We exploit the widening margin in tensor-core performance between [FP64/FP32/FP16/INT8,FP64/FP32/FP16/FP8/INT8] on NVIDIA [Ampere,Hopper] GPUs to boost the performance of output accuracy-preserving mixed-precision computation of Genome-Wide Association Studies (GWAS) of 305K patients from the UK BioBank, the largest-ever GWAS cohort studied for genetic epistasis using a multivariate approach. Tile-centric adaptive-precision linear algebraic techniques motivated by reducing data motion gain enhanced significance with low-precision GPU arithmetic. At the core of Kernel Ridge Regression (KRR) techniques for GWAS lie compute-bound cubic-complexity matrix operations that inhibit scaling to aspirational dimensions of the population, genotypes, and phenotypes. We accelerate KRR matrix generation by redesigning the computation for Euclidean distances to engage INT8 tensor cores while exploiting symmetry. We accelerate solution of the regularized KRR systems by deploying a new four-precision Cholesky-based solver, which, at 1.805 mixed-precision ExaOp/s on a nearly full Alps system, outperforms the state-of-the-art CPU-only REGENIE GWAS software by five orders of magnitude.

We present a generic multi-agent deep reinforcement learning framework for dynamic multi-domain service provisioning in large-scale networks. We formulate both the assignment of a given sub-VNF-FG to a particular domain and its placement within the assigned local domain as a two-stage graph matching problem. To this purpose, we leverage graph attention networks in combination with hierarchical deep reinforcement learning. The learning process is additionally bootstrapped through self-supervised pre-training for both domain assignment and placement stages. The initial policies are further fine-tuned by the different agents along the learning process to address the evolving states of the domains. We show that our approach provides competitive real-time VNF-FG embedding results while achieving load balancing across the domains without compromising their privacy and autonomy in addition to satisfying QoS constraints. Our intelligent framework also paves the way for novel approaches that can benefit from the inherent graph structure of the problem.

Model selection is an essential task for many applications in scientific discovery. The most common approaches rely on univariate linear measures of association between each feature and the outcome. Such classical selection procedures fail to take into account nonlinear effects and interactions between features. Kernel-based selection procedures have been proposed as a solution. However, current strategies for kernel selection fail to measure the significance of a joint model constructed through the combination of the basis kernels. In the present work, we exploit recent advances in post-selection inference to propose a valid statistical test for the association of a joint model of the selected kernels with the outcome. The kernels are selected via a step-wise procedure which we model as a succession of quadratic constraints in the outcome variable.