oscillatory properties of third-order ...joics.org/gallery/joics-4062.pdfoscillatory properties of...

13
OSCILLATORY PROPERTIES OF THIRD-ORDER NEUTRAL DELAY DIFFERENCE EQUATIONS. S. Revathy 1 and R. Kodeeswaran 2 1 Department of Mathematics, Selvam College of Technology, Namakkal, Tamilnadu, India. 2 Department of Mathematics, Kandaswami Kandar’s College, P.Velur, Namakkal, Tamilnadu, India. 1 E-Mail Address :[email protected], 2 E-mail Address: [email protected] ABSTRACT: The aim of this paper is to investigate the oscillatory behavior of solutions of third-order linear neutral delay difference equation of the term , 0 , 0 ) ( ) ( ) ( ) ( ) ( 0 1 2 t t t x t p t y t c t c where ). ( ) ( ) ( ) ( t x t q t x t y By using comparison principles with associated first and second-order delay difference inequalities. Examples are given to illustrate the main results. KEYWORDS: linear difference equation, delay, third -order 2010 Mathematics Subject Classification : 39A10. INTRODUCTION This research, we are considered with oscillation for the third order linear neutral delay difference equation of the term , 0 , 0 ) ( ) ( ) ( ) ( ) ( 0 1 2 t t t x t p t y t c t c (1) where ). ( ) ( ) ( ) ( t x t q t x t y We make the below assumptions: (LH1): ) ( 1 t c and ) ( 2 t c are the sequences of positive integers; (LH2): ) (t p and ) (t q are the positive real sequences such that q(t) β‰₯ q 0 > 1 (t) β‰  0; (LH3): Οƒ, are positive integers, such that < ; (LH4): t+ βˆ’Οƒβ‰€t and t+ βˆ’Οƒβ‰₯tβˆ’Οƒ For the sake of simplicity, we define the operators E 0 y = y, E 1 y=c 1 βˆ†y, E 2 y=c 2 βˆ†(c 1 (βˆ†y)), E 3 y = βˆ†(c 2 βˆ†(c 1 (βˆ†y))), Journal of Information and Computational Science Volume 10 Issue 9 - 2020 ISSN: 1548-7741 www.joics.org 438

Upload: others

Post on 11-Feb-2021

7 views

Category:

Documents


0 download

TRANSCRIPT

  • OSCILLATORY PROPERTIES OF THIRD-ORDER

    NEUTRAL DELAY DIFFERENCE EQUATIONS.

    S. Revathy 1 and R. Kodeeswaran 2

    1 Department of Mathematics, Selvam College of Technology, Namakkal, Tamilnadu, India. 2 Department of Mathematics, Kandaswami Kandar’s College, P.Velur, Namakkal,

    Tamilnadu, India.

    1 E-Mail Address :[email protected], 2 E-mail Address: [email protected]

    ABSTRACT: The aim of this paper is to investigate the oscillatory behavior of solutions of

    third-order linear neutral delay difference equation of the term

    ,0,0)()()()()( 012 tttxtptytctc

    where ).()()()( txtqtxty By using comparison principles with associated first and

    second-order delay difference inequalities. Examples are given to illustrate the main results.

    KEYWORDS: linear difference equation, delay, third -order

    2010 Mathematics Subject Classification : 39A10.

    INTRODUCTION

    This research, we are considered with oscillation for the third order linear neutral

    delay difference equation of the term

    ,0,0)()()()()( 012 tttxtptytctc (1)

    where ).()()()( txtqtxty We make the below assumptions:

    (LH1): )(1 tc and )(2 tc are the sequences of positive integers;

    (LH2): )(tp and )(tq are the positive real sequences such that q(t) β‰₯ q0 > 1 π‘Žπ‘›π‘‘ 𝑝(t) β‰  0;

    (LH3): Οƒ, are positive integers, such that 𝜎 < ;

    (LH4): t + βˆ’ Οƒ ≀ t and t + βˆ’ Οƒ β‰₯ t βˆ’ Οƒ

    For the sake of simplicity, we define the operators

    E0y = y, E1y = c1βˆ†y, E2y = c2βˆ†(c1(βˆ†y)), E3y = βˆ†(c2βˆ†(c1(βˆ†y))),

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org438

    mailto:[email protected]:[email protected]

  • and assume without further mention that E3y is of noncanonical type, (ie)

    βˆ‘1

    𝑐1(𝑠)< ∞

    ∞

    𝑠=𝑑0

    π‘Žπ‘›π‘‘ βˆ‘1

    𝑐2(𝑠)< ∞.

    ∞

    𝑠=𝑑0

    (2)

    By a solution of equation (1), we mean a real sequence {π‘₯(𝑑)} defined for t β‰₯ t0 and

    satisfies equation (1). We taken only those solutions {π‘₯(𝑑)} for equation (1) which satisfy this

    𝑠𝑒𝑝{|π‘₯(𝑑)|: 𝑑 β‰₯ 𝑇} > 0 for all 𝑑 β‰₯ 𝑇 and assuming that (1) possesses such solutions. A

    solution of equation (1) is called oscillatory if it is neither eventually positive nor eventually

    negative; otherwise it is called non oscillatory.

    We say that (1) have property V2 if any solution π‘₯(𝑑) of (1) is either oscillatory of

    satisfy this limπ‘‘β†’βˆž

    π‘₯(𝑑) = 0.

    Oscillation problems for third-order difference equations have been investigated in

    recent years, see for example, [2-10 , 12-14 ] and the references contained therein.

    In [13 ] author consider the following equation

    βˆ†(π‘Žπ‘›βˆ†(𝑏𝑛(βˆ†π‘₯𝑛)𝛼)) + 𝑝𝑛(βˆ†π‘₯𝑛+1)

    𝛼 + π‘žπ‘›π‘“(π‘₯𝜎(𝑛)) = 0, 𝑛 β‰₯ 𝑛0, (3)

    and established some criteria for the oscillation of certain third-order difference equations

    using comparison principles with a suitable couple of first order difference equations.

    The above observation motivated us to study oscillation criteria for third order neutral

    delay difference with noncanonical operators. In section 2, we present the oscillation of all

    solutions of equation (1) and section 3, we provide some examples to illustrative the main

    result.

    2. Main Results

    For the sake of convenience, we list the following notations to be used in the research

    paper.

    ΞΌ1(t) = βˆ‘1

    c1(s)

    tβˆ’1

    s=t1

    , ΞΌ2(t) = βˆ‘1

    c2(s)

    tβˆ’1

    s=t1

    , ΞΌ(n) = βˆ‘ΞΌ2(s)

    c1(s)

    tβˆ’1

    s=t1

    ,

    ψ1(𝑑) = βˆ‘1

    𝑐1(𝑠)

    ∞

    𝑠=𝑑

    , ψ2(𝑑) = βˆ‘1

    𝑐2(𝑠)

    ∞

    𝑠=𝑑

    , ψ(𝑑) = βˆ‘Οˆ1(𝑠)

    𝑐2(𝑠)

    ∞

    𝑠=𝑑

    ,

    πœ‡(𝑑, 𝑑1) = βˆ‘1

    𝑐1(𝑠)

    π‘‘βˆ’1

    𝑠=𝑑1

    βˆ‘1

    𝑐2(𝑒)

    π‘‘βˆ’1

    𝑒=𝑠

    , πœ‡(𝑑, 𝑑1) = βˆ‘1

    𝑐1(𝑠)βˆ‘

    1

    𝑐2(𝑒)𝑒𝛽,

    π‘‘βˆ’1

    𝑒=𝑠

    π‘‘βˆ’1

    𝑠=𝑑1

    where Ξ² is a constant satisfying

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org439

  • 0 β‰€π‘ž0𝛽

    π‘ž0 βˆ’ 1≀

    𝑑𝑝(𝑑)πœ‡(𝑑, 𝑑 + βˆ’ 𝜎)

    π‘ž(𝑑 + βˆ’ 𝜎) (4)

    Lemma 1: Suppose that (LH1) – (LH3) are satisfied and π‘₯(𝑑) is an eventually positive

    solution of equation (1). Then

    𝑦(𝑑) > π‘₯(𝑑) β‰₯1

    π‘ž(𝑑 + )[𝑦(𝑑 + ) βˆ’

    𝑦(𝑑 + 2 )

    π‘ž(𝑑 + 2 )] (5)

    and the corresponding sequence 𝑦(𝑑) belongs to one of following cases;

    𝑦(𝑑) ∈ 𝐺1 ⇔ 𝑦 > 0, 𝐸1𝑦 < 0, 𝐸2𝑦 < 0,

    𝑦(𝑑) ∈ 𝐺2 ⇔ 𝑦 > 0, 𝐸1𝑦 < 0, 𝐸2𝑦 > 0,

    𝑦(𝑑) ∈ 𝐺3 ⇔ 𝑦 > 0, 𝐸1𝑦 > 0, 𝐸2𝑦 > 0,

    𝑦(𝑑) ∈ 𝐺4 ⇔ 𝑦 > 0, 𝐸1𝑦 > 0, 𝐸2𝑦 < 0,

    eventually.

    Proof: Choose 𝑑1 > 𝑑0 such that π‘₯(𝑑 βˆ’ 𝜎) > 0 π‘Žπ‘›π‘‘ π‘₯(𝑑 βˆ’ ) > 0. From the definition of 𝑦,

    we see that 𝑦(𝑑) > π‘₯(𝑑) > 0 and

    π‘₯(𝑑) =𝑦(𝑑 + ) βˆ’ π‘₯(𝑑 + )

    π‘ž(𝑑 + )β‰₯

    1

    π‘ž(𝑑 + )(𝑦(𝑑 + ) βˆ’

    𝑦(𝑑 + 2 )

    π‘ž(𝑑 + 2 ))

    for 𝑑 β‰₯ 𝑑1. Obviously, 𝐸3𝑦(𝑑) is non-increasing, since 𝐸3𝑦(𝑑) = βˆ’π‘(𝑑)π‘₯(𝑑 βˆ’ 𝜎) ≀ 0.

    Hence 𝐸1𝑦(𝑑) and 𝐸2𝑦(𝑑) are of one sign eventually, which implies that four cases 𝐺1 βˆ’ 𝐺4

    are possible for 𝑦(𝑑).

    Next we state an auxiliary criterion for the nonexistence of positive increasing

    solutions of (1). As will be shown later, this condition is already included in those eliminating

    solutions from the class 𝐺1. In the proof, we will take advantage of the useful fact

    limπ‘‘β†’βˆž

    πœ‡(𝑑 + )

    πœ‡(𝑑)= lim

    π‘‘β†’βˆž

    πœ‡1(𝑑 + )

    πœ‡1(𝑑)= 1, (6)

    which follows from (2).

    Lemma 2: Suppose that (LH1) – (LH3) are satisfied. If

    βˆ‘Οˆ2(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑠=𝑑0

    = ∞, (7)

    then 𝐺3 = 𝐺4 = πœ‘.

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org440

  • Proof: For the sake of contradiction, let (7) be satisfied but 𝑦 πœ– 𝐺3 βˆͺ 𝐺4. Choose 𝑑1 > 𝑑0

    such that π‘₯(𝑑) > 0, π‘₯(𝑑 βˆ’ 𝜎) > 0 π‘Žπ‘›π‘‘ π‘₯(𝑑 βˆ’ ) > 0.

    Assume that 𝑦 πœ– 𝐺3. Since 𝐸2𝑦 is decreasing, we have

    𝐸1𝑦(𝑑) β‰₯ βˆ‘1

    𝑐2(𝑠)

    π‘‘βˆ’1

    𝑠=𝑑1

    𝐸2𝑦(𝑠) β‰₯ 𝐸2𝑦(𝑑)πœ‡2(𝑑).

    Thus,

    βˆ† (𝐸1𝑦(𝑑)

    πœ‡2(𝑑)) =

    𝐸2𝑦(𝑑)πœ‡2(𝑑) βˆ’ 𝐸1𝑦(𝑑)

    𝑐2(𝑑)πœ‡22(𝑑 + 1)

    ≀ 0.

    Therefore,)1(

    )(

    2

    1

    t

    tyE

    is non-increasing and moreover, this fact yields

    𝑦(𝑑) β‰₯ βˆ‘πœ‡2(𝑑)

    𝑐1(𝑠)πœ‡2(𝑑)

    π‘‘βˆ’1

    𝑠=𝑑1

    𝐸1𝑦(𝑠) β‰₯𝐸1𝑦(𝑑)πœ‡(𝑑)

    πœ‡2(𝑑)π‘“π‘œπ‘Ÿ 𝑑 β‰₯ 𝑑1.

    Consequently,)(

    )(

    t

    ty

    is non-increasing, since

    βˆ† (𝑦(𝑑)

    πœ‡(𝑑)) =

    𝐸1𝑦(𝑑)πœ‡(𝑑) βˆ’ 𝑦(𝑑)πœ‡2(𝑑)

    𝑐1(𝑑)πœ‡2(𝑑 + 1)≀ 0.

    From 𝑑 + 2 β‰₯ 𝑑 + , we have

    𝑦(𝑑 + 2 ) β‰€πœ‡(𝑑 + 2 )

    πœ‡(𝑑 + )𝑦(𝑑 + ) (8)

    using this in (5), we find that

    π‘₯(𝑑) β‰₯𝑦(𝑑 + )

    π‘ž(𝑑 + )[1 βˆ’

    πœ‡(𝑑 + 2 )

    πœ‡(𝑑 + )π‘ž(𝑑 + 2 )] , 𝑑 β‰₯ 𝑑1

    By virtue of (LH2) and (6), there is 𝑑2 β‰₯ 𝑑1 such that for any constant πœ€ ∈ (0, π‘ž0 βˆ’ 1) and 𝑑 β‰₯

    𝑑2,

    πœ‡(𝑑 + 2 )

    πœ‡(𝑑 + )π‘ž(𝑑 + 2 )≀

    1 + πœ€

    π‘ž0,

    which implies,

    π‘₯(𝑑) β‰₯𝑦(𝑑 + )

    π‘ž(𝑑 + )[1 βˆ’

    1 + πœ€

    π‘ž0] > 0. (9)

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org441

  • Combining (9) with (1) we have

    0 β‰₯ 𝐸3𝑦(𝑑) + (1 βˆ’1 + πœ€

    π‘ž0)

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)𝑦(𝑑 + βˆ’ 𝜎),

    β‰₯ 𝐸3𝑦(𝑑) + π‘˜ (1 βˆ’1 + πœ€

    π‘ž0)

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎), (10)

    where we used the fact that 𝑦 is increasing, and set π‘˜ = 𝑦(𝑑2 + βˆ’ 𝜎) < 𝑦(𝑑 + βˆ’ 𝜎).

    Summing (10) from 𝑑2 to 𝑑 βˆ’ 1, we get

    𝐸2𝑦(𝑑) ≀ 𝐸2𝑦(𝑑2) βˆ’ π‘˜ (1 βˆ’1 + πœ€

    π‘ž0) βˆ‘

    𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎).

    π‘‘βˆ’1

    𝑠=𝑑2

    (11)

    On the other hand, from (2) and (7), it follows that

    βˆ‘π‘(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑠=𝑑0

    = ∞,

    which in view of (11), contradicts the positivity of 𝐸2𝑦. Now, assume that 𝑦 ∈ 𝐺4 for 𝑑 β‰₯ 𝑑1.

    Using the monotonicity of 𝐸1𝑦, we find

    𝑦(𝑑) β‰₯ βˆ‘1

    𝑐1(𝑠)

    π‘‘βˆ’1

    𝑠=𝑑1

    𝐸1𝑦(𝑠) β‰₯ 𝐸1𝑦(𝑑)πœ‡1(𝑑).

    Thus, one can see that

    βˆ† (𝑦(𝑑)

    πœ‡1(𝑑)) =

    𝐸1𝑦(𝑑)πœ‡1(𝑑) βˆ’ 𝑦(𝑑)

    𝑐1(𝑑)πœ‡12(𝑑 + 1)

    ≀ 0,

    which implies that )(

    )(

    1 t

    ty

    is non-increasing. Hence,

    𝑦(𝑑 + 2 ) β‰€πœ‡1(𝑑 + 2 )

    πœ‡1(𝑑 + )𝑦(𝑑 + ).

    we use (6) to arrive at (10), which holds for any πœ€ > 0 π‘Žπ‘›π‘‘ 𝑑 β‰₯ 𝑑2 π‘“π‘œπ‘Ÿ 𝑑2 β‰₯ 𝑑1 sufficiently

    large. Summing (10) from 𝑑2 to 𝑑 βˆ’ 1, we have

    βˆ’βˆ†(𝐸1𝑦(𝑑)) β‰₯ π‘˜ (1 βˆ’1 + πœ€

    π‘ž0)

    1

    𝑐2(𝑑)βˆ‘

    𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑2

    .

    Summing the above inequality again from 𝑑2 to 𝑑 βˆ’ 1, we find that

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org442

  • 𝐸1𝑦(𝑑) ≀ 𝐸1𝑦(𝑑2) βˆ’ π‘˜ (1 βˆ’1 + πœ€

    π‘ž0) βˆ‘

    1

    𝑐2(𝑒)

    π‘‘βˆ’1

    𝑒=𝑑2

    βˆ‘π‘(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎).

    π‘’βˆ’1

    𝑠=𝑑2

    Letting 𝑑 π‘‘π‘œ ∞ changing the summation and using (7) we obtain

    0 ≀ 𝐸1𝑦(∞) ≀ 𝐸1𝑦(𝑑2) βˆ’ π‘˜ (1 βˆ’1 + πœ€

    π‘ž0) βˆ‘

    1

    𝑐2(𝑒)

    ∞

    𝑒=𝑑2

    βˆ‘π‘(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘’βˆ’1

    𝑠=𝑑2

    = 𝐸1𝑦(𝑑2) βˆ’ π‘˜ (1 βˆ’1 + πœ€

    π‘ž0) βˆ‘

    𝑝(𝑠)ψ2(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑒=𝑑2

    = βˆ’βˆž,

    a contradiction. The proof is complete.

    Theorem 1: Suppose that (LH1) – (LH3) are satisfied. If

    βˆ‘Οˆ(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑠=𝑑0

    = ∞, (12)

    then (1) has property V2.

    Proof: Assume that π‘₯(𝑑) is a non-oscillatory solution of (1). Without loss of generality, we

    make it eventually positive. We suppose that π‘₯(𝑑) > 0, π‘₯(𝑛 βˆ’ 𝜎) > 0 π‘Žπ‘›π‘‘ π‘₯(𝑛 βˆ’ ) > 0. By

    conclusion of Lemma 1, 𝑦 ∈ 𝐺𝑖 , 𝑖 = 1,2,3, … . π‘“π‘œπ‘Ÿ 𝑑 β‰₯ 𝑑1. First it is easy to see that in view of

    (2), condition (12) implies

    βˆ‘Οˆ2(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑠=𝑑0

    = βˆ‘π‘(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑠=𝑑0

    = ∞.

    Thus by Lemma 2, 𝐺3 = 𝐺4 = πœ‘ and so either 𝑦 πœ– 𝐺1 π‘œπ‘Ÿ 𝑦 πœ– 𝐺2. Using (LH2) and the fact that

    𝑦 is non-increasing in (5), we have,

    π‘₯(𝑑) β‰₯𝑦(𝑑 + )

    π‘ž(𝑑 + )[1 βˆ’

    1

    π‘ž(𝑑 + 2 )] β‰₯ (1 βˆ’

    1

    π‘ž0)

    𝑦(𝑑 + )

    π‘ž(𝑑 + ) (13)

    On the other hand, since βˆ†π‘¦ < 0 there is a constant 𝑙 > 0 such that

    limπ‘‘β†’βˆž

    𝑦(𝑑) = 𝑙 < ∞

    If 𝑙 > 0, there exists a 𝑑2 β‰₯ 𝑑1 such that 𝑦(𝑑) β‰₯ 𝑙 π‘“π‘œπ‘Ÿ 𝑑 β‰₯ 𝑑2. Hence, from (13), we see that

    π‘₯(𝑑) β‰₯𝑙(π‘ž0 βˆ’ 1)

    π‘ž0

    1

    π‘ž(𝑑 + ), 𝑑 β‰₯ 𝑑2.

    Using this in (1), we find

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org443

  • 𝐸3𝑦(𝑑) +𝑙(π‘ž0 βˆ’ 1)

    π‘ž0

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)≀ 0, 𝑑 ≀ 𝑑2. (14)

    If we assume that 𝑦 πœ– 𝐺1, then by summing (14) from 𝑑2 to 𝑑 βˆ’ 1 we have

    βˆ’βˆ†(𝐸1𝑦(𝑑)) β‰₯𝑙(π‘ž0 βˆ’ 1)

    π‘ž0

    1

    𝑐2(𝑑)βˆ‘

    𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑2

    .

    Summing the above inequality from𝑑2 to 𝑑 βˆ’ 1, we obtain

    βˆ’βˆ†π‘¦(𝑑) β‰₯𝑙(π‘ž0 βˆ’ 1)

    π‘ž0

    1

    𝑐1(𝑑)βˆ‘

    1

    𝑐2(𝑒)

    π‘‘βˆ’1

    𝑒=𝑑2

    βˆ‘π‘(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘’βˆ’1

    𝑠=𝑑2

    . (15)

    Summing (15) from 𝑑2 to t-1, letting 𝑑 to ∞ and changing the summation in the resulting

    inequality, and taking (12), we get

    𝑙 = 𝑦(∞) ≀ 𝑦(𝑑2) βˆ’π‘™(π‘ž0 βˆ’ 1)

    π‘ž0βˆ‘

    1

    𝑐1(𝑣)

    ∞

    𝑣=𝑑2

    βˆ‘1

    𝑐2(𝑒)

    π‘£βˆ’1

    𝑒=𝑑2

    βˆ‘π‘(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘’βˆ’1

    𝑠=𝑑2

    (16)

    = 𝑦(𝑑2) βˆ’π‘™(π‘ž0 βˆ’ 1)

    π‘ž0βˆ‘

    ψ(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    ∞

    𝑠=𝑛2

    = βˆ’βˆž,

    is contradiction. Hence, 𝑙 = 0 in this case.

    If we take 𝑦 ∈ 𝐺2, then by summing (14) from 𝑑2 to 𝑑 βˆ’1 and using (12) we arrive at

    𝐸2𝑦(𝑑) ≀ 𝐸2𝑦(𝑑2) βˆ’π‘™(π‘ž0 βˆ’ 1)

    π‘ž0βˆ‘

    𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑2

    β†’ βˆ’βˆž π‘Žπ‘  𝑑 β†’ ∞, (17)

    which contradicts the positivity of 𝐸2𝑦 and so 𝑙 = 0. Since 𝑦(𝑑) β‰₯ π‘₯(𝑑), we find limπ‘‘β†’βˆž

    π‘₯(𝑑) =

    0. This proof is complete.

    In the following result, we present a criterion for nonexistence of 𝐺1type solutions, based on

    comparison of the studied Equation (1) with an associated first - order delay difference

    inequality. The given criterion also excludes solutions from classes 𝐺3and 𝐺4.

    Lemma 3: Suppose that (LH1) – (LH4) are satisfied. If

    lim infπ‘‘β†’βˆž

    βˆ‘π‘(𝑠)ψ(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑+πœ‰βˆ’πœŽ

    > ∞, (18)

    then 𝐺1 = 𝐺3 = 𝐺4 = πœ‘.

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org444

  • Proof: For the sake of contradiction, let (18) be satisfied but 𝑦 ∈ 𝐺1 βˆͺ 𝐺3 βˆͺ 𝐺4. Choose 𝑑1 >

    𝑑0 such that π‘₯(𝑑) > 0, π‘₯(𝑛 βˆ’ 𝜎) > 0 π‘Žπ‘›π‘‘ π‘₯(𝑛 βˆ’ ) > 0. Assume first that𝑦 ∈ 𝐺1. As in the

    proof of Theorem 1, we arrive at (13), which in view of (1) yields

    𝐸3𝑦(𝑑) +π‘ž0 βˆ’ 1

    π‘ž0

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)𝑦(𝑑 + βˆ’ 𝜎) ≀ 0. (19)

    Define the function

    𝑀(𝑑) = ψ1(𝑑)𝐸1𝑦(𝑑) + 𝑦(𝑑) (20)

    From

    𝑦(𝑑) β‰₯ βˆ’ βˆ‘1

    𝑐1(𝑠)𝐸1𝑦(𝑠) β‰₯ βˆ’πΈ1𝑦(𝑑)ψ1(𝑑)

    ∞

    𝑠=𝑑

    = βˆ’πΈ1𝑦(𝑑 + 1)ψ1(𝑑) (21)

    and

    βˆ†π‘€(𝑑) = ψ1(𝑑)Ξ”(𝐸1𝑦(𝑑)) =ψ1(𝑑)

    𝑐2(𝑑)𝐸2𝑦(𝑑) < 0,

    we see that 𝑀(𝑑) is a strictly decreasing eventually positive function. Using the definition of

    𝑀 in (19), we have

    Ξ” (𝑐2(𝑑)

    ψ1(𝑑)Δ𝑀(𝑑)) +

    π‘ž0 βˆ’ 1

    π‘ž0

    𝑝(𝑑)𝑦(𝑑 + βˆ’ 𝜎)

    π‘ž(𝑑 + βˆ’ 𝜎)≀ 0.

    Hence 𝑀 is a solution of the second-order delay difference inequality

    βˆ† (𝑐2(𝑑)βˆ†π‘€(𝑑)

    ψ1(𝑑)) +

    π‘ž0 βˆ’ 1

    π‘ž0 𝑝(𝑑)𝑀(𝑑 + βˆ’ 𝜎)

    π‘ž(𝑑 + βˆ’ 𝜎)≀ 0. (22)

    Similarly as before, we define the function u by

    𝑒(𝑑) =ψ(𝑑)𝑐2(𝑑)

    ψ1(𝑑)Δ𝑀(𝑑) + 𝑀(𝑑)

    From

    Δ𝑒(𝑑) = Ξ” (𝑐2(𝑑)Δ𝑀(𝑑)

    ψ1(𝑑)) ψ(𝑑)

    = 𝐸3𝑦(𝑑)ψ(𝑑) ≀ 0

    and

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org445

  • 𝑀(𝑑) β‰₯ βˆ’ βˆ‘Οˆ1(𝑠)𝑐2(𝑠)

    𝑐2(𝑠)ψ1(𝑠)

    ∞

    𝑠=𝑑

    Δ𝑀(𝑠) β‰₯ βˆ’π‘2(𝑑)

    ψ1(𝑑)Δ𝑀(𝑑)ψ(𝑑)

    = βˆ’π‘2(𝑑 + 1)

    ψ1(𝑑 + 1)Δ𝑀(𝑑 + 1)ψ(𝑑), (23)

    We conclude that 𝑒 is eventually positive and non-increasing. Using the definition of

    𝑒 in (22), it is easy to see that 𝑒 satisfies the first-order delay difference inequality

    Δ𝑒(𝑑) +π‘ž0 βˆ’ 1

    π‘ž0

    𝑝(𝑑)ψ(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)𝑒(𝑑 + βˆ’ 𝜎) ≀ 0. (24)

    However, by [1 Theorem 6.20.5], condition (18) ensures that the above inequality does not

    posses a positive solution, which is a contradiction.

    To show that also 𝐺3 = 𝐺4 = πœ‘, it suffices to note that (12) is necessary for validity of

    (18) since otherwise, the left-hand side of (18) would equal zero. The conclusion then

    immediately follows from Theorem 1. The proof is complete.

    Lemma 4: Suppose that (LH1) – (LH4) are satisfied and (7) holds. If for any 𝑑1 β‰₯ 𝑑0 large

    enough,

    limπ‘‘β†’βˆž

    𝑠𝑒𝑝 βˆ‘ (ψ(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)βˆ’ (

    π‘ž0π‘ž0 βˆ’ 1

    )ψ1(𝑠)

    4ψ(𝑠)𝑐2(𝑠))

    π‘‘βˆ’1

    𝑠=𝑑1

    >π‘ž0

    π‘ž0 βˆ’ 1 (25)

    then 𝐺1 = 𝐺3 = 𝐺4 = πœ‘.

    Proof: For the sake of contradiction, let (18) be satisfied but 𝑦 ∈ 𝐺1 βˆͺ 𝐺3 βˆͺ 𝐺4. Choose 𝑑1 >

    𝑑0 such that π‘₯(𝑑) > 0, π‘₯(𝑑 βˆ’ 𝜎) > 0 π‘Žπ‘›π‘‘ π‘₯(𝑑 βˆ’ ) > 0. Assume that 𝑦 ∈ 𝐺1. We proceed as

    in the proof of Lemma 3 to obtain (22), where 𝑀 is given by (20). Consider the function ρ

    defined by

    𝜌(𝑑) =𝑐2(𝑑)βˆ†π‘€(𝑑)

    ψ1(𝑑)𝑀(𝑑) , (26)

    Clearly, 𝜌 < 0, from (23), it is easy to see that

    βˆ’1 ≀ ψ(𝑑)𝜌(𝑑) < 0 (27)

    Using (22) together with (26), we have

    Ξ”πœŒ(𝑑) = Ξ” (𝑐2(𝑑)Δ𝑀(𝑑)

    ψ1(𝑑))

    1

    𝑀(𝑑)βˆ’

    𝑐2(𝑑 + 1)[Δ𝑀(𝑑 + 1)]2

    ψ1(𝑑 + 1)𝑀2(𝑑 + 1)

    ≀ βˆ’ (π‘ž0 βˆ’ 1

    π‘ž0)

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)

    𝑀(𝑑 + βˆ’ 𝜎)

    𝑀(𝑑)βˆ’

    ψ1(𝑑 + 1)𝜌2(𝑑 + 1)

    𝑐2(𝑑 + 1) (28)

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org446

  • ≀ βˆ’ (π‘ž0 βˆ’ 1

    π‘ž0)

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)βˆ’

    ψ1(𝑑 + 1)𝜌2(𝑑 + 1)

    𝑐2(𝑑 + 1).

    Multiplying both side of (28) by ψ(𝑑) and summing the resulting inequality from 𝑑1 to 𝑑 βˆ’

    1, we have

    ψ(𝑑)𝜌(𝑑) ≀ ψ(𝑑1)𝜌(𝑑1) + βˆ‘πœŒ(𝑠 + 1)ψ1(𝑠 + 1)

    𝑐2(𝑠 + 1)

    π‘‘βˆ’1

    𝑠=𝑑1

    βˆ’π‘ž0 βˆ’ 1

    π‘ž0βˆ‘

    ψ(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑1

    βˆ’ βˆ‘Οˆ1(𝑠 + 1)𝜌

    2(𝑠 + 1)ψ(𝑠)

    𝑐2(𝑠 + 1)

    π‘‘βˆ’1

    𝑠=𝑑1

    = ψ(𝑑1)𝜌(𝑑1) βˆ’π‘ž0 βˆ’ 1

    π‘ž0βˆ‘

    ψ(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑1

    + βˆ‘Οˆ1(𝑠 + 1)ψ(𝑠)

    𝑐2(𝑠 + 1)

    π‘‘βˆ’1

    𝑠=𝑑1

    [𝜌(𝑠 + 1)

    ψ(𝑠)βˆ’ 𝜌2(𝑠 + 1)]

    ≀ βˆ’ (π‘ž0 βˆ’ 1

    π‘ž0) βˆ‘ [

    ψ(𝑠)𝑝(𝑠)

    π‘ž(𝑠 + βˆ’ 𝜎)βˆ’ (

    π‘ž0π‘ž0 βˆ’ 1

    )ψ1(𝑠 + 1)

    4ψ(𝑠)𝑐2(𝑠 + 1)]

    π‘‘βˆ’1

    𝑠=𝑑1

    However, in view of (27), this inequality contradicts (25). Hence 𝐺1 = πœ‘. By Lemma 2, 𝐺3 =

    𝐺4 = πœ‘ due to (7). The proof is complete.

    Corollary 1: Suppose that (LH1) - (LH3) are satisfied and (7) holds. If there is a constant 𝐢𝓀

    such that

    ψ2(𝑑)𝑝(𝑑)𝑐2(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)ψ1(𝑑)β‰₯ 𝐢𝓀 >

    π‘ž04(π‘ž0 βˆ’ 1)

    , (29)

    then 𝐺1 = 𝐺3 = 𝐺4 = πœ‘.

    To attain the oscillation of all solutions, it remains to eliminate the solutions of 𝐺2

    type.

    Lemma 5: Suppose that (LH1) – (LH4) are satisfied. If

    limπ‘‘β†’βˆž

    𝑠𝑒𝑝 βˆ‘π‘(𝑠)πœ‡(𝑑 + βˆ’ 𝜎, 𝑠 + βˆ’ 𝜎)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑+πœ‰βˆ’πœŽ

    >π‘ž0

    π‘ž0 βˆ’ 1. (30)

    then 𝐺2 = πœ‘.

    Proof: For sake of the contradiction, let (30) be satisfied 𝑦 ∈ 𝐺2. Choose 𝑑1 > 𝑑0 such that

    π‘₯(𝑑) > 0, π‘₯(𝑑 βˆ’ 𝜎) > 0 π‘Žπ‘›π‘‘ π‘₯(𝑑 βˆ’ ) > 0. Using (13) in (1), we obtain

    𝐸3𝑦(𝑑) +π‘žπ‘œ βˆ’ 1

    π‘ž0

    𝑝(𝑑)

    π‘ž(𝑑 + βˆ’ 𝜎)𝑦(𝑑 + βˆ’ 𝜎) ≀ 0. (31)

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org447

  • Using the monotonicity of 𝐸2𝑦, one can easily verify that

    βˆ’πΈ1𝑦(𝑒) β‰₯ 𝐸1𝑦(𝑣) βˆ’ 𝐸1𝑦(𝑒) = βˆ‘πΈ2𝑦(𝑠)

    𝑐2(𝑠)

    π‘£βˆ’1

    𝑠=𝑒

    β‰₯ 𝐸2𝑦(𝑣) βˆ‘1

    𝑐2(𝑠)

    π‘£βˆ’1

    𝑠=𝑒

    (32)

    for 𝑣 β‰₯ 𝑒 β‰₯ 𝑑1. Summing the latter inequality from 𝑒 π‘‘π‘œ 𝑣 βˆ’ 1, we obtain,

    𝑦(𝑒) β‰₯ 𝐸2𝑦(𝑣) βˆ‘1

    𝑐1(𝑠)

    π‘£βˆ’1

    𝑠=𝑒

    βˆ‘1

    𝑐2(π‘₯)

    π‘£βˆ’1

    π‘₯=𝑠

    = 𝐸2𝑦(𝑣)πœ‡(𝑣, 𝑒). (33)

    setting 𝑒 = 𝑠 + βˆ’ 𝜎 and 𝑣 = 𝑑 + βˆ’ 𝜎 in (33), we find

    𝑦(𝑠 + βˆ’ 𝜎) β‰₯ 𝐸2𝑦(𝑑 + βˆ’ 𝜎)πœ‡(𝑑 + βˆ’ 𝜎, 𝑠 + βˆ’ 𝜎). (34)

    On the other hand, summing (31) from 𝑑 + βˆ’ 𝜎 to 𝑑 βˆ’ 1 and using (34), we see that

    𝐸2𝑦(𝑑 + βˆ’ 𝜎) β‰₯ 𝐸2𝑦(𝑑 + βˆ’ 𝜎) βˆ’ 𝐸2𝑦(𝑑) β‰₯π‘ž0 βˆ’ 1

    π‘ž0βˆ‘

    𝑝(𝑠)𝑦(𝑠 + βˆ’ 𝜎)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑+πœ‰βˆ’πœŽ

    β‰₯π‘ž0 βˆ’ 1

    π‘ž0𝐸2𝑦(𝑑 + βˆ’ 𝜎) βˆ‘

    𝑝(𝑠)πœ‡(𝑑 + βˆ’ 𝜎, 𝑠 + βˆ’ 𝜎)

    π‘ž(𝑠 + βˆ’ 𝜎).

    π‘‘βˆ’1

    𝑠=𝑑+πœ‰βˆ’πœŽ

    Dividing the above inequality by 𝐸2𝑦(𝑑 + βˆ’ 𝜎) and taking the limsup on both sides of the

    resulting inequality as 𝑑 β†’ ∞, we obtain a contradiction with (30). The proof is complete.

    Lemma 6: Suppose that (LH1) – (LH4) are satisfied and let Ξ² be a constant satisfying (4)

    eventually. If

    limπ‘‘β†’βˆž

    𝑠𝑒𝑝(𝑑 + βˆ’ 𝜎)𝛽 βˆ‘π‘(𝑠)οΏ½ΜƒοΏ½(𝑑 + βˆ’ 𝜎, 𝑠 + βˆ’ 𝜎)

    π‘ž(𝑠 + βˆ’ 𝜎)

    π‘‘βˆ’1

    𝑠=𝑑+πœ‰βˆ’πœŽ

    >π‘ž0

    π‘ž0 βˆ’ 1, (35)

    then 𝐺2 = πœ‘.

    Proof: Setting 𝑒 = 𝑑 + βˆ’ 𝜎 and 𝑣 = 𝑑 in (33), we obtain

    𝑦(𝑑 + βˆ’ 𝜎) β‰₯ 𝐸2𝑦(𝑑)πœ‡(𝑑, 𝑑 + βˆ’ 𝜎) = 𝐸2𝑦(𝑑 + 1)πœ‡(𝑑, 𝑑 + βˆ’ 𝜎) (36)

    By (4), (31) and (36), we have

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org448

  • βˆ† (𝑑𝛽𝐸2𝑦(𝑑)) = π›½π‘‘π›½βˆ’1𝐸2𝑦(𝑑 + 1) + 𝑑

    𝛽𝐸3𝑦(𝑑)

    ≀ π›½π‘‘π›½βˆ’1𝐸2𝑦(𝑑 + 1) βˆ’ (π‘ž0 βˆ’ 1

    π‘ž0)

    𝑑𝛽𝑝(𝑑)𝑦(𝑑 + βˆ’ 𝜎)

    π‘ž(𝑑 + βˆ’ 𝜎)

    ≀ π›½π‘‘π›½βˆ’1𝐸2𝑦(𝑑 + 1) βˆ’ (π‘ž0 βˆ’ 1

    π‘ž0)

    𝑑𝛽𝑝(𝑑)𝐸2𝑦(𝑑 + 1)πœ‡(𝑑, 𝑑 + βˆ’ 𝜎)

    π‘ž(𝑑 + βˆ’ 𝜎)

    = π‘‘π›½βˆ’1𝐸2𝑦(𝑑 + 1) [𝛽 βˆ’ (π‘ž0 βˆ’ 1

    π‘ž0)

    𝑑𝑝(𝑑)πœ‡(𝑑, 𝑑 + βˆ’ 𝜎)

    π‘ž(𝑑 + βˆ’ 𝜎)] ≀ 0.

    That is 𝑑𝛽 𝐸2𝑦(𝑑 + 1) is eventually non-increasing. From here, we obtain that

    βˆ’πΈ1𝑦(𝑒) β‰₯ 𝐸1𝑦(𝑣) βˆ’ 𝐸1𝑦(𝑒) = βˆ‘πΈ2𝑦(𝑠)𝑠

    𝛽

    𝑠𝛽𝑐2(𝑠)

    π‘£βˆ’1

    𝑠=𝑒

    β‰₯ 𝐸2𝑦(𝑣)𝑣𝛽 βˆ‘

    1

    𝑠𝛽𝑐2(𝑠)

    π‘£βˆ’1

    𝑠=𝑒

    (37)

    for 𝑣 β‰₯ 𝑒 β‰₯ 𝑑1. Proceeding as in the proof of Lemma 5 with (32) replaces by (37), one arrives

    at a contradiction with (35). The proof is complete.

    Theorem 2: Suppose that (LH1) – (LH4) are satisfied. If (18) (or (25)) and (30) (or (35))

    hold, then (1) is oscillatory.

    References

    [1] R.P. Agarwal, Difference Equations and Inequalities. Theory, Methods and

    Applications, Marcel Dekker, Inc., New York, 2000.

    [2] R.P. Agarwal , M. Bohner, S.R. Grace, D. O’Regan, Discrete Oscillation Theory,

    Hindawi Publishing Corporation, New York, 2005.

    [3] R.P. Agarwal , S.R. Grace, Oscillation of certain third-order difference equations,

    Comput. Math. Appl.42(3-5)(2001), 379-384

    [4] R.P. Agarwal, S.R. Grace, D. O’ Regan, On the oscillation of certain third-order

    difference equations, Adv. Difference Equ.3(2005),345-367.

    [5] M.F.Atlas, A. Tiryaki, A. Zafer, Oscillation of third-order nonlinear delay difference

    equations, Turkish J. Math.36 (3) (2012), 422-436.

    [6] M. Bohner, C. Dharuman, R. Srinivasan, E. Thandapani, Oscillation criteria for third

    –order nonlinear functional difference equations with damping, Appl. Math. Inf.

    Sci.11 (3) (2017), 669-676.

    [7] S. R. Grace, R.P. Agarwal, J.R. Graef, Oscillation criteria for certain third order

    nonlinear difference equations, Appl. Anal. Dicrete Math. 3(1) (2009), 27-38.

    [8] J. R. Graef, E. Thandapani, Oscillatory and asymptotic behavior of solutions of third

    order delay difference equations, Funkcial. Ekvac. 42 (3) (1999), 355-369.

    [9] Horng-Jaan Li and Cheh-Chih Yeh, Oscillation Criteria for Second-Order Neutral

    Delay

    Difference Equations, Computers Math. Applic. Vol 36, No.10-12, pp.123-132,1998

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org449

  • [10] W.G., Kelley, A.C. Peterson, Difference Equations; An Introduction with

    Application, New

    York. Academic Press 1991.

    [11] Martin Bohner, C.Dharuman, R.Srinivasan and E.Thandapani, Oscillation Criteria

    for Third-

    Order Nonlinear Functional Difference Equations With Damping, Appl. Math. Inf.

    Sci. 11,

    No. 3, 1-8 (2017).

    [12] S .H. Saker, Oscillation of third-order difference equations, Port. Math., 61(2004),

    249-257.

    [13] E. Thandapani, S. Pandian, R. K. Balasubramaniam, Oscillatory behavior of

    solutions of

    third order quasilinear delay difference equations, Stud. Univ. Zilina Math. Ser. 19

    (1)

    (2005), 65-78.

    [14] E. Thandapani, S. Selvarangam, Oscillation theorems of second order quasilinear

    neutral

    difference equations. J. Math. Comput. Sci. 2012;2:866-879

    [15] Yadaiah Arupula, V. Dharmaiah, Oscillation of third order nonlinear delay

    difference

    equation, Int. J. Math& Appl., 6 (3) (2018), 181-191.

    Journal of Information and Computational Science

    Volume 10 Issue 9 - 2020

    ISSN: 1548-7741

    www.joics.org450