SEARCH · Search NASA

Results for “Model Inference”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 955 records · Page 53

All sources NASA NTRS DOE OSTI

DOE OSTI2020

Fast inference of Boosted Decision Trees in FPGAs for particle physics

We describe the implementation of Boosted Decision Trees in the hls4ml library, which allows the translation of a trained model into FPGA firmware through an automated conversion process. Thanks to its fully on-chip implementation, hls4ml performs inference of Boosted Decision Tree models with extremely low latency. With a typical latency less than 100 ns, this solution is suitable for FPGA-based real-time processing, such as in the Level-1 Trigger system of a collider experiment. These developments open up prospects for physicists to deploy BDTs in FPGAs for identifying the origin of jets, better reconstructing the energies of muons, and enabling better selection of rare signal processes.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2021

A second order yield-temperature relation for accurate inference of burn-averaged quantities in multi-species plasmas

Measured yields and ion temperatures inferred from the fusion product energy spectra can be used as metrics for the performance of an ICF implosion. This can be used to infer species separation, thermal decoupling, flows, or other effects that can cause the inferred ion temperatures to deviate from the true underlying thermal temperature and the yield ratio to deviate from the expected value. Direct inference of the impact of these effects on observed temperatures and yields can be difficult to uncover due to the underlying dependence on the shape and time evolution of the temperature and density profiles of the fusing plasma. Due to differences in the temperature dependence of the reactivities, different fusion products are emitted from different regions and times within the implosion. In order to properly account for this, a second-order analytical expression relating the apparent temperatures and yield ratios is developed. This expression can be coupled to models of yield and/or temperature altering effects to infer their burn-averaged impact on an implosion. The second-order expression shows significant improvement over lower-order expressions in synthetic data studies. Demonstrations of its applications to synthetic data coupled with models of ion thermal decoupling and radial flows are presented. In the case of thermal decoupling, both first and second-order expressions show reasonable levels of accuracy. To consistently infer the amplitude of radial flow with a <10% error, the second-order equation is required.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2026

FolpsD: combining EFT and phenomenological approaches for joint power spectrum and bispectrum analyses

We present a theoretical model for the power spectrum and bispectrum of galaxy clustering that exploits the complementarity between small-scale power spectrum information and large-scale bispectrum measurements. We extend the FOLPS code by combining its one-loop EFT galaxy power spectrum with a tree-level galaxy bispectrum projected onto the tripolar spherical harmonics (Sugiyama) basis. To access additional small-scale information, we also consider a line-of-sight damping factor in both statistics, mirroring approaches commonly used in studies of redshift-space distortions. We test the model using DESI DR2 galaxy mocks. Even without damping, the joint analysis of the EFT power spectrum and bispectrum significantly improves constraints and reduces parameter degeneracies relative to power spectrum analyses alone. For LRG-like samples, including the damping further extends the range beyond $k\sim 0.3 \,h \text{Mpc}^{-1}$ in the power spectrum and $k \sim 0.24 \,h \text{Mpc}^{-1}$ in the bispectrum without introducing statistically significant parameter biases. This leads to up to $\sim 30\%$ tighter constraints on $A_s$ and $ω_{cdm}$. For low signal-to-noise tracers such as QSOs, however, the damping parameters are weakly constrained and can absorb noise fluctuations, leading to shifts in inferred parameters. Similar limitations may arise in models where cosmological information is encoded in power-spectrum shape features degenerate with the damping, such as scenarios with massive neutrinos. In contrast, for $w_0w_a$CDM we obtain $15\%$ and $21\%$ tighter constraints on $w_0$ and $w_a$, respectively, yielding a deviation from constant dark energy at slightly more than the $1σ$ level using full-shape information alone. The code is publicly available at https://github.com/cosmodesi/FolpsD

Bansal, P. [Michigan U., MCTP; Michigan U.] (ORCID↗

DOE OSTI2023

Examination of nucleon distribution with Bayesian imaging for isobar collisions

Relativistic collision of isobaric systems is found to be valuable in differentiating the nucleon distributions for nuclei with the same mass number. In recent contrast experiment of $^{96}_{44}$Ru + $^{96}_{44}$Ru versus $^{96}_{40}$Zr + $^{96}_{40}$Zr collisions at $\sqrt{s_{NN}}$ = 200 GeV, the ratios of multiplicity distribution, elliptic flow, triangular flow, and radial flow are precisely measured and found to be significantly different from unity, indicating the difference in the shapes of the isobar pair. Here, in this work, we investigate the feasibility of nuclear structure reconstruction from heavy-ion collision observables. We perform Bayesian Inference with employing the Monte-Carlo Glauber model as an estimator of the mapping from nuclear structure to the final state observables and to provide the mock data for reconstruction. By varying combination of observables included in the mock data, we find it plausible to infer Woods–Saxon parameters from the observables. We also observe that single-system multiplicity distribution for the isobar system, rather than their ratio, is crucial to simultaneously determine the nuclear structure for the isobar system.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2017

Final Report: Quantification of Uncertainty in Extreme Scale Computations (QUEST)

QUEST (\url{www.quest-scidac.org}) is a SciDAC Institute that is focused on uncertainty quantification (UQ) in large-scale scientific computations. Our goals are to (1) advance the state of the art in UQ mathematics, algorithms, and software; and (2) provide modeling, algorithmic, and general UQ expertise, together with software tools, to other SciDAC projects, thereby enabling and guiding a broad range of UQ activities in their respective contexts. QUEST is a collaboration among six institutions (Sandia National Laboratories, Los Alamos National Laboratory, the University of Southern California, Massachusetts Institute of Technology, the University of Texas at Austin, and Duke University) with a history of joint UQ research. Our vision encompasses all aspects of UQ in leadership-class computing. This includes the well-founded setup of UQ problems; characterization of the input space given available data/information; local and global sensitivity analysis; adaptive dimensionality and order reduction; forward and inverse propagation of uncertainty; handling of application code failures, missing data, and hardware/software fault tolerance; and model inadequacy, comparison, validation, selection, and averaging. The nature of the UQ problem requires the seamless combination of data, models, and information across this landscape in a manner that provides a self-consistent quantification of requisite uncertainties in predictions from computational models. Accordingly, our UQ methods and tools span an interdisciplinary space across applied math, information theory, and statistics. The MIT QUEST effort centers on statistical inference and methods for surrogate or reduced-order modeling. MIT personnel have been responsible for the development of adaptive sampling methods, methods for approximating computationally intensive models, and software for both forward uncertainty propagation and statistical inverse problems. A key software product of the MIT QUEST effort is the MIT Uncertainty Quantification library, called MUQ (\url{muq.mit.edu}).

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

The charge reduction rate for multiply charged polymer ions via ion–ion recombination at atmospheric pressure

The charge reduction of multiply charged macromolecular ions via recombination with small ions in the gas phase is commonly employed to modulate the charge on macromolecules prior to mass spectrometric and mobility analyses. We employ a recently developed continuum-Molecular Dynamics (MD) calculation approach to determine the recombination rate coefficient of multiply charged (1 to 7 excess positive charged) polyethylene glycol ions (mass of 4600 Da) with smaller singly charged anions, modeled as NO 2 – ions. The continuum-MD approach accounts explicitly for the influence of the background gas on the recombination process, accounts explicitly for ion translational, vibrational, and rotational motion, and enables recombination rate coefficient calculation in nitrogen near atmospheric pressure, wherein neither low pressure nor high pressure recombination theories are strictly applicable. Continuum-MD simulations yield recombination rate coefficients near 3.9 × 10 –14 m 3 s –1 for singly charged ions, increasing to 3.0 × 10 –11 m 3 s –1 for the +7 ion. Pre-existing collision rate coefficient expressions for rigid ions are found to be within a factor of 2–5 of calculations for all charge states, but their use requires knowledge of an appropriate collision distance, which is not well-defined for flexible polymer ions. Continuum-MD-inferred rate coefficients are incorporated into a model of charge reduction, and the charge state distribution versus anion concentration determined with it is compared to charge reduction measurements made via atmospheric pressure differential mobility analysis. Furthermore, good agreement is observed between simulations and experiments; although results are extremely sensitive to the recombination rate coefficients, experimental results are bound by models utilizing rates within a factor of 2 (0.5–2.0×) of the continuum-MD rates.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

DOE OSTI2016

The Radiative Forcing Model Intercomparison Project (RFMIP): Experimental protocol for CMIP6

The phrasing of the first of three questions motivating CMIP6 – “How does the Earth system respond to forcing?” – suggests that forcing is always well-known, yet the radiative forcing to which this question refers has historically been uncertain in coordinated experiments even as understanding of how best to infer radiative forcing has evolved. The Radiative Forcing Model Intercomparison Project (RFMIP) endorsed by CMIP6 seeks to provide a foundation for answering the question through three related activities: (i) accurate characterization of the effective radiative forcing relative to a near-preindustrial baseline and careful diagnosis of the components of this forcing; (ii) assessment of the absolute accuracy of clear-sky radiative transfer parameterizations against reference models on the global scales relevant for climate modeling; and (iii) identification of robust model responses to tightly specified aerosol radiative forcing from 1850 to present. Complete characterization of effective radiative forcing can be accomplished with 180 years (Tier 1) of atmosphere-only simulation using a sea-surface temperature and sea ice concentration climatology derived from the host model's preindustrial control simulation. Assessment of parameterization error requires trivial amounts of computation but the development of small amounts of infrastructure: new, spectrally detailed diagnostic output requested as two snapshots at present-day and preindustrial conditions, and results from the model's radiation code applied to specified atmospheric conditions. In conclusion, the search for robust responses to aerosol changes relies on the CMIP6 specification of anthropogenic aerosol properties; models using this specification can contribute to RFMIP with no additional simulation, while those using a full aerosol model are requested to perform at least one and up to four 165-year coupled ocean–atmosphere simulations at Tier 1.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2017

Comoving Stars in Gaia DR1: An Abundance of Very Wide Separation Comoving Pairs

The primary sample of the Gaia Data Release 1 is the Tycho-Gaia Astrometric Solution (TGAS): ≈2 million Tycho-2 sources with improved parallaxes and proper motions relative to the initial catalog. This increased astrometric precision presents an opportunity to find new binary stars and moving groups. We search for high-confidence comoving pairs of stars in TGAS by identifying pairs of stars consistent with having the same 3D velocity using a marginalized likelihood ratio test to discriminate candidate comoving pairs from the field population. Although we perform some visualizations using (bias-corrected) inverse parallax as a point estimate of distance, the likelihood ratio is computed with a probabilistic model that includes the covariances of parallax and proper motions and marginalizes the (unknown) true distances and 3D velocities of the stars. We find 13,085 comoving star pairs among 10,606 unique stars with separations as large as 10 pc (our search limit). Some of these pairs form larger groups through mutual comoving neighbors: many of these pair networks correspond to known open clusters and OB associations, but we also report the discovery of several new comoving groups. Most surprisingly, we find a large number of very wide (>1 pc) separation comoving star pairs, the number of which increases with increasing separation and cannot be explained purely by false-positive contamination. Our key result is a catalog of high-confidence comoving pairs of stars in TGAS. We discuss the utility of this catalog for making dynamical inferences about the Galaxy, testing stellar atmosphere models, and validating chemical abundance measurements.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2025

From sequence to protein structure and conformational dynamics with artificial intelligence/machine learning

The 2024 Nobel Prize in Chemistry was awarded in part for de novo protein structure prediction using AlphaFold2, an artificial intelligence/machine learning (AI/ML) model trained on vast amounts of sequence and three-dimensional structure data. AlphaFold2 and related models, including RoseTTAFold and ESMFold, employ specialized neural network architectures driven by attention mechanisms to infer relationships between sequence and structure. At a fundamental level, these AI/ML models operate on the long-standing hypothesis that the structure of a protein is determined by its amino acid sequence. More recently, AlphaFold2 has been adapted for the prediction of multiple protein conformations by subsampling multiple sequence alignments. Herein, we provide an overview of the deterministic relationship between sequence and structure, which was hypothesized over half a century ago with profound implications for the biological sciences ever since. We postulate that protein conformational dynamics are also determined, at least in part, by amino acid sequence and that this relationship may be leveraged for construction of AI/ML models dedicated to predicting protein conformational ensembles. Accordingly, we describe a conceptual model architecture, which may be trained on sequence data in combination with conformationally sensitive structural information, coming primarily from nuclear magnetic resonance (NMR) spectroscopy. Notwithstanding certain limitations in this context, NMR offers abundant structural heterogeneity conducive to conformational ensemble prediction. As NMR and other data continue to accumulate, sequence-informed prediction of protein structural dynamics with AI/ML has the potential to emerge as a transformative capability across the biological sciences.

Artificial intelligence↗

DOE OSTI2024

(U) The Effect of ENDF/B-VIII.1 on Jezebel

The 239 Pu Jezebel critical assembly of 1954-55 was modeled in great detail and reevaluated using ENDF/B-VII.1 cross sections, and a new simplified (homogeneous one-dimensional spherical) benchmark was derived. The new simplified model was similar to that derived by Hansen and Paxton in 1969, but its radius was modified so that its k eff was the average of the C/E’s (where C is the calculated k eff and E is the inferred experimental k eff ) of the four detailed models. When ENDF/B-VIII.0 data were released, a mismatch between the detailed and simplified models was introduced due to changes in the nickel cross sections, because the detailed models contain nickel but the simplified model does not. A revision of the simplified model realigned the detailed and simplified models. ENDF/B-VIII.1 data have now been released. The simplified and detailed Jezebel models have been run using preliminary ENDF/B-VIII.1 ACE files. In Sec. II, it is shown that the detailed and simplified models are still aligned. No revision of the simplified model is called for. These results were computed using MCNP6.3.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2017

Lanthanides or Dust in Kilonovae: Lessons Learned from GW170817

The unprecedented optical and near-infrared lightcurves of the first electromagnetic counterpart to a gravitational-wave source, GW170817, a binary neutron star merger, exhibited a strong evolution from blue to near-infrared (a so-called “kilonova” or “macronova”). The emerging near-infrared component is widely attributed to the formation of r-process elements that provide the opacity to shift the blue light into the near-infrared. An alternative scenario is that the light from the blue component gets extinguished by dust formed by the kilonova and subsequently is re-emitted at near-infrared wavelengths. We test here this hypothesis using the lightcurves of AT 2017gfo, the kilonova accompanying GW170817. We find that of the order of 10{sup −5} M{sub ⊙} of carbon is required to reproduce the optical/near-infrared lightcurves as the kilonova fades. This putative dust cools from ∼2000 K at ∼4 days after the event to ∼1500 K over the course of the following week, thus requiring dust with a high condensation temperature, such as carbon. We contrast this with the nucleosynthetic yields predicted by a range of kilonova wind models. These suggest that at most 10{sup −9} M{sub ⊙} of carbon is formed. Moreover, the decay in the inferred dust temperature is slower than that expected in kilonova models. We therefore conclude that in current models of the blue component of the kilonova, the near-infrared component in the kilonova accompanying GW170817 is unlikely to be due to dust.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2020

The Langmuir probe system in the Wendelstein 7-X test divertor

Here, the design and evaluation of the Langmuir probe system used in the first divertor operation phase of Wendelstein 7-X is described. The probes are integrated into the target plates and have individually facetted surfaces to keep the angle of incidence of the magnetic field within an appropriate range for different magnetic configurations. Multiple models for the derivation of plasma parameters from current–voltage characteristics are introduced. These are analyzed with regard to their assumptions and limitations, generalized, and adapted to our use case. A detailed comparison is made to determine the most suitable model. It is found that the choice of model has a large impact, for example, resulting in a change in the inferred temperatures of up to a factor two. This evaluation is implemented in a Bayesian modeling framework and automated to allow for joint analysis with other diagnostics and a replacement of ad hoc assumptions. We rigorously treat parameter uncertainties, revealing strong correlations between them. General and flexible model formulations permit an expansion to additional effects.

47 OTHER INSTRUMENTATION↗

DOE OSTI2024

Accelerating data acquisition with FPGA-based edge machine learning: a case study with LCLS-II

New scientific experiments and instruments generate vast amounts of data that need to be transferred for storage or further processing, often overwhelming traditional systems. Edge machine learning (EdgeML) addresses this challenge by integrating machine learning (ML) algorithms with edge computing, enabling real-time data processing directly at the point of data generation. EdgeML is particularly beneficial for environments where immediate decisions are required, or where bandwidth and storage are limited. In this paper, we demonstrate a high-speed configurable ML model in a fully customizable EdgeML system using a field programmable gate array (FPGA). Our demonstration focuses on an angular array of electron spectrometers, referred to as the ‘CookieBox,’ developed for the Linac Coherent Light Source II project. The EdgeML system captures 51.2 Gbps from a 6.4 GS s −1 analog to digital converter and is designed to integrate data pre-processing and ML inside an FPGA. Our implementation achieves an inference latency of 0.2 µs for the ML model, and a total latency of 0.4 µs for the complete EdgeML system, which includes pre-processing, data transmission, digitization, and ML inference. The modular design of the system allows it to be adapted for other instrumentation applications requiring low-latency data processing.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

μ⋆ masses: weak-lensing calibration of the Dark Energy Survey Year 1 redMaPPer clusters using stellar masses

ABSTRACT We present the weak-lensing mass calibration of the stellar-mass-based μ⋆ mass proxy for redMaPPer galaxy clusters in the Dark Energy Survey Year 1. For the first time, we are able to perform a calibration of μ⋆ at high redshifts, z > 0.33. In a blinded analysis, we use ∼6000 clusters split into 12 subsets spanning the ranges 0.1 ≤ z < 0.65 and μ⋆ up to ${\sim} 5.5 \times 10^{13} \, \mathrm{M}_{\odot }$, and infer the average masses of these subsets through modelling of their stacked weak-lensing signal. In our model, we account for the following sources of systematic uncertainty: shear measurement and photometric redshift errors, miscentring, cluster-member contamination of the source sample, deviations from the Navarro–Frenk–White halo profile, halo triaxiality, and projection effects. We use the inferred masses to estimate the joint mass–μ⋆–z scaling relation given by $\langle M_{200c} | \mu _{\star },z \rangle = M_0 (\mu _{\star }/5.16\times 10^{12} \, \mathrm{M_{\odot }})^{F_{\mu _{\star }}} ((1+z)/1.35)^{G_z}$. We find $M_0= (1.14 \pm 0.07) \times 10^{14} \, \mathrm{M_{\odot }}$ with $F_{\mu _{\star }}= 0.76 \pm 0.06$ and Gz = −1.14 ± 0.37. We discuss the use of μ⋆ as a complementary mass proxy to the well-studied richness λ for: (i) exploring the regimes of low z, λ < 20 and high λ, z ∼ 1; and (ii) testing systematics such as projection effects for applications in cluster cosmology.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2024

Predictions of rhizosphere microbiome dynamics with a genome-informed and trait-based energy budget model

Abstract Soil microbiomes are highly diverse, and to improve their representation in biogeochemical models, microbial genome data can be leveraged to infer key functional traits. By integrating genome-inferred traits into a theory-based hierarchical framework, emergent behaviour arising from interactions of individual traits can be predicted. Here we combine theory-driven predictions of substrate uptake kinetics with a genome-informed trait-based dynamic energy budget model to predict emergent life-history traits and trade-offs in soil bacteria. When applied to a plant microbiome system, the model accurately predicted distinct substrate-acquisition strategies that aligned with observations, uncovering resource-dependent trade-offs between microbial growth rate and efficiency. For instance, inherently slower-growing microorganisms, favoured by organic acid exudation at later plant growth stages, exhibited enhanced carbon use efficiency (yield) without sacrificing growth rate (power). This insight has implications for retaining plant root-derived carbon in soils and highlights the power of data-driven, trait-based approaches for improving microbial representation in biogeochemical models.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2026

Bayesian Optimization of Catalysis with In-Context Learning

Large language models (LLMs) can perform accurate classification with zero or few examples through in-context learning (ICL), allowing the model to observe query-relevant examples at inference time and eliminating the need for additional weight updates to generalize beyond its original training data. We extend this capability to regression with uncertainty estimation using frozen LLMs (e.g., GPT-4o, Gemini), enabling Bayesian optimization (BO) in natural language without explicit model training or feature engineering. We apply this to materials discovery by representing materials as synthesis and testing procedures for use in natural language prompts. This Bayesian, design-first approach prioritizes optimization toward target material properties before detailed characterization, in contrast to conventional experimental workflows that often emphasize characterization of suboptimal materials. On benchmarks like aqueous solubility and oxidative coupling of methane (OCM), BO-ICL matches or outperforms Gaussian processes. In live experiments on the reverse water–gas shift (RWGS) reaction, BO-ICL identifies multimetallic catalysts that approach equilibrium CO yield within 6 and 10 iterations from a pool of 3,700 and 360,000 candidates, respectively. Our method redefines materials representation and accelerates discovery, with broad applications across catalysis, materials science, and AI.

Calibration↗

DOE OSTI2021

Bayesian analysis of a future β decay experiment's sensitivity to neutrino mass scale and ordering

Bayesian modeling techniques enable sensitivity analyses that incorporate detailed expectations regarding future experiments. A model-based approach also allows one to evaluate inferences and predicted outcomes, by calibrating (or measuring) the consequences incurred when certain results are reported. In this work, we present procedures for calibrating predictions of an experiment's sensitivity to both continuous and discrete parameters. Using these procedures and a new Bayesian model of the β-decay spectrum, we assess a high-precision β-decay experiment's sensitivity to the neutrino mass scale and ordering for one assumed design scenario. We find that such an experiment could measure the electron-weighted neutrino mass within ~40 meV after 1 year (90% credibility). Neutrino masses > 500 meV could be measured within ≈5 meV. Using only β decay and external reactor neutrino data, we find that next-generation β-decay experiments could potentially constrain the mass ordering using a two-neutrino spectral model analysis. By calibrating mass ordering results, we identify reporting criteria that can be tuned to suppress false ordering claims. In some cases, a two-neutrino analysis can reveal that the mass ordering is inverted, an unobtainable result for the traditional one-neutrino analysis approach.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2022

Prediction of DIII-D Pedestal Structure from Externally Controllable Parameters

The sharp increase of pressure at the edge of a high confinement mode (H-mode) plasma, the pedestal, strongly impacts overall plasma performance. Predicting the pedestal is a necessity to control and optimize tokamak operations. An experimental data-driven machine learning (ML) approach is presented that predicts the pedestal heights and widths of electron density (ne) and electron temperature (Te) profiles as well as the separatrix ne from externally controllable parameters such as the plasma shape, heating method and power, and gas puff rate and integrated gas puff. The OMFIT framework was used with DIII-D data to efficiently, robustly, and automatically build a database of pedestal parameters to train machine learning models. Database creation was enabled by the search engine tool for DIII-D data, TokSearch, which parallelizes data fetching, enabling fast searches through basic signals of thousands of DIII-D shots and selection of relevant time intervals. Principal Component Analysis (PCA) separated the database into three clusters that represent classes of plasma shapes that are regularly used in DIII-D. The most important parameters for setting the pedestal structure were plasma current (Ip), toroidal magnetic field (Bφ), neutral beam heating power (PNBI) and shaping quantities. The Deep Jointly Informed Neural Networks (DJINN) algorithm was applied to identify suitable neural network (NN) architectures that appropriately capture the features of the pedestal database. Separate NNs were implemented for each pedestal parameter, and ensembling methods were used to improve the prediction accuracy and allowed estimation of the prediction uncertainty. The pedestal predictions of the test dataset lie within the measurement uncertainties of the pedestal parameters. The NN outperformed simple Linear Regression (LR) analysis, indicating non-linear dependencies in the pedestal structure. The presented achievements illustrate a promising path for future research, using feature extraction to infer experimental trends and thereby improve pedestal models as well as deploying NN for a fast pedestal prediction in DIII-D scenario development.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗