Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “boosting”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,369 records · Page 76

A New Pivoting and Iterative Text Detection Algorithm for Biomedical Images

There is interest to expand the reach of literature mining to include the analysis of biomedical images, which often contain a paper's key findings. Examples include recent studies that use Optical Character Recognition (OCR) to extract image text, which is used to boost biomedical image retrieval and classification. Such studies rely on the robust identification of text elements in biomedical images, which is a non-trivial task. In this work, we introduce a new text detection algorithm for biomedical images based on iterative projection histograms. We study the effectiveness of our algorithm by evaluating the performance on a set of manually labeled random biomedical images, and compare the performance against other state-of-the-art text detection algorithms. We demonstrate that our projection histogram-based text detection approach is well suited for text detection in biomedical images, and that the iterative application of the algorithm boosts performance to an F score of .60. We provide a C++ implementation of our algorithm freely available for academic use.

60 APPLIED LIFE SCIENCES↗

Modeling of Electrochemistry and Steam-Methane Reforming Performance for Simulating Pressurized Solid Oxide Fuel Cell Stacks

This paper examines the electrochemical and on-cell steam-methane reforming performance of the solid oxide fuel cell when subjected to pressurization. Pressurized operation boosts the Nernst potential and decreases the activation polarization, both of which serve to increase cell voltage and power while lowering the heat load and operating temperature. A model considering the activation polarization in both the fuel and air electrodes was adopted to address this effect on the electrochemical performance. Both the increase in methane conversion kinetics and the increase in equilibrium methane concentration, which are competing effects of pressurization on steam-methane reforming, are considered in a new rate expression. The models were then applied in simulations to preview how the distributions of reforming rate, temperature, and current density can potentially be changed within stacks operating at elevated pressure. A generic 10 cm counter-flow stack model was created and used for the simulations of pressurized operation. The predictions showed improved thermal and electrical performance with increased operating pressure. The average and maximum cell temperatures decreased by 3% while the cell voltage increased by 9% as the operating pressure was increased from 1 to 10 atmospheres.

electrochemistry↗

Mesoporous Textured Fe-N-C Electrocatalysts as Highly Efficient Cathodes for Proton Exchange Membrane Fuel Cells

A new platinum group metal (PGM)-free proton exchange membrane fuel cell (PEMFC) cathode catalyst materials, synthesized using the VariPore™ method by Pajarito Powder, LLC, are characterized for their structure and activity. The physico-chemical analysis of the iron-nitrogen-carbon (Fe-N-C) electrocatalysts show mesoporous carbon material effectively doped with iron and nitrogen. The materials have an average pore size of 7-8 nm and high specific surface area. The Fe-N-C catalysts exhibit good oxygen reduction reaction (ORR) activity in 0.5 M H2SO4 electrolyte with high half-wave potential and sustainable electrochemical stability over 10,000 repeated potential cycles with insignificant losses in their activities. As cathode catalysts in a PEMFC, the Fe-N-C materials deliver remarkably good fuel cell performance at low overpotential approaching that of the commercial Pt catalyst. The high ORR electrocatalytic activity of these Fe-N-C catalysts is credited to the synergy between nitrogen-moieties, specifically pyrrolic-N, pyridinic-N, and graphitic-N, and iron in addition to the high mesoporosity that facilitate an effective reaction path in boosting the electrocatalytic activity and stability.

DIRECT ENERGY CONVERSION↗

Stability and oxygen-storage characteristics of Al-substituted YBaCo{sub 4}O{sub 7+{delta}}

A series of Al-substituted YBa(Co{sub 1-x}Al{sub x}){sub 4}O{sub 7+{delta}} samples was synthesized and characterized with respect to the capability to store large amounts of oxygen at low temperatures (at 200-400 deg. C) and the phase decomposition upon heating under oxidizing conditions at higher temperatures (above 550 deg. C). It was revealed that increasing the Al-substitution level up to x{approx}0.10 boosts up the phase-decomposition temperature from {approx}550 to {approx}700 deg. C, while the unique oxygen absorption/desorption characteristics remain nearly the same as those of the pristine YBaCo{sub 4}O{sub 7+{delta}} phase. The maximum amount of excess oxygen absorbed by the Al-substituted YBa(Co{sub 1-x}Al{sub x}){sub 4}O{sub 7+{delta}} samples was determined to be as large as {delta}{approx}1.45 for x=0.10 (in 100 atm O{sub 2} at 320 deg. C). Isothermal annealing experiments carried out for the same x=0.10 phase at 300 deg. C revealed that it could be reversibly charged and discharged with 1.2 oxygen atoms per formula unit by switching the gas flow from N{sub 2} to O{sub 2} and vice versa. - Graphical abstract: A small amount of aluminum replacing cobalt in YBaCo{sub 4}O{sub 7+{delta}} works efficiently for boosting up the phase-decomposition temperature (T{sub D}), but yet retains the unique low-temperature oxygen-storage characteristics of the phase.

36 MATERIALS SCIENCE↗

Enhanced thermoelectric performance in zinc substituted p-type filled skutterudites CeFe{sub 4-x}Zn{sub x}Sb{sub 12}

In this study, Zn-substituted polycrystalline skutterudites CeFe{sub 4-x}Zn{sub x}Sb{sub 12} (x=0, 0.05, 0.1, 0.2, 0.3) were successfully prepared by a traditional melting-annealing method. The solubility of Zn in Fe site is {approx}1.2%, exceeding which trace amount of ZnSb phase can be detected in the XRD. This ZnSb impurity phase, with size of several hundred nanometers for the sample with x=0.2 but showing surprisingly small size of {approx}10 nm for the sample with x=0.3, selectively distributes on the grain boundaries. In particular, the introduction of Zn in Fe site effectively improves the Seebeck coefficient in a manner of enhancement in hole effective mass, but it has negligible influence on both electrical and thermal conductivities though the hole concentration is increased. Consequently the corresponding improvement in power factor leads to an improved thermoelectric figure of merit (ZT) of 0.9 at 800 K for the sample with x=0.1, which is {approx}15% higher than that of Zn-free sample. This study demonstrates a favorable effect of Zn iso-substitution and opens a new strategy to improve the thermoelectric properties of p-type Fe-based skutterudites beyond the sole phonon engineering. - Graphical abstract: (a)-(c) ZnSb nanoinclusions emerge when Zn exceeds its solubility limit. (d), (e) The introduction of Zn boosts the Seebeck coefficient and thus enhances the ZT value. Highlights: Black-Right-Pointing-Pointer Zn is successfully employed to substitute Fe atom for the first time. Black-Right-Pointing-Pointer ZnSb nanoinclusions emerge when Zn exceeds its solubility limit {approx}0.12. Black-Right-Pointing-Pointer The introduction of Zn boosts the Seebeck coefficient and enhances the ZT value.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Thermoelectric performance of nanostructured p-type Zr₀.₅Hf₀.₅Co₀.₄Rh₀.₆Sb 1–x Sn x half-Heusler alloys

Several compositions of the p-type half-Heusler alloys Zr₀.₅Hf₀.₅Co₀.₄Rh₀.₆Sb 1–x Sn x (0≤x≤0.4) were synthesized by mechanically alloying high purity elemental powders using hardened steel jars and balls on a high energy shaker mill. Powder X-ray diffraction (XRD) investigations of several aliquots taken after regularly spaced milling time suggested that single phase products with half-Heusler (HH) structure can be obtained after 10 h. However, XRD and transmission electron microscopy (TEM) studies of several specimens obtained from compacted polycrystalline powders of Z₀.₅Hf₀.₅Co₀.₄Rh₀.₆Sb 1–x Sn x alloys using a uniaxial hot press (HP) revealed the presence of CoSb inclusions with various sizes embedded inside the HH matrix. Hall effect, electrical conductivity, and thermopower data collected between 300 K and 775 K on several compositions suggested that electronic transport in the synthesized Zr₀.₅Hf₀.₅Co₀.₄Rh₀.₆Sb 1–x Sn x /CoSb composites strongly depends on the average size and/or mole fraction of the embedded CoSb inclusions rather than the fraction (x) of Sn substituting for Sb. Among the samples investigated, the nanocomposite with x=0.2, which contains nanometer-scale CoSb inclusions, showed the largest power factor (800 μW/K² m at 775 K) and the lowest lattice thermal conductivity (~2.2 W/m K at 775 K) leading to a six-fold enhancement in the figure of merit when compared to the Zr₀.₅Hf₀.₅Co₀.₄Rh₀.₆Sb₀.₉₉Sn₀.₀₁ bulk matrix. - Graphical abstract: CoSb nanoinclusions embedded into a p-type Zr₀.₅Hf₀.₅Co₀.₄Rh₀.₆Sb 1–x Sn x half-Heusler matrix simultaneously boost the thermopower and carrier mobility leading to a drastic enhancement of the power factor of the resulting bulk nanostructured materials. Highlights: • The phase composition of half-Heusler (HH) alloy is very sensitive to the synthesis method. • Mechanical alloying (MA) of elements results in bulk HH matrix with CoSb inclusions. • CoSb nanoinclusions simultaneously boost the thermopower and carrier mobility with the HH matrix. • The MA process reduces lattice thermal conductivity due to high density of grain boundaries.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Structural, optical, and ferromagnetic characterization of Sm-doped LaOCl nanocrystalline synthesized by solvothermal route: Significant effect of hydrogen post treatment

Pure and Sm-doped lanthanum oxychloride (LaOCl) nanomaterials were synthesized by solvothermal route followed by a subsequent heat treatment process. The objective of the present work is to study and develop conditions required to create stable room-temperature ferromagnetic (RT-FM) properties in LaOCl. To achieve that aim, magnetic samarium Sm{sup 3+} ions were used as dopant sources for stable FM properties. Systematic structural, optical, and magnetic properties of undoped and Sm-doped LaOCl samples were investigated as function of post-annealing conditions (temperature and atmosphere). The optical absorption properties were studied by diffuse reflection spectroscopy (DRS). The magnetic measurements reveal that Sm-doped LaOCl nanopowders have partial RT-FM properties due to the doped ions. The variations of magnetic properties with pre-annealing temperature were investigated. Furthermore, the electronic medium of host LaOCl crystalline lattice, which carries the spin-spin (S.S) exchange interaction between localised dopant Sm{sup 3+}(4f{sup 5}) spins, was developed by annealing in hydrogen gas (hydrogenation). It was established that annealing in hydrogen atmosphere boosts the RT-FM properties so that the saturation magnetisation could be increased by more than 100%. Physical explanations and discussions were given in this paper. Thus, it was proved that the magnetic properties could be tailored to diamagnetic LaOCl compound by Sm-doping and post treatment under H{sub 2} atmosphere. Therefore, LaOCl nanocrystals could be used as a potential candidate for optical phosphor applications with magnetic properties. - Graphical abstract: M-H dependence of Sm-doped LaOCl powders. Study the effect of hydrogenation. - Highlights: • Synthesis of Sm-doped LaOCl nanoparticles. • DM LaOCl transforms to FM with dilute concentration of Sm doping. • Annealing under H{sub 2} atmosphere induces drastic boost in the FM properties. • Saturation magnetization attained 29 memu/g with little doping and hydrogenation.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Dosimetric Impact of Surgical Clips in Electron Beam Treatment of Breast Cancer

Titanium clips are commonly used to delineate the location of the tumor bed during breast cancer surgery. Electron beams are frequently used to boost radiation dose to the breast cavity. This research investigated the effect on such boost treatments as a result of metallic clip perturbation potentially exhibited through attenuation and scatter processes using measurements with Gafchromic film and treatment planning simulation with both generalized Gaussian pencil beam and electron Monte Carlo algorithms. Results showed that the potential effect of clip interference is unidentifiable from both algorithms. Dosimetry with Gafchromic external beam therapy film did detect dose perturbations caused by the titanium clips in proximal plane, 0.23 mm away, resulting in 3.0% backscatter and 2.4% attenuation at 6 MeV and 2.0% backscatter and 6.7% attenuation at 9 MeV. The noise contribution of the film scanner is estimated to be about 0.4% and nearly 2% uncertainty in film calibration. As a result, we conclude that the magnitude of dose perturbations from clip is negligible for clinical findings.

61 RADIATION PROTECTION AND DOSIMETRY↗

Postoperative Intensity Modulated Radiation Therapy in High Risk Prostate Cancer: A Dosimetric Comparison

The aim of this study was to compare intensity-modulated radiation therapy (IMRT) with 3D conformal technique (3D-CRT), with respect to target coverage and irradiation of organs at risk for high dose postoperative radiotherapy (PORT) of the prostate fossa. 3D-CRT and IMRT treatment plans were compared with respect to dose to the rectum and bladder. The dosimetric comparison was carried out in 15 patients considering 2 different scenarios: (1) exclusive prostate fossa irradiation, and (2) pelvic node irradiation followed by a boost on the prostate fossa. In scenario (1), a 3D-CRT plan (box technique) and an IMRT plan were calculated and compared for each patient. In scenario (2), 3 treatment plans were calculated and compared for each patient: (a) 3D-CRT box technique for both pelvic (prophylactic nodal irradiation) and prostate fossa irradiation (3D-CRT only); (b) 3D-CRT box technique for pelvic irradiation followed by an IMRT boost to the prostatic fossa (hybrid 3D-CRT and IMRT); and (c) IMRT for both pelvic and prostate fossa irradiation (IMRT only). For exclusive prostate fossa irradiation, IMRT significantly reduced the dose to the rectum (lower Dmean, V50%, V75%, V90%, V100%, EUD, and NTCP) and the bladder (lower Dmean, V50%, V90%, EUD and NTCP). When prophylactic irradiation of the pelvis was also considered, plan C (IMRT only) performed better than plan B (hybrid 3D-CRT and IMRT) as respect to both rectum and bladder irradiation (reduction of Dmean, V50%, V75%, V90%, equivalent uniform dose [EUD], and normal tissue complication probability [NTCP]). Plan (b) (hybrid 3D-CRT and IMRT) performed better than plan (a) (3D-CRT only) with respect to dose to the rectum (lower Dmean, V75%, V90%, V100%, EUD, and NTCP) and the bladder (Dmean, EUD, and NTCP). Postoperative IMRT in prostate cancer significantly reduces rectum and bladder irradiation compared with 3D-CRT.

61 RADIATION PROTECTION AND DOSIMETRY↗

The use of intensity-modulated radiation therapy photon beams for improving the dose uniformity of electron beams shaped with MLC

Electrons are ideal for treating shallow tumors and sparing adjacent normal tissue. Conventionally, electron beams are collimated by cut-outs that are time-consuming to make and difficult to adapt to tumor shape throughout the course of treatment. We propose that electron cut-outs can be replaced using photon multileaf collimator (MLC). Two major problems of this approach are that the scattering of electrons causes penumbra widening because of a large air gap, and available commercial treatment planning systems (TPSs) do not support MLC-collimated electron beams. In this study, these difficulties were overcome by (1) modeling electron beams collimated by photon MLC for a commercial TPS, and (2) developing a technique to reduce electron beam penumbra by adding low-energy intensity-modulated radiation therapy (IMRT) photons (4 MV). We used blocks to simulate MLC shielding in the TPS. Inverse planning was used to optimize boost photon beams. This technique was applied to a parotid and a central nervous system (CNS) clinical case. Combined photon and electron plans were compared with conventional plans and verified using ion chamber, film, and a 2D diode array. Our studies showed that the beam penumbra for mixed beams with 90 cm source to surface distance (SSD) is comparable with electron applicators and cut-outs at 100 cm SSD. Our mixed-beam technique yielded more uniform dose to the planning target volume and lower doses to various organs at risk for both parotid and CNS clinical cases. The plans were verified with measurements, with more than 95% points passing the gamma criteria of 5% in dose difference and 5 mm for distance to agreement. In conclusion, the study has demonstrated the feasibility and potential advantage of using photon MLC to collimate electron beams with boost photon IMRT fields.

61 RADIATION PROTECTION AND DOSIMETRY↗

Simplified field-in-field technique for a large-scale implementation in breast radiation treatment

We wanted to evaluate a simplified 'field-in-field' technique (SFF) that was implemented in our department of Radiation Oncology for breast treatment. This study evaluated 15 consecutive patients treated with a simplified field in field technique after breast-conserving surgery for early-stage breast cancer. Radiotherapy consisted of whole-breast irradiation to the total dose of 50 Gy in 25 fractions, and a boost of 16 Gy in 8 fractions to the tumor bed. We compared dosimetric outcomes of SFF to state-of-the-art electronic surface compensation (ESC) with dynamic leaves. An analysis of early skin toxicity of a population of 15 patients was performed. The median volume receiving at least 95% of the prescribed dose was 763 mL (range, 347-1472) for SFF vs. 779 mL (range, 349-1494) for ESC. The median residual 107% isodose was 0.1 mL (range, 0-63) for SFF and 1.9 mL (range, 0-57) for ESC. Monitor units were on average 25% higher in ESC plans compared with SFF. No patient treated with SFF had acute side effects superior to grade 1-NCI scale. SFF created homogenous 3D dose distributions equivalent to electronic surface compensation with dynamic leaves. It allowed the integration of a forward planned concomitant tumor bed boost as an additional multileaf collimator subfield of the tangential fields. Compared with electronic surface compensation with dynamic leaves, shorter treatment times allowed better radiation protection to the patient. Low-grade acute toxicity evaluated weekly during treatment and 2 months after treatment completion justified the pursuit of this technique for all breast patients in our department.

61 RADIATION PROTECTION AND DOSIMETRY↗

A dosimetric evaluation of dose escalation for the radical treatment of locally advanced vulvar cancer by intensity-modulated radiation therapy

The purpose of this planning study was to determine whether intensity-modulated radiation therapy (IMRT) reduces the radiation dose to organs at risk (OAR) when compared with 3D conventional radiation therapy (3D-CRT) in patients with vulvar cancer treated by irradiation. This study also investigated the use of sequential IMRT boost (seq-IMRT) and simultaneous integrated boost (SIB-IMRT) for dose escalation in the treatment of locally advanced vulvar cancer. Five vulvar cancer patients treated in the postoperative setting and 5 patients treated with definitive intent (def-group) were evaluated. For the postoperative group, 3D-CRT and IMRT plans to a total dose (TD) of 45 Gy were generated. For the def-group, 4 plans were generated: a 3D-CRT and an IMRT plan to a TD of 56.4 Gy, a SIB-IMRT plan to a TD of 56 Gy, and a SIB-IMRT with dose escalation (SIB-IMRT-esc): TD of 67.2 Gy. Mean dose and dose-volume histograms were compared using Student's t-test. IMRT significantly (all p < 0.05) reduced the D{sub mean}, V30, and V40 for all OAR in the adjuvant setting. The V45 was also significantly reduced for all OAR except the bladder. For patients treated in the def-group, all IMRT techniques significantly reduced the D{sub mean}, V40, and V45 for all OAR. The mean femur doses with SIB-IMRT and SIB-IMRT-esc were 47% and 49% lower compared with 3D-CRT. SIB-IMRT-esc reduced the doses to the OAR compared with seq-3D-CRT but increased the D{sub max.} for the small bowel, rectum, and bladder. IMRT reduces the dose to the OAR compared with 3D-CRT in patients with vulvar cancer receiving irradiation to a volume covering the vulvar region and nodal areas without compromising the dosimetric coverage of the target volume. IMRT for vulvar cancer is feasible and an attractive option for dose escalation studies.

61 RADIATION PROTECTION AND DOSIMETRY↗

Chemical composition based machine learning model to predict defect formation in additive manufacturing

With a goal of exploiting additive manufacturing to improve the manufacturing of existing reactor materials, we developed a chemical composition-based machine learning model to predict the printability of any given alloy in laser powder bed fusion (L-PBF) using experimental data from peer-reviewed literature. We defined printability as the ability to avoid defects like cracking, balling, porosity, and lack of fusion, that are caused by thermal stresses (during solidification or liquation), molten pool disintegration into disconnected small beads or lack of heat input respectively. Our models predict the tendency of balling defect formation and porosity percentage for a given composition, under a given set of processing conditions. To predict the likelihood of balling defect, three models: a random forest classifier, a gradient boost regressor and a neural network were trained on a dataset containing both traditional alloys and high entropy alloys. The neural network model showed the highest accuracy of 92.3 % in predicting the balling defect formation. A random forest regressor, gradient boost regressor and neural network were trained and tested on a dataset of various alloys to predict porosity. The random forest regressor showed the best predictions with an R 2 score of 0.97. The models also revealed the relative importance of the input descriptors on defect-formation tendency. Of particular significance was the identification of carbon as an important element in determining the occurrence of balling and percent porosity in alloys like steel, as well as being moderately important to the percentage porosity in other alloys as well as steel. Manganese was also identified as a key descriptor for the percentage of porosity in steel and other alloys. Manganese’s low thermal conductivity and consistent presence in the dataset is the likely cause for its contribution. Carbon’s role is attributable to its relatively high specific heat and high melting temperature. In conclusion, our model serves as a swift, chemistry-based tool to design experiments and find modified compositions better suited for additive manufacturing.

36 MATERIALS SCIENCE↗

High rate and long cycle life in Li-O 2 batteries with highly efficient catalytic cathode configured with Co 3 O 4 nanoflower

The reaction mechanism of non-aqueous Li-O 2 batteries is based on the deposition and decomposition of Li 2 O 2 . The polarization of Li-O 2 batteries can be rapidly increased by operation under a high rate condition, resulting in the early capacity fade of the cells. Therefore, a well-designed catalyst with a unique structure and excellent catalytic ability is an important way to boost the round-trip performance of Li-O 2 batteries, especially under high current density. In this work, a unique nanoflower structure assembled with Co 3 O 4 nanosheets is synthesized by using 2-methylimidazole (2-MIM) as a structural directing agent. X-ray photoelectron spectroscopy (XPS) and Raman spectra reveal abundant oxygen vacancies on the surface of the Co 3 O 4 nanoflower, which are beneficial for oxygen reduction and evolution reactions and long round-trip lifetime. Density functional theory results demonstrate that Co 3 O 4 catalyst with oxygen vacancies could promote the wetting of Li 2 O 2 on substrate and formation of a Li 2 O 2 nanofilm, thereby boosting the discharge capacity of Li-O 2 batteries. On account of the synergistic effect of abundant oxygen vacancies, the unique structure, and excellent oxygen evolution reaction, Co 3 O 4 nanoflower-based cells could deliver ultralong lifetime of 276 and 248 cycles with a discharge capacity of 1000 mAh g(-1) under charge/discharge current densities of 0.5 A g -1 and 1 A g -1 , respectively. Finally, this study has shed light on a new strategy for catalyst preparation for long lifetime Li-O 2 batteries.

25 ENERGY STORAGE↗

A new method to measure the U-235 content in fresh LWR fuel assemblies via fast-neutron passive self-interrogation

The purpose of this paper is to provide a solution to a decades old safeguards problem in the verification of the fissile concentration in fresh light water reactor (LWR) fuel assemblies. The problem is that the burnable poison (e.g. Gd 2 O 3 ) addition to the fuel rods decreases the active neutron assay for the fuel assemblies. This paper presents a new innovative method for the verification of the 235 U linear mass density in fresh LEU fuel assemblies that is insensitive to the burnable poison content. The technique makes use of the 238 U atoms in the fuel rods to self-interrogate the 235 U mass. The innovation for the new approach is that the 238 U spontaneous fission (SF) neutrons from the rods induces fission reactions (IF) in the 235 U that are time correlated with the SF source neutrons. Thus, the coincidence gate counting rate benefits from both the nu-bar of the 238 U SF (2.07) and the 235 U IF (2.44) for a fraction of the IF reactions. Whereas, the 238 U SF background has no time-correlation boost. The higher the detection efficiency, the higher the correlated boost because background neutron counts from the SF are being converted to signal doubles. This time-correlation in the IF signal increases signal/background ratio that provides a good precision for the net signal from the 235 U mass. The hard neutron energy spectrum makes the technique insensitive to the burnable poison loading where a Cd or Gd liner on the detector walls is used to prevent thermal-neutron reflection back into the fuel assembly from the detector. Here, we have named the system the fast-neutron passive collar (FNPC).

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Novel data analysis method for obtaining better performance from a complex 3D-printed collimator

Additively manufactured scattered beam collimators are increasingly being employed to boost the sample to cell peak signal ratio in high pressure neutron diffraction studies because of manufacturing versatility and performance improvements. We study how the measured diffraction pattern is affected by the presence of a collimator downstream of the sample, and develop a novel protocol that provides more effective background rejection. This protocol takes into account critical performance-determinants that were identified in this study, namely: (i) effectively identifying the collimator pattern on the detector; (ii) understanding the dependence of this pattern on sample and cell composition; and (iii) accurately identifying and differentiating the different regions of the pattern on the detector based on the dependency of the cell or sample and finally (iv) resolving the intensities at regions of the detector where neutrons scattered from the sample are preferentially represented, in order to boost the sample to cell peak signal ratio. Application of this novel analysis protocol is shown to increase the collimator performance over the traditional method.

3D printing↗

Performance optimization, modeling and analysis of sparse matrix-matrix products on multi-core and many-core processors

Sparse matrix-matrix multiplication (SpGEMM) is a computational primitive that is widely used in areas ranging from traditional numerical applications to recent big data analysis and machine learning. Although many SpGEMM algorithms have been proposed, hardware specific optimizations for multi- and many-core processors are lacking and a detailed analysis of their performance under various use cases and matrices is not available. In this work, we firstly identify and mitigate multiple bottlenecks with memory management and thread scheduling on Intel Xeon Phi (Knights Landing or KNL). Specifically targeting many-core processors, we develop a hash-table-based algorithm and optimize a heap-based shared-memory SpGEMM algorithm. We examine their performance together with other publicly available codes. Different from the literature, our evaluation also includes use cases that are representative of real graph algorithms, such as multi-source breadth-first search or triangle counting. Our hash-table and heap-based algorithms are showing significant speedups from libraries in the majority of the cases while different algorithms dominate the other scenarios with different matrix size, sparsity, compression factor and operation type. We wrap up in-depth evaluation results and make a recipe to give the best SpGEMM algorithm for target scenario. We build the performance model for hash-table and heap-based algorithms, which supports the recipe. A critical finding is that hash-table-based SpGEMM gets a significant performance boost if the nonzeros are not required to be sorted within each row of the output matrix. Finally, we integrate our implementations into a large-scale protein clustering code named HipMCL, accelerating its SpGEMM kernel by up to 10X and achieving an overall performance boost for the whole HipMCL application by 2.6X.

97 MATHEMATICS AND COMPUTING↗