arXiv subjects

Abhirup Datta

Publications and source records attributed to Abhirup Datta.

153 records · Page 9Linked to original sources

arXiv2016

Non-separable Dynamic Nearest-Neighbor Gaussian Process Models for Large spatio-temporal Data With an Application to Particulate Matter Analysis

Particulate matter (PM) is a class of malicious environmental pollutants known to be detrimental to human health. Regulatory efforts aimed at curbing PM levels in different countries often require high resolution space-time maps that can identify red-flag regions exceeding statutory concentration limits. Continuous spatio-temporal Gaussian Process (GP) models can deliver maps depicting predicted PM levels and quantify predictive uncertainty. However, GP based approaches are usually thwarted by computational challenges posed by large datasets. We construct a novel class of scalable Dynamic Nearest Neighbor Gaussian Process (DNNGP) models that can provide a sparse approximation to any spatio-temporal GP (e.g., with non-separable covariance structures). The DNNGP we develop here can be used as a sparsity-inducing prior for spatio-temporal random effects in any Bayesian hierarchical model to deliver full posterior inference. Storage and memory requirements for a DNNGP model are linear in the size of the dataset thereby delivering massive scalability without sacrificing inferential richness. Extensive numerical studies reveal that the DNNGP provides substantially superior approximations to the underlying process than low rank approximations. Finally, we use the DNNGP to analyze a massive air quality dataset to substantially improve predictions of PM levels across Europe in conjunction with the LOTOS-EUROS chemistry transport models (CTMs).

stat.AP↗

arXiv2016

Effects Of The Ionosphere On Ground-Based Detection Of The Global 21 CM Signal From The Cosmic Dawn And The Dark Ages

Detection of the global HI 21 cm signal from Cosmic Dawn and Epoch of Reionization is the key science driver for several ongoing ground-based and future ground/space-based experiments. The crucial spectral features in the global 21 cm signal (turning points) occur at low radio frequencies <100 MHz. In addition to the human-generated RFI, Earth's ionosphere drastically corrupts low-frequency radio observations from the ground. In this paper, we examine the effects of time-varying ionospheric refraction, absorption and thermal emission at these low radio frequencies and their combined effect on any ground-based global 21 cm experiment. It should be noted that this is the first study of the effect of a dynamic ionosphere on global 21 cm experiments. The fluctuations in the ionosphere are influenced by solar activity with flicker noise characteristics. The same characteristics are reflected in the ionospheric corruption to any radio signal passing through the ionosphere. As a result, any ground based observations of the faint global 21 cm signal are corrupted by flicker noise (or "$1/f$" noise, where "$f$" is the dynamical frequency) which scales as $ν^{-2}$ (where $ν$ is the frequency of observation) in the presence of a bright galactic foreground ($\propto ν^{-s}$, where $s$ is radio spectral index). Hence, the calibration of the ionosphere for any such experiment is critical. Any attempt to calibrate the ionospheric effects will be subject to the inaccuracies in the current ionospheric measurements using GPS ionospheric measurements, riometer measurements, ionospheric soundings, etc. Even considering an optimistic improvement in the accuracy of GPS-TEC (Total Electron Content) measurements, we conclude that the detection of the global 21 cm signal below 100 MHz is best done from above the Earth's atmosphere in orbit of the Moon.

astro-ph.CO↗

arXiv2016

Hierarchical Nearest-Neighbor Gaussian Process Models for Large Geostatistical Datasets

Spatial process models for analyzing geostatistical data entail computations that become prohibitive as the number of spatial locations become large. This manuscript develops a class of highly scalable Nearest Neighbor Gaussian Process (NNGP) models to provide fully model-based inference for large geostatistical datasets. We establish that the NNGP is a well-defined spatial process providing legitimate finite-dimensional Gaussian densities with sparse precision matrices. We embed the NNGP as a sparsity-inducing prior within a rich hierarchical modeling framework and outline how computationally efficient Markov chain Monte Carlo (MCMC) algorithms can be executed without storing or decomposing large matrices. The floating point operations (flops) per iteration of this algorithm is linear in the number of spatial locations, thereby rendering substantial scalability. We illustrate the computational and inferential benefits of the NNGP over competing methods using simulation studies and also analyze forest biomass from a massive United States Forest Inventory dataset at a scale that precludes alternative dimension-reducing methods.

stat.ME↗

arXiv2016

CoCoLasso for High-dimensional Error-in-variables Regression

Much theoretical and applied work has been devoted to high-dimensional regression with clean data. However, we often face corrupted data in many applications where missing data and measurement errors cannot be ignored. Loh and Wainwright (2012) proposed a non-convex modification of the Lasso for doing high-dimensional regression with noisy and missing data. It is generally agreed that the virtues of convexity contribute fundamentally the success and popularity of the Lasso. In light of this, we propose a new method named CoCoLasso that is convex and can handle a general class of corrupted datasets including the cases of additive measurement error and random missing data. We establish the estimation error bounds of CoCoLasso and its asymptotic sign-consistent selection property. We further elucidate how the standard cross validation techniques can be misleading in presence of measurement error and develop a novel corrected cross-validation technique by using the basic idea in CoCoLasso. The corrected cross-validation has its own importance. We demonstrate the superior performance of our method over the non-convex approach by simulation studies.

math.ST↗

arXiv2015

Bayesian Inference for High Dimensional Changing Linear Regression with Application to Minnesota House Price Index Data

In many applications, the dataset under investigation exhibits heterogeneous regimes that are more appropriately modeled using piece-wise linear models for each of the data segments separated by change-points. Although there have been much work on change point linear regression for the low dimensional case, high-dimensional change point regression is severely underdeveloped. Motivated by the analysis of Minnesota House Price Index data, we propose a fully Bayesian framework for fitting changing linear regression models in high-dimensional settings. Using segment-specific shrinkage and diffusion priors, we deliver full posterior inference for the change points and simultaneously obtain posterior probabilities of variable selection in each segment via an efficient Gibbs sampler. Additionally, our method can detect an unknown number of change points and accommodate different variable selection constraints like grouping or partial selection. We substantiate the accuracy of our method using simulation experiments for a wide range of scenarios. We apply our approach for a macro-economic analysis of Minnesota house price index data. The results strongly favor the change point model over a homogeneous (no change point) high-dimensional regression model.

math.ST↗

arXiv2015

Trajectory Design From GTO To Lunar Equatorial Orbit For The Dark Ages Radio Explorer (DARE) Spacecraft

The trajectory design for the Dark Ages Radio Explorer (DARE) mission con-cept involves launching the DARE spacecraft into a geosynchronous transfer orbit (GTO) as a secondary payload. From GTO, the spacecraft then transfers to a lunar orbit that is stable (i.e., no station-keeping maneuvers are required with minimum perilune altitude always above 40 km) and allows for more than 1,000 cumulative hours for science measurements in the radio-quiet region located on the lunar farside.

astro-ph.IM↗

arXiv2014

How Much Can We Learn From A Merging Cold Front Cluster? : Insights From X-ray Temperature and Radio Maps of Abell 3667

The galaxy cluster Abell 3667 is an ideal laboratory to study the plasma processes in the intracluster medium (ICM). High resolution Chandra X-ray observations show a cold front in Abell 3667. At radio wavelengths, Abell 3667 reveals a double radio-relic feature in the outskirts of the cluster. These suggest multiple merger events in this cluster. In this paper, we analyze the substantial archival X-ray observations of Abell 3667 from ChandraX-ray Observatory and compare these with existing radio observations as well as state-of-the-art AMR (Adaptive Mesh Refinement) MHD cosmological simulations using Enzo. We have used two temperature map making techniques, Weighted Voronoi Tessellation and Adaptive Circular Binning, to produce the high resolution and largest field-of-view temperature maps of Abell 3667. These high fidelity temperature maps allow us to study the X-ray shocks in the cluster using a new 2-dimensional shock-finding algorithm. We have also estimated the Mach numbers from the shocks inferred from previous ATCA radio observations. The combined shock statistics from the X-ray and radio data are in agreement with the shock statistics in a simulated MHD cluster. We have also studied the profiles of the thermodynamic properties across the cold front using 447 ksec from the combined Chandra observations on Abell 3667. Our results show that the stability of the cold front in Abell 3667 can be attributed to the suppression of the thermal conduction across the cold front by a factor of 100 - 700 compared to the classical Spitzer value.

astro-ph.HE↗

arXiv2014

X-ray Observations of Complex Temperature Structure in the Cool-core cluster Abell 85

X-ray observations were used to examine the complex temperature structure of Abell 85, a cool-core galaxy cluster. Temperature features can provide evidence of merging events which shock heat the intracluster gas. Temperature maps were made from both \textit{Chandra} and \textit{XMM-Newton} obervations. The combination of a new, long-exposure \textit{XMM} observation and an improved temperature map binning technique produced the highest fidelity temperature maps of A85 to date. Hot regions were detected near the subclusters to the South and Southwest in both the \textit{Chandra} and \textit{XMM} temperature maps. The presence of these structures implies A85 is not relaxed. The hot regions may indicate the presence of shocks. The Mach numbers were estimated to be $\sim$1.9 at the locations of the hot spots. Observational effects will tend to systematically reduce temperature jumps, so the measured Mach numbers are likely underestimated. Neither temperature map showed evidence for a shock in the vicinity of the presumed radio relic near the Southwest subcluster. However, the presence of a weak shock cannot be ruled out. There was tension between the temperatures measured by the two instruments.

astro-ph.CO↗

arXiv2011

Imaging of the CCS 22.3 GHz emission in the Taurus Molecular Cloud complex

Thioxoethenylidene (CCS) is an abundant interstellar molecule, and a good tracer of high density and evolutionary stage of dense molecular clouds. It is also a suitable candidate for Zeeman splitting observations for its high splitting factor and narrow thermal linewidths. We report here EVLA 22.3 GHz observations of three dense molecular cores TMC-1, TMC-1C and L1521B in the Taurus Molecular Cloud complex to image the CCS 2_1-1_0 transition. For all three sources, the clumpy CCS emission is most likely tracing the starless cores. However, these compact structures account for only ~ 1-13% of the integrated emission detected in single-dish observations, indicating the presence of significant large scale diffuse emission in favorable conditions for producing CCS.

astro-ph.GA↗