LATED

Measure emission lines you can't spectroscopically resolve, straight from photometry.

You have JWST (or Roman) photometry of a galaxy, and one of the bands looks too bright: an emission line is boosting it. lated turns that excess into a number. It fits a simple continuum + emission-line model through the real filter curves and reports the line fluxes, rest-frame equivalent widths and line ratios (for example [O III] 5007 / Hβ) that the photometry implies.

When a line is genuinely there, you get a measurement with uncertainties. When it isn't, you get an honest upper limit, not a fabricated value. That distinction is the whole point of the tool.

Use it two ways: a point-and-click web app (no coding), or a small Python API for scripting and batch work.

Install

pip install lated

That is everything: the Python library, the command line, and the web app. Python 3.9+ (depends only on NumPy, SciPy, pydantic and FastAPI).

The web app

lated serve            # open http://127.0.0.1:8321

Pick a redshift, tick the bands you have, paste your fluxes, and press run. You get a verdict card (measurement or limit), an SED figure and a corner plot, all exportable to PDF. Start from one of the built-in literature examples to see the whole workflow in a single click. Everything on the board is editable: the line list, which lines are free / tied / off, the nebular conditions that set the Balmer ratios (Osterbrock & Ferland 2006 Case A/B by temperature and density), the continuum slope, dust, and the error budget.

From Python

from lated import FitConfig, fit

photometry = {"F200W": (0.0136, 0.0019),   # {band: (flux, error)} in microJansky
              "F277W": (0.0249, 0.0016),
              "F356W": (0.0040, 0.0014)}

res = fit(photometry, FitConfig.fixed_slope(z=3.193))
print(res.ratios[0].status)                # 'upper_limit', 'measured', ...

That is the whole idea. The example notebook walks through reading the results, plotting the SED and corner, editing line roles, choosing nebular conditions, and the posterior error budget.

What you get back

Every line ratio carries a status, so a number is never mistaken for a measurement it isn't:

Each line also reports the faintest flux your photometry could have detected, so every non-detection comes with its depth.

How it works

Per band, the predicted flux is the photon-weighted mean of

f_λ(λ) = C (λ/λ0)^β  +  Σ_l  F_l · δ(λ − λ_l (1+z))

integrated through the real transmission curve and compared to your photometry with a Gaussian likelihood; amplitudes are solved by error-weighted non-negative least squares over a grid of slopes. Every parameter has an explicit role (free, fixed, or tied) that you can change, and invalid setups (unknown bands, tie cycles, non-positive errors, under-constrained fits) are raised as errors rather than silently ignored.

Command line

lated serve                 # run the web app
lated fit input.json        # fit from a JSON file
lated bands                 # list the bundled filter curves
lated lines                 # show the default line registry

The web app also exposes a small REST API (/api/meta, /api/filters/{name}, /api/fit) with interactive docs at /docs.

Citing

If lated contributes to your research, please cite Li et al. (2026, LATED Paper I).

Development

git clone https://github.com/lmytime/lated
cd lated
pip install -e ".[test]"
PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 pytest tests/ -q

Issues and contributions are welcome at github.com/lmytime/lated. Released under the MIT License.