Stephen M. Doe

Sherpa : a mission-independent data analysis application

The ever-increasing quality and complexity of astronomical data underscores the need for new and powerful data analysis applications. This need has led to the development of Sherpa, a modeling and fitting program in the CIAO software package that enables the analysis of multi-dimensional, multi-wavelength data. In this paper, we present an overview of Sherpas features, which include: support for a wide variety of input and output data formats, including the new Model Descriptor List (MDL) format; a model language which permits the construction of arbitrarily complex model expressions, including ones representing instrument characteristics; a wide variety of fit statistics and methods of optimization, model comparison, and parameter estimation; multi-dimensional visualization, provided by ChIPS; and new interactive analysis capabilities provided by embedding the S-Lang interpreted scripting language. We conclude by showing example Sherpa analysis sessions.

STATISTICAL CHARACTERIZATION OF THE CHANDRA SOURCE CATALOG

The first release of the Chandra Source Catalog (CSC) contains ~95,000 X-ray sources in a total area of 0.75% of the entire sky, using data from ~3900 separate ACIS observations of a multitude of different types of X-ray sources. In order to maximize the scientific benefit of such a large, heterogeneous data set, careful characterization of the statistical properties of the catalog, i.e., completeness, sensitivity, false source rate, and accuracy of source properties, is required. Characterization efforts of other large Chandra catalogs, such as the ChaMP Point Source Catalog or the 2 Mega-second Deep Field Surveys, while informative, cannot serve this purpose, since the CSC analysis procedures are significantly different and the range of allowable data is much less restrictive. We describe here the characterization process for the CSC. This process includes both a comparison of real CSC results with those of other, deeper Chandra catalogs of the same targets and extensive simulations of blank-sky and point-source populations.

The Chandra X-ray Observatory data processing system

Raw data from the Chandra X-ray Observatory are processed by a set of standard data processing pipelines to create scientifically useful data products appropriate for further analysis by end users. Fully automated pipelines read the dumped raw telemetry byte stream from the spacecraft and perform the common reductions and calibrations necessary to remove spacecraft and instrumental signatures and convert the data into physically meaningful quantities that can be further analyzed by observers. The resulting data products are subject to automated validation to ensure correct pipeline processing and verify that the spacecraft configuration and scheduling matched the observers request and any constraints. In addition, pipeline processing monitors science and engineering data for anomalous indications and trending, and triggers alerts if appropriate. Data products are ingested and stored in the Chandra Data Archive, where they are made available for downloading by users. In this paper, we describe the architecture of the data processing system, including the scientific algorithms that are applied to the data, and interfaces to other subsystems. We place particular emphasis on the impacts of design choices on system integrity and maintainability. We review areas where algorithmic improvements or changes in instrument characteristics have required significant enhancements, and the mechanisms used to effect these changes while assuring continued scientific integrity and robustness. We discuss major enhancements to the data processing system that are currently being developed to automate production of the Chandra Source Catalog.

Iris: an Extensible Application for Building and Analyzing Spectral Energy Distributions

Iris is an extensible application that provides astronomers with a user-friendly interface capable of ingesting broad-band data from many different sources in order to build, explore, and model spectral energy distributions (SEDs). Iris takes advantage of the standards defined by the International Virtual Observatory Alliance, but hides the technicalities of such standards by implementing different layers of abstraction on top of them. Such intermediate layers provide hooks that users and developers can exploit in order to extend the capabilities provided by Iris. For instance, custom Python models can be combined in arbitrary ways with the Iris built-in models or with other custom functions. As such, Iris offers a platform for the development and integration of SED data, services, and applications, either from the users system or from the web. In this paper we describe the built-in features provided by Iris for building and analyzing SEDs. We also explore in some detail the Iris framework and software development kit, showing how astronomers and software developers can plug their code into an integrated SED analysis environment.

Managing distributed software development in the Virtual Astronomical Observatory

The U.S. Virtual Astronomical Observatory (VAO) is a product-driven organization that provides new scientific research capabilities to the astronomical community. Software development for the VAO follows a lightweight framework that guides development of science applications and infrastructure. Challenges to be overcome include distributed development teams, part-time efforts, and highly constrained schedules. We describe the process we followed to conquer these challenges while developing Iris, the VAO application for analysis of 1-D astronomical spectral energy distributions (SEDs). Iris was successfully built and released in less than a year with a team distributed across four institutions. The project followed existing International Virtual Observatory Alliance inter-operability standards for spectral data and contributed a SED library as a by-product of the project. We emphasize lessons learned that will be folded into future development efforts. In our experience, a well-defined process that provides guidelines to ensure the project is cohesive and stays on track is key to success. Internal product deliveries with a planned test and feedback loop are critical. Release candidates are measured against use cases established early in the process, and provide the opportunity to assess priorities and make course corrections during development. Also key is the participation of a stakeholder such as a lead scientist who manages the technical questions, advises on priorities, and is actively involved as a lead tester. Finally, frequent scheduled communications (for example a bi-weekly tele-conference) assure issues are resolved quickly and the team is working toward a common vision.