Search papers, labs, and topics across Lattice.
The paper introduces PDHCG-CQP, a GPU-accelerated first-order solver designed for large-scale conic convex quadratic programming, which leverages a restarted averaged primal-dual hybrid gradient method. It achieves local linear convergence under strict complementarity by utilizing a conic quadratic proximal subproblem solved via projected gradient iterations, ensuring robust performance across various problem types. Extensive experiments demonstrate that PDHCG-CQP outperforms existing first-order solvers, efficiently scaling to 8 GPUs and handling instances with up to 4.4 billion stored primal coordinates.
Achieving state-of-the-art robustness in first-order solvers, PDHCG-CQP scales effectively to massive problem sizes with multi-GPU support.
We present PDHCG-CQP, a GPU-accelerated first-order solver for large-scale conic convex quadratic programming. PDHCG-CQP supports affine constraints and Cartesian products of nonnegative, second-order, rotated second-order, exponential, and three-dimensional power cones. At its core is a restarted averaged primal-dual hybrid gradient (PDHG) method, whose primal update is computed inexactly by solving a conic quadratic proximal subproblem with projected gradient iterations. We establish local linear convergence of the restarted averaged scheme with both exact and inexact primal proximal evaluations under a uniform local quadratic-growth condition on the smoothed primal-dual gap. We further show that this condition holds under strict complementarity by exploiting a rotated second-order-cone lifting together with local primal and dual regularity conditions. Our C/CUDA implementation combines matrix-free linear algebra, batched cone projections, adaptive inner solves, reflected-Halpern acceleration, and fully device-resident KKT residual computations. It also supports multi-GPU execution through a two-dimensional partitioning of the problem data. Extensive experiments on standard and large-scale quadratic programming (QP), convex quadratically constrained quadratic programming (QCQP), second-order cone programming (SOCP), and quasilinear Fisher equilibrium benchmarks demonstrate that PDHCG-CQP achieves state-of-the-art robustness among first-order solvers while scaling efficiently to 8 GPUs and instances with up to $4.4\times10^8$ stored primal coordinates. PDHCG-CQP is open source and available at https://github.com/Lhongpei/PDHCG.
Stanford HAI