We consider stochastic transition matrices from large social and information
networks. For these matrices, we describe and evaluate three fast methods to
estimate one column of the matrix exponential. The methods are designed to
exploit the properties inherent in social networks, such as a power-law degree
distribution. Using only this property, we prove that one of our algorithms has
a sublinear runtime. We present further experimental evidence showing that all
of them run quickly on social networks with billions of edges and accurately
identify the largest elements of the column.
Authors' comments: 41 pages. Updated version (11/20/13) published in the proceedings of
WAW13. Update (1/19/14) fixes error in runtime bound. Update (5/3/2014)
introduces two new algorithms. Update (3/1/15) accepted for publication in
journal of Internet Math; generalizes power law degree distributions
theorems. Codes available at
http://www.cs.purdue.edu/homes/dgleich/codes/nexpokit
We present the results of a search of infrared excess candidates in a
comprehensive (29\,000 stars) magnitude limited sample of dwarf stars, spanning
the spectral range F2-K0, and brighter than V$=$15 mag. We searched the sample
within the {\em WISE} all sky survey database for objects within 1 arcsecond of
the coordinates provided by SIMBAD database and found over 9\,000 sources
detected in all {\em WISE} bands. This latter sample excludes objects that are
flagged as extended sources and those images which are affected by various
optical artifacts. For each detected object, we compared the observed W4/W2
(22$\mu$m/4.6$\mu$m) flux ratio with the expected photospheric value and
identified 197 excess candidates at 3$\sigma$. For the vast majority of
candidates, the results of this analysis represent the first reported evidence
of an IR excess. Through the comparison with a simple black-body emission
model, we derive estimates of the dust temperature, as well as of the dust
fractional luminosities. For more than 80% of the sample this temperature is
higher than 120 K, suggesting the presence of warm circumstellar dust.
Complementary observations at longer wavelengths (far-IR and sub-mm) are
required for better characterising and explaining the origin of this emission.
Authors' comments: Accepted by MNRAS. 6 pages, 8 figures. The electronic table will be
available upon request to the authors
We present radio continuum mapping, optical imaging and spectroscopy of the
newly discovered double-peaked broad-lined AGN WISE J233237.05-505643.5 at
redshift z = 0.3447. This source exhibits an FR-I and FR-II hybrid-morphology,
characterized by bright core, jet, and Doppler-boosted lobe structures in ATCA
continuum maps at 1.5, 5.6, and 9 GHz. Unlike most FR-II objects, W2332-5056 is
hosted by a disk-like galaxy. The core has a projected 5" linear radio feature
that is perpendicular to the curved primary jet, hinting at unusual and complex
activity within the inner 25 kpc. The multi-epoch optical-near-IR photometric
measurements indicate significant variability over a 3-20 year baseline from
the AGN component. Gemini-South optical data shows an unusual double-peaked
emission-line features: the centroids of the broad-lined components of H-alpha
and H-beta are blueshifted with respect to the narrow lines and host galaxy by
~ 3800 km/s. We examine possible cases which involve single or double
supermassive black holes in the system, and discuss required future
investigations to disentangle the mystery nature of this system.
Authors' comments: 15 pages in emulateapj format, including 4 tables and 8 figures. To
be published in ApJ
We calculate the angular correlation function for a sample of 170,000 AGN
extracted from the Wide-field Infrared Survey Explorer (WISE) catalog, selected
to have red mid-IR colors (W1 - W2 > 0.8) and 4.6 micron flux densities
brighter than 0.14 mJy). The sample is expected to be >90% reliable at
identifying AGN, and to have a mean redshift of z=1.1. In total, the angular
clustering of WISE-AGN is roughly similar to that of optical AGN. We
cross-match these objects with the photometric SDSS catalog and distinguish
obscured sources with (r - W2) > 6 from bluer, unobscured AGN. Obscured sources
present a higher clustering signal than unobscured sources. Since the host
galaxy morphologies of obscured AGN are not typical red sequence elliptical
galaxies and show disks in many cases, it is unlikely that the increased
clustering strength of the obscured population is driven by a host galaxy
segregation bias. By using relatively complete redshift distributions from the
COSMOS survey, we find obscured sources at mean redshift z=0.9 have a bias of b
= 2.9 \pm 0.6 and are hosted in dark matter halos with a typical mass of
log(M/M_odot)~13.5. In contrast, unobscured AGN at z~1.1 have a bias of b = 1.6
\pm 0.6 and inhabit halos of log(M/M_odot)~12.4. These findings suggest that
obscured AGN inhabit denser environments than unobscured AGN, and are difficult
to reconcile with the simplest AGN unification models, where obscuration is
driven solely by orientation.
Authors' comments: Accepted for publication in ApJ. 13 pages, 15 figures
In this paper we present a catalog which includes 141 bright candidates
($\leq10.27$ mag, V band) showing the infrared (IR) excess at 22 $\mu$m. Of
which, 38 stars are known IR excess stars or disk, 23 stars are double or
multiple stars and 4 are Be stars. While the remaining more than 70 stars are
identified as the 22 $\mu$m excess candidates in our work. The criterion of
selecting candidates is $K_s-[22]_{\mu m}$. All these candidates are selected
from \emph{WISE} All-sky data cross-correlated with \emph{Hipparcos} Main
Catalog and the likelihood-ratio technique is employed. Considering the effect
of background, we introduce the \emph{IRAS} 100 $\mu$m level to exclude the
high background. We also estimated the coincidence probability of these
sources. In addition, we presented the optical to mid-infrared SEDs and optical
images of all the candidates, and gave the observed optical spectra of 6 stars
with NAOC's 2.16-m telescope. To measure for the dust amount around each star,
the fractional luminosity is also provided. We also test whether our method of
selecting IR excess stars can be used to search for extra-solar planets, we
cross-matched our catalog with known IR-excess stars having planets but none is
matched. Finally, we give the fraction of stars showing IR-excess for different
spectral type of main-sequence stars.
Authors' comments: 45 pages, 16 figures, 4 tables. Accepted for publication in ApJS
Using time-resolved, mid-infrared data from WISE and geometric parallaxes
from HST for four Galactic RR Lyrae variables, we derive the following
Population II Period-Luminosity (PL) relations for the WISE [W1], [W2] and [W3]
bands at 3.4, 4.6 & 12 um, respectively:
M[W1] = -2.44 (+/- 0.95) x logP - 1.26 (+/- 0.25) sigma = 0.10
M[W2] = -2.55 (+/- 0.89) x logP - 1.29 (+/- 0.23) sigma = 0.10
M[W3] = -2.58 (+/- 0.97) x logP - 1.32 (+/- 0.25) sigma = 0.10
The slopes and the scatter around the fits are consistent with a smooth
extrapolation of those same quantities from previously-published K-band
observations at 2.2 um, where the asymptotic (long-wavelength) behavior is
consistent with a Period-Radius relation having a slope of 0.5. No obvious
correlation with metallicity (spanning 0.4 dex in [Fe/H]) is found in the
residuals of the four calibrating RR Lyrae stars about the mean PL regression
line.
Authors' comments: 10 pages, 5 figures, 6 data tables. Accepted to the Astrophysical
Journal
This letter reports some improvement of the formalism and analysis of our
very recent work \cite{SR1} on the studies of Isgur-Wise function for
heavy-light mesons following higher dimensional string inspired potential
model.
Large-scale regression problems where both the number of variables, $p$, and
the number of observations, $n$, may be large and in the order of millions or
more, are becoming increasingly more common. Typically the data are sparse:
only a fraction of a percent of the entries in the design matrix are non-zero.
Nevertheless, often the only computationally feasible approach is to perform
dimension reduction to obtain a new design matrix with far fewer columns and
then work with this compressed data.
$b$-bit min-wise hashing (Li and Konig, 2011) is a promising dimension
reduction scheme for sparse matrices which produces a set of random features
such that regression on the resulting design matrix approximates a kernel
regression with the resemblance kernel. In this work, we derive bounds on the
prediction error of such regressions. For both linear and logistic models we
show that the average prediction error vanishes asymptotically as long as $q
\|\beta^*\|_2^2 /n \rightarrow 0$, where $q$ is the average number of non-zero
entries in each row of the design matrix and $\beta^*$ is the coefficient of
the linear predictor.
We also show that ordinary least squares or ridge regression applied to the
reduced data can in fact allow us fit more flexible models. We obtain
non-asymptotic prediction error bounds for interaction models and for models
where an unknown row normalisation must be applied in order for the signal to
be linear in the predictors.
Authors' comments: 40 pages
We present a scalable and effective classification model to train multi-class
boosting for multi-class classification problems. Shen and Hao introduced a
direct formulation of multi- class boosting in the sense that it directly
maximizes the multi- class margin [C. Shen and Z. Hao, "A direct formulation
for totally-corrective multi- class boosting", in Proc. IEEE Conf. Comp. Vis.
Patt. Recogn., 2011]. The major problem of their approach is its high
computational complexity for training, which hampers its application on
real-world problems. In this work, we propose a scalable and simple stage-wise
multi-class boosting method, which also directly maximizes the multi-class
margin. Our approach of- fers a few advantages: 1) it is simple and
computationally efficient to train. The approach can speed up the training time
by more than two orders of magnitude without sacrificing the classification
accuracy. 2) Like traditional AdaBoost, it is less sensitive to the choice of
parameters and empirically demonstrates excellent generalization performance.
Experimental results on challenging multi-class machine learning and vision
tasks demonstrate that the proposed approach substantially improves the
convergence rate and accuracy of the final visual detector at no additional
computational cost compared to existing multi-class boosting.
Authors' comments: 12 pages
We study star formation rate (SFR) indicators for Wide-field Infrared Survey
Explorer (WISE) 22 \mu m selected, star-forming galaxies at 0.01 < z < 0.3 in
the Sloan Digital Sky Survey. Using extinction-corrected H\alpha\ luminosities
and total infrared luminosities as reference SFR estimates, we calibrate WISE
mid-infrared (MIR) related SFR indicators. Both 12 and 22 \mu m monochromatic
luminosities correlate well with the reference SFR estimates, but tend to
underestimate SFRs of metal-poor galaxies (at lower than solar metallicity),
consistent with previous studies. We mitigate this metallicity dependence using
a linear combination of observed H\alpha\ and WISE MIR luminosities for SFR
estimates. The combination provides robust SFR measurements as Kennicutt et al.
(2009) applied to Spitzer data. However, we find that the coefficient a in
L_H\alpha(obs) + a L_MIR increases with SFR, and show that a non-linear
combination of observed H\alpha\ and MIR luminosities gives the best SFR
estimates with small scatters and with little dependence on physical
parameters. Such a combination of H\alpha\ and MIR luminosities for SFR
estimates is first applied to WISE data. We provide several SFR recipes using
WISE data applicable to galaxies with 0.1 <~ SFR (M_sun yr^-1) <~ 100.
Authors' comments: 14 pages, 2 tables, 6 figures, to appear in ApJ
Two decoder structures for coded modulation over the Gaussian and flat fading
channels are studied: the maximum likelihood symbol-wise decoder, and the
(suboptimal) bit-wise decoder based on the bit-interleaved coded modulation
paradigm. We consider a 16-ary quadrature amplitude constellation labeled by a
Gray labeling. It is shown that the asymptotic loss in terms of pairwise error
probability, for any two codewords caused by the bit-wise decoder, is bounded
by 1.25 dB. The analysis also shows that for the Gaussian channel the
asymptotic loss is zero for a wide range of linear codes, including all
rate-1/2 convolutional codes.
We discuss two independent methods of solution of a master equation whose
biased jump transition rates account for long jumps of L\'{e}vy-stable type and
nonetheless admit a Boltzmannian (thermal) equilibrium to arise in the large
time asymptotics of a probability density function $\rho (x,t)$. Our main goal
is to demonstrate a compatibility of a {\it direct} solution method (an
explicit, albeit numerically assisted, integration of the master equation) with
an {\it indirect} path-wise procedure, recently proposed in [Physica {\bf A
392}, 3485, (2013)] as a valid tool for a dynamical analysis of non-Langevin
jump-type processes. The path-wise method heavily relies on an accumulation of
large sample path data, that are generated by means of a properly tailored
Gillespie's algorithm. Their statistical analysis in turn allows to infer the
dynamics of $\rho (x,t)$. However, no consistency check has been completed so
far to demonstrate that both methods are fully compatible and indeed provide a
solution of the same dynamical problem. Presently we remove this gap, with a
focus on potential deficiencies (various cutoffs, including those upon the jump
size) of approximations involved in solution protocols.
Authors' comments: 11 pages, 7 figures. text modified and expanded, to appear in Centr.
Eur. J. Phys. (2014)
The Wide-field Infrared Survey Explorer (WISE) observed 52 Centaurs and
Scattered Disk Objects in the thermal infrared, including the 15 discoveries
that were new. We present analyses of these observations to estimate sizes and
mean optical albedos. We find mean albedos of 0.08 +/- 0.04 for the entire data
set. Thermal fits yield average beaming parameters of 0.9 +/- 0.2 that are
similar for both SDO and Centaur sub-classes. Biased cumulative size
distributions yield size-frequency distribution power law indices ~ -1.7 +/-
0.3. The data also reveal a relation between albedo and color at the 3-sigma
level. No significant relation between diameter and albedos is found.
Authors' comments: Accepted by ApJ, not proofed. 4 Tables, 8 Figures
In the future International Linear Collider (ILC) experiment, high
performance tracking is essential to its physics program including precision
Higgs studies. One of major challenges for a detector such as the proposed
International Large Detector (ILD) is to provide excellent momentum resolution
in a magnetic filed with small (but non-negligible) non-uniformity. The
non-uniform magnetic field implies deviation from a helical track and hence
requires the extension of a helical track model used for track fitting in a
uniform magnetic field. In this paper, a segment-wise helical track model is
introduced as such an extension. The segment-wise helical track model
approximates the magnetic field between two nearby measurement sites to be
uniform and steps between the two sites along a helix. The helix frame is then
transformed according to the new magnetic field direction for the next step, so
as to take into account the non-uniformity of the magnetic field. Details of
the algorithm and mathematical aspects of the segment-wise helical track model
in a Kalman-filter-based track fitting in the non-uniform magnetic field are
elaborated. The new track model is implemented and successfully tested in the
framework of the Kalman filter tracking software package, KalTest, which was
originally developed for tracking in a uniform magnetic field.
Authors' comments: 21 pages, 6 figures, 2 tables. Modified according to reviewer's
comments
2735.
Nearby M, L, and T Dwarfs Discovered by the Wide-field Infrared Survey
Explorer (WISE)
Maggie A. Thompson, J. Davy Kirkpatrick, Gregory N. Mace, Michael C. Cushing, Christopher R. Gelino, Roger L. Griffith, Michael F. Skrutskie, Peter R. M. Eisenhardt et al.
In our effort to complete the census of low-mass stars and brown dwarfs in
the immediate Solar Neighborhood, we present spectra, photometry, proper
motions, and distance estimates for forty-two low-mass star and brown dwarf
candidates discovered by the Wide-field Infrared Survey Explorer (WISE). We
also present additional follow-up information on twelve candidates selected
using WISE data but previously published elsewhere. The new discoveries include
fifteen M dwarfs, seventeen L dwarfs, five T dwarfs, and five objects of other
type. Among these discoveries is a newly identified "unusually red L dwarf"
(WISE J223527.07+451140.9), four peculiar L dwarfs whose spectra are most
readily explained as unresolved L+T binary systems, and a T9 dwarf (WISE
J124309.61+844547.8). We also show that the recently discovered red L dwarf
WISEP J004701.06+680352.1 (Gizis et al. 2012) may be a low-gravity object and
hence young and potentially low mass (< 25 MJup).
Authors' comments: 87 pages with 22 figures; accepted for publication in the 01 July
2013 issue of Publications of the Astronomical Society of the Pacific
Using albedos from WISE/NEOWISE to separate distinct albedo groups within the
Main Belt asteroids, we apply the Hierarchical Clustering Method to these
subpopulations and identify dynamically associated clusters of asteroids. While
this survey is limited to the ~35% of known Main Belt asteroids that were
detected by NEOWISE, we present the families linked from these objects as
higher confidence associations than can be obtained from dynamical linking
alone. We find that over one-third of the observed population of the Main Belt
is represented in the high-confidence cores of dynamical families. The albedo
distribution of family members differs significantly from the albedo
distribution of background objects in the same region of the Main Belt, however
interpretation of this effect is complicated by the incomplete identification
of lower-confidence family members. In total we link 38,298 asteroids into 76
distinct families. This work represents a critical step necessary to debias the
albedo and size distributions of asteroids in the Main Belt and understand the
formation and history of small bodies in our Solar system.
Authors' comments: Accepted to ApJ. Full version of Table 3 to be published
electronically in ApJ
The structural properties of fluids whose molecules interact via potentials
with a hard core plus two piece-wise constant sections of different widths and
heights are presented. These follow from the more general development
previously introduced for potentials with a hard core plus $n$ piece-wise
constant sections [Condens. Matter Phys. {\bf 15}, 23602 (2012)] in which use
was made of a semi-analytic rational-function approximation method. The results
of illustrative cases comprising eight different combinations of wells and
shoulders are compared both with simulation data and with those that follow
from the numerical solution of the Percus-Yevick and hypernetted-chain integral
equations. It is found that the rational-function approximation generally
predicts a more accurate radial distribution function than the Percus-Yevick
theory and is comparable or even superior to the hypernetted-chain theory. This
superiority over both integral equation theories is lost, however, at high
densities, especially as the widths of the wells and/or the barriers increase.
Authors' comments: 10 pages, 11 figures; v2: Old Fig. 1 removed, new text on the
correlation length, 7 new references added, plus other minor changes
This paper considers belief propagation algorithm over pair-wise graphical
models to develop low complexity, iterative multiple-input multiple-output
(MIMO) detectors. The pair-wise graphical model is a bipartite graph where a
pair of variable nodes are related by an observation node represented by the
bivariate Gaussian function obtained by marginalizing the posterior joint
probability density under the Gaussian input assumption. Specifically, we
consider two types of pair-wise models, the fully connected and ring-type. The
pair-wise graphs are sparse, compared to the conventional graphical model in
[18], insofar as the number of edges connected to an observation node (edge
degree) is only two. Consequently the computations are much easier than those
of maximum likelihood (ML) detection, which are similar to the belief
propagation (BP) that is run over the fully connected bipartite graph. The link
level performance for non-Gaussian input is evaluated via simulations, and the
results show the validity of the proposed algorithms. We also customize the
algorithm with Gaussian input assumption to obtain the Gaussian BP run over the
two pair-wise graphical models and, for the ring-type, we prove its convergence
in mean to the linear minimum mean square error (MMSE) estimates. Since the
maximum a posterior (MAP) estimator for Gaussian input is equivalent to the
linear MMSE estimator, it shows the optimality, in mean, of the scheme for
Gaussian input.
Authors' comments: arXiv admin note: substantial text overlap with arXiv:1011.4792
Combining the Wide-Field Infrared Survey Explorer All-Sky Release with the
2MASS Point Source Catalog, we detect a nearby, moderately metal-poor stellar
debris stream spanning 24 degrees across the southern sky. The stream, which we
designate Alpheus, is at an estimated distance of ~1.9 kpc. Its position,
orientation, width, estimated metallicity, and to some extent its distance, are
in approximate agreement with what one might expect of the leading tidal tail
of the southern globular cluster NGC 288.
Authors' comments: 8 pages, 4 figures, accepted for publication in the Astrophysical
Journal Letters
We present the first determination of absolute magnitudes for the red clump
(RC) stars with the Wide-field Infrared Survey Explorer (WISE). We used
recently reduced parallaxes taken from the Hipparcos catalogue and identified
3889 RC stars with the WISE photometry in the Solar neighbourhood. Mode values
estimated from the distributions of absolute magnitudes and a colour of the RC
stars in WISE photometry are M_{W1}=-1.635(0.026), M_{W3}=-1.606(0.024) and
(W1-W3)_0=-0.028(0.001) mag. These values are consistent with those obtained
from the transformation formulae using 2MASS data. Distances of the RC stars
estimated by using their M_{W1} and M_{W3} absolute magnitudes are in agreement
with the ones calculated by the spectrophotometric method, as well. These WISE
absolute magnitudes can be used in astrophysical researches where distance
plays an important role.
Authors' comments: 21 pages, 11 figures and 3 tables, accepted for publication in New
Astronomy
