Quantum optimal control of multilevel dissipative quantum systems with reinforcement learning

Model-Free Quantum Control with Reinforcement Learning

“…In recent years, multiple theoretical proposals have emerged around applying reinforcement learning to quantum control problems such as quantum state preparation [19][20][21][22][23][24][25][26][27] and feedback stabilization [28], the construction of quantum gates [29][30][31], design of quantum error correction protocols [32][33][34][35], and control-enhanced quantum sensing [36,37]. However, these proposals are focused on recasting the problem in a way that would avoid facing quantum observability.…”

“…This is possible only in simulated environments, for example by providing the learning agent with full knowledge of the system's wavefunction, which supplies enough information for decision making [19, 22-24, 26, 28, 32, 36, 37]. Moreover, since in the simulation the distance to the target state or operation is known at every step of the quantum trajectory, it can be used to construct a steady reward signal to guide the learning algorithm [22][23][24]36], thereby alleviating the well-known delayed reward assignment problem. Taking RL a step closer towards quantum observability, some works do not give the agent access to the wave-function, but still use it for calculation of fidelities and expectation values in different parts of the training pipeline [20,25,27,48,49], which would require a prohibitive amount of averaging in experiment.…”

“…We propose a modular circuit-based approach for training a reinforcement learning agent for continuous quantum control tasks in a completely model-free way, thereby adapting quantum control policies to the specific system in which they are deployed. Given a continuously parameterized control circuit, the agent learns its parameters through trial-end-error interaction with the controlled quantum system, without any human-provided knowledge of the system's wave-function or model of the system's dynamics, unlike in related works [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] which we survey in Section II.…”

Deep Reinforcement Learning for Quantum State Preparation with Weak Nonlinear Measurements

“…During the learning phase, the agent discovers from scratch novel strategies, in a systematic procedure which at first resembles trial and error but later begins to build on insights acquired earlier. That RL in general and deep RL in particular is a powerful approach in quantum physics has by now been demonstrated in a variety of tasks in different areas: in most works so far, these tasks did not yet require real-time feedback involving decision-making * riccardo.porotti@mpl.mpg.de based on physical measurements, but RL already proved itself to be a versatile tool even in those settings [15][16][17][18][19][20][21][22][23][24][25][26]. These publications are part of a larger drive towards the use of machine learning tools for quantum experiments (e.g.…”

Reinforcement learning-enhanced protocols for coherent population-transfer in three-level quantum systems

“…More recently machine learning techniques have emerged as a viable option for finding alternative optimal control schemes. In particular reinforcement learning (RL) has been employed in the context of state preparation [29,30], circuit architecture design [31] and control of multi-level systems [32]. In the context of three level systems, deep neural network based RL has been used along with state monitoring to learn optimal pulse shapes for driving fields [33,34].…”