Search NASA⌕ Search

SEARCH · Search NASA

Results for “Domain Distributed Framework”

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 307 records · Page 17

Climatespark: an In-Memory Distributed Computing Framework for Big Climate Data Analytics

The unprecedented growth of climate data creates new opportunities for climate studies, and yet big climate data pose a grand challenge to climatologists to efficiently manage and analyze big data. The complexity of climate data content and analytical algorithms increases the difficulty of implementing algorithms on high performance computing systems. This paper proposes an in-memory, distributed computing framework, ClimateSpark, to facilitate complex big data analytics and time-consuming computational tasks. Chunking data structure improves parallel I/O efficiency, while a spatiotemporal index is built for the chunks to avoid unnecessary data reading and preprocessing. An integrated, multi-dimensional, array-based data model (ClimateRDD) and ETL operations are developed to address big climate data variety by integrating the processing components of the climate data lifecycle. ClimateSpark utilizes Spark SQL and Apache Zeppelin to develop a web portal to facilitate the interaction among climatologists, climate data, analytic operations and computing resources (e.g., using SQL query and Scala/Python notebook). Experimental results show that ClimateSpark conducts different spatiotemporal data queries/analytics with high efficiency and data locality. ClimateSpark is easily adaptable to other big multiple- dimensional, array-based datasets in various geoscience domains.

Hu, Fei↗

Assessing the Sensitivity of Pourbaix Diagrams to Computational Protocols: Electrochemical Stability of Ni Oxides as a Case Study

Pourbaix diagrams stand as a useful tool in assessing and visualizing materials’ electrochemical stability and are widely used for electrocatalyst design. However, their reliability hinges on the accuracy of the chemical potentials of involved phases, which may bear uncertainties and can be significantly impacted by decision-making steps in the computational protocol. Here, this study introduces a robust sensitivity analysis framework, exemplified through a detailed examination of the computational Pourbaix diagram of Ni, the oxides of which are used as high-activity and cost-friendly catalysts for many electrochemical reactions. Quantities of interest derived from the Pourbaix diagram include the appearance and stability domain of the catalytically active Ni oxide phases along with the onset electrochemical potentials of phase transitions. These metrics can guide the design of operational conditions for Ni oxide electrocatalysts. We find that the employed DFT exchange-correlation functional has the most significant influence on the computed Pourbaix diagram. Uncertainties on crystal structures, along with their related ab initio energetics, are also found to affect the size of the phase stability domain. Higher-order coupling among input parameters is found to play a crucial role in influencing the appearance and distribution of Ni phases in the diagram. Our findings suggest a need to consider variations and uncertainties associated with the computational procedures on predicted Pourbaix diagrams for materials design.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Electronic structure and photophysics of a supermolecular iron complex having a long MLCT-state lifetime and panchromatic absorption

Exploiting earth-abundant iron-based metal complexes as high-performance photosensitizers demands long-lived electronically excited metal-to-ligand charge-transfer (MLCT) states, but these species suffer typically from femtosecond timescale charge-transfer (CT)-state quenching by low-lying nonreactive metal-centered (MC) states. Here, we engineer supermolecular Fe(II) chromophores based on the bis(tridentate-ligand)metal(II)-ethyne-(porphinato)zinc(II) conjugated framework, previously shown to give rise to highly delocalized low-lying 3 MLCT states for other Group VIII metal (Ru, Os) complexes. Electronic spectral, potentiometric, and ultrafast pump–probe transient dynamical data demonstrate that a combination of a strong σ-donating tridentate ligand and a (porphinato)zinc(II) moiety with low-lying π*-energy levels, sufficiently destabilize MC states and stabilize supermolecular MLCT states to realize Fe(II) complexes that express 3 MLCT state photophysics reminiscent of their heavy-metal analogs. The resulting Fe(II) chromophore archetype, FeNHCPZn, features a highly polarized CT state having a profoundly extended 3MLCT lifetime (160 ps), 3 MLCT phosphorescence, and ambient environment stability. Density functional and domain-based local pair natural orbital coupled cluster [DLPNO-CCSD(T)] theory reveal triplet-state wavefunction spatial distributions consistent with electronic spectroscopic and excited-state dynamical data, further underscoring the dramatic Fe metal-to-extended ligand CT character of electronically excited FeNHCPZn. This design further prompts intense panchromatic absorptivity via redistributing high-energy absorptive oscillator strength throughout the visible spectral domain, while maintaining a substantial excited-state oxidation potential for wide-ranging photochemistry––highlighted by the ability of FeNHCPZn to photoinject charges into a SnO 2 /FTO electrode in a dye-sensitized solar cell (DSSC) architecture. Concepts enumerated herein afford opportunities for replacing traditional rare-metal–based emitters for solar-energy conversion and photoluminescence applications.

74 ATOMIC AND MOLECULAR PHYSICS↗

The Super Tuesday Outbreak: Forecast Sensitivities to Single-Moment Microphysics Schemes

Forecast precipitation and radar characteristics are used by operational centers to guide the issuance of advisory products. As operational numerical weather prediction is performed at increasingly finer spatial resolution, convective precipitation traditionally represented by sub-grid scale parameterization schemes is now being determined explicitly through single- or multi-moment bulk water microphysics routines. Gains in forecasting skill are expected through improved simulation of clouds and their microphysical processes. High resolution model grids and advanced parameterizations are now available through steady increases in computer resources. As with any parameterization, their reliability must be measured through performance metrics, with errors noted and targeted for improvement. Furthermore, the use of these schemes within an operational framework requires an understanding of limitations and an estimate of biases so that forecasters and model development teams can be aware of potential errors. The National Severe Storms Laboratory (NSSL) Spring Experiments have produced daily, high resolution forecasts used to evaluate forecast skill among an ensemble with varied physical parameterizations and data assimilation techniques. In this research, high resolution forecasts of the 5-6 February 2008 Super Tuesday Outbreak are replicated using the NSSL configuration in order to evaluate two components of simulated convection on a large domain: sensitivities of quantitative precipitation forecasts to assumptions within a single-moment bulk water microphysics scheme, and to determine if these schemes accurately depict the reflectivity characteristics of well-simulated, organized, cold frontal convection. As radar returns are sensitive to the amount of hydrometeor mass and the distribution of mass among variably sized targets, radar comparisons may guide potential improvements to a single-moment scheme. In addition, object-based verification metrics are evaluated for their utility in gauging model performance and QPF variability.

Molthan, Andrew L.↗

Multi-head physics-informed neural networks for learning functional priors and uncertainty quantification

In numerous applications, the integration of prior knowledge and historical information is essential, particularly for tasks requiring the solution of ordinary or partial differential equations (ODEs/PDEs) in data-sparse or noisy environments. For instance, achieving accurate solutions to time-dependent PDEs with limited initial condition measurements necessitates an effective strategy for embedding prior knowledge. Hard-parameter sharing architectures in neural networks (NNs) have demonstrated success in both traditional and scientific machine learning domains, facilitating the learning of informative representations. Here, in this study, we introduce a novel, yet efficient, method to enhance physics-informed neural networks (PINNs) by incorporating a multi-head structure that enables the learning of functional priors from both empirical data and governing physical laws. This prior information can then be used to address data sparsity and high-level noise in solving ODE/PDE problems with uncertainty quantification (UQ). The approach, termed Multi-Head PINN (MH-PINN), consists of a shared body NN and multiple head NNs, each corresponding to an individual PINN instance. Our framework for functional prior learning is carried out in two stages: (1) training the MH-PINNs to develop a shared body NN alongside multiple head NNs, and (2) employing these trained head NNs to estimate a prior distribution through a normalizing flow-based density estimator. The learned functional prior can then be applied as a regularization mechanism in deterministic contexts or as an informative prior within a Bayesian inference framework, aiding in the resolution of subsequent ODE/PDE tasks. We evaluate the efficacy of MH-PINNs across five benchmark problems, including a high-dimensional parametric PDE, all characterized by data sparsity or substantial noise levels. Our findings reveal that MH-PINNs deliver accurate solutions and robust UQ, demonstrating adaptability across a range of complex and challenging scenarios.

Bayesian inference↗

Parallel quantum computing simulations via quantum accelerator platform virtualization

Quantum circuit execution is a central task in quantum computation. Due to inherent quantum-mechanical constraints, quantum computing workflows often involve a considerable number of independent measurements over a large set of slightly different quantum circuits. Here we discuss a simple model for parallelizing such quantum circuit executions that is based on introducing a large array of virtual quantum processing units (mapped to HPC nodes in our case) as a parallel quantum computing platform. Implemented within the XACC framework, the model can readily take advantage of its backend-agnostic features, enabling parallel quantum computing/simulation over any target backend supported by XACC. We illustrate the performance of this approach by demonstrating strong scaling in two pertinent domain science problems, namely in computing the gradients for the multi-contracted variational quantum eigensolver and in data-driven quantum circuit learning, where we vary the number of qubits and the number of circuit layers. Here, the latter simulation leverages the cuQuantum library to run efficiently on GPU-accelerated HPC platforms.

97 MATHEMATICS AND COMPUTING↗

Electric Vehicles Charging Time Constrained Deliverable Provision of Secondary Frequency Regulation

Aggregation of electric vehicles (EVs) is a promising technique for providing secondary frequency regulation (SFR) in highly renewable energy-penetrated power systems. Equipped with energy storage devices, EV aggregation can provide reliable SFR. However, the main challenge is to guarantee reliable intra-interval SFR capacities and inter-interval delivery following the automatic generation control (AGC) signal. Furthermore, aggregated EV SFR provision will be further complicated by the EV charging time anxiety because SFR provision might extend EV's charging time. This paper proposes a deliverable EV SFR provision with a charging-time-constrained control strategy. First, a charging-time-constrained EV aggregation is proposed to address the uncertainty of EV capacity based on the state-space model considering the charging-time restriction of EV owners. Second, a real-time economic dispatch and time domain simulation (RTED-TDS) cosimulation framework is proposed to verify financial results and the dynamic performance of the EV SFR provision. Last, the proposed charging time-constrained EV aggregation is validated on the IEEE 39-bus system. In conclusion, the results demonstrate that with charging time-constrained EV aggregation, the dynamic performance of the system can be improved with a marginal increase in total cost. More importantly, the charging time constraint can be respected in the proposed SFR provision of the EV aggregation.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Versatility and Complexity: Common and Uncommon Facets of LysR-Type Transcriptional Regulators

LysR-type transcriptional regulators (LTTRs) form one of the largest families of bacterial regulators. They are widely distributed and contribute to all aspects of metabolism and physiology. Most are homotetramers, with each subunit composed of an N-terminal DNA-binding domain followed by a long helix connecting to an effector-binding domain. LTTRs typically bind DNA in the presence or absence of a small-molecule ligand (effector). In response to cellular signals, conformational changes alter DNA interactions, contact with RNA polymerase, and sometimes contact with other proteins. Many are dual-function repressor–activators, although different modes of regulation may occur at multiple promoters. This review presents an update on the molecular basis of regulation, the complexity of regulatory schemes, and applications in biotechnology and medicine. The abundance of LTTRs reflects their versatility and importance. While a single regulatory model cannot describe all family members, a comparison of similarities and differences provides a framework for future study.

59 BASIC BIOLOGICAL SCIENCES↗

The NASA Eulerian Snow on Sea Ice Model (NESOSIM) v1.0: initial model development and analysis

The NASA Eulerian Snow On Sea Ice Model (NESOSIM) is a new, open-source snow budget model that is currentlyconfigured to produce daily estimates of the depth and density of snow on sea ice across the Arctic Ocean throughthe accumulation season. NESOSIM has been developed in a three-dimensional Eulerian framework and includes two(vertical) snow layers and several simple parameterizations (accumulation, wind packing, advection_divergence, blowingsnow lost to leads) to represent key sources and sinks of snow on sea ice. The model is forced with daily inputs of snowfall and near-surface winds (from reanalyses), sea ice concentration (from satellite passive microwave data) and sea ice drift (from satellite feature tracking) during the accumulation season (August through April). In this study, we present the NESOSIM formulation, calibration efforts, sensitivity studies and validation efforts across an Arctic Ocean domain (100 km horizontal resolution). The simulated snow depth and density are calibrated with in situ data collected on drifting ice stations during the 1980s. NESOSIM shows strong agreement with the in situ seasonal cycles of snow depth and density, and shows good (moderate) agreement with the regional snow depth (density) distributions. NESOSIM is run for a contemporary period (2000 to 2015), with the results showing strong sensitivity to the reanalysis derived snowfall forcing data, with the Modern-Era Retrospective analysis for Research and Applications (MERRA) and the Japanese Meteorological Agency 55-year reanalysis (JRA-55) forced snow depths generally higher than ERA-Interim,and the Arctic System Reanalysis (ASR) generally lower. We also generate and force NESOSIM with a consensus "median" daily snowfall dataset from these reanalyses.The results are compared against snow depth estimates derived from NASA's Operation IceBridge (OIB) snow radar data from 2009 to 2015, showing moderate_strong correlations and root mean squared errors of 10 cm depending on the OIB snow depth product analyzed, similar to the comparisons between OIB snow depths and the commonly used modified Warren snow depth climatology. Potential improvements to this initial NESOSIM formulation are discussed in the hopes of improving the accuracy and reliability of these simulated snow depths and densities.

Petty, Alek A.↗

Collisional transport coefficients of dense high-temperature plasmas within the quantum Landau-Fokker-Planck framework

In this paper, we extend the long-established formulas for the transport coefficients of classical plasmas inside the dense plasma regime for temperatures and densities where the classical Landau equation breaks down but its quantum extension that includes quantum degeneracy effects is valid. To this end, the quantum Landau kinetic equation is solved by the Chapman-Enskog method. The mathematical derivation is done in full generality, i.e., for multicomponent systems and to all orders of the polynomials expansion used to approximate the distribution functions. We apply the general results to two important examples, the electron gas model and an electron-ion plasma model consisting of one type of ions of any charge. We discuss the combined effects of the Pauli exclusion principle, of the electron-electron, and of the electron-ion collisions on the transport coefficients and on the convergence of the Chapman-Enskog method. For the electron gas model, the effect of the Pauli exclusion principle on the transport coefficients rapidly becomes non-negligible outside the domain of validity of the classical Landau equation. For the electron-ion plasmas, the effect of the Pauli exclusion principle depends sensitively on the ion charge Z and varies non-monotonically with Θ. For instance, for ion charge Z = 1, the electrical conductivity is increased by up to ~30% compared to its classical value over the range of degeneracy parameters studied, the thermal conductivity is reduced by up to ~9%, and the shear viscosity coefficient is increased by up to ~13%. Finally, in the Lorentz gas ( Z → ∞ ) limit, the electrical conductivity is reduced by up to ~14% compared to its classical value over the range of degeneracy parameters studied, the thermal conductivity is reduced by up to ~39%, and the shear viscosity coefficient is not affected.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Locating the Missing Superconducting Electrons in the Overdoped Cuprates La 2 – x Sr x CuO 4 [Locating the Missing Superconducting Electrons in Overdoped Cuprates]

Overdoped high-temperature cuprate superconductors have often been understood within the standard BCS framework of superconductivity. However, measurements in a variety of overdoped cuprates indicate that the superfluid density is much smaller than expected from BCS theory and decreases smoothly to zero as the doping is increased. Here, we combine time-domain THz spectroscopy with kHz range mutual inductance measurements on the same overdoped La 2–x Sr x CuO 4 films to determine the total, superfluid, and uncondensed spectral weight as a function of doping. A significant fraction of the carriers remains uncondensed in a wide Drude-like peak as T → 0, while the superfluid density remains linear in temperature. These observations are seemingly inconsistent with existing, realistic theories of impurity scattering suppressing the superfluid density in a BCS-like d-wave superconductor. Furthermore, our large measurement frequency range gives us a unique look at the low frequency spectral weight distribution, which may suggest the presence of quantum phase fluctuations as the critical doping is approached.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Transitioning GlideinWMS, a multi domain distributed workload manager, from GSI proxies to tokens and other granular credentials

GlideinWMS is a distributed workload manager that has been used in production for many years to provision resources for experiments like CERN’s CMS, many Neutrino experiments, and the OSG. Its security model was based mainly on GSI (Grid Security Infrastructure), using X.509 certificate proxies and VOMS (Virtual Organization Membership Service) extensions. Even when other credentials, like SSH keys, were used to authenticate with resources, proxies were also added all the time, to establish the identity of the requestor and the associated memberships or privileges. This single credential was used for everything and was, often implicitly, forwarded wherever needed. The addition of identity and access tokens and the phase-out of GSI forced us to reconsider the security model of GlideinWMS, to handle multiple credentials which can differ in type, technology, and functionality. Both identity tokens and access tokens are supported. GSI proxies even if no more mandatory, are still used, together with various JWT (JSON Web Token) based tokens and other certificates. The functionality of the credentials, defined by issuer, audience, and scope, also differ: a credential can allow access to a computing resource, or can protect the GlideinWMS framework from tampering, or can grant read or write access to storage, can provide an identity for accounting or auditing, or can provide a combination of any the formers. Furthermore, the tools in use do not include automatic forwarding and renewal of the new credentials so credential lifetime and renewal requirements became part of the discussion as well. In this paper, we will present how GlideinWMS was able to change its design and code to respond to all these changes.

97 MATHEMATICS AND COMPUTING↗

Quantified uncertainty in thermodynamic modeling for materials design

Phase fractions, compositions and energies of the stable phases as a function of macroscopic composition, temperature, and pressure (X-T-P) are the principle correlations needed for the design of new materials and improvement of existing materials. They are the outcomes of thermodynamic modeling based on the CALculation of PHAse Diagrams (CALPHAD) approach. Here, the accuracy of CALPHAD predictions vary widely in X-T-P space due to experimental error, model inadequacy and unequal data coverage. In response, researchers have developed frameworks to quantify the uncertainty of thermodynamic property model parameters and propagate it to phase diagram predictions. In most previous studies, uncertainty was represented as intervals on phase boundaries (with respect to composition or temperature) and was unable to represent the uncertainty in invariant reactions or in the stability of phase regions. In this work, we propose a suite of tools that leverages samples from the multivariate model parameter distribution to represent uncertainty in forms that surpass previous limitations and are well suited to materials design. These representations include the distribution of phase diagrams and their features, as well as the dependence of phase stability and the distributions of phase fraction, composition, activity and Gibbs energy on X-T-P location - irrespective of the total number of components. Most critically, the new methodology allows the material designer to interrogate a certain composition and temperature domain and get in return the probability of different phases to be stable, which can positively impact materials design.

36 MATERIALS SCIENCE↗

Robust Vector BOTDA Signal Processing with Probabilistic Machine Learning

This paper presents a novel probabilistic machine learning (PML) framework to estimate the Brillouin frequency shift (BFS) from both Brillouin gain and phase spectra of a vector Brillouin optical time-domain analysis (VBOTDA). The PML framework is used to predict the Brillouin frequency shift (BFS) along the fiber and to assess its predictive uncertainty. We compare the predictions obtained from the proposed PML model with a conventional curve fitting method and evaluate the BFS uncertainty and data processing time for both methods. The proposed method is demonstrated using two BOTDA systems: (i) a BOTDA system with a 10 km sensing fiber and (ii) a vector BOTDA with a 25 km sensing fiber. The PML framework provides a pathway to enhance the VBOTDA system performance.

47 OTHER INSTRUMENTATION↗

Topological Structures in Unconventional Scenario for 2D Cuprates

Numerous experimental data point to cuprates as d-d charge transfer unstable systems whose description implies the inclusion of the three many-electron valence states CuO{sub 4}{sup 7−,6−,5−} (nominally Cu{sup 1+,2+,3+}) on an equal footing as a well-defined charge triplet. We introduce a minimal model to describe the charge degree of freedom in cuprates with the on-site Hilbert space reduced to only the three states and make use of the S = 1 pseudospin formalism. The formalism constitutes a powerful method to study complex phenomena in interacting quantum systems characterized by the coexistence and competition of various ordered states. Overall, such a framework provides a simple and systematic methodology to predict and discover new kinds of orders. In particular, the pseudospin formalism provides the most effective way to describe different topological structures, in particular, due to a possibility of a geometrical two-vector description of the on-site states. We introduce and analyze effective pseudospin Hamiltonian with on-site and inter-site charge correlations, two types of a correlated one-particle transfer and two-particle, or the composite boson transfer. The latter is of a principal importance for the HTSC perspectives. The 2D S = 1 pseudospin system is prone to a creation of different topological structures, which form topologically protected inhomogeneous distributions of the eight local S = 1 pseudospin order parameters. We present a short overview of localized topological structures, typical for S = 1 (pseudo)spin systems, focusing on unexpected antiphase domain walls in parent cuprates and so-called quadrupole skyrmion, which are believed to be candidates for a topological charge excitation in parent or underdoped cuprates. Puzzlingly, these unconventional structures can be characterized by an uniform distribution of the mean on-site charge, that makes these invisible for X-rays. Quasi-classical approximation and computer simulation are applied to analyze localized topological defects and evolution of the domain structures in “negative-U” model under charge order-superfluid phase transition.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Computation of Domain-Averaged Shortwave Irradiance by a One-Dimensional Algorithm Incorporating Correlations between Optical Thickness and Direct Incident Radiation

A one-dimensional radiative transfer algorithm that accounts for correlations between the optical thickness and the incident direct solar radiation is developed to compute the domain-averaged shortwave irradiance profile. It divides the direct irradiance into four components and treats the direct irradiance in two separate, clear and cloudy columns to account for the fact that clouds attenuate the direct irradiance more than clear-sky. The horizontal inhomogeneity of clouds in the cloudy column is treated by the gamma weighted two-stream approximation, which assumes that the optical thickness of clouds follows a gamma distribution. The algorithm inputs the cloud fraction, cumulative cloud fraction as a function of height, and a parameter expressing the shape of the probability density function of the cloud optical thickness distribution in addition to inputs required for a two-stream radiative transfer model. These cloud property inputs can be obtained using ground- and satellite-based instruments. Therefore, the algorithm can treat realistic cloud overlap features and horizontal inhomogeneity of clouds in a framework of one- dimensional radiative transfer. Heating rates computed by the algorithm using cloud fields generated by cloud resolving models agree with those computed with a Monte Carlo model. If optical properties in computational layers that divide a vertically extensive cloud are correlated, the irradiance profile computed by the algorithm further improves.

Kato, S.↗

A flexible numerical approach for quantification of epistemic uncertainty

In the field of uncertainty quantification (UQ), epistemic uncertainty often refers to the kind of uncertainty whose complete probabilistic description is not available, largely due to our lack of knowledge about the uncertainty. Quantification of the impacts of epistemic uncertainty is naturally difficult, because most of the existing stochastic tools rely on the specification of the probability distributions and thus do not readily apply to epistemic uncertainty. And there have been few studies and methods to deal with epistemic uncertainty. A recent work can be found in [J. Jakeman, M. Eldred, D. Xiu, Numerical approach for quantification of epistemic uncertainty, J. Comput. Phys. 229 (2010) 4648–4663], where a framework for numerical treatment of epistemic uncertainty was proposed. The method is based on solving an encapsulation problem, without using any probability information, in a hypercube that encapsulates the unknown epistemic probability space. If more probabilistic information about the epistemic variables is known a posteriori, the solution statistics can then be evaluated at post-process steps. In this paper, we present a new method, similar to that of Jakeman et al. but significantly extending its capabilities. Most notably, the new method (1) does not require the encapsulation problem to be in a bounded domain such as a hypercube; (2) does not require the solution of the encapsulation problem to converge point-wise. In the current formulation, the encapsulation problem could reside in an unbounded domain, and more importantly, its numerical approximation could be sought in L{sup p} norm. These features thus make the new approach more flexible and amicable to practical implementation. Both the mathematical framework and numerical analysis are presented to demonstrate the effectiveness of the new approach.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Enabling probabilistic learning on manifolds through double diffusion maps

Here, we present a generative learning framework for probabilistic sampling that extends Probabilistic Learning on Manifolds (PLoM), which is designed to generate statistically consistent realizations of a random vector in a finite-dimensional Euclidean space, informed by a (representative) set of observations. In its original form, PLoM constructs a reduced-order probabilistic model by combining three main components: (a) kernel density estimation to approximate the underlying probability measure, (b) Diffusion Maps to characterize the manifold of the data, and (c) a reduced-order Itô Stochastic Differential Equation (ISDE) to sample from the learned distribution. However, its sampling dynamics are posed in the ambient space and the retained number of reduced coordinates is chosen by projection-reconstruction error. In practice, this often (i) requires more coordinates than the data’s intrinsic dimension to achieve stable sampling and (ii) lacks a smooth, basis-independent lifting back to the data domain; moreover, standard Diffusion Maps emphasize harmonic eigenfunctions and can miss non-harmonic latent structure. We address these limitations by decoupling geometry learning from sampling: a first Diffusion Maps pass identifies non-harmonic coordinates on which we formulate a full-order ISDE directly in the latent space, while Double Diffusion Maps captures multiscale geometric features and Geometric Harmonics (GH) learns a smooth lifting map to the ambient variables that is independent of the particular diffusion basis. This hybrid design preserves the system’s dynamical richness with a compact geometric representation and enables principled out-of-sample inference. The effectiveness and robustness of the proposed method are illustrated through two numerical studies: one based on data generated from two-dimensional Hermite polynomial functions and another based on high-fidelity simulations of a detonation wave in a reactive flow.

Double diffusion maps↗