All issues

Search results

rms_vis returns zero instead of a width in float32's last binade

Status: Open.

#235 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 8, 2026

The config advertises three values that setup rejects, and one it silently ignores

Status: Open.

#234 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 8, 2026

A config tabascal writes out cannot be run again: the gain widths are converted in place

Status: Open.

#233 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 8, 2026

Non-finite visibilities in the MS DATA column reach reductions that assume they are finite

Status: Open.

#232 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 8, 2026

The truth readers open a zarr without the guard written for it

Status: Open.

#231 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 8, 2026

The RFI prior's default width is a maximum where the key says typical

Status: Open.

#227 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 7, 2026

General config validation: report the keys a run never reads

Status: Open.

#226 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 7, 2026

Decide whether the RIC covariance scale should be a config knob

Status: Open.

#224 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 7, 2026

Estimate the astronomical prior width from the RFI model's own view of which samples are clean

Status: Open.

#220 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 7, 2026

Give the astronomical prior width a baseline-length envelope

Status: Open.

#219 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 7, 2026

init: sample draws a latent the model's prior does not have, so a run starts at half the prior's power

Status: Open.

#217 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 6, 2026

Decide what JAX settings tabascal owns: precision and matmul are still assigned over the user

Status: Open.

#215 In epfl-radio-astro/tabascal;

· chrisfinlay opened on Sep 6, 2026