GLANCED-IO: Taming IO Optimization for Deep Learning at Scale
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.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
We discuss over 120 laboratory experiments pertaining to the identification of the new absorption band discovered by Trafton et al. (1991) at 4705.2 cm-1 (2.1253 micrometers) in the spectrum of Io. It is shown that this band is not due to overtones or combinations of the fundamental bands associated with the molecules (or their chemical complexes) already identified on Io, namely, SO2, H2S, and H2O. Thus, this band is due to a new, previously unidentified, component of Io. Experiments also demonstrate that the band is not due to molecular H2 frozen in SO2 frosts. Since the frequency of this band is very close to the first overtone of the nu 3 asymmetric stretching mode of CO2, we have investigated the spectral behavior of CO2 under a variety of conditions appropriate for Io. The profile of the Io band is not consistent with the rotational envelope expected for single, freely rotating, gaseous CO2 under Io-like conditions. It was found that pure, solid CO2 and CO2 intimately mixed in a matrix of solid SO2 and H2S produce bands with similar widths (5-10 cm-1), but that these bands consistently fall at frequencies about 10-20 cm-1 (approximately 0.007 micrometer) lower than the Io band. CO2 in SO2 : H2S ices also produces several additional bands that are not in the Io spectra. The spectral fit improves, however, as the CO2 concentration in SO2 increases, suggesting that CO2-CO2 interactions might be involved. A series of Ar : CO2 and Kr : CO2 matrix isolation experiments, as well as laboratory work done elsewhere, show that CO2 clustering shifts the band position to higher frequencies and provides a better fit to the Io band. Various laboratory experiments have shown that gaseous CO2 molecules have a propensity to cluster between 80 and 100 K, temperatures similar to those found on the colder regions of Io. We thus tentatively identify the newly discovered Io band at 4705.2 cm-1 (2.1253 micrometers) with CO2 multimers or "clusters" on Io. Whether these clusters are buried within an SO2 frost, reside on the surface, or are in a residual, steady-state "atmospheric aerosol" population over local coldtraps is not entirely clear, although we presently favor the latter possibility. The size of these clusters is not well defined, but evidence suggests groups of more than four molecules are required. The absorption strength of the 2 nu 3 CO2 cluster overtone determined in the laboratory, in conjunction with the observed strength of the Io band, suggests that the disk-integrated abundance of CO2 is less than 1% that of the SO2. Studies of the sublimation behavior of CO2 indicate that it probably resides predominantly in the cooler areas (< 100 K) of Io. The relative constancy of the Io feature over a variety of orbital phases suggests that the polar regions may contain much of the material. Some consequences of the physical properties of CO2 under conditions pertinent to Io are discussed. The presence of CO2 clusters on Io could be verified by the detection of any one of several other infrared bands associated with the CO2 molecule, of which the strongest are the nu 3 12CO2 asymmetric stretch fundamental near 2350 cm-1 (4.25 micrometers) and the nu 2 bending mode fundamental near 660 cm-1 (15.1 micrometers). Weaker bands that may also be detectable include the nu 3 13CO2 asymmetric stretch fundamental near 2280 cm-1 (4.39 micrometers), the 2 nu 2 + nu 3 combination/overtone band near 3600 cm-1 (2.78 micrometers), and the nu 1 + nu 3 combination band near 3705 cm-1 (2.70 micrometers).
Systematic explorations of new compounds in the cadmium iodate system by hydrothermal reactions led to two layered iodates, namely, Cd(IO{sub 3})X (X=Cl, OH). Cd(IO{sub 3})Cl crystallizes in the orthorhombic space group Cmca (No. 64) whereas Cd(IO{sub 3})(OH) crystallizes in the orthorhombic space group Pnma (No. 62). Cd(IO{sub 3})Cl displays a unique double layered structure composed of {sup 1}{sub ∞}[Cd−O{sub 3}Cl]{sub n} chains. Cadmium octahedrons form a 1D chain along the a-axis through edge sharing, and such chains are further interconnected via IO{sub 3} groups to form a special double layer on (020) plane. Cd(IO{sub 3})(OH) also exhibits a layered structure that is composed of cadmium cations, IO{sub 3} groups and hydroxyl ions. Within a layer, chains of CdO{sub 6} edge-shared octahedra are observed along the b-axis. And these chains are connected by IO{sub 3} groups into a layer parallel to the bc plane. Spectroscopic characterizations, elemental analysis, and thermogravimetric analysis for the reported two compounds are also presented. - Graphical abstract: Two new layered cadmium iodates Cd(IO{sub 3})X (X=Cl, OH) are reported. Cd(IO{sub 3})Cl features a unique double layered structure whereas Cd(IO{sub 3})(OH) displays an ordinary layered structure. - Highlights: • Two new layered cadmium iodates Cd(IO{sub 3})X (X=Cl, OH) are reported. • Cd(IO{sub 3})Cl features a unique double layered structure. • Cd(IO{sub 3})(OH) displays an ordinary layered structure. • The spectroscopic and thermal properties have been studied in detail.
In this report, we describe work awarded to Science Applications International Corporation, for the period 6/l/99 to 5/31/00. During this time period, we have investigated the interaction of Io, Jupiter's innermost Galilean satellite, with the Io plasma torus, and the role this interaction plays in producing ultraviolet (UV) emissions from neutral oxygen and sulfur. Io, the innermost of Jupiter's Galilean satellites, plays a unique role in the jovian magnetosphere. Neutral material that escapes from Io is ionized to form the lo torus, a dense, heavy-ion plasma that corotates with Jupiter and interacts with Io. Io supplies not only the torus, but is a major source of plasma for the entire magnetosphere. Ionization and charge-exchange of neutrals near lo strongly influences the plasma interaction, and Io's neutral atmosphere plays an important role in the generation of currents that couple Io to Jupiter. There have been no in situ measurements of the neutral density near Io, but remote observations of neutrals near lo have been performed for many years. Recent observations from the Hubble Space Telescope (HST) have shown detailed structure in UV emissions from neutral species near Io. Electron-impact of the neutrals by the Io torus plasma is the primary mechanism responsible for exciting these emissions. Previously, we have modeled the Io plasma environment using three-dimensional magnetohydrodynamic (MHD) simulations, and we have shown that the interaction between Io and the plasma torus plays an important role in producing the morphology of the observed emissions. In the past year, we have extended these studies to use both UV observations and Galileo particle and field measurements to investigate the Io interaction.
Time-resolved spectra of Io have been obtained with the Faint Object Spectrograph on the Hubble Space Telescope in January 1992 at times centered on the passage of Io into Jupiter's shadow. Two different eclipse observations covered 1100-1600 A and 2250-3300 A. In the far ultraviolet(far-UV) range, emission lines of atomic sulfur and oxygen from Io's atmosphere (similar to those previously detected with the International Ultraviolet Explorer (IUE)) have been observed from Io in sunlight, and the spatial extent of the emitting region has been resolved for the first time: this is 0.5-1 Io radii (R(sub Io)) above the surface. The emission lines are typically 1kR in brightness while Io is in sunlight, and decrease to a few hundred Rayleighs within 20 min or less of Io's passing into shadow. If the emissions are produced in Io's ionosphere, the decrease in shadow appears consistent with the collisional slowing and recombination of photoelectrons in 100-1000 s, with recombination an important quenching process if the dominant ion is molecular (i.e., SO2(+)) condensation, with the residual emission in shadow due either to plasma impact of gas above the hot volcanic calderas or electron impact on S and O. In the near-UV range, we have not detected any airglow emissions from Io's atmosphere in shadow, with the main limitation being a high level of scattered light from Jupiter. We derive a 3 sigma upper limit to the 2560 A SO emission feature of 1 KR, which is close to what is expected from electron impact on SO2 based on the obs erved brightness of the FUV S and O lines in shadow. A high signal-to-noise spectrum of Io's albedo in sunlight reveals a spectral shape similar to laboratory spectra of SO2 frost reflectivity, and the relative albedo spectrum changed as Io passed into eclipse and part of the disk was in shadow. No specific SO2 gas absorption features appear in the albedo spectrum, although there could be substantial gas absorption near 2800 A if the individual lines are narrow and saturated. Finally, we present preliminary models for the near-UV spectrum of Io as functions of SO2 frost areal coverage and SO2 gas density.
This paper presents an overview of electron densities obtained near Io from the Galileo plasma wave instrument during the first four flybys of Io. These flybys were Io, which was a downstream wake pass that occurred on December 7, 1995; I24, which was an upstream pass that occurred on October 11, 1999; I25, which was a south polar pass that occurred on November 26, 1999; and I27, which was an upstream pass that occurred on February 22, 2000. Two methods were used to measure the electron density. The first was based on the frequency of upper hybrid resonance emissions, and the second was based on the low-frequency cutoff of electromagnetic radiation at the electron plasma frequency. For three of the flybys, Io, I25, and I27, large density enhancements were observed near the closest approach to Io. The peak electron densities ranged from 2.1 to 6.8 x 10(exp 4) per cubic centimeters. These densities are consistent with previous radio occultation measurements of Io's ionosphere. No density enhancement was observed during the I24 flyby, most likely because the spacecraft trajectory passed too far upstream to penetrate Io's ionosphere. During two of the flybys, I25 and I27, abrupt step-like changes were observed at the outer boundaries of the region of enhanced electron density. Comparisons with magnetic field models and energetic particle measurements show that the abrupt density steps occur as the spacecraft penetrated the boundary of the Io flux tube, with the region of high plasma density on the inside of the flux tube. Most likely the enhanced electron density within the Io flux tube is associated with magnetic field lines that are frozen to Io by the high conductivity of Io's atmosphere, thereby enhancing the escape of plasma along the magnetic field lines that pass through Io's ionosphere.
The session "Io, with a Dash of Titan" contained the following reports:A Deep Lava Channel on Io ; Volcanic Activity at Tvashtar Catena, Io; Characterization of Activity at Loki from Galileo and Ground-based Observations; Active and Passive Lava Resurfacing Processes on Io: A Comparative Study of Loki Patera and Prometheus; Loki Patera as the Surface of a Magma Sea; Lava Lakes on Io: New Perspectives from Modeling; Global Distribution of Active Volcanism on Io as Known at the End of the Galileo Mission Relationships Between Paterae, Mountains, and Hotspots on Io from a Global Database; Lithospheric Recycling on Io: The Role of Delamination; Building Mountains on Io: Implications for Io s Lithosphere ; The Effect of Large Melt Fraction on the Deformation Behavior of Peridotite: Implications for the Rheology of Io s Mantle; Matching Various Observations of Io with DSMC Modeling: Plume, Plume Shadow, Sodium Field Around Pele; and How Does Titan Retain a Finite Orbital Eccentricity?
Magnetometer data from Galileo's close encounters with Io do not establish absolutely either the existence or absence of an internal magnetic moment because the measurements were made in regions where plasma currents contribute sizable magnetic perturbations. Data from an additional encounter where the closest approaches were made beneath Io's south polar regions, were lost. The recent passes enhance our understanding of the interaction of Io and its flux tube with the torus, and narrows the limits on possible internal sources of magnetic fields. Simple field-draping arguments account for some aspects of the observed rotations. Analyses in terms of both a magnetized and an unmagnetized Io are considered. Data from the February 2000 pass disqualify a strongly magnetized Io (surface equatorial field stronger than the background field) but do not disqualify a weakly magnetized Io (surface equatorial field of the order of Ganymede's but smaller than the background field at Io). Models imply that if Io is magnetized, its magnetic moment is not absolutely antialigned with the rotation axis. The inferred tilt is consistent with contributions from an inductive field on the order of those observed at Europa and Callisto. The currents would flow in the outer mantle or aesthenosphere if an induced field is present. Wave perturbations differing on flux tubes that do or do not link directly to Io and its ionosphere suggest the following: (1) the latter flux tubes are almost stagnant in Io's frame; and (2) a unipolar inductor correctly models the currents linking Io to Jupiter's ionosphere.
The mechanisms by which orbital resonance locks are maintained among Io, Europa and Ganymede are analyzed, and the effects of the dissipative tides in Jupiter and Io on their establishment and evolution are considered. According to the proposed model, initially all three satellites are in orbits far from the 2:1 commensurabilities or the three body lock. The tide raised on Io damps down the free eccentricity; only modest tidal heating occurs. Subsequently the dissipative tide raised on Jupiter by Io causes Io's orbit to spiral outwards; Io approaches the 2:1 commensurability with Europa. Io's forced eccentricity increases rapidly to a critical value, and thereafter the resonant interaction causes Europa's orbit to expand at half that of Io's orbit. A fluid core is probably formed as the result of tidal heating. Finally Europa approaches the 2:1 commensurability, angular momentum is transferred from Europa's orbit to Ganymede's, and a steady state is attained. On the basis of the observed three-body resonance amplitude (0.066 deg), it is concluded that three-body resonance may have formed less than 500 million years ago; a measurement of the rate of tidal heating in Io may determine the Q value of both Io and Jupiter. Upper and lower limits for the Jovian Q value of 2 million and 20,000 are suggested.
The physics and geology of Io have been extensively studied, but there has been little discussion of the chemistry. Relatively little is known about Io chemistry, but there are constraints. Further, it will be a long time before improvements will result from direct observation, given the severe difficulties with the Galileo mission. Via laboratory simulation experiments, plausible thermochemical and photochemical processes which determine the nature and amounts of surface constituents of Io are explored. The well-known density of Io shows that the planet overall is rocky. Because the orbit of Io is well within the magnetosphere of Jupiter and because Io only has a thin, transient SO2 atmosphere, the surface is continually sputtered with magnetospheric ions. Complex processes ionize and accelerate the Io surface atoms to keV and MeV energies. Remarkably, only S, O, and Na ions were found by Voyager. Sputtering also produces an atomic cloud of Na and S (O not observable) with a trace of K. Both gaseous and solid SO2 are known from spectroscopic studies. A trace of H2S and possibly CO2 are present. Geologic features are interpreted in terms of elemental S, but there is no direct evidence for this constituent. We thus have a rocky planet which does not have rocks on the surface. Our general goal is to understand the cycling of Na, S, and O through the crust and atmosphere on present-day Io and to understand how Io evolved to this state. A specific objective was to determine the phases on the surface which are the source of the Na in the atmosphere of Io.
Observations of (O I) 6300-A emission near Io have been obtained in 1990, 1991, and 1992 by the National Solar Observatory staff using the solar-stellar spectrography on the McMath-Pierce telescope at Kitt Peak. High-resolution spectra with a resolving power of about 1.2 x 10(exp 5) were obtained with an integration time of 10-15 min each. The viewing aperture for the observations was 5.2 arc sec x 5.2 arc sec centered on Io, with spatial resolution limited within this area by seeing conditions. Observations thus far have been reduced to obtain average brightness values over the aperture which range from approximately 200 to 1000 R for a number of different Io phase angles and Io system III longitudes. The (O I) 6300-A emission brightness exhibits an east-west asymmetry, where the average intensity for Io phase angles in the west (receding ansa) is 1.5 times brighter than in the east (approaching ansa). Similar east-west intensity ratios have also been observed for neutrals near Io and ions in the plasma torus for a number of other optical and ultraviolet emission lines which are excited by electron impact. In addition to the east-west asymmetry, the (O I) 6300-A emission brightness exhibits a strong dependence on the Io system III longitude angle, with a maximum value occurring in the range 200 deg +/- 50 deg. Earlier IUE observations of ultraviolet emission lines of O and S near Io obtained over a number of years have measured the east-west asymmetry, but the long IUE integration times of approximately 7-14 hours masked any detection of system III variability. For the (O I) 6300-A emission, the O(1D) state may be excited by electron impact of atomic oxygen and by electron impact dissociation of SO. The molecule SO may be present at the exobase or may be produced above the exobase as the dominant product of SO2 dissociation by electron impact. Preliminary assessment indicates that production of O(1D) by molecular dissociation may be more important. The (O I) 6300-A emission may therefore provide a remote signature for monitoring (1) the upward transport rates of molecular species in Io's atmosphere, (2) the relative abundance and time-variable dissociation of SO2 and/or SO at the exobase and in the corona of Io, and (3) the spatial distribution of these escaping molecular and atomic species and their ion production rates in the planetary magnetosphere.
We have investigated the interaction of Io, Jupiter's innermost Galilean satellite, with the Io plasma torus. The interaction of Io with the plasma surrounding it has been a subject of interest for almost 30 years, dating from the discovery by Bigg (1964) that radio emissions from the Jovian magnetosphere are controlled by Io's position. Since that time, both ground-based and spacecraft observations have shown that Io is a unique satellite that influences the Jovian magnetosphere in important ways. In particular, material from Io is a major source of plasma for the magnetosphere, and the energy that this plasma harnesses from Jupiter's co-rotating magnetic field is an important power source for the magnetosphere. It is apparent that the local interaction of the torus plasma with Io plays a key role in the formation, composition, and energetics of the Io torus; the interaction is also highly nonlinear. We have modeled this interaction using time-dependent three-dimensional magnetohydrodynamic (MHD) simulations. During this past year, we have used NASA support to develop a new MHD code to study the interaction. As part of the Galileo spacecraft's recent successful insertion into orbit around Jupiter, the spacecraft passed within 900 km of Io's surface. Our calculations have focused on using Galileo particles and fields data to examine a question that was not resolved by the Voyager observations: Does Io have an intrinsic magnetic field? In this progress summary, we describe our efforts on this problem to date.
Traditional workload managers do not have the capacity to consider how IO contention can increase job runtime and even cause entire resource allocations to be wasted. Whether from bursts of IO demand or parallel file systems (PFS) performance degradation, IO contention must be identified and addressed to ensure maximum performance. In this paper, we present AI4IO (AI for IO), a suite of tools using AI methods to prevent and mitigate performance losses due to IO contention. AI4IO enables existing workload managers to become IO-aware. Currently, AI4IO consists of two tools: PRIONN and CanarIO. PRIONN predicts IO contention and empowers schedulers to prevent it. CanarIO mitigates the impact of IO contention when it does occur. We measure the effectiveness of AI4IO when integrated into Flux, a next-generation scheduler, for both small- and large-scale IO-intensive job workloads. Our results show that integrating AI4IO into Flux improves the workload makespan up to 6.4%, which can account for more than 18,000 node-h of saved resources per week on a production cluster in our large-scale workload.
Because of relative motion between the innermost Galilean satellite Io and Jupiter's ionosphere, a current is drawn from the ionosphere that can be a source of both deposition on, and sputtering from, the surface of Io. It is shown that the ions in this current strike Io in a localized region in the quadrant bounded by a line connecting Io and Jupiter and a tangent line extended in the direction of Io's orbital motion. If these ions are the principal source of sodium that is sputtered from Io, then this current provides a simple explanation of the observation of a localized area from which sodium ions escape from Io. The geometry of this current may also affect the optical surface of Io. Several experimental tests are suggested that can determine the compatibility of this hypothesis with the directly observable properties of Io's surface.