Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “MATRICES”

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 343 records · Page 19
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

FUN-BioCROP model with litter decomposition parameters derived from the LIDET dataset

This repository contains the code and data necessary to run the FUN-BioCROP (Fixation and Uptake of Nitrogen-Bioenergy Carbon, Rhizosphere, Organisms, and Protection) model with litter decomposition parameters derived from a modified Monte Carlo simulation that used the Long-term Intersite Decomposition Experiment Team dataset. Related publication:Juice, S.M., Ridgeway, J.R., Hartman, M.D., Parton, W.J., Berardi, D.M., Sulman, B.N., Allen, K.E., & Brzostek, E.R. Reparameterizing litter decomposition using a simplified Monte Carlo method improves litter decay simulated by a microbial model and alters bioenergy soil carbon estimates. Description of Files: FUNBioCROP_LIDET Study.Rmd R code with FUN-BioCROP model that can be run with 10 different sets of parameters for litter decomposition (Baseline, LIDET, or eight other best parameter sets identified in the modified Monte Carlo simulation. CORPSE Functions_Bioenergy_V2.R Code with CORPSE model functions, called by FUNBioCROP_LIDET Study.Rmd Model Input Data: bulk.csv, bulk_till.csv, rhizo.csv, rhizo_till.csv, litter.csv Initial C and N (kg C or N/m2) pool values for each soil compartment, final values from spin up. All five files have the same columns: (Column - Description - Units) uFastC - Unprotected fast decomposing carbon - kg carbon/m2 uSlowC - Unprotected slow decomposing carbon - kg carbon/m2 uNecroC - Unprotected necromass carbon - kg carbon/m2 pFastC - Protected fast decomposing carbon - kg carbon/m2 pSlowC - Protected slow decomposing carbon - kg carbon/m2 pNecroC - Protected necromass carbon - kg carbon/m2 livingMicrobeC - Carbon in living microbial biomass - kg carbon/m2 uFastN - Unprotected fast decomposing nitrogen - kg nitrogen/m2 uSlowN - Unprotected slow decomposing nitrogen - kg nitrogen/m2 uNecroN - Unprotected necromass nitrogen - kg nitrogen/m2 pFastN - Protected fast decomposing nitrogen - kg nitrogen/m2 pSlowN - Protected slow decomposing nitrogen - kg nitrogen/m2 pNecroN - Protected necromass nitrogen - kg nitrogen/m2 inorganicN - Inorganic nitrogen - kg nitrogen/m2 CO2 - Carbon in carbon dioxide - kg carbon/m2 livingMicrobeN - Nitrogen in living microbial biomass - kg nitrogen/m2 Model Input Data: FluxTower_AvgSoilT.csv: Average daily soil temperature (oC) at 10 cm depth at University of Illinois Urbana-Champaign (UIUC) Energy Farm flux tower from 7/2008-3/2016. (One year of averaged data) Model Input Data: FluxTower_AvgSoilVWC.csv: Average daily soil volumetric water content (VWC) at 10 cm depth at UIUC Energy Farm flux tower from 7/2008-3/2016. (One year of averaged data) Model Input Data: input_CCS_LIDET Study.csv: This file has daily data to run FUN-BioCROP (Column - Description - Units): yr - calendar year - year doy - day of year (1 to 365) (no leap year) - day anpp - aboveground NPP (DayCent) - kg C/m2/day bnpp - belowground NPP (DayCent) - kg C/m2/day aglivc - live aboveground biomass carbon (DayCent) - kg C/m2 bglivcj - live juvenile fine root biomass carbon (DayCent) - kg C/m2 bglivcm - live mature fine root biomass carbon (DayCent) - kg C/m2 aglivn - live aboveground biomass nitrogen (DayCent) - kg N/m2 bglivnj - live juvenile fine root biomass nitrogen (DayCent) - kg N/m2 bglivnm - live mature fine root biomass nitrogen (DayCent) - kg N/m2 nyr - simulation year - year cult - indicates a cultivation event (0 or 1) crop - indicates a new crop (0 or 1) fert - indicates a fertilizer event (0 or 1) frst - indicates the first day of the growing season (0 or 1) harv - indicates a harvest event (0 or 1) last - indicates the end of the growing season (0 or 1) croptype - crop type (0=none; 1=alfalfa; 2=corn; 3=grass clover pasture; 4=soybean; 5=wheat) cropsrl - crop specific root length - mm/g root cultrhizmix - fraction of rhizosphere mixed with bulk soil during cultivation (0.0-1.0) - fraction cultlitmix - fraction of litter mixed with bulk soil during cultivation (0.0-1.0) - fraction harvremov - fraction of above ground biomass removed during harvest (0.0-1.0) - fraction fertamt - fertilization amount - g N/m2 lifehist - plant life history (0 = annual, 1 = perennial) froot_turnover_c - amount of C in fine root turnover - kg C/m2 froot_turnover_n - amount of N in fine root turnover - kg N/m2 agrd_turnover_c - amount of C in aboveground biomass turnover - kg C/m2 agrd_turnover_n - amount of N in aboveground biomass turnover - kg N/m2 leaf_litter_fastfrac - Fast decomposing fraction of leaf litter (0.0-1.0) - fraction root_litter_fastfrac - Fast decomposing fraction of root litter (0.0-1.0) - fraction root_diameter - root diameter - mm root_length - root length - mm root/m2 rhizo_frac - fraction of total soil volume that is rhizosphere (0.0 - 1.0) - fraction date - date in format YYYY-MM-DD Instructions: Save the model code ("FUN-BioCROP_LIDET Study.Rmd") and accompanything files (data streams and CORPSE function code) in the same folder. In model code "Chunk 3: Load CORPSE Data Streams" set the working directory (setwd) to the folder with the files saved in step #1. In "Chunk 5: Define LIDET parameter sets" select the litter decomposition parameter set to be used in the run, and comment out all other sets. If changing any parameter values, edit them in "Chunk 6: Load parameters." Run all chunks up to and including "Chunk 10: Prepare Data for Export." In "Chunk 11: Export Output Data" edit data frames for export and filenames, as necessary. "Chunk 12: Graph Total Soil C" makes a figure of C remaining over the model run period. Description of each model chunk (in file FUN-BioCROP_LIDET Study.Rmd): Chunk 1: Remove all functions, clear memory. Removes all functions from R environment, clears the memory. Chunk 2: Load Packages. Loads packages necessary to run the code. Chunk 3: Load CORPSE Data Streams. Sets the working directory and loads the data files necessary to run CORPSE. Chunk 4: Load CORPSE Functions. Loads the R script with CORPSE functions from the working directory, "CORPSE Functions_Bioenergy_V2.R". Chunk 5: Define LIDET parameter sets. Has ten different parameter sets for litter decomposition tested in this study: Baseline parameters, LIDET parameters, and the other 8 best performing parameter sets identified in the modified Monte Carlo. To run the model, all but one parameter set must be commented out. Chunk 6: Load Parameters. Loads all fixed parameters to run the model. Data frame with definitions of parameters is in the CORPSE function script "CORPSE Functions_Bioenergy_V2.R" Chunk 7: Prepare Data Streams. Takes data streams loaded in Chunk 3 and puts them in the format necessary to run the model. The model is coded to run at least two sites at a time, so if only one site is being run it must be run in duplicate. Individual data tables of daily values are created in this chunk from the input data file. Chunk 8: Set Initial Conditions. Creates data tables of soil C and N pools for each soil compartment (rhizo_till, rhizo, bulk_till, bulk, litter) and loads initial values into the data tables. Creates lists for each soil compartment to hold model output. Chunk 9: Load FUN Data and Set Up Matrices. Uses DayCent data to calculate FUN input data: root and leaf N demand, total N demand, plant CN, leaf N available for retranslocation, and litter production. Creates matrices for FUN model outputs. Chunk 10: Run Model. Runs the model. Chunk 11: Prepare Data for Export. Combines data from each day saved as lists into data frames for each soil compartment. Adds values from all soil compartments together to calculate total soil values, creates separate data frames for each soil C and N pool (e.g., protected slow C) for the total soil value. Adds different C and N pools together to calculate total soil C and N for all layers. Creates data frame of ratio of protected to unprotected SOC. Organizes FUN data for export. Chunk 12: Export Results. Exports CSV files of model results to the working directory. Chunk 13: Graph Total Soil C. Makes figure of C remaining over time. Related Links: Original FUN-BioCROP model: https://github.com/BrzostekEcologyLab/FUN-BioCROP LIDET dataset: https://andlter.forestry.oregonstate.edu/data/abstract.aspx?dbcode=TD023

Juice, Stephanie↗
DOE OSTI2020

Approximate Inverse Chain Preconditioner: Iteration Count Case Study for Spectral Support Solvers

As the growing availability of computational power slows, there has been an increasing reliance on algorithmic advances. However, faster algorithms alone will not necessarily bridge the gap in allowing computational scientists to study problems at the edge of scientific discovery in the next several decades. Often, it is necessary to simplify or precondition solvers to accelerate the study of large systems of linear equations commonly seen in a number of scientific fields. Preconditioning a problem to increase efficiency is often seen as the best approach; yet, preconditioners which are fast, smart, and efficient do not always exist. Following the progress of [1], we present a new preconditioner for symmetric diagonally dominant (SDD) systems of linear equations. These systems are common in certain PDEs, network science, and supervised learning among others. Based on spectral support graph theory, this new preconditioner builds off of the work of [2], computing and applying a V-cycle chain of approximate inverse matrices. This preconditioner approach is both algebraic in nature as well as hierarchically-constrained depending on the condition number of the system to be solved. Due to its generation of an Approximate Inverse Chain of matrices, we refer to this as the AIC preconditioner. We further accelerate the AIC preconditioner by utilizing precomputations to simplify setup and multiplications in the con-text of an iterative Krylov-subspace solver. While these iterative solvers can greatly reduce solution time, the number of iterations can grow large quickly in the absence of good preconditioners. Initial results for the AIC preconditioner have shown a very large reduction in iteration counts for SDD systems as compared to standard preconditioners such as Incomplete Cholesky (ICC) and Multigrid (MG). We further show significant reduction in iteration counts against the more advanced Combinatorial Multigrid (CMG) preconditioner. We have further developed no-fill sparsification techniques to ensure that the computational cost of applying the AIC preconditioner does not grow prohibitively large as the depth of the V-cycle grows for systems with larger condition numbers. Our numerical results have shown that these sparsifiers maintain the sparsity structure of our system while also displaying significant reductions in iteration counts.1 2

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Study on Fracture of Fiber-Reinforced Polymeric Composites Using Spiral Notch Torsion Test

Fiber-reinforced polymer-matrix composites have been widely applied in various areas due to their excellent mechanical properties. In this study, carbon fiber-reinforced polymer-matrix composites and glass fiber-reinforced polymer-matrix composites failure tests under quasi static torsion have been demonstrated by using spiral notch torsion test methodology. Three types of cylindrical specimens with different fiber orientations were prepared and tested. For both composites, after increase stage, A type exhibited limited stress variations before the sudden failure, whereas B type showed gradual failure with torque drops. Displacement contours, stress contours and energy release rates were acquired using the anisotropic elastic finite element models. Microscopic characterization exhibited that the failure of A type occurred along interfaces between fibers and matrices, whereas circular cracks in B type propagated alternatively between polymeric matrices and interfaces. Type C failure is similar to type A whereas with a curvature observed on the fracture surface.

Du, Fen↗
DOE OSTI2020

Approximate Inverse Chain Preconditioner: Iteration Count Case Study for Spectral Support Solvers

As the growing availability of computational power slows, there has been an increasing reliance on algorithmic advances. However, faster algorithms alone will not necessarily bridge the gap in allowing computational scientists to study problems at the edge of scientific discovery in the next several decades. Often, it is necessary to simplify or precondition solvers to accelerate the study of large systems of linear equations commonly seen in a number of scientific fields. Preconditioning a problem to increase efficiency is often seen as the best approach; yet, preconditioners which are fast, smart, and efficient do not always exist. Following the progress of [1], we present a new preconditioner for symmetric diagonally dominant (SDD) systems of linear equations. These systems are common in certain PDEs, network science, and supervised learning among others. Based on spectral support graph theory, this new preconditioner builds off of the work of [2], computing and applying a V-cycle chain of approximate inverse matrices. This preconditioner approach is both algebraic in nature as well as hierarchically-constrained depending on the condition number of the system to be solved. Due to its generation of an Approximate Inverse Chain of matrices, we refer to this as the AIC preconditioner. We further accelerate the AIC preconditioner by utilizing precomputations to simplify setup and multiplications in the con-text of an iterative Krylov-subspace solver. While these iterative solvers can greatly reduce solution time, the number of iterations can grow large quickly in the absence of good preconditioners. Initial results for the AIC preconditioner have shown a very large reduction in iteration counts for SDD systems as compared to standard preconditioners such as Incomplete Cholesky (ICC) and Multigrid (MG). We further show significant reduction in iteration counts against the more advanced Combinatorial Multigrid (CMG) preconditioner. We have further developed no-fill sparsification techniques to ensure that the computational cost of applying the AIC preconditioner does not grow prohibitively large as the depth of the V-cycle grows for systems with larger condition numbers. Our numerical results have shown that these sparsifiers maintain the sparsity structure of our system while also displaying significant reductions in iteration counts.1 2

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Nonlinear Matrix Approximation with Radial Basis Function Components

We introduce and investigate matrix approximation by decomposition into a sum of radial basis function (RBF) components. An RBF component is a generalization of the outer product between a pair of vectors, where an RBF function replaces the scalar multiplication between individual vector elements. Even though the RBF functions are positive definite, the summation across components is not restricted to convex combinations and allows us to compute the decomposition for any real matrix that is not necessarily symmetric or positive definite. We formulate the problem of seeking such a decomposition as an optimization problem with a nonlinear and non-convex loss function. Several modern versions of the gradient descent method, including their scalable stochastic counterparts, are used to solve this problem. We provide extensive empirical evidence of the effectiveness of the RBF decomposition and that of the gradient-based fitting algorithm. While being conceptually motivated by singular value decomposition (SVD), our proposed nonlinear counterpart outperforms SVD by drastically reducing the memory required to approximate a data matrix with the same L2 error for a wide range of matrix types. For example, it leads to 2 to 6 times memory save for Gaussian noise, graph adjacency matrices, and kernel matrices. Moreover, this proximity-based decomposition can offer additional interpretability in applications that involve, e.g., capturing the inner low-dimensional structure of the data, retaining graph connectivity structure, and preserving the acutance of images.

Rebrova, Elizaveta↗
DOE OSTI2021

Structure Perception in 3D Point Clouds: Preprint

Understanding human perception is critical to the design of ef- fective visualizations. The relative benefits of using 2D versus 3D techniques for data visualization is a complex decision space, with varying levels of uncertainty and disagreement in both the liter- ature and in practice. This study aims to add easily reproducible, empirical evidence on the role of depth cues in perceiving structures or patterns in 3D point clouds. We describe a method to synthesize a 3D point cloud that contains a 3D structure, where 2D projec- tions of the data strongly resemble a Gaussian distribution. We performed a within-subjects structure identification study with 128 participants that compared scatterplot matrices (canonical 2D projections) and 3D scatterplots under three types of motion: rota- tion, xy-translation, and z-translation. We found that users could consistently identify three separate hidden structures under ro- tation, while those structures remained hidden in the scatterplot matrices and under translation. This work contributes a set of 3D point clouds that provide definitive examples of 3D patterns per- ceptible in 3D scatterplots under rotation but imperceptible in 2D scatterplots.

data analysis↗
DOE OSTI2022

Fast simultaneous feasibility testing and applications for electrical power grids

Fast simultaneous feasibility testing (SFT) for management of an electrical power grid is achieved through various innovations. The computation problem relates to evaluation of candidate solutions for external power flows into a power grid, with respect to predetermined constraints and contingencies. Storage and computations are reduced by formulating the problem in terms of transactional nodes (e.g. third party connections for generators and loads) instead of the larger number of bus nodes. Further advantages are achieved by precomputing matrices that can be reused across multiple SFT invocations, organizing matrices and operations to reduce storage and computation, and eliminating branches that have no contingency violations. The fast SFT further enables new applications for integrating SFT with a Security-Constrained Unit Commitment (SCUC) optimizer for efficient single pass solutions; operating SFT as a service to multiple SCUC solvers; or integration with Security-Constrained Economic Dispatch (SCED) optimization.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Synthesis and Characterization of Polymeric Pore Former Microspheres using Modified Polymer Resin

Porosity in polymer parts is often desired to improve mechanical properties. Individual pores must be small enough that bulk mechanical properties are retained but significant enough to reduce the density of the material. Filler materials can be incorporated into polymer matrices and subsequently removed to create voids. The materials used to create the voids are known as pore-formers. The present research aims to create spherical pores in a silicone matrix by inserting and removing core materials or fillers post-cure. A method was developed to incorporate pore formers into an uncured polymer resin and remove the materials from the matrix. Polyethylene Glycol (PEG) and Polylactic Acid (PLA) microspheres are created and incorporated into silicone matrices as pore formers. Microspheres with controllable size distributions are formed via ultrasonic spray pyrolysis and emulsion formation. Microspheres are removed from the matrix by solvent extraction and calcination. PLA microspheres were modified by hydroxyl addition to the surface to study their solubility in water. PEG-PLA block copolymers are also synthesized and investigated, comparing their physical and chemical properties to the monomers to identify whether the block copolymer microspheres will be more easily removable than their monomeric microspheres. Moreover, other properties including thermal stability, rheology, and mechanical response are studied. The present dissertation provides a comprehensive understanding of the synthesis and characterization of PEG and PLA-based pore formers, toward the future development of feedstock for additive manufacturing porous structures.

Damptey, Ransford↗
DOE OSTI2021

A Laboratory Assessment and Optimization of Nanoparticle-Stabilized CO2 Foams for Hydraulic Fracturing Applications

In this research, we have conducted a comprehensive assessment and optimization of aqueous CO2 foams for fracturing applications. The performance of CO2 foams is evaluated using three major matrices: foam stability, foam apparent viscosity, and leakoff. The response of these performance matrices to different internal factors and external conditions is evaluated. These factors and conditions include foam quality, nanoparticle concentration, salinity, shear rate, surfactant type and concentration, polymer concentration, and temperature. Before foam generation, the colloidal stability of the nanosilica dispersion is evaluated at different types of electrolytes and under high temperatures. With a good salinity resistance and thermal stability, the nanosilica is then used for the entire set of foam generation experiments. The foam configuration with optimized parameters is used for the leakoff test, which serves as a supplementary section to further demonstrate the advantages of the proposed foam system. Then the final commercialization section provides a general view of what should be considered to commercialize the research results. In general, it is found that the aqueous CO2 foam system has demonstrated a good potential to be used as a fracturing fluid. The design of the foam system needs to consider formations properties and formation fluid properties etc. to realize the optimal performance of the system on a laboratory scale.

02 PETROLEUM↗
DOE OSTI2020

Potential Advantages to Employing an Onsite Laboratory at FUSRAP Sites Contaminated with Radionuclides and Metals - 20353

Various factors that support an evaluation of the relative costs, advantages and disadvantages of utilizing an onsite laboratory or an offsite laboratory for remediation of Formerly Utilized Sites Remedial Action Program (FUSRAP) sites contaminated with radionuclides and metals are presented. In terms of how much more quickly an onsite laboratory provides results versus an offsite laboratory, the results of this study provide tools for calculating the amount of cost savings per day for excavation of contaminated soils for various scenarios including: - State, municipal or local property remediation where the remediation schedule becomes time-critical due to the extra cost and inconvenience to the town or municipality caused by the need for increased security and safety personnel, possible road closures and detours, and the inconvenience to the public due to modified or decreased services of State or municipal facilities; - Residential property where the remediation schedule becomes time-critical due to the inconvenience to the homeowner as well as the importance of maintaining good community relations; - Commercial property where the remediation schedule becomes time-critical due to the hardship and potential loss of revenue endured by the property owner; - Open area where treatment of excavation water is required, and - Open area where treatment of excavation water is not required. For every scenario, the number of matrices, radionuclide and metal contaminants of concern and the project duration are important factors. The costs for the onsite laboratory included the capital costs associated with setting up the lab such as trailer(s) or building(s), power and water hookup, analytical testing equipment, balances, computers, desks and bookcases; support equipment such as hot plates, filtration assemblies, safety equipment, ovens, etc.; and supplies such as glassware, pipettes, filters, etc. and ongoing costs such as labor, chemicals and reagents, standards, cost of the laboratory obtaining certification and maintenance. Offsite laboratory sample testing costs are generally very competitive and typically substantially lower than onsite laboratory costs if only the sample testing costs are considered. The great disadvantage of using an offsite laboratory is that they often cannot guarantee rapid turnaround times for results which creates project delays associated with the remediation process including delays determining the completeness of the characterization and remediation processes, delays in backfilling a property and additional costs for treating wastewater (excavation water, stormwater or both). These costs will vary greatly depending upon the area of the country in which the project is located due to the wide range of labor costs for the onsite laboratory and field workers, as well as the contaminants of interest, matrices and most importantly the size of the impacted area and the number of properties located in that area. This paper provides guidelines for calculating the cost savings that may be realized if an onsite laboratory can generate results more quickly than an offsite laboratory by estimating the costs associated with the remediation process for the various scenarios noted above such as work crew labor costs, excavator and dump truck equipment rentals, wastewater treatment costs (if applicable) and the additional costs incurred by homeowners, commercial property owners and towns/municipalities that are reimbursed by the potentially responsible parties. Key assumptions involve the typical quickest turnaround time provided by the onsite and offsite laboratories. Costs for the operation of a wastewater treatment system and the remediation of a site area or property were provided by personnel who have been involved in the operation of a large FUSRAP site in New Jersey for many years. This study provides valuable information for determining the most cost-effective approach for analysis of project samples and therefore may be used on upcoming FUSRAP projects. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2022

Nuclear data uncertainty quantification for the nuclide inventory of a Calvert Cliffs spent fuel sample

The impact of nuclear data cross section uncertainties and covariance matrices on the nuclide vector of spent nuclear fuel was investigated; This exercise was carried out for the Calvert Cliffs fuel assembly D047 benchmark available in the SFCOMPO database. Sample P irradiated in rod MKP109 for 4 cycles up to a burnup of approximately 44 GWd/MTU was selected for the analysis. Nuclear data uncertainties were taken from the most recent libraries released by evaluation projects JEFF, ENDF/B and JENDL, and were propagated using the SANDY code via a stochastic sampling approach. This paper provides a quantification of the uncertainty on the concentration of several actinides and fission products relevant for spent fuel management. Uncertainties generally below 5 % were predicted for the concentrations of most of the uranium, neptunium and plutonium isotopes relevant for SNF applications. Curium isotopes carry larger uncertainties that might exceed 10 %. The contribution of cross section uncertainties on the concentrations of fission products was found to be marginal with the exception of a few nuclides. These results can be significantly affected by the lack of evaluated covariance matrices for the capture cross section of several fission products. Burnup tracers such as {sup 148}Nd and {sup 137}Cs have negligible uncertainties because of the power normalisation imposed in every stochastic calculation.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2024

autoGEMM: Pushing the Limits of Irregular Matrix Multiplication on Arm Architectures

This paper presents an open-source library that pushes the limits of performance portability for irregular General Matrix Multiplication (GEMM) on the widely-used Arm architectures. Our library, autoGEMM, is designed to support a wide range of Arm processors: from edge devices to HPC-grade CPUs. autoGEMM generates optimized kernels for various hardware configurations by auto-combining fragments of autogenerated micro-kernels that employ hand-written optimizations to maximize computational efficiency. We optimize the kernel pipeline by tuning the register reuse and the data load/store overlapping. In addition, we use a dynamic tiling scheme to generate balanced tile shapes. Finally, we position autoGEMM on top of the TVM framework where our dynamic tiling scheme prunes the search space for TVM to identify the optimal combination of parameters for code optimization. Evaluations on five different classes of Arm chips demonstrate the advantages of autoGEMM. For small matrices, autoGEMM achieves 98% of peak and up to 2.0x speedup over state-of-the-art libraries such as LIBXSMM and LibShalom. For irregular matrices (i.e. tall skinny and long rectangles), autoGEMM is 1.3-2.0x faster than widely-used libraries such as OpenBLAS and Eigen. autoGEMM is publicly available at: https://github.com/wudu98/autoGEMM.

Wu, Du↗
DOE OSTI2026

Online Alpha Monitoring of High Cs-137 Hanford and SRS High Level Waste with Tensioned Metastable Fluid Detectors

The Department of Energy’s Hanford and Savannah River Sites maintain millions of gallons of caustic supernate and salt high activity waste in their high-level waste (HLW) tank farm inventories. The Savannah River Site is currently treating this waste with a calixarene-based solvent extraction of Cs-137 to reduce these inventories. Hanford is employing an at-tank crystalline silico titanate (CST) based solid phase extraction methodology to reduce their liquid HLW inventories. Due to the high solubility of Cs-137 and the relative insolubility of the actinides in these caustic waste forms, the beta to alpha radioactivity ratio can often exceed six orders of magnitude in the feed solutions to these treatment processes. This unique characteristic leads to significant technical challenges in making rapid gross alpha measurements in the presence of the overwhelming beta, gamma, as well as dissolved sodium salt in these HLW matrices. Conventional radioanalytical techniques, such as liquid scintillation analysis or gas flow proportional counting require significant radiochemistry preparation prior to the radiometric measurements for gross alpha activity. These required pretreatments render these technologies untenable for rapid quantification of gross alpha activity that could be required to support on or at-line measurements ensuring a waste stream will meet regulatory requirements. The radiation measurement properties of Tensioned Metastable Fluid Detectors (TMFDs) have been studied by Purdue University’s Taleyarkhan research group for well over a decade. Fluids tensioned to the appropriate degree will rupture when struck by radiation, resulting in a measurable cavitation event. The negative pressure generating this tension can be adjusted by centrifugal rotation or by acoustic means in such a way that these cavitation events can be generated from alpha radiation but will not be generated by beta or gamma radiation. Purdue University and the Savannah River National Laboratory are currently collaborating to develop a gamma/beta blind, spectroscopic alpha measurement system based on the Tensioned Metastable Fluid Detector technology to provide a potential solution for performing rapid gross alpha measurements on these high gamma/beta sample matrices. Measurements using the Indirect Drive Acoustically Tensioned Metastable Fluid Detectors developed as part of this collaboration were performed with an alpha emitting radionuclide. Successful determination of gross alpha activity was observed, indicating a potential pathway for rapid gross alpha measurements in remote-handled shielded cells or in process situations requiring online alpha monitoring. Measurements using this system have been conducted on high beta activity solutions, demonstrating the beta blind capability of this system. Measurements are currently underway to test the system’s capability to measure gross alpha activity on Savannah River Site high level waste high Cs-137 samples that have been measured by the SRNL radiochemistry team. This work was supported by the DOE EM Technology Development program.

DiPrete, David [Savannah River National Laboratory↗
DOE OSTI2020

Method used in Tokunaga, T. et al. 2018 for simulating time- and depth- dependent temperature variations in soil and sediment in Rifle, Colorado, 2013-2016.

This data package contains temperature, moisture content, matric potential, and carbon dioxide (CO2) concentration data from the Rifle, Colorado field site and from a 2 meter tall laboratory soil/sediment column designed to simulate field conditions. In addition, redox potential data from the laboratory column are included. The purpose of the data and paper is to demonstrate how a laboratory column was constructed to reproduce the dynamics of nonisothermal conditions in the field, such that CO2 production above a variable-depth water table can be studied in the laboratory. The data package contains 3 data files corresponding to the figures in the related paper (Tokunaga, T. et al. 2018) and 1 README file with the paper information.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Accelerating structural dynamics simulations with localised phenomena through matrix compression and projection‐based model order reduction

In this work, a novel approach is introduced for accelerating the solution of structural dynamics problems in the presence of localised phenomena, such as cracks. For this category of problems, conventional projection-based Model Order Reduction (MOR) methods are either limited with respect to the range of system configurations that can be represented or require frequent solutions of the Full Order Model (FOM) to update the low-dimensional spaces, in which solutions are represented. In the proposed approach, low-dimensional spaces, constructed for the healthy structure, are enriched with appropriately selected columns of the flexibility matrix of the system. It can be shown that these spaces contain the solution to the original problem for the static case, while their dimension is much smaller. In order to allow their online construction for arbitrary localised features, the full flexibility matrix of the system should be available. To this end, a hierarchical representation is used for the matrices involved, allowing to compute the flexibility matrix efficiently and with reduced memory requirements. The resulting method offers significant speedups, without sacrificing the flexibility and accuracy of the full order model. The performance and limitations of the approach are studied through a series of examples in structural dynamics.

fracture mechanics↗
DOE OSTI2021

Analysis of the ratio of ℓ 1 and ℓ 2 norms in compressed sensing

We study the ratio of ℓ 1 and ℓ 2 norms ( ℓ 1 / ℓ 2 ) as a sparsity-promoting objective in compressed sensing. We first propose a novel criterion that guarantees that an s-sparse signal is the local minimizer of the ℓ 1 / ℓ 2 objective; our criterion is interpretable and useful in practice. We also give the first uniform recovery condition using a geometric characterization of the null space of the measurement matrix, and show that this condition is satisfied for a class of random matrices. We also present analysis on the robustness of the procedure when noise pollutes data. Numerical experiments are provided that compare ℓ 1 / ℓ 2 with some other popular non-convex methods in compressed sensing. Finally, we propose a novel initialization approach to accelerate the numerical optimization procedure. We call this initialization approach support selection, and we demonstrate that it empirically improves the performance of existing ℓ 1 / ℓ 2 algorithms.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

A High-Throughput Solver for Marginalized Graph Kernels on GPU

Here, we present the design and optimization of a solver for efficient and high-throughput computation of the marginalized graph kernel on General Purpose GPUs. The graph kernel is computed using the conjugate gradient method to solve a generalized Laplacian of the tensor product between a pair of graphs. To cope with the large gap between the instruction throughput and the memory bandwidth of the GPUs, our solver forms the graph tensor product on-the-fly without storing it in memory. This is achieved by using threads in a warp cooperatively to stream the adjacency and edge label matrices of individual graphs by small square matrix blocks called tiles, which are then staged in registers and the shared memory for later reuse. Warps across a thread block can further share tiles via the shared memory to increase data reuse. We exploit the sparsity of the graphs hierarchically by storing only non-empty tiles using a coordinate format and nonzero elements within each tile using bitmaps. We propose a new partition-based reordering algorithm for aggregating nonzero elements of the graphs into fewer but denser tiles to further exploit sparsity. We carry out extensive theoretical analyses on the graph tensor product primitives for tiles of various density and evaluate their performance on synthetic and real-world datasets. Our solver delivers three to four orders of magnitude speedup over existing CPU-based solvers such as GraKeL and GraphKernels. The capability of the solver enables kernel-based learning tasks at unprecedented scales.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Link Scheduling in Satellite Networks via Machine Learning Over Riemannian Manifolds

Low Earth Orbit (LEO) satellites play a crucial role in enhancing global connectivity, serving a complementary solution to existing terrestrial systems. In wireless networks, scheduling is a vital process that allocates time-frequency resources to users for interference management. However, LEO satellite networks face significant challenges in scheduling their links towards ground users due to the satellites’ mobility and overlapping coverage. This paper addresses the dynamic link scheduling problem in LEO satellite networks by considering spatio-temporal correlations introduced by the satellites’ movements. The first step in the proposed solution involves modeling the network over Riemannian manifolds, thanks to their representation as symmetric positive definite matrices. We introduce two machine learning (ML)-based link scheduling techniques that model the dynamic evolution of satellite positions and link conditions over time and space. To accurately predict satellite link states, we present a recurrent neural network (RNN) over Riemannian manifolds, which captures spatio-temporal characteristics over time. Furthermore, we introduce a separate model, the convolutional neural network (CNN) over Riemannian manifolds, which captures geometric relationships between satellites and users by extracting spatial features from the network topology across all links. Simulation results demonstrate that both RNN and CNN over Riemannian manifolds deliver comparable performance to the fractional programming-based link scheduling (FPLinQ) benchmark. Remarkably, unlike other ML-based models that require extensive training data, both models only need 30 training samples to achieve over 99% of the sum rate while maintaining similar computational complexity relative to the benchmark.

42 ENGINEERING↗