paper / tqchenml / Jun 10
This review details advancements in quantum hardware for simulating condensed matter physics, covering phenomena like strong correlation, topology, and non-equilibrium dynamics. It primarily examines gate-based digital quantum simulations while noting the role of analog experiments. The survey encompasses various hardware platforms and highlights how current noisy simulations are vital for developing techniques for future fault-tolerant quantum simulations.
quantum-computingcondensed-matterphysics-simulationdigital-quantum-simulationnoisy-quantum-simulations
“Quantum hardware is increasingly capable of simulating complex condensed matter phenomena such as strongly-correlated, topological, and non-equilibrium physics.”
paper / tqchenml / Jun 10
This paper introduces a training-free acceleration framework for 3D diffusion models applied to low-count PET denoising. The authors propose a global skipping strategy that initializes reverse diffusion from an intermediate step via a noise-consistent transformation, plus a local feature reuse shortcut that reuses slowly-varying U-Net features across neighboring steps. The method achieves over 10x acceleration with improved or comparable reconstruction quality across multiple PET tracers (18F-FDG, 68Ga-DOTATATE, 18F-PSMA), and blinded reader studies confirm enhanced clinical confidence and perceived diagnostic quality.
diffusion-modelspet-denoisingmodel-accelerationmedical-imagingcomputer-vision
“The proposed Global-Local Skipping Strategy achieves over an order of magnitude acceleration in 3D PET diffusion model inference.”
paper / tqchenml / Jun 3 / failed
paper / tqchenml / May 12 / failed
paper / tqchenml / May 12 / failed