Skip to main content

Full text of "On Abstract Strichartz Estimates and the Strauss Conjecture for Nontrapping Obstacles"

See other formats



1. Introduction. 

"q^" ■ AND YI ZHOU 




The purpose of this paper is to show how local energy decay estimates for certain linear 
\^ ' wave equations involving compact perturbations of the standard Laplacian lead to optimal 

global existence theorems for the corresponding small amplitude nonlinear wave equations 
with power nonlinearities. To achieve this goal, at least for spatial dimensions n = 3 and 
p ^ ' , 4, we shall show how the aforementioned linear decay estimates can be combined with 

"abstract Strichart" estimates for the free wave equation to prove corresponding estimates 
for the perturbed wave equation when n > 3. As we shall see, we are only partially 
' successful in the latter endeavor when the dimension is equal to two, and therefore, at 

, present, our applications to nonlinear wave equations in this case are limited. 

Let us start by describing the local energy decay assumption that we shall make 
throughout. We shall consider wave equations on the exterior domain H. C R" of a 
. compact obstacle: 



^ ■ (1.1) 

' [d^ ~ ^sMt, x) = F{t, x), (t, x) eR+xfl 
dtu{0, • ) = g 

^{Bu){t,x) = 0, on M+ X dfl, 

where for simplicity we take B to either be the identity operator (Dirichlet-wave equation) 
I or the inward pointing normal derivative d^, (Neumann- wave equation). We shall also 

assume throughout that the spatial dimension satisfies n > 2. 

■ The operator Ag is the Laplace-Beltrami operator associated with a smooth, time 

, independent Riemannian metric gjk{x) which we assume equals the Euclidean metric 5jk 

• for \x\ > R, some R. The set is assumed to be either all of R", or else = R"\/C 

where /C is a compact subset of < i? with smooth boundary. 

We can now state the main assumption that we shall make. 

The first author was supported in part by the Grant-in-Aid for Young Scientists (B) (No. 18740069), 
The Ministry of Education, Culture, Sports, Science and Technology, Japan, and he would like to thank 
the Department of Mathematics at the Johns Hopkins University for the hospitality and financial support 
during his visit where part of this research was carried out. The second, third and fourth authors were 
supported by the National Science Foundation. The fifth author was supported by project 10728101 
of NSFC and the "111" project and Doctoral Programme Foundation of the Ministry of Education of 




Hypothesis B. Fix the boundary operator B and the exterior domain U, d M" as above. 
We then assume that given Rq > 

<ll/llH. + ll.9lli.+ / \\Fis,-)\\i.ds, 

whenever u is a solution of (II. ip with data {f{x),g{x)) and forcing term F{t, x) that both 
vanish for \x\ > Rq. 

Here A < B means that A is bounded by a constant times B, and, in what foUows, 
the constant might change at each occurrence. Also, ||/i||ffi(|2;|<_Ro) denotes the L^-norm 
of h and Wxh over the set {x £ ft : \x\ < Rq}. 

Let us review some important cases where the assumption (|1.2p is vahd. First of 
all, results from Vainberg [39], combined with the propagation of singularity results of 
Melrose and Sjostrand [21], imply that if A is the standard Euclidean Laplacian and O 
is nontrapping, then if m is a solution of (II. 1|) with data of fixed compact support and 
forcing term F = 0, then with u' — {dtu, V xu), 

\W{t, ■ < a(i)||w'(0, • )||l2, 

where a{t) — 0((1 + t)^*^"^^^) for either the Dirichlet-wave equation or the Neumann- 
wave equation when n > 3. For n = 2, if d^l is assumed to be nonempty one has 
a{t) = 0((log(2 + i))~^(l + for the Dirichlet-wave equation. Here we have used 

that, due to the Dirichlet boundary conditions and the fundamental theorem of calculus, 
the local norm can be controlled by the local norm of the gradient. Since these 
bounds yield a{t) e L^(M+), we conclude that Hypothesis B is valid in these cases. We 
remark that when = and Ag — A, then a ~ t~^ for large t (see [29 ), and so, in 
this case, a ^ L^(M+). Proofs of these results for n > 3 can be found in Melrose [23] 
and Ralston [29], while the result for the Dirichlet-wave equation for n = 2 follows from 
Vainberg [39] (see §4 and Remark 4 on p. 40). 

For the case where Ag is assumed to be a time- independent variable coefficient com- 
pact perturbation of A and f2 is assumed to be nontrapping with respect to the metric 
associated with Ag, one also has that (|1.2[) is valid for the Dirichlet-wave equation for all 
n > 3 as well for n = 2 if 917 7^ 0. See Taylor f3g and Burq [1]. 

Having described the main assumption about the linear problem, let us now describe 
the nonlinear equations that we shall consider. They are of the form 

' (9j2 - Ag)w(t, a;) = Fp {u{t, x)) , {t, x) eR+xQ. 

(1.3) Ibu^O, on M+ X dil 

u{0,x)^f{x), dtu{0,x) ^ g{x), x £ il, 

with B as above. We shall assume that the nonlinear term behaves like when u is 
small, and so we assume that 

(1.4) H'\diFp{u) \ < \u\P 


^ We are very grateful to Jim Ralston for patiently explaining these results and their history to us. 



when u is small. 

We shall be assuming that the data (and some of its derivatives) are small in certain 
Sobolev norms that we now describe. 

As in the earlier works that proved global Strichartz estimates ([4], [25], [33]), we shall 
restrict ourselves to the case where the Sobolev index 7 is smaller than n/2. One reason 
for this is that the Strichartz estimates that seem to arise in applications always have 
7 < 1. Another reason is that when I7I < n/2 , multiplication by a smooth function 
P G C^(M") is continuous from ij'''(R") to H'^{M.") and the two norms are equivalent on 
functions with fixed compact support. Recall that H''{W'-) is the homogeneous Sobolev 
space with norm given by 

WfWhiu^) - II (^)vir,.(«„) = (2-)-"/j i?r/(o r^e, 

while the inhomogeneous Sobolev space H''{W^) has norm defined by 

ii/iih.(r-.) = II (1 - A)^/v iil.(„„) = (2^)-" / I (1 + ler^'m I'at 

with / denoting the Fourier transform and A denoting the standard Laplacian. 

Let us now describe the Sobolev spaces on ft that we shall consider. Let /3 be a smooth 
cutoff on R" with /3 and 1 — /? respectively supported where |x| < 2R and |x| > R. Let 
f2' be the embedding of n < 2R} into the torus obtained by periodic extension of 
n n [-2R, 2i?]", so that dQ' = dQ. We define 

= + 11(1 - /?)/lk-'(M") 

= ll/3/lkj(0') + ll(l-/3)/llij.(K"), \l\<n/2. 

The spaces Hq{Q') are defined by a spectral decomposition of Ag|o/ subject to the 
boundary condition B. In the homogeneous spaces H'^{il) it is assumed that (1 — (3)f 
belongs to F''(M"), so that the Sobolev embedding Hl{n) ^ LP{n) holds with p = 
2n/{n — 27). From this, it is verified that the Sobolev spaces on are independent of the 
choice of (3 and R, and thus the Hg{il) and Hg(n) norms are equivalent on functions of 
fixed bounded support. We note that H^'^ {Vl) is the dual of H'^{il), and Hg''{n) is dual 
to Hg{n) for I7I < n/2. Also, for 7 a nonnegative integer, 


The Sobolev spaces as defined are verified to be an analytic interpolation scale of spaces. 
The above definition then agrees, for nonnegative integer 7, with the subspace of H'>'{n) 
such that i?(Ai/) = for all j for which the trace is well defined, and for general 7 by 
duality and interpolation. Finally, for every 7 the set of functions / G C^{Q) such that 
B{A^gf) = for all j > is dense in the norm. 

Our hypotheses regarding the data in p.3p will only involve certain 76 (0, ^), while 
the ones in the abstract Strichartz estimates to follow only involve certain 7 < (n— l)/2. 



In practice the useful Strichartz-type estimates always involve 7 G (0, 1]. This is the case 
for the mixed-norm Strichartz estimates of Keel and Tao jTBj and others for the case 
ri = R", A = Ag, as well as for the mixed-norm estimates for that we shall state. 

The data (/, g) in Theorem I 1 . ll below will have second derivatives belonging to {fl) x 
Hl^\n), where e (0, i), thus will locally belong to if xi/^+T (IT). The boundary 

condition for to locally belong to Hl~^'^{n) x H^'^'^iQ) for 7 e (0, i) is the same as 

for Hg{n) X H^{il), which for the Dirichlet case is /|ao = g\dn — 0, and for Neumann 
is diyfldn — 0. These are the assumptions placed on the data {f,g) in Theorem 

If we let 

{Z} — {di, Xjdk — Xkdj : I < I < n, 1 < j < k < n} 
then we can now state our existence theorem for (jl.3[) . 

Theorem 1.1. Let n = S or 4, and fix C M" and boundary operator B as above. 
Assume further that Hypothesis B is valid. 

Let p — Pc be the positive root of 

(1.5) (n-l)p2_(n + l)p-2 = 0, 

and fix Pc < p < {n -\- 3)/(n — 1). Then if 

(1-6) 7=1-^, 

there is an Sq > depending on fJ, B and p so that (|1.3|) has a global solution satisfying 
{Z'^u{t, ■ ), dtZ°'u{t, ■ )) e H'^ X H'^ ; |ct| < 2, t G M+, whenever the initial data satisfies 
the boundary conditions of order 2, and 

(1-7) E(ll^"/lli^S(o) + ll^"9bs-(f^))<^ 


with < £ < eo- 

In the case where $7 — M" and Ag = A it is known that p > Pc is necessary for global 
existence (see John [16], Glassey [12] and Sideris [31]). In this case under a somewhat 
more restrictive smallness condition global existence was established by John [IB] for the 
case where n = 3, then Glassey [T^] for n = 2, Zhou [IHI for n = 4, Lindblad and Sogge 
[22] for n < 8 and then Georgiev, Lindblad and Sogge [TT] for all n (see also Tataru [37)). 
For obstacles, when n = 4, Ag = A the results in Theorem 11.11 for the Dirichlet-wave 
equation outside of nontrapping obstacles under a somewhat more restrictive smallness 
assumption was obtained in [S]. 

Also, when = R'^, Ag = A, it was shown in Sogge [35, that, for the spherically 
symmetric case, the variant of the condition (|1.7p saying that the H^{R^) x H'^~^{E?) 
norm of the data be small with 7 as in (|1.6p is sharp. Further work in this direction (for 
the non-obstacle case) was done by Hidano [13], [H] and Fang and Wang [TU] . 

It is not difhcult to see that the condition (|1.7p is sharp in the sense that there are no 
global existence results for 7 > ^ — . To do this we use well known results concerning 
blowup solutions for (9^ — A)w — \v\^ , p > 0, in R+ x R" (see Levine [19]). Specifically, 
we shall use the fact that given 5 > Q one can find data (fo, vi) vanishing for |a;| < i? 
so that the solution of (9^ — A)i; — u(0, • ) = VQ,dtv{0, ■) = vi blows up within 



time 6. Next, let us assume that the above global existence results for (il, B, Ag) held 
for this nonlinearity and some 7 > § — in l|l-7p . Then, if A is sufficiently large, the 
Hi X H^-^ norm of (A-^/tf-i) vo( • /A), A-i-2/(p-i)wi( • /A)) would be bounded by its 
ij'^(R")xi/^-i(R") norm, which equals A"/2-2/(P^i)-T||(t;o, wi)||^^(g„^^^^_i(^„). Since 
this goes to zero as A ^ cx) for 7 > ^ — , we conclude that if the above existence results 
held for this value of 7 then we would obtain a global solution of {d^ — ^g)u\ — \u\\p, 
ux{t,x) = 0,{t,x) £R+xdn with initial data (A-2/(p-i)wo( • /A), A"i^2/(p-i)„^(- . /;^))^ 
Since vq and vi vanish for |a;| < i?, by finite propagation speed, if 5 > is small and 
fixed, then for large A if we extend ux to be zero on fl'^ then the resulting function 
would agree with the solution of the Minkowski space wave equation {df — A)v\ = \v\\p 
on [0,6\] X R" with data (A-2/(p-i)vo( • /A), A-i-2/(p-i)wi( • /A)). By scaling v{t,x) = 
X^/(p^^)vx{Xt, Xx) would then solve the Minkowski space equation {dt — A)v = \v\p on 
[0,(5] X M" with initial data {vq{x),vi{x)). As we noted before, we can always choose 
{vq,vi) so that this is impossible for a given i5 > 0, which allows us to conclude that 
the above existence results do not hold if the Sobolev exponent 7 in (|1.7p is larger than 

n _ 2 
2 p-1 ■ 

As a final remark, we point out that we have restricted ourselves to the case where 
p < {n + 3)/(n — 1) because of the techniques that we shall employ. However, since 
the solutions obtained are small, the above existence theorem leads to small-data global 
existence of (|1.3p when p is larger than or equal to the conformal power (rt + 3)/(n — 1). 

To prove Theorem 11.11 we shall use certain "abstract Strichartz estimates" which we 
now describe. Earlier works ([1], ^5\, (3^) have focused on establishing certain mixed 
norm, LjL^ estimates on M+ x for solutions of (|1.1[) . For certain applications, such 
as obtaining the Strauss conjecture in various settings, it is convenient to replace the 
norm with a more general one. To this end, we consider pairs of normed function spaces 
X(R") and X{n). The spaces are Realizable, in that ||/||x « WPfWx + 11(1 - l3)f\\x 
for smooth, compactly supported f3, with /? = 1 on a neighborhood of R"\51 in case 
X = X{il). Finally, we assume that 

(1-8) ll(l-/3)/llx(o)«ll(l-/3)/llx(E") 

for such f3. Weighted mixed spaces, as well as (ij^(]R"), 77^(51)) , are the examples 
used in the proof of Theorem 11.11 

We shall let || • denote the dual norm (respectively over R" and H.) so that 

\u\\x = sup 


; V dx 

An important example for us is when 

\W\\x II |a;|"u||Lp, 
for a given 1 < p < 00 and |a| < n/p, in which case the dual norm is 

ii«ii^' = iii^r"^^iiL.'> 

with p' denoting the conjugate exponent. 



We shall consider time Lebesgue exponents q >2 and assume that we have the global 
Minkowski abstract Strichartz estimates 

(1-9) II«IIl|x(rxr'.) ^ MO, ■)IIh7(r-.) + l|5tw(o, •)I1h7-i(R") > 

assuming that 

(1.10) {df-A)v^O inMxM". 



MlLfX(/xM") = ( J^\Ht,-)\\'xdt) 

We shall also consider analogous norms on / x / C K, 

\\u\\^x{ixQ) = ( / \\u{t, ■)\\x{n)'^^ 



In addition to Hypothesis B and (|1.9|) . we shall assume that we have the local abstract 
Strichartz estimates for fl: 

(1-11) II"IIl?x([o,i]xJ2) < \\.f\\Hi(n) + \\9\\Hl-\n)^ 

assuming that u solves (jl.ip with vanishing forcing term, i.e., 

(1.12) {df - Ag)ii in [0, 1] x fl. 

Definition 1.2. When ()1.9p and ()1.11|1 hold we say that (X, 7,g) is an admissible triple. 

We can now state our main estimate. 
Theorem 1.3. Let n>2 and assume that {X,j,q) is an admissible triple with 

(1.13) q>2 and j £ [-'-^, ''^]. 

Then if Hypothesis B is valid and if u solves (jl.ip with {&f — Ag)u = 0, we have the 
global abstract Strichartz estimates 

(1-14) l|w||Lfx(Exo) ^ WfWiiKn) + \\9\\hi~\uy 

The condition on 7 in (|1.13p is the one to ensure that 7 and 1 — 7 are both < (n — l)/2, 
which is what the proof seems to require. Unfortunately, for n = 2, this forces 7 to be 
equal to 1/2, while a larger range of 7 € (0, 1) is what certain applications require. For 
this reason, we are unable at present to show that the Strauss conjecture for obstacles 
holds when n = 2. See the end of the next section for further discussion. 

Corollary 1.4. Assume that (X, 7, g) and {Y^ \ — 7,r) are admissible triples and that 
Hypothesis B is valid. Also assume that (jl.l4p holds for (X, 7, 5) and {Y^l — 7,r), and 
that < 7 < 1 • Then we have the following global abstract Strichartz estimates for the 
solution of (jl.ip 

(1-15) l|w|lL?x(R+xn) < WfWiiKn) + \\9\\Hi-\n) + ll^llLr'F'(R+xn)' 

where r' denotes the conjugate exponent to r and \\ • ||y is the dual norm to \\ ■ \\y . 



For simplicity, in the corollary we have limited ourselves to the case where < 7 < 1 
since that is all that is needed for the applications. 

Let us give the simple argument that shows that (|1.15p follows from p.l4p . To prove 
(|1.15p , we may assume by (I1.14p that the initial data vanishes. By (|1.14p and the Duhamel 
formula, if P = Ag is the square root of minus the Laplacian (with the boundary 
conditions B), then we need show 

[ sin{{t~ s)P)p-^F{s, ■)ds 

< ll^^l 

LJ' Y'(R+xn) 

Since q > r', an application of the Christ-Kiselev lemma (cf. [7], [33], [36l chapter 4]) 

shows that it suffices to prove the estimate 

/ sm{{t- s)P)p-^F{s, ■)ds 

After factorization of the sin function, it suffices by (|1.14p to show that 

L?X(K+xO) - ll^llir'>"(K+x^^)- 

/ cos( sP)F( s, ■)ds + / P-^ sin( sP)F(s, ■)ds 

Jo H-'-Hn) Jo 


L';' Y' 

This, however, is the dual version of ()1.14p for {Y, 1 — 7, r). □ 

As a special case of (jl.lSp when the spaces X and Y are the standard Lebesgue spaces, 
we have the following 

Corollary 1.5. Suppose that n > 3 and that Hypothesis B is valid. Suppose that q,q > 2, 
r,f > 2 and that 

Inn In 

- + - - - 7 = - + - - 2 

q r Z q' r' 


1 2 

n — 1 71 — 1 

q r q r 2 
Then if the local Strichartz estimate (jl.lip holds respectively for the triples (L'"(f2), 7, g) 
and (L^(ri), 1 — 7, g), it follows that when u solves ()l.ip 

II"IIl?l;(k+xJ7) < WfWiiKn) + ll5llijj-i(o) + W^Wli'l^ {w^^ny 
These results also hold for n = 2 under the above assumption, provided that 7 ~ 1/2. 

These estimates of course are the obstacle versions of the mixed-norm estimates for M" 
and Ag = A. When n > 3 (and (jl.lip is valid) they include all the ones in the Keel-Tao 
theorem [18], excluding the cases where either g or g is 2. For the Dirichlet-wave operator 
{B = Id) these results were proved in odd dimensions by Smith and Sogge |33j and then 
by Burq [3] and Metcalfe [IS] for even dimensions. The Neumann case was not treated, 
but it follows from the same proof. Unfortunately, the known techniques seem to only 
apply to the case of 7 = 1/2 when n — 2, and Hypothesis B seems also to require B = Id 
and 3r2 ^ in this case. The restriction that 7 = 1/2 when n = 2 comes from the 
second part of (1.13), while for n > 3 this is not an issue due to the fact that the Sobolev 
exponents 7 in Corollary 1.5 always satisfy < 7 < 1. Also, at present, the knowledge of 
the local Strichartz estimates (jl.lll) when X = L''(f7) is limited. When fl is the exterior 



of a geodesically convex obstacle, they were obtained by Smith and Sogge [32] . Recently, 
there has been work on proving local Strichartz estimates when X = L^{Q,) for more 
general exterior domains ([5], [6], [3], [34]), but only partial results for a more restrictive 
range of exponents than the ones described in Corollary 11.51 have been obtained. 

2. Proof of Abstract Strichartz Estimates. 

As mentioned before, we shall prove (|1.14[) by adapting the arguments from [4], [25] 
and [53]. We shall assume that p.2p is valid for (fi, Ag) throughout. A key step in the 
proof of Theorem 11.31 will be to establish the following result that is implicit in [1] . 

Proposition 2.1. Let w solve the inhomogeneous wave equation in Minkowski space 

j {df - A)w ^ F on M+ X M" 
1 ^14=0 = dtw\t=Q = 0. 

Assume as above that (|1.9p is valid whenever v is a solution of the homogeneous wave 

sr that q > 2 and 7 > — 

F{t,x) = if |a;|>2i?, 

equation (|1.10p . Assume further that q > 2 and 7 > — -^^y^- Then, if 




At the end of this section we shall show that when n = 2 the assumption that 7 > 1/2 
when n = 2 is necessary even in the model case where X = L''(M") with 2/g + l/r = 1/2 
and l/q + 2/r — 1 — ■y. 

To prove Proposition l2.11 wc shall use our free space hypothesis (|1.9p and the following 
result from 33J. 

Lemma 2.2. Fix (3 € C^(R") and assume that Then 

/3(-)(e^*""/)(i, •) 




if \D\ = 

As was shown in [35], this lemma just follows from an application of Plancherel's 
theorem and the Schwarz inequality. The assumption that 7 < (n — l)/2 is easily seen 
to be sharp. 

To prove Proposition 12.11 we note that since we are assuming that g > 2, by the 
Christ-Kiselev lemma [7], it suffices to show that 


s*(*-^)l^l|i?|-i/3(-)G(s, ■)ds 



assuming that /3 e C^(R' 
inequality is majorized by 


If we apply (|1.9p . we conclude that the left side of this 


\D\-'+'^f3i-)Gis, ■)ds 



Since ||(1 - A)'^^-'^y^G{s, Olb = \\G{s, OIIh^-i, it suffices to see that 

^-ts\D\ |£)|-i+7^( . )(i _ A)'^^-''^/^H{s, ■ ) ds 

By duality, this is equivalent to the statement that 

(2-2) II (1 - A)(i--)/^/3( . )e-l^l|i?|-^+-^/.|L.(«^,«„) < \\hh.,^.y 

Since we are assuming that 7 > — -^^-j^, we have that 1 — 7 < Therefore, (|2.2p follows 

from Lemma [221 completing the proof of Proposition [27T1 □ 

To prove Theorem ll.3l we also need a similar result for solutions of the wave equation 
(HID for (r!,S,Ag). 

Proposition 2.3. Let u solve and assume that 

(2.3) f{x) = g{x) = = 0, when \x\ > 2R. 

Then if {X, 7, q) is an admissible triple with q > 2 and 7 > —^^-5-^ we have 

(2.4) lkllL?x(R+xo) < WJWhi + Mnr' + W^Wl-.hZ-'- 

The key ingredients in the proof are Proposition 12.11 and the following variant of 
(fT^ . which holds for all 7 e R, provided dM]) holds, and P G Cf'(E") equals 1 on a 
neighborhood of M"\i7: 

(2.5) WMIl^hi + WPdMl^a^Hi-^ + WMIl^.hi + \\P9tu\\^2^y^-i 

^ \\.f\\Hl + \\9\\hI-' + \\P\\l-;HI-'- 

The Lj estimates in p.5|) on u follow from ()1.2|) and elliptic regularity arguments for 
7 e Z, and by interpolation for the remaining 7 e M. The LJ" estimates then follow from 
energy estimates, duality, and elliptic regularity. 

To prove jMl), let us fix /3 e Co°°(IR") satisfying f3{x) = 1, < 3i? and write 

u = V + w, where v = (3u, w = {1 — j3)u. 

Then w solves the free wave equation 

\w\t=a = dtw\t=Q = 0. 

An application of Proposition 12.11 shows that HwUl^x is dominated by ||pu||^2^T if p 
equals one on the support of (3. Therefore, by (|2.5p . is dominated by the right 

side of (EH). 

As a result, we are left with showing that if u = j3u then 

(2-6) M\Llx(Vi+xn) ^ WIWhi + II.9|Ihj-i + II^IIl^^j-I: 

assuming, as above, that (|2.3p holds. To do this, fix (/j e C|^((— 1,1)) satisfying 
'Y^'Y=-oo — j) = 1. For a given j G N, let vj = (p{t — j)v. Then Vj solves 

' {df - Ag)v, = -^{t - j)[A, /3]u + [df,^{t - j)]Pu + vit - j)F 
Bvj{t,x)^0, xedVl 
[z;,(0, .)=ft«,(0, O^O, 



while Vq = V — JZ'jLi '"j solves 

r - Ag)«o = -^[A, I3]u + [dl ^](3u + 

[wo|t=o = /, dtVo\t=o = g, 

if = 1 — J2jLi fi^ ^ j) if ^ > and otherwise. If we then let Gj = [d^ — ^g)vj be 
the forcing term for Uj, j = 0, 1, 2, . . . , then, by ()2.5p . we have that 


killL?X(R+xn) ^ 

By the local Strichartz estimates ()l.lip and Duhamel, we get for j — 1,2,... 


/ \\Gj{s, ■)\\H-'-^ds <\\Gj\\^2Hi-^, 
Jo " " 

using Schwarz's inequality and the support properties of the Gj in the last step. Similarly, 

I|wo|1l«X(R+xO) ^ WIWhI + ll.9ll_H-J-i + I|Go!Il2^7-i • 

Since g > 2, we have 


and so we get 

as desired, which finishes the proof of Proposition 12.31 □ 

End of Proof of Theorem ll.3t Recall that we are assuming that {d'i - Ag)u = 0. By 
Proposition l2.3l we may also assume that the initial data for u vanishes when \x\ < 3R/2. 
We then fix /3 e C^(K") satisfying = 1, \x\ < R and P{x) = 0, \x\ > 3i?/2 and 

u — uo — V = {1 — [i)uo + (/3uo — v) , 

where wq solves the Cauchy problem for the Minkowski space wave equation with initial 
data defined to be (/, 5) if x S and otherwise. By the free estimate (|1.9p and (jl.Sp . 
we can restrict our attention to ii — (3uq — v. But 

(a?- Ag)u = -[A,/3]uo = G 

is supported in i? < |x| < 2i?, and satisfies 


(2.7) / \\G{t,-)r^-,^,dt<\\frH-, +\\g\\% 

by Lemma [2.21 and the fact that G vanishes on a neighborhood of Oil. Note also that u 
has vanishing initial data. Therefore, since Proposition 12.31 tells us that ||w||^9jf(-K_|_xR") 
is dominated by the left side of (|2.7p . the proof is complete. □ 

For future reference, we note that the preceeding steps establish the following gener- 

-3 n~-1 

1 ' 2 

alization of (|2.5p . assuming that 7 e [— -^V^, '^V"]' ^^'^ ^{^) — foi' l^^l > 



(2.8) ||u||i=o^T + \\dtu\\^^jj-y-i + WPuWlihi + Il/39tu||i2^j-i 

<\\f\\Hl + \\9\\H'^^ + \\F\\L.Hl-^. 

In particular, / and g have no support restrictions. To see that (|2.8p holds, first consider 
bounding the terms \\(3dlu\\j2f,i-D for j — 0,1. For these terms, it suffices by (12. 5p to 
consider F — Q and f,g = near Decomposing u = (l — P)uo + m as above, we may 
use (|2.5p and (|2.7p to deduce the bounds in (|2.8p for u. These bounds now yield 

11(9^ - Ag)(i - f3)u\\^^^. . + \\{df - A,)I3u\\l?h1-' ^ WfWHl + ll^llHr^ + \\F\\l^-hz-^ ■ 

The bounds on f3u now follow from (|2.5p . Finally, (1 — /3)u satisfies the Minkowski 
wave equation on R x M", with initial data in H'^ x H''^^, and driving force F G LilH^~^ 
which vanishes for |a::| > R. The contribution to u from its initial data satisfies the 
bounds as a result of homogeneous Sobolev bounds for the Minkowski wave group. The 
contribution from F is bounded using Lemma 12.21 and duality. 

Let us conclude this section by showing that when n — 2 the restriction in Proposi- 
tion 12.11 that 7 > 1/2 is necessary in the case where X = L''(M^). In this case, by the 
standard mixed-norm Strichartz estimates (see e.g. [TB]), the hypotheses of the Proposi- 
tion are satisfied when < 7 < 3/4, l/q + 2/r — 1 — 7 and 2/q+l/r < 1/2. 

Since the hypotheses are satisfied, if the Proposition were valid for a given 7 and 
X = L'^(R2) as above, then the L^L^(M+ x R^) norm of 

WFit,.)^f [ .-<'^^^^^-^FMd^ds 

would have to be bounded by the L^H^^^ norm of F if F{t,x) = when |a:| > 1. We 
shall take to be a product hT{s)(3{x) where /3 e C°°(R^) vanishes for \x\ > 1 but 
satisfies /3(0) — 1, while hr is an odd function supported in [— T, T]. For this choice of F 
we have 

wF{t,x) = -if e^'<cosm)hTm)md^m, \it>T. 

Fix a nonzero function p G C°°(R) supported in (1/2, 1). If we take /it to be the odd 
function which equals T^^/'^ p{s/T) for positive s, then since hx has a non-zero norm 
which is independent of T, if Proposition [2Tl] were valid for an LfL'^ space as above, then 
it would follow that 

wF{t, x) = / e*-« cosm)h,{m)m d^m 

= -*r-V2 f e^^<cos{^\^\)h,m${aT)dM 


would belong to LjLl,{[T, 00) x R^) with a bound independent of T. An easy calculation 
shows that this norm equals 







Since our assumption that /3(0) — 1 implies that the last factor on the right tends to 
a positive constant, we conclude that if the conclusion of Proposition 12.11 were valid for 
X = L'"(R^), then we would need that 

->- + - = l-7. 
2 q r 

This means that when n = 2, the assumption that 7 > 1/2 in Proposition l2.1l is necessary. 

3. The Strauss conjecture for nontrapping obstacles when n = 3,4. 

Let us start the proof of Theorem 1 1.1 1 by going over the Minkowski space results that 
will be used. These will form the assumption (|1.9p of Theorem 11.31 

Lemma 3.1. Let u solve the Minkowski wave equation 

{dt - A)u = F, (t, x) e R X M" 

u(0, •) = /, dMO,-)^g. 
Then, for 2 <p < 00, and 7 satisfying 


we have the following estimate 

11 n 1 1 n 

<7<7r , and - < 1 — 7 < — , 

2 p ' 2 p 2 '2 


\x\ 2 p 'm 


l/llijT(K") + ll.9llH-f-i(R") 


\X\ 2 

+ 1-7 p 


Here, and in what follows, we are using the mixed-norm notation with respect to the 
volume element 

for finite exponents and 

\h{ruj)\P d(T{uj)^ r"-^drj 

/illi-LS sup( / \h{ruj)\P d<T{uj) 


We first note that, by the trace lemma for the unit sphere and scaling, we have 


sup r 2 


|w(ra;)p da{uj) 



1 n 
' 2 2 ' 

where dcr denotes the unit measure on S*" . Consequently, 

sup r 2 " 


(e**l^l(^)(M|^da(c.)J <Mh^^m^) 



which is equivalent to 



n — 1 


< a < 0. 



Note that by applying (j3.3p to the Fourier transform of w, we see that it is equivalent 
to the uniform bounds 

1/2 1 n 

<A-^+11|xr«|U.(M.), A>0, 2<^<2' 

|{)(Acj)P da{uj 


which by duality is equivalent to 



A' "^\\h\\L2js^-i), 

for A > and fixed 1/2 < s < n/2. Using this estimate we can obtain 



V'h'^iR+xm'-) <\\\D\' ^ip\\L2{R^), -<s<-, 

2 2 

for, by after Plancherel's theorem with respect to the i-variable, we find that the square 
of the left side of (|3.6p equals 


dx dp 


p'^-'^\0{puj)fp'^-da{uj)dp^\\ \Dr hWhim^), 

using p.5p in the first step. 

If we interpolate between (|3.4p and (|3.6p we conclude that, for 2 < q < oo, 


^ II II 11 "1 

2 g 

2 g 

This estimate in turn implies that if v solves the Cauchy problem {d^ — A)v = in 
Ml X M" then 


\x\ 2 1 


<ll«(o, •)llij.(E.) + ll9t«(o, Ollif.-M 

The estimate dual to p.3p is 

(3.9) < |||xrt+i->||^,^. 

By the Duhamel formula and (|3.8p - p.9p . we then have 

11 n 1 


2 q ' 2 q 


a; 2 p 'u 


l"(0, - ^H-, 



|a;|-t+i-T(a2_ A)w 

provided that 7 and 1— 7 satisfy the condition in IjS.Sp for q equal to p and 00, respectively, 
i.e., dSI]). □ 

A calculation shows that if 

n 2 



2 p~r 

Pc<P< (n + 3)/(n- 1) 



then (|3.ip holds: p > Pc is needed for the first part, and p < {n + 3)/(n — 1) for the 
second. Additionally, as far as the powers of go in (|3.10p . we have 

/n n + 1 \ / (n + 1) — (n — l)p \ n ^ n 2 

(3.12) p 7 =P ^ \, ^ --77 + 1-7, if7- 

2 p '/^V p(p-l) ^ 2 " '2p-l 

As a result, by the arguments to follow, p.2p is strong enough to show that for the 
non-obstacle, Minkowski space case, i.e. = M", Ag = A, if 2 < n < 4 then for 
Pc < p < (n + 3)/(n — 1), the equation (|1.3p has a global solution for small data as 
described in Theorem 11.31 

To prove the obstacle version of this result for n — 3 and 4 we shall use a slightly 
weaker inequality for which it will be easy to show that we have the corresponding local 
Strichartz estimates (|l.lip for (il, Ag). To this end, if R is chosen so that dfl is contained 
in \x\ < R and A = Ag for \x\ > R then we define X = Xj^q{R^) to be the space with 
norm defined by 

(3.13) = ||^||L=^(|a;|<2fl) + II ^ H L'f.Ll{\x\>2R) , if ^{^^ j;) 

We then prove the following obstacle variant of (|3.2p . 
Lemma 3.2. For solutions of (|l.ip if n > 3 and p > 2: 

L-L-LUR+y.{\x\>2R}) ^ H "H if («+ x {.GO: |.| <2K}) 



+ ll.9llHr 

-t + l-7^ 

Lji^LJ (R+x{|x|>2_R}) 

+ ll^ll 4 

LlL^^^'' (R+ X {x£n:\x\<2R}) 

provided that (|3.ip holds. 

By (|3.8p and Lemma [^2] we have that the assumption p.9p of Theorem 1 1.31 is valid if 
1/2 - l/q < 7 < ?i/2 - 1/g and 2 < ^ < oo, i.e. 

(3.15) ||w||L?Jf^,(K+xR'.) < lk(0, •)llij^(R-) + l|9tl'(0, ■)IIh^-1(R-)> 

if (9^2 - A)t; = in M+ x K", 

under the additional assumption that 7 < (n — l)/2 (which is the case for (|3.1ip V Indeed 
the contribution of the second part of the norm in p.l3p is controlled by p.Sp . To handle 
the contribution of the first term in the right side of (|3.13p we note that if /3 G C5*'(M") 
equals one when < 3i? then Sobolev estimates yield 

Mt, ■)\\l'-,{\x\<2R) < \\Pi-)v{t, •)IIht(r..)- 

Thus, ||'u||l9ls7(h^x||2;|<_r}) is controlled by the right side of (|3.15p for g = 2, by Lemma 

2.21 Since this is also the case for g = cx) by energy estimates, by interpolation we 
conclude that we can control the contribution of the first term in the right side of (|3.13p 
to (|3.15p . which finishes the proof of (|3.15p . 

Since the dual norm of || is || |a;|~"/i||p', by Corollary 1.4, we would get (|3.14p 

from (|3.15p and Hypothesis B if we could show that for q > 2 

II"IIl?x^,,([o,i]xo) < ll/lliji + II.9|Ih^-i, 



whenever u solves with F = Q, and, as above, 1/2 — 1/q < 7 < n/2 — 1/q. By 

the finite propagation speed of the wave equation, it is clear that the contribution of the 
second term in the right side of (|3.13p will enjoy this estimate. As before, the first term 
satisfies it because of Sobolev estimates. This completes the proof of (|3.14p . □ 

Let us also observe a related estimate 

(3.16) ||m|Il~HJ(R+xO) + ll^t"llL~ij2"i(R+xn) + ll"llLt°-L;^(R+xn) + L?ff J (R+ xO) 



LjLiLj (R^x{|x|>2i?,}) 
+ ll^ll 4 

assuming that (|3.ip holds. Indeed, this is a direct consequence of (|2.8p and the Duhamel 
formula, together with the inclusion Hg{fl) ^ L^'>{il.), and the following consequence of 
(|3.9p . and the dual estimate to Sobolev embedding H^'^{fl) ^ L'^'^-'< (fl), 

(3.17) < II kr-+^-5lU^.i(N>2«) + bll,4-.(.,,,.|<,«) ■ 

To prove Theorem 11.11 we shall require a variation of the last two estimates involving 
the vector fields 

{T} = {duZ} 

where, as before, {Z} are the vector fields {di, Xjdk — Xkdj : 1 < i < n,l < j < k < n}. 
Note that all the {F} commute with Og — df ~ Ag when |a;| > R because dfl C {x : 
\x\ < R} and A = Ag for |a;| > R. 

The main estimate we require is the following. 

Lemma 3.3. With p and 7 as in Lemma \3.2l u solving p.ip with n>3, and {f,g,F) 
satisfying Hg x Hg x Hg boundary conditions, then 



x{\x\>2R}) + l|r""llLfL^^(R+x{2:ea:|x|<2fl}) 


^ E(ll^"/llHS + ll^".9bs- 


+ E (II l-l-^^-r"FL.,,,.(«,.,.,,,,„ + ll^"^ll...;^-.(R,x{.eo:„<2«i). 


The boundary conditions on {f,g,F) imply that d^u is locally in (il), j — 

0, 1, 2, which will be implicitly used in elliptic regularity arguments. We will also use the 
fact that the Cauchy data for T^u is bounded in H]^ x H]j~^ by the right hand side of 
(I3.18P for |a| < 2. This is clear if F" is replaced by Z". On the other hand, the Cauchy 
data for dtu is {g, Ag/ + F{0, ■ )). We may control 

(liz"5b. +ll^"Ag/||^-,_.) < J2 (ll^"/bs + ll^"5brO- 

|a|<l |a|<2 



Recall that 7 e (0, i), so that H'^{fl) = H'^{n). To control the term F(0, • ), we recall 
that r = {dt, Z}, and use the bound 

(3.19) W^'^Ph-Hl-'i^.-^) ^ E \\^"P\\LiHr\M,.n) 

\a\<l \a\<2 

which by p.l7p is seen to be dominated by the right hand side of p.l8p . Similar consid- 
erations apply to the Cauchy data for dfu. 

Let us now give the argument for (|3.18p . We first fix /3o G satisfying Pq = 1 for 
\x\ < R and supp/3o C {\x\ < 2R}. Then the first step in the proof of p.lSp will be to 
show that 


J2 kl^"'^"^(l - ^0)r"u||LfL?L2(B+x{Ix|>2fi}) + 11(1 - M^^uIIl'^l:,-' iR+x{xen:\x\<2 


+ E (II l-l---r"^ll..^.^.^(«.x(„>2.» + lir"^ll,,,:'.-.(,^.,,,,.„<,,,, 

|q I <2 

Since the T commute with \3g when > /?, we have 

□,,((1 - /5o)r"u) = (1 - /?o)r"F - [/3o, Ag]r"u . 

We can therefore write (1 — /3o)r"u a.s v + w where DgV = (1 — /3o)r"F and v has initial 
data ((1 - /3o)r"u(0, ■ ),dt{l - /3o)r"w(0, •)), while DgW = -[/3o,Ag]r"u and w has 
vanishing initial data. If we do this, it follows by (|3.14|) that if for \a\ < 2 we replace the 
term involving (1 — Po)T°'u by v in the left side of ()3.20p . then the resulting expression 
is dominated by the right side of (|3.20p . If we use (|2.4p . we find that if we replace 
(1 — Po)T°'u by w then the resulting expression is dominated by 

(3.21) J2 II [/3o,Ag]r"^||^.^.-i <^||/3ia^u||^.^.+.-, 

|a|<2 J<2 

assuming that /3i equals one on supp(/3o) and is supported in |a;| < 2R. As a result, we 
would be done with the proof of p.20p if we could show that the right hand side of p.2ip 
is dominated by the right side of (|3.20p . By (|3.16p we control \\(3idfu\\i^2jji by the right 
hand side of (|3.20p . On the other hand, 

WPidtuWl^ -,+1 < \\l3id^u\\L2Hi ||/3iu||^2jjT+2 

SO it suffices to dominate ||/3iu||^2^7+2 . Since AgU ~ &fu — F, then if P2 equals one on 
supp(/3i) and is supported in the set where |x| < 2R, we may use elliptic regularity and 
the equation to bound 

II/3i^IIl?HJ+^ ~ ll/32AgU||i2^7 + \\f32u\\L2Hl 

< \\p2d^u\\L2Hl + \\P2U\\L^2HI + \\P2F\\l2hZ- 



The first two terms are dominated as above using (|3.16p . On the other hand, Sobolev 
embedding and duality yields 

(3.22) < E ll^"^IL?ri-.(«,x{..0:|.|<2K}) 


II t,x llLjL°i-T(ffi+x{2;6a:|x|<2i?})' 


To finish the proof of p.lSp . we need to show that the analog of (|3.20p is valid when 
(1 — Po) is replaced by Pq. Since the coefficients of F are bounded on supp(/3o), if (3i 
equals one on supp(/3o) and is supported in < 2R, then by Sobolev embedding 

\a\<2 j<2 


The terms in LfH'''^'^"^ are dominated as above. To control the L'^W^'^^'^ terms, and 
conclude the proof of (|3.18p . we establish the following estimate: 

(3.23) E W^^^Wlthi + WdtT-uW^^^-,^-^^ < ^ + ll^"5lUj-(o)) 

\a\<2 \a\<2 

y ( II brt + (l-r)r"F|Uliii2 (M^y.(lx\>2m) + \\^"F\\ ^' ) 
^ V" ' llL,L,^L^(R+x{\x\>2H}) II L} L^-' (W^ xf xen:\x\<2B.}) J 


The inequality where r"u is replaced by (1 — /3o)T°'u in (|3.23p follows by energy 
estimates on R", since the right hand side dominates ||(1 — /9o)r"F||^i^^_i , together 
with p.Sp using the bound (|3.2ip to handle the commutator term. If T^u is replaced 
on the left hand side by /JoT^u, the result is dominated by J2j<3 ll/^ic^t ""H^ooji^s+t-j . For 
the case j = 0, 1, we write □g(/3i-u) — PiF — [Ag,/3i]u, and use (|2.5p with the Duhamel 
formula to bound 

\\M\l-^hi+^ + ||/3i9tw||^o.^7+i 

< II/3i/IIhi+2 + ||/3i.9||^^+i + 11/32^11^2^^+2 + \\PiF\\l^h-'+^- 

B B t B t B 

The term on the right involving u is controlled previously; on the other hand, since F 
satisfies the H'^^ boundary conditions, then 

|a|<2 *^ 

To handle the terms for j = 2, 3 we use the equation to bound 

E Widluw^^^.,-,-, < E (||/3ia^Agu||^^^^.-, + ||/3ia,^F||^^^^-,-, 

j=2.3 j=0.1 

The terms involving AgU are dominated by ||/329j w||^oo^7+2-j with j = 0, 1. The terms 
involving F are controlled for j = 1 by p.l9p . and for j = by observing that (|3.22p 
holds with replaced by L^. This completes the proof of (|XT5)l and ([X^ . □ 



We shall now use these estimates to prove Theorem 1 1.1 1 

Proof of Theorem We assume Cauchy data (/, g) satsifying the smallness condi- 
tion (|1.7p . and let uq solve the Cauchy problem (jl.ip with F ~ 0. We iteratively define 
Wfe, for fc > 1, by solving 

f {df - Ag)uk(t,x) = Fp{uk-i{t,x)) , {t,x) eR+xn 
uiO, • ) = / 
dtu{0, ■)= g 

^{Bu){t,x) = 0, on R+ X dil. 
Our aim is to show that if the constant e > in (|1.7p is small enough, then so is 


B 1.^+ 


+ II \x\ ^ " '^r"'"'=ILfL^L2(R_^x{i^[>2i?}) + \\^°'M\L^Ll-'(m+x{xen:\x\<2B})) 

for every fc = 0, 1, 2, . . . 

For fc = 0, it follows by p.lSp and (I3.23P that Mq < Cqs, with Co a fixed constant. 
More generally, ([3?T8l) and ((3?23l) yield that 

1-7-pa TP f„,. . ^ II 

l.Ll{R+ X {\x\>2B.}) 

(3.24) Mk<Coe + CoJ2 ( II l^r^^'"^r"^;,K_i)||^j^, 


+ ||r"i^pK_i)|| ^ 4 ) 

Note that our assumption (II. 4p on the nonlinear term i^p implies that for small v 

a|<2 |a|<2 |a|<l 

Furthermore, since Uk will be locally of regularity Hg'^^ C L°° and Fp vanishes at 0, it 
follows that Fp{uk) satisfies the B boundary conditions if Uk does. 

Since the collection F contains vectors spanning the tangent space to 5*"^^, by Sobolev 
embedding for n = 3, 4 we have 

II"(^-)IIls=+ E lirM^-)IUi< E lir%(r-)IU- 

q|<1 |q|<2 

Consequently, for fixed t,r > Q 

E ||F"FpK_i(t,r.))IUi< E l|r"^fc-i(^,r-)ll!. ■ 

a|<2 |a|<2 

By p.l2p . the first summand in the right side of (|3.24p is dominated by CiM^_^ . 

We next observe that, since > 2 and n < 4, it follows by Sobolev embedding on 
{nn \x\ < 2R} that 

\\y\\L°--{x£n:\x\<2R) + E l|r"^lli*(^eJ2:|x|<2_R) ^ E 1 1 ^""lU^^ (a:ef2: |xl <2_R) • 
a|<l |a|<2 



Since s'i_^ < 2, it holds for each fixed t that 

\a\<2 |a|<2 

The second summand in the right side of (|3.24p is thus also dominated by CiM^_^ , and 
we conclude that Mk < CqE + 2Co CiM^_^. For e sufficiently small, then 

(3.25) Mk<2Coe, fc = 1,2,3,... 

To finish the proof of Theorem 11.11 we need to show that Uk converges to a solution of 
the equation (|1.3|) . For this it suffices to show that 

+ \\uk - '"fe-l|lLPL=T-(R+x{2;ef2: \x\<2R.}) 

tends geometrically to zero as fc ^ oo. Since \Fp{v) — Fp{w)\ < |ti — w|( + jiylP"-*^ ) 

when V and w are small, the proof of (|3.25p can be adapted to show that, for small e > 0, 
there is a uniform constant C so that 


which, by ()3.25|) . implies that Ak < ^A^-i for small e. Since Ai is finite, the claim 
follows, which finishes the proof of Theorem 11.11 □ 


[1] M. Ben-Artzi: Regularity and smoothing for some equations of evolution, in "Nonlincr Partial 
Differential Equations and Applications" (If. Brczis and J. L. Lions, eds.), London, 1994, pp. 

[2] M. Ben-Artzi and S. Klainerman: Decay and regularity for the Schrddinger equation, J. Anal. 
Math. 58 (1992), 25-37. 

[3] M. Blair, H. Smith and C. D. Sogge: Strichartz estimates for the wave equation on manifolds with 

boundary, arXiv:0805.4733 
[4] N. Burq: Global Strichartz estimates for nontrapping geometries: About an article by H. Smith 

and C. Sogge, Comm. Partial Differential Equations 28 (2003), 1675-1683. 
[5] N. Burq, G. Lebeau and F. Planchon: Global existence for energy critical waves in 3-D domains, 

J. Amer. Math. Soc. 21 (2008), 831-845. 
[6] N. Burq and F. Planchon: Global existence for energy critical waves in 3-d domains : Neumann 

boundary conditions, arXiv:0711.0275 
[7] M. Christ and A. Kiselev: Maximal functions associated to filtrations, J. Funct. Anal. 179 (2001), 


[8] Y. Du, J. Metcalfe, C. D. Sogge and Y. Zhou: Concerning the Strauss conjecture and almost 
global existence for nonlinear Dirichlet-wave equations in A-dimensions, Comm. Partial Differential 
Equations 33 (2008), 1487-1506. 

[9] Y. Du and Y. Zhou: The life span for nonlinear wave equation outside of star-shaped obstacle in 
three space dimensions. Comm. Partial Differential Equations 33 (2008), 1455-1486. 
[10] D. Fang and C. Wang: Weighted Strichartz Estimates with Angular Regularity and their Applica- 
tions, arXiv:0802.0058 

[11] V. Georgiev, H. Lindblad, and C. D. Sogge: Weighted Strichartz estimates and global existence 

for semilinear wave equations, Amer. J. Math. 119 (1997), 1291-1319. 
[12] R. T. Glassey: Existence in the large for Ou = F{u) in two dimensions. Math. Z. 178 (1981), 



[13] K. Hidano: Morawetz-Strichartz estimates for spherically symmetric solutions to wave equations 
and applications to semi-linear Cauchy problems, Diflferential Integral Equations 20 (2007), 735- 

[14] K. Hidano: Small solutions to semi-linear wave equations with radial data of critical regularity, 

Rev. Mat. Iberoamericana, to appear. 
[15] T. Hoshiro: On weighted estimates of solutions to wave equations, J. Anal. Math. 72 (1997), 


[16] F. John: Blow-up of solutions of nonlinear wave equations in three space dimensions, Manuscripta 

Math. 28 (1979), 235-265 
[17] M. Keel, H. F. Smith, and C. D. Sogge: Almost global existence for some semilinear wave equations, 

J. Anal. Math. 87 (2002), 265-279. 
[18] M. Keel and T. Tao, Endpoint StrichaHz estimates, Amer. J. Math. 120 (1998), 955-980. 
[19] H. A. Levine, Instability and nonexistence of global solutions to nonlinear wave equations of the 

form Putt = -Au-\-T{u), Trans. Amer. Math. Soc. 192 (1974), 1-21. 
[20] T. T. Li and Y. Zhou: A note on the life-span of classical solutions to nonlinear wave equations 

in four space dimensions, Indiana Univ. Math. J. 44 (1995), 1207-1248. 
[21] H. Lindblad and C. D. Soggc: On existence and scattering with minimal regularity for semilinear 

wave equations, J. Funct. Anal. 130 (1995), 357-426. 
[22] H. Lindblad and C. D. Sogge: Long-time existence for small amplitude semilinear wave equations, 

Amer. J. Math. 118 (1996), 1047-1135. 
[23] R. B. Melrose: Singularities and energy decay in acoustical scattering, Duke Math. J. 46 (1979), 


[24] R. B. Melrose and J. Sjostrand: Singularities of boundary value problems. I, Comm. Pure Appl. 

Math. 31 (1978), 593-617. 
[25] J. Metcalfe: Global Strichartz estimates for solutions to the wave equation exterior to a convex 

obstacle, Trans. Amer. Math. Soc. 356, (2004), 4839-4855. 
[26] C. S. Morawctz: Decay for solutions of the exterior problem for the wave equation. Comm. Pure 

and Appl. Math. 28 (1975), 229-264. 
[27] C. S. Morawetz, J. Ralston and W. Strauss: Decay of solutions of the wave equation outside 

nontrapping obstacles. Comm. Pure Appl. Math. 30 (1977), 87-133. 
[28] P. D. Lax and R. S. Philips: Scattering Theory (Revised Edition), Academic Press Inc., 1989. 
[29] J. Ralston: Note on the decay of acoustic waves, Duke Math. J. 46 (1979), 799-804. 
[30] Y. Shibata and Y. Tsutsumi: Global existence theorem for nonlinear wave equation in exterior 

domain, Lecture Notes in Num. Appl. Anal., Vol. 6 (1983), 155-196. 
[31] T. C. Sideris: Nonexistence of global solutions to semilinear wave equations in high dimensions, 

J. Differential Equations 52 (1984), 378-406. 
[32] H. F. Smith and C. D. Sogge: On the critical semilinear wave equation outside convex obstacles, 

J. Amer. Math. Soc. 8 (1995), 879-916. 
[33] H. F. Smith and C. D. Sogge: Global Strichartz estimates for nontrapping perturbations of the 

Laplacian, Comm. Partial Differential Equations 25, (2000), 2171-2183. 
[34] H. F. Smith and C. D. Sogge: On the norm of spectral clusters for compact manifolds with 

boundary, Acta Math. 198, (2007), 107-153. 
[35] C. D. Sogge: Lectures on nonlinear wave equations. International Press, Boston, MA 1995. 
[36] C. D. Sogge: Lectures on nonlinear wave equations, 2nd edition. International Press, Boston, MA, 


[37] D. Tataru: Strichartz estimates in the hyperbolic space and global existence for the semilinear 

wave equation. Trans. Amer. Math. Soc. 353 (2001), 795-807. 
[38] M. Taylor: Grazing rays and reflection of singularities of solutions to wave equations, Comm. 

Pure Appl. Math. 29 (1976), 1-38. 
[39] B. R. Vainberg: The short-wave asymptotic behavior of the solutions of stationary problems, and 

the asymptotic behavior as t oo of the solutions of nonstationary problems, Russian Math. 

Surveys 30 (1975), 1-58. 

[40] Y. Zhou: Cauchy problem for semilinear wave equations with small data in four space dimensions, 
J. Partial Differential Equations 8 (1995), 135-144. 

abstract strichartz estimates 
Department of Mathematics, Mie University 

Department of Mathematics, University of North Carolina, Chapel Hill 
Department of Mathematics, University of Washington, Seattle 
Department of Mathematics, Johns Hopkins University 
School of Mathematical Science, Fudan University