Hindawi Publishing Corporation Mathematical Problems in Engineering Volume 2013, Article ID 321032, 9 pages http://dx.doi.org/10.1155/2013/321032
Research Article Inversion of General Cyclic Heptadiagonal Matrices
A. A. Karawia
Mathematics Department, Faculty of Science, Mansoura University, Mansoura 35516, Egypt
Correspondence should be addressed to A. A. Karawia; [email protected]
Received 23 December 2012; Revised 26 February 2013; Accepted 27 February 2013
AcademicEditor:JoaoB.R.DoVal
Copyright Β© 2013 A. A. Karawia. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We describe a reliable symbolic computational algorithm for inverting general cyclic heptadiagonal matrices by using parallel 2 computing along with recursion. The computational cost of it is 21π β 48π β 88 operations. The algorithm is implementable to the Computer Algebra System (CAS) such as MAPLE, MATLAB, and MATHEMATICA. Two examples are presented for the sake of illustration.
1. Introduction heptadiagonal matrices of the form (1) and for solving linear systems of the form: The πΓπgeneral cyclic heptadiagonal matrices take the form: π»π=π , (2)
π π π1 π1 π΄1 πΆ1 0 0 β β β 0 π΅1 π1 where π=(π₯1,π₯2, ..., π₯π) and π =(π1,π2, ..., ππ) . To the best of our knowledge, the inversion of a general π π π π΄ πΆ 0 β β β β β β β β β π΅ ( 2 2 2 2 2 2 ) ( ) cyclic heptadiagonal matrix of the form (1)hasnotbeen ( ) ( π΅ π π π π΄ πΆ 0 β β β β β β 0 ) considered. Very recently in [5], the inversion of a general ( 3 3 3 3 3 3 ) ( ) ( ) cyclic pentadiagonal matrix using recursion is studied with- ( ) ( . ) out imposing any restrictive conditions on the elements of ( π· π΅ π π π π΄ πΆ 0 ... . ) ( 4 4 4 4 4 4 4 . ) ( . ) thematrix.Also,inthispaperwearegoingtocompute ( . ) ( 0 dddddddd . ) the inverse of a general cyclic heptadiagonal matrix of the π»=( ) , ( . . ) ( . . ) form (1) without imposing any restrictive conditions on the ( . dddddddd . ) ( 0 β β β 0 π· π΅ π π π π΄ πΆ ) elements of the matrix π» in (1). Our approach is mainly ( πβ3 πβ3 πβ3 πβ3 πβ3 πβ3 πβ3 ) ( ) ( ) basedongettingtheelementsofthelastfivecolumnsof ( 0 β β β β β β 0 π· π΅ π π π π΄ ) β1 ( πβ2 πβ2 πβ2 πβ2 πβ2 πβ2) π» ( ) in suitable forms via the Doolittle LU factorization [10] ( ) (π΄ 0 β β β β β β 0 π· π΅ π π π ) along with parallel computation [7]. Then the elements of πβ1 πβ1 πβ1 πβ1 πβ1 πβ1 β1 the remaining (π β 5) columns of π» may be obtained π π΄ 0 β β β β β β 0 π· π΅ π π π π π π π π using relevant recursive relations. The inversion algorithm ( ) (1) of this paper is a natural generalization of the algorithm presented in [5]. The development of a symbolic algorithm is considered in order to remove all cases where the numerical algorithm fails. Many algorithms for solving banded linear where πβ₯7. systems need to pivoting, for example Gaussian elimination The inverses of cyclic heptadiagonal matrices are usually algorithm [10β12]. Overall, pivoting adds more operations required in science and engineering applications, for more to the computational cost of an algorithm. These additional details, see special cases, [1β9]. The motivation of the current operations are sometimes necessary for the algorithm to work paper is to establish efficient algorithms for inverting cyclic at all. 2 Mathematical Problems in Engineering
π΄ { πβ1 The paper is organized as follows. In Section 2,new { , if π=1, { πΌ1 symbolic computational algorithm, that will not break, is { π π { 1 1 constructed. In Section 3, two illustrative examples are given. { , if π=2, { πΌ2 Conclusions of the work are given in Section 4. { π π +π π { 2 2 1 1 { , if π=3, { πΌ3 { π π +π π +π πΆ { πβ1 πβ1 πβ2 πβ2 πβ3 πβ3 2. Main Results { , if π=4(1) πβ5, π = πΌ π { π { In this section we will focus on the construction of new { π·πβ1 βππβ5ππβ5 βππβ6ππβ6 βππβ7πΆπβ7 { , π=πβ4, { πΌ if symbolic computational algorithms for computing the deter- { πβ4 { minant and the inverse of general cyclic heptadiagonal { π΅ βπ π βπ π βπ πΆ { πβ1 πβ4 πβ4 πβ5 πβ5 πβ6 πβ6 , π=πβ3, matrices. The solution of cyclic heptadiagonal linear systems { πΌ if { πβ3 of the form (2) will be taken into account. Firstly we begin { { π βπ π βπ π βπ πΆ with computing the πΏπ factorization of the matrix π».Itisas { πβ1 πβ3 πβ3 πβ4 πβ4 πβ5 πβ5 , π=πβ2, πΌ if in the following: { πβ2 π { π , π=1, { πΌ if { 1 π»=πΏπ, (3) { (β π βπ΄ ) { 1 1 π {β , if π=2, { πΌ2 { (β π +β π ) where { 2 2 1 1 {β , if π=3, { πΌ { 3 1 0 β β β β β β β β β β β β β β β β β β β β β 0 { (βπβ1ππβ1 +βπβ2ππβ2 +βπβ3πΆπβ3) {β , if π=4(1) πβ4, π2 1 0 β β β β β β β β β β β β β β β β β β 0 πΌ βπ ={ π π3 π3 1 0 β β β β β β β β β β β β β β β 0 { ( π· ) { π·π ββπβ4ππβ4 ββπβ5ππβ5 ββπβ6πΆπβ6 ( 4 π π 1 0 β β β β β β β β β β β β 0) { , if π=πβ3, ( 4 4 ) { πΌ ( πΌ1 ) { πβ3 ( ) { ( . .) { π΅π ββπβ3ππβ3 ββπβ4ππβ4 ββπβ5πΆπβ5 ( . dd d d d d d d .) { , π=πβ2, ( . .) { πΌ if ( ) { πβ2 πΏ=( . .) , { ( . dd d d d d d d .) { π ββπβ2 β π€ ( π· ) { π π=1 π π ( 0β β β 0 πβ3 π π 1 dd0) { , if π=πβ1, ( πβ3 πβ3 ) { πΌ ( πΌπβ6 ) πβ1 ( π· ) ( 0 β β β β β β 0 πβ2 π π 1 d 0) π , π=1, ( πβ2 πβ2 ) { 1 if πΌπβ5 {βπ V +π΅ , π=2, π π π β β β β β β π π π 10 { 2 1 2 if 1 2 3 πβ2 πβ1 πβ2 { β β β β β β β β β β β β β β β 1 {βπ3V2 βπ3V1, if π=3, 1 2 3 πβ3 πβ2 πβ1 { { π·π ( ) {βππVπβ1 βππVπβ2 β Vπβ3, if π=4(1) πβ4, { πΌ (4) { πβ3 { π·πβ3 V = βππβ3Vπβ4 βππβ3Vπβ5 β Vπβ6 +πΆπβ3, if π=πβ3, πΌ π π πΆ 00β β β 0 π€ V π { πΌ 1 1 1 1 1 1 { πβ6 0πΌ2 π2 π2 πΆ2 0β β β 0 π€2 V2 { { π·πβ2 00πΌ3 π3 π3 πΆ3 β β β 0 π€3 V3 {βπ V βπ V β V +π΄ , π=πβ2, ( 000πΌπ π d 0π€ V ) { πβ2 πβ3 πβ2 πβ4 πΌ πβ5 πβ2 if ( 4 4 4 4 4 ) { πβ5 ( ) { ( . . ) { πβ2 ( . ddddd d d d . ) { ( ) {ππβ1 β βππVπ, if π=πβ1, π=( . . ) . (5) ( . ddddd d d d . ) { π=1 ( . . ) ( 0 β β β β β β β β β β β β 0 πΌπβ3 ππβ3 π€πβ3 Vπβ3) π΅ , π=1, ( 0 β β β β β β β β β β β β β β β 0 πΌ π€ V ) { 1 if πβ2 πβ2 πβ2 {βπ π€ , π=2, 0 0 β β β β β β β β β β β β β β β 0 πΌ V { 2 1 if πβ1 πβ1 {βπ π€ βπ π€ , π=3, 0 0 0 β β β β β β β β β β β β β β β 0 πΌ { 3 2 3 1 if π { π· ( ) {βππ€ βππ€ β π π€ , π=4(1) πβ5, { π πβ1 π πβ2 πβ3 if { πΌπβ3 { π· π€ = βπ π€ βπ π€ β πβ4 π€ +πΆ , π=πβ4, The elements in the matrices L and U in (4)and(5)satisfy π { πβ4 πβ5 πβ4 πβ6 πΌ πβ7 πβ4 if { πβ7 { { π· {βπ π€ βπ π€ β πβ3 π€ +π΄ , π=πβ3, { πβ3 πβ4 πβ3 πβ5 πΌ πβ6 πβ3 if π , π=1, { πβ6 { 1 if { {π βππ , π=2, { π·πβ2 { 2 2 1 if βπ π€ βπ π€ β π€ +π , π=πβ2, {π βππ βπ π , π=3, { πβ2 πβ3 πβ2 πβ4 πΌ πβ5 πβ2 if { 3 3 2 3 1 if πβ5 { π·βπ {π βππ βππ β πΆ , π=4(1) πβ2, π2 { π π πβ1 π πβ2 πΌ πβ3 if { , if π=2, πΌ = πβ3 { πΌ π { πβ2 { 1 { { {ππβ1 β βπ€πππ, if π=πβ1, { (π βπ π ) { 3 3 1 , π=3, { π=1 ππ = { if { πβ1 { πΌ2 { { {π β βV β , π=π, { (π βππ β(π·/πΌ )π ) π π π if { π π πβ2 π πβ3 πβ3 { π=1 , if π=4(1) πβ2, { πΌπβ1 Mathematical Problems in Engineering 3
π΅3 { , π=3, Set π3 =β(π1 βπ1 +π2 βπ2)/πΌ3 { πΌ if π = 1 π { (π΅ β(π·/πΌ )π ) β =β(β βπ +β βπ )/πΌ { π π πβ3 πβ3 Set 3 1 1 2 2 3 , if π=4(1) πβ2, { πΌπβ2 Set V3 =βπ3 β V1 βπ3 β V2
Set π€3 =βπ3 βπ€2 βπ3 βπ€1 {π1, if π=1, ππ = {π2 βπ2π1, if π=2, {ππ βππππβ1 βπππΆπβ2, if π=3(1) πβ3, Step 2. Compute and simplify. π 4 πβ2 π΄1, if π=1, For from to do ππ ={ π΄π βπππΆπβ1, if π=2(1) πβ4. π =(π΅ βπ· βπ /πΌ )/πΌ (6) π π π πβ3 πβ3 πβ2 ππ =(ππ βπ·π βππβ3/πΌπβ3 βππ βππβ2)/πΌπβ1 We also have: ππ β2=π΄πβ2 βππβ2 βπΆπβ3 π ππβ1 =ππβ1 βππβ1 βππβ2 βππβ1 βπΆπβ3 Det (π») = βπΌπ. (7) π=1 πΌπ =ππ βπ·π βπΆπβ3/πΌπβ3 βππ βππβ2 βππ βππβ1
Remark 1. It is not difficult to prove that the LU decomposi- If πΌπ =0then πΌπ =π‘end if tion (3) exists only if πΌπ =ΜΈ 0, π = 1(1)π β1 (pivoting elements). Moreover the cyclic heptadiagonal matrix π» of the form (1) End do has an inverse if, in addition, πΌπ =0ΜΈ .Pivotingcanbeomitted by introducing auxiliary parameter π‘ in Algorithm 1 given Step 3. Compute and simplify. later. So no pivoting is included in our algorithm. For π from 4 to πβ5do
At this point it is convenient to formulate our first result. π =β(π βπΆ +π βπ +π βπ )/πΌ It is a symbolic algorithm for computing the determinant π πβ3 πβ3 πβ2 πβ2 πβ1 πβ1 π of a cyclic heptadiagonal matrix π» of the form (1)and π€π =β(π·π βπ€πβ3/πΌπβ3 +ππ βπ€πβ2 +ππ βπ€πβ1) can be considered as natural generalization of the symbolic algorithm DETCPENTA in [5]. End do
Algorithm 1. To compute Det(π») for the cyclic heptadiagonal π» matrix in (1), we may proceed as follows. Step 4. Compute and simplify.
Step 1. Set πΌ1 =π1 For π from 4 to πβ4do If πΌ1 =0then πΌ1 =π‘(π‘is just a symbolic name) end if βπ =β(βπβ3 βπΆπβ3 +βπβ2 βππβ2 +βπβ1 βππβ1)/πΌπ Set π1 =π1, π1 =π΄1 Vπ =β(π·π β Vπβ3/πΌπβ3 +ππ β Vπβ2 +ππ β Vπβ1) Set π1 =π΄πβ1/πΌ1 Set V1 =π1 End do Set π€1 =π΅1 Step 5. Compute simplify. Set β1 =ππ/πΌ1
Set π€1 =π΅1 ππβ4 =(π·πβ1βππβ5βππβ5βππβ6βππβ6βππβ7βπΆπβ7)/πΌπβ4 Set π2 =π2/πΌ1 ππβ3 =(π΅πβ1βππβ4βππβ4βππβ5βππβ5βππβ6βπΆπβ6)/πΌπβ3 Set π3 =π΅3/πΌ1 ππβ2 =(ππβ1βππβ3βππβ3βππβ4βππβ4βππβ5βπΆπβ5)/πΌπβ2 Set πΌ2 =π2 βπ2 βπ1 π€πβ4 =πΆπβ4βπ·πβ4βπ€πβ7/πΌπβ7βππβ4βπ€πβ6βππβ4βπ€πβ5 If πΌ2 =0then πΌ2 =π‘end if π€πβ3 =π΄πβ3βπ·πβ3βπ€πβ6/πΌπβ6βππβ3βπ€πβ5βππβ3βπ€πβ4 Set π2 =βπ1 βπ1/πΌ2 π€πβ2 =ππβ2βπ·πβ2βπ€πβ5/πΌπβ5βππβ2βπ€πβ4βππβ2βπ€πβ3 Set V2 =π΅2 βπ2 β V1 βπβ3 =(π·πββπβ4βππβ4ββπβ5βππβ5ββπβ6βπΆπβ6)/πΌπβ3 Set π€2 =βπ2 βπ€1 βπβ2 =(π΅πββπβ3βππβ3ββπβ4βππβ4ββπβ5βπΆπβ5)/πΌπβ2 Set β2 =(π΄π ββ1 βπ1)/πΌ2 Vπβ3 =πΆπβ3 βπ·πβ3 βVπβ6/πΌπβ6 βππβ3 βVπβ5 βππβ3 βVπβ4 Set π2 =π2 βπ2 βπ1 V =π΄ βπ· βV /πΌ βπ βV βπ βV π =(π βπ βπ )/πΌ πβ2 πβ2 πβ2 πβ5 πβ5 πβ2 πβ4 πβ2 πβ3 Set 3 3 3 1 2 πβ2 Vπβ1 =ππβ1 ββπ=1 ππVπ Set πΌ3 =π3 βπ3 βπ1 βπ3 βπ2 πβ2 If πΌ3 =0then πΌ3 =π‘end if πΌπβ1 =ππβ1 ββπ=1 π€πππ 4 Mathematical Problems in Engineering
00000 If πΌπβ1 =0then πΌπβ1 =π‘end if 00000 β =(π β βπβ2 β π€ )/πΌ (. . . . .) πβ1 π π=1 π π πβ1 (. . . . .) ( ) (. . . . .) (. . . . .) πΌ =π β βπβ1 V β ( ) π π π=1 π π = (. . . . .) , (. . . . .) (. . . . .) (00001) If πΌπ =0then πΌπ =π‘end if ( ) (00010) π 00100 Step 6. Compute Det(π») =β ( πΌπ)π‘=0. π=1 01000 The symbolic Algorithm 1 will be referred to as 10000 DETCHEPTA. The computational cost of this algorithm ( ) is 52π β 195 operations. The new algorithm DETCHEPTA (9) is very useful to check the nonsingularity of the matrix π» we get when we consider, for example, the solution of the cyclic (π) (πβ1) (πβ2) (πβ3) (πβ4) heptadiagonal linear systems of the form (2). π1 π1 π1 π1 π1 Now, when the matrix π» is nonsingular, its inversion is (π) (πβ1) (πβ2) (πβ3) (πβ4) ( π2 π2 π2 π2 π2 ) ( ) computed as follows. ( ) ( ) Let ( . . . . . ) ( . . . . . ) ( ) ( . . . . . ) π»β1 =[π ] =[ , ,..., ], ( . . . . . ) ππ 1β€π,πβ€π Col1 Col2 Colπ (8) ( ) ( . . . . . ) ( . . . . . ) ( ) (π(π) π(πβ1) π(πβ2) π(πβ3) π(πβ4)) π π»β1 π=1,2,...,π ( πβ4 πβ4 πβ4 πβ4 πβ4 ) where Colπ denotes th column of , . ( ) πΏπ π» ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) Since the Doolittle factorization of the matrix in (π π π π π ) ( πβ3 πβ3 πβ3 πβ3 πβ3 ) (1)isalwayspossiblethenwecanuseparallelcomputa- ( ) = ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) tionstogettheelementsofthelastfivecolumnsColπ (π π π π π ) (π ,π ,...,π )π π=π,πβ1,πβ2,πβ3 πβ4 π»β1 ( πβ2 πβ2 πβ2 πβ2 πβ2 ) 1,π 2,π π,π , ,and of ( ) as follows [5]. (π) (πβ1) (πβ2) (πβ3) (πβ4) ππβ1 ππβ1 ππβ1 ππβ1 ππβ1 (10) Solving in parallel the standard linear systems whose coefficient matrix πΏ is given by (4) (π) (πβ1) (πβ2) (πβ3) (πβ4) ( ππ ππ ππ ππ ππ )
00 0 0 0 π(π) π(πβ1) π(πβ2) π(πβ3) π(πβ4) 1 1 1 1 1 00 0 0 0 (. . . . . ) ( π(π) π(πβ1) π(πβ2) π(πβ3) π(πβ4)) (. . . . . ) ( 2 2 2 2 2 ) ( ) ( ) (. . . . . ) ( ) (. . . . . ) ( ) ( ) ( . . . . . ) = . . . . . , ( . . . . . ) (. . . . . ) ( . . . . . ) (. . . . . ) ( ) (00 0 0 1) ( ) ( ) ( ) (00 0 1βπ) ( . . . . . ) ( πβ3) ( . . . . . ) 00 1βπ π ( . . . . . ) πβ2 6 ( ) 01βπ π π ( ) πβ2 4 5 ( ) (1ββ π π π ) ( . . . . . ) πβ1 1 2 3 ( . . . . . ) πΏ ( ) where ( ) ( ) ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) π1 =ββπβ2 +βπβ1ππβ2, (ππβ4 ππβ4 ππβ4 ππβ4 ππβ4 ) ( ) ( ) π =ββ +β π ( ) 2 πβ3 πβ2 πβ2 (π(π) π(πβ1) π(πβ2) π(πβ3) π(πβ4)) ( πβ3 πβ3 πβ3 πβ3 πβ3 ) +β π ββ π π , ( ) πβ1 πβ3 πβ1 πβ2 πβ2 ( ) ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) (π π π π π ) π3 =ββπβ4 +βπβ3ππβ3 +βπβ2ππβ2 ( πβ2 πβ2 πβ2 πβ2 πβ2 ) ( ) ( ) ββπβ2ππβ2ππβ3 +βπβ1ππβ4 ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) ππβ1 ππβ1 ππβ1 ππβ1 ππβ1 ββπβ1ππβ3ππβ3 ββπβ1ππβ2ππβ2 (π) (πβ1) (πβ2) (πβ3) (πβ4) ( ππ ππ ππ ππ ππ ) +βπβ1ππβ2ππβ2ππβ3, Mathematical Problems in Engineering 5
(π π +ππ +πΆπ +π€π + V π ) π4 =βππβ3 +ππβ2ππβ2, π π+1,π π π+2,π π π+3,π π πβ1,π π π,π ππ,π =β , πΌπ π5 =βππβ4 +ππβ3ππβ3 π=πβ5(β1) 1, +ππβ2ππβ2 βππβ2ππβ2ππβ3, β π =β πβ1 , π6 =βππβ2 +ππβ2ππβ3. π,πβ1 πΌπ (11) (1 β Vπβ1ππ,πβ1) ππβ1,πβ1 = , πΌπβ1 π =β(ππ +ππ +πΆπ Hence, solving the following standard linear systems whose π,πβ1 π π+1,πβ1 π π+2,πβ1 π π+3,πβ1 π β1 coefficient matrix is given by (5) +π€πππβ1,πβ1 + Vπππ,πβ1)(πΌπ) , π=πβ5(β1) 1,
(ββπβ2 +βπβ1ππβ2) π1,π π1,πβ1 π1,πβ2 π1,πβ3 π1,πβ4 ππ,πβ2 = , πΌπ π2,π π2,πβ1 π2,πβ2 π2,πβ3 π2,πβ4 (π + V π ) ( . . . . . ) πβ2 πβ1 π,πβ2 ( . . . . . ) ππβ1,πβ2 =β , ( ) πΌ ( . . . . . ) πβ1 ( . . . . . ) ( ) (1 β π€ π β V π ) ( . . . . . ) π = πβ2 πβ1,πβ2 πβ2 π,πβ2 , π ( . . . . . ) πβ2,πβ2 πΌ ( ) πβ2 (ππβ4,π ππβ4,πβ1 ππβ4,πβ2 ππβ4,πβ3 ππβ4,πβ4) ( ) ππ,πβ2 =β(ππππ+1,πβ2 +ππππ+2,πβ2 +πΆπππ+3,πβ2 (ππβ3,π ππβ3,πβ1 ππβ3,πβ2 ππβ3,πβ3 ππβ3,πβ4) ( ) β1 (ππβ2,π ππβ2,πβ1 ππβ2,πβ2 ππβ2,πβ3 ππβ2,πβ4) +π€πππβ1,πβ2 + Vπππ,πβ2)(πΌπ) , ππβ1,π ππβ1,πβ1 ππβ1,πβ2 ππβ1,πβ3 ππβ1,πβ4 π=πβ5(β1) 1, ππ,π ππ,πβ1 ππ,πβ2 ππ,πβ3 ππ,πβ4 ( ) (ββπβ3 +βπβ2ππβ2 ββπβ1 (ππβ2ππβ2 βππβ3)) ππ,πβ3 = , (π) (πβ1) (πβ2) (πβ3) (πβ4) πΌπ π1 π1 π1 π1 π1 (π π βπ β V π ) (π) (πβ1) (πβ2) (πβ3) (πβ4) (12) πβ2 πβ2 πβ3 πβ1 π,πβ3 ( π2 π2 π2 π2 π2 ) ππβ1,πβ3 = , ( ) πΌ ( ) πβ1 ( . . . . . ) ( . . . . . ) (ππβ2 +π€πβ2ππβ1,πβ3 + Vπβ2ππ,πβ3) ( . . . . . ) π =β , ( ) πβ2,πβ3 πΌ ( . . . . . ) πβ2 ( . . . . . ) ( ) (1 β π π βπ€ π β V π ) ( . . . . . ) π = πβ3 πβ2,πβ3 πβ3 πβ1,πβ3 πβ3 π,πβ3 , ( . . . . . ) πβ3,πβ3 = ( ) πΌπβ3 (π(π) π(πβ1) π(πβ2) π(πβ3) π(πβ4)) ( πβ4 πβ4 πβ4 πβ4 πβ4 ) ( ) ππ,πβ4 ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) (π π π Q π ) ( πβ3 πβ3 πβ3 πβ3 πβ3 ) ββπβ4 +βπβ3ππβ3 ββπβ2 (ππβ2 βππβ2ππβ3) ( ) = ( ( (π) (πβ1) (πβ2) (πβ3) (πβ4)) πΌ (π π π π π ) π ( πβ2 πβ2 πβ2 πβ2 πβ2 ) ( ) β (π βπ π βπ (π βπ π )) (π) (πβ1) (πβ2) (πβ3) (πβ4) + πβ1 πβ4 πβ3 πβ3 πβ2 πβ2 πβ2 πβ3 ), ππβ1 ππβ1 ππβ1 ππβ1 ππβ1 πΌπ (π) (πβ1) (πβ2) (πβ3) (πβ4) ( ππ ππ ππ ππ ππ ) ππβ1,πβ4 (βπ +π π βπ (π βπ π )βV π ) = πβ4 πβ3 πβ3 πβ2 πβ2 πβ2 πβ3 πβ1 π,πβ4 , πΌπβ1 gives the five columns Colπ, π=π,πβ1,πβ2,πβ3,andπβ4 in the forms: ((ππβ2 βππβ2ππβ3)+π€πβ2ππβ1,πβ4 + Vπβ2ππ,πβ4) ππβ2,πβ4 =β , πΌπβ2 (π +π π +π€ π + V π ) π =β πβ3 πβ3 πβ2,πβ4 πβ3 πβ1,πβ4 πβ3 π,πβ4 , 1 πβ3,πβ4 πΌ ππ,π = , πβ3 πΌπ ππβ4,πβ4 =(1βππβ4ππβ3,πβ4 βππβ4ππβ2,πβ4 Vπβ1ππ,π ππβ1,π =β , β1 πΌπβ1 βπ€πβ4ππβ1,πβ3 β Vπβ4ππ,πβ4)(πΌπβ4) , 6 Mathematical Problems in Engineering
π =β(ππ +ππ +πΆπ π,πβ4 π π+1,πβ4 π π+2,πβ4 π π+3,πβ4 Step 1. If πΆπ =0for any π = 1,2,...,πβ5set πΆπ =π‘(π‘ is just β1 a symbolic name). +π€πππβ1,πβ4 + Vπππ,πβ4)(πΌπβ4) , π΅ =0 π=6,7,...,π,π΅ =π‘ π=πβ5(β1) 1. Step 2. If π for any π . (13) Step 3. Use the DETCHEPTA algorithm to check the non- singularity of the matrix π».Ifthematrixπ» is singular then The remaining (π β 5) columns are obtained by using the OUTPUT (βThe matrix π» is singularβ); stop. β1 fact π»π» =πΌπ where πΌπ is the identity matrix. They are as in the following: Step 4. For π = 1,2,...,π, compute and simplify the π ,π ,π π π components π,π π,πβ1 π,nβ2, π,πβ3,and π,πβ4 of the columns Colπβ5 =(πΈπβ2 βπ΄πβ4Colπβ4 πΆπ,π = π,πβ1,πβ2,πβ3,andπβ4,respectively,byusing (13). βπ βπ πβ3Colπβ3 πβ2Colπβ2 (14) Step 5. For π = 1,2,...,π, compute and simplify the β1 components ππ,πβ5 by using (14). βππβ1Colπβ1 βπ΅πColπ)(πΆπβ5) , Step 6. For π=πβ6,πβ7,...,1,do Colπ =(πΈπ+3 βπ΄π+1Colπ+1 βππ+2Colπ+2
βππ+3Colπ+3 βππ+4Colπ+4 For π=1,2,...,π,do β1 (15) βπ΅π+5Colπ+5 βπ·π+6Colπ+6)(πΆπ) , Compute and simplify the components ππ,π by using π=πβ6(β1) 1, (15). where πΈπ is the ith unit vector. End do Remark 2. Equations (14)and(15) suggest an additional βπβ5πΆ =0ΜΈ assumption π=1 π , which is only formal and can be End do omitted by introducing auxiliary parameter π‘ in Algorithm 2 given later. Step 7. Substitute the actual value π‘=0in all expressions to obtain the elements, ππ,π,π,π=1,2,...,π. Now we formulate a second result. It is a symbolic computational algorithm to compute the inverse of a general The symbolic Algorithm 2 will be referred to as CHINV cyclic heptadiagonal matrix of the form (1) when it exists. algorithm. The computational cost of CHINV algorithm is 2 21π β 48π β 88 operations. The Algorithms 2.2, 2.3, and 2.2 πΓπ Algorithm 2. To find the inverse matrix of the general in [5, 13, 14], respectively, are now special cases of the CHINV π» cyclic heptadiagonal matrix in (1) by using the relations algorithm. (13)β(15). INPUT. Order of the matrix π and the components,
π·π,π΅π,ππ,ππ,ππ,π΄π,πΆπ,π=1,2,...,π, 3. Two Illustrative Examples π· =π· =π· =πΆ =πΆ = where 1 2 3 πβ2 πβ1 Inthissectionwegivetwoexamplesforthesakeofillustra- πΆ =0 π . tion.
OUTPUT. Inverse matrix. Example 1. Consider the 10 Γ 10 cyclic heptadiagonal linear system
1β11β200 0 02β1 π₯1 2 1111β100001 π₯2 15 21β11230000 π₯ 33 ( ) ( 3 ) ( ) (2β2315β60000) ( π₯ ) ( 0 ) ( ) ( 4 ) ( ) (0111111200) ( π₯ ) ( 43 ) ( ) ( 5 ) = ( ) . (0 0 β1 β1 β1 β1 β1 β1 1 0 ) ( π₯ ) (β24) (16) ( ) ( 6 ) ( ) (000222231β3) ( π₯7 ) ( 47 ) 0000β2β21135 π₯8 70 300003134β1 π₯9 78 (2400002341) (π₯10) ( 94 ) Mathematical Problems in Engineering 7
By using the coefficient matrix of the system (16) and applying the CHINV algorithm, we get the inverse of the coefficient β1 matrix (π» ):
12664 2921 1898 2957 23399 24419 2069 6676 13714 6316 β β β β β β β 32715 32715 10905 32715 32715 32715 6543 32715 32715 32715 ( 2686 4169 902 1118 8006 8801 41 2366 6391 4571 ) ( ) ( β β β ) ( 32715 32715 10905 32715 32715 32715 6543 32715 32715 32715 ) ( ) ( ) ( 5417 4693 344 241 5842 3292 280 758 2792 1658 ) ( β β β ) ( 10905 10905 3635 10905 10905 10905 2181 10905 10905 10905) ( ) ( ) ( 293 5347 591 146 1393 5698 853 1577 1838 1982 ) ( β β ) ( ) ( 10905 10905 3635 10905 10905 10905 2181 10905 10905 10905) ( ) ( ) ( 6344 4964 2983 1598 12259 9484 433 2621 6374 1676 ) ( β β β ) ( 32715 32715 10905 32715 32715 32715 6543 32715 32715 32715) ( ) π»β1 = ( ) . ( 938 357 788 339 1023 513 56 297 413 477 ) (17) ( β β β ) ( 3635 3635 3635 3635 3635 3635 727 3635 3635 3635 ) ( ) ( ) ( 1178 908 254 604 13232 14012 385 497 1658 3718 ) ( ) ( β β β β β β ) ( 10905 10905 3635 10905 10905 10905 2181 10905 10905 10905 ) ( ) ( ) ( 6356 4969 1382 778 3539 1306 922 1904 5891 194 ) (β β β β β β ) ( 10905 10905 3635 10905 10905 10905 2181 10905 10905 10905 ) ( ) ( ) ( 16382 10738 3244 1216 14092 27832 2725 8078 11282 5153 ) ( β β β ) ( ) ( 32715 32715 10905 32715 32715 32715 6543 32715 32715 32715) 808 3617 1174 44 5947 1868 938 4658 193 1643 β β β β β 32715 32715 10905 32715 32715 32715 6543 32715 32715 32715 ( )
By using (17) and after simple calculations we can obtain the Example 2. Consider the 8Γ8cyclic heptadiagonal matrix solution of cyclic heptadiagonal linear system (16):
0β11β20 0 2β1 1111β1001 21β112300 ( ) π (2β2315β600) π=(π₯1,π₯2,π₯3,π₯4,π₯5,π₯6,π₯7,π₯8,π₯9,π₯10) π»=( ) . (01111112) (19) π 0 0 β1β1β1β1β1β1 = (1, 2, 3, 4, 5, 6, 7, 8, 9, 10) . 30022223 (18) (2 4 0 0 β2 β2 1 1 ) 8 Mathematical Problems in Engineering
By applying the CHINV algorithm, it yields
438 1061 402 133 β135 1361 388 β156 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ( ) ( β72 β9 (2π‘ + 33) 2 (42π‘ + 253) 6π‘ + 19 4 (24π‘ + 169) 2 (6π‘ β 71) β (78π‘ + 595) 12 (6π‘ + 43) ) ( ) ( 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ) ( ) ( ) ( 693 3 (3π‘ + 580) β14 (3π‘ β 25) β (3π‘ β 190) 3 (57π‘ + 317) β9 (25π‘ β 69) β36 (5π‘ + 17) β12 (3π‘+41) ) ( ) ( 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ) ( ) ( ) ( β1125 321π‘ β 539 27 (4π‘ β 11) 39π‘ β 76 β (471π‘ + 2861) β (141π‘ + 4456) 25 (6π‘ + 1) 5 (6π‘ + 121) ) ( ) ( ) ( 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ) π»β1 = ( ) ( ) ( 6 β (108π‘ + 721) 66π‘ + 455 36π‘ + 247 2 (69π‘ + 469) 72π‘ + 509 β (30π‘ + 199) β (6π‘ + 43) ) ( ) ( 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ) ( ) ( ) ( 334 β2 (13π‘ β 316) 2 (12π‘ + 235) β (40π‘ + 171) 30 (3π‘ + 17) β (80π‘ β 493) β4 (16π‘ + 35) β3 (14π‘ + 135) ) ( ) ( ) ( 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ) ( ) ( ) ( 20 β (214π‘ + 1409) β8 (9π‘ + 59) β (26π‘ + 171) β45 (6π‘ + 41) β5 (98π‘ + 655) 46π‘ + 331 126π‘ + 851 ) ( 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 ) 72 9 (2π‘ + 33) β2 (42π‘ + 253) β (6π‘ + 19) 3 (114π‘ + 769) 426π‘ + 3125 78π‘ + 595 β12 (6π‘ + 43)
( 438 π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 438π‘ + 2983 )π‘=0 (20) 438 1061 402 7 135 1361 388 156 β β 2983 2983 2983 157 2983 2983 2983 2983 ( 72 297 506 1 676 142 595 516 ) ( ) (β β β β ) ( 2983 2983 2983 157 2983 2983 2983 2983 ) ( ) ( ) ( 693 1740 350 10 951 621 612 492 ) ( β β ) ( 2983 2983 2983 157 2983 2983 2983 2983 ) ( ) ( ) ( 1125 539 297 4 2861 4456 25 605 ) (β β β β β β ) ( 2983 2983 2983 157 2983 2983 2983 2983 ) ( ) = ( ) . ( 6 721 455 13 938 509 199 43 ) ( β β β ) ( ) ( 2983 2983 2983 157 2983 2983 2983 2983 ) ( ) ( ) ( 334 632 470 9 510 493 140 405 ) ( β β β ) ( 2983 2983 2983 157 2983 2983 2983 2983 ) ( ) ( ) ( 20 1409 472 9 1845 3275 331 851 ) ( β β β β β ) 2983 2983 2983 157 2983 2983 2983 2983 72 297 506 1 2307 3125 595 516 β β β ( 2983 2983 2983 157 2983 2983 2983 2983 )
4. Conclusions [4]M.Batista,βmethodforsolvingcyclicblockpenta-diagonal systems of linear equations,β Mathematical Software,2008, In this work new recursive computational algorithms have http://arxiv.org/abs/0803.0874. been developed for computing the determinant and inverse [5] M. El-Mikkawy and E.-D. Rahmo, βSymbolic algorithm for of general cyclic heptadiagonal matrices and solving linear inverting cyclic pentadiagonal matrices recursively: derivation systems of cyclic heptadiagonal type. The algorithms are and implementation,β Computers & Mathematics with Applica- reliable, are computationally efficient, and will not fail. The tions,vol.59,no.4,pp.1386β1396,2010. algorithms are natural generalizations of some algorithms in [6]S.N.NeossiNguetchueandS.Abelman,βAcomputational current use. algorithm for solving nearly penta-diagonal linear systems,β Applied Mathematics and Computation,vol.203,no.2,pp.629β 634, 2008. References [7] T. Sogabe, βNew algorithms for solving periodic tridiagonal and periodic pentadiagonal linear systems,β Applied Mathematics [1] A. A. Karawia, βAcomputational algorithm for solving periodic and Computation,vol.202,no.2,pp.850β856,2008. penta-diagonal linear systems,β Applied Mathematics and Com- putation,vol.174,no.1,pp.613β618,2006. [8] X.-G. Lv and J. Le, βA note on solving nearly penta-diagonal linear systems,β Applied Mathematics and Computation,vol. [2]A.D.A.HadjandM.Elouafi,βAfastnumericalalgorithmfor 204, no. 2, pp. 702β712, 2008. the inverse of a tridiagonal and pentadiagonal matrix,β Applied [9] X.-L. Zhao and T.-Z. Huang, βOn the inverse of a general Mathematics and Computation,vol.202,no.2,pp.441β445, pentadiagonal matrix,β Applied Mathematics and Computation, 2008. vol. 202, no. 2, pp. 639β646, 2008. [3] I. M. Navon, βA periodic pentadiagonal systems solver,β Com- [10] J. W. Demmel, Applied Numerical Linear Algebra, Society for munications in Applied Numerical Methods,vol.3,pp.63β69, Industrial and Applied Mathematics (SIAM), Philadelphia, Pa, 1987. USA, 1997. Mathematical Problems in Engineering 9
[11] G. Engeln-MullgesΒ¨ and F. Uhlig, Numerical Algorithms with C, Springer, Berlin, Germany, 1996. [12] M. B. Allen III and E. L. Isaacson, Numerical Analysis For Applied Science,JohnWileyandSons,1997. [13] M. El-Mikkawy and E.-D. Rahmo, βA new recursive algorithm for inverting general tridiagonal and anti-tridiagonal matrices,β Applied Mathematics and Computation,vol.204,no.1,pp.368β 372, 2008. [14] M. El-Mikkawy and E.-D. Rahmo, βA new recursive algo- rithm for inverting general periodic pentadiagonal and anti- pentadiagonal matrices,β Applied Mathematics and Computa- tion,vol.207,no.1,pp.164β170,2009. Advances in Advances in Journal of Journal of Operations Research Decision Sciences Applied Mathematics Algebra Probability and Statistics Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
The Scientific International Journal of World Journal Differential Equations Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
Submit your manuscripts at http://www.hindawi.com
International Journal of Advances in Combinatorics Mathematical Physics Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
Journal of Journal of Mathematical Problems Abstract and Discrete Dynamics in Complex Analysis Mathematics in Engineering Applied Analysis Nature and Society Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
International Journal of Journal of Mathematics and Mathematical Discrete Mathematics Sciences
Journal of International Journal of Journal of Function Spaces Stochastic Analysis Optimization Hindawi Publishing Corporation Hindawi Publishing Corporation Volume 2014 Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014