follow loads to meet marginal power demands, accommodate renewables and support the grid. Essential to success in this new role is a plant’s ability to maintain high reliability, efficiency and flexibility. However, these changing operational demands put significant stresses on power plant components that directly impact the ability to deliver these requirements. In the past, plants have independently monitored performance using physics-based solutions (heat rate, boiler losses, etc.) and reliability using data-driven analytics (advanced pattern recognition). The Hybrid Analytics solution integrates both approaches to enhance fault detection, provide automated diagnostics and estimates of remaining time to act, and provide critical, enhanced and accurate real-time information to coal plant operators. As a result, coal plants can improve efficiency, reliability, and flexibility and can cost-effectively maintain their key role in the power delivery system.

01 COAL, LIGNITE, AND PEAT↗
DOE OSTI2023

Creating an Advanced Sensor Network to calculate real-time, mass-weighted flue gas composition and air heater leakage of a coal-fired utility boiler under dynamic operating conditions

Utilization of renewable energy sources to minimize the environmental impact of energy production has changed the way utility boilers operate, requiring frequent load cycling between full load and partial loads as low as 30%. Dynamic operation of coal-fired utility boilers significantly reduces boiler efficiency when compared to steady state at full load. Data-driven plant optimization has shown success with coal-fired utility boilers under dynamic operating conditions. The purpose of this work was to create an Advanced Sensor Network (ASN) to provide more extensive real-time data to inform dynamic plant optimization of Net Unit Heat Rate (NUHR). The ASN consists of gas sampling grids in the convective pass of the boiler and downstream of the air heater. These sampling grids allow for quantification of spatial variation of flue gas within the boiler and calculation of mass-weighted composition of flue gas through the combination of composition, velocity, and temperature measurements. The comparison of O 2 between the inlet and outlet of the air heater is used to calculate air leakage in real time. Flue gas composition and air heater leakage are both important factors in boiler efficiency and NUHR. Further, the results of this work support the value of mass-weighted averages for determining flue gas composition accurately. The measurements from the ASN show increased composition stratification during dynamic operation, with an average standard deviation 38% higher than observed during steady-state operation. Air heater leakage was also observed to increase from 2.8% to 5.1% following a load change. Prior to the installation of the ASN, these data would not have been available for dynamic control. These real-time data will be leveraged to calculate and optimize for NUHR during dynamic operation in future work.

42 ENGINEERING↗
DOE OSTI2022

Negative hydrogen ion sources for particle accelerators: Sustainability issues and recent improvements in long-term operations

High brightness, negative hydrogen ion sources are used extensively in scientific facilities operating worldwide. Negative hydrogen beams have become the preferred means of filling circular accelerators and storage rings as well as enabling efficient extraction from cyclotrons. Several well-known facilities now have considerable experience with operating a variety of sources such as RF-, filament-, magnetron- and Penning-type H- ion sources. These facilities include the US Spallation Neutron Source (SNS), Japan Proton Accelerator Research Complex (J-PARC), Rutherford Appleton Laboratory (RAL-ISIS), Los Alamos Neutron Science Center (LANSCE), Fermi National Accelerator Laboratory (FNAL), Brookhaven National Laboratory (BNL), numerous installations of D-Pace (licenced by TRIUMF) ion sources used mainly on cyclotrons and, most recently, the CERN-LINAC-1 injector. This report first summarizes the current performance of these ion sources in routine, daily operations with attention toward source service-periods and availability metrics. Sustainability issues encountered at each facility are also reported and categorized to identify areas of common concern and key issues. Recent ion source improvements to address these issues are also discussed as well as plans for meeting future facility upgrade requirements.

Welton, Robert F.↗
DOE OSTI2023

DeepONet-grid-UQ: A trustworthy deep operator framework for predicting the power grid’s post-fault trajectories

This paper proposes a novel data-driven method for the reliable prediction of the power grid’s post-fault trajectories, i.e., the power grid’s dynamic response after a disturbance or fault. Here, the proposed method is based on the recently proposed concept of Deep Operator Networks (DeepONets). Unlike traditional neural networks that learn to approximate functions, DeepONets are designed to approximate nonlinear operators, i.e., mappings between infinite-dimensional spaces. Under this operator framework, we design a novel and efficient DeepONet that (i) takes as inputs the trajectories collected before and during the fault and (ii) outputs the predicted post-fault trajectories. In addition, we endow our method with the much-needed ability to balance efficiency with reliable/trustworthy predictions via uncertainty quantification. To this end, we propose and compare two novel methods that enable quantifying the predictive uncertainty. First, we propose a Bayesian DeepONet (B-DeepONet) that uses stochastic gradient Hamiltonian Monte-Carlo to sample from the posterior distribution of the DeepONet trainable parameters. Then, we design a Probabilistic DeepONet (Prob-DeepONet) that uses a probabilistic training strategy to enable quantifying uncertainty at virtually no extra computational cost. Finally, we validate the proposed methods’ predictive power and uncertainty quantification capability using the New York-New England power grid model.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

Impact of Non-Steady State Operation on Cooling Water Consumption at Coal- and Natural Gas-Fired Power Plants

Increased renewable energy penetration in the electricity grid in coming decades will result in more frequent cycling at thermal power plants. Simultaneously, thermal power plants may face water scarcity with declining availability of cooling water. Therefore, to enhance thermal power plants’ resiliency to water shortages it is important to understand how non-steady conditions due to cycling will impact cooling water consumption and withdrawal intensity. Non-steady operation at power plants has been previously shown to decrease power plant thermal efficiency. Energy balance models have also demonstrated that a decrease in thermal efficiency is expected to increase cooling water consumption intensity. Furthermore, past work has used operating hours data to show idling and cycling gaps where cooling systems operate more than corresponding generators. As a result, an increase in cycling behavior may impact cooling water consumption and withdrawal intensity. This study uses data from the Energy Information Administration (EIA) and Environmental Protection Agency (EPA) to quantify the impact of cycling cooling water consumption intensity for recirculating cooling systems and withdrawal intensity for once-through cooling systems using energy balance and statistical approaches. In a novel application of a fixed effects model to study the effect of temperature on cooling water consumption and withdrawal intensity, this study finds temperature was consistently expected to increase consumption intensity and withdrawal intensity. Non-steady state conditions do not increase cooling water consumption intensity with statistical significance across unit types. However, additional validation of cooling water data is required to confirm these observed trends due to the sensitivity of these findings to model form and concerns with data quality of the dependent variables, cooling water consumption and withdrawal.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2022

Reverse-mode differentiation in arbitrary tensor network format: with application to supervised learning.

This paper describes an efficient reverse-mode differentiation algorithm for contraction operations of tensor networks that may have arbitrary and unconventional network topologies. The approach leverages the tensor contraction tree of Evenbly and Pfeifer (2014), which provides an instruction set for the contraction sequence of a network. We show that this tree can be efficiently leveraged for differentiation of a full tensor network contraction using a recursive scheme that exploits (1) the bilinear property of contraction and (2) the property that trees have single path from root to leaves. While differentiation of tensor-tensor contraction is already possible in most automatic differentiation packages, we show that exploiting these two additional properties in the specific context of contraction sequences can improve efficiency. Following a description of the algorithm and computational complexity analysis, we investigate its utility for gradient-based supervised learning for low-rank function recovery and for fitting real-world unstructured datasets. We demonstrate improved performance over alternating least-squares optimization approaches and the capability to handle heterogeneous and arbitrary tensor network formats. When compared to alternating minimization algorithms, we find that the gradient-based approach requires a smaller oversampling ratio (number of samples compared to number model parameters) for recovery. This increased efficiency extends to fitting unstructured data of varying dimensionality and when employing a variety of tensor network formats. Here, we show improved learning using the hierarchical Tucker method over the tensor-train in high-dimensional settings on a number of benchmark problems.

97 MATHEMATICS AND COMPUTING↗