Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Simulation of distributed computing systems”

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 469 records · Page 26

Implementation of a Discrete Dipole Approximation Scattering Database into Community Radiative Transfer Model

The Community Radiative Transfer Model (CRTM) is a fast model that requires bulk optical properties of hydrometeors in the form of lookup tables to simulate all-sky satellite radiances. Current cloud scattering lookup tables of CRTM were generated using the Mie-Lorenz theory thus assuming spherical shapes for all frozen habits, while actual clouds contain frozen hydrometeors with different shapes. The Discrete Dipole Approximation (DDA) technique is an effective technique for simulating the optical properties of non-spherical hydrometeors in the microwave region. This paper discusses the implementation and validation of a comprehensive DDA cloud scattering database into CRTM for the microwave frequencies. The original DDA database assumes total random orientation in the calculation of single scattering properties. The mass scattering parameters required by CRTM were then computed from single scattering properties and water content dependent particle size distributions. The new lookup tables eliminate the requirement for providing the effective radius as input to CRTM by using the cloud water content for the mass dimension. A collocated dataset of short-term forecasts from Integrated Forecast System of the European Centre for Medium-Range Weather Forecasts and satellite microwave data was used for the evaluation of results. The results overall showed that the DDA lookup tables, in comparison with the Mie tables, greatly reduce the differences among simulated and observed values. The Mie lookup tables especially introduce excessive scattering for the channels operating below 90 GHz and low scattering for the channels above 90 GHz.

Discrete dipole↗

Detector Interface for Streaming, Control, and Open-source integration (DISCO) v1.0.0

This suite consists of a multi-package ecosystem featuring detector emulators, EPICS areaDetector drivers, and remote server frameworks designed for the Advanced Light Source (ALS). Engineered for high-bandwidth devices—including VFCCD, Timepix3, Timepix4, and related pixel detectors—the software simulates hardware, wraps vendor SDKs into remote-callable servers, and integrates with open-source control systems. Key Capabilities: Distributed SDK Architecture: Server packages wrap hardware-specific SDKs, allowing areaDetector drivers to execute remote framework calls. This isolates proprietary libraries from the EPICS IOC, enhancing stability and enabling distributed computing across beamline networks. Device Support: Custom drivers for VFCCD, the Timepix family, and similar sensors optimize the data path from hardware control to high-speed transport. Full-Stack Emulation: Sophisticated emulator packages allow end-to-end pipeline testing and software development without requiring physical hardware or beam time. Integrated Workflows: Supports high-bandwidth streaming for real-time analysis and robust, metadata-rich file-based workflows (e.g., HDF5/NeXus). By standardizing interfaces across heterogeneous hardware, this suite reduces technical debt. It provides the ALS with a scalable, open-source solution to manage massive data rates within a unified control environment.

Mahl, Johannes [Lawrence Berkeley National Laborat↗

CFD modeling of natural circulation in LiCl-KCl molten salt closed loop

Characterizing flow within a molten salt closed-loop system is crucial for assessing system requirements, evaluating performance, and identifying potential flaws. Direct flow measurement using instrumentation is challenging due to extreme environmental conditions and the limitations associated with measuring molten salt flow under natural convection. Here, this study aims to provide comprehensive insights into the thermal-hydraulic behavior of a closed loop, with a particular focus on temperature distribution and velocity prediction. The Computational Fluid Dynamics (CFD) model demonstrated the capability to effectively simulate and predict both temperature distributions and flow velocities within the molten salt loop. The CFD model's predictive capability was validated by its ability to replicate temperature measurements under varying boundary conditions. The analysis revealed that the CFD model tends to underpredict temperatures in the cold leg and overpredict them in the hot leg, highlighting the need for continuous model refinement and acknowledging the limitations of using a steady-state approach. Furthermore, the potential of using external temperature measurements to estimate internal molten salt temperatures and predict flow velocity was explored, revealing that this approach could introduce up to a 5.5% error in flow velocity calculations. Line probes mapping temperature distributions across the tube's cross-section and molten salt provided valuable insights into temperature gradients, emphasizing the need for a thermal conductivity equation for molten salt with lower uncertainty to achieve more accurate temperature predictions of the system.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗

CFD Analysis of the 24-inch JIRAD Hybrid Rocket Motor

A series of multispecies, multiphase computational fluid dynamics (CFD) analyses of the 24-inch diameter joint government industry industrial research and development (JIRAD) hybrid rocket motor is described. The 24-inch JIRAD hybrid motor operates by injection of liquid oxygen (LOX) into a vaporization plenum chamber upstream of ports in the hydroxyl-terminated polybutadiene (HTPB) solid fuel. The injector spray pattern had a strong influence on combustion stability of the JIRAD motor so a CFD study was initiated to define the injector end flow field under different oxidizer spray patterns and operating conditions. By using CFD to gain a clear picture of the flow field and temperature distribution within the JIRAD motor, it is hoped that the fundamental mechanisms of hybrid combustion instability may be identified and then suppressed by simple alterations to the oxidizer injection parameters such as injection angle and velocity. The simulations in this study were carried out using the General Algorithm for Analysis of Combustion SYstems (GALACSY) multiphase combustion codes. GALACSY consists of a comprehensive set of droplet dynamic submodels (atomization, evaporation, etc.) and a computationally efficient hydrocarbon chemistry package built around a robust Navier-Stokes solver optimized for low Mach number flows. Lagrangian tracking of dispersed particles describes a closely coupled spray phase. The CFD cases described in this paper represent various levels of simplification of the problem. They include: (A) gaseous oxygen with combusting fuel vapor blowing off the walls at various oxidizer injection angles and velocities, (B) gaseous oxygen with combusting fuel vapor blowing off the walls, and (C) liquid oxygen with combusting fuel vapor blowing off the walls. The study used an axisymmetric model and the results indicate that the injector design significantly effects the flow field in the injector end of the motor. Markedly different recirculation patterns are observed in the vaporization chamber as the oxygen velocity and/or spray pattern is varied. The ability of these recirculation patterns to stabilize the diffusion flame above the surface of the solid fuel gives a plausible explanation for the experimentally determined combustion stability characteristics of the JIRAD motor, and suggests how combustion stability can be assured by modifications to the injector design.

Liang, Pak-Yan↗

RLC4CLR (Reinforcement Learning Controller for Critical Load Restoration Problems)

RLC4CLR demonstrates using a reinforcement learning controller (RLC) to solve a critical load restoration (CLR) problem, which improves the grid resilience after a substation outage event. RLC4CLR consists of two parts. (1) RL environment: This environment encapsulates the CLR problem to be solved and provides interfacing functions to follow the standard OpenAI Gym format. A power system simulator, i.e., OpenDSS, is included to provide the power flow solution. Controller inputs and outputs (RL state and action) as well as the reward are defined in this environment as well. In summary, the RL environment is the problem formulation from which the RL agent can learn. (2) RL training script: The training script enables the RL agent to learn its control policy by interacting with the RL environment. For RL training, an open-sourced RL library, i.e., RLlib, is leveraged which is based on a distributed computing framework (Ray). The training script is designed to be able to be run on both local machine or the NREL HPC system. Other components of RLC4CLR include input data, e.g., grid model (standard IEEE test feeders), and other files used for results analysis.

Zhang, Xiangyu↗

Multidimensional Modeling of Mixture Formation in a Hydrogen-Fueled Heavy-Duty Optical Engine With Direct Injection

Hydrogen (H 2 ), as a carbon-free fuel, is considered as one of the most promising solutions to reduce the carbon footprint of hard-to-decarbonize energy and transportation sectors. As such, hydrogen-fueled internal combustion engines (H 2 ICEs) have recently been receiving increasing attention, particularly in applications such as on-road/off-road heavy-duty transport and combined heat and power. The direct injection (DI) of gaseous hydrogen into the combustion chamber offers great potential for achieving high power density and high engine efficiency, while mitigating the risk of backfire and reducing pre-ignition. However, the numerical simulation of H 2 DI system remains a formidable challenge associated with the high computational cost of reproducing compressible supersonic flow and shocks in narrow injector passages and in near-nozzle regions. In general, there is a lack of well-established and validated practices for the modeling of high-pressure H 2 DI in large-bore engines. Here, to this end, this study focuses on computational fluid dynamics (CFD) modeling of the mixture formation process in a heavy-duty optical engine employing a medium-pressure H 2 DI system. Both large eddy simulations (LES) and Reynolds Averaged Navier–Stokes (RANS) simulations are performed and evaluated against optical data. Gaseous hydrogen is injected into the combustion chamber via a centrally located outward opening hollow-cone injector at a pressure of 40 bar. Simulations are carried out for two injection timings, namely, −120 and −60 °CA. The numerical predictions for H 2 distribution in different horizontal and vertical planes during the compression stroke are systematically compared against optical data obtained through planar laser-induced fluorescence (PLIF) measurements. Overall, the LES approach using the Dynamic Structure model is found to have good predictive capabilities for the early jet penetration in terms of length and shape, as well as the later H 2 distributions. However, the unsteady RANS approach with the renormalization group $k - ϵ$ model, which is widely used by industry to model heavy-duty ICEs, significantly underpredicts the H 2 mixing, even at similar mesh resolution to that used in LES. These results indicate that there is a need for the improvement of mixing submodels within the RANS approach when applied to H 2 DI simulations.

LES↗

CAMERA: A method for cost-aware, adaptive, multifidelity, efficient reliability analysis

Estimating probability of failure in aerospace systems is a critical requirement for flight certification and qualification. Failure probability estimation involves resolving tails of probability distributions, and Monte Carlo sampling methods are intractable when expensive high-fidelity simulations have to be queried. Here, we propose a method to use models of multiple fidelities that trade accuracy for computational efficiency. Specifically, we propose the use of multifidelity Gaussian process models to efficiently fuse models at multiple fidelity, thereby offering a cheap surrogate model that emulates the original model at all fidelities. Furthermore, we propose a novel sequential acquisition function based experiment design framework that can automatically select samples from appropriate fidelity models to make predictions about quantities of interest at the highest fidelity. We use our proposed approach in an importance sampling setting and demonstrate our method on the failure level set and probability estimation on synthetic test functions and two real-world applications, namely, the reliability analysis of a gas turbine engine blade using a finite element method and a transonic aerodynamic wing test case using Reynolds-averaged Navier-Stokes equations. We show that our method predicts the failure boundary and probability more accurately and at a fraction of the computational cost compared with using just a single expensive high-fidelity model. Finally, we show that our sequential approach is guaranteed to asymptotically converge to the true failure boundary with high probability.

97 MATHEMATICS AND COMPUTING↗

A potential model study of the nucleon's charge and mass radius

We study the charge and mass distributions within a nucleon and compute the associated squared radii based on a potential model approach. Different constituent quark configurations such as Δ, Y, and quark-diquark are considered and compared, with model parameters calibrated by experimental measurements of the proton and neutron charge radius. The results suggest that while the charge radius is dictated by quark dynamics, the mass radius is strongly influenced by nonperturbative QCD contributions to a nucleon's mass that are not sensitive to the constituent quarks. As a result, the mass radius could become substantially different from the charge radius. The obtained nucleon mass distributions of different configurations are further used for simulations of the initial conditions in heavy ion collisions. Furthermore, the computed eccentricities ε 2 and ε 3 are found to demonstrate a considerable sensitivity to the input nucleon profiles, especially to the mass radius in the peripheral region as well as for systems with fewer participants.

Gallimore, Daniel [Indiana University, Bloomington↗

AENET–LAMMPS and AENET–TINKER : Interfaces for accurate and efficient molecular dynamics simulations with machine learning potentials

Machine-learning potentials (MLPs) trained on data from quantum-mechanics based first-principles methods can approach the accuracy of the reference method at a fraction of the computational cost. To facilitate efficient MLP-based molecular dynamics and Monte Carlo simulations, an integration of the MLPs with sampling software is needed. Here, we develop two interfaces that link the atomic energy network (ænet) MLP package with the popular sampling packages TINKER and LAMMPS. The three packages, ænet, TINKER, and LAMMPS, are free and open-source software that enable, in combination, accurate simulations of large and complex systems with low computational cost that scales linearly with the number of atoms. Scaling tests show that the parallel efficiency of the ænet–TINKER interface is nearly optimal but is limited to shared-memory systems. The ænet–LAMMPS interface achieves excellent parallel efficiency on highly parallel distributed memory systems and benefits from the highly optimized neighbor list implemented in LAMMPS. We demonstrate the utility of the two MLP interfaces for two relevant example applications: the investigation of diffusion phenomena in liquid water and the equilibration of nanostructured amorphous battery materials.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Dose Rate Analysis of the WCS Consolidated Interim Storage Facility

This report describes a confirmatory calculation for a proposed consolidated interim storage facility (CISF) to support the US Nuclear Regulatory Commission’s review of the license application submitted by the Interim Storage Partners. The scope of this report is limited to an assessment of the annual dose from direct radiation associated with Phase 1 of the Waste Control Specialists (WCS) CISF and the determination of the minimum controlled area boundary based on the 10 CFR 72.104 limit of 25 mrem annual dose to the whole body. The confirmatory dose rate calculations used the source term and shielding calculation capabilities of the SCALE 6.2.3 computer code system. The calculation method employed in the confirmatory calculations uses a detailed Monte Carlo radiation transport simulation from source to dose rate. This method differs from the two-step method used in the safety analysis report (SAR), which requires determination of the photon and neutron energy and angular distributions on the cask external surface and then the use of this surface source in a new radiation transport calculation to determine dose rate as a function of distance from the storage facility. A simulation was made using the complete site geometry (all casks present) but with only one cask containing source, and a total of 467 independent calculations were performed to obtain the dose rate maps produced by each storage cask. Basic input data and assumptions used in the SAR (e.g., cask design parameters and design basis assembly characteristics) were used in this confirmatory calculation for consistency with the SAR. Local atmospheric conditions for skyshine calculations were simulated using US standard atmosphere data, which describe average air temperature, pressure, and density as a function of altitude. The determined minimum required distances from the WCS Phase 1 CISF to a contour defined by the 25 mrem annual dose limit were approximately 623 m (0.387 miles) in the N-NE direction, 594 m (0.369 miles) in the SSW direction, and 533 m (0.331 miles) in the E-SE direction and the W-NW direction. The annual dose results of the confirmatory calculation are in relatively good agreement with the annual dose values presented in the WCS SAR. The annual dose produced by the Phase 1 CISF at the nearest site boundary, 0.75 miles away from the CISF, was estimated by this study to be approximately 6.02E-02 ± 2.0E-04 mrem, which is significantly smaller than the 25 mrem annual dose limit provided in 10 CFR 72.104. The SAR annual dose rate value for the nearest site boundary was 7.52E-2 mrem, which included 7.73E-03 mrem due to postulated leakage of the FO-, FC-, and FF-canisters. A comparison with the dose rate values reported in the SAR for 16 detector locations, D1 through D16, on the protected area boundary is also included in this report. This comparison shows that the SAR determined higher annual dose values for detectors D1 through D14 by 12% to 45% and lower annual dose values for detectors D15 and D16 by 18% and 8%, respectively, as compared with this ORNL confirmatory calculation. Sensitivity studies have been performed to support a comprehensive review of the application. In these studies, the effects on dose rate of the modeled air volume, density, and humidity were analyzed. The sensitivity calculations showed that the air volume used in the Monte Carlo simulations was adequate and that a larger air volume would not further increase the dose rate due to skyshine contributions. The sensitivity analysis showed that simulation of local atmospheric conditions is important for the accuracy of dose rate estimates at large distances from the facility. Air humidity was demonstrated to have negligible effects on dose rate values for the simulated local atmospheric conditions.

61 RADIATION PROTECTION AND DOSIMETRY↗

An evaluative model of system performance in manned teleoperational systems

Manned teleoperational systems are used in aerospace operations in which humans must interact with machines remotely. Manual guidance of remotely piloted vehicles, controling a wind tunnel, carrying out a scientific procedure remotely are examples of teleoperations. A four input parameter throughput (Tp) model is presented which can be used to evaluate complex, manned, teleoperations-based systems and make critical comparisons among candidate control systems. The first two parameters of this model deal with nominal (A) and off-nominal (B) predicted events while the last two focus on measured events of two types, human performance (C) and system performance (D). Digital simulations showed that the expression A(1-B)/C+D) produced the greatest homogeneity of variance and distribution symmetry. Results from a recently completed manned life science telescience experiment will be used to further validate the model. Complex, interacting teleoperational systems may be systematically evaluated using this expression much like a computer benchmark is used.

Haines, Richard F.↗

Assessment of the Mars Science Laboratory Entry, Descent, and Landing Simulation

On August 5, 2012, the Mars Science Laboratory rover, Curiosity, successfully landed inside Gale Crater. This landing was only the seventh successful landing and fourth rover to be delivered to Mars. Weighing nearly one metric ton, Curiosity is the largest and most complex rover ever sent to investigate another planet. Safely landing such a large payload required an innovative Entry, Descent, and Landing system, which included the first guided entry at Mars, the largest supersonic parachute ever flown at Mars, and a novel and untested Sky Crane landing system. A complete, end-to-end, six degree-of-freedom, multi-body computer simulation of the Mars Science Laboratory Entry, Descent, and Landing sequence was developed at the NASA Langley Research Center. In-flight data gathered during the successful landing is compared to pre-flight statistical distributions, predicted by the simulation. These comparisons provide insight into both the accuracy of the simulation and the overall performance of the vehicle.

Way, David W.↗

Feeding Ten Billion People Is Possible Within Four Terrestrial Planetary Boundaries

Global agriculture puts heavy pressure on planetary boundaries, posing the challenge to achieve future food security without compromising Earth system resilience. On the basis of process-detailed, spatially explicit representation of four interlinked planetary boundaries (biosphere integrity, land-system change, freshwater use, nitrogen flows) and agricultural systems in an internally consistent model framework, we here show that almost half of current global food production depends on planetary boundary transgressions. Hotspot regions, mainly in Asia, even face simultaneous transgression of multiple underlying local boundaries. If these boundaries were strictly respected, the present food system could provide a balanced diet (2,355 kcal per capita per day) for 3.4 billion people only. However, as we also demonstrate, transformation towards more sustainable production and consumption patterns could support 10.2 billion people within the planetary boundaries analysed. Key prerequisites are spatially redistributed cropland, improved water–nutrient management, food waste reduction and dietary changes. Adoption of the Sustainable Development Goals by all nations in 2015 is the first ever commitment to a world development path that safeguards the stability of the Earth system as a prerequisite for meeting universal human standards1. The longstanding challenge of achieving food security through sustainable agriculture is particularly acute in this context as world agriculture is a leading cause for the current transgressions of multiple planetary boundaries (PBs) globally and regionally2–5. The PB framework is a comprehensive scientific attempt to synoptically define our planet’s biogeophysical limits to anthropogenic interference. It suggests bounds to nine interacting processes that together delineate a Holocene-like Earth system state. The Holocene is chosen as the reference state as it is the only period known to provide a safe operating space for a world population of several billion people, and according to a precautionary principle, the PBs are set in sufficient distance from processes that may critically undermine Earth system resilience and global sustainability. A challenging question, thus, is whether human development goals such as food security can be met while maintaining multiple PBs along with their subglobal manifestations. Further PB transgressions could jeopardize the chances of providing sufficient food for a world population projected to be wealthier and reach >9 billion by 2050. This conundrum portrays a tradeoff between Earth’s biophysical carrying capacity and humankind’s rising food demand, calling in response for radical rethinking of food production and consumption patterns6–9. Yield gap closures, avoidance of excessive input use, shifts towards less resource-demanding diets, food waste reductions and efficient international trade are crucial options for sustainably increasing the food supply10–15. For example, enhancing water-use efficiency on irrigated and rain-fed farms can triple or quadruple crop yields in low-performing systems, suggesting possible global gains of >20% (ref. 16). Even higher gains appear feasible through globally optimized configurations of the land-use pattern17, and cutting food losses by half could generate food for another billion people18. Thus, collective large-scale implementation of such options could sustain food for a further growing world population19. Yet achieving this within a safe operating space as defined by PBs requires not only a halt to but actually a reversal of existing PB transgressions. Previous studies suggest that such a reconciliation might be possible, but these were based on aggregate representations of PBs (not accounting for the spatial patterns of limits, transgressions and interactions) or considered only one boundary in isolation17,20–23. Here, we systematically quantify to what extent current food production depends on local to global transgressions of the PBs for biosphere integrity, land-system change, freshwater use and nitrogen (N) flows, along with the potential of a range of solutions to avoid these transgressions and still increase food supply (Table 1). To this end, we configured an internally consistent process-based model of the terrestrial biosphere including agriculture (LPJmL) with multiple spatially distributed PBs and their interactions. LPJmL is among the longest-established and best-evaluated biosphere models, showing robust performance regarding simulation of, for example, carbon, water and crop yield dynamics (Supplementary Figs. 1 and 2 and Supplementary Table 1; see ref. 24 for a comprehensive benchmarking and Supplementary Methods for more detail on model evaluations). In principle following established definitions4, we refine the computation of some PBs with respect to their regional patterns and interactions (Methods), providing globally gridded precautionary limits to human interference with the Earth system at a level of great detail. In particular, we account for the evidence that many PBs need to be represented spatially explicitly4 to cover their

Gerten, Dieter↗

Distributed deep learning training using silicon photonic switched architectures

The scaling trends of deep learning models and distributed training workloads are challenging network capacities in today’s datacenters and high-performance computing (HPC) systems. We propose a system architecture that leverages silicon photonic (SiP) switch-enabled server regrouping using bandwidth steering to tackle the challenges and accelerate distributed deep learning training. In addition, our proposed system architecture utilizes a highly integrated operating system-based SiP switch control scheme to reduce implementation complexity. To demonstrate the feasibility of our proposal, we built an experimental testbed with a SiP switch-enabled reconfigurable fat tree topology and evaluated the network performance of distributed ring all-reduce and parameter server workloads. The experimental results show up to 3.6× improvements over the static non-reconfigurable fat tree. Our large-scale simulation results show that server regrouping can deliver up to 2.3× flow throughput improvement for a 2× tapered fat tree and a further 11% improvement when higher-layer bandwidth steering is employed. The collective results show the potential of integrating SiP switches into datacenters and HPC systems to accelerate distributed deep learning training.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A Management Information System Model for Program Management

The development of a model to simulate the information system of a program management type of organization is reported. The model statistically determines the following parameters: type of messages, destinations, delivery durations, type processing, processing durations, communication channels, outgoing messages, and priorites. The total management information system of the program management organization is considered, including formal and informal information flows and both facilities and equipment. The model is written in General Purpose System Simulation 2 computer programming language for use on the Univac 1108, Executive 8 computer. The model is simulated on a daily basis and collects queue and resource utilization statistics for each decision point. The statistics are then used by management to evaluate proposed resource allocations, to evaluate proposed changes to the system, and to identify potential problem areas. The model employs both empirical and theoretical distributions which are adjusted to simulate the information flow being studied.

Shipman, D. L.↗

Presumed PDF Modeling of Early Flame Propagation in Moderate to Intense Turbulence Environments

The present paper describes the results obtained from a one-dimensional time dependent numerical technique that simulates early flame propagation in a moderate to intense turbulent environment. Attention is focused on the development of a spark-ignited, premixed, lean methane/air mixture with the unsteady spherical flame propagating in homogeneous and isotropic turbulence. A Monte-Carlo particle tracking method, based upon the method of fractional steps, is utilized to simulate the phenomena represented by a probability density function (PDF) transport equation. Gaussian distributions of fluctuating velocity and fuel concentration are prescribed. Attention is focused on three primary parameters that influence the initial flame kernel growth: the detailed ignition system characteristics, the mixture composition, and the nature of the flow field. The computational results of moderate and intense isotropic turbulence suggests that flames within the distributed reaction zone are not as vulnerable, as traditionally believed, to the adverse effects of increased turbulence intensity. It is also shown that the magnitude of the flame front thickness significantly impacts the turbulent consumption flame speed. Flame conditions studied have fuel equivalence ratio s in the range phi = 0.6 to 0.9 at standard temperature and pressure.

Carmen, Christina↗

DeepClimGAN: A High-Resolution Climate Data Generator

Earth system models (ESMs), which simulate the physics and chemistry of the global atmosphere, land, and ocean, are often used to generate future projections of climate change scenarios. These models are far too computationally inten- sive to run repeatedly, but limited sets of runs are insufficient for some important applications, like adequately sampling distribution tails to characterize extreme events. As a com- promise, emulators are substantially less expensive but may not have all of the complexity of an ESM. Here we demon- strate the use of a conditional generative adversarial network (GAN) to act as an ESM emulator. In doing so, we gain the ability to produce daily weather data that is consistent with what ESM might output over any chosen scenario. In particular, the GAN is aimed at representing a joint proba- bility distribution over space, time, and climate variables, enabling the study of correlated extreme events, such as floods, droughts, or heatwaves.

Extreme events, Machine Learning, deep learning↗