1502.06294
Stacking for machine learning redshifts applied to SDSS galaxies
Zitlau, et al
Show how all tested machine learning technique base algorithms benefit from at least one additional 'stacking' (a layering of an output which is input in a subsequent learning round). Apply the method to both unsupervised machine learning techniques based on self-organizing maps, and supervised machine learning methods based on decision trees. Explore a range of stacking architectures, such as the number of layers and the number of base learners per layer. Explore the effectiveness of stacking even when using a successful algorithm such as AdaBoost. Observe a signifiant improvement of between 1.9% and 21% on all computed metrics when stacking is applied to weak learners (such as SOM and decision trees). When applied to strong learning algorithms (such as AdaBoost) the ratio of improvement shrinks, but still remains positive and is between 0.4% and 2.5% for the explored metrics and comes at almost no additional computational cost.
1602.06305
The need for laboratory work to aid in the understanding of exoplanetary atmospheres
Forty, et al
While exoplanet science currently has a wide range of sophisticated models that can be applied to the tide of forthcoming observations, the trajectory for preparing these models for the upcoming observational challenges is unclear. Thus, the ability to maximize the insights gained from the next generation of observatories is not certain. In many cases, uncertainties in a path towards model advancement stems from insufficiencies in the laboratory data that serve as critical inputs to atmospheric physical and chemical tools. Outline a number of areas where laboratory or ab initio investigations could fill critical gaps in the ability to model exoplanet atmospheric opacities, clouds, and chemistry. Specifically highlighted are needs for : (1) molecular opacity libelists with parameters for a diversity of broadening gases, (2) extended databases for collision-induced absorption and dimer opacities, (2) high spectral resolution opacity data for relevant molecular species, (4) laboratory studies of haze and condensate formation and optical properties, (5) significantly expanded database of chemical reaction rates, and (6) measurements of gals photo-adsorption cross sections at high temperatures. Hope that by meeting these needs, the next two decades of exoplanet science can be made as productive and insightful as the previous tow decades.
1602.06306
Cosmic voids in the SDSS DR12 BOSS galaxy sample: the Alcock-Paczynski test
Mao, et al
AP test on SDSS DR12: also use 1000 mock galaxy catalogs that match the geometry, density and clustering properties of the BOSS sample to characterize the statistical uncertainties of the measurements and take into account systematic errors such as RSD. For both BOSS data and mock catalogs, use ZOBOV algorithm to identify voids, stack together all voids with effective radii of 30-100 Mpc/h in the redshift range 0.43-0.7. Accurately measure the shape of the stacked voids. Tests with the mock catalogs show that the stacked void ellipticity is measured with a precision of 2.6%. Find that the stacked voids in redshift space are slightly squashed long the LoS, which is consistent with previous studies. Repeat this measurement of stacked void shape the BOSS data assuming several values of Omega_m within the flat LCDM model, and we compare to the mock catalogs in redshift space in order to perform the AP test. Obtain a constraint of Omega_m=0.38+0.18-0.15 at the 68% CL from the AP test. Discuss the various sources of statistical and systematic noise that affect the constraining power of this method. In particular, find that the measured ellipticity of stacked voids scales more weakly with cosmology than the standard AP prediction, leading to significantly weaker constraints. Discuss how AP constraints will improve in future surveys with larger volumes and densities.
1602.06711
3D weak lensing: modified theories of gravity
Pratten, et al
WL promises to be a particularly sensitive probe of both the growth of LSS as well as the fundamental relation between matter density perturbations and metric perturbations, thus providing a powerful tool with which we may constrain modified theories of gravity (MG) on cosmological scales. Future deep, wide-field WL surveys will provide an unprecedented opportunity to constrain deviations from GR. Employing a 3D analysis based on the spherical Fourier-Bessel (sFB) expansion, investigate the extent to which MG theories will be constrained by a typical 3D WL survey configurations including noise from the intrinsic ellipticity distribution sigma_epsilon of source galaxies. Focus on two classes of screened theories of gravity: i) f(R) chameleon models and ii) environmentally dependent dilation models. Use one-loop perturbation theory combined with halo models in order to accurately model the evolution of matter power-spectrum with redshift in these theories. Using a Fisher information matrix based approach, show that for an all-sky spectroscopic survey, the parameter f_R0 can be constrained in the range f_R0<5e-6(9e-6) for n=1(2) with a 3 sigma CL. This can be achieved by using relatively low order angular harmonics ell<100. Including higher order harmonics ell>100 an further tighten the constraints, making them comparable to current solar-system constraints. Also employ PCA in order to study the parameter degeneracies in the MG parameters. Results can trivially be extended to other MG theories, such as the K-mouflage models. The confusion from IA correlation and modification of the matter power-spectrum at small scale due to feedback mechanisms is briefly discussed.
1602.06854
Computer-generated visual morphology catalog of ~3,000,000 SDSS galaxies
Kuminski, Shamir
Agreement rate of ~98% with Galaxy Zoo debased 'super clean' dataset. Find ~900,000 spiral galaxies, and ~600,000 elliptical galaxies.
No comments:
Post a Comment