SEARCH · Search NASA

Results for “ENSEMBLE MODELING”

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 127 records · Page 7

All sources NASA NTRS DOE OSTI

NASA NTRS2011

Using Multi-Scale Modeling Systems and Satellite Data to Study the Precipitation Processes

In recent years, exponentially increasing computer power has extended Cloud Resolving Model (CRM) integrations from hours to months, the number of computational grid points from less than a thousand to close to ten million. Three-dimensional models are now more prevalent. Much attention is devoted to precipitating cloud systems where the crucial 1-km scales are resolved in horizontal domains as large as 10,000 km in two-dimensions, and 1,000 x 1,000 km2 in three-dimensions. Cloud resolving models now provide statistical information useful for developing more realistic physically based parameterizations for climate models and numerical weather prediction models. It is also expected that NWP and mesoscale model can be run in grid size similar to cloud resolving model through nesting technique. Recently, a multi-scale modeling system with unified physics was developed at NASA Goddard. It consists of (l) a cloud-resolving model (Goddard Cumulus Ensemble model, GCE model), (2) a regional scale model (a NASA unified weather research and forecast, WRF), (3) a coupled CRM and global model (Goddard Multi-scale Modeling Framework, MMF), and (4) a land modeling system. The same microphysical processes, long and short wave radiative transfer and land processes and the explicit cloud-radiation, and cloud-land surface interactive processes are applied in this multi-scale modeling system. This modeling system has been coupled with a multi-satellite simulator to use NASA high-resolution satellite data to identify the strengths and weaknesses of cloud and precipitation processes simulated by the model. In this talk, the recent developments and applications of the multi-scale modeling system will be presented. In particular, the results from using multi-scale modeling system to study the precipitating systems and hurricanes/typhoons will be presented. The high-resolution spatial and temporal visualization will be utilized to show the evolution of precipitation processes. Also how to use of the multi-satellite simulator tqimproy~ precipitation processes will be discussed.

Tao, Wei-Kuo↗

NASA NTRS2011

Microphysics in the Multi-Scale Modeling Systems with Unified Physics

In recent years, exponentially increasing computer power has extended Cloud Resolving Model (CRM) integrations from hours to months, the number of computational grid points from less than a thousand to close to ten million. Three-dimensional models are now more prevalent. Much attention is devoted to precipitating cloud systems where the crucial 1-km scales are resolved in horizontal domains as large as 10,000 km in two-dimensions, and 1,000 x 1,000 km2 in three-dimensions. Cloud resolving models now provide statistical information useful for developing more realistic physically based parameterizations for climate models and numerical weather prediction models. It is also expected that NWP and mesoscale model can be run in grid size similar to cloud resolving model through nesting technique. Recently, a multi-scale modeling system with unified physics was developed at NASA Goddard. It consists of (l) a cloud-resolving model (Goddard Cumulus Ensemble model, GCE model), (2) a regional scale model (a NASA unified weather research and forecast, WRF), (3) a coupled CRM and global model (Goddard Multi-scale Modeling Framework, MMF), and (4) a land modeling system. The same microphysical processes, long and short wave radiative transfer and land processes and the explicit cloud-radiation, and cloud-surface interactive processes are applied in this multi-scale modeling system. This modeling system has been coupled with a multi-satellite simulator to use NASA high-resolution satellite data to identify the strengths and weaknesses of cloud and precipitation processes simulated by the model. In this talk, the microphysics developments of the multi-scale modeling system will be presented. In particular, the results from using multi-scale modeling system to study the heavy precipitation processes will be presented.

Tao, Wei-Kuo↗

NASA NTRS2012

Microphysics in Multi-scale Modeling System with Unified Physics

Recently, a multi-scale modeling system with unified physics was developed at NASA Goddard. It consists of (1) a cloud-resolving model (Goddard Cumulus Ensemble model, GCE model), (2) a regional scale model (a NASA unified weather research and forecast, WRF), (3) a coupled CRM and global model (Goddard Multi-scale Modeling Framework, MMF), and (4) a land modeling system. The same microphysical processes, long and short wave radiative transfer and land processes and the explicit cloud-radiation, and cloud-land surface interactive processes are applied in this multi-scale modeling system. This modeling system has been coupled with a multi-satellite simulator to use NASA high-resolution satellite data to identify the strengths and weaknesses of cloud and precipitation processes simulated by the model. In this talk, a review of developments and applications of the multi-scale modeling system will be presented. In particular, the microphysics development and its performance for the multi-scale modeling system will be presented.

Tao, Wei-Kuo↗

NASA NTRS2011

Geospace Environment Modeling 2008-2009 Challenge: Ground Magnetic Field Perturbations

Acquiring quantitative metrics!based knowledge about the performance of various space physics modeling approaches is central for the space weather community. Quantification of the performance helps the users of the modeling products to better understand the capabilities of the models and to choose the approach that best suits their specific needs. Further, metrics!based analyses are important for addressing the differences between various modeling approaches and for measuring and guiding the progress in the field. In this paper, the metrics!based results of the ground magnetic field perturbation part of the Geospace Environment Modeling 2008 2009 Challenge are reported. Predictions made by 14 different models, including an ensemble model, are compared to geomagnetic observatory recordings from 12 different northern hemispheric locations. Five different metrics are used to quantify the model performances for four storm events. It is shown that the ranking of the models is strongly dependent on the type of metric used to evaluate the model performance. None of the models rank near or at the top systematically for all used metrics. Consequently, one cannot pick the absolute winner : the choice for the best model depends on the characteristics of the signal one is interested in. Model performances vary also from event to event. This is particularly clear for root!mean!square difference and utility metric!based analyses. Further, analyses indicate that for some of the models, increasing the global magnetohydrodynamic model spatial resolution and the inclusion of the ring current dynamics improve the models capability to generate more realistic ground magnetic field fluctuations.

Pulkkinen, A.↗

NASA NTRS2014

Operational Dust Prediction

Over the last few years, numerical prediction of dust aerosol concentration has become prominent at several research and operational weather centres due to growing interest from diverse stakeholders, such as solar energy plant managers, health professionals, aviation and military authorities and policymakers. Dust prediction in numerical weather prediction-type models faces a number of challenges owing to the complexity of the system. At the centre of the problem is the vast range of scales required to fully account for all of the physical processes related to dust. Another limiting factor is the paucity of suitable dust observations available for model, evaluation and assimilation. This chapter discusses in detail numerical prediction of dust with examples from systems that are currently providing dust forecasts in near real-time or are part of international efforts to establish daily provision of dust forecasts based on multi-model ensembles. The various models are introduced and described along with an overview on the importance of dust prediction activities and a historical perspective. Assimilation and evaluation aspects in dust prediction are also discussed.

assimilation↗

NASA NTRS2020

Global Carbon Cycle and Climate Feedbacks in the NASA GISS ModelE2.1

We present results from the NASA GISS ModelE2.1‐G‐CC Earth System Model with coupled climate‐carbon cycle simulations that were submitted to the sixth phase of the Coupled Model Intercomparison Project (CMIP6) Coupled Climate‐Carbon Cycle MIP (C4MIP). Atmospheric CO2 concentration and carbon budgets for the land and ocean in the historical simulations were generally consistent with observations. Low simulated atmospheric CO2 concentrations during 1850‐1950 were due to excess uptake from prescribed land cover change, which erroneously replaced arid shrublands with higher biomass crops, and assumed high 2004 LAI values in vegetated lands throughout the historical simulation. At the end of the historical period, slightly higher simulated CO2 than observed resulted from the land being an insufficient net carbon sink, despite the net effect of CO2 fertilization and warming‐induced increases to leaf photosynthetic capacity. The global ocean carbon uptake agreed well with the observations with the largest discrepancies in the low latitudes. Future climate projection at 2091‐2100 agreed with CMIP5 models in the northward shift, of temperate deciduous forest climate and expansion across Eurasia along 60 °N latitude, and dramatic regional biome shifts from drying and warming in continental Europe. Carbon feedback parameters were largely similar to the CMIP5 model ensemble. For our model, the variation of land feedback parameters within the uncertainty arises from the fertilization feedback being less sensitive due to lack of increased vegetation growth, and the comparably more negative ocean carbon‐climate feedback is due to the large slowdown of the Atlantic overturning circulation.

Carbon cycle↗

NASA NTRS2021

Future Climate Change Significantly Alters Interannual Wheat Yield Variability Over Half of Harvested Areas

Climate change affects the spatial and temporal distribution of crop yields, which can critically impair food security across scales. A number of previous studies have assessed the impact of climate change on mean crop yield and future food availability, but much less is known about potential future changes in interannual yield variability. Here, we evaluate future changes in relative interannual global wheat yield variability (the coefficient of variation (CV)) at 0.25° spatial resolution for two representative concentration pathways (RCP4.5 and RCP8.5). A multi-model ensemble of crop model emulators based on global process-based models is used to evaluate responses to changes in temperature, precipitation, and CO2. The results indicate that over 60% of harvested areas could experience significant changes in interannual yield variability under a high-emission scenario by the end of the 21st century (2066–2095). About 31% and 44% of harvested areas are projected to undergo significant reductions of relative yield variability under RCP4.5 and RCP8.5, respectively. In turn, wheat yield is projected to become more unstable across 23% (RCP4.5) and 18% (RCP8.5) of global harvested areas—mostly in hot or low fertilizer input regions, including some of the major breadbasket countries. The major driver of increasing yield CV change is the increase in yield standard deviation, whereas declining yield CV is mostly caused by stronger increases in mean yield than in the standard deviation. Changes in temperature are the dominant cause of change in wheat yield CVs, having a greater influence than changes in precipitation in 53% and 72% of global harvested areas by the end of the century under RCP4.5 and RCP8.5, respectively. This research highlights the potential challenges posed by increased yield variability and the need for tailored regional adaptation strategies.

Climate change↗

NASA NTRSDate not supplied

Science Workflows using Kamodo

Kamodo is a powerful python software package based on data functionalization. Once a given data set is functionalized, a large variety of capabilities are easily accessible in Kamodo, including unit conversions, custom analysis via function composition, interactive publication quality visualizations, and LaTeX encoding. The entirety of capabilities available in Kamodo are easily applied to both simulated and observed data across the multiple domains of Heliophysics and even in other disciplines. This work includes a variety of science workflows using Kamodo in combination with other resources, including with other python software packages, that expand the utility of Kamodo even further. These workflows include model-data comparisons, ensemble modeling examples, satellite mission planning examples, and other applications, all of which are freely available on CCMC’s Kamodo Github page for the community to adapt to their own uses (https://github.com/nasa/Kamodo). We invite the community to use these workflows and to contribute their own to share.

software↗

NASA NTRSDate not supplied

Science Workflows using Kamodo

python↗

DOE OSTI2023

Lithium-Ion Battery Diagnostics Using Electrochemical Impedance via Machine-Learning

Diagnosing battery states such as health, state-of-charge, or temperature is crucial for ensuring the safety and reliability of electrochemical energy storage systems. While some states, such as temperature, may be measured using cheap sensors, accurate diagnosis of battery health metrics usually requires time-consuming performance measurements, making them infeasible for use in real-world operation. These health metrics can be measured during lab-testing and then estimated on-line using predictive life models or via state observer algorithms such as Kalman filters, but these predictive methods should be supplemented by actual measurement of battery health whenever possible to ensure reliability. Rapid measurement of battery health may be done by various types of fast diagnostic techniques such as electrochemical impedance spectroscopy (EIS), which can be performed in only a few minutes and require only a fraction of the energy and power needed for a full charge and discharge measurement. But there is a substantial challenge for estimating battery health using EIS data, as EIS is sensitive to cell temperature, state-of-charge, current, and resting time in addition to health. Thus, utilizing EIS data to predict battery capacity requires correcting for all these additional variables, a task that is extremely difficult to handle analytically. This talk utilizes machine-learning methods to estimate the effectiveness of battery capacity prediction from EIS data, leveraging a data set of hundreds of EIS measurements recorded at varying temperature and state-of-charge throughout a 500-day aging study of 32 commercial, large-format NMC-Graphite lithium-ion batteries. Using EIS as input to machine-learning models is complicated by the nonlinear response of impedance to battery health, temperature, and state-of-charge, as well as the collinearity between the impedance response at neighboring frequencies, which can easily lead to overfit models. To train robust models, features from EIS data need to be extracted from the data or some subset of critical frequencies selected. Many approaches for extracting and selecting features from EIS data from electrochemical analysis and machine-learning fields were identified for analysis: using the entire raw spectra; selection of one, two, or many frequencies from the entire spectra; selecting interesting points from the EIS measurement using domain knowledge; fitting EIS with an equivalent-circuit model; calculating statistics on the raw impedance values; and reducing the dimensionality of the data using unsupervised linear (principal component analysis) and non-linear (uniform manifold approximation and projection) methods. These approaches were rigorously compared using a machine-learning pipeline approach, training linear, Gaussian process, and random forest regression models and quantifying performance using cross-validation as well as a held-out test set. An artificial neural network model trained on the raw spectra was also tested. Promising pipelines were fine-tuned via Bayesian hyperparameter optimization using cross-validation loss and training with class-specific weights to counter data set imbalance. The most reliable method for utilizing impedance in this work was the selection of two optimal frequencies through an exhaustive search, resulting in about 2% mean absolute error on test data for both Gaussian process and random forest model architectures. Interrogation of a variety of models reveals critical frequencies of 100 Hz and 103 Hz for this data set, though the optimal set of frequencies is not necessarily intuitive, i.e., the best performing models are not simply those that use impedance at frequencies that have the highest correlation to the relative discharge capacity. The best performing model is an ensemble model, which is able to predict battery capacity with 1.9% mean absolute error for unseen cells using impedance recorded at a variety of temperatures and states-of-charge.

battery↗

NASA NTRS2013

Analyzing Tropical Waves Using the Parallel Ensemble Empirical Model Decomposition Method: Preliminary Results from Hurricane Sandy

In this study, we discuss the performance of the parallel ensemble empirical mode decomposition (EMD) in the analysis of tropical waves that are associated with tropical cyclone (TC) formation. To efficiently analyze high-resolution, global, multiple-dimensional data sets, we first implement multilevel parallelism into the ensemble EMD (EEMD) and obtain a parallel speedup of 720 using 200 eight-core processors. We then apply the parallel EEMD (PEEMD) to extract the intrinsic mode functions (IMFs) from preselected data sets that represent (1) idealized tropical waves and (2) large-scale environmental flows associated with Hurricane Sandy (2012). Results indicate that the PEEMD is efficient and effective in revealing the major wave characteristics of the data, such as wavelengths and periods, by sifting out the dominant (wave) components. This approach has a potential for hurricane climate study by examining the statistical relationship between tropical waves and TC formation.

PEEMD↗

DOE OSTI2022

An adaptive knowledge-based data-driven approach for turbulence modeling using ensemble learning technique under complex flow configuration: 3D PWR sub-channel with DNS data

This work describes a new approach to increase the accuracy of Reynolds-averaged Navier–Stokes (RANS) in modeling turbulence flow leveraging the machine learning technique. Traditionally, different turbulence models for Reynolds stress are developed for different flow patterns based on human knowledge. Each turbulence model has a certain application domain and prediction uncertainty. In recent years, with the rapid improvements of machine learning techniques, researchers start to develop an approach to compensate for the prediction discrepancy of traditional turbulence models with statistical models and data. However, the approach has deficiencies in several aspects. For example, the amount of human knowledge introduced to the statistical model couldn’t be controlled, which makes the statistical model learn from a very naïve stage and limits its application. In this work, a new approach is developed to address those deficiencies. Here, the new approach uses the “ensemble learning” technique to control the amount of human knowledge introduced into the statistical model. Therefore, the new approach could be adaptive to the multiple application domains. In conclusion, according to the results of case study, the new approach shows higher accuracy than both traditional turbulence models and the previous machine learning approach.

42 ENGINEERING↗

NASA NTRS2019

Estimating Alpine Snow Depth by Combining Multi-Frequency Passive Radiance Observations with Ensemble Snowpack Modeling

This paper presents a physically-based snow depth retrieval algorithm adapted for deep mountainous snowpack and airborne multifrequency (10.7, 18.7, 37.0 and 89.0 GHz) passive microwave (PM) radiance observations from a single flight. The algorithm employs a single forecast-analysis cycle of a traditional sequential assimilation scheme. It uses an ensemble of multi-layer snowpack model runs to resolve snow microstructure and melt-refreeze crusts, andmicrowave radiative transfer models to relate snow properties to microwave measurements.Snow depth was retrieved at a 120 m spatial resolution over three 1 km2 Intensive Study Area (ISA) within the Rabbit Ears Meso-Cell Study Area (MSA) from the NASA Cold Land Processes Experiment (CLPX) in Colorado (United States) for one date in February 2003. When evaluated against in situ observations, root mean square error (RMSE) of the snow depth from the assimilation was 13.3 cm for areas with low (<5 %) forest cover, which was a reduction of 48 % in the RMSE compared with the modeled snow depth when the PM observations were not assimilated; indicating a ~5% relative error of the posterior snow depth with respect to the average snow depth (200 cm) measured at these pixels. For pixels with forest cover ranging from 5-15% and 15-30%, results were improved (R2 increased from 0.65 to 0.71 and from 0 to 0.38, respectively) by introducing a forest radiative transfer model during the assimilation.

Airborne Multifrequency↗

NASA NTRS2015

Rising Temperatures Reduce Global Wheat Production

Crop models are essential tools for assessing the threat of climate change to local and global food production. Present models used to predict wheat grain yield are highly uncertain when simulating how crops respond to temperature. Here we systematically tested 30 different wheat crop models of the Agricultural Model Intercomparison and Improvement Project against field experiments in which growing season mean temperatures ranged from 15 degrees C to 32◦ degrees C, including experiments with artificial heating. Many models simulated yields well, but were less accurate at higher temperatures. The model ensemble median was consistently more accurate in simulating the crop temperature response than any single model, regardless of the input information used. Extrapolating the model ensemble temperature response indicates that warming is already slowing yield gains at a majority of wheat-growing locations. Global wheat production is estimated to fall by 6% for each degree C of further temperature increase and become more variable over space and time.

agriculture↗

NASA NTRSDate not supplied

An Ensemble Investigation of the Causes for Regional Air-Quality Model Critical Load Exceedances Prediction Variability in European and North American Domains Using Diagnostics From Phase 4 of the Air Quality Model Evaluation International Initiative

We summarize tentative findings from multi air quality model ensembles for the years 2009 and 2010 in Europe (EU), and 2010 and 2016 in North America (NA), under AQMEII-4. The model predictions of sulphur and nitrogen deposition were used to estimate exceedances of critical loads for acidification and eutrophication, to show the extent to which the ensemble members agree in the magnitude and the trend of ecologically meaningful impacts. Model exceedance variability was analyzed using AQMEII-4 diagnostics. Evaluation against concentration and wet deposition observations, coupled with these diagnostics, identified specific process representations as the causes for variability between model predictions and for reduced model performance. All models predicted reductions in ecosystem acidification impacts in North America between the years 2010 and 2016, in accord with SO2 emissions reduction legislation which started in 2010 (SO2 SIP) However, all models in EU and NA domains had net negative biases for wet deposition of sulphur and nitrogen relative to observations. The wet S deposition average mean bias for the NA ensemble was -0.17 eq ha-1 d-1, and for the EU ensemble -1.15 eq ha-1 d-1. The NA daily wet deposition average mean bias for NH4+ was -0.37 eq ha-1d-1; EU -1.19 eq ha-1 d-1. The daily NA wet NO3- deposition average mean bias was -0.24 eq ha-1d-1; EU -0.69 eq ha-1 d-1. The members of the ensemble diverged (factor of 10) in their North American predictions for Ndep and consequently their eutrophication exceedances. The models with the highest eutrophication predictions also predicted the highest levels of gas-phase ammonia dry deposition (standard deviation of ammonia dry deposition flux across ensemble members was larger than the ensemble average). These models also had negative biases of predicted ammonia concentrations; average mean biases of -0.63 (satellite NH3) and -0.85 ppbv (surface NH3) compared to ensemble averages of -0.30 and -0.34 ppbv. Diagnostics showed that these differences resulted from the manner in which bidirectional ammonia fluxes were parameterized within these models. The second largest source of NA eutrophication prediction variability were models with positive biases in particulate ammonium and nitrate concentrations, and higher particle nitrogen deposition levels ( particle ammonium concentration bias +0.35 ug m-3; ensemble bias +0.15 ug m-3). We believe two factors may have led to these latter overestimates: higher levels of fine mode particle nitrate formation compared to other models (due to the use of an inorganic heterogeneous chemistry algorithm which did not take base cation chemistry into account), and updates to particle dry deposition velocities carried out in the absence of concurrent updates to wet scavenging algorithms. The relative importance of dry gas, dry particulate, and wet deposition towards total sulphur and nitrogen deposition totals differed between EU and North American domains, though all models had negative biases in wet deposition as noted above. Parallel and subsequent work suggests that multiphase hydrometeor scavenging may improve model wet deposition performance. An increased research focus is recommended for four model processes: multiphase hydrometeor scavenging, ammonia bidirectional fluxes, base cation chemistry and emissions, and particle dry deposition.

regional air-quality model↗

DOE OSTI2020

Precipitation Probability and Its Future Changes From a Global Cloud-Resolving Model and CMIP6 Simulations

This study investigates precipitation probability and its future changes using simulations from the Non-hydrostatic Icosahedral Atmospheric Model (NICAM), a global cloud-resolving model, and Coupled Model Intercomparison Project Phase 6 (CMIP6) models. For the precipitation probability comparison between models and Tropical Rainfall Measuring Mission (TRMM) during 1998–2008, the difference is greatly reduced by regridding to a common coarser resolution, but the difference signs largely remain. Both NICAM and CMIP6 model ensemble mean show less global-mean nonprecipitation fractions than observations, but NICAM agrees better with observations. For global-mean light precipitation probability, NICAM (CMIP6 model ensemble mean) is smaller (larger) than TRMM. NICAM shows higher heavy precipitation probability than TRMM in native resolutions, but it is close to TRMM and falls in the CMIP6 model range after regridding. For the precipitation probability changes from 1979–2008 to 2075–2104, NICAM predicts an increase in nonprecipitation and extreme heavy (>100 mm/day) precipitation probability but a slight decrease in the probability for precipitation <100 mm/day. After regridding to 2.5°, the precipitation intensity where the probability starts to increase becomes ~20 mm/day and the relative increases in nonprecipitation and heavy precipitation probability become much larger. Future heavy precipitation probability relative increases from CMIP6 models are much larger than NICAM at their native resolutions, but NICAM falls in the CMIP6 model range after regridding, indicating robust future increase (decrease) in heavy (light) precipitation probability regardless of underlying convection treatments. Discrepancy, however, exists in the projected nonprecipitation fraction changes: NICAM shows that nonprecipitation probability will increase over most of low and middle latitudes where CMIP6 models show less increases and even decreases.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS2014

Using Ensemble Short-Term Initialized Coupled NASA GEOS5 Climate Model Integrations to Study Convective Bias Growth

The representation of convective processes, particularly deep convection in the tropics, remains a persistent problem in climate models. In fact structural biases in the distribution of tropical rainfall in the CMIP5 models is hardly different than that of the CMIP3 versions. Given that regional climate change at higher latitudes is sensitive to the configuration of tropical forcing, this persistent bias is a major issue for the credibility of climate change projections. In this study we use model output from integrations of the NASA Global Earth Observing System Five (GEOS5) climate modeling system to study the evolution of biases in the location and intensity of convective processes. We take advantage of a series of hindcast experiments done in support of the US North American Multi-Model Ensemble (NMME) initiative. For these experiments a nine-month forecast using a coupled model configuration is made approximately every five days over the past 30 years. Each forecast is started with an updated analysis of the ocean, atmosphere and land states. For a given calendar month we have approximately 180 forecasts with daily means of various quantities. These forecasts can be averaged to essentially remove "weather scales" and highlight systematic errors as they evolve. Our primary question is to ask how the spatial structure of daily mean precipitation over the tropics evolves from the initial state and what physical processes are involved. Errors in parameterized convection, various water and energy fluxes and the divergent circulation are found to set up on fast time scales (order five days) compared to errors in the ocean, although SST changes can be non-negligible over that time. For the month of June the difference between forecast day five versus day zero precipitation looks quite similar to the difference between the June precipitation climatology and that from the Global Precipitation Climatology Project (GPCP). We focus much of our analysis on the influence of SST gradients, associated PBL baroclinicity enabled by turbulent mixing, the ensuing PBL moisture convergence, and how changes in these processes relate to convective precipitation bias growth over this short period.

Cohen, Charlie↗

DOE OSTI2023

Single-Point Extrapolation to the Complete Basis Set Limit through Deep Learning

Machine learning (ML) offers an attractive method for making predictions about molecular systems while circumventing the need to run expensive electronic structure calculations. Once trained on ab initio data, the promise of ML is to deliver accurate predictions of molecular properties that were previously computationally infeasible. In this work, we develop and train a graph neural network model to correct the basis set incompleteness error (BSIE) between a small and large basis set at the RHF and B3LYP levels of theory. Our results show that, when compared to fitting to the total potential, an ML model fitted to correct the BSIE is better at generalizing to systems not seen during training. We test this ability by training on single molecules while evaluating on molecular complexes. We also show that ensemble models yield better behaved potentials in situations where the training data is insufficient. However, even when only fitting to the BSIE, acceptable performance is only achieved when the training data sufficiently resemble the systems one wants to make predictions on. As a result, the test error of the final model trained to predict the difference between the cc-pVDZ and cc-pV5Z potential is 0.184 kcal/mol for the B3LYP density functional, and the ensemble model accurately reproduces the large basis set interaction energy curves on the S66x8 dataset.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗