Keeping Chandra Analysis on the Cutting Edge: CIAO 4.18—Advances, Updates, and a Farewell

Antonella Fruscione and the Chandra SDS Group

This article is, in many ways, bittersweet. On the one hand, it marks the release of another update to the Chandra data analysis software—CIAO 4.18, Sherpa 4.18, CALDB 4.12.3, SAOImage DS9 v8.7b2, and MARX 6.0.1—continuing a long tradition of steady evolution in support of Chandra science. On the other hand, this release coincides with a moment of transition for the project, as we say goodbye to one of the central figures behind Chandra's data analysis system and infrastructure: our group leader, Jonathan McDowell.

The tools that enable Chandra science evolve quietly but steadily, and the latest updates are a good example. This cycle was not about prominent new features, but about practical, behind-the-scenes work that reduces problems in common analysis tasks and makes workflows more robust. At the same time, this update offers an opportunity to reflect on the people whose vision and dedication built the foundations for these tools' continued growth.

CIAO 4.18 and related tool updates

CIAO 4.18 (https://cxc.harvard.edu/ciao/) is the core of the release, with changes focused on modern data handling and platform compatibility. The most relevant improvement for everyday analysis is the initial support for 8-byte integer table values across many CIAO tools that manipulate table data. Such values are formally called K-type in the FITS standard (and informally known as “long long”), and familiar tools such as dmcopy, dmlist, dmstat, dmsort, and dmextract can now read, filter, and bin K-type columns without truncation or conversion surprises when working with large integers or high-precision data elements. This capability was implemented at the CXC Data Model level, ensuring consistent handling of K-type values throughout CIAO.

Several targeted bug fixes improve precision in specific analysis threads. The acis_process_events tool received a correction to the temperature-dependent CTI adjustment that affected a small number of events (notably in CC mode), correcting event energies in those cases. The glvary tool was also updated to fix a bug in the variability and probability calculations when events are outside the good time intervals, increasing robustness for multi-detector variability analyses.

An animation of a workflow in SAOImage DS9. A Chandra events file is shown as a sky image, centered on an off-axis source extended by the Chandra PSF. The animation begins with the creation of a region file, which is then moved to roughly align with the position of the source. The option to “Enable Elastic Ellipses


Fig. 1: Demonstration of DAX elastic ellipse task in DS9 with an off-axis Chandra source. When the task is activated, the region resizes, repositions, and reorients to match the data based on the second-order image moments.

In addition to core tools, several scripts bundled with CIAO 4.18 received notable updates aimed at streamlining common tasks. A new utility, acis_split_evt_by_fptemp, automates the segmentation of ACIS event files experiencing large focal-plane temperature variations. The dax package, which provides access to CIAO tools and Sherpa capabilities from within SAOImageDS9 (DS9), now includes interactive region-analysis tasks that update counts, light curves, or spectra as regions are moved or edited (Fig. 1), while the new sherpa_contrib.stats.kaastra17 module provides an approximation to help assess the adequacy of CStat best-fit models. Together, these updates demonstrate a continued emphasis on practical, interactive analysis features that reduce manual steps.

A MacOS finder window. There are two folders, named ciao-4.18 and Applications, as well as a README.txt. A green arrow on the window points from the ciao folder to the Applications folder, and blue text underneath reads “Drag ciao folder to Applications.”


Fig. 2: The new CIAO 4.18 macOS installer. The disk image provides a drag-and-drop installation that places the CIAO application bundle under /Applications, a familiar process for many macOS users.

CIAO 4.18 is now distributed with Python 3.12 and updated off-the-shelf components. It can be installed using the traditional ciao-install script, as conda packages, or via a new macOS click-to-install option (a .DMG file) that places the CIAO bundle under /Applications (Fig. 2). The distribution includes SAOImageDS9 v8.7b2 (https://ds9.si.edu/), which adds a configurable autosave option, improves SAMP Hub error checking and performance (particularly on multi-user systems), and includes documentation about cloud-computing capabilities for remote environments.

A snippet of Python code and a corner plot. The code snippet imports arviz and the sherpa–sim package mcmc_to_arviz. It then runs mcmc_to_arviz and uses the arviz command plot_pair to make a corner plot from an input MCMC result. The corner plot shows five parameters plotted against each other in contours, with histograms of the parameter distributions themselves.


Fig. 3: Code example and sample output demonstrating Sherpa's new interface to Arviz, which allows users to easily make corner plots to explore parameter space from Markov Chain Monte Carlo (MCMC) based analysis.

Sherpa 4.18 (https://cxc.harvard.edu/sherpa/index.html) advances the modeling and fitting application with several usability and statistical enhancements, including support for external optimizers (for example, from SciPy and optimagic), the new cstatnegativepenalty statistic to stabilize fits in challenging parameter regimes, simplified plotting of multiple datasets with a single command, and improved export of MCMC results for use with ArviZ in Bayesian analyses (Fig. 3). On the calibration side, CALDB 4.12.3 (https://cxc.harvard.edu/caldb/index.html) delivers important updates to ACIS contamination and HRC time-dependent QE products. Finally, MARX 6.0/6.0.1 (https://chandra-marx.github.io/) modernized the simulator by removing long-unused legacy components, tightening internal algorithms that previously could introduce small image shifts, improving handling of rayfile sources, and fixing a minor imaging issue affecting some sky orientations.

Screenshot of a <em>Chandra</em> thread titled “From Sound to Image: How to Visualize Audio Waveforms with CIAO and DS9. It includes a link to download the notebook for the thread. The main component of the screenshot is a series of spectrograms, all showing the same data but colored in different ways, including different colormaps, intensity scalings, and mapping of intensity to color. At the bottom of the screenshot is an 18-second audio clip.


Fig. 4: The new thread “From Sound to Image: How to Visualize Audio Waveforms with CIAO and DS9” shows different ways to create artistic effects using several image processing techniques available in CIAO. These techniques are applied here to an image of the audio waveform from the STS-93 countdown and launch.

As always, documentation and analysis threads were updated alongside the tools to reflect all the changes (Fig. 4).

A Milestone Release and a Farewell: Jonathan McDowell Retires

This release cycle also marks an important transition for the Chandra project. Jonathan McDowell (Fig. 5), long-time member of the Science Data Systems (SDS) group, retired in January 2026 after nearly four decades at the Center for Astrophysics.

Photograph of Jonathan McDowell. He is sitting at a table, looking toward the camera. He wears a blue shirt with a stellar nebula print underneath a black blazer. Jonathan is smiling, with both hands raised slightly, and he has a smile on his lips.


Fig. 5: Jonathan McDowell, the long-time leader of the Chandra Science Data Systems group.

Jonathan's influence on Chandra has been profound. As an SDS member and then group leader, he helped shape the mission's data systems and analysis infrastructure from its earliest days. He created the foundations of the CIAO Data Model and authored the definitive reference documentation on the Chandra Data Analysis Coordinate Systems. Over many years, he provided continual guidance and management for CIAO and related software, helping ensure that scientific rigor, internal consistency, and long-term usability remained central design principles. He also emphasized interoperability, encouraging design choices that have helped keep CIAO useful beyond Chandra and in many cases applicable to data from other observatories.

The stability and reliability that users now expect from the Chandra software environment are, in no small part, evidence of Jonathan's approach to data systems: careful, detailed, and deeply informed by the needs of the scientific community.

Beyond Chandra, Jonathan is known worldwide for Jonathan's Space Report (https://planet4589.org), a long-running and highly respected chronicle of global spaceflight activity that he will continue to publish in retirement. His career and broader impact were also highlighted in the popular press, with features in the New York Times and New Scientist.

The CXC community—and especially all of us, the members of the Science Data Systems group—are deeply grateful for Jonathan's leadership, insight, and dedication. His legacy, embedded in the systems, the software, and the people he mentored and led, will continue to support Chandra science for years to come.

In this sense, CIAO 4.18 stands as both a technical milestone and a reminder of continuity. The software continues to evolve, addressing new challenges while supporting familiar workflows, just as the mission itself continues to deliver new science. At the same time, this release encourages us to recognize the individuals whose long-term commitment made that continuity possible. As Chandra analysis tools move forward, they do so on foundations carefully built and thoughtfully maintained over decades.