Submitted |
|
ABSTRACT:
Accelerated transformed primal--dual (ATPD) methods are proposed for smooth convex optimization problems with affine equality constraints. In the strongly convex regime, ATPD achieves an accelerated linear convergence rate, while in the convex regime, it attains an accelerated sublinear rate. The acceleration mechanism is unified through an exponentially stable ATPD flow at the continuous level, whose discretization yields practical algorithms. The resulting methods match the known lower bounds on first-order oracle complexity in terms of gradient evaluations and matrix--vector products. Numerical experiments confirm the theoretical results and demonstrate the efficiency of the proposed methods.