Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi Linear systems resulting from pandiagonal magic squares Received: 26/9/2004 Accepted: 9/4/2006 Saleem Al-Ashhab* & Alla Al-Zahawi** ملخص ننسس ن عيرننم ن ثأرظ ننم ثني ننم يثنمننا م م ن د م نني يسراسب نسس ثن ميين ثت ثنان ي نا ج ي نم.مدثيسث ر سننرس ف ننا اننحث ثنما ن سسث ب ع ثن ع سالت يثن ج ايل يه .م ال ثنرظ م مش ل Abstract In this paper we consider linear systems of certain types, having arbitrary dimensions, and prove a rule about the number of free parameters in their solution set. 1. Introduction: A pandiagonal magic square is a n×n matrix a1 a2 ……… an an+1 an+2 ……… a2n …………. …………. …………. …………. …... an2 a n 2 n 1 a n 2 n 2 a , a ,..., a 2 n are distinct real numbers (usually integers) where the entries 1 2 satisfying the following system of equations: a 1 a 2 .......... a n a n 1 a n 2 ..... a 2n S S .... .... a n 2 n 1 a n 2 n 2 .. a n 2 S * ** ….………..(1 – a ) Assistant professor, department of mathmatics, Al-albayt University. Researcher in Mathematics. Al-Manarah, Vol. 13, No. 6, 2007 . 179 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi a1 a n 1 ......... a n 2 n 1 S a 2 a n 2 ......... a n 2 n 2 S ..... ..... a n a 2n .......... a n 2 S a1 a n 2 ............ a n 2 …...……….(1 – b) S a 2 a n 3 ............ a n 2 n 1 S ...... ..... a n a n 1 ............ a n 2 1 S ……………(1 – c) a 1 a 2n ........... a n 2 n 2 S a 2 a n 1 ........... a n 2 n 3 S .... .... a n a 2n -1 ........... a n 2 n 1 S ……………(1 – d) where S is a real constant (the so-called magic sum). In this system the group (1 – a) represents the summation of the entries in each row of the matrix. The group (1 – b) represents the summation of the entries in each column of the matrix. The group (1 – c) represents the summation of the entries in each right (extended) diagonal of the matrix. The group (1 – d) represents the summation of the entries in each left (extended) diagonal of the matrix. We will prove that the linear system (1) will have a solution, which contains n*n – 4*n + 3 free parameters, if n is odd, and n*n – 4*n + 4 free parameters, if n is even. Al-Manarah, Vol. 13, No. 6, 2007 . 180 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi 2. The case of odd pandiagonal magic square: We consider here the linear system (1), where n is of the form 2k+1 (k = 2, 3, …), since there is no pandiagonal magic square in the case k = 1. We illustrate first our theoretical approach by considering the case k = 2, i.e. We consider the square a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 In this case the system (1) takes the form: a1 a 2 a 3 a 4 a 5 S a 6 a 7 a 8 a 9 a 10 S ………..(2 – a ) a 11 a 12 a 13 a 14 a 15 S a 16 a 17 a 18 a 19 a 20 S a 21 a 22 a 23 a 24 a 25 S a 1 a 6 a 11 a 16 a 21 S a 2 a 7 a 12 a 17 a 22 S a 3 a 8 a 13 a 18 a 23 S .……….(2 – b) a 4 a 9 a 14 a 19 a 24 S a 5 a 10 a 15 a 20 a 25 S a 1 a 7 a 13 a 19 a 25 S a 2 a 8 a 14 a 20 a 21 S a 3 a 9 a 15 a 16 a 22 S …..……(2 – c) a 4 a 10 a 11 a 17 a 23 S a 5 a 6 a 12 a 18 a 24 S Al-Manarah, Vol. 13, No. 6, 2007 . 181 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi a 1 a 10 a 14 a 18 a 22 S a 2 a 6 a 15 a 19 a 23 S a 3 a 7 a 11 a 20 a 24 S ………(2 – d) a 4 a 8 a 12 a 16 a 25 S a 5 a 9 a 13 a 17 a 21 S In the system (2 – a), …, (2 – d), there are three equations, which are linearly dependent on the other equations. These equations are: the last equation of the group (2 – b), the last equation of the group (2 – c) and the last equation of the group (2 – d). Indeed, the last equation of the group (2 – b) is the result of subtraction of the sum of all other equations of the group (2 – b) from the sum of all equations of the group (2 – a). The last equation of the group (2 – c) is the result of subtraction of the sum of all other equations of the group (2 – c) from the sum of all equations of the group (2 – a). The last equation of the group (2 – d) is the result of subtraction of the sum of all other equations of the group (2 – d) from the sum of all equations of the group (2 – a). In order to prove that there are no other dependent equations, we write down the matrix of coefficients of the system after removing the previous mentioned equations: 1 1 1 1 1 1 1 1 1 1 1 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 1 1 1 1 0 1 0 1 0 1 0 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0 1 0 01 0 0 1 1 1 1 1 0 1 1 0 0 1 0 0 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0 0 1 0 1 0 0 1 1 1 1 1 1 0 1 0 0 1 0 0 0 1 Al-Manarah, Vol. 13, No. 6, 2007 . 182 ……(3) Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi We then prove that this matrix has full rank. To establish this we consider the transpose of this matrix: 1 1 1 1 1 1 1 1 0 0 0 1 1 1 0 0 1 1 1 1 0 1 1 0 1 1 0 1 1 0 1 0 0 1 0 1 1 1 1 1 0 10 0 1 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 1 0 1 0 1 1 1 1 0 1 0 1 0 1 and prove that this matrix has full rank. In order to do this we study the equation of linear dependence between the columns of the last matrix: b1 * C1 b 2 * C 2 ... b17 * C17 0 where C1 , C2 ,..., C17 denotes the columns of the matrix and b1 , b 2 ,..., b17 are real numbers. The componentwise form of this equation is: b1 b 6 b10 b14 0 b1 b 7 b11 b15 0 b1 b 8 b12 b16 0 b1 b 9 b13 b17 0 b1 0 Al-Manarah, Vol. 13, No. 6, 2007 . …………..(4 – a) 183 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi b 2 b 6 b15 0 b 2 b 7 b10 b16 0 …………..(4 – b) b 2 b 8 b11 b17 0 b 2 b 9 b12 0 b 2 b13 b14 0 b 3 b 6 b13 b16 0 b 3 b 7 b17 0 b 3 b 8 b10 0 b 3 b 9 b11 b14 0 b 3 b12 b15 0 …………..(4 – c) b 4 b 6 b12 b16 0 b 4 b 7 b13 0 b 4 b 8 b14 0 b 4 b 9 b10 b15 0 b 4 b11 b16 0 b 5 b 6 b11 0 b 5 b 7 b12 b14 0 b 5 b 8 b13 b15 0 b 5 b 9 b16 0 b 5 b10 b17 0 …………..(4 – d) …………..(4 – e) We conclude from the last equation of the group (4 – a) that b1 0 . Substituting this value in the other equations of the group (4 – a), and adding up leads to: 17 b j 6 j ………….(5) 0 Since the summation of the equations in (4 – b) yields 17 5b 2 b j 0 . j 6 Al-Manarah, Vol. 13, No. 6, 2007 . 184 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi Hence, we deduce from (5) that b 2 0 . In the same manner we obtain b3 b4 b5 0 . Using our knowledge about b 1 , b 2 ,..., b 5 we are capable of rewriting the system (4) like this: b 6 b 10 b 14 0 b 7 b 11 b 15 0 b 8 b 12 b 16 0 b 9 b 13 b 17 0 b 6 b 15 …………..(6 – a) 0 b 7 b 10 b 16 0 b 8 b 11 b 17 0 b 9 b 12 0 b 13 b 14 0 b 6 b 13 b 16 0 b 7 b 17 0 b 8 b 10 0 b 9 b 11 b 14 0 b 12 b 15 0 …………..(6 – b) …………..(6 – c) b 6 b 12 b 16 0 b 7 b 13 0 b 8 b 14 0 b 9 b 10 b 15 0 b 11 b 16 0 b 6 b 11 0 b 7 b 12 b 14 0 b 8 b 13 b 15 0 b 9 b 16 0 b 10 b 17 0 Al-Manarah, Vol. 13, No. 6, 2007 . …………..(6 – d) …………..(6 – e) 185 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi When comparing the left side of the first equation in (6 – b) with the left side of the first equation in (6 – e) we obtain b11 b15 In the same manner we get the following relations b 10 b 14 b 12 b 16 b 13 b 17 We substitute now these values in the system (6) obtaining the system b 6 2b 14 0 b 7 2b 15 0 b 8 2b 16 0 b 9 2b 17 0 ……………(7) If we compare the first equation in (7) with the first equation in (6 – b), we get b 15 2b 14 . If we compare the second equation in (7) with the second equation in (6 – c) and so on, we obtain the following relations b 14 2b 16 , b 16 2b 17 , b 17 2b 15 We can then rewrite them as the linear system 2b 14 b 15 0 2b 15 b 17 0 2b 16 b 14 0 2b 17 b 16 0 The matrix of coefficients of this system is 2 1 0 0 0 2 0 1 1 0 2 0 0 0 1 2 Al-Manarah, Vol. 13, No. 6, 2007 . 186 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi which is strictly diagonally dominant. Thus, it is invertible (cf. [1]) and, hence, the last linear system has the trivial solution, only. Due to the previous relations between the variables b1 , b 2 ,..., b17 we conclude that all of them are zero. Setting all variables a 1 , a 2 ,..., a 25 to S represents a solution of the 5 nonhomogenous linear system (2). Since we have 25 variables and 20 equations, the solution of the nonhomogenous linear system (2) has according to our analysis 25 – 20 + 3 = 8 free parameters (cf. [2]). In order to prove this result in general, we follow the same steps as in the case of 5x5 square. We have now to study the equation of linear dependence between the columns of the transpose matrix of (3) in general: b1 * C1 b 2 * C 2 ... b 4n -3 * C 4n -3 0 where C1 , C 2 ,..., C 4n - 3 denotes the columns of the matrix and b1 , b 2 ,..., b 4n - 3 are real numbers. The componentwise form of this equation is: b1 b n 1 b 2n b 3n -1 b1 b n 2 b 2n 1 b 3n .... 0 0 b1 b 2n 1 b 3n - 2 b 4n - 3 0 b1 0 .………..(8 – 1) b 2 b n 1 b 3n 0 b 2 b n 2 b 2n b 3n 1 0 ........ b 2 b 2n -1 b 3n -3 0 b2 b 3n -2 b 3n -1 0 .………..(8 – 2) … … … Al-Manarah, Vol. 13, No. 6, 2007 . 187 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi b n b n 1 b 2n 1 0 b n b n 2 b 2n 2 b 3n -1 0 .... b n b 2n -1 bn b 4n - 4 0 b 2n b 4n - 3 0 ..…………..(8 – n) We conclude from the last equation of the group (8 – 1) that b1 0 . Substituting this value in the other equations of the group (8 – 1) and adding up leads to: 4n - 3 bj 0 j n 1 Since the summation of the equations in (8 – 2) yields nb 2 4n - 3 bj 0 j n 1 we deduce that b 2 0 . In the same manner we obtain: b 3 b 4 .... b n 0 Using our knowledge about b1 , b 2 ,..., b n we are capable of rewriting the system (8) like this: b n 1 b 2n b 3n-1 0 b n 2 b 2n1 b 3n 0 ..…………..(9 – 1) .... b 2n1 b 3n-2 b 4n-3 0 b n 1 b 3n 0 b n 2 b 2n b 3n1 0 ........ b 2n-1 b 3n- 3 ..…………..(9 – 2) 0 b 3n- 2 b 3n-1 0 … … … Al-Manarah, Vol. 13, No. 6, 2007 . 188 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi b n 1 b 2n1 0 b n 2 b 2n 2 b 3n-1 0 ..………….. (9 – n) .... b 2n-1 b 4n-4 0 b 4n-3 0 b 2n When comparing the left side of the first equation in (9 – 2) with the left side of the first equation in (9 – n) we obtain b 2n 1 b 3n In the same manner we get the following system b 2n b 3n 1 b 2n 1 b 3n ...... ...... b 3n 2 b 4n 3 We substitute now these values in the system (9) obtaining the system b n 1 2b 3n -1 0 b n 2 2b 3n 0 .... b 2n -1 2b 4n - 3 0 ..………….. (10) If we compare the first equation in (10) with the first equation in (9 – 2), we get b 3n 2b 3n -1 . If we compare the second equation in (10) with the second equation in (9 – 3) and so on, we obtain the following system Al-Manarah, Vol. 13, No. 6, 2007 . 189 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi 2b 3n 1 b 3n 2b 3n 0 b 3n 2 0 ...... 2b 7n 5 b 4n 3 0 2 2b 7n 3 b 3n 1 0 2 2b 7n 1 b 3n 1 0 2 ...... 2b 4n 3 b 4n 4 0 The matrix of coefficients of this system is 0 0 0 2 1 0 0 0 0 0 0 0 2 0 1 0 ..... 0 0 0 0 0 - 2 0 0 0 0 0 1 0 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 0 0 0 0 0 0 2 0 0 0 0 0 0 0 1 0 0 0 0 0 2 0 0 0 0 ....... 0 0 0 0 1 2 0 0 ……(11) which is strictly diagonally dominant. Thus, it is invertible (cf. [1]) and, hence, the last linear system has the trivial solution, only. Due to the previous relations between the variables b1 , b 2 ,..., b 4n - 3 we conclude that all of them are zero. Setting all variables a1 , a 2 ,..., a n 2 to S represents a solution of the n nonhomogenous linear system (1). Since we have n*n variables and 4*n equations, the solution of the nonhomogenous linear system (1) has according to our analysis n*n – 4*n + 3 free parameters (cf. [2]). Al-Manarah, Vol. 13, No. 6, 2007 . 190 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi 3. The case of double-even pandiagonal magic square: We consider the linear system (1 – a), (1 – b), (1 – c) and (1 – d), where n is of the form 2k (k = 2, 4, …). In this system there are four equations, which are linearly dependent on the other equations. These equations are the same three equations as in the odd squares beside the (n-1)th equation of the group (1 – d), which is the result of subtraction of the sum of all other odd-ranked equations of the group (1 – d) from the sum of all odd-ranked equations of the group (1 – a). Since k = 2 yields a very special case, we start illustrating it: As in the case of odd squares, we consider the matrix of coefficients of the system after removing the previous mentioned equations. This matrix will be the same as the matrix (3) after deleting the last row. We then consider the transpose of this matrix, which has now one column less. We prove that this matrix has full rank by studying the equation of linear dependence between the columns of the matrix: b1 * C1 b 2 * C2 ... b12 * C12 0 where C1 , C2 ,..., C12 denotes the columns of the matrix and b1 , b 2 ,..., b12 are real numbers. The component wise form of the last equation is: b 1 b 5 b 8 b 11 0 b 1 b 6 b 9 b 12 0 b 1 b 7 b 10 0 b1 0 ……………..(12 – 1) b 2 b 5 b 12 0 b2 b6 b8 0 b2 b7 b9 0 .……………..(12 – 2) b 2 b 10 b 11 0 b 3 b 5 b 10 0 b3 b6 0 b 3 b 7 b 8 b 11 0 b 3 b 9 b 12 .……………..(12 – 3) 0 Al-Manarah, Vol. 13, No. 6, 2007 . 191 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi b4 b5 b9 0 b 4 b 6 b 10 b 11 0 b 4 b 7 b 12 0 b4 b8 0 …….……....(12 – 4) We conclude from the last equation of the group (12 – 1) that b1 0 . Substituting this value in the other equations of the group (12 – 1) and adding up leads to: 12 b j 5 j 0 ….........……….(13) Since the summation of the equations in (12 – 2) yields 12 4b 2 b j 0 j5 we deduce from (13) that b 2 0 . In the same manner we obtain: b3 b4 0 . Using our knowledge about b 1 , b 2 , b 3 and b 4 we are capable of rewriting the system (12) like this: b 5 b 8 b 11 0 b 6 b 9 b 12 0 b 7 b 10 0 b 5 b 12 0 b6 b8 0 b7 b9 0 ….……………..(14 – 1) ….……………..(14 – 2) b 10 b 11 0 b 5 b 10 0 b6 0 b 7 b 8 b 11 0 b 9 b 12 ….……………..(14 – 3) 0 Al-Manarah, Vol. 13, No. 6, 2007 . 192 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi b5 b9 0 b 6 b 10 b 11 0 b 7 b 12 0 b8 0 ….……………..(14 – 4) From the group (14 – 3) and the group (14 – 4 ) we conclude that b 6 0 and b 8 0 . When we substitute these values in the system (14) and do some comparisons, e. g. when equating the left side of the first equation in each of the groups (14 – 1), …, (14 – 4) we obtain: b 9 b 10 b 11 b 12 ………………....(15) Doing the same with the second equation we get: b 9 b 12 b 10 b 11 0 ………………..(16) We rewrite the equations in (16) in the following manner: b 12 b 9 , b 10 b 11 Using these relations, we obtain from (15) the equation 2b 9 0 . Thus, b 9 0 . Due to the previous relations between the variables b1 , b 2 ,..., b12 we conclude that all of them are zero. Setting all variables a 1 , a 2 ,..., a 16 to S represents a solution of the 4 nonhomogenous linear system (1). Since we have 16 variables and 16 equations, the solution of the nonhomogenous linear system (1) has according to our analysis 16 – 16 + 4 = 4 free parameters (cf. [2]). Now, we treat the general case. As in the case of odd squares we consider the matrix of coefficients of the system after removing the previous mentioned dependent equations. This matrix will be the same as the matrix (3) after deleting the last row. We then consider the transpose of this matrix, which has now one column less. We prove that this matrix has full rank by studying the equation of linear dependence between the columns of the matrix: b1 * C1 b 2 * C2 ... b 4n - 4 * C4n - 4 0 .……………(17) Al-Manarah, Vol. 13, No. 6, 2007 . 193 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi where C1 , C2 ,..., C4n - 4 denotes the columns of the matrix and b1, b 2 ,..., b 4n - 4 are real numbers. The componentwise form of equation (17) is: b1 b n 1 b 2n b 3n -1 b1 b n 2 b 2n 1 b 3n 0 0 .... b1 b 2n - 2 b 3n - 3 b 4n - 4 0 b1 b 2n -1 b 3n - 2 0 0 b1 b 2 b n 1 b 3n b 2 b n 2 b 2n .... b 3n 1 0 0 b 3n - 2 b 3n -1 0 b2 ……………..(18 – 1) .…………..…..(18 – 2) … … … b n b n 1 b 2n 1 0 b n b n 2 b 2n 2 b 3n -1 0 .... b 2n bn 0 …….…………..(18 – n) We conclude from the last equation of the group (18 – 1) that b1 0 . Substituting this value in the other equations of the group (18 – 1) and adding up leads to: 4n - 4 bj 0 j n 1 ……...………….(19) Since the summation of the equations in (18 – 2) yields nb 2 4n - 4 bj 0 j n 1 Al-Manarah, Vol. 13, No. 6, 2007 . 194 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi Hence, we deduce from (19) that b 2 0 . In the same manner we obtain: b 3 b 4 .... b n 0 . Using our knowledge about b1 , b 2 ,..., b n we are capable of rewriting the system (18) like this: b n 1 b 2n b 3n -1 b n 2 b 2n 1 b 3n 0 0 .... b 2n - 2 b 3n - 3 b 4n - 4 0 b 2n -1 b 3n - 2 0 b n 1 b 3n b n 2 b 2n .... b 3n 1 0 ….……………..(20 – 1) 0 b 3n - 2 b 3n -1 0 ...….…………..(20 – 2) … … … b n 1 b 2n 1 0 b n 2 b 2n 2 b 3n -1 0 .... b 2n 0 From the group (20 …….…………..(20 – n) n ) and the group (20 – n) we conclude that b 3 0 2 n 2 and b 2n 0 . When we substitute these values in the system (20) and do some comparisons, e. g. when equating the left side of the first equation in each of the groups (20 – 1), …, (20 – n) we obtain: b 2n 1 b 2n 2 b 2n 3 b 4n - 4 b 2n 4 b 4n - 5 .... b 3n - 2 b 3n 1 b 3n b 3n -1 Al-Manarah, Vol. 13, No. 6, 2007 . 195 ………………..(21) Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi Doing the same with the second equation, we get: b 2n 1 b 3n b 2n 2 b 3n -1 b 2n 3 b 2n 4 b 2n 5 b 4n - 4 b 2n 6 b 4n - 5 .... b 3n - 2 b 3n 3 b 3n 2 b 3n 1 Continuing these comparisons till, we reach the ….(22) n th equation in each of the 2 groups (20 – 1), …, (20 – n) we get: b 2n 1 b 4n - 4 b 2n 2 b 4n - 5 ... b 3n - 3 b 3n -1 b 2n 0 …...(23) Remember that in the case k = 2 this equation will be also the second equation, and that we have to deal with two equations, only. We rewrite the equations in (23) in the following manner: b 3n -1 b 3n - 2 b 3n - 3 b 3n .... .... b 4n - 4 b 2n 1 ………...(24) Using these relations, we obtain from the equations in (21), (22) and similar equations the following linear system: 2b 2n 1 b 2n 3 0 2b 2n 3 b 2n 7 0 ...... 2b 5 2 2b 5 2 2b 5 2 2b 5 2 n 3 n 1 n 1 n 3 b 3n-5 0 0 b 2n 3 0 b 2n 7 0 ...... 2b 3n-3 b 3n-5 0 ………………...(25) Al-Manarah, Vol. 13, No. 6, 2007 . 196 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi The matrix of coefficients of this system is 1 0 0 0 0 0 0 2 -2 0 1 0 0 0 0 0 ...... 0 0 0 0 0 2 0 0 0 0 0 1 0 0 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0 0 0 0 0 0 0 -2 0 0 0 0 0 0 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0 1 0 0 0 0 0 0 -2 0 0 0 0 0 0 0 1 0 0 0 0 0 -2 0 0 0 0 0 ...... 0 0 0 0 0 0 1 2 which is strictly diagonally dominant. Thus, it is invertible and, hence, the last linear system has the trivial solution, only. Due to the previous relations between the variables b1, b 2 ,..., b 4n - 4 we conclude that all of them are zero. Setting all variables a1 , a 2 ,..., a n 2 to S represents a solution of the n nonhomogenous linear system (1). Since we have n*n variables and 4*n equations, the solution of the nonhomogenous linear system (1) has according to our analysis n*n – 4*n + 4 free parameters (cf. [2]). 4. The case of simple even pandiagonal magic square: We consider the linear system (1), where n is of the form 2k (k = 3, 5, …): Following the same steps as in the case of double-even squares we reach the same conclusion about the values of b1 , b 2 ,..., b n . Further, we obtain the system (20) again for their values, from which we deduce the fact that b 3 2 n 0 and b 2n 0 . Now, by comparing the left side of the equations in the groups (20 – 1), (20 – 2), …, (20 – n) we get the same relations (21), (22) and other similar relations. But, the comparisons of the n th equation in each group yields now: 2 b 2n1 b 4n-4 b 2n 2 b 4n-5 ... b 3n-2 b 3n-1 0 Al-Manarah, Vol. 13, No. 6, 2007 . 197 ……..……....(26) Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi We rearrange the equations in (26) as follows: b 3n-1 b 3n-2 b 3n-3 b 3n .... .... b 4n-4 b 2n 1 We substitute these relations in the sets of equations (21), (22) and the similar sets of equations. This will generate a different linear system than in the case of double-even squares. In fact, we obtain now the following linear system: 2b 2n 1 b 2n 3 0 2b 2n 3 b 2n 7 0 ...... 2b 5 2 2b 5 2 2b 5 2 n 2 n n2 b 3n-3 0 b 2n 1 0 b 2n 5 0 ...... 2b 3n-3 b 3n-5 0 When we write the coefficients of this system in the form of a matrix, we obtain the matrix (11). This is a strictly diagonally dominant matrix. Like in the first case we are so done with the proof. Notice: This paper is extracted from the Master thesis “A Study on Pandiagonal Magic Squares” by the student Alla Al - zahawi and supervisor Saleem Alashhab, Al-albayt University, May 2004. Al-Manarah, Vol. 13, No. 6, 2007 . 198 Linear systems resulting from……......….………..Saleem Al-Ashhab & Alla Al-Zahawi References: [1] W.S. Andrews, Magic Squares and Cubes, 1908. [2] W.W. Rouse Ball, Mathematical Recreations and Essays, 1911. [3] Anton H., Elementary Linear Algebra, 7th edition, Jon Wiley, New York, 1994. [4] Horen, R. and Johnson, C., Matrix Analysis, 9th edition, Cambridge University Press, New York, 1996. [5] S. Al-ashhab, The Magic and Semi Magic 4x4 Squares Problem, Dirasat; October 1998,Volume 25, No. 3, pp 445-450. [6] S. Al-ashhab, The book “Theory of Magic Squares / Mathematically and Programming” in Arabic, 2000. Al-Manarah, Vol. 13, No. 6, 2007 . 199
© Copyright 2024 Paperzz