Search NASA⌕ Search

SEARCH · Search NASA

Results for “mathematics computing”

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 1,693 records · Page 94

Green AI: Insights Into Deep Learning's Looming Energy Efficiency Crisis

As demands grow to integrate artificial intelligence into every aspect of industry, commerce, and life, deep learning's exploding energy cost has become a looming crisis, making AI systems a salient energy-efficiency challenge. One might expect that doubling a neural network's size would halve its error rate, or at least allow it to achieve greater performance given the same amount of time and energy. I will present clear and substantial scientific evidence which indicates that not only is this intuition wildly wrong, but that neural networks scale so poorly that to increase deep learning performance by only a small fraction can easily require an order of magnitude or more increase in computational resources and energy. Further, the marginal trade-off price of to increase model performance rapidly explodes as performance targets are increased. To address this challenge, I will provide a toolkit of techniques that can be applied today to mitigate the inefficiency of modern deep learning. And, I will conclude by illuminating a practical path forward towards efficient, Green AI.

artificial intelligence↗

Application of the discrete generalized multigroup method to ultra-fine energy mesh in infinite medium calculations

The Discrete Generalized Multigroup (DGM) method uses discrete Legendre orthogonal polynomials to expand the energy dependence of the multigroup neutron transport equation. This allows a solution on a fine energy mesh to be approximated for a cost comparable to a solution on a coarse energy mesh. The DGM method is applied to an ultra-fine energy mesh (14,767 groups) to avoid using self-shielding methodologies without introducing the cost usually associated with such energy discretization. Results show DGM to converge to the reference ultra-fine solution after a small number of recondensation steps for multiple infinite medium compositions. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Internal neutronics-temperature coupling in Serpent 2 - Reactivity differences resulting from choice of material property correlations

This paper describes the unique way of simultaneously solving the power and temperature distributions of a nuclear system with the Monte Carlo neutron transport code Serpent 2. The coupled solution is achieved through the implementation of an internal temperature solver and material property correlations in the code. The program structure is reviewed concerning the temperature solver and the internal correlations as well as the internal coupling between these two and the neutron transport part. To estimate the reactivity differences resulting from correlation choices a simple pin-cell case has been calculated. It is established, that some correlation choices may result in difference in reactivity of approximately 100 pcm. (authors)

97 MATHEMATICS AND COMPUTING↗

Proteus-MOC: A 3D deterministic solver incorporating 2D method of characteristics

A new transport solution methodology was developed by combining the two-dimensional method of characteristics with the discontinuous Galerkin method for the treatment of the axial variable. The method, which can be applied to arbitrary extruded geometries, was implemented in PROTEUS-MOC and includes parallelization in group, angle, plane, and space using a top level GMRES linear algebra solver. Verification tests were performed to show accuracy and stability of the method with the increased number of angular directions and mesh elements. Good scalability with parallelism in angle and axial planes is displayed. (authors)

97 MATHEMATICS AND COMPUTING↗

Decision Points and Practical Considerations for AI Projects

In this presentation, I will present business-relevant decisions, risks, and considerations for practical implementations of AI projects. I will use energy efficiency and renewable energy AI projects at NREL as examples and case-studies highlighting the journey from concept to implementation. First, I present challenges, questions, and trade-offs related to system inputs: the data. Next, I will examine issues with system behavior and trust, presenting examples, risks, and mitigation strategies. Finally, I will discuss challenges to effective widespread deployment of AI systems including energy, compute, and time requirements.

AI↗

General method for exponential-type equations for eight- and nine-point prismatic arrays

The results of three-parameter experiments are commonly interpreted by the trilinear equation for eight data in a prismatic array. Ifa center point estimate is available, the eight-and nine-point arrays can be represented by new exponential-type equations. The equations are easy to generate, they are invariant under data translation, and they estimate curvature coefficients.

97 MATHEMATICS AND COMPUTING↗

Total Order Reliability in PSA: Importance of Basic Events and Systems

The purpose of this work is twofold. First, to formalize the properties of the total order reliability importance measure for PSA models. Second, to extend the definition of the total order importance measure to groups of basic events. This allows one to obtain the importance of systems and to address the relevance of interactions among systems.

97 MATHEMATICS AND COMPUTING↗

Simulating Microbial Community Patterning Using Biocellion

Mathematical modeling and computer simulation are important tools for understanding complex interactions between cells and their biotic and abiotic environment: similarities and differences between modeled and observed behavior provide the basis for hypothesis forma- tion. Momeni et al. [5] investigated pattern formation in communities of yeast strains engaging in different types of ecological interactions, comparing the predictions of mathematical modeling and simulation to actual patterns observed in wet-lab experiments. However, simu- lations of millions of cells in a three-dimensional community are ex- tremely time-consuming. One simulation run in MATLAB may take a week or longer, inhibiting exploration of the vast space of parameter combinations and assumptions. Improving the speed, scale, and accu- racy of such simulations facilitates hypothesis formation and expedites discovery. Biocellion is a high performance software framework for ac- celerating discrete agent-based simulation of biological systems with millions to trillions of cells. Simulations of comparable scale and accu- racy to those taking a week of computer time using MATLAB require just hours using Biocellion on a multicore workstation. Biocellion fur- ther accelerates large scale, high resolution simulations using cluster computers by partitioning the work to run on multiple compute nodes. Biocellion targets computational biologists who have mathematical modeling backgrounds and basic C++ programming skills. This chap- ter describes the necessary steps to adapt the original Momeni et al.'s model to the Biocellion framework as a case study.

metabolic interaction↗

A micro-computer based system to compute magnetic variation

A mathematical model of magnetic variation in the continental United States (COT48) was implemented in the Ohio University LORAN C receiver. The model is based on a least squares fit of a polynomial function. The implementation on the microprocessor based LORAN C receiver is possible with the help of a math chip, Am9511 which performs 32 bit floating point mathematical operations. A Peripheral Interface Adapter (M6520) is used to communicate between the 6502 based micro-computer and the 9511 math chip. The implementation provides magnetic variation data to the pilot as a function of latitude and longitude. The model and the real time implementation in the receiver are described.

Kaul, R.↗

Using Sequence-Specific Chemical and Structural Properties of DNA to Predict Transcription Factor Binding Sites

An important step in understanding gene regulation is to identify the DNA binding sites recognized by each transcription factor (TF). Conventional approaches to prediction of TF binding sites involve the definition of consensus sequences or position-specific weight matrices and rely on statistical analysis of DNA sequences of known binding sites. Here, we present a method called SiteSleuth in which DNA structure prediction, computational chemistry, and machine learning are applied to develop models for TF binding sites. In this approach, binary classifiers are trained to discriminate between true and false binding sites based on the sequence-specific chemical and structural features of DNA. These features are determined via molecular dynamics calculations in which we consider each base in different local neighborhoods. For each of 54 TFs in Escherichia coli, for which at least five DNA binding sites are documented in RegulonDB, the TF binding sites and portions of the non-coding genome sequence are mapped to feature vectors and used in training. According to cross-validation analysis and a comparison of computational predictions against ChIP-chip data available for the TF Fis, SiteSleuth outperforms three conventional approaches: Match, MATRIX SEARCH, and the method of Berg and von Hippel. SiteSleuth also outperforms QPMEME, a method similar to SiteSleuth in that it involves a learning algorithm. The main advantage of SiteSleuth is a lower false positive rate.

59 BASIC BIOLOGICAL SCIENCES↗

Computationally Efficient Training of Deep Neural Networks via Transfer Learning

Transfer learning is a highly successful approach to train deep neural networks for customized image recognition tasks. Transfer learning leverages deep learning models that are previously trained on massive datasets and re-trains the network for a novel image recognition dataset. Typically, the advantage of transfer learning has been measured in sample efficiency, but instead, we investigate the computational efficiency of transfer learning. A good pre-trained model provides features that can be used as input to a new classifier (usually the top layers of a neural network). We show that if a good pre-trained model is selected, that training a new classifier can be much more computationally efficient than training a deep neural network without transfer learning. The first step in transfer learning is to select a pre-trained model to use as a feed-forward network for generating features. Here this selection process either uses human intuition/convenience or a methodical but computationally intensive validation loop. Instead, we would prefer a method to select the pre-trained model that will produce the best transfer results for the least amount of computation. To this end, we provide a computationally efficient metric for the fit between a pre-trained model and a novel image recognition task. The better the fit, the less computation will be needed to re-train the pre-trained model for the novel task. As machine learning becomes ubiquitous, highly-accurate trained models will proliferate and computationally efficient transfer learning methods will enable rapid development of new image recognition models.

97 MATHEMATICS AND COMPUTING↗

Force Reconstruction Using the Inverse of the Mode-Shape Matrix [Book Chapter]

A new technique for force reconstruction is developed. To estimate the externally applied force, this technique sums the weight-scaled acceleration signals, and is referred to as the Sum of Weighted Accelerations Technique (SWAT). To obtain the scalar weights the inverse of the mode shape matrix is used. Application of this technique is illustrated with both numerical calculations using a mass-spring model and experimental data from a structure impacting a rigid barrier.

055001 -- Nuclear Fuels-- Safeguards, Inspection, ↗

Investigating key cell types and molecules dynamics in PyMT mice model of breast cancer through a mathematical model

The most common kind of cancer among women is breast cancer. Understanding the tumor microenvironment and the interactions between individual cells and cytokines assists us in arriving at more effective treatments. Here, we develop a data-driven mathematical model to investigate the dynamics of key cell types and cytokines involved in breast cancer development. We use time-course gene expression profiles of a mouse model to estimate the relative abundance of cells and cytokines. We then employ a least-squares optimization method to evaluate the model’s parameters based on the mice data. The resulting dynamics of the cells and cytokines obtained from the optimal set of parameters exhibit a decent agreement between the data and predictions. We perform a sensitivity analysis to identify the crucial parameters of the model and then perform a local bifurcation on them. The results reveal a strong connection between adipocytes, IL6, and the cancer population, suggesting them as potential targets for therapies.

59 BASIC BIOLOGICAL SCIENCES↗

A theory of supercritical wing sections, with computer programs and examples.

Mathematical methods for the design of supercritical wings, which depend on the numerical solution of the partial differential equations of two-dimensional gas dynamics, are developed. The main contribution is a computer program for the design of shockless transonic airfoils using the hodograph transformation and analytic continuation into the complex domain. The mathematical theory is described, and a manual for users of the programs is provided. Numerical examples are given and computational results are discussed, and the computer programs themselves are listed. The analysis routine can be used to ascertain whether the profiles behave well at off-design conditions, or to smooth coordinates and obtain a desirable shape more quickly when perfectly shockless flow is not essential.

Bauer, F.↗