Jakub Pawłowski, Tomasz Śmierzchalski, Fengping Jin, Bartłomiej Gardas, Sebastian Deffner, Kristel Michielsen, Zakaria Mzaouali
This repository contains code used to produce results reported in "Quantum annealers as programmable thermal machines". We investigate quantum annealers as programmable thermal machines. Building on experiments that identified reverse-annealing on D-Wave hardware as a bona fide thermal accelerator and showed how fluctuation relations and thermodynamic uncertainty relations (TURs) turn limited energy readouts into rigorous bounds on heat, work, and dissipation, we ask whether other operating modes are allowed, realizable, and how far hardware-level performance can be certified from fluctuations alone. First, we demonstrate that by tailoring the reverse-annealing schedule and the initialization, the same device can emulate the majority of thermal machine cycles: accelerator, engine, and refrigerator. We diagnose each regime directly from the sign structure of average processor and environment energy changes and of the driving work—an operational thermodynamic classification that requires no detailed bath modeling. Second, we leverage TURs to convert fluctuations of the chip’s reported energy change into lower bounds on entropy production and bounds on per-cycle heat and work, which in turn yield hardware-certified bounds on power for any chosen schedule. Crucially, these bounds rely only on energy statistics accessible on today’s machines. Together, these results establish quantum annealers as flexible simulators of thermal machines and provide a practical methodology to benchmark their energy efficiency alongside computational performance.
This work was supported by:
- Jülich Supercomputing Centre through the Jülich UNified Infrastructure for Quantum computing (JUNIQ)
- The National Science Center (NCN), Poland, under Project No. 2020/38/E/ST3/00269
- Competence Center Quantum Computing Baden-Württemberg under Project KQCBW25