Bipartite Graphs with the Maximum Sum of Squares of Degrees

In this paper we determine all the bipartite graphs with the maximum sum of squares of degrees among the ones with a given number of vertices and edges.

Sums of Squares of Polygonal Numbers

Polygonal numbers and sums of squares of primes are distinct fields of number theory. Here we consider sums of squares of consecutive (of order and rank) polygonal numbers. We try to express sums of squares of polygonal numbers of consecutive orders in matrix form. We also try to find the solution of a Diophantine equation in terms of polygonal numbers.

Expanding Annular Domain Algorithm to Estimate Domains of Attraction for Power System Stability Analysis

This paper presents an Expanding Annular Domain(EAD)algorithm combined with Sum of Squares(SOS)programming to estimate and maximize the domain of attraction(DA)of power systems.The proposed algorithm can systematically construct polynomial Lyapunov functions for power systems with transfer conductance and reliably determine a less conservative approximated DA,which are quite difficult to achieve with traditional methods.With linear SOS programming,we begin from an initial estimated DA,then enlarge it by iteratively determining a series of so-called annular domains of attraction,each of which is characterized by level sets of two successively obtained Lyapunov functions.Moreover,the EAD algorithm is theoretically analyzed in detail and its validity and convergence are shown under certain conditions.In the end,our method is tested on two classical power system cases and is demonstrated to be superior to existing methods in terms of computational speed and conservativeness of results.

Sum-of-Squares Design Method for Four-Parameter Lag-Lead Compensator

The four-parameter lag-lead compensator design has received much attention in the last two decades. However, most approaches have been either trial-and-error or only for special cases. This paper presents a non-trial-and-error design method for four-parameter lag-lead compensators. Here, the compensator design problem is formulated into a polynomial function optimization problem and solved by using the recently developed sum-of-squares （SOS） techniques. This result not only provides a useful design method but also shows the power of the SOS techniques.

Characterization of Arithmetic Functions that Preserve the Sum-of-squares Operation

We characterize all functions f ： N → C such that f（m^2 ＋ n^2） = f（m）^2 ＋ f（n）^2 for all m, n ∈ N. It turns out that all such functions can be grouped into three families, namely f ≡ 0, f（n） = ±n （subject to some restrictions on when the choice of the sign is possible） and f（n） =±l/2（again subject to some restrictions on when the choice of the sign is possible）.

Reachable Set Estimation and Safety Verification of Nonlinear Systems via Iterative Sums of Squares Programming

In this paper,the problems of forward reachable set estimation and safety verification of uncertain nonlinear systems with polynomial dynamics are addressed.First,an iterative sums of squares(SOS)programming approach is developed for reachable set estimation.It characterizes the over-approximations of the forward reachable sets by sub-level sets of time-varying Lyapunovlike functions that satisfy an invariance condition,and formulates the problem of searching for the Lyapunov-like functions as a bilinear SOS program,which can be solved via an iterative algorithm.To make the over-approximation tight,the proposed approach seeks to minimize the volume of the overapproximation set with a desired shape.Then,the reachable set estimation approach is extended for safety verification,via explicitly encoding the safety constraint such that the Lyapunov-like functions guarantee both reaching and avoidance.The efficiency of the presented method is illustrated by some numerical examples.

WEIGHTED NONLINEAR REGRESSION

Minimization of the weighted nonlinear sum of squares of differences may be converted to the minimization of sum of squares. The Gauss-Newton method is recalled and the length of the step of the steepest descent method is determined by substituting the steepest descent direction in the Gauss-Newton formula. The existence of minimum is shown.

Certifying the Global Optimality of Quartic Minimization over the Sphere

The quartic minimization over the sphere is an NP-hard problem in the general case.There exist various methods for computing an approximate solution for any given instance.In practice,it is quite often that a global optimal solution was found but without a certification.We will present in this article two classes of methods which are able to certify the global optimality,i.e.,algebraic methods and semidefinite program(SDP)relaxation methods.Several advances on these topics are summarized,accompanied with some emerged new results.We want to emphasize that for mediumor large-scaled instances,the problem is still a challenging one,due to an apparent limitation on the current force for solving SDP problems and the intrinsic one on the approximation techniques for the problem.

An Approximation Bound Analysis for Lasserre’s Relaxation in Multivariate Polynomial Optimization

Suppose f,g1,…,gm are multivariate polynomials in x∈R^(n)and their degrees are at most 2d.Consider the problem:Minimize f(x)subject to g1(x)≥0,…,gm(x)≥0.Let f_(min)(resp.,f_(max))be the minimum(resp.,maximum)of f on the feasible set S,and f_(sos)be the lower bound of_(fmin)given by Lasserre’s relaxation of order d.This paper studies its approximation bound.Under a suitable condition on g1,…,gm,we prove that(f_(max)−f_(sos))≤Q(f_(max)−f_(min))with Q a constant depending only on g1,…,gm but not on f.In particular,if S is the unit ball,Q=O(n^(d));if S is the hypercube,Q=O(n^(2d));if S is the boolean set,Q=O(n^(d)).

Sign-based Test for Mean Vector in High-dimensional and Sparse Settings

In this article, we introduce a robust sparse test statistic which is based on the maximum type statistic. Both the limiting null distribution of the test statistic and the power of the test are analysed. It is shown that the test is particularly powerful against sparse alternatives. Numerical studies are carried out to examine the numerical performance of the test and to compare it with other tests available in the literature. The numerical results show that the test proposed significantly outperforms those tests in a range of settings, especially for sparse alternatives.