To help address our exponentially-growing information processing needs, the dynamical landscape computing paradigm offers a promising platform for investigating the fundamental physics of computing and the thermodynamic performance of logic gate designs. Here, computations are performed by controlling the distinguishable minima of a potential energy landscape that is coupled to the thermal environment. The minima serve as the system’s mesoscopic memory states, while information is represented by microstate distributions in the landscape. Dynamically manipulating the memory states then corresponds to information processing. Given a candidate substrate that generates a suitably controllable landscape, it is unclear how thermodynamically-optimized parameter choices translate to electrical engineering simulations under realistic operating conditions. A common substrate that is effectively modeled with Langevin dynamics and Simulation Program with Integrated Circuit Emphasis (SPICE) simulators is the superconducting quantum interference device (SQUID). Constructed from Josephson junctions, SQUIDs can operate at GHz frequencies and at the k_B T thermal energy scale. To bridge the gap between physics and engineering, we assess whether stochastic thermodynamic predictions agree with Josephson SIMulator (JoSIM), a superconducting SPICE program, for two 1-bit computations: an information erasure that takes place in metastable equilibrium, and a reversible momentum-based NOT gate that occurs far from equilibrium. For each protocol, we find agreement in the mean work cost associated with performing the computation, and its corresponding dynamics. Establishing this connection provides opportunities to build energy-efficient superconducting computers that perform thermodynamically-optimal computations.

To be announced