Necessary and sufficient conditions for oscillation of first order nonlinear neutral differential equations

Σχετικά έγγραφα
Oscillation Criteria for Nonlinear Damped Dynamic Equations on Time Scales

J. of Math. (PRC) u(t k ) = I k (u(t k )), k = 1, 2,, (1.6) , [3, 4] (1.1), (1.2), (1.3), [6 8]

Oscillation criteria for two-dimensional system of non-linear ordinary differential equations

Nonlinear Analysis: Modelling and Control, 2013, Vol. 18, No. 4,

Linear singular perturbations of hyperbolic-parabolic type

Positive solutions for a multi-point eigenvalue. problem involving the one dimensional

( ) ( t) ( 0) ( ) dw w. = = β. Then the solution of (1.1) is easily found to. wt = t+ t. We generalize this to the following nonlinear differential

Multiple positive periodic solutions of nonlinear functional differential system with feedback control

Vol. 40 No Journal of Jiangxi Normal University Natural Science Jul. 2016

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

r t te 2t i t Find the derivative of the vector function. 19. r t e t cos t i e t sin t j ln t k Evaluate the integral.

Appendix. The solution begins with Eq. (2.15) from the text, which we repeat here for 1, (A.1)

Every set of first-order formulas is equivalent to an independent set

Managing Production-Inventory Systems with Scarce Resources

ST5224: Advanced Statistical Theory II

Approximation of the Lerch zeta-function

2 Composition. Invertible Mappings

SCITECH Volume 13, Issue 2 RESEARCH ORGANISATION Published online: March 29, 2018

( ) ( ) ( ) Fourier series. ; m is an integer. r(t) is periodic (T>0), r(t+t) = r(t), t Fundamental period T 0 = smallest T. Fundamental frequency ω

Concrete Mathematics Exercises from 30 September 2016

Uniform Convergence of Fourier Series Michael Taylor

= e 6t. = t 1 = t. 5 t 8L 1[ 1 = 3L 1 [ 1. L 1 [ π. = 3 π. = L 1 3s = L. = 3L 1 s t. = 3 cos(5t) sin(5t).

Existence of travelling wave solutions in delayed reaction diffusion systems with applications to diffusion competition systems

Fractional Colorings and Zykov Products of graphs

Nowhere-zero flows Let be a digraph, Abelian group. A Γ-circulation in is a mapping : such that, where, and : tail in X, head in

The choice of an optimal LCSCR contract involves the choice of an x L. such that the supplier chooses the LCS option when x xl

ORDINAL ARITHMETIC JULIAN J. SCHLÖDER

Example Sheet 3 Solutions

A Simple Version of the Lucas Model

4.6 Autoregressive Moving Average Model ARMA(1,1)

Reservoir modeling. Reservoir modelling Linear reservoirs. The linear reservoir, no input. Starting up reservoir modeling

On Strong Product of Two Fuzzy Graphs

Analysis of optimal harvesting of a prey-predator fishery model with the limited sources of prey and presence of toxicity

OSCILLATION CRITERIA FOR SECOND ORDER HALF-LINEAR DIFFERENTIAL EQUATIONS WITH DAMPING TERM

1. Introduction and Preliminaries.

C.S. 430 Assignment 6, Sample Solutions

A Note on Intuitionistic Fuzzy. Equivalence Relation

Tridiagonal matrices. Gérard MEURANT. October, 2008

Matrices and Determinants

On shift Harnack inequalities for subordinate semigroups and moment estimates for Lévy processes

Electronic Companion to Supply Chain Dynamics and Channel Efficiency in Durable Product Pricing and Distribution

On local motion of a general compressible viscous heat conducting fluid bounded by a free surface

HOMEWORK 4 = G. In order to plot the stress versus the stretch we define a normalized stretch:

Lecture 21: Properties and robustness of LSE

Commutative Monoids in Intuitionistic Fuzzy Sets

Statistical Inference I Locally most powerful tests

EE512: Error Control Coding

Other Test Constructions: Likelihood Ratio & Bayes Tests

Homework 3 Solutions

University of Washington Department of Chemistry Chemistry 553 Spring Quarter 2010 Homework Assignment 3 Due 04/26/10

Congruence Classes of Invertible Matrices of Order 3 over F 2

Finite Field Problems: Solutions

arxiv: v1 [math.ap] 10 Apr 2017

Coefficient Inequalities for a New Subclass of K-uniformly Convex Functions

Section 8.3 Trigonometric Equations

2. Let H 1 and H 2 be Hilbert spaces and let T : H 1 H 2 be a bounded linear operator. Prove that [T (H 1 )] = N (T ). (6p)

The third moment for the parabolic Anderson model

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

6.1. Dirac Equation. Hamiltonian. Dirac Eq.

FRACTIONAL INTEGRATION OF THE PRODUCT OF BESSEL FUNCTIONS OF THE FIRST KIND. Abstract

5. Choice under Uncertainty

Multi-dimensional Central Limit Theorem

On Generating Relations of Some Triple. Hypergeometric Functions

The Student s t and F Distributions Page 1

F19MC2 Solutions 9 Complex Analysis

Solution Series 9. i=1 x i and i=1 x i.

Cubic Γ-n normed linear spaces

CHAPTER 25 SOLVING EQUATIONS BY ITERATIVE METHODS

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

Multi-dimensional Central Limit Theorem

Lecture 2. Soundness and completeness of propositional logic

Homomorphism in Intuitionistic Fuzzy Automata

is the home less foreign interest rate differential (expressed as it

Reminders: linear functions

ON LOCAL MOTION OF A COMPRESSIBLE BAROTROPIC VISCOUS FLUID WITH THE BOUNDARY SLIP CONDITION. Marek Burnat Wojciech M. ZajĄczkowski. 1.

A General Note on δ-quasi Monotone and Increasing Sequence

Practice Exam 2. Conceptual Questions. 1. State a Basic identity and then verify it. (a) Identity: Solution: One identity is csc(θ) = 1

Oscillation of nonlinear second-order neutral delay differential equations

Almost all short intervals containing prime numbers

Exercises 10. Find a fundamental matrix of the given system of equations. Also find the fundamental matrix Φ(t) satisfying Φ(0) = I. 1.

Jordan Journal of Mathematics and Statistics (JJMS) 4(2), 2011, pp

TRM +4!5"2# 6!#!-!2&'!5$27!842//22&'9&2:1*;832<

( y) Partial Differential Equations

( )( ) ( ) ( )( ) ( )( ) β = Chapter 5 Exercise Problems EX α So 49 β 199 EX EX EX5.4 EX5.5. (a)

Areas and Lengths in Polar Coordinates

Areas and Lengths in Polar Coordinates

The challenges of non-stable predicates

The Euler Equations! λ 1. λ 2. λ 3. ρ ρu. E = e + u 2 /2. E + p ρ. = de /dt. = dh / dt; h = h( T ); c p. / c v. ; γ = c p. p = ( γ 1)ρe. c v.

SOME PROPERTIES OF FUZZY REAL NUMBERS

On a four-dimensional hyperbolic manifold with finite volume

ANSWERSHEET (TOPIC = DIFFERENTIAL CALCULUS) COLLECTION #2. h 0 h h 0 h h 0 ( ) g k = g 0 + g 1 + g g 2009 =?

MINIMAL CLOSED SETS AND MAXIMAL CLOSED SETS

A Two-Sided Laplace Inversion Algorithm with Computable Error Bounds and Its Applications in Financial Engineering

9.09. # 1. Area inside the oval limaçon r = cos θ. To graph, start with θ = 0 so r = 6. Compute dr

Math221: HW# 1 solutions

Abstract Storage Devices

Generating Set of the Complete Semigroups of Binary Relations

CRASH COURSE IN PRECALCULUS

ΕΡΓΑΣΙΑ ΜΑΘΗΜΑΤΟΣ: ΘΕΩΡΙΑ ΒΕΛΤΙΣΤΟΥ ΕΛΕΓΧΟΥ ΦΙΛΤΡΟ KALMAN ΜΩΥΣΗΣ ΛΑΖΑΡΟΣ

Problem Set 3: Solutions

Transcript:

J. Mah. Anal. Appl. 321 (2006) 553 568 www.elsevier.com/locae/jmaa Necessary sufficien condiions for oscillaion of firs order nonlinear neural differenial equaions X.H. ang a,, Xiaoyan Lin b a School of Mahemaical Sciences Compuing echnology, Cenral Souh Universiy, Changsha, Hunan 410083, PR China b Deparmen of Mahemaics, Huaihua Universiy, Huaihua, Hunan 418008, PR China Received 17 March 2005 Available online 19 Sepember 2005 Submied by William F. Ames Absrac In his paper, we prove ha every soluion of he firs order nonlinear neural differenial equaion ] m x() px( τ) + q() x( σj ) βj sign x( σ 1 ) ] = 0,, oscillaes if only if ( m ) ] q(s)exp τ 1 ln p β j 1 s ds =, when ( m β j 1) ln p<0, q(s)ds =, when ( m β j 1) ln p>0, where p, τ>0, β j > 0, σ j 0, j = 1, 2,...,m, q C(, ), 0, )). 2005 Published by Elsevier Inc. his work is parially suppored by he NNSF (No. 10471153) of China. * Corresponding auhor. E-mail addresses: angxh@mail.csu.edu.cn (X.H. ang), xiaoyanlin98@homail.com (X. Lin). 0022-247X/$ see fron maer 2005 Published by Elsevier Inc. doi:10.1016/j.jmaa.2005.07.078

554 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 Keywords: Firs order neural differenial equaion; Superlinear; Sublinear; Oscillaion 1. Inroducion Consider he firs nonlinear neural delay differenial equaion ] m x() px( τ) + q() x( σj ) βj sign x( σ 1 ) ] = 0,, (1.1) where p,τ > 0, β j > 0, σ j 0, j = 1, 2,...,m, q C(, )), 0, )). When m β j = 1, he oscillaory behavior of soluions of Eq. (1.1) is similar o he linear neural delay differenial equaion, which has been sudied by many auhors, see 1,2,5,7,8] he references cied herein. When m β j 1, we only find hree papers 3,4,6] which deal wih he oscillaory behavior of soluions of some special forms of Eq. (1.1). he resuls obained here are he following. heorem 1.1. 3] Assume ha p = 0 m β j > 1. hen he following conclusions hold: (i) If here exiss λ>0 such ha m β j e λσ j < 1, (1.2) lim inf q()exp( e λ ) ] > 0, (1.3) hen every soluion of Eq. (1.1) oscillaes. (ii) If, for large, q(s) 0, s, + σ ], (1.4) where σ = max{σ 1,σ 2,...,σ m }, ha exiss μ>0 such ha m β j e μσ j > 1, (1.5) lim sup q()exp( e μ ) ] <, (1.6) hen Eq. (1.1) has an evenually posiive soluion. heorem 1.2. 4] Assume ha p = 1 m β j < 1. hen every soluion of Eq. (1.1) oscillaes if only if s β q(s)ds =, where β = m β j. (1.7)

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 555 heorem 1.3. 4] Assume ha p = 1 m β j > 1. hen every soluion of Eq. (1.1) oscillaes if only if sq(s)ds =. (1.8) heorem 1.4. (6], see also 1, heorem 3.4.6]) Assume ha p<1 m β j < 1. hen every soluion of Eq. (1.1) oscillaes if only if q(s)ds =. Besides he several cases menioned in he above four heorems, we find no resuls in he lieraure on he oscillaion of soluions of Eq. (1.1) in he following wo cases: (i) p (0, 1) (1, ) m β j > 1; (ii) p (1, ) m β j < 1. In his paper, we shall esablish some necessary sufficien condiions for oscillaion of soluions of Eq. (1.1) in he above case (i) case (ii), respecively. As is cusomary, a soluion x() of Eq. (1.1) is said o be oscillaory if i has arbirarily large zeros. Oherwise, i is said o be nonoscillaory. hroughou of his paper, we denoe β = m β j. 2. he superlinear case m β j > 1 heorem 2.1. Assume ha <p<1 m β j > 1. hen every soluion of Eq. (1.1) oscillaes if only if ( m ) ] q(s)exp τ 1 ln p β j 1 s ds =. (2.1) Proof. Sufficiency. Lex() be a nonoscillaory soluion of Eq. (1.1). We may wihou loss of generaliy assume ha x() > 0for 1 for some 1. Se z() = x() px( τ). (2.2) hen i follows from (1.1), (2.1) (2.2) ha m z () = q() x( σj ) βj 0 ( 0), 2 = 1 + ρ, (2.3) here in he sequel, ρ = max{τ,σ 1,...,σ m }. his shows ha z() is nonincreasing on 2, ). Hence, z() > 0, 2, (see 1, Lemma 5.1.4]), from (2.2) (2.3), we have n x() = p i z( iτ) + p n+1 x( nτ τ) i=0 (1.9)

556 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 = 1 τ n p i z( iτ) 1 τ i=0 +ρ (n+1)τ+ρ n iτ+ρ i=0 (i+1)τ+ρ p (+ρ s)/τ z(s) ds 1 τ 2 + nτ 2 + (n + 1)τ, n = 1, 2,..., which yields x( σ j ) 1 τ 1 τ +ρ σ j 2 +ρ 3 p (+ρ s)/τ z(s) ds +ρ 2 +ρ p (+ρ σ j s)/τ z(s) ds 1 τ p (+ρ s)/τ z(s) ds, p (+ρ s)/τ z(s) ds, 2 +ρ p (+ρ s)/τ z(s) ds 3 := 2 + 2ρ, j = 1, 2,...,m. (2.4) Subsiuing (2.4) ino (2.3), we obain z 1 ] β () q() p (+ρ s)/τ z(s) ds, τ 3. (2.5) 3 Se y() = z(s) ds, 3. (2.6) 3 hen i follows from (2.5) ha z () τ β p β(+ρ)/τ q() y() ] β, 3, (2.7) so z(s) τ β Subsiuing his ino (2.6), we have s y() τ β 3 τ β 3 p β(u+ρ)/τ q(u) y(u) ] β du, s 3. s p β(u+ρ)/τ q(u) y(u) ] β duds u p β(u+ρ)/τ ] β q(u) y(u) dsdu 3 = τ β+1 p β(s+ρ)/τ( p 3/τ ) ] β q(s) y(s) ds, ln p 3. (2.8) 3

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 557 Se w() = p β(+ρ)/τ( p /τ p 3/τ ) q() y() ] β, 3. (2.9) hen (2.8) yields ( ) β ( τ β+1 ) β w() w(s)ds p β(+ρ)/τ( p /τ p 3/τ ) q(), ln p 3. (2.10) 3 Choose a 1 > 3 such ha 1 3 w(s)ds > 0. hen from (2.10), we have ( τ β+1 ln p 1 I follows ha ) β w() 1 ( 3 p β(+ρ)/τ( p /τ p 3/τ ) q()d 3 w(s)ds ) β ( d = 1 1 β 1 ( 1 1 ) 1 β w(s)ds, > 1. β 1 3 w(s)ds ) 1 β ( 3 w(s)ds ) 1 β ] 1 p β(+ρ)/τ( p /τ p 3/τ ) q()d <. (2.11) Noe ha p /τ p 3/τ p /τ 1 p ( 1 3 )/τ ], 1. hen (2.11) implies ha 1 p (β 1)s/τ q(s)ds <, (2.12) which conradics o (2.1) so he sufficiency is proved. Necessiy. We only need o prove ha he condiion q(s)exp τ 1 ln p(β 1)s ] ds < (2.13) implies ha Eq. (1.1) has an evenually posiive soluion. Noe ha s = τ ln p p βu/τ q(u)duds p βu/τ ( q(u) p u/τ p /τ ) du

(p /τ y () ) = τ β p β( ρ τ)/τ q() y() ] β 0,. (2.20) 558 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 τ p (β 1)s/τ τ q(s)ds = ln p ln p Hence, from (2.13), we have Choose > + ρ such ha ( ) 2 β p βρ/τ τ s q(s)exp τ 1 ln p(β 1)s ] ds. p βu/τ q(u)duds <. (2.14) Define he sequence of funcions {y n ()} as follows: y 0 () = 2,, y n+1 () = 1 + τ β p βρ/τ s p β(u τ)/τ q(u)duds < 1. (2.15) s p β(u τ)/τ q(u) y n (u) ] β duds, (2.16), n= 1, 2,.... (2.17) By (2.15) (2.17) by inducion, i is easy o verify ha 1 y n+1 () y n () y 0 () = 2,, n= 1, 2,.... hen he limi lim n y n () = y() exiss for, ) 1 y() 2for, ). Applying Lebesgue s monoone convergence heorem o (2.17), we obain y() = 1 + τ β p βρ/τ I follows ha p /τ y () = τ β p βρ/τ s p β(u τ)/τ q(u) y(u) ] β duds,. (2.18) p β(s τ)/τ q(s) y(s) ] β ds,, (2.19) I follows from (2.15) ha here exiss a 1 > + τ such ha τ β p βρ/τ p 1/τ 1 p β(s τ)/τ ] β q(s) y(s) ds < τ 1. (2.21) If q() 0forlarge, hen x() = e ln p/τ is an evenually posiive soluion of Eq. (1.1). So, in he sequel, we only consider he case when q() 0 evenually, so p β(s τ)/τ q(s) y(s) ] β ds > 0,. (2.22)

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 559 u(s) ds + y( 1 ), 1. (2.25) Se u() = p /τ y (),. (2.23) hen from (2.19) (2.23), we have 0 <u()<τ 1 p 1/τ, u () 0, 1, (2.24) y() = 1 Define a funcion v() as follows: τ 1 u( 1 + τ)( 1 ), 1 1 + τ, v() = p v( + τ) u( + τ)], < 1, (2.26) u() + pv( τ), 1 + iτ < 1 + (i + 1)τ, i = 1, 2,... I is easy o see ha v() coninues on, ) v() > 0for> 1 v() = u() + pv( τ), 1 + τ. (2.27) From (2.24) (2.26), we have v() u( 1 + τ) u( 1 )<τ 1 p 1/τ, 1 1 + τ. (2.28) hen from (2.24), (2.25), (2.27) (2.28) by using he fac y() 1for,wehave n 1 n 1 v() = p i u( iτ) + p n v( nτ) p i u( iτ) + p n τ 1 p 1/τ 1 τ = 1 τ i=0 n 1 iτ i=0 (i+1)τ nτ i=0 p ( τ s)/τ u(s) ds + τ 1 p n+ 1/τ p ( τ s)/τ u(s) ds + τ 1 p n+ 1/τ 1 p ( τ s)/τ u(s) ds + τ 1 p ( τ)/τ 1 = τ τ p( τ)/τ u(s) ds + 1 1 1 ] 1 τ p( τ)/τ u(s) ds + y( 1 ) = 1 τ p( τ)/τ y(), 1 1 + nτ 1 + (n + 1)τ, n = 1, 2,.... Noe ha y () 0for, i follows from he above ha v( σ j ) 1 τ p( τ σ j )/τ y( σ j ) 1 τ p( τ σ j )/τ y(), 1 + 2ρ, j = 1, 2,...,m. (2.29) ]

560 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 By (2.19), (2.23), (2.27) (2.29), we obain v() = pv( τ)+ u() = pv( τ)+ τ β p βρ/τ pv( τ)+ τ β p βρ/τ pv( τ)+ his shows ha he inequaliy v() pv( τ)+ q(s) q(s) p β(s τ)/τ q(s) p β(s τ)/τ q(s) y(s) ] β ds m τp ( s+τ+σ j )/τ v(s σ j ) ] β j ds m v(s σj ) ] β j ds, 1 + 2ρ. m v(s σ j ) βj sign v(s σ 1 ) ] ds, 1 + 2ρ, (2.30) has a posiive soluion v() on 1 + 2ρ, ). Similar o he proof of 1, Lemma 5.1.5], we can prove ha he corresponding equaion x() = px( τ)+ q(s) m x(s σ j ) βj sign x(s σ 1 ) ] ds, 1 + 2ρ, (2.31) has also a posiive soluion x() on 1 + 2ρ, ). Obviously, x() is also he evenually posiive soluion of Eq. (1.1), so he necessiy is proved. he proof is complee. heorem 2.2. Assume ha p>1 m β j > 1. hen every soluion of Eq. (1.1) oscillaes if only if q(s)ds =. (2.32) Proof. Sufficiency. Lex() be a nonoscillaory soluion of Eq. (1.1). We may wihou loss of generaliy assume ha x() > 0for 1 for some 1. Se z() as in (2.2). hen i follows from p>1 (2.32) ha z() < 0 z () 0for 2 for some 2 > 1, see 1, heorem 3.2.9]. Choose a posiive ineger n such ha nτ ρ. hen from (2.2), we have so x() > 1 p n z( + nτ), 2, (2.33) x( σ j )> 1 p n z( + nτ σ j ) 1 p n z(), 2 + ρ, j = 1, 2,...,m. Subsiuing his ino (2.3), we have ( z () q() 1 ) β p n z(), 2 + ρ. I follows ha

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 561 2 +ρ q(s)ds p nβ 2 +ρ ( z(s) ) βz (s) ds = p nβ (β 1) 1( z( 2 + ρ) ) 1 β ( z() ) 1 β ] <p nβ (β 1) 1 z( 2 + ρ) ] 1 β, 2 + ρ, so 2 +ρ q(s)ds <, which conradics o (2.32) so he sufficiency is proved. Necessiy. We only need o prove ha he condiion q(s)ds < (2.34) implies ha Eq. (1.1) has an evenually posiive soluion. By (2.34), we can choose > +τ +σ such ha q(s)ds p 1. 2 (2.35) Define he sequence of funcions {x n ()} as follows: x 0 () = 1,, p 1 { p 1 2 + x n ( + τ) x n+1 () = + +τ q(s) m x n (s σ j )] β j ds},, x n+1 ( ), <. (2.36) (2.37) n = 1, 2,... By (2.35) (2.37) by inducion, i is easy o verify ha p 1 2p x n+1() x n () x 0 () = 1,, n= 1, 2,.... hen he limi lim n x n () = x() exiss for, ) (p 1)/2p x() 1for, ). Applying Lebesgue s monoone convergence heorem o (2.37), we obain { +τ x() = p 1 p 1 + x( + τ)+ q(s) 2 m x(s σj ) ] } β j ds,. (2.38) I is easy o see ha x() is also he evenually posiive soluion of Eq. (1.1), so he necessiy is proved. he proof is complee.

562 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 3. he sublinear case m β j < 1 heorem 3.1. Assume ha p>1 m β j < 1. hen every soluion of Eq. (1.1) oscillaes if only if ( m ) ] q(s)exp τ 1 ln p β j 1 s ds =. (3.1) Proof. Sufficiency. Lex() be a nonoscillaory soluion of Eq. (1.1). We may wihou loss of generaliy assume ha x() > 0 for all 1. Se z() = px( τ) x(). hen i follows from (1.1) (3.1) ha m z () = q() x( σ j ) βj 0 ( 0), 2 = 1 + ρ. (3.3) his shows ha z() is nondecreasing on 2, ). Hence, z() > 0, 2, (see 1, heorem 3.2.9]), from (3.2) (3.3), we have x() = 1 τ which yields p i z( + iτ) 1 τ i=1 i=1 p i +iτ +(i 1)τ p (s+τ )/τ z(s) ds = 1 τ p( τ)/τ x( σ j ) 1 τ p( τ σ j )/τ σ j z(s) ds z(s) ds 1 τ +iτ i=1 +(i 1)τ z(s) ds, 2, 1 τ p( τ ρ)/τ (3.2) p (s+τ )/τ z(s) ds z(s) ds, 2 + ρ, j = 1, 2,...,m. (3.4) Subsiuing (3.4) ino (3.3), we obain z 1 β () q() τ p( τ ρ)/τ z(s) ds], 3 = 2 + ρ. (3.5) Se y() = hen i follows from (3.5) ha z(s) ds, 3. (3.6) z () τ β p β( τ ρ)/τ q() y() ] β, 3, (3.7)

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 563 so s z(s) τ β 3 Subsiuing his ino (3.6), we have p β(u τ ρ)/τ ] β q(u) y(u) du, s 3. Se y() τ β τ β s 3 p β(u τ ρ)/τ ] β q(u) y(u) duds p β(u τ ρ)/τ q(u) y(u) ] β = τ β+1 p β(τ+ρ)/τ ln p u dsdu p (β 1)s/τ q(s) y(s) ] β ds, 3. (3.8) w() = p (β 1)/τ q() y() ] β, 3. (3.9) hen (3.8) yields w() ( β ( τ w(s)ds) β+1 p β(τ+ρ)/τ ) β p (β 1)/τ q(), 3. (3.10) ln p Inegraing (3.10) from 3 o,wehave ( τ β+1 p β(τ+ρ)/τ I follows ha ln p ) β 3 p (β 1)/τ q()d 3 w() = 1 1 β ( ( 3 w(s)ds) β d w(s)ds ) 1 β. 3 p (β 1)/τ q()d <, (3.11) which conradics o (3.1) so he sufficiency is proved. Necessiy. We only need o prove ha he condiion q(s)exp τ 1 ln p(β 1)s ] ds < (3.12) implies ha Eq. (1.1) has an evenually posiive soluion. Noe ha

564 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 s p βu/τ q(u)duds = Hence, from (3.12), we have s τ ln p = τ ln p p (β 1)s/τ q(s)ds q(s)exp τ 1 ln p(β 1)s ] ds. p βu/τ q(u)duds <. (3.13) Choose > + ρ such ha ( ) 2 β s p β(u+ρ)/τ q(u)duds < 1. (3.14) τ If q() 0for, hen x() = e ln p/τ is an evenually posiive soluion of Eq. (1.1). So, in he sequel, we only consider he case when q() 0for. Define he sequence of funcions {y n ()} as follows: y 0 () = 2,, y n+1 () = 1 + τ β, n= 1, 2,.... s p β(u+ρ)/τ q(u) y n (u) ] β duds, By (3.14) (3.16) by inducion, i is easy o verify ha 1 y n+1 () y n () y 0 () = 2,, n= 1, 2,.... (3.15) (3.16) hen he limi lim n y n () = y() exiss for, ) 1 y() 2for, ). Applying Lebesgue s monoone convergence heorem o (3.16), we obain y() = 1 + τ β s p β(u+ρ)/τ q(u) y(u) ] β duds,. (3.17) I follows ha p /τ y () = τ β p β(s+ρ)/τ q(s) y(s) ] β ds,, (3.18) Se ( p /τ y ()) = τ β p β(+ρ)/τ q() y() ] β 0,. (3.19) u() = p /τ y (),. (3.20)

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 565 hen from (3.17), (3.18), (3.19) (3.20), we have u() 0, u () 0,, (3.21) y() = u(s) ds + 1, 1. (3.22) From (3.21) (3.22), we find p i u( + iτ) 1 τ i=1 = 1 τ +ρ+iτ i=1 +ρ+(i 1)τ +ρ = 1 τ p(+ρ)/τ p (+ρ s)/τ u(s) ds +ρ 1 τ p(+ρ)/τ p (+ρ s)/τ u(s) ds u(s) ds +ρ u(s) ds + 1 = 1 τ p(+ρ)/τ y( + ρ),. (3.23) Noe ha q() 0for, so we can choose 1 > + ρ such ha u( 1 ) = τ β 1 I follows from (3.21) ha p β(s+ρ)/τ q(s) y(s) ] β ds > 0. (3.24) u() > 0, u () 0, 1. (3.25) Define a funcion v() as follows: v() = p i u( + iτ),. (3.26) i=1 By (3.23) (3.25), i is easy o see ha v() coninues on, ) v() > 0for, v() = 1 p u( + τ)+ v( + τ) ],, (3.27) v() 1 τ p(+ρ)/τ y( + ρ),. (3.28) I follows from (3.26) he fac ha y () 0for ha ]

566 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 v( σ j ) 1 τ p(+ρ σ j )/τ y( + ρ σ j ) 1 τ p(+ρ σ j )/τ y(), 1,j= 1, 2,...,m. (3.29) hen, by (3.18), (3.20), (3.24), (3.27) (3.29), we obain v() = 1 ] v( + τ)+ u( + τ) p = 1 +τ v( + τ)+ τ β p β(s+ρ)/τ q(s) y(s) ] ] β ds p = 1 u( 1 ) + v( + τ)+ τ β p +τ 1 +τ 1 u( 1 ) + v( + τ)+ τ β p 1 u( 1 ) + v( + τ)+ p his shows ha he inequaliy { v() 1 u( 1 ) + v( + τ)+ p +τ 1 +τ 1 1 q(s) q(s) ] p β(s+ρ)/τ ] β q(s) y(s) ds p β(s+ρ)/τ q(s) m τp ( s ρ+σ j )/τ v(s σ j ) ] ] β j ds m v(s σj ) ] ] β j ds, 1. m v(s σj ) βj sign v(s σ 1 ) ] } ds, 1, (3.30) has a posiive soluion v() on 1, ). Similar o he proof 1, Lemma 5.1.5], we can prove ha he corresponding equaion { x() = 1 +τ m u( 1 ) + x( + τ)+ q(s) x(s σj ) βj sign x(s σ 1 ) ] } ds, p 1 1, (3.31) has also a posiive soluion x() on 1, ). Obviously, x() is also he evenually posiive soluion of Eq. (1.1), so he necessiy is proved. he proof is complee. 4. Remarks Combining heorems 1.4, 2.1, 2.2 3.1, we have he following corollaries. Corollary 4.1. Assume ha ( m β j 1) ln p<0. hen every soluion of Eq. (1.1) oscillaes if only if ( m ) ] q(s)exp τ 1 ln p β j 1 s ds =. (4.1)

X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 567 Corollary 4.2. Assume ha ( m β j 1) ln p>0. hen every soluion of Eq. (1.1) oscillaes if only if q(s)ds =. (4.2) Now we consider he firs nonlinear neural delay differenial equaion ] m x() p()x( τ) + q() x( σj ) βj sign x( σ 1 ) ] = 0,, (4.3) where p C(, ), 0, )), τ, β j > 0 q() are he same as in Eq. (1.1). In view of he proof of heorems 2.1 2.3, we have he following heorems. heorem 4.1. Assume ha m β j > 1. hen he following conclusions hold: (i) If here exiss a p 1 (0, 1) such ha p 1 p() 1,, (4.4) ( m ) ] q(s)exp τ 1 ln p 1 β j 1 s ds =, (4.5) hen every soluion of Eq. (4.3) oscillaes; (ii) If here exiss a p 2 (0, 1) such ha 0 p() p 2,, (4.6) ( m ) ] q(s)exp τ 1 ln p 2 β j 1 s ds <, (4.7) hen Eq. (4.3) has an evenually posiive soluion; (iii) If here exis p 3,p 4 (1, ) such ha p 3 p() p 4,, (4.8) hen every soluion of Eq. (4.3) oscillaes if only if q(s)ds =. (4.9) heorem 4.2. Assume ha m β j < 1. hen he following conclusions hold: (i) If here exiss p 1 (0, 1) such ha 0 p() p 1,, (4.10)

568 X.H. ang, X. Lin / J. Mah. Anal. Appl. 321 (2006) 553 568 hen every soluion of Eq. (4.3) oscillaes if only if q(s)ds = ; (4.11) (ii) If here exiss a p 2 (1, ) such ha 1 p() p 2,, (4.12) ( m ) ] q(s)exp τ 1 ln p 2 β j 1 s ds =, (4.13) hen every soluion of Eq. (4.3) oscillaes; (iii) If here exiss a p 3 (1, ) such ha p() p 3,, (4.14) ( m ) ] q(s)exp τ 1 ln p 3 β j 1 s ds <, (4.15) hen Eq. (4.3) has an evenually posiive soluion. References 1] L.H. Erbe, Q. Kong, B.G. Zhang, Oscillaion heory for Funcional Differenial Equaions, Dekker, New York, 1995. 2] I. Gyori, G. Ladas, Oscillaion heory of Delay Differenial Equaions wih Applicaions, Clarendon Press, Oxford, 1991. 3] X.H. ang, Oscillaion for firs order nonlinear delay differenial equaions, J. Mah. Anal. Appl. 264 (2001) 510 521. 4] X.H. ang, X.Q. Li, Necessary sufficien condiions for oscillaion of nonlinear neural differenial equaions, Hunan Ann. Mah. 17 (1997) 57 60 (in Chinese). 5] X.H. ang, J.S. Yu, Firs order nonlinear differenial inequaliies wih deviaing argumens, Appl. Mah. J. Chinese Univ. Ser. B 15 (2000) 21 27. 6] L.W. Wang, Oscillaion of firs order nonlinear neural differenial equaions, Aca Mah. Appl. Sinica 14 (1991) 348 359. 7] J.S. Yu, Firs order nonlinear differenial inequaliies wih deviaing argumens, Aca Mah. Sinica 33 (1990) 152 159. 8] B.G. Zhang, J.S. Yu, Oscillaion nonoscillaion for neural differenial equaions, J. Mah. Anal. Appl. 172 (1993) 11 23.