foundax/CITATION.cff at main · FhG-IISB/foundax · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
cff-version: 1.2.0
message: "If you use foundax in academic work, please cite the accompanying paper."
title: "foundax: A Unified JAX Model Zoo for Operator Learning and PDE Foundation Models"
version: "0.2.0"
date-released: 2026-05-31
license: MIT
repository-code: "https://github.com/FhG-IISB/foundax"
url: "https://github.com/FhG-IISB/foundax"

abstract: >
  A small Equinox-based collection of JAX models for operator learning and PDE surrogates:
  core architectures (FNO, U-Net, DeepONet, transformer, GNOT family, …), the KAN family,
  and wrappers around eight vendored foundation models (Poseidon, MORPH, MPP, Walrus,
  BCAT, PDEformer-2, DPOT, PROSE). Integrates with the jNO library for physics-informed
  training.

keywords:
  - JAX
  - Equinox
  - neural operators
  - Kolmogorov-Arnold Networks
  - foundation models
  - PDE surrogates
  - scientific machine learning

authors:
  - family-names: Armbruster
    given-names: Leon
    affiliation: "Fraunhofer IISB, Germany"
    orcid: "https://orcid.org/0009-0007-7602-362X"

preferred-citation:
  type: article
  authors:
    - family-names: Armbruster
      given-names: Leon
    - family-names: Ramesh
      given-names: Rathan
    - family-names: Kruse
      given-names: Georg
    - family-names: Straub
      given-names: Christopher
  title: "jNO: A JAX Library for Neural Operator and Foundation Model Training"
  year: 2026
  doi: "10.48550/arXiv.2605.10159"
  url: "https://arxiv.org/abs/2605.10159"
  note: "Preprint on arXiv (arXiv:2605.10159). foundax is the model-zoo component described in this paper."