BAT-GLIMPSE

Gamma-ray Localization using Imaging and Mosaic Techniques for Pointing and Slew Epochs

BAT-GLIMPSE is a fully autonomous Swift/BAT localization pipeline for transient gamma-ray sources. It combines coded-mask imaging and mosaic techniques to recover precise arcminute-scale positions during both normal pointing observations and spacecraft slews, filling the gap left by onboard trigger outages and enabling rapid follow-up of GRBs and multi-messenger counterparts.

1. Installation

Requirements

HEASoft needs to be installed. We recommend to install it via Conda as described here, doing

conda create -n henv heasoft \
  -c https://heasarc.gsfc.nasa.gov/FTP/software/conda/ \
  -c conda-forge

In the same environment, install BAT-GLIMPSE and NITRATES as described in the next two steps. CALDB environment variables need to be defined as described here

Install via pip

The package can be installed via pip, with Python 3.10 or greater.

pip install bat-glimpse

Optional: installing NITRATES

In order to have the data setup managed by NITRATES, we need to install it locally using:

git clone git@github.com:Swift-BAT/NITRATES.git
cd NITRATES
python -m pip install -e .

If it fails, try to comment the line

voevent-parse>=1.0.3

in the file requirements.txt inside NITRATES repository, and repeat the installation. NITRATES is needed for data preparation and creation of time seeds. Even in absense of NITRATES time seeds, BAT-GLIMPSE performs its own time seeds search. Notice that when Swift is in slew only BAT-GLIMPSE produces time seeds.

Developer Mode

git clone git@github.com:samueleronchini/BAT-GLIMPSE.git
cd BAT-GLIMPSE
python -m pip install -e .

Running on SciServer

BAT-GLIMPSE can run with minimal installation steps on SciServer. Create a container with HEASARC version >= 6.36, and in the pre-installed heasoft conda env run

pip install bat-glimpse

CALDB environment variables should be already loaded.

2. How to run

bat-glimpse --workdir </path/to/workdir> --trigtime <trigtime>

You can also run the package module directly:

python -m batglimpse --workdir </path/to/workdir> --trigtime <trigtime>

Main options

• --workdir: required working directory for inputs and outputs.
• --trigtime: trigger time in YYYY-MM-DDTHH:MM:SS.sss format.
• --tmin and --tmax: explicit ad-hoc analysis window.
• --pipe: imaging or mosaic for ad-hoc analysis.
• --ext_obsid: override the GUANO obsid.
• --healpix_nside: mosaic HEALPix resolution.
• --skyview_nprocs: processes used while creating skyviews.
• --mosaic_nprocs: processes used while mosaicing.
• --trig_instr: Name of the triggering instrument

For --trig_instr use IGWN when it's a GW. This allows to create the preliminary maps with the partial coding distribution. Otherwise, by default the code searches for a Fermi-GBM map.

External Map Search

Fermi localization

• Uses gdt.missions.fermi if installed.
• Computes a Fermi trigger ID from trigtime.
• Downloads the localization and renames it to ext_loc_fermi_glg_*.fits.

GW localization (IGWN only)

• Queries GraceDb for the most relevant FITS sky map.
• Downloads it to workdir/ext_loc_*.fits.

Branch A: Ad-hoc analysis (tmin/tmax/pipe provided)

• If pipe == imaging:
  • Runs imaging() once for [tmin, tmax].
• If pipe == mosaic:
  • Runs mosaic() once for [tmin, tmax].
• If max SNR >= 6:
  • Sorts CSVs and posts results to Slack/Telegram.

Branch B: Default analysis (no explicit window)

1. NITRATES time seeds:
   • If time_seeds.csv exists and is non-empty:
     • Sort by snr and take top 10 seeds.
     • Call imaging() with those time windows.
   • If empty or missing, log and continue.
2. SNR check:
   • Reads SNR values from CSVs.
   • Posts to Slack/Telegram if SNR >= 6.
3. Custom seed search:
   • Runs cust_seeds(); if seeds are found, refine and re-image/mosaic:
     • If max SNR already >= 20, skip refinement.
     • For each seed within +/- 20 s:
       • Refine the seed center and duration.
       • If duration <= 0.2 s: run imaging().
       • If 0.2 s <= duration < 15 s:
         • Run mosaic() if interval intersects a slew interval.
         • Otherwise run imaging().

3. Algorithm Details

imaging(t0, event, workdir, ...)

• Energy range: 15-350 keV.
• Creates skyview with:
  • aperture=CALDB:DETECTION
  • pcodethresh=0.01
• Source detection parameters:
  • snrthresh=5.5, srcdetect=yes
• Filters detections:
  • NAME contains UNKNOWN.
  • SNR > 5 and CENT_SNR > 5.
• Writes imaging.csv with RA, Dec, SNR, CENT_SNR, partial coding, detect status, dt/duration, and energy bounds.

mosaic(t0, event, workdir, ...)

• Energy range: 15-350 keV.
• Initial duration dt_0 = tmax - tmin.
• Builds time bins in 0.2 s steps; uses a 3-bin fallback for short windows.
• Creates skyviews in parallel, then filters to those with:
  • sky_img, pcode_img, and bkg_stddev_img present.
• If 0 valid skyviews: double dt_0 and retry.
• If 1 valid skyview: fall back to that skyview.
• Otherwise mosaic with ba.parallel.mosaic_skyview().
• Detects sources with snrthresh=5.5.
• Accepts sources with psffwhm_separation > 1.
• Writes mosaic.csv with RA, Dec, SNR, t_start, t_end, and energy bounds.

4. Run examples

Always pass trigger time explicitly.

python run_bat_glimpse.py \
  --workdir /absolute/path/to/workdir \
  --trigtime 2020-03-25T03:18:35.000

Optional ad-hoc window:

python run_bat_glimpse.py \
  --workdir /absolute/path/to/workdir \
  --trigtime 2020-03-25T03:18:35.000 \
  --pipe imaging \
  --tmin -2.0 \
  --tmax 6.0

Mosaic run with explicit process counts:

python run_bat_glimpse.py \
  --workdir /absolute/path/to/workdir \
  --trigtime 2020-03-25T03:18:35.000 \
  --pipe mosaic \
  --tmin -5.0 \
  --tmax 15.0 \
  --healpix_nside 512 \
  --skyview_nprocs 8 \
  --mosaic_nprocs 8

Contact

Samuele Ronchini - Personal Website - samuele.ronchini@gssi.it

Acknowledgements

• If you make use of BAT-GLIMPSE, please cite the following papers and software:
  • Ronchini et al. 2026
  • https://doi.org/10.5281/zenodo.20689240
  • https://doi.org/10.5281/zenodo.7916508
  • Parsotan et. al. 2023
  • Parsotan et. al. 2025
• The BAT-GLIMPSE logo is designed by @HannahRonchiniArt