Fine-grained Differentiable Physics: A Yarn-level Model for Fabrics

Authors: Deshan Gong, Zhanxing Zhu, Andrew J. Bulpitt, He Wang

ICLR 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Through comprehensive evaluation and comparison, we demonstrate our model s explicability in learning meaningful physical parameters, versatility in incorporating complex physical structures and heterogeneous materials, data-efficiency in learning, and high-fidelity in capturing subtle dynamics. [...] 4 EXPERIMENTS
Researcher Affiliation Academia Deshan Gong1, Zhanxing Zhu2,3, Andrew J. Bulpitt1 and He Wang1 1School of Computing, University of Leeds 2School of Informatics, University of Edinburgh 3Peking University
Pseudocode No The paper describes methods through textual explanation and equations but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code Yes Code is available in: https://github.com/realcrane/Fine-grained-Differentiable-Physics-A-Yarn-level-Model-for-Fabrics.git
Open Datasets No We employ a traditional indifferentiable yarn-level simulator (Cirio et al., 2014) to generate the ground-truth data, and build a dataset of fabrics with three types of yarns and three types of woven patterns.
Dataset Splits No The paper specifies training on the first 5, 10, or 25 frames and testing on the whole 50 frames, but does not explicitly mention a separate validation split.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions software for data generation ('Cirio et al., 2014') and algorithms ('implicit Euler', 'Stochastic Gradient Descent'), but does not specify version numbers for any key software components or libraries used in their implementation.
Experiment Setup Yes We use Stochastic Gradient Descent and run 70 epochs for training. The simulation is conducted for 500 steps with h = 0.001s.