Abstract Computation requires energy. Landauer’s principle establishes that erasing one bit of information irreversibly dissipates at least kBT ln 2 of thermal energy. Modern computers operate 10^6 to 10^9 times above this fundamental bound. In 1973, Bennett demonstrated that any computation can be made thermodynamically reversible through procedures that preserve information, thereby approaching the Landauer limit. However, a critical question remained unanswered for fifty years: how quickly does reversible compilation converge to the theoretical minimum, and what ensemble size achieves practical energy savings? We resolve this problem through the Fluctuation-Dissipation Compilation Theorem, which provides the first explicit convergence rate for thermodynamically optimal computing. Reversible compilation of any algorithm dissipates average energy equal to the thermal energy scale multiplied by the logical entropy change, plus corrections that decrease as the inverse square root of the ensemble size. This N^(−1/2) convergence rate, derived by combining fluctuation theorems from statistical mechanics with the central limit theorem, is provably optimal—no compilation strategy can achieve faster convergence. The proof unifies three historically independent disciplines: statistical mechanics through the Jarzynski equality and Crooks fluctuation theorem, information theory through Shannon entropy bounds, and compilation theory through Bennett’s reversible computing framework. By formulating thermodynamic compilation as a rigorous resource theory, we establish that reversible transformations exactly conserve entropy while irreversible erasure operations incur fundamental energy costs bounded by information-theoretic limits. Critically, we provide finite-ensemble performance guarantees essential for practical implementations, characterizing not just average energy dissipation but complete statistical distributions including fluctuations and tail probabilities. Numerical experiments on Grover search and quantum Fourier transforms validate our predictions. Simulations confirm the predicted N^(−1/2) convergence rate within 3% error and demonstrate energy reductions exceeding 99.9% compared to irreversible implementations. These results establish that thermodynamically optimal compilation is achievable with realistic ensemble sizes for near-term quantum hardware operating under thermal constraints. This framework bridges fundamental physics and practical computing. Applications span quantum processors cooled to millikelvin temperatures where every picojoule of dissipation limits performance, battery-powered Internet-of-Things sensors requiring decade-long autonomous operation, data centers consuming one percent of global electricity where incremental efficiency improvements yield billion-dollar savings, and deep-space missions operating on hundred-watt radioisotope generators where every watt conserved enables additional science. By establishing thermodynamic compilation as a rigorous engineering discipline with explicit finite-size guarantees, this work transforms Bennett’s asymptotic theoretical insight into a practical methodology for approaching fundamental physical limits in real computing systems
Abstract Computation requires energy. Landauer’s principle establishes that erasing one bit of information irreversibly dissipates at least kBT ln 2 of thermal energy. Modern computers operate 10^6 to 10^9 times above this fundamental bound. In 1973, Bennett demonstrated that any computation can be made thermodynamically reversible through procedures that preserve information, thereby approaching the Landauer limit. However, a critical question remained unanswered for fifty years: how quickly does reversible compilation converge to the theoretical minimum, and what ensemble size achieves practical energy savings? We resolve this problem through the Fluctuation-Dissipation Compilation Theorem, which provides the first explicit convergence rate for thermodynamically optimal computing. Reversible compilation of any algorithm dissipates average energy equal to the thermal energy scale multiplied by the logical entropy change, plus corrections that decrease as the inverse square root of the ensemble size. This N^(−1/2) convergence rate, derived by combining fluctuation theorems from statistical mechanics with the central limit theorem, is provably optimal—no compilation strategy can achieve faster convergence. The proof unifies three historically independent disciplines: statistical mechanics through the Jarzynski equality and Crooks fluctuation theorem, information theory through Shannon entropy bounds, and compilation theory through Bennett’s reversible computing framework. By formulating thermodynamic compilation as a rigorous resource theory, we establish that reversible transformations exactly conserve entropy while irreversible erasure operations incur fundamental energy costs bounded by information-theoretic limits. Critically, we provide finite-ensemble performance guarantees essential for practical implementations, characterizing not just average energy dissipation but complete statistical distributions including fluctuations and tail probabilities. Numerical experiments on Grover search and quantum Fourier transforms validate our predictions. Simulations confirm the predicted N^(−1/2) convergence rate within 3% error and demonstrate energy reductions exceeding 99.9% compared to irreversible implementations. These results establish that thermodynamically optimal compilation is achievable with realistic ensemble sizes for near-term quantum hardware operating under thermal constraints. This framework bridges fundamental physics and practical computing. Applications span quantum processors cooled to millikelvin temperatures where every picojoule of dissipation limits performance, battery-powered Internet-of-Things sensors requiring decade-long autonomous operation, data centers consuming one percent of global electricity where incremental efficiency improvements yield billion-dollar savings, and deep-space missions operating on hundred-watt radioisotope generators where every watt conserved enables additional science. By establishing thermodynamic compilation as a rigorous engineering discipline with explicit finite-size guarantees, this work transforms Bennett’s asymptotic theoretical insight into a practical methodology for approaching fundamental physical limits in real computing systems
Abstract Whether quantum computational speedups translate to fundamental thermodynamic energy savings remains unresolved. We establish the first rigorous connection between computational complexity and thermodynamic costs through DEMON(I,E) complexity classes that classify information-processing protocols by information complexity I(n) and erasure energy E(n). Four theorems prove: (1) any g-bit erasure dissipates >= g X k_B T ln 2, generalizing Landauer's principle; (2) measurement disturbance D reduces extractable work to (I-D)k_B T ln 2; (3) Grover-based quantum demons achieve provably optimal Theta(sqrt{N}) thermodynamic advantage over classical search, tight by the BBBV lower bound; and (4) quantum walks on d-dimensional lattices exhibit a dense hierarchy with continuous scaling exponent (d-1)/(2d). Computational validation across six orders of magnitude confirms predictions with <1% error. Experimental predictions include 20.5X energy reduction for trapped-ion Grover search (N=1024, T=1 mK) and 11X effective cooling for superconducting circuit QED heat-bath protocols. This work unifies 158 years of Maxwell demon research, connects algorithmic complexity to physical energy costs, and enables experimental verification of quantum thermodynamic advantages beyond computational speedup.
Quantum state engineering and quantum computation rely on information erasure procedures that, up to some fidelity, prepare a quantum object in a pure state. Such processes occur within Landauer's framework if they rely on an interaction between the object and a thermal reservoir. Landauer's principle dictates that this must dissipate a minimum quantity of heat, proportional to the entropy reduction that is incurred by the object, to the thermal reservoir. However, this lower bound is only reachable for some specific physical situations, and it is not necessarily achievable for any given reservoir. The main task of our work can be stated as the minimisation of heat dissipation given probabilistic information erasure, i.e., minimising the amount of energy transferred to the thermal reservoir as heat if we require that the probability of preparing the object in a specific pure state | φ 1 ⟩ be no smaller than p φ 1 max − δ . Here p φ 1 max is the maximum probability of information erasure that is permissible by the physical context, and δ ⩾ 0 the error. To determine the achievable minimal heat dissipation of quantum information erasure within a given physical context, we explicitly optimise over all possible unitary operators that act on the composite system of object and reservoir. Specifically, we characterise the equivalence class of such optimal unitary operators, using tools from majorisation theory, when we are restricted to finite-dimensional Hilbert spaces. Furthermore, we discuss how pure state preparation processes could be achieved with a smaller heat cost than Landauer's limit, by operating outside of Landauer's framework.
The physical nature of computation imposes fundamental limits on the energy efficiency of information processing, as formalized by Landauer's principle, which establishes a minimum thermodynamic cost for information erasure. This dissertation investigates near-Landauer-limit computation using superconducting Quantum Flux Parametrons (QFPs). We first establish the theoretical foundations of physical computation, including logical and thermodynamic reversibility, stochastic thermodynamics, and the relationship between information, entropy, and energy dissipation. Building upon these principles, we develop and analyze computational protocols based on QFPs and Coupled QFPs (CQFPs). Numerical simulations demonstrate thermodynamically efficient bit erasure, bit-flip operations, and a superconducting realization of the Szilard Engine. We further demonstrate universal combinational logic through a NAND gate, as well as signal fan-out through a copier gate and memory through a D-latch, providing the essential primitives required for scalable digital computation. A central result of this work is the demonstration that momentum, conventionally regarded as a source of error in superconducting logic, can instead be exploited as a computational resource. By operating QFP circuits in the underdamped regime, momentum-assisted protocols achieve simultaneous improvements in speed, fidelity, and energy efficiency. These results establish QFP-based superconducting circuits as a promising platform for near-Landauer-limit computation and provide a foundation for future momentum-assisted superconducting dynamical computing architectures.
Quantum error correction protects logical qubits from decoherence but requires repeated syndrome extraction whose fundamental thermodynamic cost has not been characterized exactly. We prove five theorems and two corollaries that supply this characterization for stabilizer codes. First, a no-go result establishes that every error-correction cycle requires strictly positive work, ruling out passive error correction. Second, a duality identity shows that the combined measurement and erasure cost per syndrome-extraction round is determined solely by the number of stabilizer generators; only the split between measurement and erasure depends on the error rate. Third, a coherence dissipation theorem quantifies the additional irreversible work generated when projective stabilizer measurements destroy quantum coherence in the data register. CSS codes eliminate this cost for generators aligned with the noise channel’s eigenbasis, yielding a thermodynamic criterion for code selection under non-Pauli noise. Fourth, a Crooks fluctuation theorem for the full syndrome cycle yields a Jarzynski equality and exponential tail bounds on rare work fluctuations. Fifth, a variance formula captures the effect of syndrome correlations among stabilizers that share physical qubits. We evaluate all results across eleven codes, from a three-qubit repetition code to a 144-qubit quantum low-density parity-check code, establishing a Pareto trade-off between thermodynamic cost and error-suppression capability. Comparison with four experimental platforms—two superconducting, one trapped-ion, and one neutral-atom—shows that current hardware operates five to eleven orders of magnitude above the Landauer limit, with control and readout hardware, not syndrome extraction, as the dominant source of dissipation.
Autonomous quantum thermal machines have recently been proposed as physics-based computing substrates where logical inputs and outputs are encoded in temperature gradients. While such ``thermodynamic neurons'' exhibit a clear trade-off between computational fidelity and heat dissipation, the fundamental information-theoretic limits of temperature-encoded computation remain uncharacterized. Here, we derive rigorous bounds linking average error probability, channel capacity, and entropy production for finite-capacity thermal reservoirs operating far from equilibrium. We prove that the minimal dissipation required to achieve a target average error probability ⟨ξ⟩ diverges as ⟨ξ⟩ approaches a fundamental minimum error floor $\epsMin$ imposed by finite-reservoir thermal fluctuations. We further establish a thermodynamic channel capacity that saturates at high dissipation, and quantify the minimal dissipation required for cascaded networks to maintain target fidelity. We demonstrate that the required dissipation grows linearly O(L) with network depth for shallow networks, but diverges as the network depth L approaches a fundamental maximum limit $L_{\max} = \epsNet/\epsMin$ imposed by the minimum error floor. Furthermore, under strong noise amplification conditions, the required dissipation can grow up to O(L3). Our framework bridges stochastic thermodynamics, finite-time information theory, and autonomous computation, providing rigorous design principles for energy-efficient analog thermodynamic hardware.
The emergence of irreversibility in physical processes, despite the fundamentally reversible nature of quantum mechanics, remains an open question in physics. This thesis explores the intricate relationship between quantum mechanics and thermodynamics, with a particular focus on minimizing entropy production in finite-time processes. By employing tools from quantum information theory and geometric thermodynamics, we tackle the challenge of deriving irreversible thermodynamic behavior from the reversible microscopic framework of quantum mechanics. We begin with a comprehensive review of the laws of thermodynamics, setting the stage for the subsequent analyses. We introduce novel developments in quantum thermodynamics through a generalized framework for geometric thermodynamics, which enables the derivation of finite-time corrections beyond the Markovian regime. Building on this foundation, we extend Landauer's principle by incorporating a finite-time correction that highlights the necessity of strong coupling for optimal information erasure processes. This result underscores the emergence of Planckian time as a fundamental speed limit to thermalization. Additionally, we explore how collective effects can be harnessed to reduce energy dissipation in thermodynamic operations, revealing that classical correlations between systems can significantly mitigate dissipation, though this may pose new questions regarding the third law of thermodynamics. Finally, we optimize thermodynamic processes in mesoscopic systems, including quantum dot engines and information engines. These findings not only enhance our understanding of the fundamental limits of irreversibility but also open new avenues for research. Future works will focus on fully exploiting collective effects, aligning these with the third law of thermodynamics, and understanding the thermodynamic consistency of master equations.