Chemistry & Biochemistry Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/2752
Browse
2 results
Search Results
Item NONEQUILIBRIUM STATISTICAL PHYSICS OF FEEDBACK-CONTROLLED AND AUTONOMOUS INFORMATION-THERMODYNAMIC SYSTEMS(2024) Bhattacharyya, Debankur; Jarzynski, Christopher; Chemical Physics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This thesis investigates the nonequilibrium dynamics of a variety of systems evolving under control protocols. A control protocol can involve feedback based on measurements performed by an external agent, or it can be a predefined protocol that does not rely on explicit measurements of the system’s state. In the context of information thermodynamics, the former setup belongs to the paradigm of non-autonomous or feedback-controlled Maxwell's demons, and the latter to the paradigm of autonomous demons. The thesis begins with a framework for analyzing non-autonomous feedback control, when the control protocol is applied by an agent making continuous measurements on the system. A multiple-timescales perturbation theory, applicable when there exists an appropriate separation of timescales, is developed. This framework is applied to a classical two-state toy model of an information engine – a device that uses feedback control of thermal fluctuations to convert heat into work. Additionally, quantum trajectory simulations are used to study a feedback-controlled model of Maxwell's demon in a double quantum dot system. Next, a modeling scheme for converting feedback-controlled Maxwell's demons to autonomous (non-feedback) systems is developed. This scheme explicitly accounts for the thermodynamic costs of information processing, by incorporating an information reservoir, modeled as a sequence of bits. This modeling scheme is then applied for converting the classical analogue of the non-autonomous double quantum dot Maxwell's demon, discussed previously, to an autonomous model. Using analytical, semi-analytical and stochastic simulation-based approaches, it is shown that the obtained model can act either as an information engine, or as a “Landauer eraser”, and then the phase diagrams that identify these regimes of behavior are constructed. Finally, fast-forward shortcuts to adiabaticity for classical Floquet-Hamiltonian systems is developed, and applied to a periodically driven asymmetric double well (without feedback control). Tools from dynamical systems theory are then used to characterize the system’s angle-variable dynamics.Item Topics in equilibrium and nonequilibrium thermodynamics: computing crystalline free energies and engineering Maxwell’s demon.(2015) Lu, Zhiyue; Jarzynski, Christopher; Chemical Physics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation covers two separate topics in statistical physics. The first part of the dissertation focuses on computational methods of obtaining the free energies (or partition functions) of crystalline solids. We describe a method to compute the Helmholtz free energy of a crystalline solid by direct evaluation of the partition function. In the many-dimensional conformation space of all possible arrangements of N particles inside a periodic box, the energy landscape consists of localized islands corresponding to different solid phases. Calculating the partition function for a specific phase involves integrating over the corresponding island. Introducing a natural order parameter that quantifies the net displacement of particles from lattices sites, we write the partition function in terms of a one-dimensional integral along the order parameter, and evaluate this integral using umbrella sampling. We validate the method by computing free energies of both face-centered cubic (FCC) and hexagonal close-packed (HCP) hard sphere crystals with a precision of $10^{-5}k_BT$ per particle. In developing the numerical method, we find several scaling properties of crystalline solids in the thermodynamic limit. Using these scaling properties, we derive an explicit asymptotic formula for the free energy per particle in the thermodynamic limit. In addition, we describe several changes of coordinates that can be used to separate internal degrees of freedom from external, translational degrees of freedom. The second part of the dissertation focuses on engineering idealized physical devices that work as Maxwell's demon. We describe two autonomous mechanical devices that extract energy from a single heat bath and convert it into work, while writing information onto memory registers. Additionally, both devices can operate as Landauer's eraser, namely they can erase information from a memory register, while energy is dissipated into the heat bath. The phase diagrams and the efficiencies of the two models are solved and analyzed. These two models provide concrete physical illustrations of the thermodynamic consequences of information processing.