BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:3D NLTE Radiative Transfer with Factorized Fourier Neural Operator
 s [Astrophysics Seminar]
DTSTART:20260924T060000Z
DTEND:20260924T070000Z
DTSTAMP:20261004T210500Z
UID:indico-event-467@events.iiap.res.in
CONTACT:colloqm@iiap.res.in
DESCRIPTION:Speakers: Harsh Mathur (Rosseland Centre for Solar Physics\, I
 nstitute of Theroratical Astrophysics\, University of Oslo)\n\nAbstractThr
 ee-dimensional non-local thermodynamic equilibrium (3D NLTE) radiative tra
 nsfer is essential for interpreting modern solar and stellar atmosphere si
 mulations\, but conventional codes such as Multi3D remain computationally 
 expensive for large snapshot sequences and repeated spectral synthesis. We
  present a Fourier Neural Operator-based framework designed as a geometry-
 aware surrogate to calculate 3D NLTE population calculations from 3D stell
 ar atmospheres. Unlike earlier machine-learning surrogates that use local 
 neighborhoods to predict a center voxel or center column\, our framework l
 earns an operator from the full 3D atmospheric state to the full 3D NLTE d
 eparture coefficient volume. The model is therefore not a column-wise corr
 ection scheme\, but a volumetric predictor trained directly on whole-atmos
 phere input-output fields. During training\, it sees the surrounding atmos
 pheric structure as part of the target itself\, rather than using neighbor
 ing columns only as auxiliary context. The inputs include temperature\, de
 nsity\, electron density\, and velocity components\, while the targets are
  departure coefficients for selected atomic levels. The architecture combi
 nes horizontal spectral operators with a coordinate-conditioned vertical b
 ranch\, enabling it to capture lateral radiative coupling and depth-depend
 ent NLTE structure. Crucially\, the model explicitly incorporates physical
  horizontal grid spacings dx and dy\, and accepts a native z_scale\, inclu
 ding arbitrary non-uniform vertical grids. It is therefore not tied to fix
 ed pixel spacing\, a uniform depth index\, or a remapped common vertical c
 oordinate. This makes the framework a true volumetric surrogate for Multi3
 D: it preserves global atmospheric context\, supports native simulation ge
 ometry\, and outputs complete 3D departure coefficient cubes for downstrea
 m spectral synthesis.\n\nhttps://events.iiap.res.in/event/467/
LOCATION:Auditorium
URL:https://events.iiap.res.in/event/467/
END:VEVENT
END:VCALENDAR