Pavan Hosur

In the current Noisy Intermediate-Scale Quantum (NISQ) era, noise is widely viewed as a primary obstacle to fault-tolerant quantum computation. In this work, we flip this perspective by demonstrating that noise can be harnessed constructively in key quantum algorithms. First, drawing on the Eigenstate Thermalization Hypothesis, we show that using both classical and quantum simulations that noise generically accelerates Gibbs state preparation. In our model, we find that phase-flip noise, and more generally Haar-random noise, drives localized thermalization that yields speedups of up to a factor of ∼3.5× as compared to noiseless systems. We also show that this induces thermalization in systems that would otherwise fail to do so. Building on this, we further show that intrinsic noise can be selectively leveraged to emulate nonunitary dynamics, enabling efficient simulation of open quantum systems without encoding dissipation into unitary circuits. By preserving physical noise rather than correcting it entirely, we show that this approach reduces qubit and gate overhead and relaxes fidelity requirements on quantum hardware. Taken together, these results establish a paradigm in which noise is harnessed as a computational resource, enabling practical quantum advantages on near-term and future quantum devices.