Periodic Reporting for period 1 - TRICE QFT (TRapped Ion Coherent Execution of Quantum Fourier Transform)
Reporting period: 2015-04-01 to 2017-03-31
An adaptive correction of qubit resonance frequency was implemented in all experimental sequences. It enables to intermittently secure a precise knowledge of qubit addressing frequency while a long experimental sequence is in progress, which allows to minimize detuning of addressing frequency resulting from charging of trap electrodes and fluctuations in ambient magnetic field. This adaptive correction together with all other improvements resulting from various modifications in the experimental setup yields a significant improvement in the fidelities of quantum logic operation, such as fidelity of realizing entanglement between two ions.
Cooling of trapped ions, close to their motional ground state, was investigated for one and two ion systems using a static magnetic field gradient and a long-wavelength radiation in the microwave regime. This method features sideband cooling at a low secular frequency that is practical for trapping both ions and atoms. Both the axial modes of vibration are independently cooled and the thermal excitations of each mode are independently measured by observing one of the ions. Sympathetic cooling of a two ion crystal using microwave radiation was demonstrated for the first time.
The first proof-of-principle experiment showcasing the potential of a large-scale quantum computer in the field of quantum enhanced learning and artificial intelligence was performed. A learning agent can receive perceptual input from and react to the environment (Fig. 1). The learning aspect is facilitated by a reward system that prompts the agent to reinforce connections between the inputs and corresponding actions in its decision making process. The decision-making process of a quantum learning agent, modeled in a system of two ion qubits, was investigated following a novel approach based on projective simulation model for reinforcement learning. Here the agent’s reactions to the perceptual input were investigated by way of a quantum algorithm. The agent’s learning speed was quantified in terms of the average number of interactions with the environment until certain behavior (reactions triggering a reward) can be considered to have been learned. The result demonstrates that the decision-making process of quantum learning agent is quadratically faster compared to that of a classical learning agent. The experimental results are in excellent agreement with theoretically calculated and simulated predictions.
Apart from certain deviations from the originally proposed actions, the main objectives of the project were achieved. Major research results were acquired in the last quarter of the project. They are in the process of being written up, and will be appropriately disseminated within a few months after the project has ended. The project allowed the fellow to acquire various transferable skills in the context of career development. The project also allowed initiating and developing lasting collaborations with researchers based at other EU institutions.