Conformal Crystal Graph Transformer with Robust Encoding of Periodic Invariance
Authors: Yingheng Wang, Shufeng Kong, John M. Gregoire, Carla P. Gomes
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Through comprehensive evaluation, we verify our model s superior performance in 5 crystal prediction tasks, reaffirming the efficiency of our proposed methods. and conducting comprehensive experiments over 5 tasks on the Jarvis materials benchmark (Choudhary et al. 2020), highlighting the significant of our components and showing our model s superior performance, and subsequently verifying the effectiveness of our proposed construction and learning methods. |
| Researcher Affiliation | Academia | 1Department of Computer Science, Cornell University, USA 2Liquid Sunlight Alliance, California Institute of Technology, USA 3School of Software Engineering, Sun Yat-sen University, China |
| Pseudocode | No | The paper includes 'Figure 1: Architecture Overview' which illustrates the model's components, but it is a high-level diagram and not a structured pseudocode or algorithm block with step-by-step instructions. |
| Open Source Code | No | The paper does not provide any explicit statement or link for open-source code for the described methodology. |
| Open Datasets | Yes | We test on five crystal property prediction tasks using the JARVIS (Choudhary et al. 2020) benchmark, specifically its DFT-2021.8.18 3D version, which features 55,722 crystals. |
| Dataset Splits | Yes | We adopt data splits as per (Lin et al. 2023; Yan et al. 2022) to ensure a fair comparison. |
| Hardware Specification | No | The paper discusses computational efficiency but does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments. |
| Software Dependencies | No | The paper mentions various optimizers and baselines, but it does not specify version numbers for any software dependencies or libraries (e.g., 'PyTorch 1.9' or 'Python 3.8'). |
| Experiment Setup | Yes | Learning rates and training epochs are mildly adjusted, starting from 0.0005 and 1000 respectively, depending on the task. Specific configurations for each task can be found in the Appendix. |