This toolbox includes the Matlab implementations of the GPIS and Acc-GPIS algorithms for efficiently solving the l_1 constrained least-squares regression and nuclear-norm constrained multivariate regression tasks, proposed in the ICML 2017 paper "Gradient Projection Iterative Sketch for Large-Scale Constrained Least-Squares", as well as the Rest-Katyusha and Adaptive Rest-Katyusha algorithms for the Lasso and elastic-net regularized least-squares regression tasks, proposed in the NeurIPS 2018 paper "Rest-Katyusha: Exploiting the Solution's Structure via Scheduled Restart Schemes".
Tang, Junqi. (2019). Structure-Adaptive Large-Scale Convex Optimization Toolbox v1.0, [software]. University of Edinburgh. Institute for Digital Communications. https://doi.org/10.7488/ds/2489.