Search NASA⌕ Search

SEARCH · Search NASA

Results for “input”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

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

At least 1,747 records · Page 97

Kernel-based global sensitivity analysis obtained from a single data set

Results from global sensitivity analysis (GSA) often guide the understanding of complicated input–output systems. Kernel-based GSA methods have recently been proposed for their capability of treating a broad scope of complex systems. In this paper, we develop a new set of kernel GSA tools when only a single set of input–output data is available. Three key advances are made: (1) A new numerical estimator is proposed that demonstrates an empirical improvement over previous procedures. (2) A computational method for generating inner statistical functions from a single data set is presented. (3) A theoretical extension is made to define conditional sensitivity indices, which reveal the degree that the inputs carry shared information about the output when inherent input–input correlations are present. Utilizing these conditional sensitivity indices, a decomposition is derived for the output uncertainty based on what is called the optimal learning sequence of the input variables, which remains consistent when correlations exist between the input variables. Further, while these advances cover a range of GSA subjects, a common single data set numerical solution is provided by a technique known as the conditional mean embedding of distributions. The new methodology is implemented on benchmark systems to demonstrate the provided insights.

42 ENGINEERING↗

The potential of different artificial neural network (ANN) techniques in daily global solar radiation modeling based on meteorological data

The main objective of present study is to predict daily global solar radiation (GSR) on a horizontal surface, based on meteorological variables, using different artificial neural network (ANN) techniques. Daily mean air temperature, relative humidity, sunshine hours, evaporation, and wind speed values between 2002 and 2006 for Dezful city in Iran (32 16'N, 48 25'E), are used in this study. In order to consider the effect of each meteorological variable on daily GSR prediction, six following combinations of input variables are considered: (I)Day of the year, daily mean air temperature and relative humidity as inputs and daily GSR as output. (II)Day of the year, daily mean air temperature and sunshine hours as inputs and daily GSR as output. (III)Day of the year, daily mean air temperature, relative humidity and sunshine hours as inputs and daily GSR as output. (IV)Day of the year, daily mean air temperature, relative humidity, sunshine hours and evaporation as inputs and daily GSR as output. (V)Day of the year, daily mean air temperature, relative humidity, sunshine hours and wind speed as inputs and daily GSR as output. (VI)Day of the year, daily mean air temperature, relative humidity, sunshine hours, evaporation and wind speed as inputs and daily GSR as output. Multi-layer perceptron (MLP) and radial basis function (RBF) neural networks are applied for daily GSR modeling based on six proposed combinations. The measured data between 2002 and 2005 are used to train the neural networks while the data for 214 days from 2006 are used as testing data. The comparison of obtained results from ANNs and different conventional GSR prediction (CGSRP) models shows very good improvements (i.e. the predicted values of best ANN model (MLP-V) has a mean absolute percentage error (MAPE) about 5.21% versus 10.02% for best CGSRP model (CGSRP 5)). (author)

14 SOLAR ENERGY↗

Assessing overall building energy performance of a large population of residential single-family homes using limited field data

The U.S. Department of Energy (DOE), through the Building Energy Codes Program (BECP), supports building energy code development and implementation processes to achieve the maximum practicable, cost-effective improvements in energy efficiency while providing safe, healthy buildings for occupants. Currently, there is a lack of statistically representative data that shows how energy codes are implemented in the field and the energy savings opportunities that exist. To address this gap, DOE initiated a research study targeted at new residential single-family homes in 8 states starting in the fall of 2014. The study comprises of a baseline phase to determine current levels of compliance with the key energy provisions of the residential building energy code, identify issues, and calculate potential energy savings from improved compliance; a second phase implementing education, training and outreach activities designed to increase targeted code compliance; and a post-study phase to measure the impact of the training and outreach activities compared to the baseline phase. Initial analyses determined the code requirements with the largest energy impact and the minimum number of observations required to detect statistically significant differences in energy use. Data was collected in the field during a single site visit to avoid bias from repeat visits. However, the single site visit principle prevented energy simulation of individual homes due to the lack of all required model inputs. To address this, energy simulation was conducted using Pacific Northwest National Laboratory’s prototype building models and employing bootstrap random sampling on the data collected from all homes visited. The modeled energy consumption using the observed field data was evaluated relative to that associated with the minimum code requirements and potential savings associated with improved compliance of select items were assessed. This paper discusses the methodology developed for this field study, and presents the results and findings from the baseline phase.

Xie, YuLong↗

SymbolNet: neural symbolic regression with adaptive dynamic pruning for compression

Abstract Compact symbolic expressions have been shown to be more efficient than neural network (NN) models in terms of resource consumption and inference speed when implemented on custom hardware such as field-programmable gate arrays (FPGAs), while maintaining comparable accuracy (Tsoi et al 2024 EPJ Web Conf. 295 09036). These capabilities are highly valuable in environments with stringent computational resource constraints, such as high-energy physics experiments at the CERN Large Hadron Collider. However, finding compact expressions for high-dimensional datasets remains challenging due to the inherent limitations of genetic programming (GP), the search algorithm of most symbolic regression (SR) methods. Contrary to GP, the NN approach to SR offers scalability to high-dimensional inputs and leverages gradient methods for faster equation searching. Common ways of constraining expression complexity often involve multistage pruning with fine-tuning, which can result in significant performance loss. In this work, we propose S y m b o l N e t , a NN approach to SR specifically designed as a model compression technique, aimed at enabling low-latency inference for high-dimensional inputs on custom hardware such as FPGAs. This framework allows dynamic pruning of model weights, input features, and mathematical operators in a single training process, where both training loss and expression complexity are optimized simultaneously. We introduce a sparsity regularization term for each pruning type, which can adaptively adjust its strength, leading to convergence at a target sparsity ratio. Unlike most existing SR methods that struggle with datasets containing more than O ( 10 ) inputs, we demonstrate the effectiveness of our model on the LHC jet tagging task (16 inputs), MNIST (784 inputs), and SVHN (3072 inputs).

Tsoi, Ho Fung (ORCID:0000000225502184)↗

Robust extraction of the proton charge radius from electron-proton scattering data

Background: Extracting the proton charge radius from electron scattering data requires determining the slope of the charge form factor at Q 2 15 of zero. But as experimental data never reach that limit, numerous methods for making the extraction have been proposed, though often the functions are determined after seeing the data which can lead to confirmation bias. Purpose: To find functional forms that will allow for a robust extraction of the input radius for a wide variety of functional forms in order to have confidence in the extraction from upcoming low Q 2 experimental data such as the Jefferson Lab PRad experiment. Method: We create a general framework for inputting form-factor functions as well as various fitting functions. The input form factors are used to generate pseudo-data with fluctuations intended to mimic the binning and random uncertainty of a given set of real data. All combinations of input functions and fit functions can then be tested repeatedly against regenerated pseudo-data. Since the input radius is known, this allows us to find those functions that are robust for radius extractions in an objective fashion. Results: For the range and uncertainty of the PRad data, we find that a two-parameter rational function, a two-parameter continued fraction and the second order polynomial expansion of z can extract the input radius regardless of the input charge form factor function that is used. Conclusions: We have created an easily expandable framework to search for functional forms that allow for a robust extraction of the radius from a given binning and uncertainty of pseudo-data generated from a wide variety of trial functions. This method has enabled a successful search for the best functional forms to extract the radius from the upcoming PRad data, and can be used for other experiments.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Computer Model Calibration with Time Series Data Using Deep Learning and Quantile Regression

Computer models play a key role in many scientific and engineering problems. One major source of uncertainty in computer model experiments is input parameter uncertainty. Computer model calibration is a formal statistical procedure to infer input parameters by combining information from model runs and observational data. The existing standard calibration framework suffers from inferential issues when the model output and observational data are high-dimensional dependent data, such as large time series, due to the difficulty in building an emulator and the nonidentifiability between effects from input parameters and data-model discrepancy. To overcome these challenges, we, in this work, propose a new calibration framework based on a deep neural network (DNN) with long short-term memory layers that directly emulates the inverse relationship between the model output and input parameters. Adopting the “learning with noise” idea, we train our DNN model to filter out the effects from data-model discrepancy on input parameter inference. We also formulate a new way to construct interval predictions for DNN using quantile regression to quantify the uncertainty in input parameter estimates. Through a simulation study and real data application with the Weather Research and Forecasting Model Hydrological modeling system (WRF-Hydro), we show our approach can yield accurate point estimates and well-calibrated interval estimates for input parameters.

97 MATHEMATICS AND COMPUTING↗

SubKit

Generating a CTF (thermal-hydraulics subchannel code) model can be time consuming and error prone because of the rigid formatting of the input file and the amount of input requirements. Because of the structured input format, making small changes to a model can require large scale changes to the input file. Furthermore, a user typically has to perform calculations to develop CTF input parameters from actual model geometry. Changing model geometry requires re-calculating input parameters and copying them from wherever the calculation was performed, be it a script, spreadsheet, or piece of paper. This process complicates the quality assurance process for that model and makes changing the model more difficult. CtfDeckBuilder is an object-oriented Python interface to CTF that allows the user to build a CTF model in Python. The user no longer needs to interface with the CTF input file when using the CtfDeckBuilder. The interface reduces the information required of the user by offering defaults for most modeling options provided by CTF. The user can perform calculations directly in Python when building the model so that calculations stay with the model itself.

Salko, Jr., RobertK↗

JAva GUi for Applied Research (JAGUAR) v 3.0

JAGUAR is a Java software tool for automatically rendering a graphical user interface (GUI) from a structured input specification. It is designed as a plug-in to the Eclipse workbench to enable users to create, edit, and externally execute analysis application input decks and then view the results. JAGUAR serves as a GUI for Sandia's DAKOTA software toolkit for optimization and uncertainty quantification. It will include problem (input deck)set-up, option specification, analysis execution, and results visualization. Through the use of wizards, templates, and views, JAGUAR helps uses navigate the complexity of DAKOTA's complete input specification. JAGUAR is implemented in Java, leveraging Eclipse extension points and Eclipse user interface. JAGUAR parses a DAKOTA NIDR input specification and presents the user with linked graphical and plain text representations of problem set-up and option specification for DAKOTA studies. After the data has been input by the user, JAGUAR generates one or more input files for DAKOTA, executes DAKOTA, and captures and interprets the results

Chan, Ethan↗

Energy Productivity of the High Velocity Algae Raceway Integrated Design (ARID-HV)

The ARID raceway is an effective method to maintain temperature in the optimal growing range. However, the energy input is excessive. Thus, the ARID-HV raceway was developed in order to reduce energy input requirements. This was accomplished by improving pumping efficiency and using a serpentine flow pattern in which the water flows through channels instead of over barriers. A prototype ARID-HV raceway was installed in Tucson, Arizona in order to evaluate the flow and energy requirements of the raceway. Preliminary results show continued high energy usage, but this paper explores the possible energy reductions with the new ARID-HV design with efficient propeller pumps and properly engineered channel lengths. Channel lengths are evaluated with Manning’s equation in order to determine the maximum channel length that could be implemented with the current pumping configuration

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Incident IR Bandwidth Effects on Efficiency and Shaping for Third Harmonic Generation of Quasi-Rectangular UV Longitudinal Profiles

The photocathode of the proposed LCLS RF Photoinjector will be irradiated by uv laser light which is generated as the third harmonic of incident fundamental ir laser light. We have investigated quantitatively the effect of input ir spectral bandwidth on the exiting longitudinal intensity profiles, energy conversion efficiencies and spectral bandwidths that characterize the third harmonic generation (THG) process with a pair of crystals. These profiles, efficiencies and bandwidths include the residual fundamental and residual second harmonic light exiting the second crystal. The intrinsic acceptance bandwidth for THG is determined by crystal material and thickness as well as the type of phase matching that is used. For our case of BBO material with type I phase matching these bandwidths are approximately 0.9 nm*cm and 0.1 nm*cm for second and third harmonic generation respectively. Consequently for fixed crystal thicknesses and a fixed input ir longitudinal profile, the specified input ir bandwidth will determine the profiles, efficiencies and bandwidths exiting the second crystal. The results reported here are predictions of the SNLO code that is available as 'freeware' from the Sandia National Laboratories. It has been modified for this work. It is critical to note that this modification has enabled us to generate SNLO predictions of the 'coupled' case in which the output of the first crystal is used as input to the second crystal. Our focus is the dependence of uv longitudinal intensity profile and THG efficiency on the input ir bandwidth and crystal thicknesses. We include here cases that best illustrate input bandwidth effects. The criteria for selection of reported cases are highest efficiency generation of quasi-rectangular uv profiles with proportional intensity ripple less than 5% rms on the plateau of the pulse. Maximizing THG efficiency typically amounts to maximizing the crystal thicknesses with the longitudinal profile constraint. The specified incident ir longitudinal profile is quasi-rectangular (i.e. nonzero risetime and falltime with small intensity variation on the plateau) with a 10 psec pulse duration (FWHM). By assumption, this profile has been established upstream of the crystals at the fundamental ir wavelength. The simplest possible optical configuration is used in this work as shown in figure 1. The first crystal is the site of second harmonic generation (SHG) driven by the incident ir irradiation of central wavelength, 800nm. Downstream of the first crystal, the second crystal is the site of third harmonic generation (THG) which occurs by sum frequency mixing. Inter-crystal optics (such as a half waveplate) are assumed to be lossless at the fundamental and second harmonic wavelengths. As shown in figure 1, a portion of the incident ir irradiation is not sequestered from the first crystal for subsequent THG in the second crystal. Also, quasi-phase matching configurations and other complex compensation schemes have not been investigated at this point. The simplistic geometry better elucidates the intrinsic acceptance bandwidth limitations imposed by the crystals. Our goal in this endeavor has been to conduct a quantitative assessment of incident ir bandwidth effects on the THG process for BBO material of varied thicknesses and not, at this stage, to comply with all uv pulse specifications for the LCLS RF Photoinjector. Nonetheless, our results can be compared with LCLS photoinjector uv pulse requirements which call for a nominal 10 psec FWHM with 1 psec risetime and falltime and a nominally flat plateau (allowing for slope adjustments) with no more than a 5% rms proportional intensity variation. Furthermore, the results of this work can be used to suggest crystal thicknesses that would likely comply with all uv pulse requirements given the appropriate longitudinal profile and bandwidth for an input ir pulse.

43 PARTICLE ACCELERATORS↗

Climate Change Feedbacks from Interactions Between New and Old Carbon

Priming effects, or responses of SOM decomposition rates to inputs of new, labile carbon (C), have the potential to dramatically alter projections of ecosystem C storage. Priming effects occur in most ecosystems, are significant in magnitude, and are highly sensitive to global changes. Nevertheless, our mechanistic understanding of priming effects remains poor, and this has prevented the inclusion of these dynamics into current Earth system models (ESMs). We conducted two manipulative experiments in the field to quantify how priming effects influence SOM dynamics. Specifically, we asked: To what extent do inputs of “new” root-derived carbon (C) influence “older” C in SOM, and are the magnitude and direction of these effects sensitive to climate? We addressed these questions within the Boston-Area Climate Experiment - an old-field ecosystem that has been subjected to three precipitation treatments (ambient, -50%, and +50% of each precipitation event during the growing season) and four warming treatments (from ambient to +4°C) since 2008. In the first experiment, we installed root and fungal ingrowth cores into the plots. Each core was filled with SOM that had an isotopic signature (of its C compounds) that differed from the vegetation in the plots such that inputs of “new” C from roots/fungi could be quantified using the change in isotopic signatures of C in the cores. Further, we used cores with different mesh sizes to isolate root vs. mycorrhizal fungal inputs. We found that belowground C fluxes were dominated by root inputs (as opposed to mycorrhizal inputs), and that root-derived inputs were greatest in the plots subjected to experimental warming. Given that that the warming-induced increase in belowground C flux did not result in a net increase in soil C, we conclude that the warming treatment likely enhanced priming effects in these soils. In the second experiment, we experimentally dripped dissolved organic C compounds into soils in the BACE plots to simulate root-derived C fluxes. Specifically, we constructed artificial roots attached to an automated peristaltic pump to deliver the compounds to soil semi-continuously during the peak of the growing season. We found that changes in exudate quality had small but significant effects on microbial activities, often interacting with N availability and temperature-induced changes. These results further underscore the importance of priming effects, especially under warming conditions. Collectively, our results provide some of the first field-based estimates of how soil moisture and temperature can directly and indirectly alter root-induced changes in SOM dynamics. This exploratory project lays the groundwork for further research on priming that incorporates effects of plant species and microbial communities to global changes. Such information should enable the development of more mechanistic and better predictive models of SOM decomposition under increased greenhouse gas levels, with the ultimate goal of reducing the level of uncertainty in projections of future climate.

54 ENVIRONMENTAL SCIENCES↗

VERA and VERA-EDU 3.5 Release Notes

The Virtual Environment for Reactor Applications components included in this distribution include selected computational tools and supporting infrastructure that solve neutronics, thermal-hydraulics, fuel performance, and coupled neutronics-thermal hydraulics problems. The infrastructure components provide a simplified common user input capability and provide for the physics integration with data transfer and coupled-physics iterative solution algorithms. Neutronics analysis can be performed for 2D lattices, 2D core and 3D core problems for pressurized water reactor geometries that can be used to calculate criticality and fission rate distributions by pin for input fuel compositions. MPACT uses the Method of Characteristics transport approach for 2D problems. For 3D problems, MPACT uses the 2D/1D method which uses 2D MOC in a radial plane and diffusion or SPn in the axial direction. MPACT includes integrated cross section capabilities that provide problem-specific cross sections generated using the subgroup methodology. The code can be executed both 2D and 3D problems in parallel to reduce overall run time. A thermal-hydraulics capability is provided with CTF (an updated version of COBRA-TF) that allows thermal-hydraulics analyses for single and multiple assemblies using the simplified VERA common input. This distribution also includes coupled neutronics/thermal-hydraulics capabilities to allow calculations using MPACT coupled with CTF. The VERA fuel rod performance component BISON calculates, on a 2D or 3D basis, fuel rod temperature, fuel rod internal pressure, free gas volume, clad integrity and fuel rod waterside diameter. These capabilities allow simulation of power cycling, fuel conditioning and deconditioning, high burnup performance, power uprate scoping studies, and accident performance. Input/Output capabilities include the VERA Common Input (VERAIn) script which converts the ASCII common input file to the intermediate XML used to drive all of the physics codes in the VERA Core Simulator (VERA-CS). VERA component codes either input the VERA XML format directly, or provide a preprocessor which can convert the XML into native input. VERAView is an interactive graphical interface for the visualization and engineering analyses of output data from VERA. The python-based software is easy to install and intuitive to use, and provides instantaneous 2D and 3D images, 1D plots, and alpha-numeric data from VERA multi-physics simulations. Testing within CASL has focused primarily on Westinghouse four-loop reactor geometries and conditions with example problems included in the distribution.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Wildland Fire Induced Heating of Dome 375 Perma-Con®

AET-1 was tasked by ADEM with determining the temperature rise in the drum contents of drums stored in the Dome 375 Perma-Con® at TA-54 given a wildland fire. The wildland fire causes radiative and convective heating on the Perma-Con® exterior. The wildland fire time histories for the radiative and convective heating environment were provided to AET-1 by EES-16. If the calculated temperature rise results in a drum content temperature over 40 °C, then ADEM desires a design solution to ensure the peak temperature remains below 40 °C. An axi-symmetric FE simulation was completed to determine the peak temperature of the contents of a drum stored in the Dome 375 Perma-Con® during a wildland fire event. Three wildland fire time histories for the radiative and convective heat transfer were provided by EES-16 and were inputs for the FE simulation. The maximum drum content temperature reached was found to be 110 °C while using inputs from the SiteG_2ms_4ign_wind_from_west.xlsx time history input and not including the SWB in the model. Including the SWB in the results in a peak drum content temperature of 61 °C for the SiteG_2ms_4ign_wind_from_west.xlsx inputs. EES-16 decided that by using fuel mitigation efforts, such as mowing the grass and shrubs near the Perma-Con® they could reduce the shrub/grass fuel loading near the Perma-Con® from 1.46 kg/m 2 to 0.146 kg/m 2 and by using a less conservative fuel loading for the debris field inside the Dome 375 perimeter, reducing it from 0.58 kg/m2 to 0.058 kg/m 2 in their model. They also greatly increased the resolution of their radiation model and increased the accuracy of their model’s required convergence value. Using this refined input the maximum drum content temperature was found to be 28 °C with no SWB present in the model. Additionally, this refined input model was modified to include worst case emissivity values for the concrete, drum and Perma-Con® interior, along with adding a 91 second long residual radiative heat flux of 2,000 W/m2 to the end of the refined wildland fire input. For this case the maximum drum content temperature was found to be 32 °C. For Rev. 2 of this calculation and additional simulation was run that included a cable fire heat flux on the exterior of the Perma-Con® that was calculated by FP-DO. Including the cable fire heat flux in the model without the SWB resulted in a peak drum content temperature over time of 43 °C. Including the SWB in the simulation with the cable fire heat flux resulted in a peak drum content temperature over time of 35 °C.

42 ENGINEERING↗

PROTEUS-MOC User Manual

The PROTEUS-MOC code is a three-dimensional (3D) neutron transport code based on the finite element mesh and the method of characteristics (MOC) which combines a 2D MOC with the discontinuous Galerkin finite element method for the axial direction. Thus, for PROTEUS-MOC, a 3D geometry is represented with by the implicit extrusion of a single 2D planar geometry with different material assignments allowable on each plane. The code requires four input files for a steady-state calculation without the thermal feedback: a driver input, a mesh input, a cross section input, and a material assignment input. With thermal feedback, a T/H input file is needed. For kinetics, a kinetics driver input file is required. The PROTEUS-MOC software produces a text output file as well as a data file which includes integral and average quantities such on fluxes, powers, temperatures, densities, etc. The detailed output file can be converted to the VTK format using the post processing code so that the outputs can be visualized using a visualization software such as VisIt.

97 MATHEMATICS AND COMPUTING↗

Active Subspace Methods for Data-Intensive Inverse Problems (Final Report)

Fostered by advances in computing, simulation and data science have emerged as the third and fourth paradigms enabling scientific discovery. Many mathematical challenges arise when integrating these paradigms to produce new knowledge. One of the most important challenges is the inverse or calibration problem: find the inputs of a simulation such that its outputs agree with a given set of measurements. As both the cost of the complex simulations and the volume of measurement data sets increase, intuition-driven trial-and-error methods for calibration quickly become untenable; modern scientists need more computationally efficient and mathematically rigorous methods. When measurements and simulated predictions contain uncertainty, the Bayesian perspective on statistical inference provides a popular mathematical framework for developing computational methods to solve inverse problems. Markov chain Monte Carlo (MCMC) methods are a mainstay of this framework; they characterize a probability distribution on the space of model inputs that is conditioned on the given measurements. However, despite recent advances, MCMC methods remain impractical for expensive simulations with more than a handful of inputs. At its heart, MCMC searches the space of inputs for corresponding outputs that match measurements, and so it is afflicted with the curse of dimensionality. Additionally, MCMC methods become less efficient as the volume of the measurement data increases, particularly if the measurements are redundant. To make MCMC—and Bayesian methods, in general—amenable to today’s computer models with millions of inputs and terabytes of measurement data, one must devise methods to reduce the dimension of both the simulation inputs and the measurements. For many models used in practice, the outputs vary primarily on a low-dimensional manifold of the nominally high-dimensional input space. Recently developed active subspace methods discover this low-dimensional manifold and exploit it to enable analyses typically suited for low-dimensional settings (e.g., statistical inversion of complex systems). We hypothesize that active subspace methods can provide the necessary dimension reduction, enabling existing Bayesian methods to find solutions for a large class of otherwise intractable statistical inverse problems. We will test the hypothesis with a series of mathematical analyses and computational experiments that will proceed along three research veins: (i) integrating active subspace methods into methods for statistical inversion, (ii) addressing current challenges in active subspace methods, and (iii) incorporating methods for nonlinear dimensionality reduction into active subspace methods. We will apply the methods we develop to real inverse problems in chemical kinetics and turbulent flame modeling. Our team combines expertise in methods for inverse problems, dimension reduction, and reduced order modeling. With a history of active collaboration and thoughtful engagement, we bring enthusiasm and fresh perspectives to a well studied subject. We are excited by the potential opportunity to work together to extend and apply active subspace methods to data-intensive inverse problems.

97 MATHEMATICS AND COMPUTING↗

Active Subspace Methods for Data-Intensive Inverse Problems

Fostered by advances in computing, simulation and data science have emerged as the third and fourth paradigms enabling scientific discovery. Many mathematical challenges arise when integrating these paradigms to produce new knowledge. One of the most important challenges is the inverse or calibration problem: find the inputs of a simulation such that its outputs agree with a given set of measurements. As both the cost of the complex simulations and the volume of measurement data sets increase, intuition-driven trial-and-error methods for calibration quickly become untenable; modern scientists need more computationally efficient and mathematically rigorous methods. When measurements and simulated predictions contain uncertainty, the Bayesian perspective on statistical inference provides a popular mathematical framework for developing computational methods to solve inverse problems. Markov chain Monte Carlo (MCMC) methods are a mainstay of this framework; they characterize a probability distribution on the space of model inputs that is conditioned on the given measurements. However, despite recent advances, MCMC methods remain impractical for expensive simulations with more than a handful of inputs. At its heart, MCMC searches the space of inputs for corresponding outputs that match measurements, and so it is afflicted with the curse of dimensionality. Additionally, MCMC methods become less efficient as the volume of the measurement data increases, particularly if the measurements are redundant. To make MCMC—and Bayesian methods, in general—amenable to today’s computer models with millions of inputs and terabytes of measurement data, one must devise methods to reduce the dimension of both the simulation inputs and the measurements. For many models used in practice, the outputs vary primarily on a low-dimensional manifold of the nominally high-dimensional input space. Recently developed active subspace methods discover this low-dimensional manifold and exploit it to enable analyses typically suited for low-dimensional settings (e.g., statistical inversion of complex systems). We hypothesize that active subspace methods can provide the necessary dimension reduction, enabling existing Bayesian methods to find solutions for a large class of otherwise intractable statistical inverse problems. We will test the hypothesis with a series of mathematical analyses and computational experiments that will proceed along three research veins: (i) integrating active subspace methods into methods for statistical inversion, (ii) addressing current challenges in active subspace methods, and (iii) incorporating methods for nonlinear dimensionality reduction into active subspace methods. We will apply the methods we develop to real inverse problems in chemical kinetics and turbulent flame modeling. Our team combines expertise in methods for inverse problems, dimension reduction, and reduced order modeling. With a history of active collaboration and thoughtful engagement, we bring enthusiasm and fresh perspectives to a well studied subject. We are excited by the potential opportunity to work together to extend and apply active subspace methods to data-intensive inverse problems.

97 MATHEMATICS AND COMPUTING↗

VERAIO Software Management Plan

VERAIO is a set of utility codes used to provide a common set of input and outputs to the Virtual Environment for Reactor Applications (VERA). VERA is a collection of several different computer codes that all have a common input and output. This prevents the need to manage input and output from each individual code, allowing for ease of use and reducing errors associated with code operability. The VERAIO utilities include VERAIn, VERAView, and VERARun. Each of these utilities is described below. VERAIn is an input processor that reads an ASCII input file generated by users, parses the file, performs some error checking, and writes an XML file to be read by other VERA codes. The main purpose of VERAIn is to provide a common input to all of the VERA codes so users only need to learn one input. VERAIn is written in Perl and uses YAML configuration files to provide flexibility. VERAView is a graphical user interface (GUI) that reads a VERA HDF output file and allows users to visualize results. VERAView is written in Python. VERARun is a script that drives the VERA execution in a high performance computing (HPC) environment. Work performed at the code level supports the Quality Assurance Program Plan (QAPP) (VERA-QA-001), and VERA Software Quality Assurance Plan (VERA-QA-002).

97 MATHEMATICS AND COMPUTING↗

VERAIO Software Management Plan

VERAIO is a set of utility codes used to provide a common set of inputs and outputs to the Virtual Environment for Reactor Applications (VERA). VERA is a collection of several different computer codes that all have a common input and output. This prevents the need to manage input and output from each individual code, allowing for ease of use and reducing errors associated with code operability. The VERAIO utilities include VERAIn, VERAView, and VERARun. Each of these utilities is described below. VERAIn is an input processor that reads an ASCII input file generated by users, parses the file, performs some error checking, and writes an XML file to be read by other VERA codes. The main purpose of VERAIn is to provide a common input to all of the VERA codes, so users only need to learn one input. VERAIn is written in Perl and uses YAML configuration files to provide flexibility. VERAView is a graphical user interface (GUI) that reads a VERA hierarchical data format (HDF) output file and allows users to visualize results. VERAView is written in Python. VERARun is a script that drives the VERA execution in a high performance computing (HPC) environment. Work performed at the code level supports the quality assurance program plan (QAPP) (VERA-QA-001) and the VERA Software Quality Assurance Plan (VERA-QA-002).

97 MATHEMATICS AND COMPUTING↗