Accelerating Staple Networks for Normalizing Flows in Lattice Quantum Chromodynamics Colfax Research reported that it used CuTe DSL kernels to accelerate the staple network, the core component of its normalizing-flow approach to speeding up Monte Carlo sampling in lattice quantum chromodynamics simulations. The work targets the high computational cost of running flow models at a useful resolution, a bottleneck the company says limits the practical payoff of machine-learning acceleration for lattice QCD. The post frames the kernel work as a step toward making normalizing flows viable at production-scale lattice resolutions. Machine learning offers a promising route to accelerating lattice quantum chromodynamics QCD simulations. In particular, normalizing flows can speed up Monte Carlo sampling in lattice QCD 1, 2 also see 3 for an introductory tutorial , but there is a high computational cost to running these flow models at a useful resolution. This post describes our use of CuTe DSL kernels to accelerate the staple network, a core component of our approach to implementing these flows. The post Accelerating Staple Networks for Normalizing Flows in Lattice Quantum Chromodynamics https://research.colfax-intl.com/accelerating-staple-networks-for-normalizing-flows-in-lattice-quantum-chromodynamics/ appeared first on Colfax Research https://research.colfax-intl.com .