High-quality, usable, and effective software is essential for supporting astronomers in the discovery-focused tasks of data analysis and visualisation. As the volume, and perhaps more crucially, the velocity of astronomical data grows, the role of th
e astronomer is changing. There is now an increased reliance on automated and autonomous discovery and decision-making workflows rather than visual inspection. We assert the need for an improved understanding of how astronomers (humans) currently make visual discoveries from data. This insight is a critical element for the future design, development and effective use of cyber-human discovery systems, where astronomers work in close collaboration with automated systems to gain understanding from continuous, real-time data streams. We discuss how relevant human performance data could be gathered, specifically targeting the domains of expertise and skill at visual discovery, and the identification and management of cognitive factors. By looking to other disciplines where human performance is assessed and measured, we propose four early-stage applications that would: (1) allow astronomers to evaluate, and potentially improve, their own visual discovery skills; (2) support just-in-time coaching; (3) enable talent identification; and (4) result in user interfaces that automatically respond to skill level and cognitive state. Throughout, we advocate for the importance of user studies and the incorporation of participatory design and co-design practices into the planning, implementation and evaluation of alternative user interfaces and visual discovery environments.
Machine learning (automated processes that learn by example in order to classify, predict, discover or generate new data) and artificial intelligence (methods by which a computer makes decisions or discoveries that would usually require human intelli
gence) are now firmly established in astronomy. Every week, new applications of machine learning and artificial intelligence are added to a growing corpus of work. Random forests, support vector machines, and neural networks (artificial, deep, and convolutional) are now having a genuine impact for applications as diverse as discovering extrasolar planets, transient objects, quasars, and gravitationally-lensed systems, forecasting solar activity, and distinguishing between signals and instrumental effects in gravitational wave astronomy. This review surveys contemporary, published literature on machine learning and artificial intelligence in astronomy and astrophysics. Applications span seven main categories of activity: classification, regression, clustering, forecasting, generation, discovery, and the development of new scientific insight. These categories form the basis of a hierarchy of maturity, as the use of machine learning and artificial intelligence emerges, progresses or becomes established.
In the upcoming synoptic all--sky survey era of astronomy, thousands of new multiply imaged quasars are expected to be discovered and monitored regularly. Light curves from the images of gravitationally lensed quasars are further affected by superimp
osed variability due to microlensing. In order to disentangle the microlensing from the intrinsic variability of the light curves, the time delays between the multiple images have to be accurately measured. The resulting microlensing light curves can then be analyzed to reveal information about the background source, such as the size of the quasar accretion disc. In this paper we present the most extensive and coherent collection of simulated microlensing light curves; we have generated $>2.5$ billion light curves using the GERLUMPH high resolution microlensing magnification maps. Our simulations can be used to: train algorithms to measure lensed quasar time delays, plan future monitoring campaigns, and study light curve properties throughout parameter space. Our data are openly available to the community and are complemented by online eResearch tools, located at http://gerlumph.swin.edu.au .
In recent years, the Graphics Processing Unit (GPU) has emerged as a low-cost alternative for high performance computing, enabling impressive speed-ups for a range of scientific computing applications. Early adopters in astronomy are already benefiti
ng in adapting their codes to take advantage of the GPUs massively parallel processing paradigm. I give an introduction to, and overview of, the use of GPUs in astronomy to date, highlighting the adoption and application trends from the first ~100 GPU-related publications in astronomy. I discuss the opportunities and challenges of utilising GPU computing clusters, such as the new Australian GPU supercomputer, gSTAR, for accelerating the rate of astronomical discovery.
Astronomy is entering a new era of discovery, coincident with the establishment of new facilities for observation and simulation that will routinely generate petabytes of data. While an increasing reliance on automated data analysis is anticipated, a
critical role will remain for visualization-based knowledge discovery. We have investigated scientific visualization applications in astronomy through an examination of the literature published during the last two decades. We identify the two most active fields for progress - visualization of large-N particle data and spectral data cubes - discuss open areas of research, and introduce a mapping between astronomical sources of data and data representations used in general purpose visualization tools. We discuss contributions using high performance computing architectures (e.g: distributed processing and GPUs), collaborative astronomy visualization, the use of workflow systems to store metadata about visualization parameters, and the use of advanced interaction devices. We examine a number of issues that may be limiting the spread of scientific visualization research in astronomy and identify six grand challenges for scientific visualization research in the Petascale Astronomy Era.
Flexion-based weak gravitational lensing analysis is proving to be a useful adjunct to traditional shear-based techniques. As flexion arises from gradients across an image, analytic and numerical techniques are required to investigate flexion predict
ions for extended image/source pairs. Using the Schwarzschild lens model, we demonstrate that the ray-bundle method for gravitational lensing can be used to accurately recover second flexion, and is consistent with recovery of zero first flexion. Using lens plane to source plane bundle propagation, we find that second flexion can be recovered with an error no worse than 1% for bundle radii smaller than {Delta}{theta} = 0.01 {theta}_E and lens plane impact pararameters greater than {theta}_E + {Delta}{theta}, where {theta}_E is the angular Einstein radius. Using source plane to lens plane bundle propagation, we demonstrate the existence of a preferred flexion zone. For images at radii closer to the lens than the inner boundary of this zone, indicative of the true strong lensing regime, the flexion formalism should be used with caution (errors greater than 5% for extended image/source pairs). We also define a shear zone boundary, beyond which image shapes are essentially indistinguishable from ellipses (1% error in ellipticity). While suggestive that a traditional weak lensing analysis is satisfactory beyond this boundary, a potentially detectable non-zero flexion signal remains.
Visualisation and analysis of terabyte-scale datacubes, as will be produced with the Australian Square Kilometre Array Pathfinder (ASKAP), will pose challenges for existing astronomy software and the work practices of astronomers. Focusing on the pro
posed outcomes of WALLABY (Widefield ASKAP L-Band Legacy All-Sky Blind Survey), and using lessons learnt from HIPASS (HI Parkes All Sky Survey), we identify issues that astronomers will face with WALLABY data cubes. We comment on potential research directions and possible solutions to these challenges.
General purpose computing on graphics processing units (GPGPU) is dramatically changing the landscape of high performance computing in astronomy. In this paper, we identify and investigate several key decision areas, with a goal of simplyfing the ear
ly adoption of GPGPU in astronomy. We consider the merits of OpenCL as an open standard in order to reduce risks associated with coding in a native, vendor-specific programming environment, and present a GPU programming philosophy based on using brute force solutions. We assert that effective use of new GPU-based supercomputing facilities will require a change in approach from astronomers. This will likely include improved programming training, an increased need for software development best-practice through the use of profiling and related optimisation tools, and a greater reliance on third-party code libraries. As with any new technology, those willing to take the risks, and make the investment of time and effort to become early adopters of GPGPU in astronomy, stand to reap great benefits.
Traditional analysis techniques may not be sufficient for astronomers to make the best use of the data sets that current and future instruments, such as the Square Kilometre Array and its Pathfinders, will produce. By utilizing the incredible pattern
-recognition ability of the human mind, scientific visualization provides an excellent opportunity for astronomers to gain valuable new insight and understanding of their data, particularly when used interactively in 3D. The goal of our work is to establish the feasibility of a real-time 3D monitoring system for data going into the Australian SKA Pathfinder archive. Based on CUDA, an increasingly popular development tool, our work utilizes the massively parallel architecture of modern graphics processing units (GPUs) to provide astronomers with an interactive 3D volume rendering for multi-spectral data sets. Unlike other approaches, we are targeting real time interactive visualization of datasets larger than GPU memory while giving special attention to data with low signal to noise ratio - two critical aspects for astronomy that are missing from most existing scientific visualization software packages. Our framework enables the astronomer to interact with the geometrical representation of the data, and to control the volume rendering process to generate a better representation of their datasets.
