Search NASA⌕ Search

SEARCH · Search NASA

Results for “reducing”

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 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Precise Synthesis of 4.75 V-Tolerant LiCoO 2 with Homogeneous Delithiation and Reduced Internal Strain

The rapid advancements in 3C electronic devices necessitate an increase in the charge cutoff voltage of LiCoO 2 to unlock a higher energy density that surpasses the currently available levels. However, the structural devastation and electrochemical decay of LiCoO 2 are significantly exacerbated, particularly at ≥4.5 V, due to the stress concentration caused by more severe lattice expansion and shrinkage, coupled with heterogeneous Li + intercalation/deintercalation reactions. Herein, employing the molten-salt synthesis technique, we propose a universal morphology-shaping strategy to attain bulk reaction homogeneity and reduce internal strains, even at extremely high charge voltages. The newly designed flattened polygon prismlike LiCoO 2 (P-LCO) particle, featuring a regular symmetrical arrangement along the c-axis, demonstrates a more homogeneous Li + extraction/insertion reaction, which results in a restrained transformation to detrimental O1 phase and reduced variation in lattice volume throughout the (de)lithiation processes. This benefits the mitigation of the local stress accumulation misfit dislocations and particle cracking, ultimately maintaining the mechanical stability of the cathode. Consequently, P-LCO is capable of breaking the voltage ceiling and exhibits exceptional long-term cycling capability at an ultrahigh voltage of 4.75 V. In conclusion, this work offers a brand-new perspective for the rational design of cathode morphology to address capacity deterioration caused by inhomogeneous delithiation and internal strain, thus inspiring the development of high-energy-density cathodes with improved durability.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Molecular H 2 as the Reducing Agent in Low-Temperature Oxide Reduction Using Calcium Hydride

Low-temperature synthesis is crucial for advancing sustainable manufacturing and accessing novel metastable phases. Metal hydrides have shown great potential in facilitating the reduction of oxides at low temperatures, yet the underlying mechanism—whether driven by H - , H 2 , or atomic H—remains unclear. Here, in this study, we employ in situ electrical transport measurements and first- principles calculations to investigate the CaH 2 -driven reduction kinetics in epitaxial α-Fe 2 O 3 thin films. Intriguingly, samples in direct contact with or separated from CaH 2 powders exhibit similar apparent activation energies for H 2 reduction, although direct contact significantly increases the reduction rate. These findings indicate that molecular H 2 is the dominant reducing species in the low-temperature reduction of oxides using CaH 2 , with a key aspect of the hydrides' superior reducing power attributed to their ability to eliminate residual moisture. This work underscores the critical role of moisture control in enabling effective low-temperature oxide reduction for advanced material synthesis.

Wang, Jiayue [SLAC National Accelerator Laboratory↗
DOE OSTI2026

Role of the Character of the Excited State of Singly Reduced Rh 2 (II,II) Intermediates on Photocatalytic Activity

Singly reduced intermediates have recently been implicated as photoactive intermediates in a number of important reactions; however, their photophysical properties remain poorly understood. A series of dirhodium(II,II) complexes, cis-[Rh 2 (p-R-Form) 2 (bncn) 2 ] 2+ (bncn = benzo[c]cinnoline; p-R-Form = N,N′-di-p-R-phenylformamidinate), where R = −OCH 3 (1), −CH 3 (2), −H (3), −F (4), −Cl (5), and −CF 3 (6), and their respective radical anions were prepared and their excited state properties were investigated. Complex 3 acted as a singlemolecule photocatalyst for H 2 production with red light. Substitution on the formamidinate ligands in 1−6 affected the energy of the Rh 2 (δ*)/Form(π,nb) highest occupied molecular orbital (HOMO), consistent with the metal/ligand-to-ligand charge transfer ( 1 ML-LCT) absorption maxima and the 3 ML-LCT excited state lifetime, ranging from 1.6 ns in 1 to 54 ns in 6 in CH 3 CN. The highest turnover number for photocatalytic H 2 evolution was observed for 3, and the lowest values were for 1 and 6. The radical anion, [Rh 2 ] − , formed during photocatalysis, was shown to absorb a photon and undergo a second reduction. The lifetimes of the doublet excited states of [3] − and [6] − were 0.49 and 0.24 ns, respectively. Calculations showed that the lowest energy excited state in [3] − was 2 ML-LCT, whereas that in [6] − was a bncn − → Rh 2 (σ*) ligand-to-metal charge transfer ( 2 LMCT) state. The 2 LMCT state stabilized across the series from [1] − to [6] − , pointing at its role in modulating the photophysical properties. This work highlights the importance of the reductive quenching of [Rh 2 ] − and the generation of the doubly reduced species to effectively catalyze hydrogen evolution.

Absorption↗
DOE OSTI2024

Reduced‐Order Modeling for Linearized Representations of Microphysical Process Rates

Abstract Representing cloud microphysical processes in large scale atmospheric models is challenging because many processes depend on the details of the droplet size distribution (DSD, the spectrum of droplets with different sizes in a cloud). While full or partial statistical moments of droplet size distributions are the typical variables used in bulk models, prognostic moments are limited in their ability to represent microphysical processes across the range of conditions experienced in the atmosphere. Microphysical parameterizations employing prognostic moments are known to suffer from structural uncertainty in their representations of inherently higher dimensional cloud processes, which limit model fidelity and lead to forecasting errors. Here we investigate how data‐driven reduced‐order modeling can be used to learn predictors for microphysical process rates in bulk microphysics schemes in an unsupervised manner from higher dimensional bin distributions. Using simulations characteristic of marine stratiform clouds, we simultaneously learn lower dimensional representations of droplet size distributions and predict the evolution of the microphysical state of the system. Droplet collision‐coalescence, the main process for generating warm rain, is estimated to have an intrinsic dimension of three. This intrinsic dimension provides a lower limit on the number of degrees of freedom needed to accurately represent collision‐coalescence in models. We demonstrate how deep learning based reduced‐order modeling can be used to discover intrinsic coordinates describing the microphysical state of the system, where process rates such as collision‐coalescence are globally linearized. These implicitly learned representations of the DSD retain more information about the DSD than typical moment‐based representations.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Reduced‐Order Probabilistic Emulation of Physics‐Based Ring Current Models: Application to RAM‐SCB Particle Flux

Abstract In this work, we address the computational challenge of large‐scale physics‐based simulation models for the ring current. Reduced computational cost allows for significantly faster than real‐time forecasting, enhancing our ability to predict and respond to dynamic changes in the ring current, valuable for space weather monitoring and mitigation efforts. Additionally, it can also be used for a comprehensive investigation of the system. Thus, we aim to create an emulator for the Ring current‐Atmosphere interactions Model with Self‐Consistent magnetic field (RAM‐SCB) particle flux that not only improves efficiency but also facilitates forecasting with reliable estimates of prediction uncertainties. The probabilistic emulator is built upon the methodology developed by Licata and Mehta (2023), https://doi.org/10.1029/2022sw003345 . A novel discrete sampling is used to identify 30 simulation periods over 20 years of solar and geomagnetic activity. Focusing on a subset of particle flux, we use Principal Component Analysis for dimensionality reduction and Long Short‐Term Memory (LSTM) neural networks to perform dynamic modeling. Hyperparameter space was explored extensively resulting in about 5% median symmetric accuracy across all data sets for one‐step dynamic prediction. Using a hierarchical ensemble of LSTMs, we have developed a reduced‐order probabilistic emulator (ROPE) tailored for time‐series forecasting of particle flux in the ring current. This ROPE offers accurate predictions of omnidirectional flux at a single energy with no pitch angle information, providing robust predictions on the test set with an error score below 11% and calibration scores under 8% with bias under 2% providing a significant speed up as compared to the full RAM‐SCB run.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Boosting Barlow Twins Reduced Order Modeling for Machine Learning‐Based Surrogate Models in Multiphase Flow Problems

Abstract We present an innovative approach called boosting Barlow Twins reduced order modeling (BBT‐ROM) to enhance the reliability of machine learning surrogate models for multiphase flow problems. BBT‐ROM builds upon Barlow Twins reduced order modeling that leverages self‐supervised learning to effectively handle linear and nonlinear manifolds by constructing well‐structured latent spaces of input parameters and output quantities. To address the challenge of high contrast data in multiphase flow problems due to injection wells and faults, we employ a boosting algorithm within BBT‐ROM. This algorithm sequentially trains a set of weak models (i.e., inaccurate models), improving prediction accuracy through ensemble learning. To evaluate the performance of BBT‐ROM, we conduct three three‐dimensional multiphase flow problems, including waterflooding and geologic carbon storage (GCS), with varying numbers of input parameter cases and model domain features. The results demonstrate that BBT‐ROM excels at predicting non‐wetting phase saturation (e.g., oil or saturation) and fluid pressure, with average relative errors ranging from 0.5% to 3%. Importantly, BBT‐ROM showcases robustness when faced with limited input parameter space during GCS testing.

58 GEOSCIENCES↗
DOE OSTI2025

Improved Representations of Longwave Surface Emissivity to Reduce Surface and Atmospheric Heating Biases in Earth System Models

Abstract Many Earth system models (ESMs) approximate surface emissivity as a broadband constant. This approximation reduces the computational burden, yet omits the spectral structure of emissivity and atmospheric absorption. Neglecting spectral variation in surface emission introduces biases in longwave (LW) atmospheric fluxes and heating. Biases are strongest over surfaces with strongly varying emissivity and minimal atmospheric opacity. We examine these biases over water, ice, and snow surfaces. We partition spectral emissivity into the 16 spectral bands utilized by a single‐column atmospheric radiative transfer model (RRTMG_LW) commonly used in ESMs. We quantify flux and heating biases introduced by broadband assumptions relative to the spectrally resolved case for standard atmospheric profiles over each surface type. Current assumptions tend to overestimate upwelling surface fluxes; for example, the greybody assumption overestimates flux by 1.6 W/m 2 (0.52%) at the bottom of a mid‐latitude winter atmosphere over ice, and by 2.33 (1.0%) at the top of atmosphere. The blackbody assumption tends to artificially cool Earth's surface, stabilizing the lower troposphere. Interestingly, the optimal broadband emissivity can deviate from the Planck‐weighted mean by up to 3% depending on surface type and atmospheric profile. We investigate bias sensitivity to surface temperature, cloud water path, and atmospheric water vapor. Bias is most sensitive to water vapor content, and least sensitive to cloud water path. Lastly, we show that a modified greybody method with updated broadband values can reduce total surface flux bias up to 1.69 , comparable to a five‐band approach and at a fraction of the computational cost.

Manzo, L. [Department of Earth System Science Univ↗
DOE OSTI2024

Integrating State Data Assimilation and Innovative Model Parameterization Reduces Simulated Carbon Uptake in the Arctic and Boreal Region

Model representation of carbon uptake and storage is essential for accurate projection of the response of the arctic-boreal zone to a rapidly changing climate. Land model estimates of LAI and aboveground biomass that can have a marked influence on model projections of carbon uptake and storage vary substantially in the arctic and boreal zone, making it challenging to correctly evaluate model estimates of Gross Primary Productivity (GPP). To understand and correct bias of LAI and aboveground biomass in the Community Land Model (CLM), we assimilated the 8-day Moderate Resolution Imaging Spectroradiometer (MODIS) LAI observation and a machine learning product of annual aboveground biomass into CLM using an Ensemble Adjustment Kalman Filter (EAKF) in an experimental region including Alaska and Western Canada. Assimilating LAI and aboveground biomass reduced these model estimates by 58% and 72%, respectively. The change of aboveground biomass was consistent with independent estimates of canopy top height at both regional and site levels. The International Land Model Benchmarking system assessment showed that data assimilation significantly improved CLM's performance in simulating the carbon and hydrological cycles, as well as in representing the functional relationships between LAI and other variables. Here, to further reduce the remaining bias in GPP after LAI bias correction, we re-parameterized CLM to account for low temperature suppression of photosynthesis. The LAI bias corrected model that included the new parameterization showed the best agreement with model benchmarks. Combining data assimilation with model parameterization provides a useful framework to assess photosynthetic processes in LSMs.

58 GEOSCIENCES↗
DOE OSTI2021

Conversion of marginal land into switchgrass conditionally accrues soil carbon but reduces methane consumption

Abstract Switchgrass is a deep-rooted perennial native to the US prairies and an attractive feedstock for bioenergy production; when cultivated on marginal soils it can provide a potential mechanism to sequester and accumulate soil carbon (C). However, the impacts of switchgrass establishment on soil biotic/abiotic properties are poorly understood. Additionally, few studies have reported the effects of switchgrass cultivation on marginal lands that have low soil nutrient quality (N/P) or in areas that have experienced high rates of soil erosion. Here, we report a comparative analyses of soil greenhouse gases (GHG), soil chemistry, and microbial communities in two contrasting soil types (with or without switchgrass) over 17 months (1428 soil samples). These soils are highly eroded, ‘Dust Bowl’ remnant field sites in southern Oklahoma, USA. Our results revealed that soil C significantly increased at the sandy-loam (SL) site, but not at the clay-loam (CL) site. Significantly higher CO2 flux was observed from the CL switchgrass site, along with reduced microbial diversity (both alpha and beta). Strikingly, methane (CH4) consumption was significantly reduced by an estimated 39 and 47% at the SL and CL switchgrass sites, respectively. Together, our results suggest that soil C stocks and GHG fluxes are distinctly different at highly degraded sites when switchgrass has been cultivated, implying that carbon balance considerations should be accounted for to fully evaluate the sustainability of deep-rooted perennial grass cultivation in marginal lands.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Metabolically diverse microorganisms mediate methylmercury formation under nitrate-reducing conditions in a dynamic hydroelectric reservoir

Brownlee Reservoir is a mercury (Hg)-impaired hydroelectric reservoir that exhibits dynamic hydrological and geochemical conditions and is located within the Hells Canyon Complex in Idaho, USA. Methylmercury (MeHg) contamination in fish is a concern in the reservoir. While MeHg production has historically been attributed to sulfate-reducing bacteria and methanogenic archaea, microorganisms carrying the hgcA gene are taxonomically and metabolically diverse and the major biogeochemical cycles driving mercury (Hg) methylation are not well understood. In this study, Hg speciation and redox-active compounds were measured throughout Brownlee Reservoir across the stratified period in four consecutive years (2016–2019) to identify the location where and redox conditions under which MeHg is produced. Metagenomic sequencing was performed on a subset of samples to characterize the microbial community with hgcA and identify possible links between biogeochemical cycles and MeHg production. Biogeochemical profiles suggested in situ water column Hg methylation was the major source of MeHg. These profiles, combined with genome-resolved metagenomics focused on hgcA-carrying microbes, indicated that MeHg production occurs in this system under nitrate- or manganese-reducing conditions, which were previously thought to preclude Hg-methylation. Using this multidisciplinary approach, we identified the cascading effects of interannual variability in hydrology on the redox status, microbial metabolic strategies, abundance and metabolic diversity of Hg methylators, and ultimately MeHg concentrations throughout the reservoir. This work expands the known conditions conducive to producing MeHg and suggests that the Hg-methylation mitigation efforts by nitrate or manganese amendment may be unsuccessful in some locations.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Extensive hydrogen incorporation is not necessary for superconductivity in topotactically reduced nickelates

Abstract A key open question in the study of layered superconducting nickelate films is the role that hydrogen incorporation into the lattice plays in the appearance of the superconducting state. Due to the challenges of stabilizing highly crystalline square planar nickelate films, films are prepared by the deposition of a more stable parent compound which is then transformed into the target phaseviaa topotactic reaction with a strongly reducing agent such as CaH 2 . Recent studies, both experimental and theoretical, have introduced the possibility that the incorporation of hydrogen from the reducing agent into the nickelate lattice may be critical for the superconductivity. In this work, we use secondary ion mass spectrometry to examine superconducting La 1−x X x NiO 2 / SrTiO 3 (X= Ca and Sr) and Nd 6 Ni 5 O 12 / NdGaO 3 films, along with non-superconducting NdNiO 2 / SrTiO 3 and (Nd,Sr)NiO 2 / SrTiO 3 . We find no evidence for extensive hydrogen incorporation across a broad range of samples, including both superconducting and non-superconducting films. Theoretical calculations indicate that hydrogen incorporation is broadly energetically unfavorable in these systems, supporting our conclusion that extensive hydrogen incorporation is not generally required to achieve a superconducting state in layered square-planar nickelates.

Science & Technology - Other Topics↗
DOE OSTI2026

Tandem bulk oxygen diffusion and surface reactions in reducible metal oxides control redox cycle dynamics

The interplay between bulk oxygen diffusion and surface reactions in reducible metal oxides is key in heterogeneous catalysts, but direct measurements of oxygen mobility, transient kinetics, and in situ spectroscopies have been lacking. Here, we reveal complex dynamic behavior of ceria-zirconia by H 2 using transient kinetics via mass spectrometry and in situ Raman and near-ambient pressure x-ray photoelectron spectroscopies. Molecular dynamics simulations with a machine learning potential delineate competitive oxygen diffusion mechanisms, with an optimal mobility at intermediate reductions. We expose a compensation between vacancy availability and lattice distortion at intermediate to high reductions and Frenkel defects at low reductions, underscoring a potential deficiency of 16 O/ 18 O exchange experiments in deducing oxygen mobility. Vacancies in proximity require electron localization on Ce atoms further away. The continuous replenishment of surface oxygen results in a varying reduction rate, with H 2 dissociation being the rate-limiting step. Multiscale transient simulations, consistent with experiments, indicate catalysts of potentially spatially varying oxidation states. The approach is broadly applicable to reducible oxide materials.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Reduced CO 2 uptake and growing nutrient sequestration from slowing overturning circulation

Current Earth system models (ESMs) project dramatic slowing (28–42% by 2100) of Atlantic Meridional Overturning Circulation and Southern Meridional Overturning Circulation (SMOC) across a range of climate scenarios, with a complete shutdown of SMOC possible by year 2300. Slowing meridional overturning circulation (MOC) differentially impacts the ocean biological and solubility carbon pumps, leaving the net impact on ocean carbon uptake uncertain. Here using a suite of ESMs, we show that slowing MOC reduces anthropogenic carbon uptake by the solubility pump but increases deep-ocean storage of carbon and nutrients by the biological pump. The net effect reduces ocean uptake of anthropogenic CO 2 . The deep-ocean nutrient sequestration will increasingly depress global-scale, marine net primary production over time. As a result, MOC slowdown represents a positive feedback that could extend or intensify peak-warmth climate conditions on multi-century timescales.

58 GEOSCIENCES↗
DOE OSTI2025

Climate change drives reduced biocontrol of the invasive spongy moth

The effects of climate change on forest-defoliating insects are poorly understood, but could severely reduce forest productivity, biodiversity and timber production. For decades following its introduction in 1869, the spongy moth ( Lymantria dispar ) severely defoliated North American forests, but the introduction of the fungal pathogen Entomophaga maimaiga in 1989 suppressed spongy moth defoliation for 27 years. E. maimaiga , however, needs cool, moist conditions, whereas climate change is bringing hot, dry conditions to the range of the insect. Here, in this work, we use an empirically verified eco-climate model to project that climate change will sharply reduce E. maimaiga infection rates, thereby increasing spongy moth defoliation. Recent rebounds in defoliation are consistent with our projections. Our work demonstrates that the effects of climate change on species interactions can have important consequences for natural ecosystems.

climate-change ecology↗
DOE OSTI2022

Observation of a reduced-turbulence regime with boron powder injection in a stellarator

Abstract In state-of-the-art stellarators, turbulence is a major cause of the degradation of plasma confinement. To maximize confinement, which eventually determines the amount of nuclear fusion reactions, turbulent transport needs to be reduced. Here we report the observation of a confinement regime in a stellarator plasma that is characterized by increased confinement and reduced turbulent fluctuations. The transition to this regime is driven by the injection of submillimetric boron powder grains into the plasma. With the line-averaged electron density being kept constant, we observe a substantial increase of stored energy and electron and ion temperatures. At the same time, the amplitude of the plasma turbulent fluctuations is halved. While lower frequency fluctuations are damped, higher frequency modes in the range between 100 and 200 kHz are excited. We have observed this regime for different heating schemes, namely with both electron and ion cyclotron resonant radio frequencies and neutral beams, for both directions of the magnetic field and both hydrogen and deuterium plasmas.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Identifying transcription factors that reduce wood recalcitrance and improve enzymatic degradation of xylem cell wall in Populus

Developing an efficient deconstruction step of woody biomass for biorefinery has been drawing considerable attention since its xylem cell walls display highly recalcitrance nature. Here, we explored transcriptional factors (TFs) that reduce wood recalcitrance and improve saccharification efficiency in Populus species. First, 33 TF genes up-regulated during poplar wood formation were selected as potential regulators of xylem cell wall structure. The transgenic hybrid aspens (Populus tremula × Populus tremuloides) overexpressing each selected TF gene were screened for in vitro enzymatic saccharification. Of these, four transgenic seedlings overexpressing previously uncharacterized TF genes increased total glucan hydrolysis on average compared to control. The best performing lines overexpressing Pt × tERF123 and Pt × tZHD14 were further grown to form mature xylem in the greenhouse. Notably, the xylem cell walls exhibited significantly increased total xylan hydrolysis as well as initial hydrolysis rates of glucan. The increased saccharification of Pt × tERF123-overexpressing lines could reflect the improved balance of cell wall components, i.e., high cellulose and low xylan and lignin content, which could be caused by upregulation of cellulose synthase genes upon the expression of Pt × tERF123. Overall, we successfully identified Pt × tERF123 and Pt × tZHD14 as effective targets for reducing cell wall recalcitrance and improving the enzymatic degradation of woody plant biomass.

09 BIOMASS FUELS↗
DOE OSTI2022

Enhancing high-fidelity nonlinear solver with reduced order model

Abstract We propose the use of reduced order modeling (ROM) to reduce the computational cost and improve the convergence rate of nonlinear solvers of full order models (FOM) for solving partial differential equations. In this study, a novel ROM-assisted approach is developed to improve the computational efficiency of FOM nonlinear solvers by using ROM’s prediction as an initial guess. We hypothesize that the nonlinear solver will take fewer steps to the converged solutions with an initial guess that is closer to the real solutions. To evaluate our approach, four physical problems with varying degrees of nonlinearity in flow and mechanics have been tested: Richards’ equation of water flow in heterogeneous porous media, a contact problem in a hyperelastic material, two-phase flow in layered porous media, and fracture propagation in a homogeneous material. Overall, our approach maintains the FOM’s accuracy while speeding up nonlinear solver by 18–73% (through suitable ROM-assisted FOMs). More importantly, the proximity of ROM’s prediction to the solution space leads to the improved convergence of FOMs that would have otherwise diverged with default initial guesses. We demonstrate that the ROM’s accuracy can impact the computational efficiency with more accurate ROM solutions, resulting in a better cost reduction. We also illustrate that this approach could be used in many FOM discretizations (e.g., finite volume, finite element, or a combination of those). Since our ROMs are data-driven and non-intrusive, the proposed procedure can easily lend itself to any nonlinear physics-based problem.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Variational multiscale reinforcement learning for discovering reduced order closure models of nonlinear spatiotemporal transport systems

Abstract A central challenge in the computational modeling and simulation of a multitude of science applications is to achieve robust and accurate closures for their coarse-grained representations due to underlying highly nonlinear multiscale interactions. These closure models are common in many nonlinear spatiotemporal systems to account for losses due to reduced order representations, including many transport phenomena in fluids. Previous data-driven closure modeling efforts have mostly focused on supervised learning approaches using high fidelity simulation data. On the other hand, reinforcement learning (RL) is a powerful yet relatively uncharted method in spatiotemporally extended systems. In this study, we put forth a modular dynamic closure modeling and discovery framework to stabilize the Galerkin projection based reduced order models that may arise in many nonlinear spatiotemporal dynamical systems with quadratic nonlinearity. However, a key element in creating a robust RL agent is to introduce a feasible reward function, which can be constituted of any difference metrics between the RL model and high fidelity simulation data. First, we introduce a multi-modal RL to discover mode-dependant closure policies that utilize the high fidelity data in rewarding our RL agent. We then formulate a variational multiscale RL (VMRL) approach to discover closure models without requiring access to the high fidelity data in designing the reward function. Specifically, our chief innovation is to leverage variational multiscale formalism to quantify the difference between modal interactions in Galerkin systems. Our results in simulating the viscous Burgers equation indicate that the proposed VMRL method leads to robust and accurate closure parameterizations, and it may potentially be used to discover scale-aware closure models for complex dynamical systems.

97 MATHEMATICS AND COMPUTING↗