This paper presents a general self-guided algorithm for direct laboratory learning of controls to manipulate quantum-mechanical systems. The primary focus is on an algorithm based on the learning of a linear laboratory input–output map from a sequence of controls, and their observed impact on the quantum-mechanical system. This map is then employed in an iterative fashion, to sequentially home in on the desired objective. The objective may be a target state at a final time, or a continuously weighted observational trajectory. The self-guided aspects of the algorithm are based on implementing a cost functional that only contains laboratory-accessible information. Through choice of the weights in this functional, the algorithm can automatically stay within the bounds of each local linear map and indicate when a new map is necessary for additional iterative improvement. Finally, these concepts can be generalized to include the possibility of employing nonlinear maps, as well as just the laboratory control instrument settings, rather than observation of the control itself. An illustrative simulation of the concepts is presented for the control of a four-level quantum system.
Most quantum systems considered for control by external fields are plagued by a serious lack of complete information about the underlying Hamiltonian. Traditional feedback control techniques are generally not appropriate due to the latter problem, as well as the ultrafast nature of typical quantum dynamics phenomena and the fact that observations of the quantum system will inevitably lead to a disturbance which may often be contradictory to the desired control. In contrast, learning control techniques have a special role to play in the manipulation of quantum dynamics phenomena. The unique capabilities of quantum systems making them amenable to learning control are (a) the ability to have very large numbers of identical systems for submission to control, (b) the high duty cycle of laboratory laser controls, and (c) the ability to observe the impact of trial controls at ultrafast time scales. Various learning algorithms have been proposed to guide this control process. The paper discusses these proposals, as well as some new perspectives.
While closed-loop control of quantum dynamics in the laboratory is proving to be broadly successful, the control mechanisms induced by the fields are often left obscure. Hamiltonian encoding (HE) was originally introduced as a method for understanding mechanisms in quantum dynamics in the context of computational simulations, based on access to the system wavefunction. As a step towards laboratory implementation of HE, this paper addresses the issues raised by the use of observables rather than the wavefunction in HE. The goal of laboratory based HE is to obtain an understanding of control mechanism through a sequence of systematic control experiments, whose collective information can identify the underlying control mechanism defined as the set of significant amplitudes connecting the initial and final states. Mechanism is determined by means of observing the dynamics of special sequences of system Hamiltonians encoded through the control field. The proposed algorithm can handle complex systems, operates with no recourse to dynamical simulations, and functions with limited understanding of the system Hamiltonian. As with the closed-loop control experiments, the HE control mechanism identification algorithm performs a new experiment each time the dynamical outcome from an encoded Hamiltonian is called for. This paper presents the basic HE algorithm in the context of physical systems described by a finite dimensional Hilbert space. The method is simulated with simple models, and the extension to more complex systems is discussed.
In quantum dynamics of many-body systems, to identify the Hamiltonian becomes more difficult very rapidly as the number of degrees of freedom increases. In order to simplify the dynamics and to deduce dynamically relevant Hamiltonian information, it is desirable to control the dynamics to lie within a reduced space. With a judicious choice for the cost functional, the closed loop optimal control experiments can be manipulated efficiently to steer the dynamics to lie within a subspace of the system eigenstates without requiring any prior detailed knowledge about the system Hamiltonian. The procedure is simulated for optimally controlled population transfer experiments in the system of two degrees of freedom. To show the feasibility of steering the dynamics to lie in a specified subspace, the learning algorithms guiding the dynamics are presented along with frequency filtering. The results demonstrate that the optimal control fields derive the system to the desired target state through the desired subspace.
This paper investigates the ability of closed loop quantum learning control techniques to meet a posed physical objective while simultaneously steering the dynamics to lie in a specified subspace. Achievement of successful control with reduced space dynamics can have a number of benefits including a more easily understood control mechanism. Judicious choices for the cost functional may be introduced such that the closed loop optimal control experiments can steer the dynamics to lie within a subspace of the system eigenstates without requiring any prior detailed knowledge about the system Hamiltonian. Learning control with reduced space dynamics takes advantage of the expected existence of a multiplicity of fields that can all give acceptable quality control outcomes. The procedure eliminates the hard demands of following a specific dynamical path by only asking that the dynamics reside in a subspace. Additional measurements characterizing the subspace are necessary to monitor the system evolution during the control field learning process. This procedure is simulated for optimally controlled population transfer experiments in systems of one and two degrees of freedom. The results demonstrate that optimal control fields can be found that successfully derive the system to the target state while staying within the desired subspace.
A quantum control landscape is defined by the value of a physical observable as a functional of the time-dependent control field
E(t) for a given quantum-mechanical system. Level sets through this landscape are prescribed by a particular value of the target observable at the final dynamical time
T, regardless of the intervening dynamics. We present a technique for exploring a landscape level set, where a scalar variable
s is introduced to characterize trajectories along these level sets. The control fields
E(s,t) accomplishing this exploration (i.e., that produce the same value of the target observable for a given system) are determined by solving a differential equation over
s in conjunction with the time-dependent Schr\"odinger equation. There is full freedom to traverse a level set, and a particular trajectory is realized by making an a priori choice for a continuous function
f(s,t) that appears in the differential equation for the control field. The continuous function
f(s,t) can assume an arbitrary form, and thus a level set generally contains a family of controls, where each control takes the quantum system to the same final target value, but produces a distinct control mechanism. In addition, although the observable value remains invariant over the level set, other dynamical properties (e.g., the degree of robustness to control noise) are not specifically preserved and can vary greatly. Examples are presented to illustrate the continuous nature of level-set controls and their associated induced dynamical features, including continuously morphing mechanisms for population control in model quantum systems.
This paper presents a novel approach to dynamic quantum control utilizing feedback learning algorithms. The core idea is to develop a system capable of real-time adjustment of control parameters based on feedback information obtained from the quantum system itself. This approach addresses the inherent challenges in precisely controlling quantum systems, which are often highly sensitive to environmental noise and difficult to model accurately. The proposed framework leverages the principles of quantum control and feedback learning to achieve optimal control performance. The system architecture incorporates a closed-loop control mechanism where the system's output is continuously monitored, and the control parameters are adjusted accordingly. Mathematical formulations detailing the control process, feedback mechanisms, and optimization strategies are provided. This work contributes to advancements in quantum control by offering a robust and adaptable method for manipulating quantum states, with potential applications in quantum computing, quantum sensing, and quantum communication. The algorithm is designed to minimize the influence of external disturbances and maximize the accuracy of state preparation and manipulation. The presented approach provides a pathway towards more reliable and efficient quantum control strategies.