多任务Lasso回归法在恒星光谱物理参量估计中的应用

多任务学习方法在机器学习、计算机视觉、人工智能领域已得到广泛关注,利用任务间的相关性,将多个任务同时学习的效果优于每个任务单独学习的情况.采用多任务Lasso回归法(Multi-task Lasso Regression)用于恒星光谱物理参量的估计,不仅可以获取不同物理参量间的共同的特征信息,而且也可以很好地保留不同物理参量的特有的补充信息.使用恒星大气模拟模型合成光谱库ELODIE中的光谱数据和美国大型巡天项目Sloan发布的SDSS实测光谱数据进行实验,模型估算精度优于相关文献中的方法,特别是对重力加速度(lg g)和化学丰度([Fe/H])的估计.实验中通过改变光谱的分辨率,施加不同信噪比(SNR)的噪声,来说明模型的稳定性强.结果表明,模型精度受光谱分辨率和噪声的影响,但噪声对其影响更大,可见,多任务Lasso回归法不仅操作简便,稳定性强,而且也提高了模型的整体预测精度.     

Doppler tomography of accretion regions in algols

The technique of Doppler Tomography has been used to image the accretion regions in five short-period Algols. There is clear evidence of gas flows along the predicted free-fall path of the gas stream as well as asymmetric disk-like structures around the mass gainer. Another source of H emission is associated with the cool magnetically active star.     

Flare ribbon expansion and energy release

We report a detailed examination about the relationship between the evolution of the Hα flare ribbons and the released magnetic energy during the April 10 2001 flare. In the Hα images, several bright kernels are observed in the flare ribbons. We identified the conjugated foot-points, by analyzing the lightcurves at each Hα kernels, and showed their connectivities during the flare. Then, based on the magnetic reconnection model, we calculated quantitatively the released energy by using the photospheric magnetic field strengths and separation speeds of the Hα flare ribbons. Finally, we examined the downward motions which are observed at the Hα kernels. We found that the stronger the red-asymmetry tends to be associated with the brighter the Hα kernel.     

Direct Determination of Galaxy Luminosity Functions in Clusters

We present R-band galaxy luminosity functions (GLFs) from aspectroscopic sample of six nearby rich galaxy clusters. In addition to individual cluster GLFs, extending to, in one case, M _R=–14, we also present composite GLFs for cluster and field galaxies toM _R=–17. All six cluster samples are consistent with the composite GLF, but there is evidence that the GLF of the quiescent population in clusters is not universal. Furthermore, the GLF of quiescent galaxies is significantly steeper in clusters than in the field. The overall GLF in clusters is consistent with that of field galaxies, except for the luminous tip, which is enhanced in clusters versus the field. This revised version was published online in July 2006 with corrections to the Cover Date.     

The science case of the PEPSI high‐resolution echelle spectrograph and polarimeter for the LBT

We lay out the scientific rationale for and present the instrumental requirements of a high‐resolution adaptiveoptics Echelle spectrograph with two full‐Stokes polarimeters for the Large Binocular Telescope (LBT) in Arizona. Magnetic processes just like those seen on the Sun and in the space environment of the Earth are now well recognized in many astrophysical areas. The application to other stars opened up a new field of research that became widely known as the solarstellar connection. Late‐type stars with convective envelopes are all affected by magnetic processes which give rise to a rich variety of phenomena on their surface and are largely responsible for the heating of their outer atmospheres. Magnetic fields are likely to play a crucial role in the accretion process of T‐Tauri stars as well as in the acceleration and collimation of jet‐like flows in young stellar objects (YSOs). Another area is the physics of active galactic nucleii (AGNs) , where the magnetic activity of the accreting black hole is now believed to be responsible for most of the behavior of these objects, including their X‐ray spectrum, their notoriously dramatic variability, and the powerful relativistic jets they produce. Another is the physics of the central engines of cosmic gamma‐ray bursts, the most powerful explosions in the universe, for which the extreme apparent energy release are explained through the collimation of the released energy by magnetic fields. Virtually all the physics of magnetic fields exploited in astrophysics is somehow linked to our understanding of the Sun's and the star's magnetic fields. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

Kinematical diagrams for conical relativistic jets

We present diagrams depicting the expected inter-dependences of two key kinematical parameters of radio knots in the parsec-scale jets of blazars, deduced from VLBI observations. The two parameters are the apparent speed (υ _app = cβ _app) and the effective Doppler boosting factor (δ _eff) of the relativistically moving radio knot. A novel aspect of these analytical computations of β-δ diagrams is that they are made for parsec-scale jets having a conical shape, with modest opening angles (ω up to 10°), in accord with the VLBI observations of the nuclei of the nearest radio galaxies. Another motivating factor is the recent finding that consideration of a conical geometry can have important implications for the interpretation of a variety of radio observations of blazar jets. In addition to uniform jet flows (i.e., those having a uniform bulk Lorentz factor, Γ), computational results are also presented for stratified jets where an ultra-relativistic central spine along the jet axis is surrounded by a slower moving sheath, possibly arising from a velocity shear.     

Real-time signal processor for pulsar studies

This paper describes the design, tests and preliminary results of a real-time parallel signal processor built to aid a wide variety of pulsar observations. The signal processor reduces the distortions caused by the effects of dispersion, Faraday rotation, doppler acceleration and parallactic angle variations, at a sustained data rate of 32 Msamples/sec. It also folds the pulses coherently over the period and integrates adjacent samples in time and frequency to enhance the signal-to-noise ratio. The resulting data are recorded for further off-line analysis of the characteristics of pulsars and the intervening medium. The signal processing for analysis of pulsar signals is quite complex, imposing the need for a high computational throughput, typically of the order of a Giga operations per second (GOPS). Conventionally, the high computational demand restricts the flexibility to handle only a few types of pulsar observations. This instrument is designed to handle a wide variety of Pulsar observations with the Giant Metre Wave Radio Telescope (GMRT), and is flexible enough to be used in many other high-speed, signal processing applications. The technology used includes field-programmable-gate-array(FPGA) based data/code routing interfaces, PC-AT based control, diagnostics and data acquisition, digital signal processor (DSP) chip based parallel processing nodes and C language based control software and DSP-assembly programs for signal processing. The architecture and the software implementation of the parallel processor are fine-tuned to realize about 60 MOPS per DSP node and a multiple-instruction-multiple-data (MIMD) capability.     

Using SNAP for Galactic &; extragalactic astronomy

SNAP, the SuperNova Acceleration Probe, is planned as a space-based telescope designed specifically to search for and monitor cosmological supernovae and weak lensing. In this paper we propose some other mission objectives which are of great importance in the fields of Galactic and extragalactic astronomy and which can be done as by-products with the same instrumentation and survey strategy as currently proposed for the main SNAP mission.     

The statistical analyses of flares detected in B band photometry of UV Ceti type stars

In this study, we present the unpublished flare data collected from 222 flares detected in the B band observations of five stars and the results derived by statistical analysis and modeling of these data. Six basic properties have been found with a statistical analysis method applied to all models and analyses for the flares detected in the B band observation of UV Ceti type stars. We have also compared the U and B bands with the analysis results. This comparison allowed us to evaluate the methods used in the analyses. The analyses provided the following results. (1) The flares were separated into two types, fast and slow flares. (2) The mean values of the equivalent durations of the slow and the fast flares differ by a factor of 16.2 ± 3.7. (3) Regardless of the total flare duration, the maximum flare energy can reach a different Plateau level for each star. (4) The Plateau values of EV Lac and EQ Peg are higher than the others. (5) The minimum values of the total flare duration increase toward the later spectral types. This value is called the Half-Life value in models. (6) Both the maximum flare rise times and the total flare duration obtained from the observed flares decrease toward the later spectral types.     

Impact of neutron star oscillations on the accelerating electric field in the polar cap of pulsar

Pulsar “standard model”, that considers a pulsar as a rotating magnetized conducting sphere surrounded by plasma, is generalized to the case of oscillating star. We developed an algorithm for calculation of the Goldreich-Julian charge density for this case. We consider distortion of the accelerating zone in the polar cap of pulsar by neutron star oscillations. It is shown that for oscillation modes with high harmonic numbers (l,m) changes in the Goldreich-Julian charge density caused by pulsations of neutron star could lead to significant altering of an accelerating electric field in the polar cap of pulsar. In the moderately optimistic scenario, that assumes excitation of the neutron star oscillations by glitches, it could be possible to detect altering of the pulsar radioemission due to modulation of the accelerating field. This work was partially supported by RFBR grant 04-02-16720, and by the grants N.Sh.-5218.2006.2 and RNP-2.1.1.5940.     

Consequences of clumping in hot-star winds

It seems more and more likely that one will have to abandon the paradigm of smooth outflows from (hot) stars in favour of a clumpy structure, possibly in a fractal-like hierarchy on all scales. Observationally, this is best established for Wolf-Rayet star winds (e.g. scaling laws, mass-spectrum, anisotropy,...), for which the consequences of clumping are discussed. These include four broad categories, which are outlined in this review: (a) an ideal laboratory for studying time-dependent astrophysical turbulence, (b) potential tracers of hot-wind global structure parameters, (c) reduced mass-loss rates, and (d) impact on massive binary studies.     

Unifying neutron star sub-populations in the supernova fallback accretion model

We employ the supernova fallback disk model to simulate the spin evolution of isolated young neutron stars(NSs). We consider the submergence of the NS magnetic fields during the supercritical accretion stage and its succeeding reemergence. It is shown that the evolution of the spin periods and the magnetic fields in this model is able to account for the relatively weak magnetic fields of central compact objects and the measured braking indices of young pulsars. For a range of initial parameters, evolutionary links can be established among various kinds of NS sub-populations including magnetars, central compact objects and young pulsars. Thus, the diversity of young NSs could be unified in the framework of the supernova fallback accretion model.     

Radio synthesis imaging of anisotropic angular broadening in the solar wind

We present Very Large Array observations at wavelengths of 2, 3.5, 6, and 20 cm, of angular broadening of radio sources due to the solar wind in the region 2–16 solar radii. Angular broadening is anisotropic with axial ratios in the range 2–16. Larger axial ratios are observed preferentially at smaller solar distances. Assuming that anisotropy is due to scattering blobs elongated along magnetic field lines, the distribution of position angles of the elliptically broadened images indicates that the field lines are non-radial even at the largest heliocentric distances observed here. At 5R⊙, the major axis scattering angle is ∼ 0.7" atλ= 6 cm and it varies with heliocentric distance asR ^-1.6. The level of turbulence, characterized by the wave structure function at a scale of 10 km along the major axis, normalized toλ = 20 cm, has a value 20 ± 7 at 5R⊙and varies with heliocentric distance asR ^-3. Comparison with earlier resu lts suggest that the level of turbulence is higher during solar maximum. Assuming a power-law spectrum of electron density fluctuations, the fitted spectral exponents have values in the range 2.8–3.4 for scale sizes between 2–35 km. The data suggests temporal fluctuations (of up to 10%) in the spectral exponent on a time scale of a few tens of minutes. The observed structure functions at different solar distances do not show any evidence for an inner scale; the upper limits are l k m at 2R⊙ and 4 km at 13R⊙. These upper limits are in conflict with earlier determinations and may suggest a reduced inner scale during solar maximum.     

Evidence for ablated flows in the shell of the nova DQ Herculis

High-resolution long-slit Hα spectra of the shell of the old nova DQ Herculis have been obtained with the William Herschel Telescope using the ISIS spectrograph. An equatorial expansion velocity of  370 ± 14   km s⁻¹  is derived from the spectra which, in conjunction with a narrow-band Hα image of the remnant, allows a distance estimate of 525 ± 28 pc. An equatorial ring which exhibits enhanced [N  ii ] emission has also been detected and the inclination angle of the shell is found to be     with respect to the line of sight. The spectra also reveal tails extending from the clumps in the shell, which have a radial velocity increasing along their length. This suggests the presence of a stellar wind, collimated in the polar direction, which ablates fragments of material from the clumps and accelerates them into its stream up to a terminal velocity of the order of 800–900 km s⁻¹.