Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userSTRIDE along Spectrahedral Vertices for Solving Large-Scale Rank-One Semidefinite Relaxations
170
0
0.0
(
0
)
Added by
Heng Yang
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
We consider solving high-order semidefinite programming (SDP) relaxations of nonconvex polynomial optimization problems (POPs) that admit rank-one optimal solutions. Existing approaches, which solve the SDP independently from the POP, either cannot scale to large problems or suffer from slow convergence due to the typical degeneracy of such SDPs. We propose a new algorithmic framework, called SpecTrahedral pRoximal gradIent Descent along vErtices (STRIDE), that blends fast local search on the nonconvex POP with global descent on the convex SDP. Specifically, STRIDE follows a globally convergent trajectory driven by a proximal gradient method (PGM) for solving the SDP, while simultaneously probing long, but safeguarded, rank-one strides, generated by fast nonlinear programming algorithms on the POP, to seek rapid descent. We prove STRIDE has global convergence. To solve the subproblem of projecting a given point onto the feasible set of the SDP, we reformulate the projection step as a continuously differentiable unconstrained optimization and apply a limited-memory BFGS method to achieve both scalability and accuracy. We conduct numerical experiments on solving second-order SDP relaxations arising from two important applications in machine learning and computer vision. STRIDE dominates a diverse set of five existing SDP solvers and is the only solver that can solve degenerate rank-one SDPs to high accuracy (e.g., KKT residuals below 1e-9), even in the presence of millions of equality constraints.
rate research
Read More
The multiple-input multiple-output (MIMO) detection problem, a fundamental problem in modern digital communications, is to detect a vector of transmitted symbols from the noisy outputs of a fading MIMO channel. The maximum likelihood detector can be formulated as a complex least-squares problem with discrete variables, which is NP-hard in general. Various semidefinite relaxation (SDR) methods have been proposed in the literature to solve the problem due to their polynomial-time worst-case complexity and good detection error rate performance. In this paper, we consider two popular classes of SDR-based detectors and study the conditions under which the SDRs are tight and the relationship between different SDR models. For the enhanced complex and real SDRs proposed recently by Lu et al., we refine their analysis and derive the necessary and sufficient condition for the complex SDR to be tight, as well as a necessary condition for the real SDR to be tight. In contrast, we also show that another SDR proposed by Mobasher et al. is not tight with high probability under mild conditions. Moreover, we establish a general theorem that shows the equivalence between two subsets of positive semidefinite matrices in different dimensions by exploiting a special separable structure in the constraints. Our theorem recovers two existing equivalence results of SDRs defined in different settings and has the potential to find other applications due to its generality.
Given all (finite) moments of two measures $mu$ and $lambda$ on $R^n$, we provide a numerical scheme to obtain the Lebesgue decomposition $mu= u+psi$ with $ ulllambda$ and $psiperplambda$. When$ u$ has a density in $L_infty(lambda)$ then we obtain two sequences of finite moments vectorsof increasing size (the number of moments) which converge to the moments of $ u$ and $psi$ respectively, as the number of moments increases. Importantly, {it no} `a priori knowledge on the supports of $mu, u$ and $psi$ is required.
We report our progress on the project for solving larger scale quadratic assignment problems (QAPs). Our main approach to solve large scale NP-hard combinatorial optimization problems such as QAPs is a parallel branch-and-bound method efficiently implemented on a powerful computer system using the Ubiquity Generator (UG) framework that can utilize more than 100,000 cores. Lower bounding procedures incorporated in the branch-and-bound method play a crucial role in solving the problems. For a strong lower bounding procedure, we employ the Lagrangian doubly nonnegative (DNN) relaxation and the Newton-bracketing method developed by the authors group. In this report, we describe some basic tools used in the project including the lower bounding procedure and branching rules, and present some preliminary numerical results. Our next target problem is QAPs with dimension at least 50, as we have succeeded to solve tai30a and sko42 from QAPLIB for the first time.
For an ideal I with a positive dimensional real variety, based on moment relaxations, we study how to compute a Pommaret basis which is simultaneously a Groebner basis of an ideal J generated by the kernel of a truncated moment matrix and nesting between I and its real radical ideal. We provide a certificate consisting of a condition on coranks of moment matrices for terminating the algorithm. For a generic delta-regular coordinate system, we prove that the condition is satisfiable in a large enough order of moment relaxations.
We propose an algorithm for solving nonlinear convex programs defined in terms of a symmetric positive semidefinite matrix variable $X$. This algorithm rests on the factorization $X=Y Y^T$, where the number of columns of Y fixes the rank of $X$. It is thus very effective for solving programs that have a low rank solution. The factorization $X=Y Y^T$ evokes a reformulation of the original problem as an optimization on a particular quotient manifold. The present paper discusses the geometry of that manifold and derives a second order optimization method. It furthermore provides some conditions on the rank of the factorization to ensure equivalence with the original problem. The efficiency of the proposed algorithm is illustrated on two applications: the maximal cut of a graph and the sparse principal component analysis problem.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
