Speaker
Description
Numerically simulating coherent spin systems, such as atoms and defects, especially under highly time-dependent potentials, is incredibly resource-intensive. Nevertheless, such modelling can be greatly helpful for developing new quantum technology. We present a numerical simulator to accelerate the solving of the time-dependent Lindblad equation, where multiple time steps are calculated in simultaneously on GPU.
We further optimised the simulator by looking at the geometry of how the quantum states and operators were being represented, as to remove as many redundancies as possible. Reducing the size of the problem like this frees memory and reduces the number of operations needed for matrix arithmetic. While researching this, we discovered a connection with quantum control theory: whether or not the system has the property of being accessible/weakly controllable, directly corresponds to how we can compress its representation when simulated on a GPU. This in-turn corresponds to the Lie algebra of the system's generators and its reachable sets.
The numerical simulator is open-source and freely available as the python package superspinsim. It is almost two orders of magnitude faster than existing simulators from qutip, quantumoptics.jl and quantumtoolbox.jl for highly time-dependent problems. It is fast enough to simultaneously simulate the spin and orbital dynamics of a nitrogen-vacancy centre in diamond without needing to use the rotating wave approximation.
| I am the presenting author | Yes |
|---|