Search NASA⌕ Search

SEARCH · Search NASA

Results for “FACTORIZATION”

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 559 records · Page 31

Combining Sparse Approximate Factorizations with Mixed-precision Iterative Refinement

The standard LU factorization-based solution process for linear systems can be enhanced in speed or accuracy by employing mixed-precision iterative refinement. Most recent work has focused on dense systems. We investigate the potential of mixed-precision iterative refinement to enhance methods for sparse systems based on approximate sparse factorizations. In doing so, we first develop a new error analysis for LU- and GMRES-based iterative refinement under a general model of LU factorization that accounts for the approximation methods typically used by modern sparse solvers, such as low-rank approximations or relaxed pivoting strategies. We then provide a detailed performance analysis of both the execution time and memory consumption of different algorithms, based on a selected set of iterative refinement variants and approximate sparse factorizations. Our performance study uses the multifrontal solver MUMPS, which can exploit block low-rank factorization and static pivoting. We evaluate the performance of the algorithms on large, sparse problems coming from a variety of real-life and industrial applications showing that mixed-precision iterative refinement combined with approximate sparse factorization can lead to considerable reductions of both the time and memory consumption.

97 MATHEMATICS AND COMPUTING↗

The Prediction Model of Risk Factors for COVID-19 Developing into Severe Illness Based on 1046 Patients with COVID-19

This study analyzed the risk factors for patients with COVID-19 developing severe illnesses and explored the value of applying the logistic model combined with ROC curve analysis to predict the risk of severe illnesses at COVID-19 patients’ admissions. The clinical data of 1046 COVID-19 patients admitted to a designated hospital in a certain city from July to September 2020 were retrospectively analyzed, the clinical characteristics of the patients were collected, and a multivariate unconditional logistic regression analysis was used to determine the risk factors for severe illnesses in COVID-19 patients during hospitalization. Based on the analysis results, a prediction model for severe conditions and the ROC curve were constructed, and the predictive value of the model was assessed. Logistic regression analysis showed that age (OR = 3.257, 95% CI 10.466–18.584), complications with chronic obstructive pulmonary disease (OR = 7.337, 95% CI 0.227–87.021), cough (OR = 5517, 95% CI 0.258–65.024), and venous thrombosis (OR = 7322, 95% CI 0.278–95.020) were risk factors for COVID-19 patients developing severe conditions during hospitalization. When complications were not taken into consideration, COVID-19 patients’ ages, number of diseases, and underlying diseases were risk factors influencing the development of severe illnesses. The ROC curve analysis results showed that the AUC that predicted the severity of COVID-19 patients at admission was 0.943, the optimal threshold was −3.24, and the specificity was 0.824, while the sensitivity was 0.827. The changes in the condition of severe COVID-19 patients are related to many factors such as age, clinical symptoms, and underlying diseases. This study has a certain value in predicting COVID-19 patients that develop from mild to severe conditions, and this prediction model is a useful tool in the quick prediction of the changes in patients’ conditions and providing early intervention for those with risk factors.

Lian, Zhichuang↗

Systems analysis of multiple regulator perturbations allows discovery of virulence factors in Salmonella

Background: Systemic bacterial infections are highly regulated and complex processes that are orchestrated by numerous virulence factors. Genes that are coordinately controlled by the set of regulators required for systemic infection are potentially required for pathogenicity. Results: In this study we present a systems biology approach in which sample-matched multi-omic measurements of fourteen virulence-essential regulator mutants were coupled with computational network analysis to efficiently identify Salmonella virulence factors. Immunoblot experiments verified network-predicted virulence factors and a subset was determined to be secreted into the host cytoplasm, suggesting that they are virulence factors directly interacting with host cellular components. Two of these, SrfN and PagK2, were required for full mouse virulence and were shown to be translocated independent of either of the type III secretion systems in Salmonella or the type III injectisome-related flagellar mechanism. Conclusions: Integrating multi-omic datasets from Salmonella mutants lacking virulence regulators not only identified novel virulence factors but also defined a new class of translocated effectors involved in pathogenesis. The success of this strategy at discovery of known and novel virulence factors suggests that the approach may have applicability for other bacterial pathogens.

59 BASIC BIOLOGICAL SCIENCES↗

Calculation of Friction Factors in Heat Pipes Using CFD in Support of the Sockeye Code

A heat pipe is two-phase heat transfer device which relies on surface tension and capillary pressure to provide a very efficient heat transfer mechanism. Currently, there is interest in designing micro nuclear power reactors using high temperature heat pipes to provide totally passive cooling. To support this, Sockeye is being developed as a heat pipe simulation capability. To support the development of Sockeye, individual phenomena must be tested. This work focuses on applying a CFD model to evaluate friction factors for use in Sockeye. A CFD model has been developed in Nek5000 which simulates each component fluid, i.e. sodium liquid and vapor, independently and captures the effects of the wick on friction factor. The model tests the effect of the wick permeability and the thickness of the wick occupied by each fluid component. The thickness of the wick occupied by each component can be directly related to the local vapor volume fraction from the Sockeye heat pipe model. Results from the CFD model indicate that friction factor in the liquid flow decreases with increasing wick permeability, while the friction factor in the vapor flow increases with increasing wick permeability. In the liquid, this is likely a direct result of the increased cross sectional area available for the flow. For the vapor, this is explained as the wick acting like a rough wall, while the increased area has only a negligible impact. Results for the tests of the thickness of the wick occupied by each fluid showed a thresholding behavior for both liquid and vapor. When the thickness of the wick occupied by each fluid was increased, it increased the effect of the wick, but only up to a certain point. Once a certain thickness was reached, the friction factors remained constant. When the fluid occupied only a very small fraction of the wick, friction factors returned to nominal values.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Update to the Finite Cloud Dose Correction Factors in MACCS

In WASH - 1400, external exposure from the finite radioactive cloud (cloudshine) is calculated by assuming that the cloud is semi-infinite, the concentration of radioactive material is uniform, and by using a correction factor to account for these approximations. This correction factor is originally based upon formulations by Healy and depends on the effective size of the plume and the distance from the plume center to the receptor. The range of the finite cloud dose correction factor table from WASH - 1400 developed using Healy formulations can be exceeded in certain situations. When the range of the table is exceeded, no extrapolation is performed; rather interpolation at the edge of the table is performed per WASH - 1400. The tabulated values of these finite cloud dose correction factors from WASH - 1400 and the interpolation at the edge of the table have been used in MACCS since its creation. An expanded table of finite cloud dose correction factors is one way to reduce the need of using interpolation at the edge of the table. The generation of an expanded finite cloud dose correction factor table for future use in MACCS is documented in this report.

54 ENVIRONMENTAL SCIENCES↗

FGR 13 Dose Conversion Factor Files

The purpose of this document is to discuss the construction of two MACCS dose conversion factor (DCF) files in some detail, an older file created in 2007 named FGR13DCF.inp and a newer file created in 2018 called FGR13GyEquiv_RevA.inp. Very briefly, the difference between the two files is that the older file follows the standard conventions of assigning a radiation weighting factor of 20 for alpha radiation for all tissues and organs; whereas, the newer file complies with the FGR 13 health effects modeling and uses modified radiation weighting factors (referred to as relative biological effectiveness (RBE) factors) for alpha radiation of 10 for breast and of 1 for red bone marrow. During an intermediate period the creation of these two DCF files, a file called FGR13GyEquiv was created and used for the SOARCA calculations. This file was not released to the MACCS user community, but it is also discussed briefly in this document. DCF files are used by MACCS to convert air and ground radionuclide concentrations to doses to an organ or to the whole body. The MACCS calculation considers duration and shielding factor for each exposure pathway including cloudshine, groundshine, inhalation, and ingestion. Dose coefficients for cloudshine and groundshine are expressed as dose rates; dose coefficients for inhalation and ingestion are expressed as committed doses. In this document, dose coefficients (newer ICRP terminology) is used to describe the values contained in dose conversion factor (older ICRP terminology) files.

63 RADIATION, THERMAL, AND OTHER ENVIRON. POLLUTAN↗

Impact of K-factor on Short Circuit Program Convergence for Inverter-Based Resources during Faults

Classical short-circuit programs that linearize the power network are no longer applicable for inverter based resources (IBRs), necessitating an iterative approach. Phasor domain programs can model the IBRs using an iterative approach considering nonlinear fault responses. In phasor domain models, the IBR can be modeled as a voltage controlled current source (VCCS) in tabular form with positive, negative, and zero sequence information for balanced and unbalance short-circuit faults. In the VCCS modeling of the IBR, positive and negative incremental reactive current, also known as the k-factor, plays an important role in short circuit program convergence. In this work, a few approaches: conventional VCCS modeling with a k-factor of 2, conventional VCCS modeling with a k-factor of 2 with modified pre-fault voltages, VCCS characteristics based on the power flow solution with a k-factor of 2, and VCCS characteristics based on the power flow solution with a k-factor of 1 are investigated for short circuit program convergence under higher IBR penetration. The IEEE 39 bus New England Test System is taken as the test system, and simulations are carried out in PSS®CAPE 15.0.26 simulation software. Simulation results demonstrate that IBR penetration is higher for the VCCS model, which corresponds to the power flow solution with k-factor 1, compared to other approaches.

42 ENGINEERING↗

SAGAbg. I. A Near-unity Mass-loading Factor in Low-mass Galaxies via Their Low-redshift Evolution in Stellar Mass, Oxygen Abundance, and Star Formation Rate

Measuring the relation between star formation and galactic winds is observationally difficult. In this work we make an indirect measurement of the mass-loading factor (the ratio between the mass outflow rate and star formation rate) in low-mass galaxies using a differential approach to modeling the low-redshift evolution of the star-forming main sequence and mass–metallicity relation. We use Satellites Around Galactic Analogs (SAGA) background galaxies, i.e., spectra observed by the SAGA Survey that are not associated with the main SAGA host galaxies, to construct a sample of 11,925 spectroscopically confirmed low-mass galaxies from 0.01 ≲ z ≤ 0.21 and measure auroral line metallicities for 120 galaxies. The crux of the method is to use the lowest-redshift galaxies as the boundary condition of our model, and to infer a mass-loading factor for the sample by comparing the expected evolution of the low-redshift reference sample in stellar mass, gas-phase metallicity, and star formation rate against the observed properties of the sample at higher redshift. We infer a mass-loading factor of η m = $0.92^{+1.76}_{-0.74}$, which is in line with direct measurements of the mass-loading factor from the literature despite the drastically different sets of assumptions needed for each approach. While our estimate of the mass-loading factor is in good agreement with recent galaxy simulations that focus on resolving the dynamics of the interstellar medium, it is smaller by over an order of magnitude than the mass-loading factor produced by many contemporary cosmological simulations.

79 ASTRONOMY AND ASTROPHYSICS↗

A Framework for Batched and GPU-Resident Factorization Algorithms Applied to Block Householder Transformations

As modern hardware keeps evolving, an increasingly effective approach to developing energy efficient and high-performance solvers is to design them to work on many small size and independent problems. Many applications already need this functionality, especially for GPUs, which are currently known to be about four to five times more energy efficient than multicore CPUs. We describe the development of one-sided factorizations that work for a set of small dense matrices in parallel, and we illustrate our techniques on the QR factorization based on Householder transformations. We refer to this mode of operation as a batched factorization. Our approach is based on representing the algorithms as a sequence of batched BLAS routines for GPU-only execution. This is in contrast to the hybrid CPU-GPU algorithms that rely heavily on using the multicore CPU for specific parts of the workload. But for a system to benefit fully from the GPU's significantly higher energy efficiency, avoiding the use of the multicore CPU must be a primary design goal, so the system can rely more heavily on the more efficient GPU. Additionally, this will result in the removal of the costly CPU-to-GPU communication. Furthermore, we do not use a single symmetric multiprocessor(on the GPU) to factorize a single problem at a time. We illustrate how our performance analysis, and the use of profiling and tracing tools, guided the development and optimization of our batched factorization to achieve up to a 2-fold speedup and a 3-fold energy efficiency improvement compared to our highly optimized batched CPU implementations based on the MKL library(when using two sockets of Intel Sandy Bridge CPUs). Compared to a batched QR factorization featured in the CUBLAS library for GPUs, we achieved up to 5x speedup on the K40 GPU.

Dong, Tingzing Tim↗

Periodic Verification of the Scaling Factor for Radwastes in Korean NPPs - 13294

According to the acceptance criteria for a low and intermediate level radioactive waste (LILW) listed in Notice No. 2012-53 of the Nuclear Safety and Security Commission (NSSC), specific concentrations of radionuclides inside a drum has to be identified and quantified. In 5 years of effort, scaling factors were derived through destructive radiochemical analysis, and the dry active waste, spent resin, concentration bottom, spent filter, and sludge drums generated during 2004 ∼ 2008 were evaluated to identify radionuclide inventories. Eventually, only dry active waste among LILWs generated from Korean NPPs were first shipped to a permanent disposal facility on December 2010. For the LILWs generated after 2009, the radionuclides are being radiochemically quantified because the Notice clarifies that the certifications of the scaling factors should be verified biennially. During the operation of NPP, the radionuclides designated in the Notice are formed by neutron activation of primary coolant, reactor structural materials, corrosion products, and fission products released into primary coolant through defects or failures in fuel cladding. Eventually, since the radionuclides released into primary coolant are transported into the numerous auxiliary and support systems connected to primary system, the LILWs can be contaminated, and the radionuclides can have various concentration distributions. Thus, radioactive wastes, such as spent resin and dry active waste generated at various Korean NPP sites, were sampled at each site, and the activities of the regulated radionuclides present in the sample were determined using radiochemical methods. The scaling factors were driven on the basis of the activity ratios between a or β-emitting nuclides and γ-emitting nuclides. The resulting concentrations were directly compared with the established scaling factors' data using statistical methods. In conclusions, the established scaling factors were verified with a reliability of within 2σ, and the scaling factors will be applied for newly analyzed LILWs to evaluate the radionuclide inventories. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Development of Concentration Ratios and Transfer Factors for Human and Ecological Probabilistic Performance Assessments - 19781

Human and ecological risk assessments utilize concentration ratios and transfer factors to calculate contaminant concentrations in plant and animal tissues in equilibrium with soil or water (concentration ratios) or based on an animal's contaminant intake from various media (transfer factors). These ratios and factors are used to evaluate human dose via food ingestion pathways and to evaluate ecological dose to an organism. In a probabilistic performance assessment, concentration ratio and transfer factor distributions are developed for every element, and these distributions are applied across the spatial and temporal domain of the assessment. As a result, it is necessary to scale the underlying data distributions by applying the central limit theorem to develop appropriate distributions of the mean for the temporal and spatial scale of the model. Herein we discuss the methodologies for identifying and ranking references, and how to develop statistical distributions of concentration ratios and transfer factors for probabilistic performance assessment models. We also discuss factors that could be pertinent to differences when selecting references and developing distributions for human versus ecological risk assessment scenarios. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

A Measurement of the Neutron Electromagnetic Form Factor Ratio from a Rosenbluth Technique with Simultaneous Detection of Neutrons and Protons

The internal structure of protons and neutrons provides insight into both the dynamical behavior of the constitute quarks and gluons, and emergent properties of the nucleons (such as mass, spin, and electromagnetic distributions). Elastic electron-nucleon scattering can probe the elastic electromagnetic form factors of the nucleon. The electric and magnetic form factors, respectively, encode information about the internal charge and magnetization distributions within the nucleon. Precision data for these form factors, over a broad range of the four-momentum transfer squared, Q^2, can benchmark theoretical models describing the strong interaction of nuclear physics. The Super BigBite Spectrometer (SBS) program in Hall A at Jefferson Lab, is a series of high-precision experiments which seek to significantly extend the Q^2 reach of previous data for the nucleon electromagnetic form factors. The first two experiments of this program are known as G_M^n and the neutron Two Photon Exchange (nTPE) and the data were collected from October 2021 to February 2022. Both experiments were conducted with the simultaneous measurement of D(e,e'n) and D(e,e'p) reactions for quasi-elastic electron-deuteron scattering. The scattered electrons were detected in the BigBite Spectrometer, which features multiple large-acceptance Gas Electron Multiplier (GEM) detectors. The Super BigBite Spectrometer provided simultaneous detection of scattered nucleons, and utilized a large acceptance dipole magnet and Hadron Calorimeter (HCal). The G_M^n experiment provides precision measurements of the neutron magnetic form factor, via the ratio method, over a Q^2 range of 3.0 to 13.5 (GeV/c)2. From this data analysis, preliminary values for G_M^n/µ_n G_D are extracted. For Q^2=4.48 (GeV/c)2 we find G_M^n/µ_n G_D=0.9546±0.0132 and for Q^2=4.476 (GeV/c)2 we find G_M^n/µ_n G_D=0.9563±0.0110. These preliminary G_M^n/µ_n G_D values are more precise than existing world data in this Q^2 regime and are consistent with the most recent parameterization of the G_M^n/µ_n G_D world data. The nTPE experiment provides a first measurement of the neutron Rosenbluth Slope and seeks to quantify the two-photon exchange(TPE) contribution to elastic electron-neutron scattering at a fixed Q^2=4.5 (GeV/c)2 with two different beam energies and scattering angle values. For data of the proton form factor ratio, µ_p G_E^p/G_M^p, significant discrepancies exist between values obtained from Rosenbluth Separation and polarization transfer measurement, particularly at large Q^2, and TPE contributions are thought to resolve this discrepancy. The impacts of TPE contributions have not yet been experimentally established for the neutron. From the data analysis presented in this dissertation, a preliminary result for the neutron Rosenbluth Slope is found as S^n=(G_E^n )^2/t_n (G_M^n )^2=0.0916±0.0476 for Q^2=4.48 (GeV/c)2. This value of the neutron Rosenbluth Slope is consistent with the world data extrapolation and the absence of large TPE corrections.

Wertz, Ezekiel [Thomas Jefferson National Accelera↗

Understanding and Modeling Pooled Rideshare Acceptance: Influential Factors, Preferred User Experiences, and Implications

This dissertation explores factors influencing pooled rideshare (PR) adoption to provide actionable insights for transportation network companies (TNCs) and policymakers. PR allows travelers to share rides with unknown passengers, offering benefits such as cost reduction and congestion relief. However, adoption remains limited due to safety concerns, privacy issues, and trust in rideshare platforms. A national U.S. survey with 5,385 respondents examined transportation preferences and barriers to PR adoption. Exploratory and confirmatory factor analyses identified five key factors influencing PR consideration—safety, service experience, privacy, traffic/environment, and time/cost. Second factor analyses examined ways to optimize PR experiences, revealing four factors—comfort/ease of use, convenience, vehicle technology/accessibility, and passenger safety. Privacy concerns, for instance, using regression analysis, were found to reduce the likelihood of PR adoption by 77%, and convenience had the potential to increase it by 156%. The Pooled Rideshare Acceptance Model (PRAM), based on the Technology Acceptance Model, assessed the impact of these factors using the Structural Equation Model (SEM). Privacy, safety, trust, and convenience had a large effect (Cohen's f2 > 0.35) on PR acceptance, while multigroup analyses (PRAMMA) explored 16 demographic variables such as gender, generation, and income, emphasizing the need for tailored strategies. Based on all the statistical analysis and workshops using descriptive statistics, 95 actionable recommendations were made from the riders' perspective. Findings highlight the importance of customized services, user experience improvements, and policy interventions to enhance PR adoption. This dissertation provides a roadmap for future research and policy development, ensuring evidence-based, practical strategies to improve PR services in the U.S. and beyond.

Gangadharaiah, Rakesh↗

Insulin-like growth factor-1 suppresses the Myostatin signaling pathway during myogenic differentiation

Myogenic differentiation is a complex and well-coordinated process for generating mature skeletal muscle fibers. This event is autocrine/paracrine regulated by growth factors, principally Myostatin (MSTN) and Insulin-like Growth Factor-1 (IGF-1). Myostatin, a member of the transforming growth factor-β superfamily, is a negative regulator of skeletal muscle growth in vertebrates that exerts its inhibitory function by activating Smad transcription factors. In contrast, IGF-1 promotes the differentiation of skeletal myoblasts by activating the PI3K/Akt signaling pathway. This study reports on a novel functional crosstalk between the IGF-1 and MSTN signaling pathways, as mediated through interaction between PI3K/Akt and Smad3. Stimulation of skeletal myoblasts with MSTN resulted in a transient increase in the pSmad3:Smad3 ratio and Smad-dependent transcription. Moreover, MSTN inhibited myod gene expression and myoblast fusion in an Activin receptor-like kinase/Smad3-dependent manner. Preincubation of skeletal myoblasts with IGF-1 blocked MSTN-induced Smad3 activation, promoting myod expression and myoblast differentiation. This inhibitory effect of IGF-1 on the MSTN signaling pathway was dependent on IGF-1 receptor, PI3K, and Akt activities. Finally, immunoprecipitation assay analysis determined that IGF-1 pretreatment increased Akt and Smad3 interaction. These results demonstrate that the IGF-1/PI3K/Akt pathway may inhibit MSTN signaling during myoblast differentiation, providing new insight to existing knowledge on the complex crosstalk between both growth factors. - Highlights: • IGF-1 inhibits Myostatin canonical signaling pathway through IGF-1R/PI3K/Akt pathway. • IGF-1 promotes myoblast differentiation through a direct blocking of Myostatin signaling pathway. • IGF-1 induces the interaction of Akt with Smad3 in skeletal myoblast.

60 APPLIED LIFE SCIENCES↗

Insulin-like growth factor-binding protein-3 inhibits IGF-1-induced proliferation of human hepatocellular carcinoma cells by controlling bFGF and PDGF autocrine/paracrine loops

Basic fibroblast growth factor (bFGF) and platelet-derived growth factor (PDGF) produced by hepatocellular carcinoma (HCC) cells are responsible for the growth of HCC cells. Accumulating evidence shows that insulin-like growth factor-binding protein-3 (IGFBP-3) suppresses HCC cell proliferation in both IGF-dependent and independent manners. It's unknown, however, whether treatment with exogenous IGFBP-3 inhibits bFGF and PDGF production in HCC cells. The present study demonstrates that IGFBP-3 suppressed IGF-1-induced bFGF and PDGF expression while it does not affect their expression in the absence of IGF-1. To delineate the underlying mechanism, western-blot and RT-PCR assays confirmed that the transcription factor early growth response protein 1 (EGR1) is involved in IGFBP-3 regulation of bFGF and PDGF. IGFBP-3 inhibition of type 1 insulin-like growth factor receptor (IGF1R), ERK and AKT activation is IGF-1-dependent. Furthermore, transient transfection with constitutively activated AKT or MEK partially blocks the IGFBP-3 inhibition of EGR1, bFGF and PDGF expression. In conclusion, these findings suggest that IGFBP-3 suppresses transcription of EGR1 and its target genes bFGF and PDGF through inhibiting IGF-1-dependent ERK and AKT activation. It demonstrates the importance of IGFBP-3 in the regulation of HCC cell proliferation, suggesting that IGFBP-3 could be a target for the treatment of HCC. - Highlights: • IGFBP-3 plays an inhibition role in IGF1-induced HCC cell growth. • IGFBP-3 inhibits bFGF and PDGF production in the IGF-dependent manner. • EGR1 is involved in IGFBP-3 regulation of bFGF and PDGF in HCC cells. • IGFBP-3 suppresses EGR1 and its target genes bFGF and PDGF through inhibiting IGF-1-dependent ERK and AKT activation.

60 APPLIED LIFE SCIENCES↗

Gravitational and higher-order form factors of the pion in chiral quark models

The gravitational form factor of the pion is evaluated in two chiral quark models and confronted with the recent full-QCD lattice data. We find good agreement for the case of the spectral quark model, which builds in the vector-meson dominance for the charge form factor. We derive a simple relation between the gravitational and electromagnetic form factors, holding in the considered quark models in the chiral limit. The relation implies that the gravitational mean squared radius is half the electromagnetic one. We also analyze higher-order quark generalized form factors of the pion, related to higher moments in the symmetric Bjorken X variable of the generalized parton distribution functions, and discuss their perturbative QCD evolution, which is needed to relate the quark-model predictions to the lattice data. The values of the higher-order quark form factors at t=0, computed on the lattice, also agree with our quark-model results within the statistical and method uncertainties.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Quantitative Analysis of the Effects of Natural and Human Factors on a Hydrological System in Zhangweinan Canal Basin

Drought is a water deficit state caused by large-scale climate change that cannot be avoided by a water resource management system. Water scarcity refers to the unsustainable utilization of water resources over a long time, which is the result of water management policies. However, water deficits caused by drought and water scarcity often occur simultaneously and are indistinguishable. This study proposes a model-based simulation framework that can quantitatively distinguish natural factors (drought) from human factors (water scarcity) in a hydrological system. The simulation was applied to the Zhangweinan Canal Basin, based on the runoff sequences from 1950 to 2004. The results show that the runoff curve number, soil depth, soil available water, soil evaporation compensation coefficient, base runoff α coefficient and the maximum canopy interception have the highest sensitivity to runoff, and that the calibrated and validated SWAT model can effectively simulate the runoff process in the Zhangweinan Canal Basin and similar areas. Abrupt changes in human activities in 1975 and water scarcity led to the disappearance of the summer peak runoff period in both wet and dry years. Human factors are the main reason for the change in the hydrological system in the study area; the runoff loss caused by human factors is four times that caused by natural factors according to the proposed variable threshold. This study proposes a model-based simulation framework that can help water resource managers to distinguish the effects of drought and water scarcity in water-stressed areas and adjust management accordingly.

54 ENVIRONMENTAL SCIENCES↗

Dense LU Factorization on Multicore Supercomputer Nodes

Dense LU factorization is a prominent benchmark used to rank the performance of supercomputers. Many implementations, including the reference code HPL, use block-cyclic distributions of matrix blocks onto a two-dimensional process grid. The process grid dimensions drive a trade-off between communication and computation and are architecture- and implementation-sensitive. We show how the critical panel factorization steps can be made less communication-bound by overlapping asynchronous collectives for pivot identification and exchange with the computation of rank-k updates. By shifting this trade-off, a modified block-cyclic distribution can beneficially exploit more available parallelism on the critical path, and reduce panel factorization's memory hierarchy contention on now-ubiquitous multi-core architectures. The missed parallelism in traditional block-cyclic distributions arises because active panel factorization, triangular solves, and subsequent broadcasts are spread over single process columns or rows (respectively) of the process grid. Increasing one dimension of the process grid decreases the number of distinct processes in the other dimension. To increase parallelism in both dimensions, periodic 'rotation' is applied to the process grid to recover the row-parallelism lost by a tall process grid. During active panel factorization, rank-1 updates stream through memory with minimal reuse. In a column-major process grid, the performance of this access pattern degrades as too many streaming processors contend for access to memory. A block-cyclic mapping in the more popular row-major order does not encounter this problem, but consequently sacrifices node and network locality in the critical pivoting steps. We introduce 'striding' to vary between the two extremes of row- and column-major process grids. As a test-bed for further mapping experiments, we describe a dense LU implementation that allows a block distribution to be defined as a general function of block to processor. Other mappings can be tested with only small, local changes to the code.

99 GENERAL AND MISCELLANEOUS↗