arXiv subjects

Philip Graff

Publications and source records attributed to Philip Graff.

12 recordsLinked to original sources

arXiv2017

GALILEO: A Generalized Low-Entropy Mixture Model

We present a new method of generating mixture models for data with categorical attributes. The keys to this approach are an entropy-based density metric in categorical space and annealing of high-entropy/low-density components from an initial state with many components. Pruning of low-density components using the entropy-based density allows GALILEO to consistently find high-quality clusters and the same optimal number of clusters. GALILEO has shown promising results on a range of test datasets commonly used for categorical clustering benchmarks. We demonstrate that the scaling of GALILEO is linear in the number of records in the dataset, making this method suitable for very large categorical datasets.

stat.ML↗

arXiv2017

Bayesian Learning of Clique Tree Structure

The problem of categorical data analysis in high dimensions is considered. A discussion of the fundamental difficulties of probability modeling is provided, and a solution to the derivation of high dimensional probability distributions based on Bayesian learning of clique tree decomposition is presented. The main contributions of this paper are an automated determination of the optimal clique tree structure for probability modeling, the resulting derived probability distribution, and a corresponding unified approach to clustering and anomaly detection based on the probability distribution.

cs.LG↗

arXiv2016

Use of gravitational waves to probe the formation channels of compact binaries

With the discovery of the binary black hole coalescence GW150914, the era of gravitational-wave astrophysics has started. Gravitational-wave signals emitted by compact binary coalescences will be detected in large number by LIGO and Virgo in the coming months and years. Much about compact binaries is still uncertain, including some key details about their formation channels. The two scenarios which are typically considered, common envelope evolution and dynamical capture, result in different distributions for the orientation of the black hole spins. In particular, common envelope evolution is expected to be highly efficient in aligning spins with the orbital angular momentum. In this paper we simulate catalogs of gravitational-wave signals in which a given fraction of events comes from common envelop evolution, and has spins nearly aligned with the orbital angular momentum. We show how the fraction of aligned systems can be accurately estimated using Bayesian parameter estimation, with 1 $σ$ uncertainties of the order of 10% after 100-200 sources are detected.

gr-qc↗

arXiv2015

Robust parameter estimation for compact binaries with ground-based gravitational-wave observations using the LALInference software library

The Advanced LIGO and Advanced Virgo gravitational wave (GW) detectors will begin operation in the coming years, with compact binary coalescence events a likely source for the first detections. The gravitational waveforms emitted directly encode information about the sources, including the masses and spins of the compact objects. Recovering the physical parameters of the sources from the GW observations is a key analysis task. This work describes the LALInference software library for Bayesian parameter estimation of compact binary signals, which builds on several previous methods to provide a well-tested toolkit which has already been used for several studies. We show that our implementation is able to correctly recover the parameters of compact binary signals from simulated data from the advanced GW detectors. We demonstrate this with a detailed comparison on three compact binary systems: a binary neutron star, a neutron star black hole binary and a binary black hole, where we show a cross-comparison of results obtained using three independent sampling algorithms. These systems were analysed with non-spinning, aligned spin and generic spin configurations respectively, showing that consistent results can be obtained even with the full 15-dimensional parameter space of the generic spin configurations. We also demonstrate statistically that the Bayesian credible intervals we recover correspond to frequentist confidence intervals under correct prior assumptions by analysing a set of 100 signals drawn from the prior. We discuss the computational cost of these algorithms, and describe the general and problem-specific sampling techniques we have used to improve the efficiency of sampling the compact binary coalescence parameter space.

gr-qc↗

arXiv2014

The First Two Years of Electromagnetic Follow-Up with Advanced LIGO and Virgo

We anticipate the first direct detections of gravitational waves (GWs) with Advanced LIGO and Virgo later this decade. Though this groundbreaking technical achievement will be its own reward, a still greater prize could be observations of compact binary mergers in both gravitational and electromagnetic channels simultaneously. During Advanced LIGO and Virgo's first two years of operation, 2015 through 2016, we expect the global GW detector array to improve in sensitivity and livetime and expand from two to three detectors. We model the detection rate and the sky localization accuracy for binary neutron star (BNS) mergers across this transition. We have analyzed a large, astrophysically motivated source population using real-time detection and sky localization codes and higher-latency parameter estimation codes that have been expressly built for operation in the Advanced LIGO/Virgo era. We show that for most BNS events the rapid sky localization, available about a minute after a detection, is as accurate as the full parameter estimation. We demonstrate that Advanced Virgo will play an important role in sky localization, even though it is anticipated to come online with only one-third as much sensitivity as the Advanced LIGO detectors. We find that the median 90% confidence region shrinks from ~500 square degrees in 2015 to ~200 square degrees in 2016. A few distinct scenarios for the first LIGO/Virgo detections emerge from our simulations.

astro-ph.HE↗

arXiv2014

SKYNET: an efficient and robust neural network training tool for machine learning in astronomy

We present the first public release of our generic neural network training algorithm, called SkyNet. This efficient and robust machine learning tool is able to train large and deep feed-forward neural networks, including autoencoders, for use in a wide range of supervised and unsupervised learning applications, such as regression, classification, density estimation, clustering and dimensionality reduction. SkyNet uses a `pre-training' method to obtain a set of network parameters that has empirically been shown to be close to a good solution, followed by further optimisation using a regularised variant of Newton's method, where the level of regularisation is determined and adjusted automatically; the latter uses second-order derivative information to improve convergence, but without the need to evaluate or store the full Hessian matrix, by using a fast approximate method to calculate Hessian-vector products. This combination of methods allows for the training of complicated networks that are difficult to optimise using standard backpropagation techniques. SkyNet employs convergence criteria that naturally prevent overfitting, and also includes a fast algorithm for estimating the accuracy of network outputs. The utility and flexibility of SkyNet are demonstrated by application to a number of toy problems, and to astronomical problems focusing on the recovery of structure from blurred and noisy images, the identification of gamma-ray bursters, and the compression and denoising of galaxy images. The SkyNet software, which is implemented in standard ANSI C and fully parallelised using MPI, is available at http://www.mrao.cam.ac.uk/software/skynet/.

astro-ph.IM↗

arXiv2013

Reconstructing the sky location of gravitational-wave detected compact binary systems: methodology for testing and comparison

The problem of reconstructing the sky position of compact binary coalescences detected via gravitational waves is a central one for future observations with the ground-based network of gravitational-wave laser interferometers, such as Advanced LIGO and Advanced Virgo. Different techniques for sky localisation have been independently developed. They can be divided in two broad categories: fully coherent Bayesian techniques, which are high-latency and aimed at in-depth studies of all the parameters of a source, including sky position, and "triangulation-based" techniques, which exploit the data products from the search stage of the analysis to provide an almost real-time approximation of the posterior probability density function of the sky location of a detection candidate. These techniques have previously been applied to data collected during the last science runs of gravitational-wave detectors operating in the so-called initial configuration. Here, we develop and analyse methods for assessing the self-consistency of parameter estimation methods and carrying out fair comparisons between different algorithms, addressing issues of efficiency and optimality. These methods are general, and can be applied to parameter estimation problems other than sky localisation. We apply these methods to two existing sky localisation techniques representing the two above-mentioned categories, using a set of simulated inspiral-only signals from compact binary systems with total mass $\le 20\,M_\odot$ and non-spinning components. We compare the relative advantages and costs of the two techniques and show that sky location uncertainties are on average a factor $\approx 20$ smaller for fully coherent techniques than for the specific variant of the "triangulation-based" technique used during the last science runs, at the expense of a factor $\approx 1000$ longer processing time.

astro-ph.IM↗

arXiv2012

BAMBI: blind accelerated multimodal Bayesian inference

In this paper we present an algorithm for rapid Bayesian analysis that combines the benefits of nested sampling and artificial neural networks. The blind accelerated multimodal Bayesian inference (BAMBI) algorithm implements the MultiNest package for nested sampling as well as the training of an artificial neural network (NN) to learn the likelihood function. In the case of computationally expensive likelihoods, this allows the substitution of a much more rapid approximation in order to increase significantly the speed of the analysis. We begin by demonstrating, with a few toy examples, the ability of a NN to learn complicated likelihood surfaces. BAMBI's ability to decrease running time for Bayesian inference is then demonstrated in the context of estimating cosmological parameters from Wilkinson Microwave Anisotropy Probe and other observations. We show that valuable speed increases are achieved in addition to obtaining NNs trained on the likelihood functions for the different model and data combinations. These NNs can then be used for an even faster follow-up analysis using the same likelihood and different priors. This is a fully general algorithm that can be applied, without any pre-processing, to other problems with computationally expensive likelihood functions.

astro-ph.IM↗

arXiv2011

An investigation into the Multiple Optimised Parameter Estimation and Data compression algorithm

We investigate the use of the Multiple Optimised Parameter Estimation and Data compression algorithm (MOPED) for data compression and faster evaluation of likelihood functions. Since MOPED only guarantees maintaining the Fisher matrix of the likelihood at a chosen point, multimodal and some degenerate distributions will present a problem. We present examples of scenarios in which MOPED does faithfully represent the true likelihood but also cases in which it does not. Through these examples, we aim to define a set of criteria for which MOPED will accurately represent the likelihood and hence may be used to obtain a significant reduction in the time needed to calculate it. These criteria may involve the evaluation of the full likelihood function for comparison.

astro-ph.IM↗

arXiv2010

Classifying LISA gravitational wave burst signals using Bayesian evidence

We consider the problem of characterisation of burst sources detected with the Laser Interferometer Space Antenna (LISA) using the multi-modal nested sampling algorithm, MultiNest. We use MultiNest as a tool to search for modelled bursts from cosmic string cusps, and compute the Bayesian evidence associated with the cosmic string model. As an alternative burst model, we consider sine-Gaussian burst signals, and show how the evidence ratio can be used to choose between these two alternatives. We present results from an application of MultiNest to the last round of the Mock LISA Data Challenge, in which we were able to successfully detect and characterise all three of the cosmic string burst sources present in the release data set. We also present results of independent trials and show that MultiNest can detect cosmic string signals with signal-to-noise ratio (SNR) as low as ~7 and sine-Gaussian signals with SNR as low as ~8. In both cases, we show that the threshold at which the sources become detectable coincides with the SNR at which the evidence ratio begins to favour the correct model over the alternative.

gr-qc↗

arXiv2010

The Mock LISA Data Challenges: from Challenge 3 to Challenge 4

The Mock LISA Data Challenges are a program to demonstrate LISA data-analysis capabilities and to encourage their development. Each round of challenges consists of one or more datasets containing simulated instrument noise and gravitational waves from sources of undisclosed parameters. Participants analyze the datasets and report best-fit solutions for the source parameters. Here we present the results of the third challenge, issued in Apr 2008, which demonstrated the positive recovery of signals from chirping Galactic binaries, from spinning supermassive--black-hole binaries (with optimal SNRs between ~ 10 and 2000), from simultaneous extreme-mass-ratio inspirals (SNRs of 10-50), from cosmic-string-cusp bursts (SNRs of 10-100), and from a relatively loud isotropic background with Omega_gw(f) ~ 10^-11, slightly below the LISA instrument noise.

gr-qc↗

arXiv2008

Thermal AGN Signatures in Blazars

Long ignored in blazars because of the dominance of the beamed radiation from the jet, the topic of thermal emissions in these objects is just beginning to be explored. While this emission is weak in most blazars compared to the dominant nonthermal jet components, there is a growing body of evidence that suggests that thermal emission is observable even in the most highly beamed objects. The emitting regions, which can include the accretion disk as well as the torus, are key parts of the central engine which also powers the jets. They also may be of critical importance in helping us decide between unified scheme models. We will review the observational evidence for thermal emissions in blazars, with an emphasis on recent work, and the spectral and variability characteristics that have been observed. The majority of the evidence for thermal emission in blazars (now observed in several objects) has come as a result of multiwavelength campaigns, where the object showed a clear bump in the optical-UV in a faint state. However, evidence exists from other avenues as well, including both purely spectral and variability based arguments as well as statistical analyses of large samples of objects. We will also discuss the impact of thermal emission on an object's overall SED, including its Comptonized signatures. Finally, we will assess the current standing of unified scheme models as respects thermal signatures and the prospects for detecting thermal emission with new telescopes and missions, and further utilizing it as a probe of the central engine of blazars.

astro-ph↗