FLITS

Fast-Look Interactive Transient Suite.

Browser-based scientific software for interactive burst inspection, masking, and measurement on filterbank data.

Known presets can supply a default SEFD when the observing setup is identifiable from the file metadata and band coverage. Today that means:

• NRT uses its preset SEFD
• GBT auto-selects a default SEFD for common Gregorian bands (for example L-band files default to 10 Jy)

For data without a known default calibration, use the Generic Filterbank preset or provide an explicit SEFD override if you want calibrated flux and fluence values.

Install With Python

Install from a local checkout:

python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install .

You can also install directly from GitHub:

pip install "git+https://github.com/DirkKuiper/flits.git"

Then run FLITS against a directory of filterbanks:

flits --data-dir /path/to/filterbanks --host 127.0.0.1 --port 8123

Open http://127.0.0.1:8123.

Notes:

• Relative file paths in the UI are resolved against FLITS_DATA_DIR when set, otherwise against the current working directory.
• The --data-dir flag is the easiest way to point FLITS at a specific directory without exporting environment variables.
• The known-filterbanks dropdown lists .fil files recursively under that data directory.

Run With Docker

The canonical container image is intended to live at ghcr.io/dirkkuiper/flits.

If you have a published image available, run:

docker run --rm -p 8123:8123 \
  -e FLITS_DATA_DIR=/data \
  -v /path/to/filterbanks:/data \
  ghcr.io/dirkkuiper/flits:latest

If you want to build locally instead:

docker build -t flits .
docker run --rm -p 8123:8123 \
  -e FLITS_DATA_DIR=/data \
  -v /path/to/filterbanks:/data \
  flits

Then open http://127.0.0.1:8123.

Notes:

• The container defaults FLITS_DATA_DIR to /data.
• Relative file paths in the UI are resolved against FLITS_DATA_DIR.
• The known-filterbanks dropdown lists .fil files recursively under FLITS_DATA_DIR.
• Absolute paths inside the container still work if you prefer to type them manually.

Run With Apptainer

Apptainer users should consume the same OCI image rather than maintain a separate container recipe.

If the published GHCR image is available:

apptainer pull flits.sif docker://ghcr.io/dirkkuiper/flits:latest
APPTAINERENV_FLITS_DATA_DIR=/data \
  apptainer exec --bind /path/to/filterbanks:/data flits.sif \
  flits --data-dir /data --host 127.0.0.1 --port 8123

If you are on an HPC system where pulling from GHCR is inconvenient, you can still build or export the Docker image elsewhere and convert it to .sif with Apptainer.

Run With Docker Compose

If your filterbanks are in the current directory:

docker compose up --build

If your filterbanks live somewhere else, point DATA_DIR at that directory:

DATA_DIR=/absolute/path/to/filterbanks docker compose up --build

You can also choose a different host port:

DATA_DIR=/absolute/path/to/filterbanks FLITS_PORT=9000 docker compose up --build

Then open http://127.0.0.1:8123 or http://127.0.0.1:9000 respectively.

Remote And HPC Use

FLITS is portable across local workstations, remote servers, and HPC systems:

• On a normal workstation, pip install ., docker run, or docker compose are all fine.
• On a remote Linux machine you access over SSH, start FLITS remotely and forward the port:

ssh -L 8123:127.0.0.1:8123 user@remote-host

Then open http://127.0.0.1:8123 locally in your browser.

• On many HPC clusters, Docker itself is not permitted. Apptainer is usually the right runtime there.
• The GitHub Actions workflow in .github/workflows/publish-image.yml is the intended way to publish the canonical OCI image to GHCR.

Code Structure

• flits/settings.py: observation presets and explicit overrides
• flits/io/filterbank.py: filterbank loading and Stokes-I extraction
• flits/signal.py: shared numerical utilities
• flits/models.py: typed metadata and measurement containers
• flits/session.py: interactive burst state and measurements
• flits/web/app.py: FastAPI server for the browser UI
• flits/web_static/: packaged frontend assets served by the app
• tests/test_session_smoke.py: smoke tests on a real local filterbank file

Measurements

• Fluence (Jy ms), when an SEFD is provided
• Peak flux density (Jy), when an SEFD is provided
• Event duration (ms)
• Spectral extent (MHz)
• Peak MJD
• 1D Gaussian fits to selected burst regions

Flux and fluence are computed from the burst-only time series using the radiometer equation with the effective unmasked bandwidth inside the selected spectral extent.