Acceleration of Variational Quantum Eigen Solver

Hybrid algorithm, Variational Quantum Eigen (VQE) solver plays key role in drug discovery, material science, chemical engineering, portfolio optimization, artificial intelligence etc. The accelerated version of VQE, implemented using NVIDIA cuQuantum SDK and executed on NVIDIA DGX-A100 system. The VQE accelerated algorithm performance compared with its CPU only implementation, for 25, 28, 30 and 34 qubits respectively. The run time increases multifold, with the increase of qubits, when executed on CPU, however for GPU enabled run, the execution time increases marginally. The relative speed up between CPU only and GPU enabled run, for one stable run, increases from 37× (25 qubits) to 951× (34 qubits). In practice, VQE is executed, number of times, until the minimum expectation value of Hamiltonian is obtained, by varying the ansaz parameters. The speed up between, CPU only and GPU enabled run, becomes significant with the increase in the iterations of VQE execution.