Long Range Frequency Tuning for QML
Michael Poppel, Jonas Stein, Sebastian Wölckert, Markus Baumann, Claudia Linnhoff-Popien · 2026
Read Long Range Frequency Tuning for QML by Michael Poppel, Jonas Stein, Sebastian Wölckert, Markus Baumann, Claudia Linnhoff-Popien (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-23409.txt.
Description
Michael Poppel and four co-authors tackle long-range frequency tuning for quantum learning. Angle-encoded variational circuits output a truncated Fourier series, but fixed unary encoding needs about as many encoding gates as the maximum frequency. Trainable-frequency circuits promise relief by learning prefactors beside the ansatz parameters, adapting the spectrum to the target.
The barrier: the prefactor gradient is suppressed by the spectral gap, pinning movement near initialization. Ternary grid initialization sets prefactors to powers of three, so every target frequency lies within half a unit of the spectrum at start. On a synthetic benchmark with shifted target frequencies, ternary initialization reaches median R-squared of zero point nine nine seven versus zero point one eight for unary. All runs beat zero point nine five, against none for unary. Even CMA-ES with twenty times the budget reaches only 25 percent.
Catalog record
| CREATOR | Michael Poppel, Jonas Stein, Sebastian Wölckert, Markus Baumann, Claudia Linnhoff-Popien |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-23409