Exploring Fundamental Limits of Reversible Computing Technologies from Nonequilibrium Quantum Thermodynamics

Landauer and Bennett famously argued that traditional non-reversible computational architectures suffer from a fundamental minimum energy dissipation (and entropy generation) that is required to carry out ordinary logically irreversible computational operations, but that alternative reversible computational architectures can circumvent this limit.

Over the years, questions have been raised regarding whether these observations remain valid when treated in a rigorous non-equilibrium thermodynamic framework.  In recent work, we found that these classic statements do indeed remain valid for practical architectures when the role of correlations is properly taken into account.  In particular, we have found that Müller’s generalized framework of catalytic thermal operations provides a rigorous basis for these statements in a non-equilibrium context.

However, an important question remains regarding what fundamental limits on entropy generation can be shown to apply even to reversible computations.  We conjecture that technology-independent limits on entropy generation in reversible computations (classical and quantum) can be formulated as a function of a number of relevant physical timescales, and outline our research plan for deriving these limits.  As a first step, we discuss how to represent classical reversible operations in terms of a Lindbladian superoperator dynamics in a quantum Markovian framework.