specsr-roman

Physics-informed super-resolution of Roman grism spectra.

Roman’s High Latitude Spectroscopic Survey will deliver slitless grism spectra (R ≈ 461 λ/µm over 1–1.93 µm) for millions of emission-line galaxies. At that resolving power the diagnostic complexes stay blended. specsr-roman super-resolves them with a three-stage network, trained with losses that weight each line by how detectable it is in the grism data.

The specsr-roman pipeline in three bands: simulating the training data (OpenUniverse2024 SEDs dispersed through the Wang+2022 grism model, extracted with grizli into 23,845 LR-HR pairs); spectral super-resolution (SR1, a conservative 1D residual CNN emitting mean and log-variance, refined by SR2, an attention network with one token per emission line); and the redshift branch (a ZHead taking the coarse spectrum plus three Roman bands and emitting P(z), whose top modes gate which lines SR2 may draw).

from specsr_roman import RomanPipeline

pipe = RomanPipeline.from_pretrained()
out = pipe.predict(flux_low, flux_low_err, phot=roman_medium_fluxes)
out.flux_sr, out.z, out.pz

Guides

Tutorials

Reference