Connection between memory performance and optical absorption in quantum reservoir computing
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914080994885632 |
|---|---|
| author | Götting, Niclas Wilksen, Steffen Steinhoff, Alexander Lohof, Frederik Gies, Christopher |
| author_facet | Götting, Niclas Wilksen, Steffen Steinhoff, Alexander Lohof, Frederik Gies, Christopher |
| contents | The fading memory property is a key requirement for reservoir computers -- a specific type of recurrent neural network with fixed internal weights. While mostly undesired in gate-based quantum computing, dissipation due to material imperfections or coupling to the environment acts as a natural mechanism intrinsically providing fading memory to reservoir computers based on dynamical open quantum systems. In this work, we unravel a connection between the physical metric of optical absorption and the performance of quantum reservoir computers in terms of their short-term memory capacity. We establish this link by considering a coherent input encoding in conjunction with tunable qubit decay, giving precise control over the fading memory in the quantum reservoir computer. Our analysis enables us to identify a sweet-spot regime for the dissipation strength at which memory performance is maximized. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_15580 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Connection between memory performance and optical absorption in quantum reservoir computing Götting, Niclas Wilksen, Steffen Steinhoff, Alexander Lohof, Frederik Gies, Christopher Quantum Physics The fading memory property is a key requirement for reservoir computers -- a specific type of recurrent neural network with fixed internal weights. While mostly undesired in gate-based quantum computing, dissipation due to material imperfections or coupling to the environment acts as a natural mechanism intrinsically providing fading memory to reservoir computers based on dynamical open quantum systems. In this work, we unravel a connection between the physical metric of optical absorption and the performance of quantum reservoir computers in terms of their short-term memory capacity. We establish this link by considering a coherent input encoding in conjunction with tunable qubit decay, giving precise control over the fading memory in the quantum reservoir computer. Our analysis enables us to identify a sweet-spot regime for the dissipation strength at which memory performance is maximized. |
| title | Connection between memory performance and optical absorption in quantum reservoir computing |
| topic | Quantum Physics |
| url | https://arxiv.org/abs/2501.15580 |