Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Mathematical programming”

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 433 records · Page 24
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Automatic Generation of High-Performance Convolution Kernels on ARM CPUs for Deep Learning

In this work, we present FastConv, a template-based code auto-generation open source library that can automatically generate high-performance deep learning convolution kernels of arbitrary matrices/tensors shapes. FastConv is based on the Winograd algorithm, which is reportedly the highest performing algorithm for the time-consuming convolution layers of convolutional neural networks. ARM CPUs cover a wide range designs and specifications, from embedded devices to HPC-grade CPUs. The leads to the dilemma of how to consistently optimize Winograd-based convolution solvers for convolution layers of different shapes. FastConv addresses this problem by using templates to auto-generate multiple shapes of tuned kernels variants suitable for skinny tall matrices. As a performance portable library, FastConv transparently searches for the best combination of kernel shapes, cache tiles, scheduling of loop orders, packing strategies, access patterns, and online/offline computations. Auto-tuning is used to search the parameter configuration space for the best performance for a given target architecture and problem size. The experiments with layer-wise evaluation on the VGG--16 model confirms a 1.25x performance gains is got by tuning the Winograd library. Integrated comparison results shows 1.02x to 1.40x, 1.14x to 2.17x, and 1.22x and 2.48x speedup is achieved over NNPACK, Arm NN, and FeatherCNN on the Kunpeng 920 beside few cases. Furthermore, problem size performance portability experiments with various convolution shapes shows that FastConv achieves 1.2x to 1.7x speedup and 2x to 22x speedup over NNPACK and ARM NN inference engine using Winograd on Kunpeng 920 . CPU performance portability evaluation on the VGG--16 show an average speedup over NNPACK of 1.42x, 1.21x, 1.26x, 1.37x, 2.26x, and 11.02x is observed on Kunpeng 920, Snapdragon 835, 855, 888, Apple M1, and AWS Graviton2, respectively.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Binary Optimal Control of Single-Flux-Quantum Pulse Sequences

We introduce a binary, relaxed gradient, trust-region method for optimizing pulse sequences for single flux quanta (SFQ) control of a quantum computer. The pulse sequences are optimized with the goal of realizing unitary gate transformations. Each pulse has a fixed amplitude and duration. Here we model this process as an binary optimal control problem, constrained by Schrödinger’s equation, where the binary variables indicate whether each pulse is on or off. We introduce a first-order trust-region method, which takes advantage of a relaxed gradient to determine an optimal pulse sequence that minimizes the gate infidelity, while also suppressing leakage to higher energy levels. The proposed algorithm has a computational complexity of O(p log(p)), where p is the number of pulses in the sequence. We present numerical results for the H and X gates, where the optimized pulse sequences give gate fidelity’s better than 99.9%, in ≈ 25 trust-region iterations.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2023

Architecture-Preserving Provable Repair of Deep Neural Networks

Deep neural networks (DNNs) are becoming increasingly important components of software, and are considered the state-of-the-art solution for a number of problems, such as image recognition. However, DNNs are far from infallible, and incorrect behavior of DNNs can have disastrous real-world consequences. This paper addresses the problem of architecture-preserving V-polytope provable repair of DNNs. A V-polytope defines a convex bounded polytope using its vertex representation. V-polytope provable repair guarantees that the repaired DNN satisfies the given specification on the infinite set of points in the given V-polytope. An architecture-preserving repair only modifies the parameters of the DNN, without modifying its architecture. The repair has the flexibility to modify multiple layers of the DNN, and runs in polynomial time. It supports DNNs with activation functions that have some linear pieces, as well as fully-connected, convolutional, pooling and residual layers. To the best our knowledge, this is the first provable repair approach that has all of these features. We implement our approach in a tool called APRNN. Using MNIST, ImageNet, and ACAS Xu DNNs, we show that it has better efficiency, scalability, and generalization compared to PRDNN and REASSURE, prior provable repair methods that are not architecture preserving.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Performing k eff Validation of As-Loaded Criticality Safety Calculations Using UNF-ST&DARDS: Applicable Experiment Selection

The general method for performing validation of as loaded criticality safety calculations using UNF ST&DARDS is presented in a paper by Clarity, which includes a description of the UNF-ST&DARDS system. Proof-of-principle analyses were performed in the summer of 2019 for MPC-32 dual purpose canisters (DPCs) containing pressurized water reactor (PWR) fuel assemblies. Summaries of these results are presented in this and a companion paper for this conference. The current paper describes the TSUNAMI-IP calculations performed to select applicable experiments for validation of 11 MPC-32 DPCs. The companion paper discusses the TSUNAMI-3D calculations used to generate sensitivity data to support the experiment selections discussed here. Experiment selection is based on the sensitivity/uncertainty (S/U) methods used to validate criticality safety calculations of as-loaded DPCs containing pressurized water reactor (PWR) spent nuclear fuel (SNF). This process has been demonstrated and is summarized in this paper. The approach is similar to that used in NUREG/CR-7109, which provides an approach for validation of PWR burnup credit (BUC), including major and minor actinides and major fission products. The premise of S/U-based validation is that applicable experiments—those having a similar bias to a given application system—will have similar sensitivities for each isotope and reaction in the two systems. It is assumed that cross sections with larger uncertainties are more likely to contain data errors which contribute to the bias. The integral index c k thus propagates the system sensitivities with the nuclear covariance data to calculate a correlation coefficient representing the similarity of the two systems. In this work, a c k value of 0.8 or higher is interpreted as identifying an experiment with sufficient similarity for use in validation. This paper presents a brief summary of the characteristics of the 11 MPC-32 DPCs used in the proof of-principle analysis for as-loaded criticality safety calculation validation and an overview of the critical experiment suite with which each of these DPC models was compared. A summary and discussion of c k results is also presented, followed by conclusions and a discussion of future work to be performed for validation of UNF ST&DARDS as-loaded criticality safety calculations.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2020

Validation of MCNP Critical Benchmark Models of PU-MET-FAST-016

A new centralized repository of high-quality MCNP models of critical benchmark experiments is currently under development at Los Alamos National Laboratory (LANL). The repository is the combined effort of the Nuclear Criticality Safety, Nuclear Data, and Monte Carlo code development/application organizations at LANL. The initial set of benchmark models in the repository are derived from the Whisper Suite provided with MCNP6.2 and are evaluated against the benchmark experiments described in the International Criticality Safety Benchmark Evaluation Project (ICSBEP) Handbook. The goal is to build a single LANL benchmark collection that is up to date with the latest ICSBEP revision, has a formal review and revision process, is contained in an open-source repository, and utilizes new Python tools for improved input and output file review. This paper describes the validation of the models associated with PU-MET-FAST-016, “Flooded 3X3X3 Arrays of 3-kg Plutonium (Pu) Metal Cylinders – Phase I”. In this critical experiment, twenty-seven 3-kg Pu metal cylinders with a height of 46.33 ± 0.15 mm and diameter of 65.25 ± 0.05 mm were arranged in a cubic array. The Pu cylinders were sealed in aluminum cans, and three cans were placed in each of nine perforated aluminum sleeves, as shown in Fig. 1. The nine sleeves were supported by an aluminum frame that allowed for control of the lateral spacing of the cylinders, both in the X and Y directions. The benchmark experiment measured thirteen configurations. Of these configurations, six were modeled using MCNP.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Sensitivity Calculations for Systems with Polyethylene Reflector Materials Using CLUTCH

The SCALE 6.2.4 code package contains four sequences for calculating $k_{eff}$ sensitivity coefficients. Two of these sequences use deterministic transport solvers: a one-dimensional (1D) capability based on XSDRN, and a two-dimensional (2D) capability based on NEWT. These sequences are restricted to the multigroup (MG) treatment of neutron energy. The three-dimensional (3D) sequences use the KENO V.a or KENO-VI Monte Carlo transport codes and can be used to calculate sensitivity coefficients with either MG or continuous-energy (CE) transport. The 3D sensitivities are ultimately reported in an MG structure, regardless of the method used in the transport calculations. If desired, the sensitivity coefficients can be reported with very fine energy resolution from a CE calculation, but they are calculated only in the MG library structure in the MG mode. CE TSUNAMI methods are available in SCALE starting in SCALE version 6.2. Sensitivity coefficients were generated using the 3D sequences as part of the generation of the SCALE 6.2.2 Validation Report; difficulties encountered when using the CLUTCH method for thick, fissionable-material reflectors were discussed and investigated as documented in a previous paper. This paper discusses the difficulties encountered in generating accurate sensitivity coefficients using the CLUTCH technique for polyethylene reflectors for two fast spectrum benchmarks. Direct perturbation (DP) calculations were performed to confirm the accuracy of the total sensitivity coefficient for important isotopes with large sensitivities in the system. Discrepancies were detected for CLUTCH-calculated sensitivity coefficients in the reflector of a critical experiment with a radial polyethylene reflector. A simple polyethylene-reflected plutonium sphere was then used to further investigate the discrepancy. Calculations performed using the iterated fission probability (IFP) method generated accurate sensitivity coefficients in both cases. The results of this study emphasize the need to confirm CLUTCH sensitivity results with DP calculations. IFP calculations are generally less efficient but more reliable than CLUTCH calculations. Improvements to the CLUTCH methodology that retain the greater efficiency but address identified difficulties are therefore potentially useful to analysts.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Hierarchical Data Format for Nuclear Data Sensitivities

The SCALE code system includes capabilities for sensitivity and uncertainty (S/U) analysis as part of its TSUNAMI code suite. The sensitivity of a quantity of interest (for example, an application’s $k_{eff}$) to nuclear data is stored as a profile in a text-based file, which is known as a sensitivity data file (SDF). The sensitivity profile can be used to calculate uncertainties, correlation coefficients, and similarity indices. One of the goals of the present work was to seek general performance improvements in the TSUNAMI code suite, starting with the TSUNAMI-IP code for calculating similarity indices. Through profiling, it was found that reading the text-based sensitivity files was a performance bottleneck in the TSUNAMI-IP code. In a typical TSUNAMI-IP calculation, an application might be compared to thousands of benchmarks, thus requiring the reading of thousands of SDFs. Reading of binary-based data is generally faster than reading text-based data. Hierarchical Data Format 5 (HDF5) is a binary-based format that also benefits from being portable, and it can be inspected with nonproprietary tools. This paper describes an HDF5-based file format that has been introduced for SDFs.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Sensitivity-based Similarity Metrics for New Experiment Design Optimization

The nuclear data used in advanced reactor simulations requires validation. Data from nuclear criticality experiments can provide this validation. New nuclear criticality experiment design requires extensive knowledge and expert judgement such that the experimental design parameters are selected in such a way to keep the experiment subcritical. To aide in this experimental design process, professionals can utilize sensitivity and uncertainty analysis. Sensitivity and uncertainty analysis relies on matching new application experiments with currently existing benchmark experiments. Currently, there is functionality in the Whisper 1.1 software package to calculate a similarity metric based on neutron multiplication factor sensitivity coefficients between a new application designed by the user and existing International Criticality Safety Benchmark Experiment Project (ICSBEP) benchmarks. The Whisper 1.1 software package is included in Monte Carlo N-Particle ® Code Version 6.21 (MCNP ® 6.2). This work is geared toward expanding this capability to new similarity metrics based on beta-effective sensitivity coefficients and reactivity coefficient sensitivity coefficients. While the investigation of these sensitivity coefficients is presented in detail in separate works at this same conference, this work will be primarily focused on studying the similarity metrics in more detail. These similarity metrics will then be incorporated into the optimization algorithms used for experiment design in EUCLID (Experiments Underpinned by Computational Learning for Improvements in nuclear Data), which is a Los Alamos National Laboratory (LANL) project designed to constrain nuclear data of interest, such that adjustments can be made to possible inaccuracies. A more detailed optimization can be subsequently performed by breaking down these similarity metrics by isotope, reaction, and energy.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2020

KENO-VI Primer: Performing Calculations using SCALE’s Criticality Safety Analysis Sequence (CSAS6) with Fulcrum

The SCALE code system developed at Oak Ridge National Laboratory is widely used and accepted around the world for criticality safety analysis. The well-known KENO-VI three-dimensional Monte Carlo criticality computer code is one of the primary criticality safety analysis tools in SCALE. The KENO-VI primer is designed to help a new user understand and use the SCALE/KENO-VI Monte Carlo code for nuclear criticality safety analysis. It assumes that the user has a college education in a technical field. There is no assumption of familiarity with Monte Carlo codes in general or with SCALE/KENO-VI in particular. The primer is designed to teach by example, with each example illustrating two or three features of SCALE/KENO-VI that are useful in criticality analysis. The primer is based on SCALE 6.2 and 6.3, which includes the Fulcrum graphical user interface. Each example uses Fulcrum to provide the framework for preparing input data and viewing output results. Starting with a Quickstart section, the primer gives an overview of the basic requirements for SCALE/KENO-VI input and allows the user to quickly run a simple criticality problem with SCALE/KENO-VI. Each following section begins with a list of basic objectives identifying the goal of the section and the individual SCALE/KENO-VI features covered in detail in the section’s sample problems. Upon completion of the primer, a new user should be comfortable using Fulcrum to set up criticality problems in SCALE/KENO-VI. The primer provides a starting point for the criticality safety analyst who uses SCALE/KENO-VI. Complete descriptions are provided in the SCALE/KENO-VI manual. Although the primer is self-contained, it is intended as a companion volume to the SCALE/KENO-VI training and documentation. The SCALE manual and training schedule are available at https://scale.ornl.gov. The primer provides specific examples of using SCALE/KENO-VI for criticality analysis; the SCALE/KENO-VI manual provides information on the use of SCALE/KENO-VI and all its modules. The primer also contains an appendix with sample input files. In addition, this primer, its errata, and sample inputs are also available at https://code.ornl.gov/scale/primers/kenovi/.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Energy Management Information Systems Technical Resources Report

This report illustrates how energy management information systems (EMIS) can enable federal agencies to improve energy performance, reduce costs, and meet energy goals. The guide discusses how to identify and implement needed EMIS functions and capabilities, covering the entire EMIS life cycle including system design, procurement, implementation, and ongoing maintenance.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2022

Bayesian Optimization Framework for Imperfect Data or Models

Conventional Bayesian optimization methods implicitly assume that the data and model being optimized are “perfect.” This assumption leads to inaccurate posterior probability distribution functions (PDFs) when applied to “imperfect” data or models. The new Bayesian optimization framework presented in this report provides a way to parameterize the effect of imperfections usually encountered in a prior PDF of generalized data or a model on the posterior PDF. The effects of imperfections are parameterized by a set of constraints imposed on the posterior expectation values of deviations between the data and the model and on their covariance matrix elements. A particular set of values for these constraints conveys an evaluator’s best estimate of the effect of imperfections on the corresponding posterior expectation values. When a prior PDF of generalized data is assumed to be normal, an expression for a posterior PDF satisfying an arbitrary set of constraints is derived analytically for linear models. An analogous iterative algorithm is given for nonlinear models. The corresponding posterior PDF should be used to estimate any posterior expectation values in the presence of imperfections parameterized by that set of constraints. A posterior PDF of a conventional Bayesian optimization method is recovered analytically when all evaluator-specified constraints are set to zero (i.e., in the absence of any imperfections). The analytical expressions derived in this report for normal PDFs and linear models were verified numerically by a Metropolis–Hastings Monte Carlo method. The methods presented herein could be applied to any kind of data or models, including differential cross-section data or integral benchmark experiments.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

EUCLID: Experiments Underpinned by Computational Learning for Improvements in Nuclear Data [Slides]

EUCLID will design validation experiments optimized to resolve compensating errors and adjust nuclear data to experiments. A big part of the success of the EUCLID proposal was due to previous work supported by NCSP (MCNP, nuclear data, and NCERC capabilities). The work performed under EUCLID will similarly benefit the NCSP mission. It will lead to new MCNP capabilities, improved nuclear data and nuclear data capabilities, and new methodology and tools that will have large impact on future NCERC experiments.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2022

Illustration of Neutronic Decoupling in Fresh Fuel Storage Environments [Slides]

Operations beyond static storage of fuel assemblies occur in SFPs (moving around the pool, inspection, etc.). These assemblies are assumed suitably distanced at 30 cm. When assemblies are separated by enough water, the neutrons from one cannot cause fission in another, which results in neutronic decoupling. This assumption was investigated for PWR and BWR assemblies as part of ORNL-TM/2021/2330.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2022

Criticality and Radiation Shielding Impacts of ENDF/B-VIII.0 [Slides]

Several impacts of ENDF/B-VIII.0 on Radiation Shielding Applications were observed by researchers. Most reaction rates, attenuation ratios, and dose rates are similar for ENDF/B-VII.1 and ENDF/B-VIII.0. A significant reduction in reactor dosimetry results was noted. Investigation continues to reveal underlying data for discrepancies.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Cumulative χ 2 Metric for VALID for ENDF/B-VII.1 and ENDF/B-VIII.0 in SCALE 6.3b9 [Slides]

Quantifying the performance of a new evaluation is difficult and doing so for a new library is significantly more challenging. Cumulative χ 2 metric is an established approach to comparing distributions. In this case, C/E results from different libraries. Andrei Trkov and Roberto Capote have published cumulative χ 2 results for ENDF/B-VIII.0. Cumulative χ 2 metric was recently calculated for VALID library maintained at ORNL.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2022

Status of the Slide Rule Update - Phase 4 Plutonium Configurations – Delayed Gamma [Slides]

The fourth phase of Slide-Rule update gives a calculation scheme for DFG doses and its application to five plutonium material, as well as comparisons between radiation transport and depletion codes such as MCNP, SCALE, COG and FISPACT. Overall, DFG dose rates calculated by each laboratory led to consistent results. Extra effort is under way to identify the cause of the remaining differences, especially for case 2, by comparing precisely the gamma source, and particularly nuclides inventories.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗