Reducing the Feature Distance between Real and Synthetic Data for Training Safeguards-Relevant Computer Vision Models
Explore the source record for details and available documents.
SEARCH · Search NASA
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.
Explore the source record for details and available documents.
Computational fluid dynamics (CFD) enables modeling, analysis, and visualization of phenomena that would usually be impossible or extremely expensive to measure experimentally.
Large-scale computational models have become common tools for analyzing complex man made systems. However, when coupled with optimization or uncertainty quantification methods in order to conduct extensive model exploration and analysis, the computational expense quickly becomes intractable. Furthermore, these models may have both continuous and discrete parame ters. One common approach to mitigating the computational expense is the use of response surface approximations. While well developed for models with continuous parameters, they are still new and largely untested for models with both continuous and discrete parameters. In this work, we describe and investigate the performance of three types of response surfaces developed for mixed variable models: Adaptive Component Selection and Shrinkage Operator, Treed Gaussian Process, and Gaussian Process with Special Correlation Functions. We focus our efforts on test problems with a small number of parameters of interest, a characteristic of many physics-based engineering models. We present the results of our studies and offer some insights regarding the performance of each response surface approximation method.
The Cellular-Potts model is a powerful and ubiquitous framework for developing computational models for simulating complex multicellular biological systems. Cellular-Potts models (CPMs) are often computationally expensive due to the explicit modeling of interactions among large numbers of individual model agents and diffusive fields described by partial differential equations (PDEs). In this work, we develop a convolutional neural network (CNN) surrogate model using a U-Net architecture that accounts for periodic boundary conditions. We use this model to accelerate the evaluation of a mechanistic CPM previously used to investigate in vitro vasculogenesis. The surrogate model was trained to predict 100 computational steps ahead (Monte-Carlo steps, MCS), accelerating simulation evaluations by a factor of 562 times compared to single-core CPM code execution on CPU. Over short timescales of up to 3 recursive evaluations, or 300 MCS, our model captures the emergent behaviors demonstrated by the original Cellular-Potts model such as vessel sprouting, extension and anastomosis, and contraction of vascular lacunae. This approach demonstrates the potential for deep learning to serve as a step toward efficient surrogate models for CPM simulations, enabling faster evaluation of computationally expensive CPM simulations of biological processes.
This paper describes the development of a computational fluid dynamics methodology to simulate the hydrodynamics of horizontal-axis tidal current turbines.
Computational models that formalize complex human behaviors enable study and understanding of such behaviors. However, collecting behavior data required to estimate the parameters of such models is often tedious and resource intensive. Thus, estimating dataset size as part of data collection planning (also known as Sample Size Determination) is important to reduce the time and effort of behavior data collection while maintaining an accurate estimate of model parameters. In this paper, we present a sample size determination method based on Uncertainty Quantification (UQ) for a specific Inverse Reinforcement Learning (IRL) model of human behavior, in two cases: 1) pre-hoc experiment design—conducted in the planning stage before any data is collected, to guide the estimation of how many samples to collect; and 2) post-hoc dataset analysis—performed after data is collected, to decide if the existing dataset has sufficient samples and whether more data is needed. Here, we validate our approach in experiments with a realistic model of behaviors of people with Multiple Sclerosis (MS) and illustrate how to pick a reasonable sample size target. Our work enables model designers to perform a deeper, principled investigation of effects of dataset size on IRL.
Efficient production of algal biofuels could reduce dependence on foreign oil by providing a domestic renewable energy source. Moreover, algae-based biofuels are attractive for their large oil yield potential despite decreased land use and natural resource (e.g., water and nutrients) requirements compared to terrestrial energy crops. Important factors controlling algal lipid productivity include temperature, nutrient availability, salinity, pH, and the light-to-biomass conversion rate. Here, computational approaches allow for inexpensive predictions of algae growth kinetics for various bioreactor sizes and geometries without the need for multiple, expensive measurement systems. Parametric studies of algal species include serial experiments that use off-line monitoring of growth and lipid levels. Such approaches are time consuming and usually incomplete, and studies on the effect of the interaction between various parameters on algal growth are currently lacking. However, these are the necessary precursors for computational models, which currently lack the data necessary to accurately simulate and predict algae growth. In this work, we conduct a lab-scale parametric study of the marine alga Nannochloropsis salina and apply the findings to our physics-based computational algae growth model. We then compare results from the model with experiments conducted in a greenhouse tank and an outdoor, open-channel raceway pond. Results show that the computational model effectively predicts algae growth in systems across varying scale and identifies the causes for reductions in algal productivities. Applying the model facilitates optimization of pond designs and improvements in strain selection.
Computer simulations that demonstrate the value of novel approaches are crucial to developing more flexible and robust power systems operations with high penetrations of renewable energy at multiple geographic and temporal scales. However, optimization-based simulations that depend on forecast data often face challenges in evaluating performance, reproducing results, and testing under realistic simulation scenarios. In this paper, we develop scientific computing best-practices for the validation and reproduction of power systems operational models. We then employ two case studies to demonstrate the proposed validation and reproduction framework.
Mathematical and computational models are widely used for the analysis and design of both physical and financial systems. Modeling the electric grid is of particular importance to China for three reasons. First, power-sector assets are expensive and long-lived, and they are critical to any country's development. China's electric load, transmission, and other energy-related infrastructure are expected to continue to grow rapidly; therefore it is crucial to understand and help plan for the future in which those assets will operate (NDRC ERI 2015). Second, China has dramatically increased its deployment of renewable energy (RE), and is likely to continue further accelerating such deployment over the coming decades. Careful planning and assessment of the various aspects (technical, economic, social, and political) of integrating a large amount of renewables on the grid is required. Third, companies need the tools to develop a strategy for their own involvement in the power market China is now developing, and to enable a possible transition to an efficient and high RE future.
Accurate and efficient computational modeling of material behavior is essential to the DOE's mission of advancing the development of devices and components needed for power generation and storage. One of the outstanding challenges in atomistic simulation of condensed systems, such as solids, liquids, and glasses, is access to experimentally meaningful length and time scales. The project developed and improved several solution methods to overcome these challenges: diffusive molecular dynamics, adaptive kinetic Monte Carlo, coarse-graining of transition state theory, finite temperature coarse-graining methods for crystalline defects, and optimization-based atomistic-to-continuum coupling methods.
In many situations across computational science and engineering, multiple computational models are available that describe a system of interest. These different models have varying evaluation costs and varying fidelities. Typically, a computationally expensive high-fidelity model describes the system with the accuracy required by the current application at hand, while lower-fidelity models are less accurate but computationally cheaper than the high-fidelity model. Outer-loop applications, such as optimization, inference, and uncertainty quantification, require multiple model evaluations at many different inputs, which often leads to computational demands that exceed available resources if only the high-fidelity model is used. This work surveys multifidelity methods that accelerate the solution of outer-loop applications by combining high-fidelity and low-fidelity model evaluations, where the low-fidelity evaluations arise from an explicit low-fidelity model (e.g., a simplified physics approximation, a reduced model, a data-fit surrogate) that approximates the same output quantity as the high-fidelity model. The overall premise of these multifidelity methods is that low-fidelity models are leveraged for speedup while the high-fidelity model is kept in the loop to establish accuracy and/or convergence guarantees. We categorize multifidelity methods according to three classes of strategies: adaptation, fusion, and filtering. The paper reviews multifidelity methods in the outer-loop contexts of uncertainty propagation, inference, and optimization.
Estimating the probability of failure for complex real-world systems using high-fidelity computational models is often prohibitively expensive, especially when the probability is small. Exploiting low-fidelity models can make this process more feasible, but merging information from multiple low-fidelity and high-fidelity models poses several challenges. Here, this paper presents a robust multi-fidelity surrogate modeling strategy in which the multi-fidelity surrogate is assembled using an active learning strategy using an on-the-fly model adequacy assessment set within a subset simulation framework for efficient reliability analysis. The multi-fidelity surrogate is assembled by first applying a Gaussian process correction to each low-fidelity model and assigning a model probability based on the model's local predictive accuracy and cost. Three strategies are proposed to fuse these individual surrogates into an overall surrogate model based on model averaging and deterministic/stochastic model selection. The strategies also dictate which model evaluations are necessary. No assumptions are made about the relationships between low-fidelity models, while the high-fidelity model is assumed to be the most accurate and most computationally expensive model. Through two analytical and two numerical case studies, including a case study evaluating the failure probability of Tristructural isotropic-coated (TRISO) nuclear fuels, the algorithm is shown to be highly accurate while drastically reducing the number of high-fidelity model calls (and hence computational cost).
Research on integrative modeling has gained considerable attention in recent years and expert opinion has been increasingly recognized as an important data source and modeling contributor. However, little research has systematically compared and evaluated expert elicitation methods in terms of their ability to link to computational models that capture human behavior and social phenomena. In this paper, we describe a decision-making process we used for evaluating and selecting a task specific elicitation method within the framework of integrative computational social-behavioral modeling. From the existing literature, we identified the characteristics of problems that each candidate method is well suited to address. A small-scale expert elicitation was also conducted to evaluate the comparative strength and weaknesses of the methods against a number of consensus-based decision criteria. By developing a set of explicit method evaluation criteria and a description characterizing decision problems for the candidate methods, we seek to gain a better understanding of the feasibility and cost-effectiveness of integrating elicitation methods with computational modeling techniques. This serves an important first step toward expanding our research effort and trajectory toward greater interdisciplinary modeling research of human behavior.
Systems biology has experienced dramatic growth in the number, size, and complexity of computational models. To reproduce simulation results and reuse models, researchers must exchange unambiguous model descriptions. We review the latest edition of the Systems Biology Markup Language (SBML), a format designed for this purpose. A community of modelers and software authors developed SBML Level 3 over the past decade. Its modular form consists of a core suited to representing reaction-based models and packages that extend the core with features suited to other model types including constraint-based models, reaction-diffusion models, logical network models, and rule-based models. The format leverages two decades of SBML and a rich software ecosystem that transformed how systems biologists build and interact with models. More recently, the rise of multiscale models of whole cells and organs, and new data sources such as single-cell measurements and live imaging, has precipitated new ways of integrating data with models. We provide our perspectives on the challenges presented by these developments and how SBML Level 3 provides the foundation needed to support this evolution.
Mixing between fuel and shell materials in ICF implosions can affect implosion dynamics and even prevent ignition. We use data from a series of separated reactant experiments on the National Ignition Facility to calibrate and test the predictive power of gas–shell mix models. Two models are used to estimate fuel–shell mix: a Reynolds-averaged turbulence model and molecular diffusion. Minor uncertainties in capsule manufacture, experimental conditions, and values for mix model parameters produce significant variation in simulation results. Using input/output pairs from 1D simulations, we train Gaussian process surrogate models to predict experimental quantities of interest. The surrogates are used to construct posteriors for mix model parameters by marginalizing over uncertainties in capsule manufacture and experimental conditions. Mix models are calibrated with a subset of experimental data (neutron yields, ion temperature, and bang time) and tested using the remaining data. In general, both the diffusion and turbulence model correctly predict experimental DT and TT neutron yields. Despite having more free parameters, the turbulence model underpredicts ion temperature at high convergence ratio. Furthermore, the simpler diffusion model correctly predicts these temperatures, suggesting nonhydrodynamic gas–shell mix. The computational model consistently overpredicts DD neutron yield, indicating possible shortcomings outside of the mix model.
Metal–organic frameworks (MOFs) have premium exceptional properties for a variety of functions, such as gas separation and storage and catalysis. The large variety of possible inorganometallic nodes and organic linkers provide an almost unlimited number of combinations for assembling MOFs, which makes the experimental characterization and examination of all potentially useful combinations practically impossible. Furthermore, experimental studies of MOFs typically fall short in uncovering crucial details regarding their mechanisms of action or the molecular details responsible for their functional properties, such as the nature of adsorbate binding or the structures of transition states. Computational modeling has, therefore, become an efficient and important tool for strategizing the functionalization of MOFs and explicating the mechanisms of their functions. Furthermore, we review the computational methodologies used for computational studies of MOFs, especially Kohn–Sham density functional theory and combined quantum mechanical and molecular mechanical methods for calculating their structural, electronic, and magnetic properties, as well as for understanding the mechanisms of MOFs' applications to magetic devices, thermal conduction, gas adsorption, separation, storage, and sensing, thermal catalysis, photocatalysis, and electrocatalysis.
Cardiovascular implantable devices alter the biofluid dynamics and biochemistry of the blood in which they are placed. These perturbations can lead to thrombus formation which may or may not be desired, depending on the application. In this work, a computational model is developed that couples biofluid dynamics and biochemistry to predict the clotting response of blood to such devices. The model consists of 28 advection–diffusion–reaction partial differential equations to track proteins in the blood involved in clotting and utilizes boundary flux terms to model the initiation of the intrinsic clotting pathway at thrombogenic device surfaces. We use this model to simulate the transient clot growth within a 2D idealized bifurcation aneurysm filled with various distributions of bare metal coils with similar packing densities. The clot model predicts initial clot formation to occur in areas along coil surfaces where flow is minimal and where time-averaged shear rates are the smallest. Among the six coil-filled aneurysm cases simulated, maximum thrombus occlusion ranged between 80.8 and 92.2% of the post-treatment aneurysm volume and was achieved 325–450 s after treatment. Finally, with further refinement and validation, the computational clotting model will be a valuable engineering tool for evaluating and comparing the relative performance of cardiovascular implantable devices.
Practical applications of synthetic self-propelled nano and microparticles for microrobotics, targeted drug delivery, and manipulation at the nanoscale are rapidly expanding. However, fabrication limitations often hinder progress, resulting in relatively simple shapes and limited functionality. Here, taking advantage of 3D nanoscale printing, chiral micropropellers powered by the hydrogen peroxide reduction reaction are fabricated. Due to their chirality, the propellers exhibit multifunctional behavior controlled by an applied magnetic field: spinning in place (loitering), directed migration in the prescribed direction, capture, and transport of polymer cargo particles. Design parameters of the propellers are optimized by computation modeling based on mesoscale molecular dynamics. It is predicted by computer simulations, and confirmed experimentally, that clockwise rotating propellers attract each other and counterclockwise repel. These results shed light on how chirality and shape optimization enhance the functionality of synthetic autonomous micromachines.