Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2411.17728 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917849575981056 |
|---|---|
| author | Zhao, Zhe Xu, Jingping Wang, Ce Yang, Yaping |
| author_facet | Zhao, Zhe Xu, Jingping Wang, Ce Yang, Yaping |
| contents | Analytic continuation aims to reconstruct real-time spectral functions from imaginary-time Green's functions; however, this process is notoriously ill-posed and challenging to solve. We propose a novel neural network architecture, named the Feature Learning Network (FL-net), to enhance the prediction accuracy of spectral functions, achieving an improvement of at least $20\%$ over traditional methods, such as the Maximum Entropy Method (MEM), and previous neural network approaches. Furthermore, we develop an analytical method to evaluate the robustness of the proposed network. Using this method, we demonstrate that increasing the hidden dimensionality of FL-net, while leading to lower loss, results in decreased robustness. Overall, our model provides valuable insights into effectively addressing the complex challenges associated with analytic continuation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_17728 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Analytic Continuation by Feature Learning Zhao, Zhe Xu, Jingping Wang, Ce Yang, Yaping Strongly Correlated Electrons Machine Learning Signal Processing Computational Physics Analytic continuation aims to reconstruct real-time spectral functions from imaginary-time Green's functions; however, this process is notoriously ill-posed and challenging to solve. We propose a novel neural network architecture, named the Feature Learning Network (FL-net), to enhance the prediction accuracy of spectral functions, achieving an improvement of at least $20\%$ over traditional methods, such as the Maximum Entropy Method (MEM), and previous neural network approaches. Furthermore, we develop an analytical method to evaluate the robustness of the proposed network. Using this method, we demonstrate that increasing the hidden dimensionality of FL-net, while leading to lower loss, results in decreased robustness. Overall, our model provides valuable insights into effectively addressing the complex challenges associated with analytic continuation. |
| title | Analytic Continuation by Feature Learning |
| topic | Strongly Correlated Electrons Machine Learning Signal Processing Computational Physics |
| url | https://arxiv.org/abs/2411.17728 |