Automatic Determination of Stellar Atmospheric Parameters Using Neural Networks and Instance-Based Machine Learning

In this article we show how machine learning methods can beeffectively applied to the problem of automatically predictingstellar atmospheric parameters from spectral information, a veryimportant problem in stellar astronomy. We apply feedforwardneural networks, Kohonen's self-organizing maps andlocally-weighted regression to predict the stellar atmosphericparameters effective temperature, surface gravity and metallicityfrom spectral indices. Our experimental results show that thethree methods are capable of predicting the parameters with verygood accuracy. Locally weighted regression gives slightly betterresults than the other methods using the original dataset asinput, while self-organizing maps outperform the other methods when significant amounts of noise are added. We also implemented a heterogeneous ensemble of predictors, combining the results given by the three algorithms. This ensemble yields better results than any of the three algorithms alone, using both the original and the noisy data.     

Boundary curves for families of planar orbits

For monoparametric familiesf(x,y)=c of planar orbits, created by a planar potentialV(x,y), we introduce the notion of the family boundary curves (FBC). All members of the familyf(x,y)=c are traced in an allowable region of thexy plane, defined by the corresponding FBC, with total energyE=E(c) varying along the family. Family boundary curves are also found for two-parametric familiesf(x,y,b)=c. The relation of equilibrium points and asymptotic orbits, possibly possessed by the potentialV(x,y), to be FBC is studied.     

HST/NICMOS Observations of M82

The irregular galaxy M82 is known as the archetypal starburst galaxy. Its proximity (3.5 Mpc) makes this galaxy an ideal laboratory for studying the properties of its starburst. The detailed morphology of the [FeII] 1.644 μm and emission Pa_α (at 1.87 μm) is revealed by the NICMOS images. The peak of the 2.2 μm continuum brightness(evolved population) lies very close to the dynamical centre. Most of the Pa_α emission (which traces the young population) is distributed in a ring of star formation (with a `hole' lacking line emission at the centre of the galaxy). These observations support the scenario in which the starburst in M82 is propagating outwards. It has long been suggested that the [FeII] emission in starburst galaxies can be used as a measure of supernova (SN) activity. M82 shows a large number of radio supernova remnants (SNRs), approximately 50, lying in the plane of the galaxy. The comparison of the positions of the bright compact [FeII] emitting regions with the location of the radio SNRs shows that there is no one-to-one spatial correspondence between the two emissions, suggesting that the radio and [FeII] emissions trace two populations of SNRs with different ages. Young (a few hundred years) SNRs are best traced by their radio emission, whereas the [FeII] stage lasts for at least a few 10⁴ yr. The compact [FeII] sources contribute only some 20 % of the total [FeII] emission observed in M82. However, much of the remaining unresolved [FeII] emission in the plane of the galaxy may arise from SNRs that expanded and merged into a general interstellar medium within a few 10⁴ yr. Presumably, as much as 70% of the total extinction-corrected [FeII]1.644 μm in M82 is associated with SNRs. The extended and diffuse [FeII] component in M82 seems to be related with the superwind above and below the disc of the galaxy. This revised version was published online in July 2006 with corrections to the Cover Date.     

Accretion disk spectra of super-luminal jet sources and ultra-luminous compact X-ray sources in nearby spiral galaxies

The Ultra-luminous Compact X-ray Sources (ULXs)in nearby spiral galaxies and the Galactic super-luminaljet sources sharethe common spectral characteristic that they haveextremely high disk temperatures which cannot be explainedin the framework of the standard accretion disk modelin the Schwarzschild metric. We have calculated an extreme Kerr disk model to examine if the Kerr disk model can instead explain the observed `too hot' accretion disk spectra.We found that the Kerr disk spectrum becomes significantly hardercompared to the Schwarzschild disk only when the disk is highlyinclined.For super-luminal jet sources, which are known to beinclined systems, the Kerr disk model may thuswork if we choose proper values for the black hole angular momentum. For the ULXs, however, the Kerr disk interpretation will be problematic,as is is highly unlikely that their accretion disks are preferentiallyinclined.     

Spectrophotometric Variability of Mira

Data from the Pulkovo spectrophotometric data base on the absolute quasimonochromatic fluxes from oCet in the 320–1080 nm range are used to determine the physical parameters of this star in different phases of its light curve. The continuum emission layer is found to expand between the phases of the cycle corresponding to the rising and falling branches of the light curve. The average expansion velocity is 32 km/s. By the time the star’s brightness has fallen by roughly three magnitudes, its radius has increased by almost a factor of three. Over this same time the temperature of the layer has fallen from 3000 K to 2200 K. For this expansion velocity, the calculated mass rate loss is ⊙ M /year.__________Translated from Astrofizika, Vol. 48, No. 2, pp. 175–189 (May 2005).     

Polarization observations of giant radio pulses from the millisecond pulsar B1937+21 at a frequency of 600 MHz

We performed polarization observations of giant radio pulses from the millisecond pulsar B1937+21. The observations were carried out in July 2002 with the 64-m Kalyazin radio telescope at a frequency of 600 MHz in two polarization channels with left-and right-hand circular polarizations (RCP and LCP). We used the S2 data acquisition system with a time resolution of 125 ns. The duration of an observing session was 20 min. We detected twelve giant radio pulses with peak flux densities higher than 1000 Jy; five and seven of these pulses appeared in the RCP and LCP channels, respectively. We found no event that exceeded the established detection threshold simultaneously in the two polarization channels. Thus, we may conclude that the detected giant pulses have a high degree of circular polarization, with the frequency of occurrence of RCP and LCP pulses being the same.     

Structures in the phase space of a four dimensional symplectic map

We numerically investigate the projections of non periodic orbits in a 4-dimensional (4-D) symplectic map composed of two coupled 2-dimensional (2-D) maps. We describe in detail the structures that are produced in different planes of projection and we find how the morphology of the 4-D orbits is influenced by the features of the 2-D maps as the coupling parameter increases. We give an empirical law that describes this influence.     

Some Problems with Slow Rotation of CP Stars

Some difficulties in explaining the slow rotation of CP stars are discussed. The most likely hypotheses are (1) a loss of angular momentum involving a magnetic field during “pre-main sequence” evolution and (2) the slow rotation existed from the very start of the creation of these stars. The braking hypothesis is supported by only one property of CP stars— the lower the mass of the star is, the greater the difference between its average rotation velocity vsini and that of normal stars. On the other hand, there is another property— the lower the rotation speeds of CP stars are, the greater their fraction among normal stars. The latter property supports the hypothesis that the lower the initial rotation speed of a star is when it is created, the greater the probability will become chemically peculiar. This property is inherent in chemically peculiar stars both with and without a magnetic field. It is proposed that the cause of the slow rotation of CP stars must be sought in the very earliest phases of their formation, as should the cause of the separation into chemically peculiar magnetic, chemically peculiar nonmagnetic, and normal stars.__________Translated from Astrofizika, Vol. 48, No. 2, pp. 229–245 (May 2005).     

Simulation of double cold cores of the 35°N section in the Yellow Sea with a wave-tide-circulation coupled model

Based on the MASNUM wave-tide-circulation coupled numerical model, the temperature structure along 35°N in the Yellow Sea was simulated and compared with the observations. One of the notable features of the temperature structure along 35°N section is the double cold cores phenomena during spring and summer. The double cold cores refer to the two cold water centers located near 122°E and 125°E from the depth of 30m to bottom. The formation, maintenance and disappearance of the double cold cores are discussed. At least two reasons make the temperature in the center (near 123°E) of the section higher than that near the west and east shores in winter. One reason is that the water there is deeper than the west and east sides so its heat content is higher. The other is invasion of the warm water brought by the Yellow Sea Warm Current (YSWC) during winter. This temperature pattern of the lower layer (from 30m to bottom) is maintained through spring and summer when the upper layer (0 to 30m) is heated and strong thermocline is formed. Large zonal span of the 35°N section (about 600 km) makes the cold cores have more opportunity to survive. The double cold cores phenomena disappears in early autumn when the west cold core vanishes first with the dropping of the thermocline position. Supported by the National Basic Research Program of China (No. G1999043809) and the National Science Foundation of China (No. 49736190).