High-Rank Irreducible Cartesian Tensor Decomposition and Bases of Equivariant Spaces
Authors: Shihao Shao, Yikang Li, Zhouchen Lin, Qinghua Cui
JMLR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Both theoretical and experimental analyses confirm the efficiency of this approach. We also benchmark the speed of our algorithm for obtaining the set of all decomposition matrices for different ranks, as shown in Table 3. We run the experiments on 28-cores Intel R Xeon R Gold 6330 CPU @ 2.00GHz. |
| Researcher Affiliation | Academia | Shihao Shao EMAIL State Key Lab of Vascular Homeostasis and Remodeling Department of Bioinformatics School of Basic Medical Sciences, Peking University Yikang Li EMAIL State Key Lab for General AI School of Intelligence Science and Technology Peking University, China Zhouchen Lin EMAIL State Key Lab for General AI School of Intelligence Science and Technology, Peking University Pazhou Laboratory (Huangpu), China Qinghua Cui EMAIL State Key Lab of Vascular Homeostasis and Remodeling Department of Bioinformatics School of Basic Medical Sciences, Peking University |
| Pseudocode | Yes | Algorithm 1 Decomposition Matrices Generation Algorithm 2 Equivariant Basis Generation Algorithm 3 Equivariant Basis Generation Linear Combinations in Spherical Spaces |
| Open Source Code | Yes | The Python code is available at https://github.com/Shihao Shao-GH/ICT-decomposition-and-equivariant-bases, where the n = 6, . . . , 9 ICT decomposition matrices are obtained in 1s, 3s, 11s, and 4m32s on 28-core Intel R Xeon R Gold 6330 CPU @ 2.00GHz, respectively. |
| Open Datasets | No | The paper focuses on theoretical contributions and algorithm benchmarking, not on training machine learning models with specific datasets. No datasets are used for the experiments described in this paper. |
| Dataset Splits | No | The paper does not involve dataset-based experiments or machine learning model training; therefore, no dataset splits are applicable or mentioned. |
| Hardware Specification | Yes | The Python code is available at https://github.com/Shihao Shao-GH/ICT-decomposition-and-equivariant-bases, where the n = 6, . . . , 9 ICT decomposition matrices are obtained in 1s, 3s, 11s, and 4m32s on 28-core Intel R Xeon R Gold 6330 CPU @ 2.00GHz, respectively. |
| Software Dependencies | Yes | The only external dependency packages are Py Torch and e3nn. The e3nn package can be easily installed using the command pip install --upgrade e3nn. The code can run on any computer with Python version > 3.0.0 and Py Torch version > 1.0. |
| Experiment Setup | Yes | Shao, Li, Lin, and Cui n = 6, . . . , 9 ICT decomposition matrices are obtained in 1s, 3s, 11s, and 4m32s on 28-core Intel R Xeon R Gold 6330 CPU @ 2.00GHz, respectively. Table 3: Benchmarking decomposition matrices generation speed. |