Learning Hamiltonians for $O(1)$ Oracle-Query Quantum State Preparation

TL;DR

This paper introduces a Hamiltonian-based quantum state preparation method that efficiently encodes classical data into quantum states with minimal quantum queries, leveraging classical training and structured Hamiltonians for near-term quantum hardware.

Contribution

The paper presents a novel Hamiltonian learning approach combined with shallow quantum circuits for efficient data encoding, reducing quantum query complexity to $O(1)$ for certain datasets.

Findings

Achieves $O(1)$ quantum queries for data encoding.

Maintains low infidelity (~10^{-5}) with polynomial resources.

Provides hardware-efficient circuits suitable for near-term devices.

Abstract

We propose a Hamiltonian-based quantum state preparation method implemented via a shallow parametrized quantum circuit. The approach learns the parameters of a diagonal Hamiltonian through a classical training phase, while the quantum circuit itself performs only fixed-depth Hamiltonian evolution and mixing operations. With oracle access to the learned Hamiltonian parameters, $N$ classical data values can be encoded into $n=\log_2{N}$ qubits using $O(1)$ quantum queries, shifting the overall computational cost to an $O(N\log{N})$ classical preprocessing stage. For structured datasets generated by an underlying function, oracle access can be avoided by expressing the Hamiltonian in the Walsh basis and retaining only a polynomial number of significant terms. In this regime, quantum state preparation is achieved in $\text{poly}(n)$ time using $\text{poly}(n)$ parameters, reaching infidelities on the order of $10^{-5}$. By restricting the Hamiltonian to one-local and two-local terms, the method naturally yields hardware-efficient circuits suitable for near-term quantum devices.