SEARCH · Engineering Papers

Results for “variable selection”

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 163 records · Page 9

All sources NASA NTRS DOE OSTI

DOE OSTI2021

MACAW v1.0

The ability to embed molecules in a numeric space is essential in order to build mathematical and machine-learning models describing molecular properties or other processes affected by molecules. MACAW is a cheminformatic tool that allows embedding small molecules into a multidimensional numeric space. In the embedding, each molecule is assigned a numeric vector that captures information of the molecule in relation to other molecules, and that vector can be used as input to mathematical models. Molecules that are more similar to each other are embedded closer in this numeric space, whereas molecules that are more different are embedded further away. One advantage of the MACAW embedding technology compared to established alterantives is that it is fast and does not require extensive computational resources or expertise. In particular, MACAW embeddings can be used as input to mathematical models without the need for variable cleaning or feature selection, saving time and simplifying their use. On the other hand, MACAW also contains methods to generate new molecules and to recommend new molecules satisfying a desired molecular property. The generation of new molecules can be biased based on an input set of molecules, effectively generating molecular diversity around it. In its turn, MACAW's molecular recommendation tool is a novel method for evolving molecules in silico towards a desired molecular specification. In this method, the biased molecular generator tool is applied iteratively in combination with a molecular selection step. As a result, in each iteration the molecules selected by the software are increasingly closer to the desired specification. Both the molecular generation and the molecular recommendation tools are very fast, efficient, and intuitive to use.

Roger, VincentBlay↗

DOE OSTI2019

Selection of glasses in support of the Sludge Batch 9 variability study for coupled operation with the Salt Waste Processing Facility

Prior to the processing of Sludge Batch 9 (SB9) with Frit 803 at the Defense Waste Processing Facility (DWPF), assessments and experimental glass work were completed to support the qualification of SB9 coupled processing with streams from the Actinide Removal Process (ARP) and Modular Caustic Side Solvent Extraction Unit (MCU). This previous work did not address coupled processing with streams from the Salt Waste Processing Facility (SWPF) because the Product Composition Control System (PCCS) models limited the TiO 2 concentration in glass to 2 weight percent (wt.%). These high activity streams from SWPF include monosodium titanate (MST) and sludge solids from the Sludge Solids Receipt Tank as well as Cs-containing strip effluent. The incorporation of these SWPF streams is expected to reach TiO2 concentrations in glass greater than 2 wt.% based on an MST addition of 0.4 g MST/L of salt solution. If double strike operation is required, then the MST addition is increased to 0.8 g MST/L of salt solution. Therefore, revisions to the models contained in PCCS were completed to allow the evaluation of glasses containing greater than 2 wt.% TiO 2 , but less than 6 wt.%. To support the implementation of the revised PCCS for SB9 – Frit 803 processing at DWPF prior to SWPF startup, the applicability of the new durability models for the SB9 variability study glasses was verified. Since the initial SB9 qualification for ARP-MCU coupled operation, additional Tank 40 analyses have been completed, which served as the basis for an updated projection of the SB9 blend composition in Tank 40. To identify a frit optimized for SB9 coupled operation with SWPF, the Savannah River National Laboratory (SRNL) performed a paper-study by utilizing Measurement Acceptance Region (MAR) assessments with DWPF PCCS. Based on these MAR assessment results, Frit 625 was recommended for processing both single and double MST strike scenarios. Savannah River Remediation (SRR) requested that SRNL perform a variability study with both Frit 803 and Frit 625 for SB9 coupled operation with SWPF. The objective of the variability study is to demonstrate that the Product Consistency Test (PCT) responses of the glasses are acceptable relative to the chemical durability of the Environmental Assessment (EA) reference glass and predictable by the current DWPF PCCS models for durability. This report documents the development of a test matrix that will support the variability study. To provide operational flexibility prior to SWPF startup, the predictability of glasses within the glass region for Frit 625-SB9 coupled operation with ARP-MCU was also confirmed.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2024

Application of Partial Least Squares Approaches to Pyroprocessing ER Data

Multivariate approaches show promise for application to process monitoring for safeguards of pyroprocessing. Past MPACT work explored the application of Principal Component Analysis (PCA) to detect off-normal conditions in pyroprocessing electrorefiner (ER) data from in the Hot Fuel Examination Facility (HFEF) at Idaho National Laboratory (INL) known as the Scalable Pyrochemical Recycling testbed (SPyRe) ER. PCA, however, does not consider the output variables. In FY24, multivariate analysis was extended from PCA to Partial Least Squares (PLS) analysis. PLS maximizes the variance between both the input signals and output variables. In the case of this work, PLS was applied in two different manners: Predictive PLS and Discriminant PLS. Predictive PLS maximizes the covariance between the process variables of the ER and the measured U concentration from in-situ voltammetry. Discriminant PLS maximizes the covariance between the process variables and a set of training process “states” such as known off-normal conditions. By projecting into the latent variable space in PLS, the process variables can be regressed onto the outputs and predictions can be made for new data sets. In this work, by applying predictive PLS, a penalized non-linear PLS approach was able to make predictions of concentration based on test and training data and detect when operations were off-normal. However, the predictive PLS does not classify the signals to which off-normal operations are attributable. Discriminant PLS can be used to classify off-normal operations but is inadequate to properly classify specific off-normal classes like power supply faults when the Discriminant PLS model is only specifically trained to detect that off-normal class. When all faults are trained against the observation data, all three operational classes are accurately classified and distinguished. Thus, future application of latent variable techniques should not select any given method, but should use a mixture of PCA, Predictive PLS, and Discriminant PLS.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

DOE OSTI2018

On computing the evolution of temperature for materials under dynamic loading

Modeling and simulation of the dynamic response of materials is important to many applications including the development of armor systems, understanding the safety of explosives, and assessing the crashworthiness of vehicles. Within such applications it is often critical to accurately compute the evolution of the temperature because it is a state variable that affects the kinetics of competing active processes within the material (e.g., dislocation motion, phase transformation, decomposition). Depending on the selection of an independent state variable, e.g. temperature or entropy, the approach for computing temperature is well understood based on the thermodynamic framework attributed to Coleman and Noll. However, different computational codes used for modeling the dynamic response of materials adopt different independent state variables. In this work, two thermodynamically consistent strategies for computing the temperature of a coupled thermodynamic state are compared and implemented into two different Lagrangian computational codes. The equivalence of these two approaches is established through the numerical solutions of several test problems. Finally, the implication of various approximations made to each of these approaches within the literature are assessed in the context of uniaxial stress conditions (split Hopkinson pressure bar experiments) and uniaxial strain conditions (plate impact experiments). Here, it is shown that the temperature rate or energy partition approaches are equivalent when implemented in their complete forms, but that several common simplifying assumptions, that are warranted in the case of uniaxial stress, lead to significant errors in the resulting Hugoniot state for plate impact.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Representative Period Selection for Robust Capacity Expansion Planning in Low-carbon Grids

With the increasing urgency to decarbonize power systems, while mitigating extreme events, capacity expansion models can play a vital role in reliably planning the expansion of power systems and facilitating the integration of renewable energy sources. Optimizing capacity expansion generally involves selecting surrogate representative days from forecasts of load and the generation profiles of variable renewable energy resources. To properly select those representative days, we propose a novel input-based approach in combination with the k-means clustering algorithm that utilize three unique operational inputs: load shedding, renewable curtailment, and transmission congestion. The proposed method allows for more robust and cost-effective capacity planning. The method is validated using a capacity expansion model and a production cost model based on California Independent System Operator (CAISO)'s decarbonization goals, and results in reduced costs and drastically lower load shedding.

Anderson, Osten P.↗

DOE OSTI2020

NMDQi Thermal-hydraulic Optimization, Analysis, and Scoping Tool (TOAST)

The Thermal-hydraulic Optimization, Analysis, and Scoping Tool (TOAST) is developed to support the thermal-hydraulic aspects for the design of optimal irradiation vehicles for the Nuclear Materials Discovery and Qualification initiative (NMDQi). TOAST is currently a MATLAB-based tool that utilizes a simplified steady-state analytical model where a radial thermal resistance circuit is utilized to account for conduction through up to 2 capsules, a gas gap, a thermal bond, a sample cladding, and a sample, as well as the convection from a water coolant outside the capsules. The geometry is discretized axially along the user-defined height of a basket with a user-defined number of points/nodes, essentially turning TOAST into a semi-2D heat transfer analysis utility for cylindrical irradiation vehicles with practically any configuration, solving for temperatures radially at multiple axial locations. TOAST utilizes a Graphical User Interface (GUI) in-which a user can define the geometrical layout, the materials utilized for each component, the coolant characteristics, and the required solution. The user can define the geometry by inputting the diameters, thicknesses, and heights for each component, whereas the materials are defined via the user-provided constant or variable thermal conductivities. The user can select the coolant's inlet temperature, pressure, and the pressure drop across the height of the problem (which is used to calculate the velocity of the flow). Finally, the user can choose to solve for a sample heat generation rate limit by inputting temperature limits for the sample and capsules or can choose to purely solve for the axial temperature distributions of each component by inputting a pre-selected heat generation rate. Either way, TOAST provides the user with axial temperature distributions of all the components including the coolant, and the heat generation rate limit as well as the maximum outlet coolant temperature. The user can also choose to do one of 6 sensitivity analyses in TOAST. The sensitivity analyses yield plots of the sensitivity of the heat generation rate limit, the maximum coolant temperature, and the pressure limit for the annular components due to perturbations in 1-2 unknown variables based on the selected sensitivity analysis. Benchmarks between computations in TOAST and equivalent 2D axis symmetric ABAQUS models are presented, showing that TOAST results are within less than 3% of ABAQUS results in most cases, and a maximum of 8% difference in some cases. The benchmarks also revealed that this uncertainty is tied to the selection of an appropriate Nusselt number correlation and appropriate thermal conductivities, which the user can do from the GUI. Regardless, TOAST is demonstrated as a computationally efficient, highly accessible, and accurate utility for optimization and scoping s of studies different irradiation vehicles.

42 ENGINEERING↗

DOE OSTI2022

Variable Active Galactic Nuclei in the Galaxy Evolution Explorer Time Domain Survey

We searched the northern hemisphere fields of the Galaxy Evolution Explorer Time Domain Survey for galaxies with UV variability indicative of active galactic nuclei (AGNs). We identified 48 high-probability candidate AGNs from a parent sample of 1819 galaxies in the NASA Sloan Atlas catalog. We further characterized these systems using optical spectroscopic diagnostics, Wide-field Infrared Survey Explorer IR color selection criteria, and spectral energy distribution modeling. Of the 48 candidates, eight were identified as AGNs from optical emission lines, two were identified by their IR colors, and 28 were identified through spectral energy decomposition. Observational biases of each selection method are discussed in connecting these AGN subsamples to another. By selecting AGNs based on UV variability, we also identified six low-mass AGN candidates, all of which would have been missed by spectroscopic selection.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2024

Nine lensed quasars and quasar pairs discovered through spatially extended variability in Pan-STARRS

We present the proof of concept of a method for finding strongly lensed quasars using their spatially extended photometric variability through difference imaging in cadenced imaging survey data. We applied the method to Pan-STARRS, starting with an initial selection of 14 107Gaiamultiplets with quasar-like infrared colours from WISE. We identified 229 candidates showing notable spatially extended variability during the Pan-STARRS survey period. These include 20 known lenses and an additional 12 promising candidates for which we obtained long-slit spectroscopy follow-up. This process resulted in the confirmation of four doubly lensed quasars, four unclassified quasar pairs, and one projected quasar pair. Only three are pairs of stars or quasar+star projections. The false-positive rate accordingly is 25%. The lens separations are between 0.81″ and 1.24″, and the source redshifts lie betweenz = 1.47 andz = 2.46. Three of the unclassified quasar pairs are promising dual-quasar candidates with separations ranging from 6.6 to 9.3 kpc. We expect that this technique is a particularly efficient way to select lensed variables in the upcomingRubin-LSST, which will be crucial given the expected limitations for spectroscopic follow-up.

Astronomy & Astrophysics↗

DOE OSTI2019

Estimation of Tropical Cyclone Intensity in the North Atlantic and Northeastern Pacific Basins Using TRMM Satellite Passive Microwave Observations

Here, a statistical passive microwave intensity estimation (PMW-IE) algorithm for estimating the intensity of tropical cyclones (TCs) in the North Atlantic and northeastern and central Pacific basins is developed and tested. The algorithm is derived from Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) 85-GHz brightness temperatures and near-surface rain-rate retrievals to provide objective estimates of current maximum sustained surface winds (Vmax) and 6-h future Vmax of TCs. The full record of TRMM data (1998–2013) including 2326 TMI overpasses of 503 TCs is separated into dependent samples (1998–2010) for model development and independent samples (2011–13) for model verification. The best track intensities are used as dependent variables in a stepwise multiple-regression approach. Separately for each basin, three regression models are derived using selected 1) 85-GHz-only variables, 2) rain-rate-only variables, and 3) combined 85-GHz and rain variables. The algorithms are assessed using independent samples and those with contemporaneous aircraft-reconnaissance measurements. Rain-only and combined models perform better than the 85-GHz-only model. Lower errors are found for estimating the 6-h future Vmax than estimating the current Vmax using all three models. This implies that it is optimal to use passive-microwave-retrieved rain variables observed a few hours earlier to estimate TC intensity. The MAE (RMSE) of 6-h future Vmax is 9 (12) kt (1 kt ≈ 0.51 m s -1 ) when testing the combined models with ATL and EPA independent samples. Aircraft-reconnaissance-based independent samples yields a MAE of 9.6 kt and RMSE of 12.6 kt for estimating 6-h future Vmax.

58 GEOSCIENCES↗

DOE OSTI2022

Evaluating Grid Strength under Uncertain Renewable Generation

The increasing displacement of synchronous generators with renewable resources such as wind and solar via power electronic interfaces causes a reduction in short-circuit strength and weak grid issues. The variation and uncertainty of renewable energy increase challenges for identifying weak grid conditions. This paper proposes an efficient method to analyze the impact of uncertain renewable energy on grid strength. The proposed method uses the probabilistic collocation method (PCM) to approximate the results of grid strength assessment under uncertain renewable generation, in order to reduce computational burden without compromising result accuracy when compared with traditional Monte Carlo simulation (MCS). To improve the accuracy of the approximation results, the proposed method integrates the K-means clustering technique with PCM to select the approximation samples of input variables. The efficacy of the proposed method is demonstrated by comparison with MCS on the modified IEEE 9-bus system and modified IEEE 39-bus system with multiple renewable generators.

grid strength↗

DOE OSTI2021

Texture and microstructure evolution in thermomechanically processed Mg-Ca and Mg-Zn-Ca alloys

Mg-Zn-Ca alloys have the potential for producing Mg alloy sheets with weaker basal textures and therefore improved formability. Thermomechanical processing (TMP) also plays a substantial role in determining the final sheet texture. The interplay of these variables, TMP and alloying, complicates comparisons of alloys across the literature. This work systematically explores the texture evolution and recrystallization behavior in Mg-Ca and Mg-Zn-Ca alloys during plane strain compression (PSC) using a Gleeble thermomechanical simulator. It is demonstrated that the basal texture intensity and texture characteristics change significantly during post-deformation annealing, particularly in the ternary alloys. It is also shown that careful selection of the TMP processing variables used during PSC is essential to producing weak textures. In particular, it is important to limit the recrystallization that occurs between compressive hits. This is achieved by adjusting solute content, strain rate, and the duration of the soak between passes.

36 MATERIALS SCIENCE↗

DOE OSTI2022

CondiDiag1.0: a flexible online diagnostic tool for conditional sampling and budget analysis in the E3SM atmosphere model (EAM)

Abstract. Numerical models used in weather and climate prediction take into account a comprehensive set of atmospheric processes (i.e., phenomena) such as the resolved and unresolved fluid dynamics, radiative transfer, cloud and aerosol life cycles, and mass or energy exchanges with the Earth's surface. In order to identify model deficiencies and improve predictive skills, it is important to obtain process-level understanding of the interactions between different processes. Conditional sampling and budget analysis are powerful tools for process-oriented model evaluation, but they often require tedious ad hoc coding and large amounts of instantaneous model output, resulting in inefficient use of human and computing resources. This paper presents an online diagnostic tool that addresses this challenge by monitoring model variables in a generic manner as they evolve within the time integration cycle. The tool is convenient to use. It allows users to select sampling conditions and specify monitored variables at run time. Both the evolving values of the model variables and their increments caused by different atmospheric processes can be monitored and archived. Online calculation of vertical integrals is also supported. Multiple sampling conditions can be monitored in a single simulation in combination with unconditional sampling. The paper explains in detail the design and implementation of the tool in the Energy Exascale Earth System Model (E3SM) version 1. The usage is demonstrated through three examples: a global budget analysis of dust aerosol mass concentration, a composite analysis of sea salt emission and its dependency on surface wind speed, and a conditionally sampled relative humidity budget. The tool is expected to be easily portable to closely related atmospheric models that use the same or similar data structures and time integration methods.

58 GEOSCIENCES↗

DOE OSTI2021

Very Large Array imaging rules out precessing radio jets in three DES–SDSS-selected candidate periodic quasars

ABSTRACT Periodic quasars have been suggested as candidates for hosting binary supermassive black holes (SMBHs), although alternative scenarios remain possible to explain the optical light-curve periodicity. To test the alternative hypothesis of precessing radio jet, we present deep 6 GHz radio imaging conducted with NSF’s Karl G. Jansky Very Large Array (VLA) in its C configuration for the three candidate periodic quasars, DES J024703.24−010032.0, DES J024944.66−000036.8, and DES J025214.67−002813.7. Our targets were selected based on their optical variability using 20 yr long multicolour light curves from the Dark Energy Survey (DES) and the Sloan Digital Sky Survey (SDSS). The new VLA observations show that all three periodic quasars are radio-quiet with the radio loudness parameters measured to be $R\equiv f_{6\, {\rm cm}}/f_{{\rm 2500}}$ of ≲1.0–1.5 and the k-corrected luminosities νLν[6 GHz] of ≲5–21 × 1039 erg s−1. They are in stark contrast to previously known periodic quasars proposed as binary SMBH candidates such as the blazar OJ 287 and PG1302−102. Our results rule out optical emission contributed from precessing radio jets as the origin of the optical periodicity in the three DES–SDSS-selected candidate periodic quasars. Future continued optical monitoring and complementary multiwavelength observations are still needed to further test the binary SMBH hypothesis as well as other competing scenarios to explain the optical periodicity.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2019

The effect of dust bias on the census of neutral gas and metals in the high-redshift Universe due to SDSS-II quasar colour selection

ABSTRACT We present a systematic study of the impact of a dust bias on samples of damped Ly α absorbers (DLAs). This bias arises as an effect of the magnitude and colour criteria utilized in the Sloan Digital Sky Survey (SDSS) quasar target selection up until data release 7 (DR7). The bias has previously been quantified assuming only a contribution from the dust obscuration. In this work, we apply the full set of magnitude and colour criteria used up until SDSS-DR7 in order to quantify the full impact of dust biasing against dusty and metal-rich DLAs. We apply the quasar target selection algorithm on a modelled population of intrinsic colours, and by exploring the parameter space consisting of redshift, ($z_{\rm{\small QSO}}$and zabs), optical extinction, and H i column density, we demonstrate how the selection probability depends on these variables. We quantify the dust bias on the following properties derived for DLAs at z ≈ 3: the incidence rate, the mass density of neutral hydrogen and metals, and the average metallicity. We find that all quantities are significantly affected. When considering all uncertainties, the mass density of neutral hydrogen is underestimated by 10–50 per cent, and the mass density in metals is underestimated by 30–200 per cent. Lastly, we find that the bias depends on redshift. At redshift z = 2.2, the mass density of neutral hydrogen and metals might be underestimated by up to a factor of 2 and 5, respectively. Characterizing such a bias is crucial in order to accurately interpret and model the properties and metallicity evolution of absorption-selected galaxies.

Krogager, Jens-Kristian↗

DOE OSTI2019

Atom Probe Tomography Interlaboratory Study on Clustering Analysis in Experimental Data Using the Maximum Separation Distance Approach

We summarize the findings from an interlaboratory study conducted between ten international research groups and investigate the use of the commonly used maximum separation distance and local concentration thresholding methods for solute clustering quantification. The study objectives are: to bring clarity to the range of applicability of the methods; identify existing and/or needed modifications; and interpretation of past published data. Participants collected experimental data from a proton-irradiated 304 stainless steel and analyzed Cu-rich and Ni–Si rich clusters. The datasets were also analyzed by one researcher to clarify variability originating from different operators. The Cu distribution fulfills the ideal requirements of the maximum separation method (MSM), namely a dilute matrix Cu concentration and concentrated Cu clusters. This enabled a relatively tight distribution of the cluster number density among the participants. By contrast, the group analysis of the Ni–Si rich clusters by the MSM was complicated by a high Ni matrix concentration and by the presence of Si-decorated dislocations, leading to larger variability among researchers. While local concentration filtering could, in principle, tighten the results, the cluster identification step inevitably maintained a high scatter. Recommendations regarding reporting, selection of analysis method, and expected variability when interpreting published data are discussed.

APT, Round Robin Studies, atom probe tomography, c↗

DOE OSTI2023

Prioritizing the selection of CMIP6 model ensemble members for downscaling projections of CONUS temperature and precipitation

Given the mismatch between the large volume of data archived for the sixth phase of the Coupled Model Intercomparison Project (CMIP6) and limited personnel and computational resources for downscaling, only a small fraction of the CMIP6 archive can be downscaled. In this work, we develop an approach to robustly sample projected hydroclimate states in CMIP6 for downscaling to test whether the selection of a single initial condition (IC) ensemble member from each CMIP6 model is sufficient to span the range of modeled hydroclimate over the conterminous United States (CONUS) and CONUS sub-regions. We calculate the pattern-centered root mean square difference of IC ensemble member anomalies relative to each model’s historical climatology for shared socioeconomic pathway (SSP) projections over 30-year time periods and compare the ratio of inter-model to intra-model variability for this metric. Regardless of SSP, inter-model variability is generally much greater than intra-model variability at the scales of the CONUS as a whole, as well as for most CONUS sub-regions. However, for some variables and scenarios, inter- and intra-model variability are similar at sub-CONUS scales, indicating that selecting a single IC ensemble member per model may be sufficient to sample the range of projected hydroclimate states in the 21st Century across CONUS, but for specific regions and variables, more careful selection of ensemble members may be necessary. Regionally resolved Taylor diagrams identify where more IC ensemble member downscaling efforts should be focused if resources are available to do so. Our results suggest that, with parsimonious sampling, the requisite computational expense of downscaling temperature and precipitation fields over the CONUS for subsequent CMIP activities may increase only marginally despite the great increase in data volumes with each successive CMIP phase.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2018

Truly horizontal finite difference scheme in the WRF model FY18 Q1 Update

A main goal of the Wind Forecasting Improvement Project 2 (WFIP2) is to improve performance of the Weather Research and Forecasting (WRF) model over complex terrain. Two methods for improving model performance were outlined in the LLNL WFIP 2 annual operating plan. The first is modification to WRF’s horizontal finite difference scheme to utilize an improved numerical stencil. A second proposed model improvement is use of the immersed boundary method within the WRF model to facilitate improved simulations over complex terrain. In Q1, work focused on implementing an alternative finite difference scheme, with a deliverable of a project update. The alternative finite difference method was implemented for the temperature variable and shown to improve our selected test case, and recommendations are made to continue the work in Q2. In Q2 the focus should be on extending the method to additional model variables, mainly pressure, as well as to real test cases, such as the WFIP 2 cold pool test case. Additionally, use of the immersed boundary method will be pursed in Q2 and Q3, with a second project update due at the end of Q3.

17 WIND ENERGY↗

DOE OSTI2022

Controlling extrapolations of nuclear properties with feature selection

Predictions of nuclear properties far from measured data are inherently inaccurate because of uncertainties in our knowledge of nuclear forces and in our treatment of quantum many-body effects in strongly-interacting systems. While the model bias can be directly calculated when experimental data is available, only an estimate can be made in the absence of such measurements. Current approaches to compute the estimated bias quickly lose predictive power when their input variables are taken far from the training region, resulting in uncontrolled uncertainties in applications such as nucleosynthesis simulations. In this letter, we present a novel technique to identify the input variables of machine learning algorithms that can provide robust estimates of model bias. Our process is based on selecting input variables, or features, based on their probability distribution functions across the entire nuclear chart. We illustrate our approach on the problem of quantifying the model bias in nuclear binding energies calculated with Density Functional Theory (DFT). We prove that building model biases with only Z and N as features leads to highly unreliable extrapolations. Conversely, we show that proper feature selection can systematically improve theoretical predictions without increasing uncertainties.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗