Simulation-Efficient Cosmological Inference with Multi-Fidelity SBI
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913920879427584 |
|---|---|
| author | Thiele, Leander Bayer, Adrian E. Takeishi, Naoya |
| author_facet | Thiele, Leander Bayer, Adrian E. Takeishi, Naoya |
| contents | The simulation cost for cosmological simulation-based inference can be decreased by combining simulation sets of varying fidelity. We propose an approach to such multi-fidelity inference based on feature matching and knowledge distillation. Our method results in improved posterior quality, particularly for small simulation budgets and difficult inference problems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_00514 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Simulation-Efficient Cosmological Inference with Multi-Fidelity SBI Thiele, Leander Bayer, Adrian E. Takeishi, Naoya Cosmology and Nongalactic Astrophysics Machine Learning The simulation cost for cosmological simulation-based inference can be decreased by combining simulation sets of varying fidelity. We propose an approach to such multi-fidelity inference based on feature matching and knowledge distillation. Our method results in improved posterior quality, particularly for small simulation budgets and difficult inference problems. |
| title | Simulation-Efficient Cosmological Inference with Multi-Fidelity SBI |
| topic | Cosmology and Nongalactic Astrophysics Machine Learning |
| url | https://arxiv.org/abs/2507.00514 |