Solvable product-type system of difference equations whose associated polynomial is of the fourth order
Stevo Stevi´c
BMathematical Institute of the Serbian Academy of Sciences Knez Mihailova 36/III, 11000 Beograd, Serbia
Operator Theory and Applications Research Group, Department of Mathematics King Abdulaziz University, P.O. Box 80203, Jeddah 21589, Saudi Arabia
Received 12 January 2017, appeared 14 March 2017 Communicated by Jeff R. L. Webb
Abstract. The solvability problem for the following system of difference equations zn+1=αzanwbn, wn+1=βwcn−1zdn−2, n∈N0,
where a,b,c,d ∈ Z, α,β ∈ C\ {0}, z−2,z−1,z0,w−1,w0 ∈ C\ {0}, is solved. In the main case whenbd 6= 0, a polynomial of the fourth order is associated to the system, and its solutions are represented in terms of the parameters, through the roots of the polynomial in all possible cases (the roots are given in terms of parameters a,b,c,d).
This is also the first paper which successfully deals with the associated polynomial (to a product-type system) of the fourth order in detail, which is the main achievement of the paper.
Keywords: system of difference equations, product-type system, solvable in closed form, polynomial of fourth order.
2010 Mathematics Subject Classification: 39A20, 39A45.
1 Introduction
Concrete nonlinear difference equations and systems is a research field of some recent interest (see, e.g., [2,4,9–15,17–45]). Among the systems, symmetric and related ones have attracted at- tention of some experts, especially after the publication of several papers by Papaschinopoulos and Schinas almost twenty years ago (see, e.g., [2,4,9–14,17,18,21,23–28,30,31,34–40,42–45]).
On the other hand, solvability of the equations and systems has re-attracted some recent attention (see, e.g., [2,15,21–36,38–45]). Some of them are solved by the method of transfor- mation (see, e.g., [15,21,22,24,38–41] and the references therein). For somewhat more complex methods see [33] and [34]. An interesting related method has been recently applied to partial difference equations in [29] and [32]. Books [1,5–8] contain many classical methods for solving difference equations and systems.
BEmail: sstevic@ptt.rs
If initial values and coefficients of product-type systems are positive they can be solved by transforming them to the linear ones with constant coefficients, by using the logarithm. If the initial values and coefficients are not positive the method is not of a special use. Therefore, the solvability of product-type systems with non-positive initial values and coefficients is a problem of interest. We started studying the problem in [36], where we showed the solvability of the system
zn+1 = w
an
zbn−1, wn+1= z
cn
wdn−1, n∈N0, (1.1)
for a,b,c,d ∈ Z and z−1,z0,w−1,w0 ∈ C\ {0}, and gave many results on the long-term behavior of solutions to (1.1) by using the obtained closed form formulas. Product-type equa- tions appeared also in the study of the difference equation in [33], as its special cases. The max-type system in [23] is solved by reducing it to a product-type one. They also appeared indirectly in the study of some max-type and related difference equations and systems, as their boundary cases (see, e.g., [19,20,37]). The study was continued in [42], in [30] where a three-dimensional system was considered, in [28] where it was noticed for the first time that some coefficients can be added to a product-type system so that the solvability is preserved, and later in [31,35,43–45] where various new details and methods are presented.
This paper continues investigating the solvability problem, by studying the following product-type system
zn+1=αzanwbn, wn+1= βwcn−1zdn−2, n ∈N0, (1.2) wherea,b,c,d∈Z,α,β∈Candz−2,z−1,z0,w−1,w0∈ C.
Clearly, the domain of undefinable solutions [24] to system (1.2) is a subset of
U ={(z−2,z−1,z0,w−1,w0)∈C5 :z−2 =0 orz−1 =0 orz0=0 orw−1 =0 orw0 =0}. Thus, from now on we will assume thatz−2,z−1,z0,w−1,w0 ∈ C\ {0}. Since the casesα = 0 and β = 0 are trivial or produce solutions which are not well-defined we will also assume thatαβ6=0.
In the main case when bd 6= 0, a polynomial of the fourth order is associated to the system, and its solutions are represented in terms of the parameters, through the roots of the polynomial in all the cases (the roots are given in terms of parametersa,b,c,d), which is the main achievement of the paper. This is the first paper which deals with the associated polynomial (to a product-type system) of the fourth order in detail. An associated polynomial of the fourth order appears yet in [42], but almost without any analysis of its roots and their influence on the solutions to the system therein.
In this paper, we will use the following standard convention∑mi=kai =0, whenm< k.
2 Auxiliary results
In this section we quote two auxiliary results which are used in the proofs of the main results.
The first one is the following lemma which is well-known (see, e.g., [6,8]). For a proof of a more general result see [35].
Lemma 2.1. Let i∈N0and
s(ni)(z) =1+2iz+3iz2+· · ·+nizn−1, n∈N (2.1) where z∈C.
Then
s(n0)(z) = 1−zn
1−z , (2.2)
s(n1)(z) = 1−(n+1)zn+nzn+1
(1−z)2 , (2.3)
for every z∈ C\ {1}and n ∈N.
The following lemma is also known, and can be proved, for example, by using the Lagrange interpolation polynomial or the calculus of residue (see, for example, [6] and [42]).
Lemma 2.2. Assume thatλj, j=1,k,are pairwise different zeros of the polynomial P(z) =αkzk+αk−1zk−1+· · ·+α1z+α0,
withαkα0 6=0.
Then
∑
k j=1λlj P0(λj) =0 for l =0,k−2, and
∑
k j=1λkj−1 P0(λj) = 1
αk.
3 Main results
The main results in this paper are formulated and proved in this section.
3.1 Solvability of system (1.2)
The first result concerns the solvability problem of system (1.2).
Theorem 3.1. Assume that a,b,c,d∈ Z,α,β∈C\ {0}and z−2,z−1,z0,w−1,w0 ∈C\ {0}. Then system(1.2)is solvable in closed form.
Proof. Case b=0. In this case system (1.2) becomes
zn+1 =αzan, wn+1= βwcn−1zdn−2, n∈N0. (3.1) From the first equation in (3.1) we get
zn= α∑
n−1 j=0 aj
za0n, n∈N, (3.2)
from which it follows that
zn=α
1−an
1−a z0an, n∈N, (3.3)
whena 6=1, and
zn= αnz0, n∈N, (3.4)
whena =1.
Employing (3.2) in the second equation in (3.1), we get wn =βαd∑
n−4 j=0aj
zda0 n−3wcn−2, (3.5)
forn≥4, from which it follows that w2n+i = βαd∑
2n+i−4 j=0 aj
zda0 2n+i−3wc2n+i−2, (3.6) forn≥2 andi=0, 1.
Assume that for somek ∈Nit has been proved that w2n+i =β∑
k−1 j=0cj
αd∑
k−1
j=0cj∑2nl=−02j+i−4al
zd∑
k−1
j=0cja2n−2j+i−3
0 wc2k(n−k)+i, (3.7) forn≥ k+1 andi=0, 1.
By using (3.6) withnreplaced byn−k and inserting into (3.7) we get w2n+i = β∑
k−1 j=0cj
αd∑
k−1
j=0cj∑2nl=−02j+i−4al
zd∑
k−1
j=0cja2n−2j+i−3 0
βαd∑
2(n−k)+i−4
j=0 aj
zda0 2(n−k)+i−3wc2(n−k−1)+ick
= β∑
kj=0cj
αd∑
kj=0cj∑2nl=−02j+i−4alzd∑
kj=0cja2n−2j+i−3
0 wc2k(+n1−k−1)+i, (3.8) forn≥ k+2 andi=0, 1, from which along with (3.6) and the method of induction it follows that (3.7) holds for everyn≥k+1 andi=0, 1.
Choosingk =n−1 in (3.7) we obtain w2n= β∑
n−2 j=0 cj
αd∑
n−2
j=0 cj∑2ni=−02j−4ai
zd∑
n−2
j=0 cja2n−2j−3
0 wc2n−1, (3.9)
and
w2n+1= β∑
n−2 j=0cj
αd∑
n−2
j=0 cj∑2ni=−02j−3ai
zd∑
n−2
j=0 cja2n−2j−2
0 wc3n−1, (3.10)
forn≥2.
From the second equation in (3.1), withn=0, 1, 2, we have
w1= βwc−1zd−2, w2 =βwc0zd−1, w3 =βw1czd0 =β1+cw−c21zcd−2zd0. (3.11) Then by using (3.11) into (3.9) and (3.10) we have
w2n= β∑
n−2 j=0cj
αd∑
n−2
j=0 cj∑2ni=−02j−4ai
zd∑
n−2
j=0 cja2n−2j−3
0 (βwc0zd−1)cn−1
= β∑
n−1 j=0cj
αd∑
n−2
j=0 cj∑2ni=−02j−4ai
w0cnzdc−1n−1zd∑
n−2
j=0 cja2n−2j−3
0 (3.12)
and
w2n+1 =β∑
n−2 j=0 cj
αd∑
n−2
j=0cj∑2ni=−02j−3ai
zd∑
n−2
j=0cja2n−2j−2
0 (β1+cwc−21zcd−2zd0)cn−1
=β∑
nj=0cj
αd∑
n−2
j=0 cj∑2ni=−02j−3ai
wc−n+11zdc−2nzd∑
n−1
j=0cja2n−2j−2
0 , (3.13)
forn≥2.
Subcase a6=16=c, c 6= a2. In this case we have w2n= β∑
n−1 j=0cj
αd∑
n−2
j=0 cj∑2ni=−02j−4ai
wc0nzd∑
n−2
j=0 cja2n−2j−3 0 zdc−1n−1
= β
1−cn 1−cαd∑
n−2
j=0 cj1−a2n1−−a2j−3
wc0nzda
a2n−2−cn−1 a2−c
0 zdc−1n−1
= β
1−cn 1−cα
d 1−a 1−cn−1
1−c −a(a2n−2−cn−1)
a2−c
wc0nzda
a2n−2−cn−1 a2−c
0 zdc−1n−1
= β
1−cn 1−cα
d(a2−c+(a+c)(1−a)cn−1−(1−c)a2n−1) (1−a)(1−c)(a2−c) w0cnzda
a2n−2−cn−1 a2−c
0 zdc−1n−1, (3.14) forn≥2, and
w2n+1 =β∑
nj=0cj
αd∑
n−2
j=0cj∑2ni=−02j−3ai
wc−n1+1zd∑
n−1
j=0cja2n−2j−2
0 zdc−2n
=β
1−cn+1 1−c αd∑
n−2
j=0cj1−a2n1−−a2j−2
wc−n1+1zd
a2n−cn a2−c
0 zdc−2n
=β
1−cn+1 1−c α
d
1−a 1−cn−1
1−c −a2(a2n−2−cn−1)
a2−c
wc−n1+1zd
a2n−cn a2−c
0 zdc−2n
=β
1−cn+1 1−c α
d(a2−c+(1−a2)cn−(1−c)a2n)
(1−a)(1−c)(a2−c) wc−n+11zd
a2n−cn a2−c
0 zdc−2n, (3.15)
forn∈N.
Subcase a26=16=c, c= a2. In this case we have w2n= β∑
n−1 j=0a2j
αd∑
n−2
j=0 a2j∑2ni=−02j−4ai
wa02nzd∑
n−2
j=0a2ja2n−2j−3 0 zda−12n−2
= β
1−a2n 1−a2 αd∑
n−2
j=0 a2j1−a2n1−−a2j−3
wa02nz0d(n−1)a2n−3zda−12n−2
= β
1−a2n 1−a2
α
1−da 1−a2n−2
1−a2 −(n−1)a2n−3
wa02nzd0(n−1)a2n−3zda−12n−2
= β
1−a2n 1−a2 α
d(1−(n−1)a2n−3−a2n−2+(n−1)a2n−1)
(a−1)2(a+1) wa02nzd0(n−1)a2n−3zda−12n−2 (3.16) forn≥2, and
w2n+1= β∑
n j=0a2j
αd∑
n−2
j=0 a2j∑2ni=−02j−3ai
wa−2n1+2zd∑
n−1
j=0a2ja2n−2j−2 0 zda−22n
= β
1−a2n+2 1−a2
αd∑
n−2
j=0 a2j1−a2n1−−a2j−2
wa−2n1+2zdna0 2n−2zda−22n
= β
1−a2n+2 1−a2
α
d
1−a 1−a2n−2
1−a2 −(n−1)a2n−2
wa−2n1+2zdna0 2n−2zda−22n
= β
1−a2n+2 1−a2
α
d(1−na2n−2+(n−1)a2n)
(a−1)2(a+1) w−a2n1+2zdna0 2n−2zda−22n, (3.17) forn≥2.
Subcase a26=1= c.In this case we have w2n =β∑
n−1 j=0 1
αd∑
n−2
j=0∑2ni=−02j−4ai
w0zd∑
n−2 j=0 a2n−2j−3
0 zd−1
=βnαd∑
n−2
j=0 1−a2n−2j−3 1−a w0zda
a2n−2−1 a2−1
0 zd−1
=βnα
d
1−a n−1−a(a2n−2−1)
a2−1
w0zda
a2n−2−1 a2−1
0 zd−1
=βnα
d(a2n−1+(n−1)(1−a2)−a) (a−1)2(a+1) w0zda
a2n−2−1 a2−1
0 zd−1, (3.18)
forn≥2, and
w2n+1= β∑
nj=01
αd∑
n−2
j=0 ∑2ni=−02j−3ai
w−1zd∑
n−1 j=0 a2n−2j−2
0 zd−2
= βn+1αd∑
n−2
j=0 1−a2n−2j−2 1−a w−1zd
a2n−1 a2−1
0 zd−2
= βn+1α
1−da n−1−a2(a2n−2−1)
a2−1
w−1zd
a2n−1 a2−1
0 zd−2
= βn+1α
d(a2n−na2+n−1) (a−1)2(a+1) w−1zd
a2n−1 a2−1
0 zd−2, (3.19)
forn∈ N.
Subcase a=−1, c=1. In this case we have w2n=βnαd∑
n−2
j=0∑2ni=−02j−4(−1)i
w0zd∑
n−2
j=0(−1)2n−2j−3
0 zd−1
=βnαd∑
n−2 j=0
1−(−1)2n−2j−3
1−(−1) w0zd0(1−n)zd−1
=βnαd(n−1)w0zd0(1−n)zd−1, (3.20) forn∈ N, and
w2n+1 =β∑
nj=01
αd∑
n−2
j=0∑2ni=−02j−3ai
w−1zd∑
n−1 j=0 a2n−2j−2
0 zd−2
=βn+1αd∑
n−2
j=0∑2ni=−02j−3(−1)i
w−1zd∑
n−1
j=0(−1)2n−2j−2
0 zd−2
=βn+1αd∑
n−2 j=0
1−(−1)2n−2j−2
1−(−1) w−1zdn0 zd−2
=βn+1w−1zdn0 zd−2, (3.21)
forn∈ N0.
Subcase a=1, c6=1. In this case, by using formula (2.3), we get w2n =β∑
n−1 j=0 cj
αd∑
n−2
j=0cj∑2ni=−02j−41
wc0nzd∑
n−2 j=0cj 0 zdc−1n−1
=β
1−cn 1−c αd∑
n−2
j=0(2n−2j−3)cj
wc0nzd
1−cn−1 1−c
0 zdc−1n−1
=β
1−cn
1−c αd (2n−1)
1−cn−1
1−c −21−ncn−1+(n−1)cn (1−c)2
wc0nzd1
−cn−1 1−c
0 zdc−1n−1
=β
1−cn 1−c αd
2n−3−(2n−1)c+cn−1+cn (1−c)2 wc0nzd
1−cn−1 1−c
0 zdc−1n−1, (3.22)
forn≥2, and
w2n+1 =β∑
n j=0cj
αd∑
n−2
j=0cj∑2ni=−02j−31
wc−n+11zd∑
n−1 j=0cj 0 zdc−2n
=β
1−cn+1 1−c αd∑
n−2
j=0(2n−2j−2)cj
wc−n1+1zd
1−cn 1−c
0 zdc−2n
=β
1−cn+1 1−c αd 2n
1−cn−1
1−c −21−ncn−(11−+(c)n2−1)cn
wc−n+11zd
1−cn 1−c
0 zdc−2n
=β
1−cn+1 1−c α
2d(n−1−nc+cn)
(1−c)2 wc−n1+1zd
1−cn 1−c
0 zdc−2n, (3.23)
forn∈ N.