Research Article A Novel Filled Function Method for Nonlinear Equations

Size: px

Start display at page:

Download "Research Article A Novel Filled Function Method for Nonlinear Equations"

Kory Hood
5 years ago
Views:

1 Applied Mathematics, Article ID 24759, 7 pages Research Article A Novel Filled Function Method for Nonlinear Equations Liuyang Yuan and Qiuhua Tang 2 College of Sciences, Wuhan University of Science and Technology, Wuhan, Hubei 43008, China 2 Industrial Engineering Department, Wuhan University of Science and Technology, Wuhan, Hubei 43008, China Correspondence should be addressed to Liuyang Yuan; yangly060@26.com Received 24 December 203; Accepted 26 May 204; Published 2 June 204 Academic Editor: Nan-Jing Huang Copyright 204 L. Yuan and Q. Tang. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. A novel filled function method is suggested for solving box-constrained systems of nonlinear equations. Firstly, the original problem is converted into an equivalent global optimization problem. Subsequently, a novel filled function with one parameter is proposed for solving the converted global optimization problem. Some properties of the filled function are studied and discussed. Finally, an algorithm based on the proposed novel filled function for solving systems of nonlinear equations is presented. The objective function value can be reduced by quarter in each iteration of our algorithm. The implementation of the algorithm on several test problems is reported with satisfactory numerical results.. Introduction Systems of nonlinear equations arise in myriad applications, for example, in engineering, physics, mechanics, applied mathematics and sciences; see [] for a more detailed description. In this paper, we consider the following box-constrained systems of nonlinear equations for short, SNE: F x =0, x X, SNE where the mapping F:R n R m is continuous, x R n is a box. Generally, systems of nonlinear equations are very difficult to solve directly. The typical methods to solve SNE are optimization-basedmethodsinwhich SNE is reformulated as an optimization problem. The most popular optimizationbased methods involve solving the following optimization problem for short OP: min s.t. f x = 2 FxT F x x X OP to find solutions of SNE. Note that the problem above is a box-constrained nonlinear least-squares problem. It is easy to see that the objective function satisfies fx 0 and global optimal solutions of problem OP with the zero objective function value corresponding to solutions of SNE. Generally speaking, the traditional optimization-based methods for solving SNE are often stuck at a stationary point or a local minimizer of the corresponding optimization problem, which is not necessarily a solution of the original system.lately,greateffortshavebeenmadetoovercome the difficulty caused by nonglobal minimizers. In Particular, some switching techniques [2 6] have been developed to escape from a stationary point or a local minimizer which is not a solution of SNE. Kanzow [3] incorporated two well-known global optimization algorithms, namely, a tunneling [7] and a filled function method [8], into a standard nonsmooth Newton-type method for solving a nonsmooth system of equations which is a reformulation of the mixed complementarity problem. Wu et al. [9] andlinetal.[0, ] also gave some filled function methods to solve a nonlinear system with box constraints. Wang et al. [2] gave a filled function method to solve an unconstrained nonlinear system. In this paper, we will propose another kind of filled function method to solve boxconstrained systems of nonlinear equations. Unlike [3], we do not use any Newton-type methods to solve SNE, and,also unlike [9, 0], a better initial point of the primal optimization

2 2 Applied Mathematics problem sometimes can not be obtained by minimizing the constructed filled function locally. Moreover, unlike [, 2], the exponential term was used in the construct of the filled function, which may increase the computability when applied to numerical optimization. Instead, in this paper, we use an efficient filled function method to solve the corresponding optimization problem, and the local minimizer of the filled function is always obtained in the interior and is always good point. The objective function value can be reduced by quarter in each iteration of our algorithm. The existence of local minimizers other than global ones makes global optimization a great challenge. As one of the main methods to solve general unconstrained or boxconstrained global optimization problems without special structural property, the filled function method has attracted extensive attention; see [8 5]. The main idea of the filled function method is to construct an auxiliary function called filled function via the current local minimizer of the original optimization problem, with the property that the current local minimizer is a local maximizer of the constructed filled function and a better initial point of the primal optimization problem can be obtained by minimizing the constructed filled function locally. However, generally speaking, the local minimizer of the filled function cannot ensure that it is a better point of the primal optimization problem. In the paper, we propose a new filled function method, which can ensure that the proposed function is an efficient filled function and the local minimizer of the new filled function on a given box set is a better point and the primal problem s objective value at this better point can be reduced by quarter in each iteration of our filled function algorithm. The numerical results obtained show that our method is applicable and efficient. The paper is organized as follows. Following this introduction, a novel filled function is proposed for the optimization problem in Section 2. The corresponding algorithm is presented in Section 3. In Section 4, several numerical examples are reported. Finally, some conclusions are drawn in Section Filled Function for the Optimization Problem Throughout this paper we make the following assumption. Assumption. SNE hasatleastonesolutioninx and the number of solutions of SNE is finite. Suppose that x is a local miminizer of problem OP,the definition of the filled function is as follows. Definition 2. A continuously differentiable function px, x is called a filled function of problem OP at x, if it satisfies the following conditions: x is a strict local maximizer of px, x on X; 2 px, x has no stationary point in the region S = {x : fx > fx /4, x X \ {x }}; 3 If x is not a global minimizer of problem OP,then px, x does have a minimizer in the region S 2 ={x: fx fx /4, x X}. These conditions of the new filled function ensure that when a descent method, for example, the steepest descent method, is employed to minimize the constructed filled function, the sequence of iteration points will not terminate at any point at which the objective function value is large than fx /4;ifx isnotaglobalminimizerofproblemop,then there must exist a minimizer of the filled function at which the objective function value is less than or equal to fx /4, namely, any local minimizer of px, x must belong to the set S 2 ={x:fx fx /4, x X}. Therefore, the present local minimizer of the objective function escapes and a better minimizer can be found by a local search algorithm starting from the minimizer of the filled function. Let LP denote the set of local minimizers of problem OP and let GP denote the set of global minimizers of problem OP. In the following, a novel filled function with one parameter satisfying Definition 2 is introduced. To begin with, we design a continuously differentiable function ht with the following properties: it is equal to 0 when t>0. More specifically, we construct ht as follows: 0, t > 0, h t ={ t 2, t 0. It is not difficult to check that ht is continuously differentiable and decreasing on R.Obviously,we have h 0, t > 0, t ={ 2t, t 0. Given x LP, the following filled function with one parameter is constructed: Fx,x,q= + x x +qhfx fx, 3 4 where the only parameter q > 0. Clearly, Fx, x,q is continuously differentiable on R n. The following theorems show that Fx, x,q satisfy Definition 2 when the positive parameter q is sufficiently large. Theorem 3. Let x LP, q>0.then,x is a strict local maximizer of Fx, x,qon X. Proof. Since x LP, there exists a neighborhood Nx,σ of x with σ>0such that fx fx,forallx Nx,σ, where Nx,σ = {x x x < σ}.then,foranyx Nx,σ, x =x, q>0,andht 0.Wehave Fx,x,q= 2 + x x <=Fx,x,q. 4 Thus, x is a strict local maximizer of Fx, x,qon X.

3 Applied Mathematics 3 Theorem 3 reveals that the proposed new filled function satisfies condition of Definition 2. Theorem 4. Let x LP, q>0.then,fx, x,qhas no stationary point in the region S ={x:fx>fx /4, x X\{x }}. Proof. Assume that x S,namely,fx > fx /4 and x =x. Then, we have F x, x,q= x x + x x 2 x x, F x, x,q x x x x = + x x 2 <0. It implies that the function Fx, x,qhas no stationary point in the region S ={x:fx>fx /4, x X \ {x }}. Theorem 4 revealsthattheproposednewfilledfunction satisfies condition 2 of Definition 2. Theorem 5. Let x LP, butx GP. AndSNE satisfies Assumption.Then,Fx, x,qdoes have a minimizer in the region S 2 ={x:fx fx /4, x X} when q>0is sufficiently large. Proof. Since x LP, butx GP, and the global minimum of fx is zero, there exists an x LOP such that fx fx /4. Bythecontinuityoffx and Assumption, there exists σ > 0 that is small enough and x Nx,σ,anditholdsfx fx fx /4 = fx, for all x Nx,σ,whereNx,σ={x X: x x <σ}. Here,wejustgivetheprooftothecasewhen x x x x. For the other case when x x > x x,the proof is similar. Therefore, for each x Nx, σ = {x X : x x < σ}, there are two cases: fx > fx = fx /4; 2 fx fx = fx /4. For case, by fx fx /4 > fx fx /4 = 0 and x x < x x, Fx,x,q= 5 + x x > + x x =Fx, x,q. 6 For case 2, by fx fx /4 fx fx /4 = 0 and x x < x x,wehavefx, x,q < Fx,x,qif and only if + x x < which is equivalent to q> + x x +qfx fx 4 2, 7 x x x x f x fx /4 2 + x x + x x >0. 8 Let q 0 = x x x x f x fx /4 2 + x x + x x. 9 Thus, there exists sufficiently large q 0 > 0 as function fx approaches fx /4. Consequently, it must has that Fx, x,q < Fx,x,q for all x Nx, σ when q>q 0. Thus, x S 2 is a minimizer of Fx, x,q when q > 0 is sufficiently large. Theorems 5 show that, for all q q 0, Fx, x,qsatisfies Condition 3 of Definition 2. Thefollowingtheoremsshow that function Fx, x,qhas some interesting properties. Theorem 6. Let x,x 2 Xand the following conditions hold: i min{fx, fx 2 } /4fx ; ii x 2 x > x x. Then, the inequality Fx,x,q > Fx 2,x,q holds for all q>0. Proof. Since min{fx, fx 2 } /4fx,then Fx,x,q= + x, x Fx 2,x,q= + x. 2 x Therefore, for all q>0, Fx,x,q>Fx 2,x,qholds. Theorem 7. Fx, x,q>0for all x X. 0 Proof. Bytheformofthefilledfunction3 and since ht 0, we have Fx, x,q>0for all x X. Remark 8. In the phase of minimizing the filled function, Theorems 3 5 guarantee that the present local minimizer x of the objective function is escaped and the minimum of the filled function will be always achieved at a point where the objective function value is not greater than the quarterofthecurrentminimumoftheobjectivefunction. Moreover, the proposed filled function does not include exponential terms. A continuously differentiable function is used in the constructed filled function, which possesses many good properties and is efficient in numerical implementation. 3. Filled Function Algorithm The theoretical properties of the proposed filled function Fx, x,qwere discussed in the last section. In this section, a global optimization method for solving problem OP is presented based on the constructed filled function 3, which leads to a solution or an approximate solution to SNE. Suppose that SNE has at least one solution and the number of solutions is finite. The general idea of the global optimization method is as follows.

4 4 Applied Mathematics Let x 0 Xbeagiven initial point. Starting from this initial point, a local minimizer x 0 of problem OP is obtained with a local minimization method Newton method, Quasi- Newton Method, or Conjugate Gradient method. If x 0 is not a global minimizer, the main task is to find a better local minimizer of problem OP. Consider the following filled function problem for short FFP: min x X Fx,x k,q, FFP where Fx, x k,qis given by 3. Let x 0 be a obtained local minimizer of problem FFP on X,andthenbyTheorem 5,wehavefx 0 fx 0 /4.Starting from this initial point x 0, we can obtain a local minimizer x of problem OP.Ifx is a global minimizer namely, fx = 0, x is the solution of the system SNE; otherwise,locally solve problem FFP.Letx be the obtained local minimizer, andthenwehavethatfx fx 0 /42.Repeatingthis process, we can finally obtain a solution of the system SNE or a sequence {x k } with fx k <fx 0 /4k, k=,2,...for such a sequence {x k }, k=,2,...,whenkis sufficiently large, x k can be regarded as an approximate solution of the system SNE. Let x Xand ε>0,andx is called a ε-approximate solution of the system SNE if x Xand fx ε. The corresponding filled function algorithm for the global optimization problem OP is described as follows. The algorithm is referred as FFSNE the filled function method for SNE. Algorithm FFSNE Step 0. Choosesmallpositivenumbersε, λ, alargepositive number q U,andaninitialvalueq 0 for the parameters q.e.g., ε=0 8, λ=0 5, q U =0 20,andq 0 =0 0. Choose a positive integer number K e.g., K=2nanddirectionse i, i =,...,K, are the coordinate directions. Choose an initial point x 0 X.Setk:= 0. If fx 0 ε,thenletx k := x 0 and go to Step 6.Otherwise, let q:=q 0 and go to Step. Step. Find a local minimizer x k of the problem OP by local search methods starting from x k.iffx k ε,gotostep 6. Step 2. Let Fx,x k,q= + x x k +qhfx fx k, 4 where ht is defined by. Set l=and u = 0.. Step 3. Consider a If l>k,setq := 0q,andgotoStep 5;otherwise,go to b. b If u λ,sety l k := x k +ue l, and go to c; otherwise, set l:=l+, u = 0., go to a. c If y l k X, go to d; otherwise, set u:=u/0,goto b. d If fy l k fx k /4, thensetx k+ := y l k, k:= k+, and go to Step ;otherwise,gotostep 4. Step 4. Search for a local minimizer of the following filled function problem starting from y l k : min x R n Fx,x k,q. 2 Once a point y k Xwith fy k fx k /4 is obtained in the process of searching, set x k+ := y k, k:= k+ and go to Step ; otherwise continue the process. Let x k be an obtained local minimizer of problem 2. If x k satisfies fx k fx k /4, then set x k+ := x k, k:=k+and go to Step ;otherwise,set u:=u/0,andgotostep 3b. Step 5. If q q U,gotoStep 2. Step 6. Let x s =x k and stop. In this algorithm, the termination criteria for minimization of Fx, x k,qin Step 4 can be interpreted as follows. The purpose of minimizing Fx, x k,qis to find a better point y k in set S 2 ={x:fx fx /4, x X}. If it is successful, that is, the solution obtained satisfies y k S 2,thenwecan turn to Step and restart to minimize the objective function with y k as a new starting point. If in the process of searching such point is not found, we can obtain the local minimizer x k of Fx, x k,q.bytheorems4 and 5, weknowthatthere must exist a local minimizer of Fx, x k,qwhich belongs to the set S 2 ={x:fx fx /4, x X}. Then,weturnto Step and minimizer fx starting from the local minimizer x k.obviously,ifλ is small enough, qu is large enough, and the direction set {e,...,e K } is large enough and x s can be obtained from algorithm FFOP within finite steps. 4. Numerical Experiment In this section, several sets of numerical experiments are presented to illustrate the efficiency of algorithm FFSNE. All the numerical experiments are implemented in Matlab200b. In our programs, the local minimizers of problem FFP and problem OP are obtained by the SQP method. Note that fx 0 6 is used as the terminate condition. The symbols used in Table 3 are given in Table. Throughout our computational experiments, the parameters in algorithm FFSNE are set as ε=0 8, λ = 0 5, q U := 0 20, q 0 := 0 0. Problem 9 test problem 4.. in [6]. Consider 4x 3 +4x x 2 +2x x 4=0, 4x x2 +4x x 2 26x 2 22=0, 5 x i 5, i=, Therearenineknownsolutionsasshownin[6]seeTable 2.

5 Applied Mathematics 5 Parameter k x 0 x k fx k Fx k Table Description The number of iterations in finding the kth local minimizer The initial point which satisfies x 0 X The kth local minimizer which we take as a solution or an approximate solution to SNE The function value of fx in finding the kth local minimizer The function value of Fx in finding the kth local minimizer Problem 0 test problem 4..2 in [6]. Consider x x 2 +x 3x 5 =0, Problem 3 test problem 4..5 in [6]. Consider 2x +x 2 +x 3 +x 4 +x 5 6=0, x +2x 2 +x 3 +x 4 +x 5 6=0, x +x 2 +2x 3 +x 4 +x 5 6=0, x +x 2 +x 3 +2x 4 +x 5 6=0, x x 2 x 3 x 4 x 5 =0 2 x i 2, i=,...,5. 8 The known solution is,,,, T and 0.96, 0.96, 0.96, 0.96,.48 T. Problem 4 test problem 4..6 in [6]. Consider 2x x 2 +x +3R 0 x 2 2 +x 2x 2 3 +R 7x 2 x 3 +R 9 x 2 x 4 +R 8 x 2 Rx 5 =0, 2x 2 x 2 3 +R 7x 2 x 3 +2R 5 x 2 3 +R 6x 3 8x 5 =0, R 9 x 2 x 4 +2x 2 4 4Rx 5 =0, x x 2 +x +R 0 x 2 2 +x 2x 2 3 +R 7x 2 x 3 +R 9 x 2 x x x x 2 x x +x x x = 0, x x x 2 x x x x x = 0, x 6 x x x 2 =0, +R 8 x 2 +R 5 x 2 3 +R 6x 3 +x 2 4 =0, x i 00, i =,...,5, x x = 0, x 2 +x2 2 =0, 9 where R = 0, R 5 = 0.93, R 6 = , R 7 = , R 8 = , R 9 = , and R 0 = The known solution in [6] is , , , , T. Problem test problem 4..3 in [6]. Consider 0 4 x x 2 =0, x 2 3 +x2 4 =0, x 2 5 +x2 6 =0, x 2 7 +x2 8 =0, x i, i=,...,8. The are sixteen known solutions of problem 6 are given in [6]. exp x +exp x 2.00 = 0, x 4.553, x The known solution is , T. 6 Problem 5 test problem in [7]. Consider 2x sin 4πx 2 x =0, x 2 0.5sin 2πx =0, 0 x, x Problem 2 test problem 4..4 in [6]. Consider The known solution is , T. 0.5 sin x x x 2 0.5x π =0, 0.25 π e2x e+ ex 2 π 2ex =0, 0.25 x,.5 x 2 2π. 7 The known solution is , T and 0.5, T. Problem 6 test problem 7 in [8]. Consider x 0.7 sin x 0.2 cos x 2 =0, x cos x +0.2sin x 2 =0, 00 x, x The known solution is , T. 2

6 6 Applied Mathematics Table 2 Sol x x Table 3: Computational results for Problems 9 6. Prob. Number x 0 k x k fx k Fx k The numerical results are listed in Table 3.FromTable 3, it is easy to see that all problems that we tested have been solved with a small number of iterations. 5. Conclusions In this paper, the filled function Fx, x,qwith one parameter is constructed for solving nonlinear equations and it has been proved that it satisfies the basic characteristics of the filled function definition. Promising computation results have been observed from our numerical experiments. In the future, the filled function method can be used to solve otherproblemssuchasnonlinearsystemsofequalitiesand inequalities and nonlinear feasibility problems with expensive functions. Conflict of Interests The authors declare that there is no conflict of interests regarding the publication of this paper. Acknowledgment This work was supported by the Natural Science Foundation of China no and no References [] S. Fucik and A. Kufner, Nonlinear Differential Equations, Scientific Publishing, New York, NY, USA, 2008.

7 Applied Mathematics 7 [2] X. Chen, L. Qi, and Y. F. Yang, Lagrangian globalization methods for nonlinear complementarity problems, Optimization Theory and Applications,vol.2,no.,pp.77 95, [3] C. Kanzow, Global optimization techniques for mixed complementarity problems, Global Optimization, vol. 6, no., pp. 2, [4] J. L. Nazareth and L. Qi, Globalization of Newton s methods for solving nonlinear equations, Numerical Linear Algebra with Applications,vol.3,pp ,996. [5] L. Qi and Y. F. Yang, NCP functions applied to Lagrangian globalization for the nonlinear complementarity problem, Global Optimization,vol.24,no.2,pp ,2002. [6] X. J. Tong, L. Qi, and Y. F. Yang, The Lagrangian globalization method for nonsmooth constrained equations, Computational Optimization and Applications,vol.33,no.,pp.89 09,2006. [7] A. V. Levy and A. Montalvo, The tunneling algorithm for the global minimization of functions, SIAM Journal on Scientific and Statistical Computing,vol.6,no.,pp.5 29,985. [8] R. P. Ge, A filled function method for finding a global minimizer of a function of several variables, Mathematical Programming,vol.46,no.2,pp.9 204,990. [9]Z.Y.Wu,M.Mammadov,F.S.Bai,andY.J.Yang, Afilled function method for nonlinear equations, Applied Mathematics and Computation,vol.89,no.2,pp ,2007. [0] Y. Lin, Y. J. Yang, and M. Mammadov, A new filled function method for nonlinear equations, Applied Mathematics and Computation,vol.20,no.2,pp.4 42,2009. [] Y. Lin and Y. Yang, Filled function method for nonlinear equations, Computational and Applied Mathematics, vol. 234, no. 3, pp , 200. [2]C.J.Wang,R.H.Luo,K.Wu,andB.S.Han, Anewfilled function method for an unconstrained nonlinear equation, Computational and Applied Mathematics,vol.235,no. 6, pp , 20. [3] Y. Lin and Y. Yang, A filled function method for box constrained nonlinear integer programming, the Korean Mathematical Society,vol.48,no.5,pp ,20. [4] L.Y.Yuan,Z.Wan,J.J.Zhang,andB.Sun, Afilledfunction method for solving nonlinear complementarity problem, Journal of Industrial and Management Optimization,vol.5,no.4,pp , [5] W. Zhu, Globally concavized filled function method for the box constrained continuous global minimization problem, Optimization Methods and Software,vol.2,no.4,pp , [6] C. A. Floudas, P. M. Pardalos, C. S. Adjiman et al., Handbook of Test Problems in Local and Global Optimization, vol. 33, Kluwer Academic, Dordrecht, The Netherland, 999. [7]L.S.Zhang,C.K.Ng,D.Li,andW.W.Tian, Anewfilled function method for global optimization, Global Optimization,vol.28,no.,pp.7 43,2004. [8] Y. Zhang and Z. H. Huang, A nonmonotone smoothing-type algorithm for solving a system of equalities and inequalities, Computational and Applied Mathematics,vol.233,no. 9, pp , 200.

8 Advances in Operations Research Advances in Decision Sciences Applied Mathematics Algebra Probability and Statistics The Scientific World Journal International Differential Equations Submit your manuscripts at International Advances in Combinatorics Mathematical Physics Complex Analysis International Mathematics and Mathematical Sciences Mathematical Problems in Engineering Mathematics Discrete Mathematics Discrete Dynamics in Nature and Society Function Spaces Abstract and Applied Analysis International Stochastic Analysis Optimization

Research Article Modified T-F Function Method for Finding Global Minimizer on Unconstrained Optimization

Mathematical Problems in Engineering Volume 2010, Article ID 602831, 11 pages doi:10.1155/2010/602831 Research Article Modified T-F Function Method for Finding Global Minimizer on Unconstrained Optimization