In this paper, we give an alternative proof to the global existence result, which is originally owing to the pioneering work of Klainerman and Christodoulou, for the Cauchy problem of quasilinear wave equations with null condition in three space dimensions. The proof can display the following three features simultaneously: the Lorentz boost operator is not employed in the generalized energy estimates; the generalized energy of the solution will be always small, which was first observed by Alinhac; and the initial data are not assumed to have a compact support. Copyright © 2016 John Wiley & Sons, Ltd.Keywords: quasilinear wave equations; three space dimensions; null condition; global existence
IntroductionWe consider the following quasilinear wave equation in 3-D:where D @ 2 t is the wave operator,3 /, and the coefficients g˛ˇare smooth real functions vanishing at the origin, that is,We always assume that the following symmetric conditions hold:which is an essential condition for energy estimate in our proof. The set g D .g˛ˇ / is said to satisfy the null condition if for any givenConsider the Cauchy problem for (1) with the initial dataFor the Cauchy problems (1) and (5), under the null condition (4), the global existence of classical solutions with small initial data was first proved by Christodoulou [1] and Klainerman [2] independently. Christodoulou's proof employed the conformal transformation method, while Klainerman used the vector fields method. Klainerman's proof relies on a pair of coupled differential inequalities for the lower-order weighted L 1 norm of the solution and the higher-order energy. It is worth noting that the higher-order energy may be not always small and may grow with the time.In [3] (Section 9.1.3), Alinhac got the global existence of classical solutions to the Cauchy problems (1) and (5) via a single generalized energy estimate, and the generalized energy can be shown to be always small. The key point of his proof is the ghost weight energy We note that in both Alinhac's proof and Klainerman's proof, they all use the full Lorentz invariance of the wave operator; therefore, the general space-time derivatives, spatial rotation, scaling and Lorentz boost operator are all used. Moreover, in Alinhac [3], it is essentially assumed that the initial data have a compact support. However, the Lorentz boost operator is not suitable for some wave systems, which are not Lorentz invariant, such as nonrealistic system with multiple wave speeds [7][8][9], nonlinear wave equations on non-flat space-time [10-13], wave-type equation with nonlocal term [14,15] and exterior problem [16]. Furthermore, the compact support assumption of the initial data is also restrictive and is not suitable for some nonlocal problems such as the incompressible elastodynamics(e.g. Sideris and Thomases [14,15], Lei and Wang [17], Lei [18] and Wang [19]). So how to remove the Lorentz boost operator and the compact support assumption of the initial data in Alinhac's proof, and in the meantime, keep the smallness of the general...