Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “inference accelerators”

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 451 records · Page 25
All sourcesNASA NTRSDOE OSTI
NASA NTRS2007

Analysis of Cometary Dust Impact Residues in the Aluminum Foil Craters of Stardust

In January 2006, the sample return capsule from NASA s Stardust spacecraft successfully returned to Earth after its seven year mission to comet Wild-2. While the principal capture medium for comet dust was low-density graded silica aerogel, the 1100 series aluminum foil (approximately 100 m thick) which wrapped around the T6064 aluminum frame of the sample tray assembly (STA) contains micro-craters that constitute an additional repository for Wild-2 dust. Previous studies of similar craters on spacecraft surfaces, e.g. the Long Duration Exposure Facility (LDEF), have shown that impactor material can be preserved for elemental and mineralogical characterization, although the quantity of impact residue in Stardust craters far exceeds previous missions. The degree of shock-induced alteration experienced by the Wild-2 particles impacting on foil will generally be greater than for those captured in the low-density aerogel. However, even some of the residues found in LDEF craters showed not only survival of crystalline silicates but even their solar flare tracks, which are extremely fragile structures and anneal at around 600 C. Laboratory hypervelocity experiments, using analogues of Wild-2 particles accelerated into flight-grade foils under conditions close to those of the actual encounter, showed retention of abundant projectile residues at the Stardust encounter velocity of 6.1 km/s. During the preliminary examination (PE) of the returned foils, using optical and electron microscopy studies, a diverse range in size and morphologies of micro-craters was identified. In this abstract we consider the state of residue preservation in a diverse range of craters with respect to their elemental composition and inferred mineralogy of the original projectiles.

Graham, G. A.↗
NASA NTRSDate not supplied

GeneLab: Current and Future Omics Data Integration Between Space Biology and HRP

For the past five years, the Biological and Physical Sciences Division has pioneered Open Science in Space Biology by funding the NASA GeneLab project. Along with the Ames Life Sciences Data Archive, GeneLab has quickly become the world leader in archiving and scientifically curating spaceflight and spaceflight relevant multi-omics data. Specifically, the GeneLab Data System has become a full enterprise solution providing advanced mining capabilities, several application programming interfaces for data federation and machine learning approaches, and delivering to the world an analytical and visualization platform which has enabled collaboration within the scientific community. Over the past three years, large meta-analysis and modeling studies have been published by the GeneLab Analysis Working Groups (AWGs), which are comprised of ~200 volunteer scientists. One natural extension of GeneLab data reuse has recently turned towards linking animal data with human data, which is the next necessary step to further validate animal models for inferring biological risks to humans conducting LEO, lunar or Martian missions. As such, data from the Human Research Program are an essential component of GeneLab and ALSDA. At the moment, simulated space radiation experiments conducted at Brookhaven National Laboratory make the most of HRP GeneLab data, and the scientific community has been eager to also link their animal spaceflight results to actual Astronaut data and human analog data. We will discuss further the current status of knowledge and future approaches to accelerate our basics understanding of the impact of space stressors on humans using latest omics technology.

omics↗
DOE OSTI2021

FlowPM: Distributed TensorFlow implementation of the FastPM cosmological N-body solver

Here, we present FlowPM, a Particle-Mesh (PM) cosmological N-body code implemented in Mesh-TensorFlow for GPU-accelerated, distributed, and differentiable simulations. We implement and validate the accuracy of a novel multi-grid scheme based on multiresolution pyramids to compute large-scale forces efficiently on distributed platforms. We explore the scaling of the simulation on large-scale supercomputers and compare it with corresponding Python based PM code, finding on an average 10x speed-up in terms of wallclock time. We also demonstrate how this novel tool can be used for efficiently solving large scale cosmological inference problems, in particular reconstruction of cosmological fields in a forward model Bayesian framework with hybrid PM and neural network forward model. We provide skeleton code for these examples and the entire code is publicly available at https://github.com/modichirag/flowpm[Formula presented].

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS1992

Electron acceleration in Tycho's and Kepler's supernova remnants - Spectral evidence of Fermi shock acceleration

First model synchrotron spectra calculated with a self-consistent nonlinear shock model of first order Fermi acceleration are presented and compared with the observed radio spectra of Tycho's and Kepler's SNR. Excellent agreement is obtained, and the correct mean spectral indices of about -0.64 are easily reproduced. The model spectra are slightly concave, with hardening toward higher energies, and there is evidence for such an effect in the data, allowing the mean magnetic field strength to be estimated in each remnant. Improvements in both theory and observation could allow accurate values of magnetic fields to be inferred from sufficiently precise integrated synchrotron spectra.

Reynolds, Stephen P.↗
NASA NTRS1994

Contributions to the Earth's obliquity rate, precession, and nutation

The precession and nutation of the Earth's equator arise from solar, lunar, and planetary torques on the oblate Earth. The mean lunar orbit plane is nearly coincident with the ecliptic plane. A small tilt out of the ecliptic is caused by planetary perturbations and the Earth's gravitational harmonic J(sub2). These planetary perturbations on the lunar orbit result in torques on the oblate Earth which contribute to precession, obliquity rate, and nutation while the J(sub 2) perturbations contribute to precession and nutation. Small additional contributions to the secular rates arise from tidal effects and planetary torques on the Earth's bulge. The total correction to the obliquity rate is -0.024sec/century, it is an observable motion in space (the much larger conventional obliquity rate is wholly from the motion of the ecliptic, not the equator), and it is not present in the IAU-adopted expressions for the orientation of the Earth's equator. The J(sub2) effects have generally been allowed for in past nutation theories and some procession theories. For the planetary effect, the contributions to the 18.6 yr nutation are -0.03 mas (milliarcseconds) for the in-phase Delta phi plus out-of-phase contributions of 0.14 mas in Delta phi and -0.03 mas in Delta epsilon. The latter terms demonstrate that out-of-phase contributions can arise by means other than dissipation. The sum of the contributions to the precession rate is considered and the inferred value of the moment of inertia combination (C-A)/C, which is used to scale the coefficients in the nutation series, is evaluated. Using an updated value for the precession rate, the rigid body (C-A)/C = 0.003 273 763 4 which, in combination with a satellite-derived J(sub2), gives a normalized polar moment of inertia C/MR(exp2) = 0.330 700 7. The planetary contributions to the precession and obliquity rates are not constant for long times causing accelerations in both quantities. Acceleration in precession also arises from tides and changing J(sub2). Contributions from the improved theory, masses, ecliptic motion, and measured values of the precession rate and obliquity are combined to give expressions (polynomials in time) for precession, obliquity, and Greenwich Mean Sidereal Time.

Williams, James G.↗
DOE OSTI2024

Predicting Missing Regions in Charged Particle Tracks Using a Sparse 3D Convolutional Neural Network

The 2x2 Demonstrator is a prototype detector for the Deep Underground Neutrino Experiment (DUNE)'s Near Detector. Both the 2x2 Demonstrator and the Near Detector itself will have inactive regions wherein there is no sensitivity to charge deposition and light signals that arise from charged particle interactions with liquid argon. In the 2x2, these inactive regions are positioned in-between the active detector modules, which introduces the challenge of inferring what charge signals ought to look like in these regions. This study explores the use of a Sparse 3D Convolutional Neural Network (ConvNet) to infer missing regions in charged particle tracks. Hits corresponding to energy depositions are voxelized into a three-dimensional (3D) grid for each track. Inactive regions within the tracks are replaced with a dense, rectangular 3D grid of voxels, ensuring consistent step sizes in X, Y, and Z directions. Voxels in these dense regions are initialized with an energy value of -1, indicating nonphysical energy or charge. The model is trained to predict which voxels should activate as part of the track and which should not, with the goal of eventually inferring the missing charge or energy values in these voxels. Results indicate that the model accurately predicts track voxels within ±1 unit in X, Y, or Z directions and effectively identifies non-track voxels, despite some overprediction. The approach shows promise in prediction of missing track regions with some accuracy.

43 PARTICLE ACCELERATORS↗
NASA NTRS2018

Expansion and Age of the X-Ray Synchrotron-Dominated Supernova Remnant G330.2+1.0

We report new Chandra observations of one of the few Galactic supernova remnants whose X-ray spectrum is dominated by nonthermal synchrotron radiation, G330.2+1.0. We find that between 2006 and 2017, some parts of the shell have expanded by about 1%, giving a free-expansion (undecelerated) age of about 1000 yr, and implying shock velocities there of 9000 km/s for a distance of 5 kpc. Somewhat slower expansion is seen elsewhere around the remnant periphery, in particular in compact knots. Because some deceleration must have taken place, we infer that G330.2+1.0 is less than about 1000 yr old. Thus, G330.2+1.0 is one of only four Galactic core-collapse remnants of the last millennium. The large size, low brightness, and young age require a very low ambient density, suggesting expansion in a stellar-wind bubble. We suggest that in the east, where some thermal emission is seen and expansion velocities are much slower, the shock has reached the edge of the cavity. The high shock velocities can easily accelerate relativistic electrons to X-ray-emitting energies. A few small regions show highly significant brightness changes by 10%–20%, both brightening and fading, a phenomenon previously observed in only two supernova remnants, indicating strong and/or turbulent magnetic fields.

Borkowski, Kazimierz J.↗
NASA NTRS1994

Contributions to the Earth's Obliquity Rate, Precession, and Nutation

The precession and nutation of the Earth's equator arise from solar, lunar, and planetary torques on the oblate Earth. The mean lunar orbit plane is nearly coincident with the ecliptic plane. A small tilt out of the ecliptic is caused by planetary perturbations and the Earth's gravitational harmonic J(sub 2). These planetary perturbations on the lunar orbit result in torques on the oblate Earth which contribute to precession, obliquity rate, and nutation while the J(sub 2) perturbations contribute to precession and nutation. Small additional contributions to the secular rates arise from tidal effects and planetary torques on the Earth's bulge. The total correction to the obliquity rate is -0.024 sec/century, it is an observable motion in space (the much larger conventional obliquity rate is wholly from the motion of the ecliptic, not the equator), and it is not present in the IAU-adopted expressions for the orientation of the Earth's equator. The effects have generally been allowed for in past nutation theories and some precession theories. For the planetary effect, the contributions to the 18.6 yr nutation are -0.03 mas (milliarcseconds) for the in-phase Delta(psi) plus out-of-phase contributions of 0.14 mas in Delta(psi) and -0.03 mas in Delta(sub epsilon). The latter terms demonstrate that out-of-phase contributions can arise by means other than dissipation. The sum of the contributions to the precession rate is considered and the inferred value of the moment of inertia combination (C-A)/C, which is used to scale the coefficients in the nutation series, is evaluated. Using an updated value for the precession rate, the rigid body (C-A)/C =0.003 273 763 4 which, in combination with a satellite-derived J(sub 2), gives a normalized polar moment of inertia C/MR(exp 2) = 0.330 700 7. The planetary contributions to the precession and obliquity rates are not constant for long times causing accelerations in both quantities. Acceleration in precession also arises from tides and changing J(sub 2) Contributions from the improved theory, masses, ecliptic motion, and measured values of the precession rate and obliquity are combined to give expressions (polynomials in time) for precession, obliquity, and Greenwich Mean Sidereal Time.

Williams, James G.↗
NASA NTRS1984

A detailed X-ray and radio comparison of Kepler's supernova remnant

High-resolution radio images of Kepler's SNR (SNR (SN 1604) at 6 and 21 cm have been obtained using the VLA. The orientation of linear polarization in the SNR implies a substantial nonrandom radial component for the magnetic field of the remnant. Based on a new X-ray picture of the SNR, obtained using the Einstein Observatory, a detailed comparison of the X-ray and radio morphology is carried out. There is a good correlation between rotation measure, obtained from the radio maps, and emission measure, an X-ray-derived quantity in the SNR shell. From the mean electron density value infered from an analysis of X-ray fluxes, a mean radial magnetic field of 14 micro G and a total magnetic field of 70 microG are estimated. The energy density of relativistic electrons in Kepler's SNR in the shell comparable to the energy density in the magnetic field; the relativistic electron-pressure is 2 percent of the thermal gas pressure. The strong correlation between radio emissivity and X-ray emissivity found over the SNR in somewhat large scale can be ascribed to acceleration of thermal electrons at the shock wave and insignificant diffusion of resultant relativistic electrons.

Matsui, Y.↗
NASA NTRS2012

Constraints on Cosmic Rays, Magnetic Fields, and Dark Matter from Gamma-ray Observations of the Coma Cluster of Galaxies with VERITAS and FERMI

Observations of radio halos and relics in galaxy clusters indicate efficient electron acceleration. Protons should likewise be accelerated and, on account of weak energy losses, can accumulate, suggesting that clusters may also be sources of very high energy (VHE; E greater than100 GeV) gamma-ray emission. We report here on VHE gamma-ray observations of the Coma galaxy cluster with the VERITAS array of imaging Cerenkov telescopes, with complementing Fermi Large Area Telescope observations at GeV energies. No significant gamma-ray emission from the Coma Cluster was detected. Integral flux upper limits at the 99 confidence level were measured to be on the order of (2-5) x 10(sup -8) photons m(sup -2) s(sup -1) (VERITAS,greater than 220 GeV) and approximately 2 x 10(sup -6) photons m(sup -2) s(sup -1) (Fermi, 1-3 GeV), respectively. We use the gamma-ray upper limits to constrain cosmic rays (CRs) and magnetic fields in Coma. Using an analytical approach, the CR-to-thermal pressure ratio is constrained to be less than 16% from VERITAS data and less than 1.7% from Fermi data (averaged within the virial radius). These upper limits are starting to constrain the CR physics in self-consistent cosmological cluster simulations and cap the maximum CR acceleration efficiency at structure formation shocks to be 50. Alternatively, this may argue for non-negligible CR transport processes such as CR streaming and diffusion into the outer cluster regions. Assuming that the radio-emitting electrons of the Coma halo result from hadronic CR interactions, the observations imply a lower limit on the central magnetic field in Coma of approximately (2-5.5)microG, depending on the radial magnetic field profile and on the gamma-ray spectral index. Since these values are below those inferred by Faraday rotation measurements in Coma (for most of the parameter space), this renders the hadronic model a very plausible explanation of the Coma radio halo. Finally, since galaxy clusters are dark matter (DM) dominated, the VERITAS upper limits have been used to place constraints on the thermally averaged product of the total self-annihilation cross section and the relative velocity of the DM particles, (sigma upsilon)

cosmic↗
DOE OSTI2022

Inferring safety critical events from vehicle kinematics in naturalistic driving environment: Application of deep learning Algorithms

Advances in sensing technology has enabled the collection of countless terabytes of second-by-second kinematics data. Such data provides opportunities for real-time monitoring of driving behavior and identification of safety critical events (SCEs) including crashes and near crashes. The concept of volatility is relevant in this context, which identifies instability and erratic variations in driving behavior prior to involvement in SCEs. This study utilized vehicle kinematics from a large-scale naturalistic driving data to develop a deep learning approach based on 1D convolutional neural networks (CNN) for inferring SCEs. The data are unique in the sense that such accurate pre-crash data at high fidelity are not available in traditional crash repositories. This study contributes to the literature by providing a first attempt at predicting responses to SCEs by developing deep learning-based CNN architectures using novel driving volatility based kinematic thresholds for a sample of 9553 events. The key contribution lies in developing a volatility-based CNN input layout that is acceptable to CNN schemes and represents the motion kinematics such as speed, acceleration and volatility measures. Several 1D-CNN architectures were developed using layers, numbers of convolutions, layer patterns, and kernels. Shallow and deep architectures were tested, revealing higher accuracy of shallow architectures in detecting SCEs. The optimal number of epochs were identified using an early stopping method while the CNN performance was improved by increasing the number of epochs. The ensemble CNN had the highest predictive accuracy of 95.6% for detection of crashes and near crashes, which was 2.5% higher than the optimal CNN using 20% hold out test data. The ensemble CNN also outperformed classical machine learning models and model performance reported in past studies on detection of SCEs. Finally, these results have implications for identification of safety hotspots and providing real-time alerts and warnings in connected and highly automated vehicle environment including society of automotive engineers levels 3–5.

42 ENGINEERING↗
NASA NTRS2010

Steady X-Ray Synchrotron Emission in the Northeastern Limb of SN 1006

We investigate time variations and detailed spatial structures of X-ray synchrotron emission in the northeastern limb of SN 1006, using two Chandra observations taken in 2000 and 2008. We extract spectra from a number of small ([approx]10'') regions. After taking account of proper motion and isolating the synchrotron from the thermal emission, we study time variations in the synchrotron emission in the small regions. We find that there are no regions showing strong flux variations. Our analysis shows an apparent flux decline in the overall synchrotron flux of [approx]4% at high energies, but we suspect that this is mostly a calibration effect, and that flux is actually constant to [approx]1%. This is much less than the variation found in other remnants where it was used to infer magnetic-field strengths up to 1 mG. We attribute the lack of variability to the smoothness of the synchrotron morphology, in contrast to the small-scale knots found to be variable in other remnants. The smoothness is to be expected for a Type Ia remnant encountering uniform material. Finally, we find a spatial correlation between the flux and the cutoff frequency in synchrotron emission. The simplest interpretation is that the cutoff frequency depends on the magnetic-field strength. This would require that the maximum energy of accelerated electrons is not limited by synchrotron losses, but by some other effect. Alternatively, the rate of particle injection and acceleration may vary due to some effect not yet accounted for, such as a dependence on shock obliquity.

Katsuda, Satoru↗
DOE OSTI2026

MIND-MAC: Multi-Level In-memory Quasi Non-Destructive MAC Operation in Compact 2T-nC FeRAM for Efficient DNN Accelerator

We present MIND-MAC, a compact 2T-nC FeRAM architecture that performs multi-level, quasi-non-destructive in-memory multiply–accumulate (MAC) for deep neural networks. By exploiting voltage-controlled partial domain switching in MFM capacitors and read-transistor amplification, the cell stores multi-bit weights and gates bit-serial inputs to produce an accumulated current on shared lines. We combine TCAD-extracted parasitics with experimentally calibrated ferroelectric models in SPICE to validate device-/circuit-level behavior, and validate multi-level sensing and QNRO with measurements on a fabricated 2T-3C test vehicle. An analytical system model maps MIND-MAC to a 6-GB main-memory in-memory compute (IMC) architecture and benchmarks VGG13 inference in 61.08 ms at 964.99 mJ. Results indicate high density, reduced rewrite overhead, and energy efficiency, positioning 2T-nC FeRAM as a promising IMC candidate for next-generation AI hardware.

36 MATERIALS SCIENCE↗
DOE OSTI2023

AutoReP: Automatic ReLU Replacement for Fast Private Network Inference

The proliferation of the Machine-Learning-As-A-Service (MLaaS) market has brought to light a number of clients’ data privacy and security concerns. One promising solution is private inference (PI) techniques using cryptographic primitives. These techniques often come with high computation and communication overhead associated with the non-linear operator such as ReLU. Several approaches have been developed in reducing the number of ReLU operations, however, they either require a heuristic threshold selection or introduce significant accuracy drop. This work presents AutoReP, a gradient-based framework for non-linear operators reduction that aims to mitigate these concerns from a systematic perspective. AutoReP automates the process of discrete selection of ReLU and polynomial functions on neurons to accelerate PI applications. We also introduce distribution-aware polynomial approximation (DaPa) to accurately approximate ReLUs under given distribution, preserving model expressivity. Our experimental results demonstrate significant accuracy improvements of 6.12% (94.31%, 12.9K ReLU budget, CIFAR-10), 8.39% (74.92%, 12.9K ReLU budget, CIFAR-100), and 9.45% (63.69%, 55K ReLU budget, Tiny-ImageNet) over current state-of-the-art methods, e.g., SNL. Morever, AutoReP is applied to EfficientNet-B2 on ImageNet dataset, and achieved 75.55% accuracy with 176.1 × ReLU budget reduction.

Peng, Hongwu↗
DOE OSTI2023

hIPPYlib-MUQ: A Bayesian Inference Software Framework for Integration of Data with Complex Predictive Models under Uncertainty

Bayesian inference provides a systematic framework for integration of data with mathematical models to quantify the uncertainty in the solution of the inverse problem. However, the solution of Bayesian inverse problems governed by complex forward models described by partial differential equations (PDEs) remains prohibitive with black-box Markov chain Monte Carlo (MCMC) methods. We present hIPPYlib-MUQ, an extensible and scalable software framework that contains implementations of state-of-the art algorithms aimed to overcome the challenges of high-dimensional, PDE-constrained Bayesian inverse problems. These algorithms accelerate MCMC sampling by exploiting the geometry and intrinsic low-dimensionality of parameter space via derivative information and low rank approximation. The software integrates two complementary open-source software packages, hIPPYlib and MUQ. hIPPYlib solves PDE-constrained inverse problems using automatically-generated adjoint-based derivatives, but it lacks full Bayesian capabilities. MUQ provides a spectrum of powerful Bayesian inversion models and algorithms, but expects forward models to come equipped with gradients and Hessians to permit large-scale solution. By combining these two complementary libraries, we created a robust, scalable, and efficient software framework that realizes the benefits of each and allows us to tackle complex large-scale Bayesian inverse problems across a broad spectrum of scientific and engineering disciplines. To illustrate the capabilities of hIPPYlib-MUQ, we present a comparison of a number of MCMC methods available in the integrated software on several high-dimensional Bayesian inverse problems. These include problems characterized by both linear and nonlinear PDEs, various noise models, and different parameter dimensions. The results demonstrate that large (~ 50×) speedups over conventional black box and gradient-based MCMC algorithms can be obtained by exploiting Hessian information (from the log-posterior), underscoring the power of the integrated hIPPYlib-MUQ framework.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1995

Observations of micro-turbulence in the solar wind near the sun with interplanetary scintillation

Velocity and density turbulence of solar wind were inferred from interplanetary scintillation (IPS) observations at 2.3 GHz and 8.5 GHz using a single-antenna. The observations were made during September and October in 1992 - 1994. They covered the distance range between 5 and 76 solar radii (Rs). We applied the spectrum fitting method to obtain a velocity, an axial ratio, an inner scale and a power-law spectrum index. We examined the difference of the turbulence properties near the Sun between low-speed solar wind and high-speed solar wind. Both of solar winds showed acceleration at the distance range of 10 - 30 Rs. The radial dependence of anisotropy and spectrum index did not have significant difference between low-speed and high-speed solar winds. Near the sun, the radial dependence of the inner scale showed the separation from the linear relation as reported by previous works. We found that the inner scale of high-speed solar wind is larger than that of low-speed wind.

Yamauchi, Y.↗
DOE OSTI2025

Optical Imaging of Laser-Driven Fast Electron Weibel-like Filamentation in Overcritical Density Plasma

We report on the measurement of filamented transport of laser-generated fast electron beams in near-critical density plasma. A relativistic intensity long-wave-infrared laser irradiated a hydrodynamically shaped helium gas flow at an electron density n e ≃ 10 25 m − 3 , generating a large flux of fast electrons that propagated beyond the critical surface. The beam-to-background electron density ratio was sufficiently high to drive growth of Weibel-like filamentation, which was measured by optical probing to extend up to 800 μ m with radii ∼ 10 μ m . Particle-in-cell simulations reproduce the main features of the filamentation generation, suggesting that collisionless processes are dominant in these interactions. Expansion of the filaments after formation infers a fast electron heated plasma temperature ∼ 400 eV in the overcritical density plasma. Published by the American Physical Society 2025

43 PARTICLE ACCELERATORS↗
NASA NTRS1993

Soft x ray properties of the Geminga pulsar

The ROSAT soft x ray spectrum and pulse profile of the Geminga pulsar are analyzed and interpreted in terms of thermal emission from the surface of the neutron star. The x ray spectrum appears to consist of two blackbody components with T(sub 1) = (5.2 +/- 1.0) x 10 (exp 5) K and T(sub 2) approximately 3 x 10(exp 6) K, respectively. The inferred ratio of surface areas, A(sub 2)/A(sub 1), is approximately 3 x 10(exp -5). Both components are highly modulated at the pulsar rotation period, but the harder x ray pulse is narrower, and leads the main (soft) x ray pulse by about 105 deg of phase. The soft x ray component is interpreted as photospheric cooling of much of the neutron star's surface area, while the small, hot region could be part of the much smaller polar cap heated by energetic particles flowing inward from the magnetospheric accelerator which is responsible for the production of Geminga's gamma rays. Geminga's gamma ray emission is consistent with outer-magnetosphere accelerator models for highly inclined dipoles. These predict the beaming of energetic gamma rays close enough to the star to give copious e(+/-) production in the stellar magnetic field and a large circumstellar pair density from pair inflow toward the surface. These pairs may quench radio emission, and also reflect most of the hard polar cap x rays back to the stellar surface by cyclotron resonance scattering. They are then reemitted from that much larger area at the lower temperature T(sub 1). The single-peaked nature of the x ray pulse and its energy-dependent phase suggest an off-center dipole geometry for the surface magnetic field. Under the assumption that the soft x ray emission comes from the full surface of a neutron star of radius R = 10 km, a distance estimate of (150-400) pc is derived. This range is consistent with the fit interstellar column density of (1.5 +/- 0.5) x 10(exp 20) cm(exp -2). Distances less than 150 pc are probably ruled out both by the lower limit on the column density, and also by the requirement that the Rayleigh-Jeans extrapolation of the soft x ray spectrum not exceed the observed blue flux of the faint optical counterpart. This distance estimate implies that Geminga's efficiency for converting spindown power into gamma-rays is near unity, and that there may be significant beaming of the gamma rays as well. These results tend to bolster the prospect that most of the unidentified high-energy gamma ray sources in the Galactic plane are pulsars, some of which may be radio quiet.

Halpern, J. P.↗