Unbalanced L1 optimal transport for vector valued measures and application to Full Waveform Inversion - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2024

Unbalanced L1 optimal transport for vector valued measures and application to Full Waveform Inversion

Résumé

Optimal transport has recently started to be successfully employed to define misfit or loss functions in inverse problems. However, it is a problem intrinsically defined for positive (probability) measures and therefore strategies are needed for its applications in more general settings of interest. In this paper we introduce an unbalanced optimal transport problem for vector valued measures starting from the $L^1$ optimal transport. By lifting data in a self-dual cone of a higher dimensional vector space, we show that one can recover a meaningful transport problem. We show that the favorable computational complexity of the $L^1$ problem, an advantage compared to other formulations of optimal transport, is inherited by our vector extension. We consider both a one-homogeneous and a two-homogeneous penalization for the imbalance of mass, the latter being potentially relevant for applications to physics based problems. In particular, we demonstrate the potential of our strategy for full waveform inversion, an inverse problem for high resolution seismic imaging.
Fichier principal
Vignette du fichier
OT1VECT.pdf (8.86 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04486680 , version 1 (04-03-2024)
hal-04486680 , version 2 (02-04-2024)

Identifiants

Citer

Gabriele Todeschi, Ludovic Métivier, Jean-Marie Mirebeau. Unbalanced L1 optimal transport for vector valued measures and application to Full Waveform Inversion. 2024. ⟨hal-04486680v2⟩
29 Consultations
3 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More