Important Announcement
PubHTML5 Scheduled Server Maintenance on (GMT) Sunday, June 26th, 2:00 am - 8:00 am.
PubHTML5 site will be inoperative during the times indicated!

Home Explore วารสาร "โครงการสัมมนาวิชาการและนำเสนอผลงานด้านคณิตศาสตร์ ครั้งที่ 7"

วารสาร "โครงการสัมมนาวิชาการและนำเสนอผลงานด้านคณิตศาสตร์ ครั้งที่ 7"

Published by อนุพงษ์ ดาปัง, 2022-01-20 03:05:13

Description: Proceeding

Search

Read the Text Version

Type of the Article: Research RE-AP 04 33 ALipnpelaicr aOtiDoEns owfitMhoVhaarniadbTleraCnosefoffirmciefnotrsSecond-Order Kitsada Warin, Kanokon Suphason and Somthawin Khunkhet* Department of Mathematics, Faculty of Science Ubon Ratchathani Rajabhat University Ubon Ratchathani, 34000, Thailand *Correspondence: [email protected] Abstract Mohand transform of derivative is derived, and its applicability demonstrated using six formulars. In this paper, we find the particuler solutions of these equations. Keywords: Mohand transform, Second-Order Linear ODEs 1. Introduction The differential equation have played a central role in every aspect of applied mathematics for every long time, their importance has increased father. Thus investi- gation and analysis of differential equation cruising in applications led to many deep mathematical problem, therefore, there are so many different technignes in order to solve. Mohand transform is derived from the classical Fourier integral. Based on the mathematical simplicity of the Mohand transform and its fundamental properties. Mo- hand transform was introduced by Mohand Mahgoub to facilitate the process of solving ordinary and partial differential equations in the time domain [1]. Typically, Fourier, Laplace, Elzaki, Aboodh, Kamal and Sumudu transforms are the convenient mathemat- ical tools for solving differential equations [2–5]. In this paper, we derive the formulale for Mohand transform of derivative and apply them in solving some types of initial value problem. 2. Preliminaries Definition 2.1. The M∫oh∞and transform of the function f(t) for t 0 was given by M {f (t)} = v2 e−vtf (t)dt = R(v). The variable v in0 this transform is used to factor the variable t the argument of the function f, and for all valnes of v for which the improper integral converges. โครงการสมั มนาวิชาการและนำเสนอผลงานดานคณิตศาสตร ครง้ั ที่ 7 มหาวิทยาลยั ราชภัฏนครสวรรค

34 Kitsada Warin et al. Theorem 2.2. [1, 8] Let R(v) be the Mohand transform of f(t) [i.e., M{f(t)} = R(v)] : i) M {f ′(t)} = vR(v) − v2f (0) ii) M {f ′′(t)} = v2R(v) − v3f (0) − v2f ′(0) iii) n∑−1 M {f (n)(t)} = vnR(v) − vn−k+1f (k)(0) k=0 iv) M {tf (t)} = 2 R(v) − d R(v) v dv Table 1. Mohand transform of some elementary function [6]: f (t) R(v) = M {f (t)} 1v t 1 t2 2! tn, n ∈ N tn, n > −1 v eat n! sin at vn−1 cos at n+1 sinh at Γ vn−1 cosh at v2 v−a av2 v2 + a2 v3 v2 + a2 av2 v2 − a2 v3 v2 − a2 3. Main Theorems Theorem 3.1. Let R(v) be the Mohand transform of f(t) [i.e., M{f(t)} = R(v)] . Then we have i) M {tf ′(t)} = −v d R(v) + R(v) dv ii) M {tf ′′(t)} = −v2 d R(v) + v2f (0) dv iii) M {t2f (t)} = d2 4d 6 dv2 R(v) − R(v) + v2 R(v) v dv iv) M {t2f ′(t)} = v d2 R(v) − 2 d R(v) + 2 dv2 dv R(v) v v) M {t2f ′′(t)} = v2 d2 R(v) dv2 โครงการสมั มนาวิชาการและนำเสนอผลงานดานคณิตศาสตร ครง้ั ที่ 7 มหาวทิ ยาลัยราชภัฏนครสวรรค

Application of Mohand Transform for Second-Order Linear ODEs with… 35 Proof. (i) Let g(t) = f′(t). Substitute g(t) into M{tf(t)}, we get M {tf ′(t)} = M {tg(t)} = 2 M {g(t)} − M ′{g(t)} v = 2 M {f ′(t)} − M ′{f ′(t)} [ v vR(v) [ ] ] = 2 vR(v) − v2f (0) − d − v2f (0) v dv Therefore, M {tf ′(t)} = −v d R(v) + R(v). dv (ii) From M{tf′(t)} = R(v) − vR′(v), and let g(t) = f′(t) and g′(t) = f′′(t). Substitute g′(t) into M{tf′′(t)}, and using (i), we get Then M {tf ′′(t)} = M {tg′(t)} = M {f ′(t)} − vM ′{f ′(t)}. = [ − v2f ] − v d [ − v2 f ] vR(v) (0) vR(v) (0) dv Therefore, M {tf ′′(t)} = −v2 d R(v) + v2f (0). dv (iii) Let g(t) = tf(t). Substitute g(t) into M{t2f(t)} and using theorem 2.2 (iv), we get M {t2f (t)} = M {tg(t)} = 2 M {tf (t)} − d M {tf (t)} v[ dv ] 2 2 R(v) − d d [ 2 R(v) − d ] = R(v) R(v) v v dv − dv v dv Thus, M {t2f (t)} = d2 − 4 d + 6 R(v). dv2 R(v) v R(v) v2 dv (iv) Let g(t) = tf′(t). Substitute into M{t2f′(t)} and using theorem 2.2 (iv) and (i), we get M {t2f ′(t)} = M {tg(t)} = 2 M {g(t)} − M ′{g(t)} v = 2 M {tf ′(t)} − M ′{tf ′(t)} v [ vR′(v)] [ vR′(v)] = 2 R(v) − − d R(v) − v dv Thus, M {t2f ′(t)} = v d2 R(v) − d R(v) + 2 R(v). dv2 2 v dv โครงการสัมมนาวิชาการและนำเสนอผลงานดานคณติ ศาสตร ครง้ั ที่ 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

36 Kitsada Warin et al. (v) Let g(t) = tf′′(t). Substitute into M{t2f′′(t)} and using theorem 2.2 (iv) and (ii), we get M {t2f ′′(t)} = M {tg(t)} = 2 M {g(t)} − M ′{g(t)} v = 2 M {tf ′′(t)} − d M {tf ′′(t)} v dv [−v2R′(v) ] [−v2R′′(v) ] = 2 + v2f (0) − d + v2f ′ (0) v dv Thus, M {t2f ′′(t)} = v2 d2 R(v). dv2 4. Application of Mohand transform of ODEs with variable co- efficients In this section, we apply the above theorem to find Mohand transform for some diffierential equation. Example 4.1. Solve the differential equation: ;t2y′′ + 4ty′ − 4y = 12t2 y(0) = 0, y′(0) = 0. By using Mohand transform onto equation and the theorem 3.1, we have v2R′′(v) + 4 [−vR′ + ] − 4R(v) = 24 R(v) v R′′(v) − vR′(v) = 24 v3 . Let G(v) = R′(v), and G′(v) = R′′(v). Then, this equation above rewrite as follow; G′(v) − 4 = 24 G(v) v3 . v This is linear equation, then we solve this equation to G(v). Thus, G(v) = −4 + C1. Since then we get v2 G(v) = R′(v), 4 R(v) = v + C1v + C2. If C1, C2 = 0, then by using the inverse Mohand transform, we get y(t) = 2t2. โครงการสมั มนาวิชาการและนำเสนอผลงานดานคณิตศาสตร ครง้ั ท่ี 7 มหาวทิ ยาลัยราชภัฏนครสวรรค

Application of Mohand Transform for Second-Order Linear ODEs with… 37 Example 4.2. [7] Sovle the differential equation, ty′′ + (1 − 2t)y′ − 2y = 0 ; y(0) = 1, y′(0) = 2 By using Mohand transform into equation and the theorem 3.1, we have −v2R′ + 2vR′ + vR(v) − 4R(v) = 0. We can write equation in the from, R′(v) − 2 − v 1 )R(v) = 0. ( − 2 v This is a linear differential equation for unknow function R(v). Then, we solve this equation for R(v) get R(v) = Cv2 . v−2 By using the inverse Mohand transform we obtain the solution in the form y(t) = Ce2t. Now we determine value of C by using given initial value y(0) = 1, then C = 1. Therefore, the solution of this equation is y(t) = e2t. 5. Conclusion In this paper, we derive five formulars for Mohand transform of derivative and apply them to the solution of ordinary differential equation with variable coefficients has been demonstrated. References [1] Mohand M. Abdelrahim Mahgoub,“The New Integral Transform Mohand Trans- form,” Advances in Theoretical and Applied Mathematics ISSN 0973-4554 Volume 12, pp. 113-120, Number 2 2017. [2] Tarig M. Elzaki, “ The New Integral Transform Elzaki Transform ” Global Journal of Pure and Applied Mathematics, ISSN 0973-1768, pp. 57-64, Number 1, 2011. โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร ครั้งที่ 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

38 Kitsada Warin et al. [3] Tarig M. Elzaki and Salih M. Elzaki, “ Application of New Transform Elzaki Transform to Partial Differential Equations,” Global Journal of Pure and Applied Mathematics, ISSN 0973-1768, pp. 65-70, Number 1, 2011. [4] Phil Dyke, “An Introduction to Laplace Transforms and Fourier Series,” Springer Undergraduate Mathematics Series, November 12, 2002. [5] Abdelilah Kamal. H. Sedeeg, “ The New Integral Transform Kamal Transform” ,Ad- vances in Theoretical and Applied Mathematics, Vol.11, No.4, pp. 451-458, 2016. [6] Sudhanshu Aggarwal, Nidhi Sharma and Raman Chauhan, “Solution of Linear Volterra Integral Equations of Second Kind Using Mohand Transform,” International Journal of Research in Advent Technology, vol. 6, November 2018. [7] Tarig. M. Elzaki1 and Salih M. Ezaki2, “On the ELzaki Transform and Ordinary Differ- ential Equation with Variable Coefficients ,” Advances in Theoretical and Applied Mathematics ISSN 0973-4554 Volume 6, pp. 41–46, Number 1 2011. [8] Sudhanshu Aggarwal1, Anjana Rani Gupta and Deepak Kumar, “Mohand Transform of Error Function ,” International Journal of Research in Advent Technology E-ISSN: 2321-9637 Vol.7, No.5, May 2019. โครงการสัมมนาวิชาการและนำเสนอผลงานดานคณติ ศาสตร ครัง้ ที่ 7 มหาวิทยาลยั ราชภฏั นครสวรรค

Type of the Article: Research RE-AP 05 39 EKqamuaatlioTnrsawnsitfhorVmarifaobrleSoCloveinfgficOierndtisnary Differential Nantida Yenchai, Prakairung Laokiom and Somthawin Khunkhet* Department of Mathematics, Faculty of Science Ubon Ratchathani Rajabhat University Ubon Ratchathani, 34000, Thailand *Correspondence: [email protected] Abstract Kamal transform of derivative is derived, and its applicability demonstrated using six for- mulars. In this paper, we find the particuler solutions of these equations. Keywords: Kamal transform, ordinary differential equations 1. Introduction An integral transform is a particular kind of mathematical operator, and integral transform is any transform T of following form T (f (t)) = ∫ t2 K (t, u)f (t)dt t1 The input of this transform is a function f(t), and the output is another T (f(t)). There are numerous useful integral transforms, each is specified by a choice of the function K of two variables, the kernel function or nucleus of the transform. The properties of integral transforms vary widely, they have some properties in common. For example, every integral transform is a linear operator, since the integral is a linear operator, and in fact if kernel is allowed to be a generalized function then all linear operators are integral transforms. Recently some applications for Sumudu transform are used to solving different types of differential equations [1]. The Laplace transform can be used to solve differential equations [2]. The Elzaki transform can be used to solve ordinary differention equation with variables coefficients [3, 4]. Kamal Transform is derived from the classical Fourier integral. Based on the mathematical simplicity of the Kamal transform and its fundamental properties, Kamal transform was introduced by Abdelilah Kamal to facilitate the process of solving ordi- nary and partial differential equations in the time domain [5]. Typically, Fourier, Laplace , Sumudu , Elzaki , Aboodh and Mahgoub transforms are the convenient mathematical tools for solving differential equations. Also, Kamal transform and some of its funda- โครงการสมั มนาวชิ าการและนำเสนอผลงานดา นคณิตศาสตร ครง้ั ที่ 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

40 Nantida Yenchai et al. mental properties can be used to solve differential equations [1, 4]. In this paper, we derive the formular for Kamal transform of derivative and apply them in solving some types of initial value problem. 2. Preliminaries Definition 2.1. The Kamal transform [5]. Given a function f(t) defined for all t ≥ 0 , the Kamal transform of f is the function G defind as follow: ∫∞ −t K[f (t)] = G(v) = f (t)e v dt, 0 for all values of v for which the improper integral converges. Theorem 2.2. [5] Let G(v) be the Kamal transform of f [ = ] : (t) K[f (t)] G(v) i) K [ (n)(t)] = 1 − 1 − 1 F ′ (0) . . . − F n−1(0), f vn G(v) vn−1 F (0) vn−2 ii) K[tf (t)] = v2 d G(v), dv iii) K [t2 f (t)] = v4 d2 G(v) + 2v3 d G(v), dv2 dv iv) K[tf ′(t)] = v d G(v) − G(v), dv v) K [t2 f ′(t)] = v3 d2 G(v), dv2 vi) K[tf ′′(t)] = d G(v) − 2 + f (0), G(v) dv v vii) K[t2f ′′(t)] = d2 G(v) − 2v d G(v). dv2 dv Table 1. Kamal transform of some elementary function [6]: f (t) K(f (t)) = G(v) 1 t v t2 v2 2!v3 tn, nϵN n!vn+1 tn, n > −1 Γ(n + 1)vn+1 v eat 1 − av โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณิตศาสตร ครงั้ ท่ี 7 มหาวิทยาลัยราชภัฏนครสวรรค

Kamal Transform for Solving Ordinary Differential Equations with … 41 sin at av2 cos at sinh at 1 + a2v2 cosh at v J0(αt) 1 + a2v2 J1(αt) av2 J2(αt) 1 − a2v2 v 1 − a2v2 ] √v [ (1 + α2v2) 1 1− √ 1 [α (1 √+ α2v2) α2v2 +√2 − 2 (1 + α2v2) ] v (1 + α2v2) 1 α2 3. Main Theorems Theorem 3.1. Let G(v) = K[f(t)] , then: (i) K[t3f (t)] = v8G′′′(v) + 6v7G′′(v) + 6v6G′(v) (ii) K[t3f ′(t)] = v5G′′′(v) + 3v4G′′(v) (iii) K[t3f ′′(t)] = v2G′′′(v) − 2v3G′′(v) − 2v2G′(v) (iv) K[tf ′′′(t)] = 1 G′(v) − 3 2 f (0) + f ′(0) v v2 G(v) + v (v) K[t2f ′′′(t)] = vG′′(v) − 4G′(v) + 6 G(v) − 2 v v2 f (0) (vi) d3 [ 1 1 ] d2 [ 1 ] dv3 v2 v ′(0) +3v2 dv2 v2 1 f (0) − f ′(0) . K[t3f ′′′(t)] = v5 G(v) − f (0) − f G(v) − v Proof. Let G(v) = K[f(t)]. (i) By theorem 2.2 (iv) and (v), we obtain K[t3f (t)] = K[tg(t)] , g(t) = t2f (t) = v2 d K[t2f (t)] dv = v2 d [v4G′′(v) + 2v3G′(v)] dv = v4 [v4G′′′(v) + 6v3G′′(v) + 6v2G′(v)] = v8G′′′(v) + 6v7G′′(v) + 6v6G′(v). โครงการสมั มนาวชิ าการและนำเสนอผลงานดานคณิตศาสตร ครงั้ ท่ี 7 มหาวทิ ยาลัยราชภัฏนครสวรรค

42 Nantida Yenchai et al. (ii) Let g(t) = t2f′(t), then substitute in (i) and using theorem 2.2 (iv) and (vii), we get K[t3f ′(t)] = K[tg(t)] = v2 d K[t2f ′(t)] dv = v2 d [ d2 ] v3 dv2 G(v) dv = v5 d3 G(v) + 3v4 d2 G(v). dv3 dv2 (iii) By theorem 2.2 (iv) and (ix), we get K[t3f ′′(t)] = K[tg(t)] , g(t) = t2f ′′(t) = v2 d g(t) dv = v2 d K[t2f ′′(t) dv d [ d2 ] 2vd = v2 G(v) − G(v) dv dv dv = v2G′′′(v) − v2 [2vG′′(v) + 2G′(v)] = v2G′′′(v) − 2v3G′′(v) − 2v2G′(v). (iv) Let g(t) = f′′(t) and g′(t) = f ,′′′(t) we find that by using theorem 2.2 (vi), K[tf ′′′(t)] = K[tg′(t)] = v2 d K[f ′′′(t)] dv [ ] = v2 d 1 1 1 f ′(0) dv v3 G(v) − v2 f (0) − v − f ′′(0) = v2 [ v3G′(v) − 3v2G(v) + 2V f ′(0) + f ′(0) ] v6 v4 v2 = 1 G′(v) − 3 + 2 + f ′(0). v v2 G(v) f (0) v (v) Let g(t) = tf′′′(t). Thus, K[t2f ′′′(t)] = K[tg(t)] = v2 d K[tf ′′′(t)] dv [ ] = v2 d 1 G′(v) 3 2 dv v − v2 G(v) + v f (0) + f ′(0) = vG′′(v) − G′(v) − 3G′(v) + 6 − 2 f (0) G(v) v2 v = vG′′(v) − 4G′(v) + 6 − 2 f (0). G(v) v2 v โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร คร้งั ที่ 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

Kamal Transform for Solving Ordinary Differential Equations with … 43 (vi) Let g(t) = f′′(t) and g′(t) = f′′′(t). Thus, K[t3f ′′′(t)] = K[t3g′(t)] = v5 d3 K [g (t)] + 3v4 d2 K [g (t)] dv3 dv2 = v5 d3 ′′(t)] + 3v4 d2 K [f ′′ (t)] = v5 ddv33[K1[f − dv2 ] d2 [ 1 ] dv3 v2 G(v) 1 − f ′(0) + 3v4 dv2 v2 G(v) − 1 − f ′(0) . f (0) f (0) v v 4. Application of Kamal transform of ODEs with variable coef- ficients In this section, we apply the above theorem to find Kamal transform for some differential equations. Example 4.1 To solve the differential equation, ty′′ + (1 − 2t)y′ − 2y = 1 ; y(0) = 1, y′(0) = 3. By using Kamal transform into equation and the theorem 2.2, we have G′(v) − 2 + 1 − 2vG′(v) = v. G(v) G(v) vv We can write equation in the from, () 12 G′(v) − v + 1 − 2v G(v) = v this is a linear differential equation for unknow function G(v). Then, we solve this equation for G(v) to get G(v) = −v + 2(1 v 2v) + 1 cv 2 − − 2v . By using the inverse Kamal transform we obtain the solution in the form y(t) = − 1 + 1 e2t + ce2t. 22 We now determine value of c by using given initial value y(0) = 1, then c = 1. Therefore, the solution of this equation is y(t) = − 1 + 3 e2t. 22 โครงการสัมมนาวิชาการและนำเสนอผลงานดานคณิตศาสตร ครั้งท่ี 7 มหาวิทยาลยั ราชภัฏนครสวรรค

44 Nantida Yenchai et al. Example 4.2 consider the Bessel equation, ty′′ + y′ + α2ty = 0 , y(0) = 1 , y′(0) = 0. Solution. Applying Kamal transform yields, G′(v) − 2 + y(0) + 1 − y(0) + α2v2G′(v) = 0 G(v) G(v) vv (1 + α2v2)G′(v) − 1 = 0 G(v) v G′(v) − v(1 1 = 0 + α2v2) G(v) ( α2v ) G′(v) − 1 − 1 + α2v2 G(v) = 0 v This is a linear differential equation, we get G(v) = √ cv , 1 + α2v2 where c is a constant of integration. By the inverse Kamal transform, we obtain y(t) = cJ0(αt). Now, we determinevalue of c by using initial value y(0) = 1, then c = 1. Therefor, the solution of this equation is y(t) = J0(αt). 5. Conclusion In this paper, we derive six formulars for Kamal transform of derivative and have applied them to the solution of ordinary differential equation with variable coefficients. References [1] G. K. Watugala, “Sumudu Transform: a new integral transform to solve differential equations and control engineering problems,” International Journal of Mathemat- ical Education in Science and Technology 24. 35-43, (2006). [2] Phil Dyke, “An Introduction to Laplace Transforms and Fourier Series,” Springer Undergraduate Mathematics Series, 1-12, 2002. [3] Tarig M. Elzaki, “The New Integral Transform Elzaki Transform .” Global Journal of Pure and Applied Mathematics, 7. 57-64, (2011). โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร คร้งั ท่ี 7 มหาวิทยาลยั ราชภฏั นครสวรรค

Kamal Transform for Solving Ordinary Differential Equations with … 45 [4] Tarig M. Elzaki and Salih M. Elzaki, “ Application of New Transform “Elzaki Trans- form” to Partial Differential Equations,” Global Journal of Pure and Applied Math- ematics ISSN 0973-1768, 7, 65-70.(2011). [5] Abdelilah Kamal . H. Sedeeg, “The New Integral Transform “Kamal Transform” Advances in Theoretical and Applied Mathematics ISSN 0973-4554 11, 451-458. (2016). [6] A.Sudhanshu, “Kamal Transform of Bessel’s Function,” International Journl of Re- search and Innovation in Applied Science (IJRIAS),3,1-4.(2018) [7] 1A.S.Khan, 2I.I.Khan, 3M.M. Janolkar, “SOLUTION OF ORDINARY DIFFERENTIAL EQUA- TIONS WITH VARIABLE COEFFICIENTS USING KAMAL TRANSFORM,” International Journal of Scientific Research and Review,7,173-178. (2018). โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร คร้งั ท่ี 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

46 Nantida Yenchai et al. โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร คร้งั ท่ี 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

Type of the Article: Research RE-PU 01 47 On the Ternary Quadratic Equation (n + 1)(x2 + y2) − (2n − 2)xy = 4nz2 Darunee Tharawngsa, Supaporn Sriracha and Supamit Pimsri* Department of Mathematics, Faculty of Science, Ubon Ratchathani Rajabhat University *Correspondence: [email protected] Abstract The solution of ternary quadratic diophantine equation (n+1)(x2 +y2)−(2n−2)xy = 4nz2 by finding the solution that is non-zero five different patterns of integral solutions. A few interesting relations between the solutions and special numbers. Keywords: Ternary, Quadratic, Diophantine equation. 1. Introduction The ternary quadratic diophantine equation there are many forms. There are many ways to find the solutions of equation. The equation on the form A(x2 + y2) − Bxy = Cz2 being attracted by many researchers such as [1–4]. In this paper, we present different pattern of non-zero distinct solutions of ternary quadratic equation (n + 1)(x2 + y2) − (2n − 2)xy = 4nz2. 2. Notations [] =n (n − 1)(m − 2) , Polygonal number of rank n with sides m. Tm,n 1+ 2 Obln = n(n + 1) , Oblong number of rank n. Gn = 2n − 1 , Gnomonic number of rank n. 3. Method of Analysis The ternary quadratic diophantine equation to be solved for its non-zero distinct integral solution is (n + 1)(x2 + y2) − (2n − 2)xy = 4nz2. (3.1) The substitution of linear transformations (u ̸= v ≠ 0) x = u + v; y = u − v. (3.2) โครงการสัมมนาวิชาการและนำเสนอผลงานดานคณติ ศาสตร ครั้งท่ี 7 มหาวิทยาลัยราชภฏั นครสวรรค

48 Darunee Tharawngsa et al. Using (3.2) in (3.1), it is written as u2 + nv2 = nz2. (3.3) Different patterns of solutions of (3.1) are illustrated below : Pattern I We can write (3.3) as u2 = n(z − v)(z + v). (3.4) Case 1 Equation (3.4) can also be rewritten as u = n(z − v) = A (3.5) . (3.6) z+v u B This is equivalent to the system of double equations  Bu − Av − Az = 0,  Au + nBv − nBz = 0.  This is satisfied by u = 2nAB, v = −A2 + nB2, z = A2 + nB2. Hence in view of (3.2), the corresponding solutions of (3.1) are given by x = x(A, B) = −A2 + nB2 + 2nAB, y = y(A, B) = A2 − nB2 + 2nAB, z = z(A, B) = A2 + nB2. Properties (1) x(1, B) + z(1, B) = 4nT3,A . (2) x(1, B) + y(1, B) − 2nOblB + 2nBGB ≡ 0 (mod n). (3) x(A, B) − y(A, B) + 2z(A, B) ≡ 0 (mod n). โครงการสมั มนาวชิ าการและนำเสนอผลงานดานคณิตศาสตร คร้งั ท่ี 7 มหาวิทยาลัยราชภฏั นครสวรรค

On the Ternary Quadratic Equation (n + 1)(x2 + y2) − (2n − 2)xy = 4nz2 49 Case 2 (3.7) Equation (3.4) can also be rewritten as (3.8) u z+v A (3.9) n(z − v) = =. (3.10) u B This is equivalent to the system of double equations  Bu + nAv − nAz = 0,  Au − Bv − Bz = 0.  This is satisfied by u = 2nAB, v = nA2 − B2, z = nA2 + B2. Hence in view of (3.2), the corresponding solutions of (3.1) are given by x = x(A, B) = nA2 − B2 + 2nAB, y = y(A, B) = −nA2 + B2 + 2nAB, z = z(A, B) = nA2 + B2. Properties (1) x(A, 1) + y(A, 1) − 2A = 2AGA. (2) x(A, 1) + z(A, 1) − 4nT3,A = 0. (3) y(A, 1) − z(A, 1) ≡ 0 (mod n). Case 3 Equation (3.4) can also be rewritten as u z−v A = =. n(z + v) u B This is equivalent to the system of double equations  Bu − nAv − nAz = 0,  Au + Bv − Bz = 0.  โครงการสัมมนาวิชาการและนำเสนอผลงานดานคณิตศาสตร คร้งั ที่ 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

50 Darunee Tharawngsa et al. This is satisfied by u = 2nAB, v = −nA2 + B2, z = nA2 + B2. Hence in view of (3.2), the corresponding solutions of (3.1) are given by x = x(A, B) = −nA2 + B2 + 2nAB, y = y(A, B) = nA2 − B2 + 2nAB, z = z(A, B) = nA2 + B2. Properties (1) z(A, B) − x(A, B) ≡ 0 (mod n). (2) y(A, 1) + z(A, 1) − 4nT3,A = 0. (3) y(A, B) − x(A, B) − 2z(A, B) ≡ 0 (mod n). Case 4 Equation (3.4) can also be rewritten as u n(z + v) A (3.11) z−v = =. (3.12) u B This is equivalent to the system of double equations  Bu + Av − Az = 0,  Au − nBv − nBz = 0.  This is satisfied by u = 2nAB, v = A2 − nB2, z = A2 + nB2. Hence in view of (3.2), the corresponding solutions of (3.1) are given by x = x(A, B) = A2 − nB2 + 2nAB, y = y(A, B) = −A2 + nB2 + 2nAB, z = z(A, B) = A2 + nB2. โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณิตศาสตร คร้งั ที่ 7 มหาวิทยาลัยราชภัฏนครสวรรค

On the Ternary Quadratic Equation (n + 1)(x2 + y2) − (2n − 2)xy = 4nz2 51 Properties (1) x(1, B) + y(1, B) − 2B = 2BGn. (2) y(1, B) + z(1, B) = 2nOblB. (3) x(A, (A + 1)) + y(A, (A + 1)) = 4nOblA. Pattern II Assume z = A2 + nB2. (3.13) (3.14) Where A, B are non-zero distinct integers. Write n as √√ n = (i n)(−i n). Use (3.13) and (3.14) in (3.3) and employing the method of factorization. Define (u + √ = √ + √ )2. (3.15) i nv) (i n)(A i nB Equating real and imaginary parts in (3.15) and using (3.2) ,the values of x and y satisfies (3.1) are given by x = x(A, B) = A2 − nB2 − 2nAB, (3.16) y = y(A, B) = −A2 + nB2 − 2nAB. (3.17) Thus (3.16), (3.17) and (3.13) represents non-zero distinct integral solutions of (3.1) in two parameters. Properties (1) y(1, B) + z(1, B) − 2nBGB ≡ 0 (mod n). (2) x(1, B) − z(1, B) + 2nOblB = 0. (3) y(1, (B + 1)) + z(1, (B + 1)) − 2nOblB = 0. 4. Conclusion We have discovered that the solution of ternary quadratic diophantine equation (n + 1)(x2 + y2) − (2n − 2)xy = 4nz2 five different patterns of integer solutions. A few interesting relations between the solutions and special numbers. โครงการสมั มนาวชิ าการและนำเสนอผลงานดา นคณิตศาสตร ครงั้ ที่ 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

52 Darunee Tharawngsa et al. References [1] K. Meena, S. Vidhyalaksmi, E. Bhuraneswari and R. Presenna, “On Ternary Quadratic Diophantine Equation 5(x2 +y2)−6xy = 20z2 ,” International Journal of Advancel Scienfific Research, vol. 1, no. 2, pp. 59-61, 2016. [2] M. A. Gropalan, S. Vidhyalaksmi, A. kavitha and D. Mary Madona, “On Ternary Quadratic Diophantine Equation 3(x2 + y2) − 2xy = 4z2 ,” International Journal of Engineering Science and Management., vol. 5, no. 2, pp. 11-18, 2015. [3] M. A. Gropalan, S. Vidhyalaksmi and T. R. Usha Rani, “On Ternary Quadratic Dio- phantine Equation 6(x2 + y2) − 8xy = 21z2 ,” Sch. J. Eng Tech., vol. 2, no. 2A, pp. 108-112, 2014. [4] M. A. Gropalan, S. Vidhyalaksmi, and U. K. Rajakshmi, “On Ternary Quadratic Dio- phantine Equation 5(x2 + y2) − 6xy = 196z2 ,” International Journal of Engineering Science and Management., vol. 3, no. 5, pp. 1-10, 2017. โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครง้ั ที่ 7 มหาวทิ ยาลัยราชภัฏนครสวรรค

Type of the Article: Research RE-PU 02 53 SFuonlucttiioonnaalnEdquSatatiboinlitoyfoPfeaxnidenr-dTiympeensional Cauchy Sutthiphong Moonmee and Choodech Srisawat* Department of Mathematics, Faculty of Science, Udon Thani Rajabhat University, Udon Thani 41000 Thailand *Correspondence: [email protected] Abstract In this paper, we study the general solution and prove the stability of the n-dimensional Cauchy functional equation of Pexider type, nn g xi = fi(xi), i=1 i=1 when g, f1, . . . , fn are functions from an abelian group to a Banach space. Keywords: stability, Cauchy pexider, general solution 1. Introduction In 1940, S.M. Ulam [4] proposed the stability problem of the additive functional equation, f (x + y) = f (x) + f (y), when f is a mapping from a group G1 to a group G2. In 1941, D.H. Hyers [1] provided a first affirmative partial answer to Ulam’s prob- lem for the case where G1 and G2 are Banach spaces. Let f : E → E′ be a function from a Banach space E to a Banach space E′ and let ε > 0 satisfy ∥f (x + y) − f (x) − f (y)∥ ≤ ε for all x, y ∈ E. Then there exists a unique additive function φ : E → E′ satisfying the inequality ∥f (x) − φ(x)∥ ≤ ε and the function φ is given by φ(x) = lim 2−nf (2nx). n→∞ โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครั้งที่ 7 มหาวิทยาลัยราชภฏั นครสวรรค

54 Sutthiphong Moonmee and Choodech Srisawat In 1978, Th. M. Rassias [5] has expanded the stability problem such that f : E → E′ satisfies the inequality ∥f (x + y) − f (x) − f (y)∥ ≤ θ ∥x∥p + ∥y∥p for all x, y ∈ E and for some constants θ ≥ 0 and 0 ≤ p < 1. Then there exists a unique additive function A : E → E′ satisfying the inequality ∥f (x) − A(x)∥ ≤ 2 2θ ∥x∥p. − 2p Next, M.S. Moslehian [2] proved the generalized Hyers-Ulam stability of the Cauchy functional equation of Pexider type, f1(x + y) = f2(x) + f3(y), when f1, f2, f3 are functions from an abelian group to a Banach space. In 2012, Renu Chugh and Ashish [3] proved the Hyers-Ulam-Rassias stability of the pexiderized Cauchy functional equation, f (x + y + z) = g(x) + h(y) + j(z), when f, g, h, j are functions from an abelian group to a Banach space. In this paper, we investigate the solution and the stability of an n-dimensional Cauchy functional equation of Pexider type, nn g xi = fi(xi), i=1 i=1 when g, f1, f2, · · · , fn are functions from an abelian group to a Banach space. 2. Preliminaries Definition 2.1. [7] Given a non-emptry set G and a binary operation “ + ”. (G, +) is a group if 1. a + (b + c) = (a + b) + c for all a, b, c ∈ G, 2. there is an element 0 ∈ G such that a + 0 = 0 + a = a for all a ∈ G and 3. if a ∈ G, then there is an element −a ∈ G such that a + (−a) = (−a) + a = 0. Definition 2.2. [7] A group (G, +) is called an abelian group if the binary operation is commutative, i.e., a + b = b + a for all a, b ∈ G. โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร คร้งั ท่ี 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

Solution and Stability of an n-dimensional Cauchy Functional Equa… 55 Definition 2.3. [1] A metric space is a pair (X, d), where X is a set and d is a metric on X (or distance function on X), that is; a real valued function defined on X × X such that for all x, y, z ∈ X we have 1. d(x, y) ≥ 0, 2. d(x, y) = 0 if and only if x = y, 3. d(x, y) = d(y, x), 4. d(x, y) ≤ d(x, z) + d(z, y). Definition 2.4. [7] The structure (V, ⊕, ⊙) is a vector space if and only if the following properties hold: 1. (V, ⊕) is an abelian group. 2. c ⊙ u ∈ V for all c ∈ R and for all u ∈ V. 3. c ⊙ (u ⊕ v) = (c ⊙ u) ⊕ (c ⊙ v) for all c ∈ R and for all u, v ∈ V. 4. (c + d) ⊙ u = (c ⊙ u) ⊕ (d ⊙ u) for all c, d ∈ R and for all u ∈ V. 5. (c · d) ⊙ u = c ⊙ (d ⊙ u) for all c, d ∈ R and for all u ∈ V. 6. There exists 1 ∈ F such that 1 ⊙ u = u for all u ∈ V. Definition 2.5. [1] Let X be a vector space. A norm on X is a real-valued function ∥·∥ on X such that the following conditions are satisfied 1. ∥x∥ ≥ 0 for all x ∈ X, 2. ∥x∥ = 0 if and only if x = 0, 3. ∥αx∥ = |α|∥x∥ for all x ∈ X and α ∈ R, 4. ∥x + y∥ ≤ ∥x∥ + ∥y∥ for all x, y ∈ X, The ordered pair (X, ∥ · ∥) is called a normed space. It should be noted that the metric induced by the normed of X is the metric d on X defined by formula d(x, y) = ∥x − y∥. Definition 2.6. [1] A sequence {xn} in a normed space is said to be Cauchy sequence if given any ε > 0, there exists k ∈ N such that ∥xn − xm∥ < ε for all m, n ≥ k. โครงการสมั มนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร ครงั้ ท่ี 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

56 Sutthiphong Moonmee and Choodech Srisawat Definition 2.7. [1] A normed space (X, ∥ · ∥) is said to be complete if every Cauchy sequence in X converges. Definition 2.8. [1] A normed space which is complete with respect to the metric induced by the norm is called a Banach space. 3. Main Results In this section, we will find the general solution and prove the Hyers-Ulam sta- bility of an n-dimensional Cauchy functional equation of Pexider type. Let n > 1 be an integer and let Λn = {1, 2, . . . , n}. Theorem 3.1. Let (G, +) be an abelian group and E be a Banach space. A mapping f : G → E satisfies nn f xi = f (xi) (3.1) i=1 i=1 for all x1, x2, . . . , xn ∈ G if and only if f (x + y) = f (x) + f (y) (3.2) for all x, y ∈ G. Proof. (=⇒) Assume that f satisfies (3.1) for all x1, x2, . . . , xn ∈ G. Substituting x1 = x2 = . . . = xn = 0 in (3.1), we get f(0) = 0. Setting x3 = x4 = . . . = xn = 0 in (3.1), we obtain f(x1 + x2) = f(x1) + f(x2) for all x1, x2 ∈ G. (⇐=) Assume that f satisfies (3.2) for all x, y ∈ G. We have n f xi = f (x1 + x2 + . . . + xn) i=1 = f (x1) + f (x2 + x3 + . . . + xn) ... = f (x1) + f (x2) + . . . + f (xn) n = f (xi) i=1 for all x1, x2, . . . , xn ∈ G. Theorem 3.2. Let (G, +) be an abelian group and E be a Banach spece. Mappings g, f1, f2, . . . , fn : G → E satisfy nn (3.3) g xi = fi(xi) i=1 i=1 โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร คร้ังท่ี 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

Solution and Stability of an n-dimensional Cauchy Functional Equa… 57 for all x1, x2, . . . , xn ∈ G if and only if there exists an additive mapping A : G → E and constants C1, C2, . . . , Cn ∈ E, for each k ∈ Λn, such that fk(x) = A(x) + Ck (3.4) for all x ∈ G and n (3.5) for all x ∈ G. g(x) = A(x) + Ci i=1 Proof. (=⇒) Assume that g, f1, f2, . . . , fn satisfy (3.3) for all x1, x2, . . . , xn ∈ G. Let Ci = fi(0) for all i ∈ Λn. Setting x1 = x, x2 = y and x3 = x4 = . . . = xn = 0 in (3.1), we get n (3.6) g(x + y) = f1(x) + f2(y) + Ci i=3 for all x, y ∈ G. Setting x = 0 in (3.6), we get n (3.7) g(y) = f2(y) + C1 + Ci i=3 for all y ∈ G. Setting y = 0 in (3.6), we obtain n (3.8) g(x) = f1(x) + Ci i=2 for all x ∈ G. By (3.6), (3.7) and (3.8), we have n (3.9) g(x + y) = g(x) + g(y) − Ci i=1 for all x, y ∈ G. Define a function A by n (3.10) A(x) = g(x) − Ci i=1 for all x ∈ G. Note that A is an additive function. Let k ∈ Λn. Setting xk = x and xi = 0 for all i ∈ Λn\\{k} in (3.3), we get g(x) = fk(x) + Ci i∈Λn\\{k} n A(x) + Ci = fk(x) + Ci i=1 i∈Λn\\{k} fk(x) = A(x) + Ck โครงการสมั มนาวิชาการและนำเสนอผลงานดานคณิตศาสตร ครั้งท่ี 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

58 Sutthiphong Moonmee and Choodech Srisawat for all x ∈ G. (⇐=) Let A : G → E be an additive function and g, f1, f2, . . . , fn : G → E be functions satisfying (3.4) and (3.5) for all x ∈ G. For any x1, x2, . . . , xn ∈ G, we get n nn g xi = A xi + Ci i=1 i=1 i=1 n = A(x1) + A(x2) + . . . + A(xn) + Ci i=1 nn = A(xi) + Ci i=1 i=1 n = A(xi) + Ci i=1 n = fi(xi) i=1 for all x1, x2, . . . , xn ∈ G as desired. Theorem 3.3. Let (G, +) be an abelian group, E be a Banach space and ε ≥ 0 be given. Let g, f1, f2, . . . , fn : G → E be mappings satisfying nn ≤ε (3.11) fi(xi) − g xi i=1 i=1 for all x1, x2, . . . , xn ∈ G. Then there exists a unique additive function A : G → E and constants C1, C2, . . . , Cn ∈ E satisfying, for each k ∈ Λn, ∥ fk(x) − A(x) − Ck∥ ≤ 8ε (3.12) for all x ∈ G and n (3.13) for all x ∈ G. g(x) − A(x) − Ci ≤ 6ε i=1 Proof. For each k ∈ Λn, we define mappings f¯k, g¯ : G → E by f¯k(x) = fk(x) − fk(0) and g¯(x) = g(x) − g(0) (3.14) for all x ∈ G. By (3.14), we obtain g¯(0) = f¯k(0) = 0 โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร ครั้งท่ี 7 มหาวิทยาลัยราชภัฏนครสวรรค

Solution and Stability of an n-dimensional Cauchy Functional Equa… 59 for each k ∈ Λn. Setting xi = 0 for all i ∈ Λn in (3.11), we get (3.15) n fi(0) − g(0) ≤ ε. i=1 By (3.11), (3.14) and (3.15), we have nn (3.16) f¯i(xi) − g¯ xi ≤ 2ε i=1 i=1 for all x1, x2, . . . , xn ∈ G. Setting x1 = x, x2 = y and xi = 0 for all i ∈ Λn\\{1, 2} in (3.16), we have ∥ f¯1(x) + f¯2(y) − g¯(x + y)∥ ≤ 2ε (3.17) for all x, y ∈ G. For each k ∈ Λn, we put xi = 0 for all i ∈ Λn\\{k} and xk = x in (3.16) to get ∥ f¯k(x) − g¯(x)∥ ≤ 2ε (3.18) (3.19) for all x ∈ G. By (3.18), we see that ∥ f¯1(x) − g¯(x)∥ ≤ 2ε and ∥ f¯2(x) − g¯(x)∥ ≤ 2ε. (3.20) (3.21) Replacing x by y in (3.20), we get ∥ f¯2(y) − g¯(y)∥ ≤ 2ε for all y ∈ G. By (3.17), (3.19) and (3.21), we obtain ∥ g¯(x) + g¯(y) − g¯(x + y)∥ = ∥ f¯1(x) + f¯2(y) − g¯(x + y) + g¯(x) − f¯1(x) + g¯(y) − f¯2(y) ∥ ≤ ∥ f¯1(x) + f¯2(y) − g¯(x + y)∥ + ∥ g¯(x) − f¯1(x)∥ + ∥ g¯(y) − f¯2(y)∥ (3.22) = 6ε for all x, y ∈ G. Setting y = x in (3.22), we have ∥ 2g¯(x) − g¯(2x)∥ = ∥ g¯(x) + g¯(x) − g¯(x + x)∥ ≤ 6ε โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร คร้งั ท่ี 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

60 Sutthiphong Moonmee and Choodech Srisawat for all x ∈ G. Multiplying by 2−1 in above inequality, we obtain ∥ g¯(x) − 2−1g¯(2x)∥ ≤ 3ε (3.23) for all x ∈ G. For each positive integer m and each x ∈ G, we use (3.23) to get m−1 ∥ g¯(x) − 2−mg¯(2mx)∥ = 2−ig¯(2ix) − 2−(i+1)g¯(2i+1x) i=0 m−1 (3.24) ≤ 2−i3ε. i=0 Consider the sequence {2−mg¯(2mx)}. For each positive integers k < l and each x ∈ G, we apply inequality (3.24) to obtain l−k−1 ∥ 2−kg¯(2kx) − 2−lg¯(2lx)∥ ≤ 2−k 2−i3ε i=0 ∞ ≤ 2−k3ε 2−i i=0 = 2−k6ε. (3.25) Since lim 2−k6ε = 0, {2−mg¯(2mx)} is a Cauchy sequence. From {2−mg¯(2mx)} ⊆ E and Eki→s ∞complete, we obtain that {2−mg¯(2mx)} converges. We can define a mapping A : G → E by A(x) = lim 2−mg¯(2mx) for all x ∈ G. As m → ∞, (3.24) becomes m→∞ ∞ ∥ g¯(x) − A(x)∥ ≤ 2−i3ε i=0 = 6ε (3.26) for all x ∈ G. By (3.18) and (3.26), we have ∥ f¯k(x) − A(x)∥ ≤ ∥ f¯k(x) − g¯(x)∥ + ∥ g¯(x) − A(x)∥ (3.27) ≤ 8ε for all x ∈ G and k ∈ Λn. Next, we will prove A is additive. Replicing x by 2mx and y by 2my in (3.22), we get ∥ g¯(2mx) + g¯(2my) − g¯ 2m(x + y) ∥ ≤ 6ε. (3.28) Multiplying the above inequality by 2−m and taking m → ∞, we see that lim 2−m∥ g¯(2mx) + g¯(2my) − g¯ 2m(x + y) ∥ ≤ lim 2−m6ε m→∞ m→∞ lim 2−mg¯ 2m(x + y) = lim 2−mg¯(2mx) + lim 2−mg¯(2my) m→∞ m→∞ m→∞ A(x + y) = A(x) + A(y) โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครงั้ ที่ 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

Solution and Stability of an n-dimensional Cauchy Functional Equa… 61 for all x, y ∈ G. We will prove the uniqueness, let A¯ : G → E be a function satisfies (3.2) and (3.26). For each positive integer m and each x ∈ G, we observe that ∥ A(x) − A¯(x)∥ = 2−m∥ A(2mx) − A¯(2mx)∥ (3.29) ≤ 2−m ∥ A(2mx) − g¯(2mx)∥ + ∥ g¯(2mx) − A¯(2mx)∥ ≤ 2−m12ε. Then, by taking m → ∞ in (3.29), we get A(x) − A¯(x) = 0 A(x) = A¯(x) for all x ∈ G. References [1] D. Daners, “Introduction to Functional Analysis,” The University of Sydney, 2017. [2] D. H. Hyers, “On the Stability of the Linear Functional Equations,” Proceedings of the National Academy of Sciences of the United States of America, vol. 27, no. 4, pp. 222-224, 1941. [3] M.S. Moslehian, “On the stability of the orthogonal Pexiderized Cauchy equation”, Dept. of Math., Ferdowsi Univ. 2004 [4] Renu Chugh and Ashish, “The stability of Pexiderized Cauchy functional equation”, International Journal of Pure and Applied Mathematics, vol 77, no. 5, pp. 649-666, 2012 [5] S. M. Ulam, Problems in Modern Mathematics, Chapter 6, John Wiley & Sons, New York, NY, USA, 1964. [6] Th. M. Rassias, “On the Stability of the Linear Mapping in Banach Spaces,” Pro- ceedings of the National Academy of Sciences of the United States of America, vol. 72, no. 2, pp. 297-300, 1978. [7] Th. W. Judson, “Abstract Algebra,” Stephen F. Austin State University, 2013. โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร คร้งั ท่ี 7 มหาวิทยาลัยราชภฏั นครสวรรค

62 Sutthiphong Moonmee and Choodech Srisawat โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครั้งท่ี 7 มหาวิทยาลัยราชภัฏนครสวรรค

Type of the Article: Research RE-PU 16 63 On the Ternary Quadratic Diophantine Equation n(x2 + y2) − (2n − 1)xy = (4n − 1 + m2)z2 Kallayanee Angkeaw, Chawewan Petdee and Supamit Pimsri* Department of Mathematics, Faculty of Science, Ubon Ratchathani Rajabhat University Ubon Ratchathani, 34000, Thailand *Correspondence: [email protected] Abstract We will find the solution to the ternary quadratic diophantine equation in the form n(x2 + y2) − (2n − 1)xy = (4n − 1 + m2)z2 which has 9 different solutions. Keywords: Ternary quadratic, Integral solutions, Diophantine equation 1. Introduction The ternary quadratic diophantine equation there are many forms. The solution to the equation has a variety of methods. The diophantine equation in the form A(x2 + y2)−Bxy = Cz2. there are many researchers interested such as [1–9]. In this paper, we present different patterns of non-zero distinct solutions of ternary quadratic diophantine equation n(x2 + y2) − (2n − 1)xy = (4n − 1 + m2)z2. 2. Method of Analysis Consider the equation n(x2 + y2) − (2n − 1)xy = (4n − 1 + m2)z2. (2.1) The substitution of the linear transformations x = u + v; y = u − v. (2.2) Using (2.2) in (2.1), it is written as u2 + (4n − 1)v2 = (4n − 1 + m2)z2. (2.3) We present below different methods of solving (2.3) and thus obtain different choices of integer solution of (2.1). โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร คร้งั ที่ 7 มหาวทิ ยาลัยราชภัฏนครสวรรค

64 Kallayanee Angkeaw et al. Pattern I We can write (2.3) as (u − mz)(u + mz) = (4n − 1)(z − v)(z + v). (2.4) Case 1 Equation (2.4) can also be rewritten as u − mz (4n − 1)(z − v) A = =. z+v u + mz B This is equivalent to the system of equations  Bu − Av − (A + mB)z = 0,  −Au − (4n − 1)Bv − [mA − (4n − 1)B]z = 0.  Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = mA2 − (4n − 1)mB2 − 2(4n − 1)AB, v = A2 − (4n − 1)B2 + 2mAB, z = −A2 − (4n − 1)B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = (m + 1)A2 − (m + 1)(4n − 1)B2 − 2[(4n − 1) − m]AB, y = (m − 1)A2 − (m − 1)(4n − 1)B2 − 2[(4n − 1) + m]AB, z = −A2 − (4n − 1)B2. Case 2 Equation (2.4) can also be rewritten as u − mz = z−v = A (4n − 1)(z + v) u + mz . B This is equivalent to the system of equations  Bu − (4n − 1)Av − [(4n − 1)A + mB]z = 0,  −Au − Bv − (mA − B)z = 0.  โครงการสมั มนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร ครง้ั ท่ี 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

On the Ternary Quadratic Diophantine Equation n(x2 + y2) − (2n − 1)… 65 Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = (4n − 1)mA2 − mB2 − 2(4n − 1)AB, v = (4n − 1)A2 − B2 + 2mAB, z = −(4n − 1)A2 − B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = (m + 1)(4n − 1)A2 − (m + 1)B2 − 2[(4n − 1) − m]AB, y = (m − 1)(4n − 1)A2 − (m − 1)B2 − 2[(4n − 1) + m]AB, z = −(4n − 1)A2 − B2. Case 3 Equation (2.4) can also be rewritten as u − mz = (4n − 1)(z + v) = A z−v u + mz . B This is equivalent to the system of equations  Bu + Av − (A + mB)z = 0,  −Au + (4n − 1)Bv − [mA − (4n − 1)B]z = 0.  Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = −mA2 + (4n − 1)mB2 + 2(4n − 1)AB, v = A2 − (4n − 1)B2 + 2mAB, z = A2 + (4n − 1)B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = −(m − 1)A2 + (m − 1)(4n − 1)B2 + 2[(4n − 1) + m]AB, y = −(m + 1)A2 + (m + 1)(4n − 1)B2 + 2[(4n − 1) − m]AB, z = A2 + (4n − 1)B2. โครงการสัมมนาวชิ าการและนำเสนอผลงานดา นคณิตศาสตร ครั้งท่ี 7 มหาวทิ ยาลัยราชภัฏนครสวรรค

66 Kallayanee Angkeaw et al. Case 4 Equation (2.4) can also be rewritten as u − mz = z+v = A. (4n − 1)(z − v) u + mz B This is equivalent to the system of equations  Bu + (4n − 1)Av − [(4n − 1)A + mB]z = 0,  −Au + Bv − (mA − B)z = 0.  Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = −(4n − 1)mA2 + mB2 + 2(4n − 1)AB, v = (4n − 1)A2 − B2 + 2mAB, z = (4n − 1)A2 + B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = −(m − 1)(4n − 1)A2 + (m − 1)B2 + 2[(4n − 1) + m]AB, y = −(m + 1)(4n − 1)A2 + (m + 1)B2 + 2[(4n − 1) − m]AB, z = (4n − 1)A2 + B2. Case 5 Equation (2.4) can also be rewritten as u + mz (4n − 1)(z − v) A = =. z+v u − mz B This is equivalent to the system of equations  Bu − Av − (A − mB)z = 0,  −Au − (4n − 1)Bv + [(4n − 1)B + mA] z = 0.  Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = −mA2 + (4n − 1)mB2 − 2(4n − 1)AB, v = A2 − (4n − 1)B2 − 2mAB, z = −A2 − (4n − 1)B2. โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร ครั้งท่ี 7 มหาวิทยาลยั ราชภัฏนครสวรรค

On the Ternary Quadratic Diophantine Equation n(x2 + y2) − (2n − 1)… 67 Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = −(m − 1)A2 + (m − 1)(4n − 1)B2 − 2 [(4n − 1) + m] AB, y = −(m + 1)A2 + (m + 1)(4n − 1)B2 − 2 [(4n − 1) − m] AB, z = −A2 − (4n − 1)B2. Case 6 Equation (2.4) can also be rewritten as u + mz z−v A = = . (4n − 1)(z + v) u − mz B This is equivalent to the system of equations  Bu − (4n − 1)Av − [(4n − 1)A − mB]z = 0,  −Au − Bv + (mA + B)z = 0.  Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = −(4n − 1)mA2 + mB2 − 2(4n − 1)AB, v = (4n − 1)A2 − B2 − 2mAB, z = −(4n − 1)A2 − B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = −(m − 1)(4n − 1)A2 + (m − 1)B2 − 2[(4n − 1) + m]AB, y = −(m + 1)(4n − 1)A2 + (m + 1)B2 − 2[(4n − 1) − m]AB, z = −(4n − 1)A2 − B2. Case 7 Equation (2.4) can also be rewritten as u + mz (4n − 1)(z + v) A z−v = =. u − mz B This is equivalent to the system of equations  Bu + Av − (A − mB)z = 0,  −Au + (4n − 1)Bv + [mA + (4n − 1)B]z = 0.  โครงการสมั มนาวิชาการและนำเสนอผลงานดานคณิตศาสตร คร้งั ที่ 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

68 Kallayanee Angkeaw et al. Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = mA2 − (4n − 1)mB2 + 2(4n − 1)AB, v = A2 − (4n − 1)B2 − 2mAB, z = A2 + (4n − 1)B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = (m + 1)A2 − (m + 1)(4n − 1)B2 + 2[(4n − 1) − m]AB, y = (m − 1)A2 − (m − 1)(4n − 1)B2 + 2[(4n − 1) + m]AB, z = A2 + (4n − 1)B2. Case 8 Equation (2.4) can also be rewritten as u + mz = z+v A = . (4n − 1)(z − v) u − mz B This is equivalent to the system of equations  Bu + (4n − 1)Av − [(4n − 1)A − mB]z = 0,  −Au + Bv + (mA + B)z = 0.  Applying the method of cross multiplication, the above system of the equation is sat- isfied by u = (4n − 1)mA2 − mB2 + 2(4n − 1)AB, v = (4n − 1)A2 − B2 − 2mAB, z = (4n − 1)A2 + B2. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = (m + 1)(4n − 1)A2 − (m + 1)B2 + 2[(4n − 1) − m]AB, y = (m − 1)(4n − 1)A2 − (m − 1)B2 + 2[(4n − 1) + m]AB, z = (4n − 1)A2 + B2. โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร ครัง้ ที่ 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

On the Ternary Quadratic Diophantine Equation n(x2 + y2) − (2n − 1)… 69 Pattern II Assume that z = A2 + (4n − 1)B2. (2.5) Substituting (2.5) in (2.3) and applying the method of factorization we get, √ − √ − 1v) = √ − √ − 1)[A2 +(4n−1)B2]2 . (2.6) (u+i 4n 1v)(u−i 4n (m+i 4n 1)(m−i 4n Equating the positive and negative factors, the resulting equations are √ = (m + √ − 1)(A + √ − 1B )2. (2.7) u + i 4n − 1v i 4n i 4n (2.8) √ = (m − √ − 1)(A − √ − 1B )2. u − i 4n − 1v i 4n i 4n Equating the real and imaginary parts in either (2.7) or (2.8), we get u = mA2 − (4n − 1)mB2 − 2(4n − 1)AB, v = A2 − (4n − 1)B2 + 2mAB. Substituting the values of u and v in (2.2), we obtain the integer solutions to (2.1) as given below x = (m + 1)A2 − (m + 1)(4n − 1)B2 − 2[(4n − 1) − m]AB, y = (m − 1)A2 − (m − 1)(4n − 1)B2 − 2[(4n − 1) + m]AB, z = A2 + (4n − 1)B2. 3. Conclusion We have found the solution to the ternary quadratic equation n(x2 + y2) − (2n − 1)xy = (4n − 1 + m2)z2 which has 9 different solutions. A few interesting relations between the solutions and special numbers are exhibited. References [1] J. Shanthi, M. A. Gopalan and S. Vidhyalakshmi , “Lattice points on the homoge- neous cone 8(x2 + y2) − 15xy = 56z2,” Sch. J. Phys. Math. Stat., vol. 1, no. 1, pp. 29–32, 2014. [2] K. Meena and Team , “Integral points on the cone 3(x2+y2)−5xy = 47z2,” Bulletin of Mathematics and Statistic Research , vol. 2, no. 1, pp. 65–70, 2014. โครงการสัมมนาวชิ าการและนำเสนอผลงานดา นคณติ ศาสตร ครัง้ ที่ 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

70 Kallayanee Angkeaw et al. [3] M. Somanath, J. Kannan and K. Raja , “Integral Solutions of an Infinite Cone α(x2 + y2) = (2α − 1)xy + (4α − 1)z2,” International Journal for Research in Applied Science and Engineering Technology (IJRASET)., vol. 4, no. 10, pp. 504–507, 2016. [4] M. A. Gopalan, M. Somanath and V. Sangeetha, “On ternary quadratic equation 5(x2 + y2) − 9xy = 19z2,” International Journal of Advanced Engineering Science and Information Technology , vol. 2, no. 6, pp. 2008–2013, 2013. [5] M. A. Gopalan, R. Anbuselvi and N. Ahila, “On ternary quadratic equation 3(x2 + y2) − 5xy = 11z2,” The International Journal Of Engineering And Science , vol. 4, no. 7, pp. 26–30, 2015. [6] M. A. Gopalan, S. Vidhyalakshmi and S. Nivetha, “On the ternary quadratic equation 4(x2 + y2) − 7xy = 31z2,” Diophantus J. Math , vol. 3, no. 1, pp. 1–7, 2014. [7] M. A. Gopalan and S. Vidhyalakshmi, “On the ternary quadratic Diophantine equa- tion 8(x2 + y2) − 15xy = 80z2,” BOMSR, vol. 2, no. 4, pp. 429–433, 2014. [8] R. Anbuselvi and K. S. Araththi, “On the ternary quadratic equation x2 + y2 − xy = 3z2,” International Educational Scientific Research Journal, vol. 3, no. 1, pp. 122– 123, 2017. [9] R. Nandhini , “On ternary quadratic equation 6(x2 +y2)−11xy = 23z2,” Asia Pacific Journal of Research, vol. 1, no. 58, pp. 142–146, 2017. โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณิตศาสตร ครงั้ ท่ี 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

ผลงานสมั มนา



Type of the Article: Seminar SE-AP 01 71 การวิเคราะหก ารถดถอยโดยใชต ัวอยา งแบบชดุ ลำดับ On Regression Analysis Using Ranked Set Sample ผแู ตง: Hani M. Samawi and Faisal M. Ababneh จัดทำโดย: ธนาภรณ อินเอ่ียม1* 1หลักสตู รสาขาวชิ าคณิตศาสตร คณะวิทยาศาสตรแ ละเทคโนโลยี มหาวิทยาลยั ราชภฏั พิบูลสงคราม *Correspondence: [email protected] บทคัดยอ สำหรับการวจิ ยั นี้ไดศกึ ษาผลของตวั อยา งจากการสมุ ตัวอยางแบบชดุ ลำดับ (RSS) ของการวิเคราะหการ ถดถอย โดยเฉพาะอยา งย่งิ การประมาณคาพารามเิ ตอรของแบบจำลองการถดถอยอยางงายรวมถงึ แสดง ใหเหน็ วาการสุมตวั อยางแบบชดุ ลำดับใหตวั ประมาณท่ีไมเอนเอียงของพารามเิ ตอรสำหรบั ตัวแบบการ ถดถอย คำสำคญั : ตวั อยา งแบบชดุ ลำดบั , การวเิ คราะหก ารถดถอย Abstract The effect of ranked set sample (RSS) on regression analysis is investigated. In particular, parameter estimation of the simple regression model fit. It is shown that RSS gives unbiased estimates of the regression model parameters. Keywords: Ranked seet sample, Regression analysis. 1. บทนำ การวิเคราะหการถดถอยเปนวิธีการท่ีใชเพ่อื ทำความเขาใจวาตวั แปรหนึ่งตวั จะข้นึ อยูกบั อกี หนงึ่ หรอื มากกวา หนึ่งตัวแปร โดยมีวตั ถปุ ระสงค คอื อธิบาย ควบคมุ และ ทำนาย ตัวอยางเชน หาก ทราบถงึ ความสมั พันธระหวา งคาใชจายในการโฆษณา และการขาย จะสามารถทำนายยอดขายได ดว ยการวเิ คราะหการถดถอยได เมื่อมีการกำหนดระดับคา ใชจา ยการโฆษณา ในตัวแบบการถดถอย ประกอบดวยหนึ่งตวั แปร (Y ) ซง่ึ เปน ตวั แปรตาม ในขณะที่ตัวแปรอืน่ ๆ (X) ท้ังหมดคอื ตัวแปร ทำนาย หรือตัวแปรตน ตัวแบบของการวเิ คราะหการถดถอยมีรูปแบบ คือ Yi = α + βXi + εi ; i = 1, 2, ..., N ประกอบดวยพารามเิ ตอร α และ β ในป 1985 Neter และคณะ ไดแสดงใหเ หน็ วา การประมาณคา การทดสอบ และการทำนายที่ ไดจากกรณีเมอ่ื ทราบคาของ X ซง่ึ เปน คาคงที่ โดยมีเง่ือนไขคือ Y เปนตัวแปรสุมที่มีแจกแจงความนา จะเปน f(Yi | Xi) ที่มีคาเฉลย่ี α +βXi และความแปรปรวนตามเงอื่ นไข σ2 และ Xi′s เปนตวั แปรสมุ ทีเ่ ปน อสิ ระตอกนั ซ่งึ การแจกแจงความนา จะเปน ไมขึ้นอยูกับพารามเิ ตอร α, β และ σ2 [8] นอกจากน้ี โครงการสัมมนาวชิ าการและนำเสนอผลงานดา นคณิตศาสตร คร้ังท่ี 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค

72 ธนาภรณ อินเอี่ยม การเปล่ียนแปลงสว นใหญที่เกดิ ข้นึ จากการตีความคาสมั ประสิทธ์ิความเชือ่ ม่นั และความเสยี่ งที่เกดิ ขน้ึ ของความผดิ พลาด เมอ่ื X ถกู สมุ โดยการสุมตัวอยา งจะสุมมาเปน คู (Xi, Yi) เมื่อคาของ Xi เปลยี่ น คา ของ Yi ก็จะเปล่ียนตาม ในกรณีใดๆ คาสังเกตไดจะถกู ใชประมาณคาพารามิเตอรท่ีไมทราบคา วิธี การประมาณคาท่ีใชในบทความนี้ คอื วธิ กี ำลงั สองนอยสุด (Ordinary least squares method) ในทางปฏิบัติขอ มลู จะถกู รวบรวมในแตละหนว ยหรอื กรณีของตวั แปรเหลา น้ี โดยใชวธิ ีการ สุมอยา งงา ย (SRS) นอกเหนอื จากวิธีการสุมตัวอยา งอยา งงาย แลว ยังมีการสุมตวั อยา งแบบชุดลำดบั (RSS) ซง่ึ ไดนำเสนอโดย McIntyre ในป 1952 ซง่ึ ไดแสดงใหเห็นถึงประสิทธิภาพของการประมาณคา เฉลย่ี ประชากร [6] สำหรบั รายละเอยี ดเพ่มิ เติมเก่ยี วกบั การสมุ ตัวอยา งแบบชดุ ลำดบั เห็นไดจากงาน ของ Kaur และคณะ (1995) และ Patil และคณะ (1999) [5] ตอมาในป 1995 Muttlak ใชการสมุ ตัวอยางแบบชดุ ลำดบั เพ่ือประมาณพารามเิ ตอรของตวั แบบการถดถอยอยา งงาย โดยสมมติวา คา X เปน คา คงท่ีท่ีทราบ อยา งไรกต็ ามภายใตสมมติของตัวแบบของเขาไมมีการปรบั ปรุงในการประมาณคา พารามเิ ตอรโมเดล [7] เพอ่ื ใหบรรลุถงึ การปรบั ปรงุ ในการประมาณพารามิเตอรของตวั แบบ จงึ จำเปน ตอ งสมมติวาใหทง้ั Y และ X เปนตวั แปรสมุ ในบทความผูเขยี นสมมติวา ตัวแปร X และ Y ถูกสมุ นอกจากน้ีผูเขยี นไดเสนอการศกึ ษา พฤตกิ รรม และประสิทธิภาพของการวเิ คราะหการถดถอยเชิงเสนอยางงายเมือ่ ใชการสุมตัวอยางแบบ ชดุ ลำดบั 2. การสุมตัวอยางแบบชดุ ลำดับไดด ลุ (Balance Ranked Set Sampling) ในการสำรวจตวั อยา งเพื่อเก็บรวบรวมขอ มูลสำหรบั ในกรณีที่ประชากรที่ศึกษามีขนาดใหญจึง ทำใหตองใชเวลา และคาใชจา ยสำหรบั การศกึ ษาเปนจำนวนมาก ในป 1995 McIntyre จงึ ไดพฒั นา วิธีการสุมตวั อยา งขึ้นมา เพื่อหาวธิ ีการสมุ ตวั อยางท่ีสามารถลดคา ใชจาย (Cost-effective Sampling Method) และประหยดั เวลาในการดำเนนิ การ [6] วธิ ีการสุมตวั อยา งแบบชดุ ลำดับไดดลุ ไดพฒั นามาจากแผนการสุม ตวั อยางแบบงายซง่ึ แผนการ สมุ ตวั อยา งแบบงา ยนนั้ มีขอ จำกดั ดา นกรอบตวั อยา ง หรอื บญั ชีรายชอ่ื หนวยประชากร สว นแผนการสมุ ตัวอยา งแบบมีชน้ั ภูมิที่ไดพฒั นามาจากแผนการสมุ ตวั อยางแบบงา ยนัน้ เปน แผนการสมุ ตัวอยา งท่ีแบง ขนาดประชากร N ออกเปน ชน้ั ภูมิ แลว จงึ ทำการสมุ ตัวอยางในแตละช้ันภูมิ ซ่งึ แผนการสมุ ตวั อยา ง แบบมชี น้ั ภมู ินี้ก็ตองมกี รอบตวั อยา ง หรอื บญั ชรี ายชื่อเชนเดยี วกับแผนการสุมตัวอยา งแบบงาย ตอ มาเม่อื มีความตองการศึกษาประชากรที่มีขนาดใหญมาก ๆ จงึ ไดพัฒนาเปนแผนการสมุ ตวั อยา งแบบกลุมขึ้น โดยทำการแบง ประชากรออกเปน กลมุ ยอ ยตามภูมิศาสตร แลว เกบ็ ขอมูลตวั อยา ง จากกลุมท่ีแบง ซงึ่ ไมจำเปน ตองมีรายชอ่ื ทุกหนว ยประชากร แตการสุมตวั อยา งแบบกลุมน้ีก็มีขอจำกัด ดานความเชอ่ื ถือไดข องคา ประมาณ ซึ่งอยูในระดบั ต่ำกวา แผนการสมุ ตวั อยา งแบบงาย และเพ่อื ตอบสนองความตอ งการในขณะนน้ั จงไดหาวธิ ีท่ีสามารถลดคา ใชจาย และเวลาใน การดำเนินการเก็บรวบรวมขอ มลู เพือ่ ใชในการปริมาณหญาของทงุ หญา ในประเทศออสเตรเลยี เม่ือตน โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครัง้ ที่ 7 มหาวิทยาลยั ราชภัฏนครสวรรค

การวเิ คราะหก ารถดถอยโดยใชต ัวอยา งแบบชดุ ลำดับ 73 ป 1952 McIntyre จงึ ไดนำเสนอวธิ ีการสุม ตัวอยา งข้นึ มาหน่ึงวธิ ี ซง่ึ ตอ มาภายหลงั Hall และ Dell ได เรียกวิธกี ารสมุ ตัวอยา งดงั กลาววา การสมุ ตัวอยา งแบบชุดลำดบั (Ranked Set Sampling) แผนการสุมตวั อยางของ McIntyre ดำเนินการสุม ตวั อยางดงั นี้ เริ่มจากการแบงทุงหญา ออก เปน ล็อต (Lost) แลวทำการสุม ตัวอยา งทงุ หญา ที่ถกู แบง เปน ล็อตมา k ล็อต จากนน้ั การเรียงลำดบั แต ละลอ็ ตตามปริมาณหญา ดวยการคาดคะเน หรอื ประมาณดวยสายตา (Eye Inspection) ซ่ึงจะทำการ สุม ตัวอยา งในลักษณะดงั กลา วมา k คร้ัง จากนน้ั ทำการเลอื กหนว ยตวั อยา ง ซึง่ ตัวอยา งจากการสมุ ครัง้ ท่ี 1 เลือกหญาในลอ็ ตลำดบั ท่ี 1 เพอ่ื ทำการตัด และนำไปชั่งนำ้ หนกั สว นตัวอยางจากการสมุ คร้ังท่ี 2 เลือกหญาในล็อตดับที่ 2 แลว ดำเนนิ การเชนเดยี วกนั กับล็อตลำดบั ที่ 1 และดำเนินการเชน น้ีไปเรื่อย ๆ จนกระทั่งครบ k คร้ัง ซึ่งจะไดตวั อยา งหญา k ล็อตการสุมตวั อยา งดังกลาว เรียกวา การสมุ ตัวอยา ง 1 รอบ (Cycle) และหาก ตองการจำนวนตัวอยา งท่ีเพิ่มขน้ึ สามารถทำการสุมตวั อยา งตามกระบวนการดงั กลาวมา m รอบจาก น้นั จงึ ทำการประมาณปรมิ าณหญา การสมุ ตัวอยา งแบบชุดลำดบั นัน้ มีข้ันตอนการสุมตวั อยา งที่ซับซอนมากกวาการสุม ตัวอยาง แบบงายน่นั คือการสมุ ตวั อยา งแบบงายจะทำการสุม ตัวอยางข้นึ มาเพยี ง n หนวยเทานัน้ จากประชากร ขนาด N หนวยแตสำหรบั การสมุ ตวั อยางแบบชุดลำดบั จะทำการสุมตวั อยา งมา nr หนว ย และ r = 1, 2, ..., k เรยี กวา ชุดตวั อยา ง ซง่ึ จะทำการสุม ตัวอยางขึน้ มา k ชดุ ตัวอยาง เรียกวาการสมุ ตัวอยา ง 1 รอบ และกำหนดให Y[r]i′ คือ หนว ยตวั อยา งลำดับที่ r ชุดตัวอยา งที่ i′ สำหรบั การสมุ ตัวอยา ง 1 รอบ ไดแ สดงดังนี้ Y[1]1 Y[1]2 ... Y[1]n1 Y[2]1 Y[2]2 ... Y[1]n2 ... ... ... ... Y[k]1 Y[k]2 ... Y[k]nk ถาขนาดในการสุมแตล ะชดุ ตวั อยา งเทากนั คือ n1 = n2 = ... = nk แลว จะเรียกการสมุ ตัวอยางแบบชดุ ลำดบั น้ีวา การสมุ ตวั อยางแบบชดุ ลำดับไดดุล (Balance Ranked Set Sampling) และถาขนาดตัวอยา งท่สี ุมมาในแตล ะชดุ ตัวอยา งมขี นาดไมเ ทา กนั แลว จะเรยี กการสมุ ตัวอยา งนวี้ า การ สมุ ตวั อยา งแบบชุดลำดบั ไมไ ดด ลุ (Unbalance Ranked Set Sampling) [2] ขนั้ ตอนการสมุ ตวั อยา งแบบชดุ ลำดบั ไดดลุ สำหรับขั้นตอนของการสุม ตัวอยางแบบชุดลำดบั ไดดุลนนั้ มีขนั้ ตอนการสุมตวั อยางดังนี้ การ สุม ตัวอยา งแบบชดุ ลำดบั ไดดุลเริ่มตน ดว ยการสมุ ตัวอยางแบบงายไมใสคนื (SRSWOR) ขนาด k หนว ย จากประชากรขนาด N หนวย ซงึ่ เรียกวา ชดุ ตัวอยา งท่ี 1 แลว นำหนว ยตวั อยาง (Sampling Units) ทง้ั k หนวยมาเรยี งลำดับจากนอ ยไปหามากดว ยการคาดคะเน หรือการประมาณดว ยสายตา (Judgment โครงการสมั มนาวชิ าการและนำเสนอผลงานดานคณิตศาสตร ครัง้ ท่ี 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

74 ธนาภรณ อนิ เอ่ยี ม Ranking) ดว ยตวั แปรที่สนใจ จากน้ันหนวยตัวอยา งท่ีอยูในลำดบั ท่ี 1 จะถูกเลือกเปน ตัวอยา ง และ ทำการวัดคาของตวั แปรทีส่ นใจ สวนหนว ยตวั อยางทเี่ หลือถกู ตัดทิง้ หรอื ไมส นใจ เม่อื สน้ิ สุดการสมุ ตัวอยา งชดุ ท่ี 1 แลว ตอ จากนัน้ สมุ ตวั อยา งแบบงายไมใสคืนขนาด k หนว ย จากประชากรขนาด N หนวย ซึ่งนั่นหมายความวา หนว ยประชากรท่ีถูกสมุ เปนตัวอยา งในชดุ ตัวอยาง ท่ี 1 ถูกนำกลบั มาใสในประชากรอกี ครงั้ หนวยประชากร k หนวยที่ถกู สุม ในครงั้ ที่ 2 เรยี กวา ชดุ ตวั อยางที่ 2 จากนนั้ นำหนวยตัวอยางมาเรียงลำดบั จากนอ ยไปหามาก หนวยตัวอยางท่ีอยูในลำดับท่ี 2 ในชดุ ตวั อยางที่ 2 น้ีจะถกู เลอื กเปนตัวอยาง แลวทำการวดั คาของตัวแปรท่ีสนใจ และหนว ยตัวอยางท่ี เหลอื ถูกตดั ทิ้งหรือไมสนใจ กระบวนการสุม ตัวอยา งน้ีดำเนินการตอ ไปเร่ือย ๆ จนกระทง่ั สมุ ตวั อยางไปจนครบ k ชุด ตวั อยา ง และเรียกกระบวนการสุมตวั อยางจนครบ k ชุดตวั อยา งนี้วา “รอบ (Cycle)” ซึง่ จะทำการ สมุ ตัวอยางทั้งหมด m รอบ และการสมุ หนวยตวั อยางในแตละชุดตัวอยา งที่ทำการสุมขน้ึ มาท้งั k ชุด ตวั อยา งในแตละรอบนัน้ การไดหนว ยตัวอยาง k หนวย ในแตละชดุ ตัวอยางไมม ีผลกระทบตอการได หนวยตวั อยางในชุดตวั อยา งอ่นื ๆ นนั่ คอื ชุดตัวอยา งท่ี 1 จะใหตัวอยา งหนวยที่ 1 ชดุ ตวั อยา งท่ี 2 จะใหตวั อยางหนวยท่ี 2 จนถึง ชดุ ตัวอยางที่ k จะใหตวั อยางหนว ยท่ี k เม่อื เสร็จส้นิ การสมุ ตัวอยางในรอบที่ 1 จะไดตัวอยา ง หนว ย และจากน้นั จึงทำการสมุ ตัวอยางในรอบท่ี 2 จะไดตวั อยา ง k หนว ย และเม่อื ทำการสุมตัวอยา งไปเรือ่ ยๆ จนถงึ รอบที่ m ซึง่ ไดต วั อยา งมา k หนวยเชน เดมิ ดังนน้ั ขอ มลู ตัวอยา งที่เกิดขนึ้ ท้งั หมดจึงมี mk หนวย เนือ่ งจากในการสุม ตัวอยางในแตล ะชดุ ตวั อยางจะสมุ จากประชากรที่มีขนาด N หนวย ทุก ครั้ง ดังน้ันหนว ยประชากรใดท่ีถกู สุมมาเปนตัวอยา งแลว ยงั สามารถถกู สุมขึ้นมาเปน หนวยตวั อยา งใน ตวั อยา งชดุ ตอไปไดสำหรับการสุมตัวอยางคร้งั ตอ ไป ดังนั้นชดุ ตวั อยา งทัง้ k ชดุ ตัวอยา งเปนอสิ ระตอ กัน และเกิดขึน้ ดว ยความนาจะเปนเทากนั เม่ือชุดตวั อยา งทงั้ k ชดุ ตัวอยาง เปน อสิ ระกนั แลว การสมุ ตัวอยางในแตละรอบยอ มเปน อสิ ระ ตอกนั และไดตวั อยา งสำหรับการสุม ตวั อยางแบบชดุ ลำดบั ไดดลุ ทม่ี ีขนาดเทา กบั mk หนวย [2] 3. การวเิ คราะหการถดถอยอยางงา ย (Simple Regression Analysis) การวเิ คราะหการถดถอยเชิงเสนอยางงา ยเปน การศกึ ษาความสมั พันธระหวางตวั แปรอิสระ 1 ตัว และ ตวั แปรตาม 1 ตวั โดยตัวแปรท้งั สองมีความสมั พันธเ ชงิ เสน รูปแบบการวิเคราะหน้ีเปน รปู แบบ พื้นฐานท่ีงายทส่ี ุดของการวิเคราะหก ารถดถอยโดยมีตัวแบบการถดถอยประชากร คือ Yi = α + βXi + εi ; i = 1, 2, ..., N โดยท่ี Yi คือ ตวั แปรตามในลำดับที่ i α และ β คือ พารามิเตอร Xi คือ ตัวแปรอิสระในลำดบั ที่ i โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครงั้ ท่ี 7 มหาวิทยาลัยราชภัฏนครสวรรค

การวิเคราะหการถดถอยโดยใชตัวอยางแบบชดุ ลำดบั 75 εi คอื ความคลาดเคลือ่ นเชงิ สุมในลำดับที่ i เปน ความคลาดเคลอ่ื นท่ีเกดิ ระหวา ง Y และ α + βXi โดยมขี อ สมมตเิ บ้ืองตนของการวิเคราะหก ารถดถอย ดงั น้ี 1. ตวั แปร X เปนตวั แปรคงท่ี หรอื กำหนดไวล วงหนา 2. ความคลาดเคล่ือน εi เปน ตัวแปรสมุ ทม่ี คี า เฉลี่ยเทา กับศนู ย หรือ E(εi) = 0 คาแปรปรวน ของ εi มคี าเทากันทกุ คาของ i หรือ V (εi) = σ2 3. εi และ εj เปน อสิ ระตอกัน 4. εi มีการแจกแจงแบบปกตทิ ี่มคี าเฉล่ียเปน ศนู ยและคาแปรปรวน = σ2 นั่นคอื εi ∼ N (0, σ2) สำหรับในทางปฏิบัตินนั้ ในบางสถานการณผูศึกษาอาจไมสามารถทำการเกบ็ รวบรวมขอมูล ของประชากรไดอยา งครบถว นดว ยขอ จำกัด ไมวา จะเปน ในเรือ่ งของการเก็บรวบรวมขอมลู บคุ ลากร ในการดำเนนิ การ รวมไปถึงในเรอื่ งของงบประมาณในการเก็บรวมรวมขอมลู ดวยเหตุนี้จงึ ตองมีการ สำรวจตัวอยาง เพื่อนำขอ มูลตวั อยา งหรอื ขอ มลู เพยี งบางสว นของประชากรมาเพอื่ ใช จากเหตผุ ลขา งตนจึงนำไปสูการเก็บขอมลู ตัวอยางมาเพือ่ ใชในการพยากรณหรือการคาด การณ ซง่ึ สมการการถดถอยของตัวอยาง (Sample Regression Function) คอื [1] Yˆi = αˆ + βˆXi ; i = 1, 2, ..., n 3.1 การประมาณคาพารามเิ ตอรข องสมการถดถอย คาพารามเิ ตอร α และ β ซึง่ เปนคาท่ีไดมาจากประชากรหรอื เปนคา ที่หาไดยากมากเนือ่ งดว ย ขอ จำกัดดังกลา วไวแลวขา งตน ดวยเหตุนี้จึงตอ งมีการสำรวจตัวอยาง เพ่อื นำขอมูลตัวอยา งหรือขอ มลู เพยี งบางสวนของประชากรมาเพอื่ ใชในการพยากรณคาของตัวแปรตามหรือ Yˆ i โดยสามารถเขยี นได ดังน้ี Yˆ1 = αˆ + βˆiXi ; i = 1, 2, ..., n ในการหาตัวประมาณ α และ β นัน้ ในท่ีน้ีจะใชวิธีกำลังสองนอยสดุ (Ordinary Least Squares : OLS) ซ่งึ เปน วธิ ีที่ตองการตวั ประมาณ α และ β ท่ีทำใหผลบวกของคา คลาดเคล่ือนยก กำลังสองมคี านอ ยทีส่ ุด [1] เน่ืองจาก Yi = α + βXi + εi และ Yˆi = αˆ + βˆXi ดงั น้นั Yi − Yˆi = ei กำหนดให Q เปน คา ผลรวมกำลังสองของความคลาดเคล่ือน ท่ีนอ ยทสี่ ุด ไดดังน้ี ∑n ∑n ; i = 1, 2, ..., n (3.1) Q = (Yi − Yˆ )2 = (Yi − αˆ − βˆXi)2 i=1 i=1 โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร ครั้งท่ี 7 มหาวทิ ยาลัยราชภฏั นครสวรรค

76 ธนาภรณ อนิ เอย่ี ม เพื่อใหไดคา Q ทนี่ อ ยทีส่ ุดน้ันตองหาอนพุ ันธย อยเทยี บกับ αˆ และ βˆ โดยสามารถเขยี นสมการ ทั้งสองไดด[ังน∑nี้ ( )] ∂Q ∂ Yi − αˆ − βˆXi ∂αˆ = ∂αˆ i=1 ) ∑n ( = −2 Yi − αˆ − βˆX1 (3.2) ∂Q ∂ [i=1 ( − αˆ − βˆXi)2] ∂βˆ ∂βˆ ∑n = Yi i=1 ∑n ( ) Yi − αˆ − βˆXi Xi = −2 (3.3) i=1 จากน้นั ใหส มการท่ี (3.2) และ (3.3) มีคา = 0 ไดดังนี้ ∑n − 2 (Yi − αˆ − βˆXi) =0 (3.4) i=1 =0 (3.5) ∑n − 2 (Yi − αˆ − βˆXi)Xi จาก (3.4) i=1 ∑n (3.6) = αˆn+βˆ Xi ∑n Yi จาก (3.5) i=1 i=1 (3.7) ∑n ∑n ∑n XiYi = αˆ Xi + βˆ (Xi)2 i=1 i=1 i=1 แกสมการเพ่ือหา αˆ และ βˆ ∑n ∑n ∑n n XiYi − Xi Yi βˆ = i=1 i=(1∑n i=1)2 (3.8) ∑n n (Xi)2 − Xi จากสมการท่ี (3.6) คอื i=1 i=1 ∑n ∑n Yi = αˆn + βˆ Xi i=1 i=1 ∑n ∑n αˆn = Yi − βˆ Xi i=1 i=1 ∑n ∑n Yi βˆ Xi αˆ = i=1 − i=1 n n αˆ = Y¯ − βˆX¯ (3.9) ซง่ึ จะไดต ัวประมาณของพารามิเตอร αˆ และ βˆ เพอื่ นำไปสรางเปนสมการถดถอยเชิงเสน Yˆi = αˆ + βˆXi 3.2 คุณสมบัติของตวั ประมาณคา พารามิเตอร 1. ถา c เปนคา คงที่ แลว E(c) = c 2. ถา c เปนคาคงท่ี และ u เปน ฟงกชนั แลว E[cu(X)] = cE[u(X)] 3. ถา c1 และ c2 เปน คาคงที่ และ u1 และ u2 เปนฟงกชัน แลว โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณติ ศาสตร คร้งั ท่ี 7 มหาวทิ ยาลยั ราชภฏั นครสวรรค

การวิเคราะหก ารถดถอยโดยใชตัวอยา งแบบชุดลำดับ 77 E[c1u1(X) + c2u2(X)] = c1E[u1(X)] + c2E[u2(X)] [4] 4. ตัวแบบการวิเคราะหการถดถอยเชิงเสนอยางงายโดยใชการสมุ ตัวอยาง แบบชุดลำดับ(RSS) สมมติวามีการเรียงลำดบั ถกู ดำเนินการโดยคา เฉลีย่ ของตวั แปร X ซง่ึ สามารถเรยี งลำดบั และ วัดคาไดงา ย และตัวแปร X ซงึ่ มีความสัมพันธกบั ตัวแปร Y ในที่น้ีตัวแปร Y อาจเปนตวั แปรที่มีราคา แพงหรือยากตอการวดั คา [3] ให Xr(i)j และ Yr(i)j เปน ทกุ ๆ ลำดับ X ท่ีนอยท่ีสดุ ใน i และคาท่ีสอดคลอ งกันคา ของ Y ทีไ่ ดจากชดุ ตัวอยางที่ r ในรอบท่ี j จากนนั้ เรามี Y[r]i = α + βX(r)j + εrj (4.1) r = 1, 2, ..., k และ j = 1, 2, ..., m ขอสมมตเิ บอื้ งตน ของตวั แบบถดถอยเปนไปตามขอ สมมติเบ้อื ง ตน ของการวเิ คราะหถดถอยเชงิ เสนอยา งงาย โดยใชวิธกี ำลงั สองนอยทสี่ ุด ซงึ่ ไดตัวประมาณดังตอไปน้ี αˆRSS = Y¯RSS − βˆRSS X¯RSS (4.2) และ ∑m ∑k ( − X¯RSS ) X(r)j Y[r]j βˆRSS = j=1 r=1 − X¯RSS )2 (4.3) ∑m ∑k ( X(r)j j=1 r=1 โดยท่ี X¯RSS 1 ∑m ∑k และ Y¯RSS 1 ∑m ∑k เปนตวั อยางชดุ ลำดบั ตามลำดับ ดัง n rm = X(r)j = Y[r]j j=1 r=1 j=1 r=1 น้นั ตวั แบบตัวอยา งคอื Yˆ [r]j = αˆRSS + βˆRSS X¯(r)j (4.4) ตวั ประมาณคา พารามิเตอรจาก (4.2) และ (4.3) สามารถหาตวั ประมาณพารามิเตอร α และ β วิธกี ำลังสองนอยท่ีสุด (Ordinary Least Squares : OLS) ซึง่ เปน วธิ ที ตี่ องการตัวประมาณ α และ β ที่ทำใหผลบวกของคาคลาดเคล่ือนยกกำลงั สองมคี า นอ ยที่สดุ จาก (4.1) Y[r]j = α + βX(r)j + εrj และจาก (4.4) Yˆ[r]j = αˆRSS + βˆRSS X(r)j ดังนั้น erj = (Y[r]j − Yˆ[r]j ) กำหนดให Q เปน คา ผลรวมกำลงั สองของความคลาดเคล่ือนท่นี อ ยที่สุด ไดด งั น้ี ∑m ∑k (Y[r]j −Yˆ[r]j )2 = ∑m ∑k (Y[r]j −αˆRSS −βˆRSS X(r)j )2 Q= (4.5) j=1 r=1 j=1 r=1 ; r = 1, 2, ..., k, j = 1, 2, ..., m โครงการสัมมนาวชิ าการและนำเสนอผลงานดานคณิตศาสตร ครงั้ ท่ี 7 มหาวิทยาลัยราชภัฏนครสวรรค

78 ธนาภรณ อินเอยี่ ม เพื่อใหไ ดค า Q ท่นี อ ยทีส่ ุดนนั้ ตอ งหาอนุพันธยอ ยเทยี บกับตัวประมาณ αˆRSS และ βˆRSS โดย สามารถเขียนสมการทั้งสองไดดังนี้  ∑m ∑k ( − βˆRSS X(r)j )2 ∂Q = ∂  Y[r]j − αˆRSS ∂αˆRSS ∂αˆRSS ∂ [j=(1 r=1 − βˆRSS X(1)1)2 ( − βˆRSS X(2)1)2 ∂αˆRSS Y[2]1 = Y[1]1 − αˆRSS + − αˆRSS + ...+ ( )2 ] Y[k]m X(k)m − αˆRSS − βˆRSS = −2 ∑m ∑k (Y[r]j − αˆRSS − βˆRSS X(r)j ) (4.6) j=1 r=1  ∑m ∑k ( − αˆRSS − βˆRSS X(r)j )2 ∂Q = ∂  Y[r]j ∂βˆRSS ∂βˆRSS j[=(1 r=1 ∂ βˆRSS )2 ( βˆRSS X(2)1)2 = ∂βˆRSS Y[1]1 − αˆRSS − X(1)1 + Y[2]1 − αˆRSS − + ...+ ( )2 ] Y[k]m X(k)m − αˆRSS − βˆRSS = −2 ∑m ∑k (Y[r]j − αˆRSS − βˆRSS X(r)j )X(r)j (4.7) j=1 r=1 จากนนั้ ใหสมการที่ (4.6) และ (4.7) มคี า = 0 ไดด ังนี้ ∑m ∑k ( ) − 2 Y[r]j − αˆRSS − βˆRSS X(r)j =0 (4.8) j=1 r=1 ∑m ∑k ( ) − 2 Y[r]j − αˆRSS − βˆRSS X(r)j X(r)j =0 (4.9) จาก 4.8 j=1 r=1 ∑m ∑k ∑m ∑k ∑m ∑k ∑m ∑k ∑m ∑k Y[r]j − X(r)j (mk)αˆRSS − X(r)j βˆRSS X(r)j X(r)j = 0 j=1 r=1 j=1 r=1 j=1 r=1 j=1 r=1 j=1 r=1 (4.10) จาก 4.9 ∑m ∑k ∑m ∑k ∑m ∑k ∑m ∑k Y[r]j −(mk)αˆRSS X(r)j −(mk)βˆRSS (X(r)j )2 = 0 (mk) X(r)j j=1 r=1 j=1 r=1 j=1 r=1 j=1 r=1 (4.11) แกสมการเพ่ือหา αˆRSS และ βˆRSS ∑m ∑k X(r)j Y[r]j − X¯RSS ∑m ∑k Y[r]j βˆ = j=1 r=1 j=1 r=1 (4.12) ∑m ∑k (X(r)j )2 − X¯RSS ∑m ∑k X(r)j j=1 r=1 j=1 r=1 X¯เมอ่ื และ∑m ∑k1 Y¯RSS 1 ∑m ∑k mk mk = X(r)j = Y[r]j j=1 r=1 j=1 r=1 โครงการสมั มนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครั้งที่ 7 มหาวิทยาลัยราชภัฏนครสวรรค

การวิเคราะหการถดถอยโดยใชต วั อยา งแบบชดุ ลำดบั 79 αˆRSS = Y¯RSS − βˆRSS X¯RSS (4.13) ซง่ึ จะไดตวั ประมาณของพารามเิ ตอร αˆRSS และ βˆRSS เพื่อนำไปสรา งเปนสมการถดถอยเชิง เสน Yˆ[r]j = αˆRSS + βˆRSSXRSS คุณสมบตั ขิ องตัวประมาณคาพารามเิ ตอร 1.E(αˆRSS) = α, E(βˆRSS) = β เมือ่2.V ar(αˆRSS)= σe2 [ X¯R2 SS ] SR2 SS = 1 ∑m ∑k ( − X¯RSS )2 n 1 + E( SR2 SS ) n X(r)j n = mr j=1 r=1 ในที่น้ีไดศกึ ษาเฉพาะตัวประมาณคาของพารามิเตอร αRSS และ βRSS และคุณสมบัติของตวั ประมาณตามลำดบั จากคุณสมบตั ิของตวั ประมาณท่ีดีน้ันตอ งเปนตวั ประมาณที่มีความไมเอนเอียงและมีความ แปรปรวนตำ่ สดุ เราสามารถตรวจสอบคุณสมบัติของตัวประมาณคาพารามิเตอร αRSS วาเปน ตัว ประมาณท่ีไมเ อนเอียงไดด งั น้ี ∑m ∑k  () E(αˆRSS) = E Y¯RSS − βˆRSSX¯RSS  ∑m ∑k    βˆRSS  Y[r]j X(r)j E (αˆRSS ) = E j=1 r=1 − E j=1 r=1 mk mk      1 E  ∑m ∑k Y[r]j  − 1 βˆRSS  ∑m ∑k X(r)j  mk mk = j=1 r=1 j=1 r=1 = 1 ∑m ∑k ( ) 1 βˆRSS ∑m ∑k mk E Y[r]j − mk X(r)j j=1 r=1 j=1 r=1 = 1 ∑m ∑k [ ] 1 βˆRSS ∑m ∑k mk αRSS + βˆRSS X(r)j − mk X(r)j j=1 r=1 j=1 r=1 ∑m ∑k ∑m ∑k ∑m ∑k αRSS βˆRSS X(r)j βˆRSS X(r)j = j=1 r=1 + j=1 r=1 − j=1 r=1 mk mk mk ∑m ∑k αRSS = j=1 r=1 mk = αRSS สวนการตรวจสอบวา ตวั ประมาณของพารามเิ ตอร βRSS วา เปนตวั ประมาณที่มีความไมเอน เอยี งน้ันสามารถทำไดเ ชน เดยี วกนั กบั ตวั ประมาณของพารามิเตอร αRSS โครงการสมั มนาวชิ าการและนำเสนอผลงานดา นคณติ ศาสตร ครง้ั ท่ี 7 มหาวิทยาลยั ราชภฏั นครสวรรค

80 ธนาภรณ อนิ เอ่ยี ม เอกสารอางอิง [1] กลั ยา วานิชยบญั ชา, การวิเคราะหสถติ ิ: สถติ ิเพื่อการตัดสินใจ. พมิ พครัง้ ที่ 4. กรุงเทพฯ: โรง พมิ พ แหง จฬุ าลงกรณวทิ ยาลยั , 2542. [2] ชฎารัตน ถาปน , การประมาณคา แบบชวงสำหรับคาเฉลีย่ ประชากรภายใตการสุมตัวอยางแบบ ชดุ ลำดบั ไดดุล, มหาวทิ ยาลัยเชียงใหม, 2555. [3] M. Hani Samawi and M. Faisal Ababneh, “On regression analysis using ranked set sample”, Journal of Statistical Research, Vol. 35, No. 2, pp. 93-105, 2001. [4] R. Hogg and E. Tanis, Probability and Statistical inference, Fifth edition, 1992. [5] A. Kaur, G. Patil, A. Sinha and C. Taillie, Ranked set sampling: an annotated bibli- ography. Environmental and ecology statistics. No. 2, pp. 25-54, 1995. [6] G. A. McIntyre, A method of unbiased selective sampling, using ranked sets. Aus- tralian J. Agricultural Research No. 3, pp. 385-390, 1952. [7] H. A. Muttlak, Parameters Estimation in a simple linear regression using rank set sampling. Biom. J. Vol. 37, No. 7, pp. 799-810, 1955. [8] J. Neter, W. Wassrman, and M. Kutnre, Applied linear statistical models. Richard D. IRWIN, INC, Homewood, Illinois, 1985. [9] G. P. Patil, A. K. Sinha and C. Taaillie, Ranked set sampling: A Bibliography. Environ. Ecolog. Staatist. No. 6, pp. 1207-1211, 1999. โครงการสัมมนาวิชาการและนำเสนอผลงานดา นคณติ ศาสตร ครั้งที่ 7 มหาวทิ ยาลยั ราชภัฏนครสวรรค