Speaker
Description
Preparing thermal and ground states of many-body Hamiltonians is a central task for quantum simulation. Algorithms with provable guarantees target fault-tolerant devices, while variational methods such as variational quantum imaginary time evolution (VarQITE) suit near-term hardware but require measuring a number of expectation values that grows with the ansatz size, with shot costs amplified by barren plateaus. We introduce PP-VarQITE, a method that combines VarQITE with Pauli propagation to classically compile shallow variational circuits approximating ground and Gibbs states. The metric matrix and evolution vector of the VarQITE update are evaluated classically with Pauli propagation, removing the iterative quantum optimization and its shot overhead. A selection rule derived from the evolution vector identifies the Pauli rotations that receive non-zero updates at initialization, yielding a Hamiltonian-guided ansatz that remains shallow and executable on current hardware. We benchmark PP-VarQITE on the 1D transverse-field Ising model, the J1-J2 Heisenberg chain, and the 2D Ising model, for up to 100 qubits (50 system qubits). The prepared states reproduce energy densities and two-point correlations of matrix product state (MPS) references, and the accuracy is retained at high inverse temperature, where direct operator-basis Pauli propagation has been reported to lose accuracy beyond 30 qubits. For the 50-qubit transverse-field Ising model, PP-VarQITE reaches comparable accuracy at 940 CX gates and depth 88, against 1188 CX gates and depth 736 for the MPS-compiled circuit.
| Topic | Quantum variational circuits and their applications |
|---|---|
| Additional Topic | Hybrid quantum-classical approaches for learning and optimization |