Search NASA⌕ Search

SEARCH · Search NASA

Results for “Ensemble computing.”

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 361 records · Page 20

Complete treatment of galaxy two-point statistics: Gravitational lensing effects and redshift-space distortions

We present a coherent theoretical framework for computing gravitational lensing effects and redshift-space distortions in an inhomogeneous universe and investigate their impacts on galaxy two-point statistics. Adopting the linearized Friedmann-Lemaitre-Robertson-Walker metric, we derive the gravitational lensing and the generalized Sachs-Wolfe effects that include the weak lensing distortion, magnification, and time delay effects, and the redshift-space distortion, Sachs-Wolfe, and integrated Sachs-Wolfe effects, respectively. Based on this framework, we first compute their effects on observed source fluctuations, separating them as two physically distinct origins: the volume effect that involves the change of volume and is always present in galaxy two-point statistics, and the source effect that depends on the intrinsic properties of source populations. Then we identify several terms that are ignored in the standard method, and we compute the observed galaxy two-point statistics, an ensemble average of all the combinations of the intrinsic source fluctuations and the additional contributions from the gravitational lensing and the generalized Sachs-Wolfe effects. This unified treatment of galaxy two-point statistics clarifies the relation of the gravitational lensing and the generalized Sachs-Wolfe effects to the metric perturbations and the underlying matter fluctuations. For near-future dark energy surveys, we compute additional contributions to the observed galaxy two-point statistics and analyze their impact on the anisotropic structure. Thorough theoretical modeling of galaxy two-point statistics would be not only necessary to analyze precision measurements from upcoming dark energy surveys, but also provide further discriminatory power in understanding the underlying physical mechanisms.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Analytical and EZmock covariance validation for the DESI 2024 results

The estimation of uncertainties in cosmological parameters is an important challenge in Large-Scale-Structure (LSS) analyses. For standard analyses such as Baryon Acoustic Oscillations (BAO) and Full-Shape two approaches are usually considered. First: analytical estimates of the covariance matrix use Gaussian approximations and (nonlinear) clustering measurements to estimate the matrix, which allows a relatively fast and computationally cheap way to generate matrices that adapt to an arbitrary clustering measurement. On the other hand, sample covariances are an empirical estimate of the matrix based on an ensemble of clustering measurements from fast and approximate simulations. While more computationally expensive due to the large amount of simulations and volume required, these allow us to take into account systematics that are impossible to model analytically. In this work we compare these two approaches in order to enable DESI's key analyses. We find that the configuration space analytical estimate performs satisfactorily in BAO analyses and its flexibility in terms of input clustering makes it the fiducial choice for DESI's 2024 BAO analysis. On the contrary, the analytical computation of the covariance matrix in Fourier space does not reproduce the expected measurements in terms of Full-Shape analyses, which motivates the use of a corrected mock covariance for DESI's 2024 Full Shape analysis.

79 ASTRONOMY AND ASTROPHYSICS↗

Active operator learning with predictive uncertainty quantification for partial differential equations

With the increased prevalence of neural operators being used to provide rapid solutions to partial differential equations (PDEs), understanding the accuracy of model predictions and the associated error levels is necessary for deploying reliable surrogate models in scientific applications. Existing uncertainty quantification (UQ) frameworks employ ensembles or Bayesian methods, which can incur substantial computational costs during both training and inference. Here, we propose a lightweight predictive UQ method tailored for Deep operator networks (DeepONets) that also generalizes to other operator networks. Numerical experiments on linear and nonlinear PDEs demonstrate that the framework’s uncertainty estimates are unbiased and provide accurate out-of-distribution uncertainty predictions with a sufficiently large training dataset. Our framework provides fast inference and uncertainty estimates that can efficiently drive outer-loop analyses that would be prohibitively expensive with conventional solvers. We demonstrate how predictive uncertainties can be used in the context of Bayesian optimization and active learning problems to yield improvements in accuracy and data-efficiency for outer-loop optimization procedures. In the active learning setup, we extend the framework to Fourier Neural Operators (FNO) and describe a generalized method for other operator networks. To enable real-time deployment, we introduce an inference strategy based on precomputed trunk outputs and a sparse placement matrix, reducing evaluation time by more than a factor of five. Our method provides a practical route to uncertainty-aware operator learning in time-sensitive settings.

97 MATHEMATICS AND COMPUTING↗

A dynamic kinetic model captures cell-free metabolism for improved butanol production

Cell-free systems are useful tools for prototyping metabolic pathways and optimizing the production of various bioproducts. Mechanistically-based kinetic models are uniquely suited to analyze dynamic experimental data collected from cell-free systems and provide vital qualitative insight. However, to date, dynamic kinetic models have not been applied with rigorous biological constraints or trained on adequate experimental data to the degree that they would give high confidence in predictions and broadly demonstrate the potential for widespread use of such kinetic models. In this work, we construct a large-scale dynamic model of cell-free metabolism with the goal of understanding and optimizing butanol production in a cell-free system. Using a combination of parameterization methods, the resultant model captures experimental metabolite measurements across two experimental conditions for nine metabolites at timepoints between 0 and 24 hours. Here, we present analysis of the model predictions, provide recommendations for butanol optimization, and identify the aldehyde/alcohol dehydrogenase as the primary bottleneck in butanol production. Sensitivity analysis further reveals the extent to which various parameters are constrained, and our approach for probing valid parameter ranges can be applied to other modeling efforts.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Well-Defined Nanoparticle Electrocatalysts for the Refinement of Theory

The relationship between experiment and theory in electrocatalysis is one of profound importance. Until fairly recently, the principal role of theory in this field was interpreting experimental results. Over the course of the past decade (roughly the period covered by this review), however, that has begun to change, with theory now frequently leading the design of electrocatalytic materials. Though rewarding, this has not been a particularly easy union. For one thing, experimentalists and theorists have to come to grips with the fact that they rely on different models. Theorists make predictions based on individual, perfect structural models, while experimentalists work with more complex and heterogeneous ensembles of electrocatalysts. As discussed in this review, computational capabilities have improved in recent years, so that theory is better represented by the structures that experimentalists are able to prepare. Likewise, synthetic chemists are able to make ever more complex electrocatalysts with high levels of control, which provide a more extensive palette of materials for testing theory. Finally, the goal of this review is to highlight research from the last ~10 years that focuses on carefully controlled electrocatalytic experiments which, in combination with theoretical predictions, bring us closer to bridging the gap between real catalysts and computational models.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

High-Performance Semiempirical Excited-State Molecular Dynamics Powered by Graphics Processing Units

Here, this Letter introduces excited-state molecular dynamics in PYSEQM, a GPU-accelerated semiempirical quantum chemistry engine implemented in PyTorch. The new module enables Born–Oppenheimer molecular dynamics (BOMD) using configuration-interaction singles and random phase approximation for excited states, allowing long trajectories and large statistical ensembles to be simulated efficiently on a single GPU. We also implement an extended Lagrangian excited-state BOMD (XL-ESMD) scheme that propagates auxiliary electronic variables, enabling relaxed ground and excited-state convergence thresholds without compromising energy conservation. The excited-state BOMD implementation scales smoothly from small chromophores to a nearly 900-atom dendrimer (taking 6.5 s per MD step). PYSEQM also supports batched execution, allowing many geometries or trajectories to be evaluated in a single GPU launch, substantially increasing throughput and making ensemble-based protocols routine. As a demonstration, we compute absorption, emission, and infrared spectra from trajectories propagated on the ground and first excited states. The XL-ESMD scheme yields identical spectra at significantly lower computational cost, establishing the role of extended Lagrangian based dynamics for efficient excited-state BOMD simulations. Beyond raw performance, PYSEQM’s PyTorch foundation provides automatic differentiation for forces, efficient GPU batching, and seamless interfacing with machine learning models. These capabilities position PYSEQM as a practical platform for machine learning-augmented excited-state dynamics and lay the foundation for future data-driven nonadiabatic excited-state dynamics modeling of ultrafast spectroscopic probes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Influence of Longer Range Transfer of Vapor Interface Modified Caging Constraints on the Spatially Heterogeneous Dynamics of Glass-Forming Liquids

Based on the elastically collective nonlinear Langevin equation (ECNLE) theory of bulk relaxation in glass-forming liquids, and our recent ideas of how interface-nucleated modification of caging constraints are spatially transferred into the interior of a thick film, we present a force-based theory for dynamical gradients in thick films with one vapor interface that includes collective elasticity effects. Quantitative applications to the foundational hard sphere fluid and polymer melts of diverse fragilities are presented. In this work, we predict a roughly exponential spatial variation of the total activation barrier which is nonperturbatively modified to similar to 10 particle diameters into the film. This leads, to leading order, to the prediction of a reduced alpha relaxation time gradient of a double-exponential form characterized by a nearly constant spatial decay length, in qualitative accord with simulations. The relaxation acceleration at the surface grows exponentially with increasing packing fraction or decreasing temperature. The rate of increase with cooling of the alpha time strongly weakens upon approaching the vapor interface, with the top two layers remaining liquid-like down to similar to ~80-85% of the bulk glass transition temperature. These spatially heterogeneous changes of the temperature variation of the alpha time result in a large and long-range gradient of the local glass transition temperature. An average interfacial layer thickness relevant to ensemble-averaged dielectric and other experiments is also computed. It is predicted to be rather large at the bulk glass transition temperature, decreasing roughly linearly with heating. Remarkably, to leading order the ratio of this layer-averaged interfacial relaxation time to its bulk analogue is invariant to chemistry, volume fraction, and temperature and of modest absolute magnitude. Spatially inhomogeneous power law decoupling of the alpha relaxation time from its bulk value is predicted, with an effective exponent that decays to zero with distance from the free surface in a nearly exponential manner, trends which are in qualitative accord with recent simulations. Finally, this behavior and the double-exponential alpha time gradient are related and can be viewed as consequences of an effective quasi-universal factorization of the total barrier in films into the product of its bulk-temperature-dependent value times a function solely of location in the film.

36 MATERIALS SCIENCE↗

Computational Identification of Slow Conformational Fluctuations in Proteins

Conformational flexibility of proteins has been linked to their designated functions. Slow conformational fluctuations occurring at the microsecond to millisecond time scale, in particular, have recently attracted considerable interest in connection to the mechanism of enzyme catalysis. Computational methods are providing valuable insights into the connection between protein structure, flexibility, and function. In this report, we present studies on identification and characterization of microsecond flexibility of ubiquitin, based on quasi-harmonic analysis (QHA) and normal-mode analysis (NMA). The results indicate that the slowest 10 QHA modes, computed from the 0.5 {micro}s molecular dynamics ensemble, contribute over 78% of all motions. The identified slow movements show over 75% similarity with the conformational fluctuations observed in nuclear magnetic resonance ensemble and also agree with displacements in the set of X-ray structures. The slowest modes show high flexibility in the {beta}1-{beta}2, {alpha}1-{beta}3, and {beta}3-{beta}4 loop regions, with functional implications in the mechanism of binding other proteins. NMA of ubiquitin structures was not able to reproduce the long time scale fluctuations, as they were found to strongly depend on the reference structures. Further, conformational fluctuations coupled to the cis/trans isomerization reaction catalyzed by the enzyme cyclophilin A (CypA), occurring at the microsecond to millisecond time scale, have also been identified and characterized on the basis of QHA of conformations sampled along the reaction pathway. The results indicate that QHA covers the same conformational landscape as the experimentally observed CypA flexibility. Overall, the identified slow conformational fluctuations in ubiquitin and CypA indicate that the intrinsic flexibility of these proteins is closely linked to their designated functions.

59 BASIC BIOLOGICAL SCIENCES↗

Efficiency in nonequilibrium molecular dynamics Monte Carlo simulations

Hybrid algorithms combining nonequilibrium molecular dynamics and Monte Carlo (neMD/MC) offer a powerful avenue for improving the sampling efficiency of computer simulations of complex systems. These neMD/MC algorithms are also increasingly finding use in applications where conventional approaches are impractical, such as constant-pH simulations with explicit solvent. However, selecting an optimal nonequilibrium protocol for maximum efficiency often represents a non-trivial challenge. This work evaluates the efficiency of a broad class of neMD/MC algorithms and protocols within the theoretical framework of linear response theory. The approximations are validated against constant pH-MD simulations and shown to provide accurate predictions of neMD/MC performance. An assessment of a large set of protocols confirms (both theoretically and empirically) that a linear work protocol gives the best neMD/MC performance. Lastly, a well-defined criterion for optimizing the time parameters of the protocol is proposed and demonstrated with an adaptive algorithm that improves the performance on-the-fly with minimal cost.

97 MATHEMATICS AND COMPUTING↗

A simulation method for the phase diagram of complex fluid mixtures

We present that the phase behavior of complex fluid mixtures is of continuing interest, but obtaining the phase diagram from computer simulations can be challenging. In the Gibbs ensemble method, for example, each of the coexisting phases is simulated in a different cell, and ensuring the equality of chemical potentials of all components requires the transfer of molecules from one cell to the other. For complex fluids such as polymers, successful insertions are rare. An alternative method is to simulate both coexisting phases in a single simulation cell, with an interface between them. The challenge here is that the interface position moves during the simulation, making it difficult to determine the concentration profile and coexisting concentrations. In this work, we propose a new method for single cell simulations that uses a spatial concentration autocorrelation function to (spatially) align instantaneous concentration profiles from different snapshots. This allows one to obtain average concentration profiles and hence the coexisting concentrations. We test the method by calculating the phase diagrams of two systems: the Widom-Rowlinson model and the symmetric blends of freely jointed polymer molecules for which phase diagrams from conventional methods are available. Excellent agreement is found, except in the neighborhood of the critical point where the interface is broad and finite size effects are important. Finally, the method is easy to implement and readily applied to any mixture of complex fluids.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Decoherence and entropy production in relativistic nuclear collisions

Short thermalization times of less than 1 fm/c for quark and gluon matter have been suggested by recent experiments at the Relativistic Heavy Ion Collider. It has been difficult to justify this rapid thermalization in first-principle calculations based on perturbation theory or the color glass condensate picture. Here, we address the related question of the decoherence of the gluon field, which is a necessary component of thermalization. We present a simplified leading-order computation of the decoherence time of a gluon ensemble subject to an incoming flux of Weizsaecker-Williams gluons. We also discuss the entropy produced during the decoherence process and its relation to the entropy in the final state that has been measured experimentally.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Ab Initio Nuclear Thermodynamics

Here, we propose a new Monte Carlo method called the pinhole trace algorithm for ab initio calculations of the thermodynamics of nuclear systems. For typical simulations of interest, the computational speedup relative to conventional grand-canonical ensemble calculations can be as large as a factor of one thousand. Using a leading-order effective interaction that reproduces the properties of many atomic nuclei and neutron matter to a few percent accuracy, we determine the location of the critical point and the liquid-vapor coexistence line for symmetric nuclear matter with equal numbers of protons and neutrons. We also present the first ab initio study of the density and temperature dependence of nuclear clustering.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Simulation and big data challenges in tuning building energy models

EnergyPlus is the flagship building energy simulation software used to model whole building energy consumption for residential and commercial establishments. A typical input to the program often has hundreds, sometimes thousands of parameters which are typically tweaked by a buildings expert to “get it right”. This process can sometimes take months. “Autotune” is an ongoing research effort employing machine learning techniques to automate the tuning of the input parameters for an EnergyPlus input description of a building. Even with automation, the computational challenge faced to run the tuning simulation ensemble is daunting and requires the use of supercomputers to make it tractable in time. In this paper, we describe the scope of the problem, particularly the technical challenges faced and overcome, and the software infrastructure developed/in development when taking the EnergyPlus engine, which was primarily designed to run on desktops, and scaling it to run on shared memory supercomputers (Nautilus) and distributed memory supercomputers (Frost and Titan). The parametric simulations produce data in the order of tens to a couple of hundred terabytes. We describe the approaches employed to streamline and reduce bottlenecks in the workflow for this data, which is subsequently being made available for the tuning effort as well as made available publicly for open-science.

Sanyal, Jibonananda↗

Short-Term Probabilistic Solar Forecasting via Reinforcement Learning over ECMWF

In this paper, we present an innovative reinforcement learning approach for short-term solar forecasting, leveraging data from the European Centre for Medium-Range Weather Forecasts (ECMWF). The methodology begins with the application of the System Advisor Model (SAM) to transform various ECMWF numerical weather prediction members into predictive photovoltaic power generation. To enhance the precision of deterministic forecasting, we introduce a dynamic model selection algorithm based on Q-learning. This algorithm dynamically identifies and utilizes the most accurate ensemble member for forecasting purposes. Furthermore, we employ a support vector regression surrogate model with a Gaussian distribution to generate probabilistic forecasts, providing a holistic view of solar energy generation uncertainty. To expedite the training process and make it more practical for real-world applications, we integrate a rolling update workflow. This innovative workflow reduces the training period from months to a mere 19 days, making our method highly efficient. Numerical results of the case study show that in comparison to benchmark models, the proposed method improves the deterministic and probabilistic solar forecasting accuracy by up to 40.84% and 48.42%, respectively.

ensemble forecasting↗

Exploring Data Set Bias and Decision Support with Predictive Uncertainty Through Bayesian Approximations and Convolutional Neural Networks

Individual seismic catalogs can contain multiscale observations from fault level to global scales and associated waveforms from discrete events reflect crustal structure across many different scales and locations. Seismic network aperture, geographic location, and observation distance may not provide informative guidance or intuition on how different catalogs will behave across models trained under different conditions. We rely on uncertainty to provide guardrails for when to trust model decisions, but understanding when our uncertainty is trustworthy is an open challenge. Here, in this work, we explore Bayesian approximation methods for assigning predictive uncertainty in seismic event classification problems. We find that computationally expensive Bayesian approximations do not outperform simple ensemble methods. We also find that when exploiting multiple seismic event catalogs, joint training with data from all the catalogs combined with Bayesian approximations and supervised training for classification can obscure bias and result in less robust uncertainty while also not providing substantial performance benefits compared to training individual models for each catalog.

58 GEOSCIENCES↗

Boundary of the Distribution of Solar Wind Proton Beta versus Temperature Anisotropy

Abstract The frequency distribution of solar wind protons, measured in the vicinity of Earth’s orbit, is customarily plotted in ( β ∥ , T ⊥ / T ∥ ) phase space. Here, T ⊥ / T ∥ is the ratio of perpendicular and parallel temperatures, and β ∥ = 8 π nT ∥ / B 2 is the ratio of parallel thermal energy to background magnetic field energy, the so-called “parallel beta,” with ⊥ and ∥ denoting directions with respect to the ambient magnetic field. Such a frequency distribution, plotted as a two-dimensional histogram, forms a peculiar rhombic shape defined with an outer boundary in the said phase space. Past studies reveal that the threshold conditions for temperature anisotropy–driven plasma instability partially account for the boundary on the high- β ∥ side. The low- β ∥ side remains largely unexplained despite some efforts. Work by Vafin et al. recently showed that certain contours of collisional relaxation frequency, ν pp , when parameterized by T ⊥ / T ∥ and β ∥ , could match the overall shape of the left-hand boundary, thus suggesting that the collisional relaxation process might be closely related to the formation of the left-hand boundary. The present paper extends the analysis by Vafin et al. and carries out the dynamical computation of the collisional relaxation process for an ensemble of initial proton states with varying degrees of anisotropic temperatures. The final states of the relaxed protons are shown to closely match the observed boundary to the left of the ( β ∥ , T ⊥ / T ∥ ) phase space. When coupled with a similar set of calculations for the ensemble in the collective instability regime, it is found that the combined collisional/collective effects provide the baseline explanation for the observation.

Astronomy & Astrophysics↗

Skill and independence weighting for multi-model assessments

We present a weighting strategy for use with the CMIP5 multi-model archive in the fourth National Climate Assessment, which considers both skill in the climatological performance of models over North America as well as the inter-dependency of models arising from common parameterizations or tuning practices. The method exploits information relating to the climatological mean state of a number of projection-relevant variables as well as metrics representing long-term statistics of weather extremes. The weights, once computed can be used to simply compute weighted means and significance information from an ensemble containing multiple initial condition members from potentially co-dependent models of varying skill. Two parameters in the algorithm determine the degree to which model climatological skill and model uniqueness are rewarded; these parameters are explored and final values are defended for the assessment. The influence of model weighting on projected temperature and precipitation changes is found to be moderate, partly due to a compensating effect between model skill and uniqueness. However, more aggressive skill weighting and weighting by targeted metrics is found to have a more significant effect on inferred ensemble confidence in future patterns of change for a given projection.

58 GEOSCIENCES↗

Atmospheric lateral and longitudinal turbulent length scales (measured at 600 to 800 meters above grade level)

NASA's Airborne Doppler Lidar System has been used to obtain a detailed 'instantaneous' mapping of horizontal spatial wind fields at 600-800 m elevations on the east side of the San Gorgonio Pass in California, in the form of checkerboard-fashion horizontal wind vectors spaced at 300 m intervals along and normal to the flight path. Spatial autocorrelations for the lateral and longitudinal components are ensemble-averaged, and integral turbulent length scales are computed for the wind fields' longitudinal and lateral directions. The flow in the region studied does not appear to be isotropic.

Cliff, W. C.↗