Stochastic Thermodynamics of Cost of Computing in Autonomous Nanomechanical Systems

This abstract has open access
Summary
Modern computers are powerful devices that have given impetus to our abilities to make discoveries in many fields of science and technology. With their ubiquitous use, the amount of energy they consume is on the rise. Current electronic computers consume energies that are orders of magnitude higher than the theoretical predicted cost, given by the Landauer Limit. The Landauer limit states that the minimum energy required to erase a bit, an operation essential for irreversible computation, at a given temperature, depends on the entropy change that occurs in the erasure. Recent experiments have verified the Landauer Limit using time-dependent potentials. This approach does not take into account the energy required to break the detailed balance and establish an arrow of time for the erasure. In this work, a theoretical nanomechanical system, that does an autonomous bit erasure, is presented and the energy cost for the bit erasure and breaking detailed balance are discussed. The system consists of two coupled nonlinear mechanical oscillators, a bistable buckled beam that acts as the “bit” and a heavy mass, here called the “power clock”, that provides energy and sets the time scale for the erasure. The system is studied using the framework of stochastic thermodynamics. The simulation results show that the system can do a bit erasure using 250 kT of energy (1 kT = 4.1 x 10^-21 J) compared to the 3900 kT of energy used to do a logical operation in modern electronic computers.
Abstract ID :
385
7 visits