Quadratic Semidefinite Programming Research Articles

A revised sequential quadratic semidefinite programming method for nonlinear semidefinite optimization

In 2020, Yamakawa and Okuno proposed a stabilized sequential quadratic semidefinite programming (SQSDP) method for solving, in particular, degenerate nonlinear semidefinite optimization problems. The algorithm is shown to converge globally without a constraint qualification, and it has some nice properties, including the feasible subproblems, and their possible inexact computations. In particular, the convergence was established for approximate-Karush-Kuhn-Tucker (AKKT) and trace-AKKT conditions, which are two sequential optimality conditions for the nonlinear conic contexts. However, recently, complementarity-AKKT (CAKKT) conditions were also considered, as an alternative to the previous mentioned ones, that is more practical. Since few methods are shown to converge to CAKKT points, at least in conic optimization, and to complete the study associated to the SQSDP, here we propose a revised version of the method, maintaining the good properties. We modify the previous algorithm, prove the global convergence in the sense of CAKKT, and show some preliminary numerical experiments.

A stabilized sequential quadratic semidefinite programming method for degenerate nonlinear semidefinite programs

In this paper, we propose a new sequential quadratic semidefinite programming (SQSDP) method for solving degenerate nonlinear semidefinite programs (NSDPs), in which we produce iteration points by solving a sequence of stabilized quadratic semidefinite programming (QSDP) subproblems, which we derive from the minimax problem associated with the NSDP. Unlike the existing SQSDP methods, the proposed one allows us to solve those QSDP subproblems inexactly, and each QSDP is feasible. One more remarkable point of the proposed method is that constraint qualifications (CQs) or boundedness of Lagrange multiplier sequences are not required in the global convergence analysis. Specifically, without assuming such conditions, we prove the global convergence to a point satisfying any of the following: the stationary conditions for the feasibility problem, the approximate-Karush-Kuhn-Tucker (AKKT) conditions, and the trace-AKKT conditions. Finally, we conduct some numerical experiments to examine the efficiency of the proposed method.

Density Functional Theory Transformed into a One-Electron Reduced-Density-Matrix Functional Theory for the Capture of Static Correlation.

Density Functional Theory (DFT), the most widely adopted method in modern computational chemistry, fails to describe accurately the electronic structure of strongly correlated systems. Here we show that DFT can be formally and practically transformed into a one-electron reduced-density-matrix (1-RDM) functional theory, which can address the limitations of DFT while retaining favorable computational scaling compared to wave function based approaches. In addition to relaxing the idempotency restriction on the 1-RDM in the kinetic energy term, we add a quadratic 1-RDM-based term to DFT's density-based exchange-correlation functional. Our approach, which we implement by quadratic semidefinite programming at DFT's computational scaling of O(r3), yields substantial improvements over traditional DFT in the description of static correlation in chemical structures and processes such as singlet biradicals and bond dissociations.

An accelerated active-set algorithm for a quadratic semidefinite program with general constraints

In this paper, we are concerned with efficient algorithms for solving the least squares semidefinite programming which contains many equalities and inequalities constraints. Our proposed method is built upon its dual formulation and is a type of active-set approach. In particular, by exploiting the nonnegative constraints in the dual form, our method first uses the information from the Barzlai–Borwein step to estimate the active/inactive sets, and within an adaptive framework, it then accelerates the convergence by switching the L-BFGS iteration and the semi-smooth Newton iteration dynamically. We show the global convergence under mild conditions, and furthermore, the local quadratic convergence under the additional nondegeneracy condition. Various types of synthetic as well as real-world examples are tested, and preliminary but promising numerical experiments are reported.

A Three-Operator Splitting Perspective of a Three-Block ADMM for Convex Quadratic Semidefinite Programming and Beyond

In recent years, several convergent variants of the multi-block alternating direction method of multipliers (ADMM) have been proposed for solving the convex quadratic semidefinite programming via its dual, which is inherently a [Formula: see text]-block separable convex optimization problem with coupled linear constraints. Among these multi-block ADMM-type algorithms, the modified [Formula: see text]-block ADMM in [Chang, XK, SY Liu and X Li (2016). Modified alternating direction method of multipliers for convex quadratic semidefinite programming. Neurocomputing, 214, 575–586] bears a peculiar feature that the augmented Lagrangian function is not necessarily to be minimized with respect to the block-variable corresponding to the quadratic term in the objective function. In this paper, we lay the theoretical foundation of this phenomenon by interpreting this modified [Formula: see text]-block ADMM as a special implementation of the Davis–Yin [Formula: see text]-operator splitting [Davis, D and WT Yin (2017). A three-operator splitting scheme and its optimization applications. Set-Valued and Variational Analysis, 25, 829–858]. Based on this perspective, we are able to extend this modified [Formula: see text]-block ADMM to a generalized [Formula: see text]-block ADMM, in the sense of [Eckstein, J and DP Bertsekas (1992). On the Douglas–Rachford splitting method and the proximal point algorithm for maximal monotone operators. Mathematical Programming, 55, 293–318], which not only applies to the more general convex composite quadratic programming problems but also admits the flexibility of achieving even better numerical performance.

Inverse semidefinite quadratic programming problem with [formula omitted] norm measure

We consider an inverse problem arising from a semidefinite quadratic programming (SDQP) problem, which is a minimization problem involving l1 vector norm with positive semidefinite cone constraint. By using convex optimization theory, the first order optimality condition of the problem can be formulated as a semismooth equation. Under two assumptions, we prove that any element of the generalized Jacobian of the equation at its solution is nonsingular. Based on this, a smoothing approximation operator is given and a smoothing Newton method is proposed for solving the solution of the semismooth equation. We need to compute the directional derivative of the smoothing operator at the corresponding point and to solve one linear system per iteration in the Newton method and its global convergence is demonstrated. Finally, we give the numerical results to show the effectiveness and stability of the smoothing Newton method for this inverse problem.

Quadratic Semidefinite Programming for Waveform-Constrained Joint Filter-Signal Design in STAP

In waveform-adaptive space-time adaptive processing (STAP), signal-dependent interference can be mitigated through joint filter-signal design. For the common signal-to-interference-plus-noise ratio (SINR) optimality criterion, these designs result in biquadratic problems, which can be efficiently solved through joint relaxation. In this paper, we consider the addition of two realistic waveform constraints to this process, constant modulus (CM) and the similarity constraint (SC), and demonstrate that they are both amenable to a quadratic semidefinite program (QSDP) via relaxation. We examine the optimality conditions of the newly constrained problems and demonstrate, both analytically and numerically, improved performance over existing iterative techniques. To increase solution robustness, additional refinement strategies for rank-inflated relaxed solutions are also provided and analyzed which extend the aforementioned performance gain.

On the equivalence of inexact proximal ALM and ADMM for a class of convex composite programming

In this paper, we show that for a class of linearly constrained convex composite optimization problems, an (inexact) symmetric Gauss–Seidel based majorized multi-block proximal alternating direction method of multipliers (ADMM) is equivalent to an inexact proximal augmented Lagrangian method. This equivalence not only provides new perspectives for understanding some ADMM-type algorithms but also supplies meaningful guidelines on implementing them to achieve better computational efficiency. Even for the two-block case, a by-product of this equivalence is the convergence of the whole sequence generated by the classic ADMM with a step-length that exceeds the conventional upper bound of $$(1+\sqrt{5})/2$$ , if one part of the objective is linear. This is exactly the problem setting in which the very first convergence analysis of ADMM was conducted by Gabay and Mercier (Comput Math Appl 2(1):17–40, 1976), but, even under notably stronger assumptions, only the convergence of the primal sequence was known. A collection of illustrative examples are provided to demonstrate the breadth of applications for which our results can be used. Numerical experiments on solving a large number of linear and convex quadratic semidefinite programming problems are conducted to illustrate how the theoretical results established here can lead to improvements on the corresponding practical implementations.

A superlinearly convergent SSDP algorithm for nonlinear semidefinite programming

In this paper, we present a sequential semidefinite programming (SSDP) algorithm for nonlinear semidefinite programming. At each iteration, a linear semidefinite programming subproblem and a modified quadratic semidefinite programming subproblem are solved to generate a master search direction. In order to avoid Maratos effect, a second-order correction direction is determined by solving a new quadratic programming. And then a penalty function is used as a merit function for arc search. The superlinear convergence is shown under the strict complementarity and the strong second-order sufficient conditions with the sigma term. Finally, some preliminary numerical results are reported.

A nonconvex ADMM for a class of sparse inverse semidefinite quadratic programming problems

ABSTRACTIn this paper, we consider a class of sparse inverse semidefinite quadratic programming problems, in which a nonconvex alternating direction method of multiplier is investigated. Under mild conditions, we establish convergence results of our algorithm and the corresponding non-ergodic iteration-complexity is also considered under the assumption that the potential function satisfies the famous Kurdyka–Łojasiewicz property. Numerical results show that our algorithm is suitable to solve the given sparse inverse semidefinite quadratic programming problems.

An Extension of Yuan’s Lemma to Fourth-Order Tensor System

Yuan’s lemma is a basic proposition on homogeneous quadratic function systems. In this note, we extend Yuan’s lemma to the fourth-order tensor system. We first give generalized definition of positive semidefinite of fourth-order tensor, and based on it, an extension of Yuan’s lemma is proposed. As an application, we establish the strong duality result of a class of quadratic semidefinite programming problems using the extended Yuan’s lemma.

An SQP-type Method with Superlinear Convergence for Nonlinear Semidefinite Programming

A sequentially semidefinite programming method is proposed for solving nonlinear semidefinite programming problem (NLSDP). Inspired by the sequentially quadratic programming (SQP) method, the algorithm generates a search direction by solving a quadratic semidefinite programming subproblem at each iteration. The [Formula: see text] exact penalty function and a line search strategy are used to determine whether the trial step can be accepted or not. Under mild assumptions, the proposed algorithm is globally convergent. In order to avoid the Maratos effect, we present a modified SQP-type algorithm with the second-order correction step and prove that the fast local superlinear convergence can be obtained under the strict complementarity and the second-order sufficient condition with the sigma term. Finally, some numerical experiments are given to show the effectiveness of the algorithm.

On the R-superlinear convergence of the KKT residuals generated by the augmented Lagrangian method for convex composite conic programming

Due to the possible lack of primal-dual-type error bounds, it was not clear whether the Karush–Kuhn–Tucker (KKT) residuals of the sequence generated by the augmented Lagrangian method (ALM) for solving convex composite conic programming (CCCP) problems converge superlinearly. In this paper, we resolve this issue by establishing the R-superlinear convergence of the KKT residuals generated by the ALM under only a mild quadratic growth condition on the dual of CCCP, with easy-to-implement stopping criteria for the augmented Lagrangian subproblems. This discovery may help to explain the good numerical performance of several recently developed semismooth Newton-CG based ALM solvers for linear and convex quadratic semidefinite programming.

QSDPNAL: a two-phase augmented Lagrangian method for convex quadratic semidefinite programming

In this paper, we present a two-phase augmented Lagrangian method, called QSDPNAL, for solving convex quadratic semidefinite programming (QSDP) problems with constraints consisting of a large number of linear equality and inequality constraints, a simple convex polyhedral set constraint, and a positive semidefinite cone constraint. A first order algorithm which relies on the inexact Schur complement based decomposition technique is developed in QSDPNAL-Phase I with the aim of solving a QSDP problem to moderate accuracy or using it to generate a reasonably good initial point for the second phase. In QSDPNAL-Phase II, we design an augmented Lagrangian method (ALM) wherein the inner subproblem in each iteration is solved via inexact semismooth Newton based algorithms. Simple and implementable stopping criteria are designed for the ALM. Moreover, under mild conditions, we are able to establish the rate of convergence of the proposed algorithm and prove the R-(super)linear convergence of the KKT residual. In the implementation of QSDPNAL, we also develop efficient techniques for solving large scale linear systems of equations under certain subspace constraints. More specifically, simpler and yet better conditioned linear systems are carefully designed to replace the original linear systems and novel shadow sequences are constructed to alleviate the numerical difficulties brought about by the crucial subspace constraints. Extensive numerical results for various large scale QSDPs show that our two-phase algorithm is highly efficient and robust in obtaining accurate solutions. The software reviewed as part of this submission was given the DOI (Digital Object Identifier) https://doi.org/10.5281/zenodo.1206980 .

Convergent prediction–correction-based ADMM for multi-block separable convex programming

The direct extension of the classic alternating direction method with multipliers (ADMMe) to the multi-block separable convex optimization problem is not necessarily convergent, though it often performs very well in practice. In order to preserve the numerical advantages of ADMMe and obtain convergence, many modified ADMM were proposed by correcting the output of ADMMe or employing proximal terms to solve inexactly the subproblems in ADMMe. In this paper, we present an efficient Prediction–Correction-based ADMM (PCB-ADMM) to solve the multi-block separable convex minimization model. The prediction step takes a special block coordinate descent (BCD) cycle to update the variable blocks, then the correction step corrects the output slightly by computing a convex combination of two points from the prediction step and previous iteration. The convergence property is obtained by using the variational inequality. The numerical experiments illustrate effectiveness of the proposed PCB-ADMM to solve the quadratic semidefinite programming and image decomposition.

Modified alternating direction method of multipliers for convex quadratic semidefinite programming

The direct extension of alternating direction method of multipliers (ADMM) is not necessarily convergent to the dual form of convex quadratic semidefinite programming (CQSDP) problems, though the convergence of ADMM is proved when two blocks of variables are alternatively updated. In this paper, we present a modified ADMM for solving the dual form of the CQSDP problems. At each iteration, the proposed method did not have to work out the sub-problem with the primal variable, compared with the existing ADMM. Assuming that the penalty parameter satisfies the condition related to the quadratic term of the primal objective function, the convergence is proved by using a fixed-point argument. Numerical results for the nearest correlation matrix problems and the random CQSDP problems demonstrate the efficiency of our proposed algorithm.

On the superlinear local convergence of a penalty-free method for nonlinear semidefinite programming

This paper is concerned with a sequentially semidefinite programming (SSDP) algorithm for solving nonlinear semidefinite programming problems (NLSDP), which does not use a penalty function or a filter. This method, inspired by the classic SQP method, calculates a trial step by a quadratic semidefinite programming subproblem at each iteration. The trial step is determined such that either the value of the objective function or the measure of constraint violation is sufficiently reduced. In order to guarantee global convergence, the measure of constraint violation in each iteration is required not to exceed a progressively decreasing limit. We prove the global convergence properties of the algorithm under mild assumptions. We also analyze the local behaviour of the proposed method while using a second order correction strategy to avoid Maratos effect. We prove that, under the strict complementarity and the strong second order sufficient conditions with the sigma term, the rate of local convergence is superlinear. Finally, some numerical results with nonlinear semidefinite programming formulation of control design problem with the data contained in COMPleib are given.

An alternating direction method for solving a class of inverse semi-definite quadratic programming problems

In this paper, we propose an alternating-direction-type numerical method to solvea class of inverse semi-definite quadratic programming problems. Anexplicit solution to one direction subproblem is given and the otherdirection subproblem is proved to be a convex quadratic programmingproblem over positive semi-definite symmetric matrix cone. We designa spectral projected gradient method for solving the quadraticmatrix optimization problem and demonstrate its convergence.Numerical experiments illustrate that our method can solve inversesemi-definite quadratic programming problems efficiently.

A Schur complement based semi-proximal ADMM for convex quadratic conic programming and extensions

This paper is devoted to the design of an efficient and convergent {semi-proximal} alternating direction method of multipliers (ADMM) for finding a solution of low to medium accuracy to convex quadratic conic programming and related problems. For this class of problems, the convergent two block semi-proximal ADMM can be employed to solve their primal form in a straightforward way. However, it is known that it is more efficient to apply the directly extended multi-block semi-proximal ADMM, though its convergence is not guaranteed, to the dual form of these problems. Naturally, one may ask the following question: can one construct a convergent multi-block semi-proximal ADMM that is more efficient than the directly extended semi-proximal ADMM? Indeed, for linear conic programming with 4-block constraints this has been shown to be achievable in a recent paper by Sun, Toh and Yang [arXiv preprint arXiv:1404.5378, (2014)]. Inspired by the aforementioned work and with the convex quadratic conic programming in mind, we propose a Schur complement based convergent semi-proximal ADMM for solving convex programming problems, with a coupling linear equality constraint, whose objective function is the sum of two proper closed convex functions plus an arbitrary number of convex quadratic or linear functions. Our convergent semi-proximal ADMM is particularly suitable for solving convex quadratic semidefinite programming (QSDP) with constraints consisting of linear equalities, a positive semidefinite cone and a simple convex polyhedral set. The efficiency of our proposed algorithm is demonstrated by numerical experiments on various examples including QSDP.

Conditional Quadratic Semidefinite Programming: Examples and Methods

The conditional quadratic semidefinite programming (cQSDP) refers to a class of matrix optimization problems whose matrix variables are required to be positive semidefinite on a subspace, and the objectives are quadratic. The chief purpose of this paper is to focus on two primal examples of cQSDP: the problem of matrix completion/approximation on a subspace and the Euclidean distance matrix problem. For the latter problem, we review some classical contributions and establish certain links among them. Moreover, we develop a semismooth Newton method for a special class of cQSDP and establish its quadratic convergence under the condition of constraint nondegeneracy. We also include an application in calibrating the correlation matrix in Libor market models. We hope this work will stimulate new research in cQSDP.