Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
PIVNO: Particle Image Velocimetry Neural Operator
Authors: Xu Jie, Xuesong Zhang, Jing Jiang, Qinghua Cui
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive evaluations demonstrate the accuracy, flexibility, and robustness of our approach across both simulated and experimental PIV datasets. Our code is at https://github.com/ZXS-Labs/PIVNO. The experimental evaluation comprises three synthetic datasets and three real-world PIV challenge tasks. |
| Researcher Affiliation | Academia | 1Beijing University of Posts and Telecommunications 2Beijing Union University 3Peking University 4Wuhan Sports University EMAIL EMAIL |
| Pseudocode | No | The paper describes algorithmic steps and equations for its components (e.g., Conv-GRU operations from (8) to (11)), but it does not present them in a clearly labeled "Pseudocode" or "Algorithm" block. |
| Open Source Code | Yes | Our code is at https://github.com/ZXS-Labs/PIVNO. |
| Open Datasets | Yes | The experimental evaluation comprises three synthetic datasets and three real-world PIV challenge tasks. Initially, supervised training and benchmark testing are conducted on Synthetic Datasets 1 and 2. Synthetic Dataset 1: This PIV dataset [23] contains five classic flow field cases... Synthetic Dataset 2: To evaluate the model s performance under large displacement and high noise conditions, we used a synthetic dataset from [25]... Synthetic Dataset 3: To evaluate the impact of the three loss terms (Ld, Ls, and Ldiv) on flow field estimation, we conducted ablation experiments using the SPID dataset [62]... Solid Body Rotation Flow: To evaluate the generalization capability of the fine-tuning strategy in real-world scenarios, we selected the solid body rotation flow [63] as a classical benchmark... Strong Vortex: We use PIVNO to process the vortex flow field images recorded by the German Aerospace Center in the DNW-LLF large wind tunnel [64]... Turbulent Jet: We evaluated the turbulent round jet dataset from Delft University of Technology [66]... |
| Dataset Splits | No | The paper mentions 'supervised training' and 'benchmark testing' on synthetic datasets 1 and 2, and 'self-supervised fine-tuning' on synthetic dataset 3 and real-world benchmarks, but does not provide specific percentages, sample counts, or explicit splitting methodology for training, validation, or test sets in the main text. |
| Hardware Specification | No | The main text of the paper does not specify any particular GPU models, CPU models, memory amounts, or other detailed computer specifications used for running the experiments. It states that detailed information about the experimental environment, including hardware, is provided in the supplementary material. |
| Software Dependencies | No | The main text of the paper does not provide specific software dependencies, such as programming languages, libraries, or solvers with their corresponding version numbers, that are needed to replicate the experiment. |
| Experiment Setup | Yes | Training samples are generated by uniformly sampling downsampling factors within the range of 1 to 4... In practice, we find that a small number of refinement steps suffices: five iterations achieve near-optimal performance... The self-supervised loss function is defined as: LP(u) = Ld(u) + λs Ls(u) + λd Ldiv(u), where Ld represents the data term, modeling the similarity of the image pairs, Ls and Ldiv are the spatial smoothing and divergence regularization terms respectively, and λs and λd are their respective weights. |