(英) |
In recent years, noisy intermediate-scale quantum computer (NISQ) has appeared, and it is expected that quantum algorithms running on NISQ will be applied to quantum chemistry and materials development.
In NISQ research, since the actual quantum computer is still developing, it is important to evaluate the performance by the simulation with classical computer. In the simulations with conventional computers, computation time and memory consumption increase exponentially with the number of qubits, so it is important for NISQ research to execute it faster with parallel computers.
In the simulation with the classical computer, since computation time and the amount of memory are
exponentially increased with the number of qubits, it is important to execute such numerical simulations fast with parallel computer.
In this study, we propose how to accelerate a simulation of quantum dynamics using a GPU.
Extending the quantum circuit simulator 'qulacs', we simulate the quantum circuit obtained by applying the Trotter-Suzuki expansion to the quantum dynamics.
It is successful that the execution time of the simulation is greatly improved by parallelizing operations that can be diagonalized.
When we apply this scheme to a time dependent dynamics of fully-connected transverse Ising model, it is confirmed that the GPU model is 100 times faster than the CPU one.
This method can also be applied to the general Hamiltonian, and is useful for evaluating quantum algorithms such as quantum annealing and quantum chemistry. |