Convolutional Neural Operators for robust and accurate learning of PDEs

Authors: Bogdan Raonic, Roberto Molinaro, Tim De Ryck, Tobias Rohner, Francesca Bartolucci, Rima Alaifari, Siddhartha Mishra, Emmanuel de Bézenac

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental CNOs are tested on a novel suite of benchmarks, encompassing a diverse set of PDEs with possibly multi-scale solutions and are observed to significantly outperform baselines, paving the way for an alternative framework for robust and accurate operator learning.
Researcher Affiliation Academia 1 Seminar for Applied Mathematics, ETH, Zurich, Switzerland 2 ETH AI Center, Zurich, Switzerland 3 Delft University of Technology, Netherlands
Pseudocode No The paper describes the architecture and components but does not provide pseudocode or algorithm blocks.
Open Source Code Yes 1The code can be found at https://github.com/bogdanraonic3/Convolutional Neural Operator
Open Datasets Yes access to training and test data is readily available for rapid prototyping and reproducibility
Dataset Splits Yes For each such hyperparameter configuration, the corresponding models are trained on the benchmark and the configuration with smallest validation error is selected and the resulting test errors are reported
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models) used for experiments.
Software Dependencies No The paper does not specify software dependencies with version numbers.
Experiment Setup No We provide a detailed description of the implementation of CNO and the training (and test) protocol for CNO as well as all the baselines in SM C.1.