Finding Inputs that Trigger Floating-Point Exceptions in GPUs via Bayesian Optimization
Explore the source record for details and available documents.
SEARCH · Engineering Papers
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.
Explore the source record for details and available documents.
Abstract Chilling temperatures (0–15°C) inhibit photosynthesis in most C 4 grasses, yet photosynthesis is chilling tolerant in the ‘Illinois’ clone of the C 4 grass Miscanthus x giganteus , a candidate cellulosic bioenergy crop. M. x giganteus is a hybrid between Miscanthus sacchariflorus and Miscanthus sinensis ; therefore chilling‐tolerant parent lines might produce hybrids superior to the current clone. Recently a collection of M. sacchariflorus from Siberia, the apparent low temperature limit of natural distribution , became available, which may be a source for chilling tolerance. The collection was screened for chilling tolerance of photosynthesis by measuring dark‐adapted maximum quantum yield of PSII photochemistry ( F v /F m ) on plants in the field in cool weather. Superior accessions were selected for further phenotyping: plants were grown at 25°C, transferred to 10°C (chilling) for 15 days, and returned to 25°C for 7 days (recovery). Two experiments assessed: (a) light‐saturated net photosynthetic rate ( A sat ) and operating quantum yield of PSII photochemistry (Φ PSII ), (b) response of net leaf CO 2 uptake ( A ) to intercellular [CO 2 ] ( c i ). Three accessions showed superior chilling tolerance: RU2012‐069 and RU2012‐114 achieved A sat up to double that of M. x giganteus prior to and during chilling, due to increased c i ‐ saturated photosynthesis ( V max ). RU2012‐069 and RU2012‐114 also maintained greater levels of Φ PSII during chilling, indicating reduced photodamage. Additionally, accession RU2012‐112 maintained a stable A sat throughout the 15‐day chilling period, while A sat continuously declined in other accessions; this suggests RU2012‐112 could outperform others in lengthy chilling periods. Plants were returned to 25°C after the chilling period; M. x giganteus showed the weakest recovery after 1 day, but a strong recovery after 1 week. This study has therefore identified important genetic resources for the synthesis of improved lines of M. x giganteus , which could facilitate the displacement of fossil fuels by cellulosic bioenergy.
Abstract To breed improved biomass cultivars of Miscanthus ×giganteus, it will be necessary to select the highest‐yielding and best‐adapted genotypes of its parental species, Miscanthus sinensis and Miscanthus sacchariflorus . We phenotyped a diverse clonally propagated panel of 569 M. sinensis and nine natural diploid M . × giganteus at one subtropical (Zhuji, China) and five temperate locations (Sapporo, Japan; Leamington, Ontario, Canada; Fort Collins, CO; Urbana, IL; and Chuncheon, Korea) for dry biomass yield and 14 yield‐component traits, in trials grown for 3 years. Notably, dry biomass yield of four Miscanthus accessions exceeded 80 Mg/ha in Zhuji, China, approaching the highest observed for any land plant. Additionally, six M. sinensis in Sapporo, Japan and one in Leamington, Canada also yielded more than the triploid M . × giganteus ‘1993‐1780’ control, with values exceeding 20 Mg/ha. Diploid M . × giganteus was the best‐yielding group at the northern sites. Genotype‐by‐environment interactions were modest among the five northern trial sites but large between Zhuji, and the northern sites. M. sinensis accessions typically yielded best at trial sites with latitudes similar to collection sites, although broad adaptation was observed for accessions from southern Japan. Genotypic heritabilities for third year yields ranged from 0.71 to 0.88 within locations. Compressed circumference was the best predictor of yield. These results establish a baseline of data for initiating selection to improve biomass yield of M. sinensis and M . × giganteus in a diverse set of relevant geographies.
We investigated the roles of covalent bonding, separation of surface oxygen, and electrolyte pH on the oxygen evolution reaction (OER) on transition metal oxides by comparing catalytic onset potentials and activities of CaCoO 3 and SrCoO 3 . Both cubic, metallic perovskites have similar Co IV intermediate spin states and onset potentials, but a substantially smaller lattice parameter and shorter surface oxygen separation make CaCoO 3 a more stable catalyst with increased OER activity. The onset potentials are similar, occurring where H + is removed from surface -OH - , but two competing surface reactions determine the catalytic activity. In one, the surface -O - is attacked by electrolyte OH - to form the surface -OOH - ; in the other, two -O - form a surface peroxide ion and an oxygen vacancy with electrolyte OH - attacking the oxygen vacancy. The second pathway can be faster if the surface oxygen separation is smaller.
An accurate record of preindustrial (pre-1900 CE) sea level is necessary to contextualize modern global mean sea level (GMSL) rise with respect to natural variability. Precisely dated phreatic overgrowths on speleothems (POS) provide detailed rates of Late Holocene sea-level rise in Mallorca. Statistical analysis indicates that sea level rose locally by 0.12 to 0.31 m (95% confidence) from 3.26 to 2.84 thousand years (ka) ago (2σ) and remained within 0.08 m (95% confidence) of preindustrial levels from 2.84 ka to 1900 CE. This sea-level history is consistent with glacial isostatic adjustment models adopting relatively weak upper mantle viscosities of ~10 20 Pa s. There is virtual certainty (>0.999 probability) that the average GMSL rise since 1900 CE has exceeded even the high average rate of sea-level rise between 3.26 and 2.84 ka inferred from the POS record. We conclude that modern GMSL rise is anomalous relative to any natural variability in ice volumes over the past 4000 years.
Not Available
We report that similar to conventional materials, most multicomponent high-entropy alloys (HEAs) lose ductility as they gain strength. In this study, we controllably introduced gradient nanoscaled dislocation cell structures in a stable single-phase HEA with face-centered cubic structure, thus resulting in enhanced strength without apparent loss of ductility. Upon application of strain, the sample-level structural gradient induces progressive formation of a high density of tiny stacking faults (SFs) and twins, nucleating from abundant low-angle dislocation cells. Furthermore, the SF-induced plasticity and the resultant refined structures, coupled with intensively accumulated dislocations, contribute to plasticity, increased strength, and work hardening. These findings offer a promising paradigm for tailoring properties with gradient dislocation cells at the nanoscale and advance our fundamental understanding of the intrinsic deformation behavior of HEAs.
Single-phase body-centered cubic (bcc) refractory medium- or high-entropy alloys can retain compressive strength at elevated temperatures but suffer from extremely low tensile ductility and fracture toughness. We examined the strength and fracture toughness of a bcc refractory alloy, NbTaTiHf, from 77 to 1473 kelvin. This alloy’s behavior differed from that of comparable systems by having fracture toughness over 253 MPa·m 1/2 , which we attribute to a dynamic competition between screw and edge dislocations in controlling the plasticity at a crack tip. Whereas the glide and intersection of screw and mixed dislocations promotes strain hardening controlling uniform deformation, the coordinated slip of <111> edge dislocations with {110} and {112} glide planes prolongs nonuniform strain through formation of kink bands. These bands suppress strain hardening by reorienting microscale bands of the crystal along directions of higher resolved shear stress and continually nucleate to accommodate localized strain and distribute damage away from a crack tip.
The epitopes defined by HIV-1 broadly neutralizing antibodies (bNAbs) are valuable templates for vaccine design, and studies of the immunological development of these antibodies are providing insights for vaccination strategies. In addition, the most potent and broadly reactive of these bNAbs have potential for clinical use. We previously described a family of 12 V1V2-directed neutralizing antibodies, CAP256-VRC26, isolated from an HIV-1 clade C-infected donor at years 1, 2, and 4 of infection (N. A. Doria-Rose et al., Nature 509:55–62, 2014, http://dx.doi.org/10.1038/nature13036). Here, we report on the isolation and characterization of new members of the family mostly obtained at time points of peak serum neutralization breadth and potency. Thirteen antibodies were isolated from B cell culture, and eight were isolated using trimeric envelope probes for differential single B cell sorting. One of the new antibodies displayed a 10-fold greater neutralization potency than previously published lineage members. This antibody, CAP256-VRC26.25, neutralized 57% of diverse clade viral isolates and 70% of clade C isolates with remarkable potency. Among the viruses neutralized, the median 50% inhibitory concentration was 0.001 μg/ml. All 33 lineage members targeted a quaternary epitope focused on V2. While all known bNAbs targeting the V1V2 region interact with the N160 glycan, the CAP256-VRC26 antibodies showed an inverse correlation of neutralization potency with dependence on this glycan. Overall, our results highlight the ongoing evolution within a single antibody lineage and describe more potent and broadly neutralizing members with potential clinical utility, particularly in areas where clade C is prevalent. Studies of HIV-1 broadly neutralizing antibodies (bNAbs) provide valuable information for vaccine design, and the most potent and broadly reactive of these bNAbs have potential for clinical use. We previously described a family of V1V2-directed neutralizing antibodies from an HIV-1 clade C-infected donor. Here, we report on the isolation and characterization of new members of the family mostly obtained at time points of peak serum neutralization breadth and potency. One of the new antibodies, CAP256-VRC26.25, displayed a 10-fold greater neutralization potency than previously described lineage members. It neutralized 57% of diverse clade viral isolates and 70% of clade C isolates with remarkable potency: the median 50% inhibitory concentration was 0.001 μg/ml. Furthermore, our results highlight the ongoing evolution within a single antibody lineage and describe more potent and broadly neutralizing members with potential clinical utility, particularly in areas where clade C is prevalent.
Explore the source record for details and available documents.
Abstract not provided
Future high-performance computing systems may face frequent failures with their rapid increase in scale and complexity. Resilience to faults has become a major challenge for large-scale applications running on supercomputers, which demands fault tolerance support for prevalent MPI applications. Among failure scenarios, process failures are one of the most severe issues as they usually lead to termination of applications. However, the widely used MPI implementations do not provide mechanisms for fault tolerance. We propose FTA-MPI (Fault Tolerance Assistant MPI), a programming model that provides support for failure detection, failure notification and recovery. Specifically, FTA-MPI exploits a try/catch model that enables failure localization and transparent recovery of process failures in MPI applications. We demonstrate FTA-MPI with synthetic applications and a molecular dynamics code CoMD, and show that FTA-MPI provides high programmability for users and enables convenient and flexible recovery of process failures.
Abstract not provided.
Thermonuclear X-ray bursts from accreting neutron stars power brief but strong irradiation of their surroundings, providing a unique way to study accretion physics. We analyze MAXI /Gas Slit Camera and Swift /XRT spectra of a day-long flash observed from IGR J17062-6143 in 2015. It is a rare case of recurring bursts at a low accretion luminosity of 0.15% Eddington. Spectra from MAXI , Chandra , and NuSTAR observations taken between the 2015 burst and the previous one in 2012 are used to determine the accretion column. We find it to be consistent with the burst ignition column of 5×10{sup 10} g cm{sup −2}, which indicates that it is likely powered by burning in a deep helium layer. The burst flux is observed for over a day, and decays as a straight power law: F ∝ t {sup −1.15}. The burst and persistent spectra are well described by thermal emission from the neutron star, Comptonization of this emission in a hot optically thin medium surrounding the star, and reflection off the photoionized accretion disk. At the burst peak, the Comptonized component disappears, when the burst may dissipate the Comptonizing gas, and it returns in the burst tail. The reflection signal suggests that the inner disk is truncated at ∼10{sup 2} gravitational radii before the burst, but may move closer to the star during the burst. At the end of the burst, the flux drops below the burst cooling trend for 2 days, before returning to the pre-burst level.
We present a z = 0.94 quasar, SDSS J004846.45-004611.9, discovered in the Sloan Digital Sky Survey III (SDSS-III) BOSS survey. A visual analysis of this spectrum reveals highly broadened and blueshifted narrow emission lines, in particular, [Ne v] λ3426 and [O III ] λ5007, with outflow velocities of 4000 km s -1 , along with unusually large [Ne v] λ3426/[Ne III ] λ3869 ratios. The gas shows higher ionization at higher outflow velocities, indicating a connection between the powerful outflow and the unusual strength of the high ionization lines. The spectral energy distribution and the i – W3 color of the source reveal that it is likely a core extremely red quasar (ERQ); a candidate population of young active galactic nuclei (AGN) that are violently blowing out gas and dust from their centers. The dominance of host galaxy light in its spectrum and its fortuitous position in the SDSS S82 region allows us to measure its star formation history and investigate variability for the first time in an ERQ. Our analysis indicates that SDSS J004846.45-004611.9 underwent a short-lived starburst phase 400 Myr ago and was subsequently quenched, possibly indicating a time lag between star formation quenching and the onset of AGN activity. We also find that the strong extinction can be uniquely attributed to the AGN and does not persist in the host galaxy, contradicting a scenario where the source has recently transitioned from being a dusty submillimeter galaxy. In our relatively shallow photometric data, the source does not appear to be variable at 0.24–2.4 μm in the rest frame, most likely due to the dominant contribution of host galaxy starlight at these wavelengths.