🔭 astrophysics

Hostless extragalactic transients in Fink: Results from the ELEPHANT pipeline

This paper evaluates the ELEPHANT pipeline within the Fink broker, demonstrating its ability to identify a high-purity sample of genuine hostless extragalactic transients with 84% accuracy while characterizing contamination sources and confirming its adaptation for the Rubin alert stream.

Original authors: R. Durgesh, P. J. Pessi, E. E. O. Ishida, J. Peloton

Published 2026-05-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: R. Durgesh, P. J. Pessi, E. E. O. Ishida, J. Peloton

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the night sky as a massive, bustling city. Most of the time, when a bright "firework" (a stellar explosion called a supernova) goes off, it happens right in front of a building (a host galaxy). Astronomers can usually see the building behind the firework.

But sometimes, a firework goes off in the middle of a dark, empty field. There is no building in sight. These are called hostless transients. They are mysterious because we don't know what kind of "neighborhood" they came from. Are they from a tiny, invisible house? Did they get kicked out of their home? Or are they just floating in deep space?

This paper is about a new digital tool called ELEPHANT (which stands for "ExtragaLactic alErt Pipeline for Hostless AstroNomical Transients") that helps astronomers find these lonely fireworks in real-time.

Here is a simple breakdown of what the paper says:

1. The Problem: Finding a Needle in a Haystack

Astronomers use powerful telescopes (like the Zwicky Transient Facility, or ZTF) that take millions of pictures of the sky every night. It's like having a security camera system for the entire universe.

The Challenge: When a new explosion happens, the computer has to decide: "Is there a galaxy behind this explosion?"
The Old Way: Humans would have to look at every single picture to check. This is too slow and too many pictures to handle.
The New Way (ELEPHANT): The authors built an automated "smart filter" inside a system called Fink. This filter acts like a bouncer at a club. It looks at the pictures and instantly flags the ones that look like they are floating in empty space.

2. How the Tool Works (The "Stamp" Analogy)

Think of the telescope images as postage stamps.

The Process: ELEPHANT takes a tiny square cutout (a stamp) from the center of the image where the explosion happened.
The Test: It compares the "Science" stamp (the new photo with the explosion) against a "Template" stamp (an old photo of the same spot before the explosion).
The Magic Trick: If there was a galaxy there, the old photo would show a faint, fuzzy blob. ELEPHANT uses a mathematical trick (looking at the "frequency" of the pixels, kind of like analyzing the texture of a fabric) to see if that faint fuzz is there.
- If the texture looks like random noise (like static on an old TV), the tool says, "No galaxy here! This is a hostless candidate!"
- If the texture shows a pattern, it says, "There's a building here. Ignore this one."

3. The Results: What Did They Find?

The team ran this tool on data from September 2023 to December 2025.

The Volume: They looked at over 3 million alerts.
The Hits: The tool flagged 877 events as "hostless." That's a tiny fraction (about 0.6%), which is good because it means the tool is being picky.
The Cleanup: Not every flag was perfect.
- False Alarms: Some of the "hostless" events were actually just Cataclysmic Variables (dying stars in our own galaxy that look like fireworks but aren't) or moving stars (stars that move so fast they look like they are leaving a trail). The tool caught these, but they aren't the "extragalactic" fireworks the team was looking for.
- The Real Deal: After cleaning out the false alarms and checking deep archives, they found 67 events that are almost certainly truly hostless.
The Mystery: For these 67 events, the "host galaxy" is so faint that even the best existing maps of the universe can't see it. It's like trying to see a firefly in a dark room with a flashlight that isn't quite bright enough. The paper estimates these invisible hosts are incredibly dim, far fainter than typical dwarf galaxies.

4. The Types of Fireworks

Among the confirmed hostless events, the most common type was Type Ia Supernovae (standard explosions used to measure the universe). The second most common were Superluminous Supernovae (massive, incredibly bright explosions).

Why does this matter? Superluminous supernovae are so bright they can outshine their entire home galaxy. This paper suggests that these bright explosions act like "beacons," helping us discover tiny, hidden galaxies that we couldn't see otherwise.

5. The Future: Getting Ready for a Bigger Camera

The paper mentions that this tool is already being updated for the Vera C. Rubin Observatory (a future telescope that will take even deeper, sharper pictures than ZTF).

The Upgrade: The new version of ELEPHANT is being tuned to handle the massive flood of data from this new telescope.
The Goal: With the new telescope, we will be able to see even fainter galaxies. The authors believe that as our "flashlights" get brighter, we will find even more of these hostless events, which will help us understand how stars and galaxies form and evolve.

Summary

In short, the authors built a smart, automated filter that scans the sky for explosions happening in "empty" space. They found hundreds of candidates, cleaned up the false alarms, and confirmed a high-quality list of truly lonely cosmic events. This tool is now ready to help the next generation of telescopes discover the faintest, most hidden corners of our universe.

Technical Summary: Hostless Extragalactic Transients in Fink

Problem Statement
The systematic study of extragalactic transients requires understanding their local environments, yet a significant fraction (estimated between 2% and 7%) of observed transients appear "hostless," lacking an identifiable underlying galaxy. This phenomenon presents two primary challenges: distinguishing between transients in extremely faint, undetected dwarf galaxies versus those genuinely unassociated with any host (e.g., runaway stars or intracluster systems), and managing the contamination from non-extragalactic sources such as Cataclysmic Variables (CVs) and high proper motion stars. Current surveys like the Zwicky Transient Facility (ZTF) are limited by magnitude depth and stamp size, often failing to detect faint hosts, which necessitates automated, real-time filtering to isolate high-purity hostless candidates for follow-up.

Methodology
The authors evaluate the ELEPHANT (ExtragaLactic alErt Pipeline for Hostless AstroNomical Transients) pipeline, implemented within the Fink alert broker. The pipeline operates on ZTF science and template stamps with minimal input requirements:

Image Processing: The pipeline crops 15×15 pixel (approx. 15×15 arcsec) regions centered on the transient. It applies sigma-clipping ( $\sigma=3$ ) to detect bright pixels. If a source is present in the science image but absent in the template (or vice versa), the candidate proceeds to statistical analysis.
Statistical Classification: The core of the pipeline utilizes a power spectrum analysis. It transforms pixel values into the frequency domain and computes the Wasserstein distance between the original image and a set of shuffled counterparts. A Kolmogorov–Smirnov (KS) test compares these distributions; a significant difference indicates the presence of spatially coherent structure (a host). If no structure is detected, the event is flagged as hostless.
Filtering Constraints: The pipeline processes alerts from the last 45 days, applies a magnitude cut ( $<19.5$ mag) for spectroscopic feasibility, requires at least three detections, and filters out asteroids using the d:roid column.
Validation & Cross-Matching: Flagged candidates were analyzed between September 1, 2023, and December 31, 2025. The authors augmented Fink's internal cross-matching with external tools Pröst (using Bayesian methods against GLADE+, DECaLS, Pan-STARRS, and SkyMapper) and Sherlock (calculating angular separations against Gaia, SDSS, 2MASS, etc.).
Visual Inspection: Candidates without catalogued hosts underwent visual inspection of Legacy Survey (LS-DR10) and Pan-STARRS images to identify faint, uncatalogued hosts.

Key Contributions and Results

Pipeline Performance: Out of 877 flagged events (representing $\sim0.6\%$ of total ZTF events in the period), the pipeline achieved an overall accuracy of 84.03%. The primary contaminants were identified as CVs (15.84% of flagged events) and high proper motion stars (3 events).
Hostless Sample Characterization:
- Confirmed Hostless: 67 events were confidently confirmed as genuinely hostless, with no detectable host in existing catalogues or archival imaging.
- Visually Identified Hosts: 51 events were found to have visually identifiable hosts that were absent from both catalogues and ZTF stamps, representing faint systems below current detection limits.
- Spectroscopic Classification: Of the 877 flagged candidates, 276 had spectroscopic classifications in the Transient Name Server (TNS). The dominant class was Type Ia Supernovae (162 events, $\sim58.7\%$ ), followed by Type I Superluminous Supernovae (SLSN-I) (30 events, $\sim10.9\%$ ).
Host Magnitude Limits: For the confirmed hostless subset, the inferred upper limits on host-galaxy absolute magnitudes ( $M_{r, upper}$ ) range from $-9.6$ to $-18$. These limits extend well below the luminosity range of typical dwarf galaxies, suggesting the hosts are intrinsically faint or absent.
LSST Adaptation: The pipeline has been adapted for the Rubin Observatory (LSST) alert stream. Key modifications include:
- Integration of cross-matches with Gaia DR3, Legacy DR8, SIMBAD, SPICY, 2MASS, and HyperLEDA.
- Implementation of astrometric filtering (fitting a first-order polynomial to RA/Dec) to distinguish moving Solar System Objects (SSOs) from extragalactic transients.
- Adjustment of stamp sizes to 30×30 pixels (approx. 6×6 arcsec) to accommodate LSST's smaller stamp footprint while maintaining spatial context.
- Early testing on the LSST stream has already identified candidates, including spectroscopically confirmed SLSN-I events (SN 2026ejf and SN 2025altn).

Significance and Claims
The paper claims that ELEPHANT is an efficient tool for the automatic identification of hostless transients in ZTF data, successfully isolating a high-purity sample of intrinsically faint or absent hosts. The authors emphasize that the framework is adaptable to the upcoming Rubin/LSST data stream, which will offer deeper limiting magnitudes ( $\sim24.7$ mag).

The significance of this work lies in its potential to enable population studies of hostless events to unprecedented depths. The authors posit that while current hostless detections are often attributed to the limiting magnitude of surveys, the deeper LSST data will likely remove these observational biases. Consequently, the surviving population of hostless candidates in the LSST era will require new theoretical frameworks to explain their origin, potentially challenging current models of stellar dynamics, galaxy formation, and the incidence of supernovae from runaway stars. The paper aims to "sparkle" this revolution by providing the necessary automated infrastructure to identify these rare events in real-time.