arXiv:2609.28760
Walsh-Transform Realization of Dense-to-Sparse Quantum State Preparation
Emad Rezaei Fard Boosari
Emad Rezaei Fard BoosariAbstract
We propose a hybrid quantum state preparation method based on the Walsh--Hadamard transform within the dense-to-sparse quantum state preparation framework. The method approximately prepares structured classical data in a quantum circuit through an indirect approach. Instead of directly preparing the dense amplitude-encoded state, the method first transforms the classical data into the Walsh domain and prepares only the largest-magnitude coefficients using a sparse state-preparation algorithm. The original state can then be approximately recovered through a parallel layer of Hadamard gates. Since this reconstruction requires no CNOT gates and has circuit depth 1, the proposed method introduces no additional CNOT or circuit-depth overhead, and its quantum preparation cost is determined by the underlying sparse state-preparation algorithm. Numerical results on representative benchmark signals demonstrate that the proposed method enables accurate state preparation for signals admitting sparse or approximately sparse representations in the Walsh domain.
AI Summary
The paper studies approximate amplitude encoding of structured classical signals using a Walsh–Hadamard realization of the existing dense-to-sparse state-preparation framework. It classically transforms an $N=2^n$-component vector, retains its $d$ largest-magnitude Walsh coefficients, prepares the normalized sparse coefficient state, and applies $H^{\otimes n}$ to reconstruct an approximation to the target state. The quantum decompression is exact, requires no CNOT gates, and adds one parallel gate layer. Preparation costs therefore follow the chosen sparse-state encoder: the quoted circuit-size bounds are $\mathcal{O}(nd)$ without ancillas and $\mathcal{O}(nd/\log(n+m)+n)$ with $m$ ancillas. Unlike the probabilistic Walsh loaders discussed in the paper, this construction uses deterministic sparse preparation without postselection or amplitude amplification. Evidence consists of classical numerical compression and reconstruction experiments on eight one-dimensional benchmark signals, with dimensions up to $N=16384$. The selected square-wave, step, and piecewise-constant examples require exactly one, two, and three Walsh coefficients. At $N=16384$ and relative coefficient-truncation error $10^{-5}$, the sinusoid requires 206 coefficients; Gaussian, sinc, and Gaussian-mixture signals require 614, 1502, and 2141, respectively. White noise requires 16375 coefficients. Additional experiments examine trace distance and sparsity versus dimension and accuracy. The stated advantage is conditional on Walsh sparsity or compressibility: unstructured inputs offer little compression, and tighter accuracy generally increases the required sparsity. Thresholding irreversibly discards information. The numerical investigation concerns one-dimensional signals; higher-dimensional data are left for future work.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 13 pages, 5 figures