Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “spatial accelerator”

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 307 records · Page 17
All sourcesNASA NTRSDOE OSTI
NASA NTRSDate not supplied

A generalizable machine learning approach to predict land surface temperature

Monitoring of land surface and atmospheric states is highly reliant on satellite data. Traditionally, data products are generated using carefully tuned and validated algorithms for low-earth orbit (LEO) sensors. However, the emerging constellation of geostationary (GEO) sensors contributes global, high temporal resolution observations which can better capture the diurnal variability of key observables like land surface temperature (LST). Using high performance computing and datasets from the NASA Earth Exchange, we exploit co-located, co-temporal observations from LEO and GEO satellites to develop a deep learning-based method for sensor-to-sensor algorithm emulation. Our model is trained on GOES-16 thermal bands to predict MODIS Terra LST and achieves a validation error <2K. Further, application of the model to unseen times of day and a second GEO sensor observing an unseen spatial domain demonstrate the generalization of the deep learning model across space, time and spectra. We anticipate that the synergies between a variety of active orbit configurations can be used to accelerate application of existing algorithms to new datasets.

Kate Marie Duffy↗
NASA NTRS2021

Rapid Cloud Removal of Dimethyl Sulfide Oxidation Products Limits SO 2 and Cloud Condensation Nuclei Production in the Marine Atmosphere

Oceans emit large quantities of dimethyl sulfide (DMS) to the marine atmosphere. The oxidation of 60 DMS leads to the formation and growth of cloud condensation nuclei (CCN) with consequent effects on Earth’s radiation balance and climate. Quantitative assessment of the impact of DMS emissions on CCN concentrations necessitates a detailed description of the oxidation of DMS in the presence of existing aerosol particles and clouds. In the unpolluted marine atmosphere, DMS is efficiently oxidized to hydroperoxymethyl thioformate (HPMTF), a stable intermediate in the chemical trajectory towards sulfur dioxide (SO 2 ) and ultimately sulfate aerosol (Veres et al. 2020). Using direct airborne flux measurements, we demonstrate that irreversible loss of HPMTF to clouds in the marine boundary layer determines the HPMTF lifetime (τ HPMTF < 2 hours) and terminates DMS oxidation to SO 2 in the cloudy marine boundary layer. When accounting for HPMTF cloud loss in a global chemical transport model, we show that SO 2 production from DMS is reduced by 35% globally and near surface (0-3km) SO 2 concentrations over the ocean are lowered by 24%. This large, previously unconsidered loss process for volatile sulfur accelerates the timescale for conversion of DMS to sulfate, while limiting new particle formation in the marine atmosphere and changing the dynamics of aerosol growth. This loss process potentially reduces the spatial scale over which DMS emissions contribute to aerosol production and growth and weakens the link between DMS emission and marine CCN production with subsequent implications for cloud formation, radiative forcing, and climate.

dimethyl sulfide↗
NASA NTRS2026

Petermann Glacier on the Brink: Progress, Challenges and Insights

Petermann Glacier, the largest marine-terminating glacier in northern Greenland based on catchment area and ice discharge, plays a key role in regulating ice discharge from the Greenland Ice Sheet into the Arctic Ocean. With an upstream catchment connected to the ice sheet interior via a deep subglacial canyon, its future stability has major implications for sea level rise. In this review, we synthesize recent advances in understanding Petermann’s dynamics across three critical interfaces: the ice–ocean, ice–atmosphere, and ice–bed boundaries. At the surface, observations show that reanalysis products underestimate air temperatures and melt, while regional climate models diverge significantly in their estimates of surface mass balance, underscoring the need for improved in situ data and models. At the ocean boundary, enhanced basal melting driven by both subglacial runoff and Atlantic water intrusions is identified as the dominant driver of recent mass loss of Petermann Glacier, with continued warming posing a serious threat to the stability of the floating tongue. At the bed, new geophysical synthesis reveals complex geology, likely spatial variability in geothermal heat flux, and the influence of the megacanyon on seasonal hydrology and velocity fluctuations. Petermann’s mass balance has been negative in recent decades without corresponding flow acceleration. However, the glacier has undergone significant calving events, and the current rifting that began in September 2025 highlights its vulnerability. This upcoming calving event underscores the timeliness of this review, as it will put Petermann Glacier’s terminus at its most retreated position since records began in 1923. The anticipated retreat of the ice tongue also reduces buttressing and brings the terminus closer to the grounding zone, and modeling studies suggest that calving within 12 km of the grounding zone could potentially trigger dynamic retreat, accelerating ice discharge, and a doubling of flow speeds. We conclude that improved observations, sustained monitoring of oceanographic and atmospheric properties, and high-resolution modeling are critical to constraining projections of Petermann Glacier’s future and its role in the stability of the Greenland Ice Sheet.

Dominik Fahrner↗
NASA NTRS1991

Acceleration of electrons outside flares - Coronal manifestation and possible origin

Imaging observations at decimetric and metric wavelengths of the solar corona are used to investigate a short-lasting noise storm and the associated changes of the underlying active-region plasma. It is shown that a new source appears in the active region in close temporal and spatial coincidence with the onset of the noise storm in the middle corona. The onset of the noise storm is delayed at longer wavelengths. At a given wavelength, the noise-storm source undergoes a systematic slow movement with a significant component perpendicular to the magnetic-field lines above the active region. The observations are tentatively attributed to the emission of nonthermal electrons in a system of expanding coronal loops, the expansion being initiated by the appearance of the new 21-cm source in the low atmosphere. The derived velocity of expansion is about 80 to 150 km/s. The electrons emitting the noise storm cannot be provided by the high-energy tail of the Maxwellian in the new active region source, but originate either from a nonthermal population in this source or from acceleration at higher altitudes, in the structures which give rise to the noise storm.

Raulin, J. P.↗
NASA NTRS1987

Two-phase measurements of a spray in the wake of a bluff body

The dynamics of spray drop interaction with the turbulent, recirculating wake of a flat disk bluff body were investigated using a phase Doppler particle analyzer to determine drop size and velocity and the gas-phase velocity. Detailed measurements obtained included spray drop size, axial and radial velocity, angle of trajectory, and size-velocity correlations. The gas-phase velocity was determined from seeding of the two-phase flow. Results showed dramatic differences in drop behavior for various size classes when interacting with the turbulent flow field. Small drops were quickly entrained and recirculated, while initially, the larger drops continued in the general direction of the spray cone. Further downstream, significant numbers of large drops recirculated, generating a bifurcated size-velocity correlation. These lateral convections and streamwise accelerations and decelerations strongly influenced the number density along with size and velocity distributions. The complex interaction of the spray with the turbulent air-flow points out the need for spatially-resolved measurements that determine drop behavior for individual size classes, rather than characterizing a spray only via simple integral quantities such as the Sauter mean diameter.

Rudoff, R. C.↗
NASA NTRS1994

Evolution of solitary density waves in stellar winds of early-type stars: A simple explanation of discrete absorption component behavior

We model the evolution of a density shell propagating through the stellar wind of an early-type star, in order to investigate the effects of such shells on UV P Cygni line profiles. Unlike previous treatments, we solve the mass, momentum, and energy conservation equations, using an explicit time-differencing scheme, and present a parametric study of the density, velocity, and temperature response. Under the assumed conditions, relatively large spatial scale, large-amplitude density shells propagate as stable waves through the supersonic portion of the wind. Their dynamical behavior appears to mimic propagating 'solitary waves,' and they are found to accelerate at the same rate as the underlying steady state stellar wind (i.e., the shell rides the wind). These hydrodynamically stable structures quantitatively reproduce the anomalous 'discrete absorption component' (DAC) behavior observed in the winds of luminous early-type stars, as illustrated by comparisons of model predictions to an extensive International Ultraviolet Explorer (IUE) time series of spectra of zeta Puppis (O4f). From these comparisons, we find no conclusive evidence indicative of DACs accelerating at a significantly slower rate than the underlying stellar wind, contrary to earlier reports. In addition, these density shells are found to be consistent within the constraints set by the IR observations. We conclude that the concept of propagating density shells should be seriously reconsidered as a possible explanation of the DAC phenomenon in early-type stars.

Waldron, Wayne L.↗
NASA NTRS1999

Correlation of Fin Buffet Pressures on an F/A-18 with Scaled Wind-Tunnel Measurements

Buffeting is an aeroelastic phenomenon occurring at high angles of attack that plagues high performance aircraft, especially those with twin vertical tails. Previous wind-tunnel and flight tests were conducted to characterize the buffet loads on the vertical tails by measuring surface pressures, bending moments, and accelerations. Following these tests, buffeting responses were computed using the measured buffet pressures and compared to the measured buffeting responses. The calculated results did not match the measured data because the assumed spatial correlation of the buffet pressures was not correct. A better understanding of the partial (spatial) correlation of the differential buffet pressures on the tail was necessary to improve the buffeting predictions. Several wind-tunnel investigations were conducted for this purpose. When compared, the results of these tests show that the partial correlation scales with flight conditions. One of the remaining questions is whether the wind-tunnel data is consistent with flight data. Presented herein, cross-spectra and coherence functions calculated from pressures that were measured on the High Alpha Research Vehicle indicate that the partial correlation of the buffet pressures in flight agrees with the partial correlation observed in the wind tunnel.

Moses, Robert W.↗
NASA NTRS2012

The Mechanisms for the Onset and Explosive Eruption of Coronal Mass Ejections and Eruptive Flares

We have investigated the onset and acceleration of coronal mass ejections (CMEs) and eruptive flares. To isolate the eruption physics, our study uses the breakout model, which is insensitive to the energy buildup process leading to the eruption. We performed 2.5D simulations with adaptive mesh refinement that achieved the highest overall spatial resolution to date in a CME/eruptive flare simulation. The ultra-high resolution allows us to separate clearly the timing of the various phases of the eruption. Using new computational tools, we have determined the number and evolution of all X- and O-type nulls in the system, thereby tracking both the progress and the products of reconnection throughout the computational domain. Our results show definitively that CME onset is due to the start of fast reconnection at the breakout current sheet. Once this reconnection begins, eruption is inevitable; if this is the only reconnection in the system, however, the eruption will be slow. The explosive CME acceleration is triggered by fast reconnection at the flare current sheet. Our results indicate that the explosive eruption is caused by a resistive instability, not an ideal process. Moreover, both breakout and flare reconnections begin first as a form of weak tearing characterized by a slowly evolving plasmoids, but eventually transition to a fast form with well-defined Alfvenic reconnection jets and rapid flux transfer. This transition to fast reconnection is required for both CME onset and explosive acceleration. We discuss the key implications of our results for CME/flare observations and for theories of magnetic reconnection.

Karpen, Judith T.↗
NASA NTRS2003

New ISTP Solar Max: A Multi-Spacecraft Study of the Flow of Ionospheric Plasma

The unique instrumentation on the Polar satellite combined with the simultaneous measurement of different parts of the magnetosphere with multiple satellites make possible the study of magnetospheric processes in a special way. In particular, the study of the ionospheric supply of plasma to the magnetosphere can by accomplished to give important results on the plasmas which drive the dynamics of the magnetosphere. This study concentrated on the period of September to December, 2001 in which the Polar orbit had precessed to the point that the line of apsides was near the equatorial plane. This unique orbital configuration enabled the TIDE instrument to measure outflowing ions across the polar cap and then transit the magnetotail lobes and observe the dramatic change in plasma characteristics as the satellite entered the plasma sheet. Contact was made with investigators on the Cluster and Geotail satellite missions and corresponding time frames were studied in the data. There were two approximate conjunctions between Polar and Geotail and data were compared to look for features which might be related. The higher energy concentration of the Geotail instrument made direct comparisons with TIDE difficult, and the Cluster measurements did not surface any cases that corresponded closely in space and time. There were, however, many interesting aspects of the Polar orbits which permitted the observation of the changing ionospheric outflowing plasma characteristics. As in earlier measurements, the ionospheric plasma could be seen flowing up the magnetic field lines out of the northern and southern polar caps. Its energy suggested a polar wind origin energized by the centrifugal acceleration of flow through the polar cusp. The roughly 10eV ions then moved out into the lobes of the magnetotail where they could be seen flowing toward the plasma sheet in both the northern and southern magnetotail lobes. There was also a double field-aligned region of warm ions observed just outside the plasmasphere, stretching toward the auroral zone and inner plasma sheet boundary. Upon entering the plasma sheet, the plasma energy jumped from 10 s of eV to greater than 1 keV. The single field aligned flows transitioned to highly variable spatially choppy energized ion distributions with a variety of pitch angle configurations. This pattern was quite repeatable in all of the Polar orbits that were examined and are compatible with the source of ions being the polar wind which is then energized to 10 s of eV by the centrifugal acceleration. These modestly energized polar wind ions are then carried to the magnetotail where they are substantially energized by the curvature drift-induced movement across the cross-tail potential of the magnetotail. This latter drift energizes the ions to the energies typically found in the plasma sheet. Subsequent drift and energization can cause the ions to become part of the ring current. The results of this study were presented at the Spring AGU meeting in 2002 and the GEM meeting in June 2003. They are the foundation for a paper that has been submitted by Matthew Huddleston to the Journal of Geophysical Research in December 2003. This work was part of the thesis that Matthew completed in finishing his Ph.D. in Physics at Vanderbilt University.

Chappell, Charles R.↗
NASA NTRS1979

Particle size and velocity measurement in flames by laser anemometer

Simultaneous droplet size and velocity measurements by a particle counting Laser Doppler Anemometer (LDA) in kerosene fuel sprays under burning and non-burning conditions are presented. Particle sizes are derived from pulse height analysis of the mean LDA signals and velocities are simultaneously determined by measuring Doppler shift frequencies. The measurements show that droplet velocity is a function of droplet diameter for burning and non-burning conditions, and spatially averaged size distributions are derived from velocity data. A comparison of results obtained under burning and non-burning conditions show changes in size distribution due to preferential vaporization of small droplets, acceleration due to thermal expansion of gases, and corresponding changes in droplet momentum.

Chigier, N. A.↗
NASA NTRS1985

Simultaneous microwave observations of solar flares at 6 and 20 cm wavelengths using the VLA

Using the Very Large Array, solar burst observations have been carried out simultaneously at 6 and 20 cm. Structural changes and preheating have been observed in the flaring regions on time scales of minutes to tens of minutes before the onset of the burst impulsive phase. The 6 cm burst sources are located close to the neutral line, or near the legs of a flaring loop. The 20 cm burst sources show complex and extended structures spatially separated from both the preburst emission and the gradual decay phase of the burst. The observations are interpreted in terms of a two-component flare model (bulk heating as well as acceleration of particles), and the physical parameters of the burst sources are derived.

Melozzi, M.↗
NASA NTRS1998

Spatially Resolved Spectroscopy of the SNR IC443

IC 443 is a supernova remnant of intermediate age, i.e. a few thousand years. It is especially interesting because part of its periphery is expanding into a molecular cloud while other sections are expanding into a typical interstellar medium of much lower density. Since the evolution of a supernova remnant through its various phases is affected by the density of the medium it expands into with the reasonable assumption that the supernova explosion was approximately symmetric we have an opportunity to observe a single object in two phases simultaneously. It was observed by ASCA in April, 1993 for a short period during the PV phase and more thoroughly in a 42 ksec exposure in March, 1994. The latter measurement provides most of the results that have been reported. Most of the analysis took place after the grant ended but is included here for completeness. The data was sent simultaneously to US and Japanese Pls. We worked independently. The software set of FTOOLs was used to construct images and spectra. They were judged to be rather unintuitive and not at all user friendly. I found I was using one FTOOL to read the header to obtain information that would only be provided to another FTOOL. The Japanese investigators were more successful. They analyzed the data and published results more rapidly. The scientific results summarized below are based primarily on their publications. Since IC 443 is an interesting example of a middle aged SNR in which a variety of processes are occurring it is one of a class. IC 443 exhibits shell-like emission in hard X-rays and extended soft X-rays with thin thermal spectra. It resembles SN 1006 in these respects. IC 443 contains hard X-rays in a semi-circular shell surrounding the thermal component. The total hard X-ray flux in the ASCA FOV is only a half of the Ginga hard component; which suggests that the hard X-rays are not confined only in the shell but some are extended larger than the ASCA FOV of eq 1 degree diameter. Japanese investigators examined the spatial structure of the thermal component and analyzed the GIS spectra with a non-equilibrium plasma model, and found no systematic variation of the interstellar absorption across the remnant. Evidence for shock acceleration of cosmic rays to high energies (10 TeV) was found by Keohane. X-ray imaging spectroscopy with ASCA reveals two regions of particularly hard emission: an unresolved source embedded in an extended emission region, and a ridge of emission coincident with the southeastern rim. Both features are located on part of the radio shell where the shock wave is interacting with molecular gas, and together they account for a majority of the emission at 7 keV. Though we would not have noticed it a priori, the unresolved feature is coincident with one resolved by the ROSAT HRI. The ASCA measurements were combined with higher energy data from the XTE and GRO missions and with radio and TeV gamma-ray data to produce a nonthermal multiwavelength spectrum for IC 443 which was fit with a cosmic ray interaction model. This model calculates the cynchrotron, bremsstrahlung, invers Compton, and neutral pion decay emission produced by locally accelerated cosmic ray interacting with ambient matter, soft photon fields, and magnetic fields.

Gorenstein, P.↗
NASA NTRS2002

Double Layers in Expanding Plasmas and Their Relevance to the Auroral Plasma Processes

When a dense plasma consisting of a cold and a sufficiently warm electron population expands, a rarefaction shock forms. In the expansion of the polar wind in the magnetosphere, it has been previously shown that when a sufficiently warm electron population also exists, in addition to the usual cold ionospheric one, a discontinuity forms in the electrostatic potential distribution along the magnetic field lines. Despite the lack of spatial resolution and the assumption of quasi-neutrality in the polar wind models, such discontinuities have been called double layers (DLs). Recently similar discontinuities have been invoked to partly explain the auroral acceleration of electrons and ions in the upward current region. By means of one-dimensional Vlasov simulations of expanding plasmas, for the first time we make here the connection between (i) the rarefaction shocks, (ii) the discontinuities in the potential distributions, and (iii) DLs. We show that when plasmas expand from opposite directions into a deep density cavity with a potential drop across it and when the plasma on the high-potential side contains two electron populations, the temporal evolution of the potential and the plasma. distribution generates evolving multiple double layers with an extended density cavity between them. One of the DLs is the rarefaction-shock (RFS) and it forms by the reflections of the cold electrons coming from the high-potential side; it supports a part of the potential drop approximately determined by the hot electron temperature.

Singh, Nagendra↗
NASA NTRS2003

Double Layers in Expanding Plasmas and Their Relevance to the Auroral Plasma Processes

When a dense plasma consisting of a cold and a sufficiently warm electron population expands, a rarefaction shock forms [Bezzerides et al., 1978]. In the expansion of the polar wind in the magnetosphere, it has been previously shown that when a sufficiently warm electron population also exists, in addition to the usual cold ionospheric one, a discontinuity forms in the electrostatic potential distribution along the magnetic field lines [Barakat and Schunk, 1984]. Despite the lack of spatial resolution and the assumption of quasi-neutrality in the polar wind models, such discontinuities have been called double layers (DLs). Recently similar discontinuities have been invoked to partly explain the auroral acceleration of electrons and ions in the upward current region [Ergun et al., 2000]. By means of one-dimensional Vlasov simulations of expanding plasmas, for the first time we make here the connection between (1) the rarefaction shocks, (2) the discontinuities in the potential distributions, and (3) DLs. We show that when plasmas expand from opposite directions into a deep density cavity with a potential drop across it and when the plasma on the high-potential side contains hot and cold electron populations, the temporal evolution of the potential and the plasma distribution generates evolving multiple double layers with an ,extended density cavity between them. One of the DLs is the rarefaction-shock (RFS) and it forms by the reflections of the cold electrons coming from the high-potential side; it supports a part of the potential drop approximately determined by the hot electron temperature. The other DLs evolve from charge separations arising either from reflection of ions coming from the low-potential side or stemming from plasma instabilities; they support the rest of the potential drop. The instabilities forming these additional double layers involve electron-ion (e-i) Buneman or ion-ion (i-i) two-stream interactions. The electron-electron two-stream interactions on the high-potential side of the RFS generate electron-acoustic waves, which evolve into electron phase-space holes. The ion population originating from the low-potential side and trapped by the RFS is energized by the e-i and i-i instabilities and it eventually precipitates into the high-potential plasma along with an electron beam. Applications of these findings to the auroral plasma physics are discussed.

Singh, Nagendra↗
NASA NTRS2024

Highly Dynamic Gamma-Ray Emissions Are Common in Tropical Thunderclouds

Thunderstorms emit fluxes of gamma rays known as gamma-ray glows, sporadically observed by aircraft, balloons and from ground. Glows are observed as increased gamma-ray emissions by tens of percent up to two orders of magnitude above the background, sometimes abruptly terminated by lightning discharges. Glows are produced by the acceleration of energetic electrons in high electric field regions within thunderclouds, and contribute to charge dissipation. Glows are considered as quasi-stationary phenomena, with durations up to a few tens of seconds and spatial scales up to 10-20 kilometers. No measurement of the full extension in space and time of a gamma ray glow region and their occurring frequency has been reported so far. Here we show that tropical thunderclouds over ocean and coastal regions commonly emit gamma rays for hours over areas up to a few thousands of square kilometers. Emission is associated with deep convective cores; it is not uniform and continuous but shows characteristic timescales of 1-10 seconds and even sub-second for individual glows. The dynamics of gamma-glowing thunderclouds starkly contradicts the quasi-stationary picture of glows, but rather resembles that of a huge gamma-glowing «boiling pot» both in pattern and behavior.

M Marisaldi↗
NASA NTRS2009

Combined Modeling of Acceleration, Transport, and Hydrodynamic Response in Solar Flares: The Numerical Model - 1

Acceleration and transport of high-energy particles and fluid dynamics of atmospheric plasma are interrelated aspects of solar flares, but for convenience and simplicity they were artificially separated in the past. We present here self consistently combined Fokker-Planck modeling of particles and hydrodynamic simulation of flare plasma. Energetic electrons are modeled with the Stanford unified code of acceleration, transport, and radiation, while plasma is modeled with the Naval Research Laboratory flux tube code. We calculated the collisional heating rate directly from the particle transport code, which is more accurate than those in previous studies based on approximate analytical solutions. We repeated the simulation of Mariska et al. with an injection of power law, downward-beamed electrons using the new heating rate. For this case, a -10% difference was found from their old result. We also used a more realistic spectrum of injected electrons provided by the stochastic acceleration model, which has a smooth transition from a quasi-thermal background at low energies to a non thermal tail at high energies. The inclusion of low-energy electrons results in relatively more heating in the corona (versus chromosphere) and thus a larger downward heat conduction flux. The interplay of electron heating, conduction, and radiative loss leads to stronger chromospheric evaporation than obtained in previous studies, which had a deficit in low-energy electrons due to an arbitrarily assumed low-energy cutoff. The energy and spatial distributions of energetic electrons and bremsstrahlung photons bear signatures of the changing density distribution caused by chromospheric evaporation. In particular, the density jump at the evaporation front gives rise to enhanced emission, which, in principle, can be imaged by X-ray telescopes. This model can be applied to investigate a variety of high-energy processes in solar, space, and astrophysical plasmas.

STOCHASTIC ACCELERATION MODEL↗
NASA NTRS1993

Leading-Edge Votex-System Details Obtained on F-106B Aircraft Using a Rotating Vapor Screen and Surface Techniques

A flight research program to study the flow structure and separated-flow origins over an F-106B aircraft wing is described. The flight parameters presented include Mach numbers from 0.26 to 0.81, angles of attack from 8.5 deg to 22.5 deg, Reynolds numbers from 22.6 x 10(exp 6) to 57.3 x 10(exp 6) and load factors from 0.9 to 3.9 times the acceleration due to gravity. Techniques for vapor screens, image enhancement, photogrammetry, and computer graphics are integrated to analyze vortex-flow systems. Emphasis is placed on the development and application of the techniques. The spatial location of vortex cores and their tracks over the wing are derived from the analysis. Multiple vortices are observed and are likely attributed to small surface distortions in the wing leading-edge region. A major thrust is to correlate locations of reattachment lines obtained from the off-surface (vapor-screen) observations with those obtained from on-surface oil-flow patterns and pressure-port data. Applying vapor-screen image data to approximate reattachment lines is experimental, but depending on the angle of attack, the agreement with oil-flow results is generally good. Although surface pressure-port data are limited, the vapor-screen data indicate reattachment point occurrences consistent with the available data.

Lamar, John E.↗
NASA NTRS1994

An implementation of a tree code on a SIMD, parallel computer

We describe a fast tree algorithm for gravitational N-body simulation on SIMD parallel computers. The tree construction uses fast, parallel sorts. The sorted lists are recursively divided along their x, y and z coordinates. This data structure is a completely balanced tree (i.e., each particle is paired with exactly one other particle) and maintains good spatial locality. An implementation of this tree-building algorithm on a 16k processor Maspar MP-1 performs well and constitutes only a small fraction (approximately 15%) of the entire cycle of finding the accelerations. Each node in the tree is treated as a monopole. The tree search and the summation of accelerations also perform well. During the tree search, node data that is needed from another processor is simply fetched. Roughly 55% of the tree search time is spent in communications between processors. We apply the code to two problems of astrophysical interest. The first is a simulation of the close passage of two gravitationally, interacting, disk galaxies using 65,636 particles. We also simulate the formation of structure in an expanding, model universe using 1,048,576 particles. Our code attains speeds comparable to one head of a Cray Y-MP, so single instruction, multiple data (SIMD) type computers can be used for these simulations. The cost/performance ratio for SIMD machines like the Maspar MP-1 make them an extremely attractive alternative to either vector processors or large multiple instruction, multiple data (MIMD) type parallel computers. With further optimizations (e.g., more careful load balancing), speeds in excess of today's vector processing computers should be possible.

Olson, Kevin M.↗