One Operator for Many Densities: Amortized Approximation of Conditioning by Neural Operators

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tsimpos, Panos, Calvello, Edoardo, Belhadji, Ayoub, Nelsen, Nicholas H.
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913113211666432
author Tsimpos, Panos
Calvello, Edoardo
Belhadji, Ayoub
Nelsen, Nicholas H.
author_facet Tsimpos, Panos
Calvello, Edoardo
Belhadji, Ayoub
Nelsen, Nicholas H.
contents Probabilistic conditioning is concerned with the identification of a distribution of a random variable $X$ given a random variable $Y$. It is a cornerstone of scientific and engineering applications where modeling uncertainty is key. This problem has traditionally been addressed in machine learning by directly learning the conditional distribution of a fixed joint distribution. This paper introduces a novel perspective: we propose to solve the conditioning problem by identifying a single operator that maps any joint density to its conditional, thus amortizing over joint-conditional pairs. We establish that the conditioning operator can be approximated to arbitrary accuracy by neural operators. Our proof relies on new results establishing continuity of the conditioning operator over suitable classes of densities. Finally, we learn the conditioning map for a class of Gaussian mixtures using neural operators, illustrating the promise of our framework. This work provides the theoretical underpinnings for general-purpose, amortized methods for probabilistic conditioning, such as foundation models for Bayesian inference.
format Preprint
id arxiv_https___arxiv_org_abs_2605_06873
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle One Operator for Many Densities: Amortized Approximation of Conditioning by Neural Operators
Tsimpos, Panos
Calvello, Edoardo
Belhadji, Ayoub
Nelsen, Nicholas H.
Machine Learning
Numerical Analysis
68T07 (Primary), 62E17, 65J15 (Secondary)
Probabilistic conditioning is concerned with the identification of a distribution of a random variable $X$ given a random variable $Y$. It is a cornerstone of scientific and engineering applications where modeling uncertainty is key. This problem has traditionally been addressed in machine learning by directly learning the conditional distribution of a fixed joint distribution. This paper introduces a novel perspective: we propose to solve the conditioning problem by identifying a single operator that maps any joint density to its conditional, thus amortizing over joint-conditional pairs. We establish that the conditioning operator can be approximated to arbitrary accuracy by neural operators. Our proof relies on new results establishing continuity of the conditioning operator over suitable classes of densities. Finally, we learn the conditioning map for a class of Gaussian mixtures using neural operators, illustrating the promise of our framework. This work provides the theoretical underpinnings for general-purpose, amortized methods for probabilistic conditioning, such as foundation models for Bayesian inference.
title One Operator for Many Densities: Amortized Approximation of Conditioning by Neural Operators
topic Machine Learning
Numerical Analysis
68T07 (Primary), 62E17, 65J15 (Secondary)
url https://arxiv.org/abs/2605.06873