• 沒有找到結果。

Asymptotic behavior of positive solutions of the nonlinear differential equation t²u''= u{^n},1 < n

N/A
N/A
Protected

Academic year: 2021

Share "Asymptotic behavior of positive solutions of the nonlinear differential equation t²u''= u{^n},1 < n"

Copied!
9
0
0

加載中.... (立即查看全文)

全文

(1)

ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu

ASYMPTOTIC BEHAVIOR OF POSITIVE SOLUTIONS OF THE NONLINEAR DIFFERENTIAL EQUATION t2u00= un

MENG-RONG LI, HSIN-YU YAO, YU-TSO LI

Abstract. In this article we study properties of positive solutions of the or-dinary differential equation t2u00= unfor 1 < n ∈ N, we obtain conditions for

their blow-up in finite time, and some properties for global solutions. Equa-tions containing more general nonlinear terms are also considered.

1. Introduction

Some interesting results on the blow-up, blow-up rates, and estimates for the life-span of solutions of the Emden-Fowler equation and the semi-linear wave equation u + f (u) = 0 have been obtained, as shown in the references.

Here we wish to study the Emden-Fowler type wave equation, i.e. solutions, independent of the space variable x, of the equation t2utt− ∆u = un for n > 1.

The existence and uniqueness of local solutions of the initial-value problem t2u00= un, 1 < n ∈ N,

u(1) = u0, u0(1) = u1,

(1.1) follow by standard arguments. Considering the transformation t = es, u(t) = v(s), we have t2u00(t) = −v

s(s) + vss(s), v(s)n = −vs(s) + vss(s) and v(0) = u(1) = u0; vs(0) = u0(1) = u1, the problem (1.1) can be transformed into

vss(s) − vs(s) = vn(s), v(0) = u0, vs(0) = u1.

(1.2) Thus, the existence of local solutions u for (1.1) in (1, T ) is equivalent to the existence of local solutions v for (1.2) in (0, ln T ). In this article, we give estimates for the life-span T∗ of positive solutions u of (1.1) in three different cases. The main results are as follows:

(a) u1= 0, u0> 0: T∗≤ ek1, k1:= s0+ 2(n+3) 8− 2 n−1v(s0) 1−n 2 , ε ∈ (0, 1); (b) u1> 0, u0> 0: (i) E(0) ≥ 0, T∗≤ ek2, k 2:= n−12 q n+1 2 u 1−n 2 0 ; (ii) E(0) < 0, T∗≤ ek3, k 3:= n−12 uu01; (c) u1< 0, u0∈ (0, (−u1) 1 n): u(t) ≤ (u0− u1− un 0) + (u1+ un0)t − u n 0ln t.

2000 Mathematics Subject Classification. 34A34, 34C11, 34C60.

Key words and phrases. Nonlinear differential equation; Emden-Fowler equation; blow-up rate. c

2013 Texas State University - San Marcos.

Submitted October 5, 2013. Published November 20, 2013.

(2)

where E(0) is defined in the next section and s0 is given by (3.3).

In Section 6, we replace the nonlinear term vn by a more general increasing function f (v).

2. Notation and some lemmas For a given function v, we use the following functions

a(s) := v(s)2, E(0) := u21− 2 n + 1u n+1 0 , J (s) := a(s) −n−1 4 ,

where u0and u1 are the given initial conditions.

By an easy calculation we can obtain the following two Lemmas; we shall omit the proof of the first lemma.

Lemma 2.1. Suppose that v ∈ C2[0, T ] is the solution of (1.2), then E(s) = vs(s)2− 2 Z s 0 vs(r)2dr − 2 n + 1v(s) n+1= E(0), (2.1) (n + 3)vs(s)2= (n + 1)E(0) + a00(s) − a0(s) + 2(n + 1) Z s 0 vs(r)2dr, (2.2) J00(s) = n 2− 1 4 J (s) n+3 n−1(E(0) − a 0(s) n + 1+ 2 Z s 0 vs(r)2dr), (2.3) J0(s)2= J0(0)2+(n − 1) 2 4 E(0)(J (s) 2(n+1) n−1 − J (0) 2(n+1) n−1 ) +(n − 1) 2 2 J (s) 2(n+1) n−1 Z s 0 vs(r)2dr. (2.4)

Lemma 2.2. For u0> 0, the positive solution v of (1.2) satisfies:

(i) If u1≥ 0, then vs(s) > 0 for all s > 0 (2.5) (ii) If u1< 0, u0∈ (0, (−u1)

1

n), then vs(s) < 0 for all s > 0. (2.6)

Proof. (i) Since vss(0) = u1+ un0 > 0, we know that vss(s) > 0 in [0, s1) and vs(s) is increasing in [0, s1) for some s1 > 0. Moreover, since v and vs are increasing in [0, s1), vss(s1) = vs(s1) + v(s1)n > vs(0) + v(0)n > 0 for all s ∈ [0, s1) and vs(s1) > vs(s) > 0 for all s ∈ [0, s1), we know that there exists a positive number s2 > 0, such that vs(s) > 0 for all s ∈ [0, s1+ s2). Continuing this process, we obtain vs(s) > 0 for all s > 0, for which the solution exists.

(ii) Since vss(0) = vs(0) + v(0)n = u1+ un0 < 0, there exists a positive number s1 > 0 such that vss(s) < 0 in [0, s1), vs(s) is decreasing in [0, s1); therefore, vs(s) < vs(0) = u1< 0 for all s ∈ [0, s1) and v(s) is decreasing in [0, s1). Moreover, since v and vs are decreasing in [0, s1), vss(s) = vs(s) + v(s)n < vs(0) + v(0)n < 0 for all s ∈ [0, s1) and vs(s1) < vs(s) < 0 for all s ∈ [0, s1), we know that there exists a positive number s2 > 0, such that vs(s) < 0 for all s ∈ [0, s1+ s2). Continuing this process, we obtain vs(s) < 0 for all s > 0 in the interval of existence. 

3. Life-span of positive solutions of (1.1) when u1= 0, u0> 0 In this section we want to estimate the life-span of a positive solution u of (1.1) if u1 = 0, u0 > 0. Here the life-span T∗ of u means that u is the solution of equation (1.1) and u exists only in [0, T∗) so that the problem (1.1) possesses a positive solution u ∈ C2[0, T).

(3)

Theorem 3.1. For u1 = 0, u0 > 0, the positive solution u of (1.1) blows up in finite time; that is, there exists T∗< ∞ so that

u(t)−1→ 0 as t → T∗.

Proof. By (2.5), we know that vs(s) > 0, a0(s) > 0 for all s > 0 provided that u1= 0, u0> 0. By Lemma 2.1, a00(s) − a0(s) = 2(vs(s)2+ v(s)n+1), (a0(s)e−s)0 = e−s(a00(s) − a0(s)) = 2e−s(vs(s)2+ v(s)n+1), a0(s)e−s= 2 Z s 0 e−r(vs(r)2+ v(r)n+1)dr ≥ 4 Z s 0 e−rvs(r)v(r) n+1 2 dr,

and a0(0) = 0, hence we have a0(s)e−s≥ 8 n + 3(v(r) (n+3)/2e−r|s r=0+ Z s 0 v(r)(n+3)/2e−rdr) = 8 n + 3(v(s) (n+3)/2e−s− v(0)(n+3)/2) + 8 n + 3 Z s 0 v(r)(n+3)/2e−rdr. Since a0(s) > 0 for all s > 0, v is increasing on (0, ∞) and

a0(s)e−s≥ 8 n + 3(v(s) (n+3)/2e−s− v(0)(n+3)/2) + 8 n + 3 Z s 0 v(0)(n+3)/2e−rdr = 8 n + 3(v(s) (n+3)/2e−s− v(0)(n+3)/2) + 8 n + 3v(0) (n+3)/2(1 − e−s), a0(s) ≥ 8 n + 3(v(s) (n+3)/2− v(0)(n+3)/2) = 8 n + 3(v(s) (n+3)/2− u(n+3)/2 0 ). (3.1) Using u1= 0 and integrating (1.2), we obtain

vs(s) = v(s) − u0+ Z s 0 v(r)ndr, vs(s) ≥ v(s) − u0+ Z s 0 v(0)ndr = v(s) − u0+ un0s, (e−sv(s))s= e−s(vs(s) − v(s)) ≥ e−s(un0s − u0), a0(s) ≥ 8 n + 3(v(s) (n+3)/2− v(0)(n+3)/2) = 8 n + 3(v(s) (n+3)/2− u(n+3)/2 0 ). (3.2)

According to (3.2), and since v0(s) > 0, v(s)(n+3)/2≥ (u

0+ un0(es− 1 − s))(n+3)/2, for all  ∈ (0, 1), we obtain

v(s)(n+3)/2≥ (u0+ un0(e s− 1 − s))(n+3)/2, v(s)(n+3)/2− 8u(n+3)/20 ≥ (u0+ un0(e s− 1 − s))(n+3)/2− 8u(n+3)/2 0 ≥ (u(n+3)/20 + u n(n+3) 2 0 (e s− 1 − s)(n+3)/2) − 8un+32 0 = ( − 8)u(n+3)/20 + u n(n+3) 2 0 (e s − 1 − s)(n+3)/2. Now, we want to find a number s0> 0 such that

es0− s 0= 1 + 8 −   u n+3 2 (1−n) 0 2/(n+3) . (3.3)

(4)

This means that there exists a number s0 > 0 satisfying (3.3) with  ∈ (0, 1) such that

v(s)(n+3)/2− 8u(n+3)/20 ≥ 0 for all s ≥ s0. From (3.1), it follows that

a0(s) ≥ 8 n + 3v(s) (n+3)/2 8 n + 3u (n+3)/2 0 = 8 −  n + 3v(s) (n+3)/2+v(s)(n+3)/2− 8u (n+3)/2 0 n + 3 ≥ 8 −  n + 3v(s) (n+3)/2, for all s ≥ s 0. For all s ≥ s0,  ∈ (0, 1), we obtain that

2v(s)vs(s) ≥ 8 −  n + 3v(s) (n+3)/2, v(s)−n+12 vs(s) ≥ 8 −  2(n + 3), 2 1 − n(v(s) 1−n 2 )s≥ 8 −  2(n + 3) and hence (v(s)1−n2 )s≤ 8 −  2(n + 3) 1 − n 2 . Integrating the above inequality, we conclude that

v(s)1−n2 ≤ v(s0) 1−n 2 − 8 −  2(n + 3) n − 1 2 (s − s0). Thus, there exists a number

s∗1≤ s0+ 2(n + 3) 8 −  2 n − 1v(s0) 1−n 2 =: k1

such that v(s)−1 → 0 for s → s∗

1, that is, u(t)−1 → 0 as t → e

k1, which implies

that the life-span T∗ of a positive solution u is finite and T∗≤ ek1. 

4. Life-span of positive solutions of (1.1) when u1> 0, u0> 0 In this section we estimate the life-span of a positive solution u of (1.1) whenever u1> 0, u0> 0.

Theorem 4.1. For u1 > 0, u0 > 0, the positive solution u of (1.1) blows up in finite time; that is, there exists a number T∗< ∞ so that

u(t)−1→ 0 as t → T∗.

Proof. We separate the proof into two parts depending on whether E(0) ≥ 0 or E(0) < 0.

(i) Assume that E(0) ≥ 0. By (2.1) and (2.5) we have vs(s)2− 2 n + 1v(s) n+1 ≥ E(0), vs(s)2≥ 2 n + 1v(s) n+1+ E(0), v s(s) ≥ r 2 n + 1v(s) n+1+ E(0).

(5)

Since E(0) ≥ 0, we obtain vs(s) ≥ r 2 n + 1v(s) n+1 2 , v(s)−n+12 · vs(s) ≥ r 2 n + 1, (v(s)1−n2 )s≤1 − n 2 r 2 n + 1. Integrating the above inequality, we obtain

v(s)1−n2 ≤ u 1−n 2 0 + 1 − n 2 r 2 n + 1s. Thus, there exists

s∗2≤ 2 n − 1 r n + 1 2 u 1−n 2 0 =: k2 such that v(s)−1→ 0 for s → s∗

2; that is, u(t)−1→ 0 as t → ek2, which means that the life-span T∗ of a positive solution u is finite and T∗≤ ek2.

(ii ) Assume that E(0) < 0. From (2.1) and (2.5) we obtain that J0(s) = −n−1

4 a(s) −n+3

4 a0(s), a0(s) > 0, vs(s) > 0 for all s > 0 and

J0(s) = −n − 1 2 s 2 n + 1+ E(0)a(s) −n+1 2 + 2a(s)− n+1 2 Z s 0 vs(r)2dr ≤ −n − 1 2 r 2 n + 1+ E(0)a(s) −n+1 2 , J (s) ≤ J (0) −n − 1 2 Z s 0 r 2 n + 1+ E(0)a(r) −n+1 2 dr.

Since E(0) < 0 and a0(s) > 0 for all s > 0, J (s) ≤ J (0) −n − 1 2 Z s 0 r 2 n + 1+ E(0)a(0) −n+1 2 dr = a(0)−n−14 −n − 1 2 r 2 n + 1+ E(0)a(0) −n+1 2 s.

Thus, there exists a number s∗3 ≤ 2 n − 1a(0) −n−1 4 ( 2 n + 1+ E(0)a(0) −n+1 2 )−12 =: k3

such that J (s∗3) = 0 and a(s)−1 → 0 for s → s∗

3; that is, u(t)−1 → 0 as t → ek3. This means that the life-span T∗ of u is finite and T∗≤ ek3. 

5. Life-span of positive solutions of (1.1) when u1< 0

Finally, we estimate the life-span of a positive solution u of (1.1) when u1< 0. Theorem 5.1. For u1< 0, u0∈ (0, (−u1)

1

n) we have

u(t) ≤ (u0− u1− un0) + (u1+ un0)t − u n 0ln t,

(6)

and in particular, if E(0) ≥ 0, we have u(t) ≤ (u 1−n 2 0 + n − 1 2 r 2 n + 1ln t) 2 1−n.

Proof. (i) By (1.2) and integrating this equation with respect to s, we get vs(s) = (u1− u0) + v(s) +R

s 0 v(r)

ndr. By (2.6), we have that v is decreasing and

vs(s) ≤ (u1− u0) + v(s) + Z s 0 v(0)ndr = (u1− u0) + v(s) + un0s, e−sv(s) − u0≤ (u1− u0) Z s 0 e−rdr + un0 Z s 0 re−rdr = (u1− u0)(1 − e−s) + un0(−se −s− e−s+ 1); that is, u(t) ≤ (u0− u1) + u1t + un0(t − 1 − ln t) = (u0− u1− un0) + (u1+ un0)t − u n 0ln t. (ii) If E(0) ≥ 0, by (2.1), we have

vs(s)2− 2 n + 1v(s) n+1= E(0) + 2 Z s 0 vs(r)2dr ≥ E(0), vs(s)2≥ E(0) + 2 n + 1v(s) n+1 2 n + 1v(s) n+1. By (2.6), we obtain that −vs(s) ≥ q 2 n+1v(s) n+1 2 , 2 n−1(v(s) 1−n 2 )s≥ q 2 n+1 and r 2 n + 1s ≤ 2 n − 1(v(s) 1−n 2 − v(0) 1−n 2 ), v(s)1−n2 ≥ (u 1−n 2 0 + n − 1 2 r 2 n + 1s). Then, we know that

v(s) ≤ (u 1−n 2 0 + n − 1 2 r 2 n + 1s) 2 1−n, for all s ≥ 0; that is, u(t) ≤ (u 1−n 2 0 + n − 1 2 r 2 n + 1ln t) 2 1−n for all t ≥ 1.  6. A generalization of Theorem 4.1

In this section we want to extent the blow-up result for the following generaliza-tion of (1.2),

vss(s) − vs(s) = f (v), v(0) = v0, vs(0) = v1,

(6.1) where f is an increasing continuous function with f (0) = 0. We have the following result.

(7)

Theorem 6.1. Suppose that f is an increasing function with f (0) = 0 and suppose v is a positive solution of (6.1). If F (v) :=R0vf (r)dr, then

¯

E(s) := vs(s)2− 2 Z s

0

vs(r)2dr − 2F (v(s)) (6.2) is constant. Furthermore, if there exists a positive constant k such that F (s) ≥ ksp+1, p > 1 for all s ≥ 0, and v

1> 0, then the life span of v is finite.

Proof. By an argument similar to that used in proving (2.1), we easily obtain that ¯

E(s) is a constant. Since f is increasing, we have

vf (v) = (v − 0) · (f (v) − f (0)) ≥ 0 for v ≥ 0, thus (v2)s− 2v2(s) ≥ 2v0(v1− v0) + 2 Z s 0 v2s(r)dr. (6.3)

By (6.2) and (6.3), we have ¯E(s) = v2

1− 2F (v0) := ¯E, and v2(s) ≥ v0v1e2s− v0(v1− v0), (6.4) (v2)s− 2v2(s) ≥ 2v0(v1− v0) + 2 Z s 0 v2s(r)dr = 2v0(v1− v0) + 2 Z s 0 ( ¯E + 2F (v(r)) + 2 Z r 0 vs(η)2dη)dr ≥ 2v0(v1− v0) + 2 ¯Es + 4k Z s 0 vp+1(r)dr ≥ 2v0(v1− v0) + 2 ¯Es + 4ks1−p( Z s 0 v2(r)dr)p+12 . (6.5) LetRs 0 v 2(r)dr := b(s), e−sb(s) = B(s). Then b(s)00− 2b(s)0 ≥ 2v0(v1− v0) + 2 ¯Es + 4ks1−pb(s) p+1 2 for s > 0 and by (6.5), we have (e−s(b(s)0− b(s)))0 = e−s(b(s)00− 2b(s)0+ b(s)) ≥ 2v0(v1− v0) + 2 ¯Es + 4ks1−pe−sb p+1 2 + e−sb(s) = 2v0(v1− v0) + 2 ¯Es + 4ks1−p(e−sb(s)) p+1 2 e p−1 2 s+ e−sb(s) ≥ 0, (6.6) (e−sb(s))00= (e−s(b(s)0− b(s)))0 ≥ 2v0(v1− v0) + 2 ¯Es + 4k( es s2) p+1 2 (e−sb(s)) p+1 2 + e−sb(s) ≥ 2v0(v1− v0) + 2 ¯Es + 22− p+1 2 k(e−sb(s)) p+1 2 + e−sb(s), B(s)00≥ 2v0(v1− v0) + 2 ¯Es + 22− p+1 2 kB(s) p+1 2 + B(s). (6.7)

From (6.4) it follows that b(s) ≥ v0v1 2 (e 2s − 1) − v0(v1− v0)s, B(s) ≥ v0v1 2 (e s− e−s) − v 0(v1− v0)se−s,

(8)

2v0(v1− v0) + 2 ¯Es + B(s)

2 ≥ 0, s ≥ s0 for some s0> 0. Therefore,

B(s)00≥B(s) 2 ≥ v0v1 5 e s, s ≥ s 0, B0(s) ≥ v0v1 5 (e s − es0) + B0(s 0) > 0, s ≥ s1 for some s1> s0. By (6.7), for all s ≥ s1, ((B(s)0)2)0= 2B(s)0B(s)00 ≥ 22−p−1 2 kB(s) p+1 2 B(s)0 = 22−p−12 k 2 p + 3(B(s) p+3 2 )0, (B0)2− B0(s1)2≥ 23−p−12 k p + 3 (B p+3 2 − B(s1) p+3 2 ), (B0)2≥ 2 3−p−12 k p + 3 (B p+3 2 − B(s1) p+3 2 ) + B0(s1)2 = 2 3−p−12 k 2(p + 3)B p+3 2 + 23− p−1 2 k 2(p + 3)B p+3 2 − 2B(s1) p+3 2  + B0(s1)2, B0≥ 2 7−p 4 √ p + 3B p+3 4 for s ≥ s2,

for some s2> s1; hence, for s ≥ s2, 4 1 − p(B 1−p 4 )0 = B p+3 −4B0(s) ≥ 2 7−p 4 √ p + 3, B(s)1−p4 ≤ B(s2) 1−p 4 −p − 1 4 27−p4 √ p + 3(s − s2) for all s ≥ s2> 0.

Thus B(s) blows up at a finite s∗. Since b(s) = esB(s), b(s) also blows up at s. Further, since v2(s) = b0(s) ≥ 2b(s), v(s) blows up at s, as well.

 Acknowledgements. Thanks are due to Professors Tai-Ping Liu, Ton Yang and Shih-Shien Yu for their continuous encouragement; to the anonymous referee for his/her helpful comments; and to Professor K. Schmitt for his comments and sug-gestions on Theorem 6.1. The authors want to thank Metta Education, Grand Hall and Auria Solar for their financial assistance.

References

[1] M. R. Li; Nichtlineare Wellengleichungen 2. Ordnung auf beschr¨ankten Gebieten. PhD-Dissertation, T¨ubingen 1994.

[2] M. R. Li; Estimates for the life-span of solutions of semilinear wave equations. Comm. Pure Appl. Anal., 2008, 7(2): 417-432.

[3] M. R. Li; On the blow-up time and blow-up rate of positive solutions of semilinear wave equations u − up= 0 in 1-dimensional space. Comm.Pure Appl. Anal., to appear.

[4] M. R. Li; On the semilinear wave equations. Taiwanese J. Math., 1998, 2(3): 329-345. [5] M. R. Li, L. Y. Tsai; On a system of nonlinear wave equations. Taiwanese J. Math., 2003,

(9)

[6] M. R. Li, L. Y. Tsai; Existence and nonexistence of global solutions of some systems of semilinear wave equations. Nonlinear Analysis, 2003, 54: 1397-1415.

[7] Meng-rong Li, Jenet Pai; Quenching problem in some semilinear wave equations. Acta Math. Sci., 2008, 28B(3): 523-529.

[8] M. R. Li; On the generalized Emden-Fowler Equation u00(t)u(t) = c

1+ c2u0(t)2with c1≥ 0,

c2≥ 0. Acta Math. Sci., 2010 30B(4): 1227-1234.

[9] M. R. Li; Blow-up results and asymptotic behavior of the Emden-Fowler Equation. Acta Math. Sci., 2007, 4: 703-734.

[10] Meng-Rong Li, Yue-Loong Chang, Yu-Tso Li; A Mathematical Model of Enterprise Com-petitive Ability and Performance through Emden-Fowler Equation (II), Acta Mathematica Scientia, 2013, 33(4): 1127-1140.

Meng-Rong Li

Department of Mathematical Sciences, National Chengchi University, Taipei, Taiwan E-mail address: liwei@math.nccu.edu.tw

Hsin-Yu Yao

Department of Mathematical Sciences, National Chengchi University, Taipei, Taiwan E-mail address: diadia0914@gmail.com

Yu-Tso Li

Department of Aerospace and Systems Engineering, Feng Chia University, Taichung, Taiwan

參考文獻

相關文件

Using the solution U of Riemann problem to obtain an e approximate solution for the solution U of balance laws..

By contrast to the often heavily edited and revised printed Chán scriptures of the Sòng period (many of them eventually being integrated in the official Buddhist canon

The five-equation model system is composed of two phasic mass balance equations, the mixture momentum equation, the mixture total energy equation, and an evolution equation for

A constant state u − is formed on the left side of the initial wave train followed by a right facing (with respect to the velocity u − ) dispersive shock having smaller

When the spatial dimension is N = 2, we establish the De Giorgi type conjecture for the blow-up nonlinear elliptic system under suitable conditions at infinity on bound

One of the main results is the bound on the vanishing order of a nontrivial solution u satisfying the Stokes system, which is a quantitative version of the strong unique

You are given the wavelength and total energy of a light pulse and asked to find the number of photons it

substance) is matter that has distinct properties and a composition that does not vary from sample