Chandra's First Decade of Discovery

Source Catalogs and Software

It's Not Just for Data Anymore: The Many Faces of the Chandra Data Archive

Glenn Becker, Smithsonian Astrophysical Observatory
Arnold Rots, Sherry Winkelman, Michael McCollough, Aaron Watry, Joan Hagler

The Chandra Data Archive (CDA) has been recognized for a decade as a deep wellspring of X-Ray data. These data are available for download via multiple interfaces and international mirror sites; however, this is far from the end of the story. The CDA also maintains a database that allows users to check on the processing status of their observational data. Researchers can submit requests for special data processing and custom queries, do detailed bibliography searches (The Chandra Bibliography Cataloging System: A Scientific Research Aid), and request and host contributed datasets. Applications coming online soon will include a crossmatch database with SDSS (The CSC-SDSS Cross-match Catalog) and a Chandra footprint service. The CDA is an archive that faces squarely forward, ready to accommodate an ever-widening range of needs.This work is supported by NASA contract NAS 8-03060.

X-ray Stacking for the Analysis of Faint Sources: A Bayesian Alternative

Alexander Blocker, Harvard University Department of Statistics
Tom Aldcroft (SAO), Paul J Green (SAO), Daryl Haggard (UWa), Anca Constantin (JMU), Scott Anderson (UWa), Dong-Woo Kim (SAO)

Stacking is a powerful and increasingly common technique for the analysis of celestial sources undetected in the x-ray band. However, it has a number of significant drawbacks. These include a lack of error bars, the loss of information on individual sources, handling of outliers (i.e., bright detections) and the problems of negative net counts (and potentially negative stacked mean intensities). We present an alternative Bayesian method based on a full probability model for the observed counts and source intensities. Using this model, we can obtain estimates of individual source intensities (including luminosities, where redshifts are available) in addition to conventional stacked values. Furthermore, all these estimates have statistically valid error bars. This new method is demonstrated using both a simulated sample with known x-ray properties and a sample of SDSS galaxies. The latter includes a binned analysis of luminosity by absolute magnitude and redshift. We provide Python implementations of our key algorithms.

ACIS Extract takes on the Carina Complex

Patrick Broos, Pennsylvania State University

The Chandra community has been using the ACIS Extract (AE) software package to extract and analyze sources in ACIS observations since y2002. AE brings significant automation to the extraction of point-like and diffuse sources, and provides a simple analysis strategy for projects that involve multiple overlapping observations. The community has in past years applied AE to some very complex Chandra projects, e.g. the Chandra Deep Field and the Galactic Center. The AE developers have recently completed their first AE analysis of similar scale--the 14,000 point sources and extensive diffuse emission in the Carina Very Large Project. That experience has significantly refined AE's capabilities and recipes; we will illustrate some of this new functionality with examples from Carina.

[PDF of this poster]

Using 'LIRA' To Quantify Diffuse Structure Around X-ray and Gamma-Ray Pulsars

Alanna Connors, Eureka Scientific
Nathan M. Stein (Harvard Statistics), David van Dyk (University of California, Irvine, Statistics), Aneta Siemiginowska (SAO), Vinay Kashyap (SAO), Mallory Roberts (Eureka Scientific)

In this poster, we exploit several capabilities of a Low-count Image Restoration and Analysis (LIRA) package, to quantify details of faint “scruffy” emission, consistent with PWN around X-ray and gamma-ray pulsars. Our preliminary results show evidence for irregular structure on scales of 1"-10" or less (i.e. <500 pc), rather than larger smooth loops. Additionally, we can show this to be visible across several energy bands.LIRA grew out of work by the California-Boston Astro-Statistics Collaboration (CBASC) on analyzing high resolution, high energy Poisson images from X-ray and gamma-ray telescopes (see Stein et. al. these proceedings; also Esch et al 2004; and Connors and van Dyk in SCMAIV). LIRA fits: a “Null” or background model shape, times a scale factor; plus a flexible Multi-Scale (MS) model; folded though an instrument response (PSF, exposure). Embedding this in a fully Poisson probability structure allows us to map out uncertainties in our image analysis and reconstruction, via many MCMC samples. Specifically, for quantifying irregular nebular structure, we exploit the Multi-Scale model's smoothing parameters at each length-scale, as “Summary Statistics” (i.e low-dimensional summaries of the probability space). When distributions of these summary statistics, from analysis of simulated “Null” data sets, are compared with those from the actual Chandra data, we can set quantitative limits on structures at different length scales. Since one can do this for very low counts, one is able to analyze and compare structure in several energy slices. This work is supported by NSF and AISR funds.

Flexibility and Extensibility of Sherpa in CIAO 4.2

Stephen Doe, SAO
Dan T. Nguyen (SAO), Brian L. Refsdal (SAO), Aneta L. Siemiginowska (SAO), Tom Aldcroft (SAO), Nina R. Bonaventura (SAO), Douglas Burke (SAO), Ian N. Evans (SAO), Janet D. Evans (SAO), Antonella Fruscione (SAO), Elizabeth Galle (SAO), John C. Houck (MIT), Nicholas Lee (SAO), Jonathan C. McDowell (SAO), Michael A. Nowak (MIT)

Sherpa is the general modeling and fitting application developed in CIAO (Chandra Interactive Analysis of Observations). It provides a flexible environment for modeling a variety of Chandra data (spectra, images, time series) as well as data from other X-ray missions. It also allows for modeling multi-frequency data simultaneously (e.g. spectral energy distributions using X-ray, optical and radio data; modeling and fitting optical images with PSFs; or any other type of data that are provided as collections of arrays). Sherpa uses the forward fitting technique, so parameterized models are fit by minimization of the statistic that describes the differences between the data and the model. Sherpa has a library of mathematical and astrophysical models including the XSPEC 12.5 models. Sherpa allows for fitting high counts data with a choice of weighted chi-square statistics. The Poisson likelihood for low counts data is available with Cash statistics and simulations to access the probability distributions. The final model parameters have confidence bounds that are calculated, and visualization of the parameter space allows for analysis of correlations between parameters or multi-modality of the parameter space. Sherpa's modular and extensible design allows for a considerable degree of user customization. Models and statistical functions of arbitrary complexity can be written in C, Fortran or Python, and can be easily imported into a running Sherpa session. We briefly describe the new Sherpa and give examples of the complex cases that can be successfully modeled in the current version. This work is supported by NASA contract NAS8-03060 (CXC).

A Monte Carlo Method for Including Chandra Instrument Response Uncertainties in Parameter Estimation Studies

Jeremy Drake, Smithsonian Astrophysical Observatory
Peter W. Ratzlaff, Vinay Kashyap and the Chandra MC Uncertainties Team

Including allowance for instrument calibration uncertainties in data analysis is technically challenging in terms of both understanding and specifying the uncertainties themselves, and in developing appropriate algorithms to employ them. Calibration uncertainties are generally correlated in complicated ways, rendering traditional methods of error analysis inappropriate. Here we describe a Monte Carlo method in which current knowledge of the response of the Chandra ACIS instrument is represented by random sampling of plausible perturbations to anominal response function. The resulting set of response functions are employed in parameter estimation exercises to assess the effects of the instrument uncertainties on derived parameter values. We use this method to assess the limiting accuracy of Chandra for understanding typical X-ray source model parameters. We briefly describe a code slated for public release that will enable end users to perform similar full error analyses on Chandra ACIS observations.

Nonparametric Estimation of Intrinsic Properties of Faint X-ray Sources

Konstantin Getman, Penn State University
Eric Feigelson (PSU), Patrick Broos (PSU), Leisa Townsley (PSU), Gordon Garmire

X-ray sources with very few counts can be identified with low-noise X-ray detectors such as the Advanced CCD Imaging Spectrometer onboard the Chandra X-ray Observatory. These sources are often too faint for parametric spectral modeling using well-established methods such as spectral fitting with XSPEC. We discuss the estimation of apparent and intrinsic broad-band X-ray fluxes and soft X-ray absorption from gas along the line-of-sight to these sources, using nonparametric methods. Apparent flux is estimated from the ratio of the source count number to the instrumental effective area averaged over the chosen band. Absorption and intrinsic flux are estimated from the comparison of the apparent median energy of the source photons and apparent source flux with those of high signal-to-noise spectra that were simulated using spectral models characteristic of much brighter sources of similar class previously studied in detail. The concept of this method is similar to the long-standing use of color-magnitude diagrams in optical and infrared astronomy. Our nonparametric method is tested against the apparent spectra of faint sources. We show that the intrinsic X-ray properties can be determined with little bias and reasonable accuracy using these observable photometric quantities without employing often uncertain methods of non-linear parametric spectral modeling. Our results are obtained for thermal spectra characteristic of stars in young stellar clusters, but similar results should hold for other classes of faint X-ray sources.

[PDF of this poster]

Implementation of the Non-equilibrium Ionization Code in ISIS

Li Ji, MIT Kavli Institute for Astrophysics & Space Research
M. Noble, J. Houck, N.S. Schulz, M. Nowak, H.L. Marshall

The non-equilibrium ionization is present in a wide range of astrophysical phenomena such as colliding winds in X-ray binaries, outflows in AGNs, and shock flows in the IGM. Challenges are significant when applying the non-equilibrium ionization code to the Chandra HETG observations. Parallelism has been used to speed up the calculations. Modular software techniques enable us to compute atomic rates directly in ISIS. We present the implementation of our non-equilibrium ionization code in ISIS based on all the available update atomic data. In addition, we demonstrate several applications to the non-equilibrium ionization plasma for the Chandra HETG observations.

Chandra Source Catalog, ChaMP and SDSS

Dong-Woo Kim, SAO
Amy Mossman(SAO), Paul Green(SAO), Pepi Fabbiano(SAO)

We present the result of cross-correlation between Chandra Source Catalog (CSC) and SDSS DR7. First, we compared CSC and ChaMP X-ray sources and found consistent results in detecting sources and their X-ray properties with a very small number of exceptional cases. By visual examination, we identified and excluded optically saturated stars with incorrect optical properties which often cause multiple matches. We have identified ~13500 CSC X-ray sources which match with SDSS optical sources. There are ~5000 X-ray sources with possible optical counterparts but with a higher probability of random coincidence. Additionally, there are ~10000 X-ray sources with no SDSS optical counterpart. Among the matched sources, 85% have fx/fo between 0.1 and 10 (consistent being AGNs) and ~10% have fx/fo less than 0.1 (likely normal galaxies). We further discuss different types of objects, in terms of their redshift distributions and spectral properties.

A new MCMC module for Bayesian Spectral Analysis in Sherpa

Jason Kramer, University of California Irvine
David A. van Dyk (University of California Irvine), Aneta Siemiginowska (SAO), Doug Burke (SAO), Brian Refsdal (SAO), Dan Nguyen (SAO), Stephen Doe (SAO)

Standard statistical methods for addressing parameter uncertainty may not always be appropriate for sophisticated spectral models. Typically relying on high-count approximations, standard methods applied to low-count data may result in misleading error bars. For example, when fitting a narrow emission line, there may be several separate likely values for its location that cannot be summarized with simple error bars. Finding possible locations requires sophisticated techniques capable of exploring the sometimes highly multi-modal posterior distribution, while testing for the presence of a line requires computational methods such as PPP (Protassov et al 2002) since standard p-values do not apply. Methods that fully explore the parameter space, such as MCMC, provide a more complete picture of parameter uncertainty. The use of MCMC for Bayesian analysis of spectral models has been explored and validated, (van Dyk et al 2004, 2006; Park, van Dyk, and Siemiginowska 2008), but up until now has required specialized software that could only accommodate a specific class of models. We are in the process of developing a new, more flexible computational module that will allow these techniques to be used with most models available in Sherpa. The computational module has been validated for a class of background-contaminated absorbed single-component models including power law, blackbody, and thermal bremsstrahlung. Statistically, our MCMC method relies on default non-informative prior distributions that do not need to be specified by the user. For computational and statistical reasons, we use normalizing transformations of the parameters, though estimates and confidence intervals are provided on the standard physical scale. For more complex models, more sophisticated techniques will have to be employed. Multiple modes may require techniques such as simulated annealing or MH with multiple proposal distributions. For samplers with high rejection rates due to strong posterior correlations, we will explore Metropolis within Gibbs using dynamic correlation reduction.

ChIPS - Chandra's Interactive, Publication-Ready Plotting Tool

Joseph Miller, SAO
D. Burke (SAO), I. Evans (SAO), J. Evans (SAO), A. Fruscione (SAO), G. Germain (SAO), J. McDowell (SAO), W. McLaughlin (NGIS), R. Milaszewski (SAO)

The Chandra Interactive Plotting System, ChIPS, is a powerful component of the CIAO data analysis system that enables users to visualize their data and construct high-quality, publication-ready plots interactively. The user can control almost every aspect of the plot layout and the properties of individual plot components such as tick positions or symbol styles. ChIPS offers a rich interactive environment to help users design and fine tune their plots. Key features of ChIPS include the ability to explore alternative presentations of their data by interactively adjusting plot parameters or plot component properties, or correct mistakes via the included undo/redo functionality, without having to redo the visualizations from the beginning. Through a Python or S-Lang interface, ChIPS provides a set of high-level user routines which hides the details of the underlying environment from the new user. At the same time, the scripting environment affords experienced users the ability to manipulate data or extend existing functionality. New to CIAO 4.2, all users will benefit from being able to interactively develop plots and then save the steps to create the final product as a script. This can then be used to recreate the visualization with additional data sets. Also new in CIAO 4.2 is the ability to integrate plot data with basic imaging. Users can combine their images (in world coordinates) with plot elements such as overlay contours, grids, or annotations to produce high-quality publication-ready output in the formats expected by the major journals.

Chandra Data: Spacecraft to Scientist in About a Day

Joy Nichols, SAO
C. Anderson (SAO), D. Morgan (SAO), A. W. Mitschang (SAO), J. Lauer (SAO), B. Sundheim (SAO), D. Huenemoerder (MIT), G. Fabbiano (SAO)

The Chandra Project has a reputation for extremely rapid delivery of science data to the observer. In fact, no other NASA major mission approaches the Chandra record. But how much care and validation is done on the data products before delivery? We present a summary of the Chandra data processing operations and procedures, along with statistics of the time lapse in delivery of the data products. The validation process is described to help users understand the quality of the delivered products.

The Chandra Variable Guide Star Catalog

J. Nichols, Smithsonian Astrophysical Observatory
J. Lauer (SAO), E. Martin (Northrop-Grumman Space Technology), D. Huenemoerder (MIT Kalvi Institute for Astrophysics and Space Research), D. Morgan (SAO)

Variable stars have been identified among the optical-wavelength lightcurves of guide stars used for pointing control of the Chandra X-ray Observatory. We present a catalog of these variable stars along with their light curves and ancillary data. Variability was detected to a lower limit of 0.03 mag in the 4000-9000 A range using the photometrically stable Aspect Camera onboard the Chandra spacecraft. The Chandra Variable Guide Star Catalog contains more than 900 objects, of which we believe at least 500 are new detections of variable stars. Types of variables in the catalog include eclipsing binaries, pulsating stars, and rotating stars. The variability was detected during the course of normal verification of each Chandra pointing and results from analysis of 75,200 guide star light curves from the entire Chandra mission. Since many guide stars were used for multiple observations, a significant number of the stars in the catalog have multiple light curves available from various times over the last 10 years. The search for variable stars continues as the Chandra mission proceeds. Supplements to the catalog will be published and added to the web interface. The Chandra Variable Guide Star Catalog is a unique project using otherwise discarded information collected during the mission.

ChaMPlane: latest results and future projects

Mathieu Servillat, Harvard-Smithsonian Center for Astrophysics
Grindlay J.E., Van den Berg M., Hong J., Zhao P., Allen B. (CfA)

The Chandra Multiwavelength Plane Survey (ChaMPlane) is based on Chandra deep X-ray observations close to the Galactic Plane and the Galactic Center. We obtained optical imaging data (NOAO 4m Mosaic), and are conducting a follow-up spectroscopy and near-infrared (NOAO 4m ISPI mainly) program in order to detect counterparts to the X-ray sources and identify their nature. Based on this dataset, we aim to identify peculiar Galactic populations of objects such as accreting white dwarfs, neutron stars and black holes in order to study the distribution and the evolution of those populations. We present here the latest results and the work in progress.

Optimization Methods in Sherpa

Aneta Siemiginowska, SAO
Dan T. Nguyen (SAO), Stephen M. Doe (SAO) and Brian L. Refsdal (SAO)

Forward fitting is a standard technique used to model X-ray data. A statistic, usually assumed weighted χ2 or Poisson likelihood (e.g. Cash), is minimized in the fitting process to obtain a set of the best model parameters. Astronomical models often have complex forms with many parameters that can be correlated (e.g. an absorbed power law). Minimization is not trivial in such setting, as the statistical parameter space becomes multimodal and finding the global minimum is hard. Standard minimization algorithms can be found in many libraries of scientific functions, but they are usually focused on specific functions. However, Sherpa designed as general fitting and modeling application requires very robust optimization methods that can be applied to variety of astronomical data (X-ray spectra, images, timing, optical data etc.). We developed several optimization algorithms in Sherpa targeting a wide range of minimization problems. Two local minimization methods were built: Levenberg-Marquardt algorithm was obtained from MINPACK subroutine LMDIF and modified to achieve the required robustness; and Nelder-Mead simplex method has been implemented in-house based on variations of the algorithm described in the literature. A global search Monte-Carlo method has been implemented following a differential evolution algorithm presented by Storn and Price (1997). We will present the methods in Sherpa and discuss their usage cases. We will focus on the application to Chandra data showing both 1D and 2D examples. This work is supported by NASA contract NAS8-03060 (CXC).

LIRA: Low-Count Image Reconstruction and Analysis

Nathan Stein, Harvard University
David van Dyk (University of California, Irvine), Alanna Connors (Eureka Scientific), Aneta Siemiginowska (SAO), Vinay Kashyap (SAO)

LIRA is a new software package for the R statistical computing language. The package is designed for multi-scale non-parametric image analysis for use in high-energy astrophysics. The code implements an MCMC sampler that simultaneously fits the image and the necessary tuning/smoothing parameters in the model (an advance from `EMC2' of Esch et al. 2004). The model-based approach allows for quantification of the standard error of the fitted image and can be used to access the statistical significance of features in the image or to evaluate the goodness-of-fit of a proposed model. The method does not rely on Gaussian approximations, instead modeling image counts as Poisson data, making it suitable for images with extremely low counts. LIRA can include a null (or background) model and fit the departure between the observed data and the null model via a wavelet-like multi-scale component. The technique is therefore suited for problems in which some aspect of an observation is well understood (e.g, a point source), but questions remain about observed departures. To quantitatively test for the presence of diffuse structure unaccounted for by a point source null model, first, the observed image is fit with the null model. Second, multiple simulated images, generated as Poisson realizations of the point source model, are fit using the same null model. MCMC samples from the posterior distributions of the parameters of the fitted models can be compared and can be used to calibrate the misfit between the observed data and the null model. Additionally, output from LIRA includes the MCMC draws of the multi-scale component images, so that the departure of the (simulated or observed) data from the point source null model can be examined visually. To demonstrate LIRA, an example of reconstructing Chandra images of high redshift quasars with jets is presented.

The Chandra Bibliography Cataloging System: A Scientific Research Aid

Sherry Winkelman, SAO
Arnold Rots (SAO), Michael McCollough (SAO), Glenn Becker (SAO), Aaron Watry (SAO), Joan Hagler (SAO)

The Chandra Data Archive (CDA) has been tracking publications based on Chandra observations in journals and on-line conference proceedings since early in the mission. Our goals are two-fold: 1) provide a means for Chandra users to search literature on Chandra-related papers to further their scientific research; and 2) provide a means for measuring the science produced from Chandra data. In this presentation we focus on our first goal and detail our classification system and procedures, bibliography search pages, as well as our interactions with the Astronomic Data Systems (ADS) to provide direct links to Chandra data from an ADS bibcode page. Our classification system is multi-tiered. Whenever possible we directly link Chandra observations to papers published about that data. We classify Chandra-related papers into 5 general categories and also have ~20 flags which indicate something about the content of the paper. Some of those flags are: whether the principle investigator of the data was an author of the paper; which instruments are cited in the paper; and whether the paper contains follow-up analysis, theory, or contains a multi-observatory analysis with Chandra data and some other dataset. We are in the process of expanding our list of flags to include more than 60 other observatories and more than 70 catalogs and surveys. We hope that by gathering more information contained within papers that we can provide a more powerful research tool for our users whether they use our search pages or ADS. This work is supported by NASA contract NAS8-03060.