Core Concepts
This work proposes variational quantum algorithms for approximately solving semidefinite programming problems, including both general and standard forms. The algorithms aim to converge to approximate stationary points of the optimization problems.
Abstract
The key highlights and insights of this content are:
The authors consider three different forms of semidefinite programming (SDP) problems: the general form (GF), the equality constrained standard form (ECSF), and the inequality constrained standard form (ICSF).
For each form, the authors reformulate the constrained optimization problems into unconstrained forms by employing a series of reductions. This allows them to express the final unconstrained forms in terms of expectation values of Hermitian operators.
The authors propose variational quantum algorithms (VQAs) to solve these unconstrained optimization problems. The VQAs utilize parameterized quantum circuits to evaluate the expectation values and a classical optimizer to update the circuit parameters.
For the ECSF case, the authors provide a rigorous analysis of the convergence rate of their proposed VQA, under the assumption that the SDPs are weakly constrained (i.e., the number of constraints is much smaller than the dimension of the input matrices).
The authors also provide numerical simulations to demonstrate the performance of their VQAs, including on a noisy quantum simulator, for applications such as MaxCut.
Overall, this work presents a framework for solving SDP problems using variational quantum algorithms, which can be advantageous when the dimension of the input matrices is large.