| E-ISSN: 2676-6167 | P-ISSN: 2538-5100

Computational Algorithms and Numerical Dimensions

www.journal-cand.com

Com. Alg. Num. Dim Vol. 1, No. 4 (2022) 130–136.

Paper Type: Original Article

The Modification of the Generalized Gauss-Seidel Iteration
Techniques for Absolute Value Equations
Rashid Ali1,* , Ashfaq Ahmad1, Ijaz Ahmad1, Asad Ali1
1 Schoolof Mathematics and Statistics, Central South University, Changsha 410083, Hunan P.R. China; [email protected];
[email protected]; [email protected]; [email protected].

Citation:
Ali, Sh., Ahmad, A., Ahmad, I., & Ali, A. (2022). The modification of the generalized gauss-seidel
iteration techniques for absolute value equations. Computational algorithms and numerical dimensions,
1(4), 130-136.

Received: 26/08/2022 Reviewed: 26/09/2022 Revised: 02/10/2022 Accept: 19/10/2022

Abstract
This paper proposes two modified generalized Gauss-Seidel iteration techniques to determine the Absolute Value
Equations (AVEs). Convergence of the new techniques is established under some appropriate conditions lastly; several
numerical examples verify the significance of the techniques.

Keywords: Gauss-seidel iteration techniques, Convergence, Absolute value equations, Numerical examples.

1 | Introduction
Licensee
In this article, we explore the AVE of the form;
Computational
Algorithms and
(1)
Numerical Dimensions.
This article is an open
access article distributed Here and represents the vector with the absolute value of each element of .
under the terms and
conditions of the Creative The AVE (1) has attracted significant interest in the optimization domain for more than twenty
Commons Attribution years. The most important reason is that they are closely related to Linear Complementarity Problems
(CC BY) license
(LCP) [1], [2] and horizontal LCP [3], which include a vast range of mathematical programming
(http://creativecommons.
problems and have a wide range of applications [4]-[10].
org/licenses/by/4.0).

In recent years, many efforts have been made to develop techniques for finding numerical solutions
to the AVE (1). For example, Salkuyeh [11] proposed the Picard-HSS approach to compute Eq. (1).
The SOR-like system is discussed in [12]. Additionally, Chen et al. [13] explored various convergence
effects for a SOR-like strategy using different parameters. In [15], the authors presented the shift
splitting technique to obtain Eq. (1) and established the convergence properties of the offered
approach. In the study of Moosaei et al. [16], two approaches are offered to determine Eq (1), the
Homotopy perturbation approach and the Newton approach with the Armijo step.
Corresponding Author: [email protected]
https://doi.org/10.22105/cand.2022.157605
 Fakharzadeh and Shams [17] offered a Mixed-Type Splitting (MTS) strategy for the calculation of Eq. (1)
and demonstrated its convergence conditions. In addition, there are several other numerical techniques in
the literature to determine Eq. (1) [18]-[24].

The present study extends the Generalized Gauss-Seidel (GGS) method [14] and gives Modified GGS
131 (MGGS) techniques, which can be improved by adding three extra parameters. In addition, we examine
the convergence of the novel techniques under appropriate situations.

An overview of this paper is provided below. In Section 2, we discuss the auxiliary results. Section 3
examines the proposed techniques and their convergence. In Sections 4 and Sections 5, we present
numerical results as well as conclusions.

2 | Preliminaries
The modification of the generalized gauss-seidel iteration techniques for absolute value equations

This section provides a helpful lemma and a few notations for later analysis.
Throughout this paper, and represent the absolute value as well as the infinity norm of
, respectively. In addition, the symbol signifies the identity matrix.

Lemma 1 ([31]). Let Then .

3 | Proposed Techniques

Here, we briefly describe the MGGS techniques. We have divided this section into two sections. The first
section consists of the MGGS I and its convergence, and the second section concerns the MGGS II and
its convergence to solve the AVEs.

3.1 | MGGS I for AVE

Let we split the A matrix as

(2)

With
.
Where and are real parameters with and where Also, and are strictly
lower-triangular, diagonal and strictly upper-triangular parts of . Based on Eq. (2), we can write Eq. (1) as
follows:

(3)

By using the iterative scheme, we define MGGS I as follows:

,
(4)

Remark. If and , then the MGGS I tranform to the GGS approach [14].

The subsequent theorem specifies the convergence of MGGS I.

Theorem 1. Consider , which represents a splitting of with entries in the diagonal of
greater than one. If

(5)

Then the sequence obtained from the MGGS I converges to the unique solution of Eq. (1). 132
Proof. The proof of is easy and is skipped here (see Theorem 3 in [14]). Let us consider
and to be two distinct solutions of Eq. (1) for the purpose of determining the uniqueness of the
solution. By applying Eq. (4), we determine

(6)
(7)

Based on Eq. (6) and Eq. (7), we can get

Ali et al.| Com. Alg. Num. Dim. 1(4) (2022) 130-136

According to Lemma 1 and Eq. (5), we have

The result is a contradiction. Consequently, .

To determine convergence, we assume that is a unique solution of Eq. (1). Accordingly, from Eq. (6)
and

We deduce

Taking the infinity norm, we get;

This inequality implies that convergence of the approach can be achieved if Eq. (5) is fulfilled.

3.2 | MGGS II for AVE

This unit offers the MGGS II. Using Eq. (1) and Eq. (2), the MGGS II is expressed as follows:
 To demonstrate the convergenceiof MGGS II, we apply the following theorem.
133
Theorem 2. Consider which represents a splitting of with entries in the diagonal terms of
greater than 1. Then the sequence obtained from the MGGS II converges to the unique solution
of Eq. (1).

Proof. From Theorem 1, the uniqueness follows directly. In order to verify convergence, we can consider
the following:
The modification of the generalized gauss-seidel iteration techniques for absolute value equations

(8)

Since

As a result, Eq. (8) can be expressed as

Thus, solves the AVE (1) system.

4 | Numericaliexperiments

This part gives four examples in order to provide an overview of the performance of the newly formulated
techniques from three points of view:

I. The iteration steps (demonstrated as ‘Itr’).

II. The CPU time in seconds (denoted as ‘Time’).
III. The residual error (represented as ‘RSV’).

The term ‘RSV’ is described as follows:

Example 1. Let
Compute , where Here the starting iterate is zero vector, and we compare 134
both the proposed techniques with the SOR-like approach outlined in [12] (shown as SL) and the special
shift splitting approach described in [15] (shown as SS). Table 1 describes the results.

Table 1. A comparison of example 1 using , and .

Methods 1000 2000 3000 4000
SL Itr 18 18 18 18
Time 3.0156 13.1249 33.9104 65.1345
RSV
SS Iter 14 14 14
14
Time 2.8128 9.0954 17.3028 29.1644
RSV

Ali et al.| Com. Alg. Num. Dim. 1(4) (2022) 130-136

MGGS I Itr 9 9 9 9
Time 2.0823 5.3422 12.3724 21.2038
RSV 9.86 6.97 5.69 4.92
MGGS II Itr 7 7 7 7
Tim 0.9208 1.8422 2.3677 6.2190
RS 1.82 1.83 1.83 1.83

Example 2. Let and with

Where and . In example 2, the same starting guess and

stopping criteria as in [17] are used. We compare the suggested techniques with the AOR procedure [25]
and the MTS scheme [17]. We illustrate our results in Table 2:

Table 2. A comparison of example 2 using , .

Methods 100 400 900
1600
Itr 97 190 336 706
AOR Time 0.4721 2.8203 3.2174 6.3887
RSV9.80 9.61 9.73 9.84
Itr 88 157 250 386
MTS Time 0.4041 1.7953 3.0219 5.7626
RSV 8.91 9.65 9.18 9.56
Itr 46 70 91
110 MGGS I Time 0.2108 0.3844 0.9534 1.8747
RSV 8.83 8.94 8.17 8.57
Itr 23 34 45 55
MGGS II Time 0.1722 0.2344 0.7603 1.4200
RSV 5.08 9.96 7.81 7.51

Example 3. Let
 And , where Based on the same initial guess and ending criteria as explained
in [26]. We compare our newly developed techniques with those presented in [26] (signified as SI with
) as well as the SOR-like strategy offered in [12] (considered as SOR).

Table 3. A comparison of example 3 using , and

135 Methods 1000 2000 3000 4000
5000
SI Itr 13 13 14 14 14
Time 3.9928 8.8680 24.4031 51.3946 73.3394
RSV 6.04 8.54 2.33 2.69
SOR Itr 12 13 13 13 13
Time 1.5136 3.3817 6.1262 7.1715 9.5262
RSV 9.45 2.69 3.29 3.80
MGGS I Itr 11 11 11 11 11
Time 1.2718 2.3422 3.0127 4.4824 6.2511
RSV 9.45 2.69 3.80 3.80
MGGS II Itr 5 6 6 6 6
The modification of the generalized gauss-seidel iteration techniques for absolute value equations

Time 0.45532 0.5203 0.9376 1.2788 3.7826

RSV 9.29 1.22 1.47

Based on the results of Table 1 to Table 3, all tested approaches can compute the solution to Eq. (1).
Nevertheless, we notice that both the ‘Itr’ as well as the ‘time’ of the suggested techniques are better than
the other known procedures. To conclude, we can state that our presented techniques are practicable and
effective for AVEs.

5 | Conclusion

We have shown two MGGS techniques for determining AVE (1) and verified that the offered strategies
converge to the AVE (1) under proper selections of the parameters. Theoretical estimation and the
numerical study of the proposed algorithms indicate that the proposed techniques are applicable for
determining AVEs.

Conflict of Interset
The authors have no conflict of interest for this submission.

References

Mangasarian, O. L. (2014). Linear complementarity as absolute value equation solution. Optimization

letters, 8(4), 1529-1534. https://doi.org/10.1007/s11590-013-0656-z
Prokopyev, O. (2009). On equivalent reformulations for absolute value equations. Computational optimization
and applications, 44(3), 363-372. https://doi.org/10.1007/s10589-007-9158-1
Mezzadri, F. (2020). On the solution of general absolute value equations. Applied mathematics letters, 107,
106462. https://doi.org/10.1016/j.aml.2020.106462
Rohn*, J. (2004). A theorem of the alternatives for the equation Ax+ B|x|= b. Linear and multilinear algebra, 52(6),
421-426. https://doi.org/10.1080/0308108042000220686
Mangasarian, O. L., & Meyer, R. R. (2006). Absolute value equations. Linear algebra and its applications, 419(2-
3), 359-367. https://doi.org/10.1016/j.laa.2006.05.004
Ali, R., Khan, M. R., Abidi, A., Rasheed, S., & Galal, A. M. (2021). Application of PEST and PEHF in magneto-
Williamson nanofluid depending on the suction/injection. Case studies in thermal engineering, 27, 101329.
https://doi.org/10.1016/j.csite.2021.101329
Mao, X., Wang, X., Edalatpanah, S. A., & Fallah, M. (2019). The monomial preconditioned SSOR method for
linear complementarity problem. IEEE access, 7, 73649-73655. DOI: 10.1109/ACCESS.2019.2920485
Najafi, H. S., & Edalatpanah, S. A. (2012). A kind of symmetrical iterative methods to solve special class of
LCP (M, q). International journal of applied mathematics and applications, 4(2), 183-189.
Ullah, I., Ali, R., Nawab, H., Uddin, I., Muhammad, T., Khan, I., & Nisar, K. S. (2022). Theoretical
analysis of activation energy effect on Prandtl–Eyring nanoliquid flow subject to melting
condition. Journal of non-equilibrium thermodynamics, 47(1), 1-12. https://doi.org/10.1515/jnet-2020-
0092
Najafi, H. S., & Edalatpanah, S. A. (2013). SOR-like methods for non-Hermitian positive definite
linear complementarity problems. Advanced modeling and optimization, 15(3), 697-704. 136
https://camo.ici.ro/journal/vol15/v15c10.pdf
Salkuyeh, D. K. (2014). The Picard–HSS iteration method for absolute value equations. Optimization
letters, 8(8), 2191-2202. https://doi.org/10.1007/s11590-014-0727-9
Ke, Y. F., & Ma, C. F. (2017). SOR-like iteration method for solving absolute value equations. Applied
mathematics and computation, 311, 195-202. https://doi.org/10.1016/j.amc.2017.05.035
Chen, C., Yu, D., & Han, D. (2020). Optimal parameter for the SOR-like iteration method for solving the
system of absolute value equations. Retrieved from https://doi.org/10.48550/arXiv.2001.05781
Edalatpour, V., Hezari, D., & Salkuyeh, D. K. (2017). A generalization of the Gauss–Seidel iteration
method for solving absolute value equations. Applied mathematics and computation, 293, 156-167.
https://doi.org/10.1016/j.amc.2016.08.020
Wu, S., & Li, C. (2020). A special shift splitting iteration method for absolute value equation. AIMS

Ali et al.| Com. Alg. Num. Dim. 1(4) (2022) 130-136

mathematics, 5(5), 5171-5183. https://www.aimspress.com/fileOther/PDF/Math/math-05-05-332.pdf
Moosaei, H., Ketabchi, S., & Jafari, H. (2015). Minimum norm solution of the absolute value
equations via simulated annealing algorithm. Afrika matematika, 26(7), 1221-1228.
https://doi.org/10.1007/s13370-014-0281-8
Fakharzadeh, A., & Shams, N. N. (2020). An efficient algorithm for solving absolute value
equations. Journal of mathematical extension, 15. https://ijmex.com/index.php/ijmex/article/view/1393
Ali, R., Kejia, P., & Ali, A. (2020). Two generalized successive overrelaxation methods for solving
absolute value equations. Mathematics, 40(4), 44-55.
Ali, R., Khan, I., Ali, A., & Mohamed, A. (2022). Two new generalized iteration methods for solving
absolute value equations using M-matrix. AIMS mathematics, 7(5), 8176-8187.
http://www.aimspress.com/aimspress-data/math/2022/5/PDF/math-07-05-455.pdf
Ali, R. (2022). The solution of absolute value equations using two new generalized gauss-seidel
iteration methods. Computational and mathematical methods, 2022.
https://doi.org/10.1155/2022/4266576
Ali, R., Ali, A., Alam, M. M., & Mohamed, A. (2022). Numerical solution of the absolute value
equations using two matrix splitting fixed point iteration methods. Journal of function spaces, 2022.
https://doi.org/10.1155/2022/7934796
Ali, R., Ali, A., & Iqbal, S. (2022). Iterative methods for solving absolute value equations. The journal
of mathematics and computer science, 26, 322-329.
Ali, R. (2022). Numerical solution of the absolute value equation using modified iteration
methods. Computational and mathematical methods, 2022.
https://downloads.hindawi.com/journals/cmm/2022/2828457.pdf
Khan, A., Iqbal, J., Akgül, A., Ali, R., Du, Y., Hussain, A., ... & Vijayakumar, V. (2022). A newton-
type technique for solving absolute value equations. Alexandria engineering journal.
https://doi.org/10.1016/j.aej.2022.08.052
Li, C. X. (2017). A preconditioned AOR iterative method for the absolute value
equations. International journal of computational methods, 14(02), 1750016.
https://doi.org/10.1142/S0219876217500165
Guo, P., Wu, S. L., & Li, C. X. (2019). On the SOR-like iteration method for solving absolute value
equations. Applied mathematics letters, 97, 107-113. https://doi.org/10.1016/j.aml.2019.03.033