Theoretical Study On Optimizing The Pulse Of Cold Atom Interferometer Using Quantum Optimal Control

As a high-precision measurement tool,the cold atom interferometer has been rapidly developed in recent years.It is widely used in the fields of gravity measurement,fine structure constant measurement,gravitational wave detection,dark energy research and basic physical law inspection.The cold atom interferometer relies on the interaction of laser pulses with atoms to achieve beam splitting,reflection and beam combining operations.However,due to the inhomogeneity of the spatial distribution of the laser electric field intensity,the speed broadening of the atomic group,and the coupling of other effects in the environment,the cold atom interferometer cannot achieve complete Rabi oscillation.These factors limit the pulse efficiency,which in turn limits the fringe contrast and sensitivity of the interferometer.Optimizing the control pulse of the cold atom interferometer,including anti-noise,shortening time consumption,etc.,can improve the interference measurement process.As an important means to improve the control performance of quantum systems,quantum optimal control is widely used in many fields such as state preparation,logic gate control,and quantum information processing.In this work,we use quantum optimal control to carry out the pulse optimization of the cold atom interferometer,and complete the numerical calculation of the pulse optimization.This work uses quantum optimal control to optimize the laser pulse sequence,achieving the goals of improving pulse efficiency and shortening the control time.The main contents of this work include:(1)Compile a quantum optimal control algorithm based on deep reinforcement learning;(2)For the far detuned Raman pulse,the deep reinforcement learning algorithm is applied to optimize its beam splitting and reflection operations,which improves the robustness of the beam splitting and reflection operations to ununiform laser intensity spatial distribution,atomic velocity broadening and other undesirable factors,and improves pulse efficiency;(3)For the near-resonant Raman pulse,the deep reinforcement learning algorithm is applied to optimize its beam splitting operation,which achieves the optimization result of shortening the beam splitting time by two orders of magnitude without reducing the fidelity of the beam splitting state.The above-mentioned work has laid a certain foundation for improving the efficiency and stability of the measurement by optimizing the laser pulse in the subsequent cold atom interference experiment.