Abstract
The classical model of an isolated selfgravitating gaseous star is given by the Euler–Poisson system with a polytropic pressure law \(P(\rho )=\rho ^\gamma \), \(\gamma >1\). For any \(1<\gamma <\frac{4}{3}\), we construct an infinite-dimensional family of collapsing solutions to the Euler–Poisson system whose density is in general space inhomogeneous and undergoes gravitational blowup along a prescribed space-time surface, with continuous mass absorption at the origin. The leading order singular behavior is described by an explicit collapsing solution of the pressureless Euler–Poisson system.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
The basic model of a Newtonian star is given by the 3-dimensional compressible Euler–Poisson system [1, 11, 62]
Here \(\rho ,\mathbf{u}, P(\rho ), \Phi \) denote the gas density, the gas velocity vector, the gas pressure, and the gravitational potential respectively. To close the system we impose the so-called polytropic equation of state:
The power \(\gamma \) is called the adiabatic exponent.
Here a star is modelled as a compactly supported compressible gas surrounded by vacuum, which interacts with a self-induced gravitational field. To describe the motion of the boundary of the star we must consider the corresponding free-boundary formulation of (). In this case, a further unknown in the problem is the support of \(\rho (t,\cdot )\) denoted by \(\Omega (t)\). We prescribe the natural boundary conditions
and the initial conditions
Here \(\mathcal {V}(\partial \Omega (t))\) is the normal velocity of the moving boundary \(\partial \Omega (t)\) and condition (1.3b) simply states that the movement of the boundary in normal direction is determined by the normal component of the velocity vector field. We refer to the system ()–() as the \(\hbox {EP}_\gamma \)-system. We point the reader to the classical text [11] where the existence of static solutions of \(\hbox {EP}_\gamma \) is studied under the natural boundary condition (1.3a).
We next impose the physical vacuum condition on the initial data:
Condition (1.5) implies that the normal derivative of the squared speed of sound \(c_s^2(\rho )=\frac{\mathrm{d}P}{\mathrm{d}\rho }(\rho )\) is discontinuous at the vacuum boundary. This condition is famously satisfied by the well-known class of steady states of the \(\hbox {EP}_\gamma \)-system known as the Lane–Emden stars. At the same time, condition (1.5) is the key assumption that guarantees the well-posedness of the Euler–Poisson system with vacuum regions.
For any \(\bar{\varepsilon }>0\) consider the mass preserving rescaling applied to the \(\hbox {EP}_\gamma \)-system:
where
It is easy to see that the above rescaling is mass-critical, that is \( M[\rho ] = M[\tilde{\rho }]. \) A simple calculation reveals that if \((\rho , \mathbf{u}, \Phi )\) solve the \(\hbox {EP}_\gamma \)-system, then the rescaled quantities \((\tilde{\rho }, \tilde{\mathbf{u}}, \tilde{\Phi })\) solve
where
Observe that for \(\bar{\varepsilon } \ll 1\) the factor \(\varepsilon \) in front of the pressure in (1.7b) is small precisely in the supercritical range \(1<\gamma <\frac{4}{3}\). The system obtained by dropping the \(\varepsilon \)-term in (1.7b) is known as the pressureless- or dust-Euler system. This gives a vague heuristics that, if one for a moment thinks of \(\varepsilon \) as a sufficiently small length scale of density concentration, the effects of the pressure term may become negligible and the leading order singular behavior will be driven by the pressure-less dynamics. On the other hand, at this stage, this scaling heuristics is at best doubtful, as the pressure term enters the equation at the top order from the point of view of the derivative count.
Parameter \(\varepsilon \) serves the purpose of a “small” parameter in our analysis. Defining \(\tilde{\Omega }(s) = \bar{\varepsilon }^{-1}\Omega (t)= \varepsilon ^{-\frac{1}{4-3\gamma }} \Omega (t)\), a homothetic image of \(\Omega (t)\), boundary conditions () take the form
and the initial conditions read
1.1 Lagrangian Coordinates
To address the problem of collapse we express () in the Lagrangian coordinates. Firstly, if we wish to follow the collapse process in its entirety until all of the stellar mass is absorbed, it is clear that the Eulerian description becomes inadequate at and after the first collapse time. In order to describe particle trajectories after the first collapse time, it is advantageous to work in a coordinate system that avoids this issue. Secondly, the free boundary is automatically fixed in Lagrangian description and thus more amenable to rigorous analysis.
For the remainder of the paper we make the assumption of radial symmetry and assume that the reference domain \(\tilde{\Omega }\) is the unit ball \(\{y\in \mathbb {R}^3\,\big | \ |y|\le 1\}\). Let the flow map \(\eta :\tilde{\Omega }\rightarrow \tilde{\Omega }(s)\) be a solution of
Here the boundary \(\partial \tilde{\Omega }\) is mapped to the moving boundary \(\partial \tilde{\Omega }(s)\). The choice of \(\eta _0\) corresponds to the initial particle labelling and represents a gauge freedom in the problem. Equation (1.10) automatically incorporates the dynamic boundary condition (1.8b) when we pull-back the problem on \(\tilde{\Omega }\)
Since the flow is spherically symmetric, \(\eta \) is parallel to the vectorfield y. We introduce the ansatz
and denote \(\chi (0,r)\) by \(\chi _0(r)\). The Jacobian determinant of \(D\eta \) expressed in terms of \(\chi \) takes the form
Since \(\partial _s \mathscr {J}= \mathscr {J}(\text {div}\tilde{\mathbf{u}})\circ \eta \), as a consequence of the continuity equation the Eulerian density \(\tilde{\rho }\) evaluated along the particle world-lines satisfies
Let
The fluid enthalpy is a function \(r\mapsto w(r)\) defined through the relationship
and this is a fundamental object in our work. Instead of specifying \(\tilde{\rho }_0\) and \(\chi _0\), throughout the paper we fix the choice of the fluid enthalpy w satisfying properties (w1)–(w3) below.
-
(w1)
We assume that \(w:\tilde{\Omega }\rightarrow \mathbb {R}_+\) is a non-negative radial function such that \([0,1)\ni r\mapsto w(r)^\alpha \) is \(C^\infty \), \(w>0\) on [0, 1) and \(w(1)=0\).
Assuming further that \(\chi _0(r)\) is uniformly bounded from below and \(C^2\), from \(\tilde{\rho }(\chi _0(1))=0\) and the physical vacuum condition (1.5), we conclude \(\nabla w\cdot \tilde{\mathbf{n}}<0\) at the boundary \(\partial \tilde{\Omega }\) of the reference domain.
-
(w2)
This leads us to the second basic assumption on w:
$$\begin{aligned} \nabla w\cdot \tilde{\mathbf{n}}\Big |_{\partial \tilde{\Omega }} = w'(1)<0. \end{aligned}$$ -
(w3)
Finally we denote the mean density of the gas by
$$\begin{aligned} G(r): = \frac{1}{r^3}\int \nolimits _0^r4\pi w^{\alpha } s^2\,\mathrm{d}s, \end{aligned}$$(1.17)and let
$$\begin{aligned} g(r) :=3\sqrt{\frac{G(r)}{2}}, \ \ r\in [0,1]. \end{aligned}$$(1.18)Clearly \(g>0\). Observe that \(G(0) = \frac{4\pi }{3}w(0)^\alpha \). We shall require that \(g:[0,1]\rightarrow \mathbb {R}\) is a smooth function such that there exist positive constants \(c_1,c_2>0\) and \(n\in \mathbb {Z}_{>0}\) so that
$$\begin{aligned} c_1 r^n \le -r\partial _r \left( \log g(r)\right)&\le c_2r^n, \ \ r\in [0,1]. \end{aligned}$$(1.19)
The purpose of the next lemma is to show that there exist choices of the enthalpy w consistent with the above assumptions.
Lemma 1.1
For any \(n\in \mathbb {N}\) there exists a choice of the enthalpy w satisfying properties (w1)–(w3). In particular, the resulting map g defined by (1.18) satisfies (1.19).
Proof
Let \(w(r) =a (1-r^n)_+\) We observe that for any \(r\in (0,1]\) \(G(r)=\frac{a^\alpha }{r^3} \int \nolimits _0^r 4\pi (1-s^n)^\alpha s^2\,\mathrm{d}s = \frac{4\pi a^\alpha }{3} - \frac{a^\alpha c_{n,\alpha }}{n}r^{n} + o_{r\rightarrow 0}(r^n)\), with \(1\lesssim c_{n,\alpha }\lesssim 1\). Note that
which implies (1.19). \(\square \)
Remark 1.2
It is evident from the proof that one can easily modify the enthalpy w in the regions away from \(r=0\) so that (1.19) is still satisfied. In fact, the family of enthalpies w which satisfy the assumptions (w1)–(w3) is infinite-dimensional.
As a simple, but important corollary of (w3), specifically (1.19), we have
Corollary 1.3
Let g be given by (1.18). Then the following properties hold:
-
(i)
the map \(r\mapsto g(r)\) is monotonically decreasing on [0, 1];
-
(ii)
in the vicinity of the origin the following Taylor expansion for g holds:
$$\begin{aligned} g(r)= g(0)-\frac{c}{n} r^n +o_{r\rightarrow 0}(r^n) \end{aligned}$$(1.20)for some constant \(c>0\);
-
(iii)
for any \(k\in \mathbb {N}\) there exists a positive constant \(c_k\) such that
$$\begin{aligned} \left| (r\partial _r)^k g(r)\right| \le c_k r^n. \end{aligned}$$(1.21)
As shown in [34], the momentum Equation (1.7b) expressed in the Lagrangian variables (s, r) reduces to a nonlinear second order degenerate hyperbolic equation for \(\chi \):
where \(r\mapsto G(r)\) is given above in (1.17) and the nonlinear pressure operator P is given by
We may explicitly relate the Eulerian density, the fluid enthalpy and the Jacobian determinant; as long as \(\mathscr {J}[\chi ]>0\) by (1.14) and (1.16) we have the fundamental formula
Remark 1.4
Without being precise about the definition of the gravitational collapse for the moment, our goal is to prove that there exists a choice of initial conditions
with a particular choice of the enthalpy w so that \(\mathscr {J}[\chi ]\) becomes zero in finite time. We shall then show that there indeed exists a density \(\tilde{\rho }_0\) satisfying the physical vacuum condition
as both the profile \(w^\alpha \) and the initial labelling of the particles \(\chi _0\) are necessary to recover the Eulerian density \(\tilde{\rho }_0\), see (1.16).
Remark 1.5
In the special case when \(\chi _0=1\), \(w^\alpha \) and \(\tilde{\rho }_0\) coincide. We refrain from imposing the initial condition \(\chi _0=1\), but we shall prove a posteriori that the initial conditions that we use for the construction of the collapsing stars indeed satisfy \(\chi _0=1+O(\varepsilon )\) in a suitable norm.
Finally, from (1.17) we have \(r\partial _r G + 3G = 4\pi w^\alpha \) and therefore
Since \(\partial _r g \le 0\), \(w^{\alpha }|_{r=1}=0\) and \(\frac{9\pi w^{\alpha }}{g^{2}}>0\) for \(r\in [0,1)\) it follows that
Bounds (1.26)–(1.27) are crucial in proving sharp coercivity properties of our high-order energies later in the article.
1.2 Pressureless Collapse
The first step in our analysis is to describe the solutions of (1.22) when \(\varepsilon =0\). We are led to the ordinary differential equation (ODE)
with initial conditions
We now give a detailed description of the dust collapse from both the Lagrangian and Eulerian perspective, as this will serve as the leading order description of the collapsing stars for the \(\hbox {EP}_\gamma \) system.
Notice that for any fixed \(r\in [0,1]\) the coefficient G(r) merely serves as a parameter in the above ODE. The total energy
is clearly a conserved quantity. We are interested in the collapsing solutions, that is solutions of (1.28)–(1.29) such that there exists a \(0<T<\infty \) so that \(\lim _{s\rightarrow T^-}\chi (s,r)=0\) for some \(r\in [0,1]\). We consider the inward moving initial velocities with \(\chi _1<0\). From the conservation of (1.30) we obtain the formula
Integrating (1.31) one sees that for every r there exists a \(0<t^*(r)<\infty \) such that \(\chi (t^*(r),r)=0\). A simple calculation reveals that for any \(r\in [0,1]\) we have the universal blow-up exponent 2/3
We may further define the first blow-up time
Observe that the Eulerian description of the solution ceases to make sense at and after time \(s\ge t^*\). On the other hand, for different values of r the Lagrangian solution may make sense even after \(t^*\). In particular, when \(t^*(r)\) is a non-constant function, we can speak of a “fragmented” or “continued” collapse, wherein particles with a different Lagrangian label r collapse at different times. This is the hallmark behavior of inhomogeneous collapse (Fig. 1).
For simplicity, we shall consider a special subclass of solutions of (1.28)–(1.29) with zero energy. Up to multiplication by a constant such profiles have the form
where g is given by (1.18). It follows that \(\chi _{\mathrm{dust}}\) becomes zero along the space-time curve
The solution is only well-defined in the region
After a simple calculation we have
In particular, \(\chi _{\mathrm{dust}}\) and \(\mathscr {J}[\chi _{\mathrm{dust}}]\) vanish along \(\Gamma \) and therefore, since the Eulerian density satisfies
the value of \(\tilde{\rho }_{\mathrm{dust}}(s,0)\) diverges to infinity at the first blow-up time \(t^*:=\frac{1}{g(0)}\). In the region \(\chi _{\mathrm{dust}}>0\), the Eulerian density \(Y\mapsto \tilde{\rho }_{\mathrm{dust}}(s,Y)\) is always well-defined away from the origin \(Y=0\). Moreover for any \(r\in [0,1]\)
Since \(r\mapsto g(r)\) is monotonically decreasing, particles that start out closer to the boundary of the star take longer to vanish into the singularity.
Remaining mass. For any time \(s\in (\frac{1}{g(0)},\frac{1}{g(1)})\) the remaining star mass is given by
where we have changed variables: \(z \rightarrow Z=\chi (s,z) z\) and used \(w^\alpha (z) =\tilde{\rho }(s, \chi (s,z) z) \mathscr {J}[\chi _{\mathrm{dust}}] \) and \(4\pi \mathscr {J}[\chi _{\mathrm{dust}}] z^2\,\mathrm{d}z = 4\pi Z^2\,\mathrm{d}Z\). Since for any \(\frac{1}{g(0)}< s<\frac{1}{g(1)}\) \(\mathscr {J}[\chi _{\mathrm{dust}}](s,r)>0\) for all \(r\in (g^{-1}\circ (\frac{1}{s}),1]\), this change of variables is justified.
Finally, the support of the collapsing dust star shrinks to zero as \(s\rightarrow \frac{1}{g(1)}\). This is clear, as the free boundary in the Eulerian description is at distance \(\chi _{\mathrm{dust}}(s,1)=(1-g(1)s)^{\frac{2}{3}}\) from the origin. As \(s\rightarrow \frac{1}{g(1)}\) the star concentrates with its mass completely absorbed at the origin:
Therefore the time \(s=\frac{1}{g(1)}\) has a natural interpretation as the end-point of star collapse for the dust example considered here.
1.3 Main Theorem and Related Works
Stellar collapse is one of the most important phenomena of both Newtonian and relativistic astrophysics. Even though extensively studied in the physics literature, very little is rigorously known about the compactly supported solutions to \(\hbox {EP}_\gamma \)-system that lead to the gravitational collapse.
-
1.
When \(P(\rho )=0\) and therefore the star content is the pressureless dust, there exists an infinite-dimensional family of collapsing dust solutions, as described in Section 1.2.
-
2.
If \(\gamma =\frac{4}{3}\) in (1.2), due to the special symmetries of the problem, “homologous” self-similar collapsing solutions exist and were discovered by Goldreich and Weber [25] in 1980. Further rigorous mathematical works about such solutions are given in [22, 24, 43]. Here all the gas contracts to a point at the same time and the dynamics is described by a reduction to a finite-dimensional system of ODEs.
-
3.
When \(\gamma >\frac{4}{3}\) it is shown in [21] that the collapse by density concentration cannot occur.
We refer to the values \(1<\gamma <\frac{4}{3}\), \(\gamma =\frac{4}{3}\), and \(\gamma >\frac{4}{3}\) of the adiabatic exponent as the mass supercritical, mass-critical, and mass subcritical cases respectively. This terminology is motivated by the invariant scaling analysis of the \(\hbox {EP}_\gamma \)-system, see for example [29].
It has been an outstanding open problem to prove or disprove the existence of collapsing solutions in the supercritical range \(1<\gamma <\frac{4}{3}\).
Theorem 1.6
(Main theorem). For any \(\gamma \in (1,\frac{4}{3})\) there exist classical solutions \(\chi (s,r)\) of (1.22) defined in \(\Xi =\{(s,r) \,\big | \, 1-g(r)s>0\}\). The solution behaves qualitatively like the collapsing dust solution \(\chi _{\mathrm{dust}}\) and in particular
Further, for any \(r\in [0,1]\),
Finally, the following three properties hold:
-
1.
(Density blows up) For any \(r\in [0,1]\)
$$\begin{aligned} \lim _{s\rightarrow \frac{1}{g(r)}} \tilde{\rho }(s,\chi (s,r)r) = \lim _{s\rightarrow \frac{1}{g(r)}} w(r)^\alpha \mathscr {J}[\chi ]^{-1} = \infty . \end{aligned}$$(1.39) -
2.
(Support shrinks to a point)
$$\begin{aligned} \lim _{s\rightarrow \frac{1}{g(1)}} \chi (s,1) = 0. \end{aligned}$$(1.40) -
3.
(Mass is continuously absorbed into the singularity)
$$\begin{aligned} \lim _{s\rightarrow \frac{1}{g(1)}} M(s) = \lim _{s\rightarrow \frac{1}{g(1)}} 4\pi \int \nolimits _{g^{-1}\circ \frac{1}{s}}^1 w(z)^\alpha z^2\,\mathrm{d}z = 0. \end{aligned}$$(1.41)
Remark 1.7
One distinctive feature of our proof is that the singularity occurs along the prescribed space-like surface \(\Gamma \) (1.34) which coincides with the blow-up surface of the underlying dust solution \(\chi _{\mathrm{dust}}\).
Remark 1.8
Our result shows a finite time density blow up and a loss of total mass during the collapse. This phenomenon is very different from the shock formation where the singularity occurs in the form of density discontinuity.
Remark 1.9
We have used the vacuum free boundary framework to deal with the dynamics of compactly supported isolated star configurations in space that are physically important. Gravitational collapse, however, is not dependent on the presence of the vacuum boundary. In fact, dust solutions describe pressureless collapse for non-compact densities and our methodology would lead to analogous results for for example densities with infinite support having sufficient decay at infinity.
Theorem 1.6 identifies an infinite-dimensional family of monotonically decreasing initial densities that lead to the gravitational collapse. This is a global characterization of the dynamics, as the region \(\Xi \) corresponds to the maximal forward development of the data at \(s=0\).
The best known class of global solutions to the \(\hbox {EP}_\gamma \) system are the famous static Lane–Emden stars [1, 11, 62]. In the range \(\frac{6}{5}<\gamma <2\) one finds compactly supported radially symmetric time-independent solutions of finite mass, whose stability still remains an outstanding open problem. In the subcritical range \(\gamma >\frac{4}{3}\) the question of nonlinear stability is open despite the promising conditional nonlinear stability result proven by Rein [48] (see [41] for rotating stars). If the solution exists globally in time when \(\gamma >\frac{4}{3}\) and the energy is strictly positive, then the support of the star must grow at least linearly in t, as shown in [44]. A similar conditional result holds when \(\gamma =\frac{4}{3}\) [21]. In the supercritical range \(\frac{6}{5}\le \gamma <\frac{4}{3}\) it has been shown by Jang [33, 34] that the Lane–Emden stars are dynamically nonlinearly unstable. Besides the stationary states and the homologous collapsing stars in the mass-critical case \(\gamma =\frac{4}{3}\), the only other global solutions of \(\hbox {EP}_\gamma \) were constructed by Hadžić and Jang [29, 31].
Since the works of Sideris [52, 53] it has been well-known that solutions of the compressible Euler equation (without gravity) develop singularities even with small and smooth initial perturbations of the steady state \((\rho ,\mathbf{u})=(1,\mathbf{0})\). This type of blow up is generally attributed to the loss of regularity in the fluid unknowns which typically results in a shock. Under the assumption of irrotationality, Christodoulou [15] gave a very precise information on the dynamic process of shock formation for the relativistic Euler equation. In the context of nonrelativistic fluids, a related result was given by Christodoulou and Miao [16], while a wider range of quasilinear wave equations is treated extensively by Speck [56], Holzegel et al. [32]. Most recently, shock formation results have been obtained even in the presence of vorticity by Luk and Speck [40], for an overview we refer the reader to [57]. A very different type of singular behavior which results in a wild nonuniqueness for the weak solutions of compressible Euler flows was obtained by Chiodaroli et al. [13], inspired by the methods of convex integration, see [20] for an overview.
The above mentioned mechanisms of singularity formation are different from the singularity exhibited in Theorem 1.6, where the density and the velocity remain smooth in the vicinity of the origin and no shocks are formed before the gravitational collapse occurs.
In the absence of gravity, a finite dimensional class of special affine expanding solutions to the vacuum free boundary compressible Euler flows was constructed by Sideris [54, 55]. Their support takes on the shape of an expanding ellipsoid. Related finite-dimensional reductions of compressible flows with the affine ansatz on the Lagrangian flow map go back to the works of Ovsiannikov [46] and Dyson [23], with different variants of the equation of state. Nonlinear stability of the Sideris motions was shown by Hadžić and Jang [30] for the range of adiabatic exponents \(1<\gamma \le \frac{5}{3}\) and it was later extended to the range \(\gamma >\frac{5}{3}\) by Shkoller and Sideris [50].
In the setting of compressible non-isentropic gaseous stars (where the equation of state (1.2) is replaced by the requirement \(p=P(\rho ,T)\), T being the internal temperature) it is possible to impose an affine ansatz (separation of variables) for the Lagrangian flow map and thus reduce the infinite-dimensional PDE dynamics to a finite-dimensional system of ODEs. The resulting solutions have space-homogeneous gas densities and the system is therefore closed—the star takes on the shape of a moving ellipsoid. For an overview we refer to [3, 4]. A number of finite-dimensional reductions in the absence of vacuum regions relying on self-similarity and scaling arguments can be found in the physics literature for example [2, 5, 6, 39, 47, 49, 51, 59, 61].
Without the free boundary, in the context of finite-time break up of \(C^1\)-solutions for the gravitational Euler–Poisson system with a fixed background we refer to [12] and references therein. There are various models in the literature where the stabilizing effects of the pressure are contrasted to the attractive effects of a nonlocal interaction; we refer the reader to [7,8,9] for a review and many references for different choices of repulsive/attractive potentials.
The analogues of the collapsing dust solutions in the general relativistic context were discovered in 1934 by Tolman [60]. In their seminal work from 1939, Oppenheimer and Snyder [45] studied in detail the causal structure of a subclass of asymptotically flat Tolman solutions with space-homogeneous density distributions, thus providing basic intuition for the concept of gravitational collapse. Nevertheless, in 1984 Christodoulou [14] showed that the causal structure of solutions described in [45] is in a certain sense non-generic in the wider family of Tolman collapsing solutions, proving thereby that for densities given as small inhomogeneous perturbations of the Oppenheimer–Snyder density, one generically obtains naked singularities. This, in particular, highlights the importance of the rigorous study of the gravitational collapse of gaseous stars with more realistic equations of state, that is with nontrivial pressure. In the absence of any matter, existence of singular solutions containing black holes has been known since 1915. This is the 1-parameter family of Schwarzschild solutions, which is embedded in the larger family of Kerr solutions. The nonlinear stability of the Kerr solution has been an important open problem in the field. Substantial progress has been made over the recent years by Dafermos, Rodnianski, Holzegel, Shlapentokh-Rothman, Taylor, see [18, 19, 58] and references therein.
1.4 Foliation by the Level Sets of \(\chi _{\mathrm{dust}}\)
We would like to build a solution of (1.22) “around” the fundamental collapsing profile (1.33). To that end it is natural to consider the change of variables
and introduce the unknown
Note that \(0\le \tau \le 1\) and \(\tau =0\) corresponds to the space-time curve \(\Gamma \), while \(\tau = 1\) represents the initial time. It is clear that the change of variables \((s,r)\mapsto (\tau ,r)\) is nonsingular since \(g(r)>0\) on [0, 1] (Fig. 2).
The operator \(r\partial _r\) expressed in the new variables is denoted by \(\Lambda \) and it reads
We also use the abbreviation
so that
From (1.22) we immediately see that the unknown \(\phi \) solves
where
is the pressure term in new variables \((\tau ,r)\). In \(( \tau , r)\)-coordinates the dust collapse solution (1.33) is denoted by \(\phi _0\), it solves
and is given explicitly by
After a simple calculation we obtain
In particular, \(\mathscr {J}[\phi _0](\tau ,r)>0\) for all \((\tau ,r)\in (0,1]\times [0,1]\), and
The connection between the above formulas and mass conservation for the dust solution is detailed in Section 1.2. From the formula (1.50), (1.44), and (1.19) we conclude that for \(0<\tau \ll 1\)
from which the scale \(r^n/\tau \) emerges naturally and will play an important role in our work.
We will prove Theorem 1.6 in the \((\tau ,r)\)-coordinate system, using (1.46) as a starting point. This is natural, as the collapse surface in the new coordinates takes on a simpler description \(\Gamma =\{\tau =0\}\).
1.5 Methodology and Outline of the Proofs
The continuity equation in Lagrangian coordinates reduces to (1.24), which implies that the blow-up points of the density coincide with the zero set of the Jacobian determinant
Therefore, the key goal of this work is to identify a class of initial data that in a suitable sense mimic the bahavior of the dust solution and we do that by showing
A natural idea is to consider the dynamic splitting
where the relative remainder R is expected to be small in an appropriate sense. A straightforward calculation gives a partial differential equation satisfied by R, which at the leading order takes the schematic form,
where one can show that
The source term \(\bar{F}\) contains, as a leading order contribution the expression
which in the region \(0<\tau \ll 1, 0<r\ll 1\) has the asymptotic form
in the zone \(\{ \frac{r^n}{\tau }\approx 1\}\), where \(0<\delta =\delta (\gamma ,n)<\frac{2}{3}\) is a quantity defined later in (1.60) with the property \(\lim _{\gamma \rightarrow \frac{4}{3}}\delta (\gamma ,n)=0\). The simplest way of interpreting the relative “strength” of each of the terms in (1.53) is to compute the associated energies by taking the inner product with \(\partial _\tau R\). Assuming that we can obtain a coercive energy contribution on the left-hand side which roughly controls
we then have to control a source term of the form
which is clearly too singular to be controlled by the quadratic form (1.54), since \(0<\delta <\frac{2}{3}\). We must therefore refine our approximate solution \(\phi _0\) so to obtain a less singular source term.
We note that we have already implicitly used the assumption \(\gamma <\frac{4}{3}\) via the scaling transformation (1.6), which resulted in the occurrence of the small parameter \(\varepsilon \) in (1.46). We want to further use \(\gamma <\frac{4}{3}\), but with a more refined dynamic splitting ansatz. Namely, our main idea is to seek a more special solution \(\phi \) of the form
where \(\phi _{\mathrm{app}}\) will be chosen as a more accurate approximate solution of the Euler–Poisson system (1.46) in hope of mitigating the issue explained above. The exponent \(m>0\) is a sufficiently large positive number, so that H is a weighted remainder, small relative to \(\tau ^m = \phi _0^{\frac{3}{2} m}\ll \phi _0\) for small values of \(\tau \).
Step 1. Hierarchy and the construction of the approximate solution \(\phi _{\mathrm{app}}\) (Section 2).
We shall find the approximate profile \(\phi _{\mathrm{app}}\) as a finite order expansion into the powers of \(\varepsilon \) around the background dust profile \(\phi _0\), that is
With the solution ansatz (1.57) we can formally Taylor expand the pressure term \(\varepsilon P[\phi _0 + \varepsilon \phi _1 + \cdots ]\) into the powers of \(\varepsilon \), thus giving us a hierarchy of ODEs satisfied by the \(\phi _j\):
Functions \(f_{j+1}\), \(j=0,1,\ldots ,M\) are explicit and generally depend nonlinearly on \(\phi _k\), \(0\le k\le j\), and their spatial derivatives (up to the second order).
The system of ODEs (1.58) can be solved iteratively as the right-hand side \(f_{j+1}\) is always known as a function of the first j iterates.Footnote 1 To show that finite sums of the form (1.57) are good approximate solutions of (1.46), we must prove that the iterates \(\phi _j\), \(j\ge 1\), are effectively “small” with respect to \(\phi _0\). The mechanism by which this is indeed true is one of the key ingredients of the paper, in both the conceptual and the technical sense. In particular we shall have to choose special solutions of (1.58), as they are in general not unique (the two general solutions of the homogeneous problem are \(\tau ^{4/3}\) and \(\tau ^{-1/3}\)), which will allow us to see the above mentioned gain. We now proceed to explain these ideas in more detail. To provide a quantitative statement, we assume that the enthalpy profile w satisfies
in a neighbourhood of the center of symmetry \(r=0\). The exponent \(n\in \mathbb {N}\) is our effective measure of flatness of the star close to the center. For a given \(\gamma \in (1,\frac{4}{3})\) we consider densities (1.59) with n so large that
With this assumption in place we prove that the iterates \(\{\phi _j\}_{j\in \mathbb {N}}\) “gain” smallness and this conclusion is summarized in the following theorem:
Theorem 1.10
Let \(M,K\in \mathbb {Z}_{>0}\) be given. There exists a sequence \(\{\phi _{j}\}_{j\in \{0,\ldots ,M\}}\) of solutions to (1.58) with \(\phi _0(\tau ,r)=\tau ^{\frac{2}{3}}\), constants \(C_{jkm}\) depending on K and M, and a \(\lambda >\frac{2}{n}\) such that for \(j\in \{1,\ldots ,M\}\) and \(\ell ,m \in \{0,1,\ldots ,K\}\) we have
Therefore the iterates \(\phi _{j}\) exhibit a crucial gain of \(\tau ^{j\delta }\) with respect to the dust profile \(\phi _0=\tau ^{\frac{2}{3}}\)! This is one manifestation of the supercriticality (that is \(1<\gamma <\frac{4}{3}\)) of the problem and it can be viewed as the gain of smallness in the singular regime \(0<\tau \ll 1\).
To motivate (1.61), we explain informally how the gain happens for \(\phi _1\). To find \(\phi _1\) we solve the ODE
For \(0\le r\ll 1\) we have \(w\approx g(r)\approx 1\). Approximating \(\Lambda \approx r^n\partial _\tau + r\partial _r\), and by (1.51) \(\mathscr {J}[\phi _0]\approx \tau (\tau +r^n)\), \(r\ll 1\), we obtain
We expect \(\phi _1\) to “gain” 2 powers of \(\tau \) with respect to the right-hand side of (1.62), and thus
with \(\delta \) defined in (1.60). Of course \(\delta \) can be positive if and only if \(\gamma <\frac{4}{3}\) and the exponent n from (1.59) is sufficiently large!
Most important consequence of Theorem 1.10 is that it leads to a source term \(\mathscr {S}(\phi _{\mathrm{app}})\) generated by \(\phi _{\mathrm{app}}\) (see Lemma 5.14) which satisfies a natural improved bound
Therefore, if \(M \gg 1,\) it is reasonable to expect that the remainder ansatz \(\tau ^{m}\frac{H}{r}\) (with \(m\ge M\delta \)) is consistent with our strategy of treating H as an error term, in the regime \(\tau \ll 1\).
Another crucial input in (1.61) is the factor \(\frac{\left( \frac{r^n}{\tau }\right) ^{\lambda -\frac{2}{n}}}{(1+\left( \frac{r^n}{\tau }\right) )^{\lambda }}\le 1\). The gain is visible only in the asymptotic regime \(r^n/\tau \ll 1\), which suggests that the scale
plays a critical role in our problem. Indeed, this gain is important in the closure of the energy estimate for H—it is used to absorb various negative powers of r which inevitably appear in our high order energy scheme intimately tied to the assumed spherical symmetry.
The proof of Theorem 1.10 is complex and delicate. It is based on the introduction of special solution operators \(S_{1}\) and \(S_{2}\) (2.28)–(2.29) for the ODE (1.58). In addition to a careful and precise tracking of the powers of \(\tau \) and \(\frac{r^{n}}{\tau }\), to see the gain of \(\tau ^{j\delta }\) one has to use different solution operators \(S_{1}\) and \(S_{2}\) for \(j\leqq \lfloor \frac{2}{3\delta }\rfloor \) and for \(j>[\frac{2}{3\delta }]\), respectively. The precise estimate (1.61) and the emergence of \(\frac{r^n}{\tau }\) as a critical quantity is intimately tied to the algebraic structure of \(f_j\), \(j=1,\ldots ,M\), which in turn possesses a rich geometric information related to the Taylor expansion of the negative powers of the Jacobian determinant \(\mathscr {J}[\phi ]\).
Step 2. Equation for the remainder H (Section 3).
Thanks to the crucial gain of \(\tau ^{j\delta }\) and in the presence of \(\tau ^{m}\) factor, now H satisfies the following quasilinear wave-like equation:
where at the leading order
The precise formulas for the right-hand side F, \(g^{00}\), \(g^{01}\), and \(c[\phi ]\) are given in (3.26)–(3.28), (3.19) respectively. In comparison to (1.53), the remarkable new feature of (1.63) is the presence of the coefficient \(\frac{m(m-1)}{\tau ^{2}}\) so that
for m sufficiently large. This leads to a coercive positive definite control of the solution at the singular surface \(\{\tau =0\}\).
Step 3. The physical vacuum and weighted energy spaces (Section 4)
Much of the difficulty in producing energy estimates for (1.63) comes from an antagonism between two different singularities present in the equation.
-
at \(\tau =0\) the coefficient \(c[\phi _{\mathrm{app}}]\) and various others formally blow up to infinity. This is the singularity associated with the collapse at the singular surface \(\tau =0\) and already explained above;
-
at \(r=1\) we have \(w=0\) and therefore the elliptic part of the quasilinear operator on the left-hand side of (1.63) does not scale like the Laplacian as \(r\rightarrow 1\). This is a well-known degeneracy associated with the presence of the vacuum boundary.
The assumption of physical vacuum can be recast as the requirement that the enthalpy \(w\ge 0\) behaves like a distance function when \(r\sim 1\), that is
Requirement (1.64) is important in establishing the well-posedness of (1.63). The local well-posedness theory for the physical vacuum problem was first developed in the Euler case [17, 37], while the well-posedness statements for the gravitational Euler–Poisson system can be found in [26, 29, 31, 34, 42]. Nevertheless, the well-posedness theory cannot be directly applied to our setting, as (1.63) differs from the above mentioned works in two important aspects: the problem has explicit singularities at \(\tau =0\) and the space time domain \((\tau ,r)\in (0,1]\times [0,1]\) is strictly larger than the domain \((s,r)\in [0,\frac{1}{g(0)})\times [0,1]\) which only covers the star dynamics up to the first stipulated collapse time \(t^*=\frac{1}{g(0)}\), see Section 1.4.
Step 4. Energy estimates and the conclusion (Sections 5, 6).
Since \(\phi _{\mathrm{app}}\sim \phi _{0}=\tau ^{2/3},\) \(\tau \)-derivatives of \(\phi _{\mathrm{app}}\) create severe singularities in \(\tau \) as \(\tau \rightarrow 0\), which leads to difficulties in our energy estimates. We must in particular abandon the use vector field \(\partial _{\tau }\) to form the natural high-order energy and instead rely on purely spatial derivatives. Due to very precise and delicate features of the approximate solution \(\phi _{\mathrm{app}}\) near the center \(r=0\) (as described in Theorem 1.10), we are forced to use polar coordinates throughout [0, 1], which results in the introduction of many novel analytic tools to control the singularity at \(r=0.\)
To motivate the definition of high-order energy spaces we isolate the leading order spatial derivatives contribution from the left-hand side of (1.63):
where
is the radial expression for the three-dimensional divergence operator. This particular form of \(L_\alpha \) suggests that we have to carefully apply high-order derivatives to (1.63) in order to avoid singularities at \(r=0\). We therefore introduce a class of operators defined as concatenations of \(\partial _r\) and \(D_r\):
and set \(\mathcal {D}_0 =1\). The operators \(\mathcal {D}_j\) are then commuted with the Equation (1.63). For some N sufficiently large, the idea is to form the energy spaces by evaluating the inner product of the commuted equation with \(\mathcal {D}_j H_\tau \), \(j=1,\ldots , N\). However, following the ideas developed in [29, 30, 37], we need to perform our energy estimates in a cascade of weighted Sobolev-like spaces. For any given \(j\in \{1,\ldots ,N\}\) the correct choice is the inner product associated with the weights \(w^{\alpha +j}\).
Definition 1.11
(Weighted spaces). For any \(i\in \mathbb {Z}_{\ge 0}\) we define weighted spaces \(L^2_{\alpha +i}\) as a completion of the space \(C_c^\infty (0,1)\) with respect to the norm \(\Vert \cdot \Vert _{\alpha +i}\) generated by the inner product
and denote the associated norm by \(\Vert \cdot \Vert _{\alpha +i}.\)
Definition 1.12
(Weighted space-time norm). For any \(0< \kappa \le 1\), \(N\in \mathbb {Z}_{>0}\), \(\kappa \le \tau \le 1\) we define the weighted space-time norm
where \(q_\nu (x) = (1+x)^\nu \), \(\nu \in \mathbb {R}\).
We see that the powers of the w-weights increase with the number of derivatives. Such spaces are carefully designed to control the motion of the free boundary at \(r=1\) and the key technical tool in our estimates is the Hardy inequality. This is natural since \(w\sim 1-r\) near \(r=1\). Similarly, the presence of \(\tau \)-weights allows us to precisely capture the degeneration of our wave operator at the singular space-time curve \(\{\tau =0\}\).
The positive function \(x\mapsto q_\nu (x)\) serves as a weight for the top order spatial derivative contributions in the above definition, with powers \(\nu =-\frac{\gamma +1}{2}\) and \(\nu =-\frac{\gamma +2}{2}\) respectively. Such weights appear in the dust Jacobian \(\mathscr {J}[\phi _0]\) and by means of expanding the true solution around \(\phi _0\), functions \(q_\nu \) appear naturally in our energies. The presence of \(q_\nu \) highlights again the importance of the characteristic scale \(r^n/\tau \) in our problem. We shall prove the following key theorem.
Theorem 1.13
(The \(\kappa \)-problem). Let \(\gamma \in (1,\frac{4}{3})\) and \(m\ge \frac{5}{2}\) be given. Set \(N=N(\gamma )=\lfloor \frac{1}{\gamma -1} \rfloor +6\). For a sufficiently large \(n=n(\gamma )\in \mathbb {Z}_{>0}\), there exist \(\sigma _*,\varepsilon _*>0\), \(M=M(m,\gamma ,n)\gg 1\) and \(C_0>0\), such that for any \(0<\sigma <\sigma _*\) and any \(0<\varepsilon <\varepsilon _*\) the following is true: for any \(\kappa \in (0,1)\) and any initial data \((H_0^\kappa ,H_1^\kappa ]\) satisfying
there exists a unique solution solution \(\tau \mapsto H^\kappa (\tau ,\cdot )\) to (1.63) on \([\kappa ,1]\) satisfying
Theorem 1.13 gives uniform-in-\(\kappa \) bounds for the sequence \(H^\kappa \) with initial data specified at time \(\tau =\kappa \). One may for example choose trivial data at \(\tau =\kappa \), that is set \(\sigma =0\) in the above theorem to generate a family of solutions \(\{H_\kappa \}_{\kappa \in (0,1]}\). As \(\kappa \rightarrow 0\) we conclude the existence of a solution H on (0, 1]. By (1.56), this gives a solution \(\phi = \phi _{\mathrm{app}}+ \tau ^m\frac{H}{r}\) of the original problem (1.46), thus allowing us to prove Theorem 1.6 (after going back to the (s, r) coordinate system). The proof of Theorem 1.13 is given in Section 5.6, while the proof of Theorem 1.6 is given in Section 6.
Remark 1.14
Note that the small parameter \(\varepsilon \) used for the construction of the approximate solution \(\phi _{\mathrm{app}}\) enters explicitly in (1.63).
Remark 1.15
As part of the proof of Theorem 1.13, we also obtain a lower bound on the parameter M—the expansion order of the approximate solution \(\phi _{\mathrm{app}}= \sum _{j=0}^M\varepsilon ^j\phi _j.\) A precise formula is given in (5.122).
Many of our energy estimates depend crucially on both the gain of a \(\tau ^\delta \)-power and a power of \(\frac{r^{n}}{\tau }\) in Theorem 1.10. The former allows us to obtain a crucial gain of integrability-in-\(\tau \) close to the singular surface \(\tau =0\), while the latter is needed to absorb negative powers of r arising from the application of the operators \(\mathcal {D}_j\) on \(\phi _{\mathrm{app}}\). This delicate interplay works out, but requires a certain “numerological” constraint, namely the coefficient n has to be large enough relative to the total number of derivatives N used in our energy scheme.
Despite the delicate tools and analysis, one term stands out and seemingly causes a major obstruction to our method. After commuting the equation with high-order operators \(\mathcal {D}_j\) and evaluating the \((\cdot ,\cdot )_{\alpha +j}\)-inner product with \(\mathcal {D}_j H_\tau \), an error term \(\mathscr {M}[H]\) defined in (4.8) emerges. A simple counting argument suggests that the number of powers of w in \(\mathscr {M}[H]\) is insufficient to close the estimates near the vacuum boundary, but we carefully exploit a remarkable algebraic structure within the term and obtain the necessary cancellation, see Lemma 5.8.
The last claim of Theorem 1.6 shows that the infinitesimal volume of the shrinking domain of our collapsing solution behaves like the infinitesimal volume of the collapsing dust profile. More importantly, using (1.24), one can conclude that the qualitative behavior on approach to the singular surface \(\tau =0\) of the Eulerian density \(\tilde{\rho }\) is the same as that of the dust density, see Section 6.
1.5.1 Plan of the Paper
Section 2 is devoted to the derivation of the hierarchy of ODEs (1.58) and the proof of Theorem 1.10. In Section 3 we derive the equation for the remainder term H. In Section 4 we introduce the high-order differentiated version of the H-equation derived in Section 3. We also define high-order energies that arise naturally from integration-by-parts and show (Section 4.2) that they are equivalent to norm \(S_N^\kappa \) from Definition 1.12. The remainder of the section is devoted to various a priori estimates and preparatory bounds. In Section 5 we prove the key energy estimates, culminating in the proof of Theorem 1.13 in Section 5.6. Finally, Theorem 1.6 is shown in Section 6. In Appendices A–C many important properties and analytic tools used in our estimates are shown. We present details of the product and chain rule within vector field classes \(\mathcal {P}\) and \(\bar{\mathcal {P}}\) (“Appendix A”), commutator identities (“Appendix B”), and the Hardy–Sobolev embeddings (“Appendix C”). Finally, for the sake of completeness, we sketch the local well-posedness argument in “Appendix D”.
1.6 Notation
-
By \(\mathbb {Z}_{\ge 0}\), \(\mathbb {Z}_{>0}\) we denote the sets of non-negative and strictly positive integers respectively.
-
\(C^0([a,b], [c,d])\) denotes the space of continuous functions \((\tau ,r)\mapsto f(\tau ,r)\) on the set \([a,b]\times [c,d]\).
-
We use \(\Vert {\cdot }\Vert _{L^2}\) to denote \(\Vert {\cdot }\Vert _{L^2([0,1];r^2dr)}\) and \(\Vert \cdot \Vert _\infty \) to denote \(\Vert {\cdot }\Vert _{L^\infty ([0,1])}\).
-
Writing \(A\lesssim B\) means that there exists a universal constant \(C>0\) such that \(A\le CB\). \(A > rsim B\) simply means \(B\lesssim A\). If we write \(A\approx B\) we mean \(A\lesssim B\) and \(A > rsim B\).
-
For a given \(a>0\) we denote the closed three-dimensional ball of radius a centered at 0 by \(B_a(0)\).
2 The Hierarchy
Formally we would like to build a solution of (1.46) as a sum of the approximate profile \(\phi _{\mathrm{app}}\) (given as a finite series expansion in the powers of \(\varepsilon \)) and the remainder term \(\theta \) which we hope to show to be suitably small. In other words, we are looking to write
Plugging (2.1) into (1.46), we will now derive a formal hierarchy of ODEs satisfied by the functions \(\phi _j\), \(j\in \{1,\ldots , M\}\). We define the source term \(S(\phi _{\mathrm{app}})\):
We first recall the formula of Faa Di Bruno (see for example [38]) which will be repeatedly used in this section. Given two functions f, g with formal power series expansions,
we can compute the formal Taylor series expansion of the composition \(h= f \circ g\) via
where \(f_n\), \(g_n\) and \(h_n\) are constants with respect to x. Faa Di Bruno’s formula gives
where
An element of \(\pi (n,k)\) encodes the partitions of the first n numbers into \(\lambda _i\) classes of cardinality i for \(i\in \{1,\ldots ,k\}\). Observe that by necessity \(\lambda _j=0\) for any \(n-k+2\le j\le n\).
Lemma 2.1
(Detailed structure of the source term \(S(\phi _{\mathrm{app}})\)). The source term \(S(\phi _{\mathrm{app}})\) given by (2.2) satisfies
where \(R_P^\varepsilon = R_P^\varepsilon [\phi _0,\phi _1,\ldots ,\phi _M]\) and \(R^\varepsilon _{M,2}=R^\varepsilon _{M,2}[\phi _0,\phi _1,\ldots ,\phi _M]\) are explicitly given by (2.17), (2.9) below, \(f_0 := 0\), and
with \(\tilde{O}_j \) and \(h_m\) given explicitly below by (2.10) and (2.15).
Proof
For any \(m\in \mathbb {N}\), \(\nu \in \mathbb {R}\), there exists a smooth function \(R^\varepsilon _{m,\nu }:\mathbb {R}^m\rightarrow \mathbb {R}\) such that
where by the formula of Faa Di Bruno
and \(R^\varepsilon _{m,\nu }\) is smooth in a neighbourhood of \(\mathbf{0}\),
and for any \(\varepsilon \in (0,1)\),
for some constant \(C_\ell >0\) which grows as \(\ell \) gets larger.
Recalling (2.1),
where
Note that from (2.5), for \(k=1,\) \(\lambda _{j}=1,\) and \(\lambda _{i}=0\) for \(i<j,\) so that the summation for \(k=1\) is given by \(-2\frac{\phi _{j}}{\phi _{0}}.\) From (2.5) again, for \(k\geqq 2,\) \(\lambda _j=0\), and therefore in the definition of \(\tilde{O}_j \), the expression depends only on \(\phi _1,\ldots \phi _{j-1}\), justifying the notation \( \tilde{O}_j = \tilde{O}_j[\phi _0,\ldots ,\phi _{j-1}]. \) Note that \(\tilde{O}_1=0\).
Our next goal is to expand the function \(\mathscr {J}[\phi _{\mathrm{app}}]^{-\gamma }\) in the powers of \(\varepsilon \). To that end we first observe that
where
where we note that \(\bar{\mathscr {J}}_0=1\) since \(\mathscr {J}_0=\mathscr {J}[\phi _0]\). The remainder \(R_{\mathscr {J}}\) is given by the formula
We have
where we use (2.7). Here \(h_0=1\) and the formula of Faa Di Bruno gives
and
Similarly
where
so we finally have
where
From the definition of the source term (2.2) it therefore follows that
with \(f_0 := 0\) and
as claimed. \(\square \)
Motivated by the previous lemma, we define the hierarchy of ODEs as
where \(\phi _{0}^{3}=\tau ^{2}\), \(f_j\) is given by (2.19) and \(\tilde{O}_j \) and \(h_j\) given by (2.10) and (2.15) respectively.
With \(\{\phi _j\}_{j=1,\ldots ,M}\) satisfying (2.20), Lemma 2.1 in particular implies that
with \(R_P^\varepsilon = R_P^\varepsilon [\phi _0,\phi _1,\ldots ,\phi _M]\) and \(R^\varepsilon _{M,2}=R^\varepsilon _{M,2}[\phi _0,\phi _1,\ldots ,\phi _M]\) given by (2.17), (2.9). Therefore, by solving the hierarchy up to order M we force the source term to be of order \(\varepsilon ^{M+1}\).
2.1 Solution Operators and Definition of \(\phi _j\), \(j\in \mathbb {Z}_{>0}\)
For any \(\gamma \in (1,\frac{4}{3})\) we define
The number N will later correspond to the total number of derivatives used in our energy estimates.
Definition 2.2
(The “gain” \(\delta \) and \(\delta ^*\)). Let \(\gamma \in (1,\frac{4}{3})\) be given and let \(\bar{\gamma } = \frac{4}{3}-\gamma \). For any natural number \(n>\frac{N+2}{2\bar{\gamma }}\) we define
Lemma 2.3
Let \(\gamma \in (1,\frac{4}{3})\) be given and fix an arbitrary natural number \(a\in \mathbb {Z}_{>0}\). Then there exists an \(n^*=n^*(\gamma , a)\) such that
In fact
Proof
For the simplicity of notation let \(j:=\left\lfloor \frac{2}{3\delta (n)}\right\rfloor \). Then it is easy to check that (2.25) is equivalent to
Since \(1<\frac{1}{3\bar{\gamma }}\) it is clear that the above inequality will be true if n is chosen sufficiently large. \(\square \)
Remark 2.4
Lemma 2.3 implies in particular \(\frac{2}{3\delta } \notin \mathbb {Z}_{>0}\) since by (2.25) \(\left\lfloor \frac{2}{3\delta }\right\rfloor <\frac{2}{3\delta }\).
Definition 2.5
(Regularity parameter \(\lambda \)). Let \(\gamma \in (1,\frac{4}{3})\) be given. Choose an \(n>n^*(\gamma ,2N(\gamma ))\) (where \(n^*(\gamma ,a)\) is given by Lemma 2.3) sufficiently large so that
Remark 2.6
A simple consequence of Lemma 2.3 and Definition 2.5 is the bound
Motivated by (2.20), consider for a moment a general inhomogeneous ODE of the form
A simple calculation shows that the previous ODE is formally equivalent to
This motivates the following definition of the solution operators:
By direct inspection, one can check that for a given f functions \(S_i[\tau ^{\frac{4}{3}},\tau ^{-\frac{8}{3}},\tau ^{\frac{4}{3}} f]\), \(i=1,2\) are solutions of \(\partial _{\tau \tau }\phi - \frac{4}{9\tau ^2}\phi = f\). We define
The above definition of the solution is designed to enforce the gain of \(\tau ^\delta \) with respect to the previous iterate for all \(j\in \{1,\ldots ,M\}\). Since \(M\gg \lfloor \frac{1}{3\bar{\gamma }}\rfloor \), the above choice of the formula at the index values \(j>\lfloor \frac{1}{3\bar{\gamma }}\rfloor \) is crucial; see Proposition 2.8 and Lemma 2.14.
2.2 Bounds on \(\phi _j\) and Proof of Theorem 1.10
The main goal of this section is the proof of Theorem 1.10. To that end, we need a number of preparatory steps. We first introduce the notation
For the remainder of the section, constants \(M,K\in \mathbb {Z}_{>0}\) are arbitrarily large fixed constants.
Lemma 2.7
(Basis of the induction). Let \(\phi _1\) be given by (2.30). Then
The main result of this section is the quantitative estimate on the space-time derivatives of the iterates \(\phi _j\).
Proposition 2.8
(Inductive step). Let \(\phi _j\) be given by (2.30). Let \(1\le I< M\) be given and assume that for any \(j\in \{1,\ldots , I\}\) and any \(\ell ,m \in \{0,1,\ldots ,K\}\) we have
Then, for any \(\ell ,m \in \{0,1,\ldots ,K\}\), the following bound holds:
where \(\lambda = \frac{2N}{n}\) is given in Definition 2.5.
Remark 2.9
The constants in the above statement depend on \(K,M\in \mathbb {Z}_{>0}\) and they generally grow as K and M get larger.
Proofs of Lemma 2.7, Proposition 2.8, and finally Theorem 1.10 are contained in Section 2.2.2. Before that we need a number of auxiliary bounds.
2.2.1 Auxiliary Lemmas
Since
it is easy to see that for any \(\ell \in \mathbb {N}\) there exist some universal constants \(k_{abc_1\ldots c_a}\ge 0\), \(a,b,c_j=1,\ldots , \ell \), \(j=1,\ldots a\), such that
Lemma 2.10
(Auxiliary estimates). Let \(\ell ,m \in \{0,1,\ldots ,K\}\) be given nonnegative integers. Under the inductive assumptions (2.34) the following estimates hold:
Proof
Proof of (2.37). By the Leibniz rule for any \(k_1,k_2\in \mathbb {N}\) and any smooth function \(\varphi \) we have
where we have used (1.20) and \(\tau \le 1\) in the last estimate. Letting \(\varphi =\phi _0 = \tau ^{\frac{2}{3}}\) above we obtain
Now for any \(\ell ,m \in \{0,1,\ldots ,K\}\), we have
where we recall \(\mathscr {J}[\phi _0]=\phi _0^2(\phi _0+\Lambda \phi _0)\). Therefore, if \(\ell =0\), we have
and if \(\ell >0\), since \((r\partial _{r})^{\beta }\phi _{0}=0\) for \(\beta \ne 0,\) we have
which leads to (2.37). \(\square \)
Proof of (2.38)
The bound is obvious. \(\square \)
Proof of (2.39)
We use the formula of Faa Di Bruno. We may write \(\mathscr {J}[\phi _0]^{-k}(t,r) = f(h(r))\) where \(f(x) = x^{-k}\) and \(h(r) = \mathscr {J}[\phi _0]\). Derivatives of \(x\mapsto f(x)\) are easily computed:
Formula of Faa Di Bruno then gives
where we refer to (2.5) for the definition of \(\pi (m,j)\). Since
and
it follows that
Therefore, since \(\sum \lambda _{i}=j\) and \(\sum i\lambda _{i}=m,\) we have
\(\square \)
Proof of (2.40)
Using the formula of Faa Di Bruno like above, replacing formally \(\partial _\tau ^m\) by \((r\partial _r)^\ell \) we obtain
where we refer to (2.5) for the definition of \(\pi (\ell ,k)\). Therefore, for \(\ell \in \mathbb {Z}_{\ge 1}\), by using the Leibniz rule, we get
Notice that for any \(d,\ell \in \mathbb {Z}_{\ge 0}, i\in \mathbb {Z}_{\ge 1}\),
where we have made use of (2.37). From this bound, the product rule, and (2.53), we conclude that
where we have used (2.37), (2.54), (2.31), the identity \(\sum _{i=1}^\ell \lambda _i = j\) which follows from the definition of the index set \(\pi (\ell ,j)\), and the trivial estimate \(x\le q_1(x)\). \(\square \)
Proof of (2.41)
By letting \(\varphi =\phi _j\), \(j\in \{1,\ldots , I\}\), in (2.44) we obtain
where we have used the inductive assumption (2.34). If \(j=0\), from (2.45) we have
Recalling \(\mathscr {J}_{k}\) from (2.12), applying the Leibniz rule and using (2.34) and (2.56)–(2.57), we obtain
\(\square \)
Proof of (2.42)
We use the Faa Di Bruno formula again. Analogously to (2.53), we obtain
where \((a)_j=a(a-1)\ldots (a-j+1)\). Notice that for any \(d,\ell \in \mathbb {Z}_{\ge 0}, i\in \mathbb {Z}_{\ge 1}\),
where we have made use of (2.41). Using this bound just like in (2.55), we obtain (2.42). \(\square \)
Proof of (2.43)
Using the formula of Faa Di Bruno, for any \(k\in \{1,\ldots , I\}\) we have
where we have used the inductive assumption (2.34), identities \(\sum _{i=1}^m\lambda _i = j\), \(\sum _{i=1}^m(i\lambda _i)=m\) from (2.5), and the additive property of \(p_{\mu ,\nu }\).
By analogy to (2.55) we have
where we have used the inductive assumption (2.34), identities \(\sum _{i=1}^m\lambda _i = j\), \(\sum _{i=1}^m(i\lambda _i)=m\) and the additive property of \(q_\nu \). \(\square \)
Lemma 2.11
Recall \(h_j\) and \(\tilde{O}_j \) from (2.15) and (2.10). Under the inductive assumptions (2.34) the following estimates hold:
Proof
Proof of (2.59). Recall (2.15). For any \(j\in \{1,\ldots , I\}\) by the Leibniz rule
where we have used (2.40), (2.41), the additive property of \(p_{\mu ,\nu }\) and the exponent of \(\tau \) is simplified from (2.5):
\(\square \)
Proof of (2.60)
Recall (2.10). By the Leibniz rule
where we have used (2.38), (2.43), additive property of \(p_{\mu ,\nu }\), the bound \(p_{\lambda , -\frac{2}{n}}\le 1\) and the bound \(\sum _{i=1}^j\lambda _i = k\ge 2\). Note that for any \(k\ge 2\) and \((\lambda _1,\ldots ,\lambda _{j+1})\in \pi (j,k)\), we have \(\lambda _{j+1}=0\). \(\square \)
Proof of (2.61)
Assume first \(j\in \{1,\ldots , I\}\). Note that
Using the bound \(|(r\partial _r)^{a} w| \lesssim r^n\) for \(a\ge 1\), by the previous identity and (2.44)
where we have used (2.59), (2.40), and the additive property of \(q_\nu , p_{\mu ,\nu }\). If on the other hand \(j=0\), then the above proof and \(h_0=1\) give
\(\square \)
Lemma 2.12
Under the inductive assumptions (2.34), for any \(\ell ,m \in \{0,1,\ldots ,K\}\) the following estimate holds:
Proof
Using the Leibniz rule and the formula (2.19), we have
The worst case is \(j-1=0\) where we have \(d=i=k=0\), as in (2.61), since \(p_{2+\lambda ,-2/n}\le p_{1,0}\) by our choices of \(\lambda \) in Definition 2.5. We now choose \(p_{1,0}\) in (2.61) to obtain:
where we have used (2.34), (2.60), (2.61), and from the definition of \(\delta \) (2.23), the exponent
and the estimates \(p_{2\lambda ,-\frac{4}{n}}\le p_{\lambda ,-\frac{2}{n}}\) and \(q_{-\gamma }(x) x^{1-\frac{2}{n}} = \frac{x^{1-\frac{2}{n}}}{(1+x)^\gamma } \le \frac{x^{\lambda -\frac{2}{n}}}{{(1+x)^{\lambda }}} = p_{\lambda ,-\frac{2}{n}}(x)\). (We remind the reader of definitions (2.32) and (2.31) of \(x\mapsto p_{\mu ,\nu }(x)\) and \(x\mapsto q_\nu (x)\) respectively.) \(\square \)
Remark 2.13
When \(j=1\) we have \(\tilde{O}_1=0\), and thus from (2.19)
In particular \(f_1\) depends only on \(\phi _0\) and the inductive assumption (2.34) is not used in the proof of (2.62).
Lemma 2.14
-
1.
Let \(0<\lambda <1\) be given and let \(\beta \) satisfy
$$\begin{aligned} \beta -\lambda + \frac{2}{n} >-1. \end{aligned}$$Then the following bound holds:
$$\begin{aligned} \left| \int \nolimits _0^{\tau } (\tau ')^\beta p_{\lambda ,-\frac{2}{n}}(\frac{ r^n}{\tau '}) \,\mathrm{d}\tau ' \right| \lesssim \tau ^{\beta +1} p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) , \end{aligned}$$(2.63)where \(x\mapsto p_{\mu ,\nu }(x)\) is defined in (2.32).
-
2.
Let b satisfy
$$\begin{aligned} b + \frac{2}{n} <-1. \end{aligned}$$Then the following bound holds:
$$\begin{aligned} \left| \int \nolimits _\tau ^{1} (\tau ')^b p_{\lambda ,-\frac{2}{n}}\left( \frac{ r^n}{\tau '}\right) \,\mathrm{d}\tau ' \right| \lesssim \tau ^{b+1} p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) . \end{aligned}$$(2.64)
Proof
Proof of part (i). Applying the change of variables \(x=\tau '/ r^n\) we have
Case 1: \( r^n\ge \tau \). We have
where the very last inequality follows from \(1\lesssim p_{\lambda ,0}\left( \frac{r^n}{\tau }\right) \) which in turn relies on \( r^n\ge \tau \).
Case 2: \( r^n\le \tau \). We have from \(\beta -\lambda +\frac{2}{n}>-1\)
where the very last inequality follows from \(1\lesssim \frac{1}{(1+\frac{ r^n}{\tau })^{\lambda }}\) which in turn relies on \( r^n\le \tau \). \(\square \)
Proof of part (ii)
By the same change of variables as in (2.65) we have
We distinguish two cases again.
Case 1: \( r^n\ge \tau \). We have from \(b+\frac{2}{n}<-1,\)
where the last inequality follows from \(1\lesssim p_{\lambda ,0}\left( \frac{r^n}{\tau }\right) \) which in turn relies on \( r^n\ge \tau \), just like in Case 1 in part (i).
Case 2: \( r^n\le \tau \). We have
where the very last inequality follows from \(1\le \frac{1}{(1+\frac{ r^n}{\tau })^{\lambda }}\) which in turn relies on \( r^n\le \tau \). The two previous estimates together with (2.66) give (2.64). \(\square \)
2.2.2 Proofs of Proposition 2.8, Lemma 2.7, and Theorem 1.10
Proof of Proposition 2.8
We first assume that \(m=0\). Let \(k\in \mathbb {N}\) and \(|f|\lesssim \tau ^{4/3}\), \(|g|\lesssim \tau ^{-8/3}\), \(|h(\tau , r)|\lesssim \tau ^{k\delta }p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \). If \(k\le \left\lfloor \frac{1}{3\bar{\gamma }}\right\rfloor \) we then have
since \(k\delta -\lambda +\frac{2}{n}>-1\), where we have first used (2.63) and then (2.64). Note that we have used the assumption \(k\le \left\lfloor \frac{1}{3\bar{\gamma }}\right\rfloor \) and Lemma 2.3 to ensure that \( -\frac{5}{3}+k\delta + \frac{2}{n} <-1 \) and therefore (2.64) is applicable in the last line of (2.67). If \(k> \left\lfloor \frac{1}{3\bar{\gamma }}\right\rfloor \) we then have
where we have used (2.63) twice. We note that for any \(k> \left\lfloor \frac{1}{3\bar{\gamma }}\right\rfloor \) we have by Lemma 2.3\(-\frac{5}{3}+k\delta -\lambda +\frac{2}{n}>-1\) where we set \(a=2N-2\) and we recall Definition 2.5 of \(\lambda \). Therefore (2.63) is applicable in the second line.
By (2.20) and (2.30), and the facts \(\phi _{0} ^{2}=\tau ^{4/3},\phi _{0}^{-4}=\tau ^{-8/3},\)
where \(i=1\) or \(i=2\) according to (2.30). Since \(\left| (r\partial _r)^{\ell _1} \left( \phi _0^2\right) \right| \lesssim \tau ^{\frac{4}{3}}\), \(\left| (r\partial _r)^{\ell _2} \left( \phi _0^{-4}\right) \right| \lesssim \tau ^{-\frac{8}{3}}\) by (2.40) and \(\left| (r\partial _r)^{\ell _4}f_{I+1}\right| \lesssim \tau ^{-\frac{4}{3}+(I+1)\delta }p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \) by (2.62), we may apply (2.67)–(2.68) to conclude
When \(m=1\) we observe by taking \(\tau \) derivative of (2.28) and (2.29) and by Lemma 2.12, 2.14
Similarly, using the Leibniz rule like above,
For \(m\ge 2\) we simply use the equation
Applying \(\partial _\tau ^{m-2}(r\partial _r)^\ell \) to (2.70) we obtain
where we have used the inductive assumption (that has been verified for all \(m'<m\)) and Lemma 2.12. This completes the proof of Proposition 2.8. \(\square \)
Proof of Lemma 2.7
It remains to show the basis of induction, that is Lemma 2.7. By Lemma 2.12 for any \(\ell ,m \in \{0,1,\ldots ,K\}\) we have the bound
By Remark 2.13, bound (2.71) does not rely on the inductive assumptions (2.34). Using an argument identical to the proof of Proposition 2.8, we conclude Lemma 2.7.
\(\square \)
Proof of Theorem 1.10
The proof follows by induction on the index \(j\in \{1,\ldots ,M\}\). The claim is shown for \(j=1\) in Lemma 2.7, while the inductive step follows from Proposition 2.8. \(\square \)
3 Remainder Equations and the Main Results
We look for a solution of (1.46) in the form
where M is to be specified later.
3.1 Derivation of the Remainder Equations
Lemma 3.1
(PDE satisfied by \(\theta \)). Let \(\phi \), \(\phi _{\mathrm{app}}\), and \(\theta \) be related by (3.1). Then the equation satisfied by \(\theta \) reads
where the source term \(S(\phi _{\mathrm{app}})\) and the expressions \(\mathfrak {K}_j[\theta ]\), \(j=1,2,3\) are given by (2.2), (3.15), (3.16), (3.17) below respectively and c is given by (3.19).
Proof
We recall the formulas (1.44), (1.45), (1.47) of \(M_g\), the operator \(\Lambda \), and the nonlinear pressure term \(P[\phi ]\) respectively. Finally, recall the fundamental formula
Let
Then
We want to find alternative expression for
Note that
Since
and
where we have used the identity \(3\phi _{\mathrm{app}}^2 + 6\phi _{\mathrm{app}}\theta +3\theta ^2 = 3\phi ^2\). We may rewrite
Similarly,
and
We may write
where
Plugging (3.10) and (3.11) into (3.5), we deduce that
Note that
By writing
and
the last line of (3.13) can be rewritten as
Observe that
which asserts that the expression is a nonlinear term. Therefore by splitting
we obtain
where
and
Note that \(\mathfrak {K}_1[\theta ]\) contains both linear and nonlinear terms in terms of \(\theta \) and we view them as linear terms with nonlinear coefficients. \(\mathfrak {K}_2[\theta ]\) and \(\mathfrak {K}_3[\theta ]\) consist of quadratic and higher terms. We have distinguished them because \(\mathfrak {K}_3[\theta ]\) needs to be estimated together with the main linear elliptic operator in higher order estimates due to the presence of nonlinear factor c.
The \(\phi \) Equation (1.46) can be written as
where the source term \(S(\phi _{\mathrm{app}})\) is given by (2.2) and
\(\square \)
Lemma 3.2
(The H equation). Let
Then H solves
where
where the source term \(S(\phi _{\mathrm{app}})\) and the expressions \(\mathfrak {K}_j[\theta ]\), \(j=1,2,3\) are given by (2.2), (3.15), (3.16), (3.17).
Proof
The proof follows by a direct verification after plugging in \(\theta = \tau ^m r^{-1} H\) in (3.18). \(\square \)
We rewrite (3.21) in the form
where
The leading order operator
will be shown to be hyperbolic due to the bound \(1\lesssim g^{00} \lesssim 1\) shown later in Lemma 4.6. We shall see that the former estimate is crucially tied to the supercriticality (\(\gamma <\frac{4}{3}\)) and the flatness assumption on the enthalpy w near \(r=0\) (that is n sufficiently large in (1.19), that is Lemma 1.1). Moreover, \(\Box \) is also manifestly quasilinear as \(c[\phi ]\) depends on the space-time derivatives of H. The twofold singular nature of \(\Box \) coming from the gravitational singularity at \(\tau =0\) and the vacuum singularity at \(r=1\) is discussed at length in Section 1.5.
The basic equation for our energy estimates is obtained by dividing (3.26) by \(g^{00}\):
We denote the first summation without \(\sup \) in Definition 1.12 of \(S_\kappa ^N\) by \(E^N\) and the second summation without the time integral by \(D^N\), that is for any \(\tau \in (0,1]\) we let
Then the space-time norm can be written as
4 High-Order Energies and Preparatory Bounds
4.1 High-Order Equations and Energies
In order to derive high-order equations, we first introduce the elliptic operators
Then for any f, h we have
where we recall the inner product \((\cdot ,\cdot )_k\) given in (1.67).
We recall here the definition of the fundamental high-order differential operators \(\mathcal {D}_j\) given in (1.66). We then define
Important role is played by the operator \(\bar{\mathcal {D}}_i\) defined as
Let \(1\le i\le N\). After applying \(\mathcal {D}_i\) to (3.31) we use Lemmas B.1–B.2 to derive the equation for \(\mathcal {D}_i H\):
Here \(\mathcal {C}_i\) contains all the commutators
where the functions \(\zeta _{ij}\) are given by (B.416) and the commutators \([\cdot ,\cdot ]\) are defined in (B.417). Furthermore,
Note that we have written for \(i\geqq 1\),
Definition 4.1
(Weighted high-order energies). For any \(0< \kappa \le 1\) and \(N\in \mathbb {N}\) we define the high-order energies
where for any \(0\leqq j \leqq N\) we have
and
Remark 4.2
It will be shown in Section 4.2, Lemma 4.6, that every summand appearing in the definition of \(\mathscr {D}_j\) above is positive in our bootstrap regime.
Proposition 4.3
Assume that H is a sufficiently smooth solution to (3.26). The the following energy identity holds:
where for any \(i\in \{1,\ldots , N\}\), the error terms \(\mathcal {R}_i\) are explicitly given by
where \(\mathcal {C}_i[H]\) is given by (4.7) and \(\mathscr {M}[H]\) by (4.8). When \(i=0\), we replace \(\mathcal {C}_{i}[H] +\bar{\mathcal {D}}_{i-1}\mathscr {M}[H]\) in the above formula by \(-\varepsilon \frac{\mathscr {N}_0[H]}{g^{00}}\).
Proof
We evaluate the \((\cdot ,\cdot )_{\alpha +i}\)-inner product of (4.6) with \(\tau ^{\gamma -\frac{5}{3}}\mathcal {D}_i H_\tau \), and use Definition 4.1. \(\square \)
4.2 A Priori Bounds and the Energy-Norm Equivalence
Assume that H is a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\). For a sufficiently small \(\sigma '<1\), to be fixed later, we stipulate the following a priori bounds.
Lemma 4.4
Assume that H is a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) hold. Then for any \(( \tau , r)\in [\kappa ,T]\times [0,1]\)
Proof
Proof of (4.14). Let \(h: =\frac{\phi }{\phi _0}\). By (3.1) and (3.20) we have
By Proposition 2.8 and the a priori assumption (4.13) for any \(( \tau , r)\in [\kappa ,T]\times [0,1]\) we have
for \(\varepsilon ,\sigma '>0\) sufficiently small. \(\square \)
Proof of (4.15)
Note that
Therefore, in view of (4.14) it suffices to prove
Recall that
Moreover,
From (4.25), and Proposition 2.8 with the crude bound \(p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \lesssim 1\) and the bound
we have
Now the bound (4.15) follows from (4.29), (4.28), (4.14), and (4.13). \(\square \)
Proof of (4.16)
By (3.1), (3.20), and Proposition 2.8 we have
where we have used the a priori bounds (4.13), the crude bound \(\varepsilon \tau ^\delta p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \lesssim 1\) and the assumption \(m\ge \frac{5}{2}\). \(\square \)
Proof of (4.17)
This is similar to the proof of (4.16). With \( r\partial _r{\phi _0} =0\), applying \( r\partial _r\) we obtain
where we have used (4.13) in the last line and the crude bound \(p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \lesssim 1\). \(\square \)
Proof of (4.18)
By (3.1), (3.20), and Proposition 2.8 we have
where we have used the a priori bounds (4.13), \(\sigma '< 1\), \(p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \lesssim 1\), and the assumption \(m\ge \frac{5}{2}\). \(\square \)
Proof of (4.19)
By (3.1), (3.20), and Proposition 2.8, for any \(\ell =0,1,2\), we have
where we have used the a priori bounds (4.13) and the assumption \(m\ge \frac{5}{2}\). \(\square \)
Proof of (4.20)
\(\square \)
Proof of (4.21)
From the definition of \(\Lambda \) we have
where we have used the crude bound \(\varepsilon \tau ^\delta p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \lesssim 1\). \(\square \)
Proof of (4.22)
From the definition of \(\Lambda \) we have
where we have used (4.16), (4.17), and (4.18). \(\square \)
Proof of (4.23) and (4.24)
By (4.25) and (4.13), we have (4.23). To show (4.24) we first observe that \(| \Lambda h|+|\tau \partial _\tau \Lambda h| \lesssim \tau ^\delta \), which is a simple consequence of the bounds shown above. We recall here \(h = \frac{\phi }{\phi _0}\). Now the bound follows from (4.26), (4.29), (4.13). \(\square \)
Lemma 4.5
Assume that H is a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) hold. Then for any \(( \tau , r)\in [\kappa ,T]\times [0,1]\)
Proof
Proof of (4.30). Recall the definition of \(c[\phi ]\) (3.19). By (4.15) we have \(\mathscr {J}[\phi ]\approx \mathscr {J}[\phi _0]\approx \tau ^2 q_1\left( \frac{r^n}{\tau }\right) \), where we have used (4.28) to infer the last equivalence. By (4.15) \(\phi ^4 \approx \tau ^{\frac{8}{3}}\). Therefore
\(\square \)
Proof of (4.31)
Since \(\partial _\tau \mathscr {J}[\phi ] = 2\phi \phi _\tau (\phi + \Lambda \phi ) + \phi ^2(\phi _\tau + \partial _\tau \Lambda \phi )\), bounds (4.16), (4.20), and (4.21) imply
\(\square \)
Proof of (4.32)
Since \( r\partial _r\mathscr {J}[\phi ] = 2\phi r\partial _r\phi (\phi + \Lambda \phi ) + \phi ^2( r\partial _r\phi + r\partial _r\Lambda \phi )\), bounds (4.19), (4.20), and (4.22) imply
\(\square \)
Proof of (4.33)
From the definition of \(c[\phi ]\) it is easy to check the identity \(\partial _\tau c[\phi ] = c[\phi ] \left( 4 \frac{\phi _\tau }{\phi } - (\gamma +1)\frac{\partial _\tau \mathscr {J}[\phi ]}{\mathscr {J}[\phi ]}\right) \). Therefore
where we have used (4.16), (4.31), and (4.15). \(\square \)
Proof of (4.34)
Like in the proof of (4.33) we have
where we have used bounds (4.19), (4.30), and (4.32). \(\square \)
Lemma 4.6
Assume that H is a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) hold. Then for any \(( \tau , r)\in [\kappa ,T]\times [0,1]\) the following bounds hold:
Proof
Proof of (4.35). By definition (3.27) of \(g^{00}\) it suffices to check that \(\left\| c[\phi ] (\partial _r g)^2\right\| _{C^0([\kappa ,T]\times [0,1])} \lesssim 1\). By (4.30) and the bound \(\left| \partial _r g\right| \lesssim r^{n-1}\) for all \( r\in [0,1]\) (by (1.21)) we have
where we recall \(\delta = \frac{8}{3}-2\gamma -\frac{2}{n}>0\) and \(x\mapsto \frac{x^{2-\frac{2}{n}}}{(1+x)^{\gamma +1}}\) is clearly bounded for all \(x\ge 0\) and any \(\gamma >1\). This proves (4.35). \(\square \)
Proof of (4.36)
From (3.27) we have
where we have used (4.34), (4.30). \(\square \)
Proof of (4.37)
Like above, we need to show \(\left| \partial _\tau c[\phi ]\right| r^{2n-2} \lesssim \tau ^{\delta -1}\). Applying (4.33), it then follows
\(\square \)
Proof of (4.38)
From (3.28) we have
where we have used (4.34), (4.30). The bound for \(\left| \frac{g^{01}}{r}\right| \) follows analogously. \(\square \)
Proof of (4.39)
It is clear that
where we have used (4.35), (4.38), (4.36) and \(g^{01}w^{-1} = -\varepsilon \gamma c[\phi ] \frac{M_g }{r}\), \(M_g \) defined in (1.44). Note that a negative power of w is fortunately cancelled away as one positive power of w is contained in the definition of \(g^{01}\). \(\square \)
Proof of (4.40)
It clearly suffices to show \(\partial _\tau \left( \frac{d(\tau ,r)^2}{g^{00}}\right) \lesssim \varepsilon \tau ^{\delta -1}\). Observe that \(\partial _\tau \left( d^2\right) = \frac{4}{3} \left( \frac{\phi _{\mathrm{app}}}{\phi _0}\right) ^{-4} \partial _\tau \left( \frac{\phi _{\mathrm{app}}}{\phi _0}\right) \). Since \(\partial _\tau \left( \frac{\phi _{\mathrm{app}}}{\phi _0}\right) = \sum _{j=1}^M\varepsilon ^j\partial _\tau \left( \frac{\phi _j}{\phi _0}\right) \), it follows that \(\left| \partial _\tau \left( \frac{\phi _{\mathrm{app}}}{\phi _0}\right) \right| \lesssim \varepsilon \tau ^{\delta -1}\). Therefore \(\left| \partial _\tau \left( d(\tau ,r)^2\right) \right| \lesssim \varepsilon \tau ^{\delta -1}\). Together with (4.37) the claim follows. \(\square \)
Proof of (4.41)
Observe the identity \(\tau ^{\gamma -\frac{5}{3}}c[\phi _0] = g^{-2}\left( \tau +\frac{2}{3}M_g \right) ^{-\gamma -1}\). Taking a \(\tau \)-derivative we obtain
where we have used (1.26). Since \(1\lesssim g, G \lesssim 1\), the claim follows. \(\square \)
A corollary of Lemma 4.6 is the proof of equivalence between the norms and energies given respectively by Definitions 1.12 and 4.1.
Proposition 4.7
Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\). We assume that the a priori bound (4.13) are valid on \([\kappa ,T]\) for some sufficiently small \(\sigma '\). Then there exists a \(\kappa \)-independent constant \(C>0\) such that
4.2.1 Vector Field Classes \(\mathcal {P}\) and \(\bar{\mathcal {P}}\)
We now introduce a set of auxiliary, admissible vector fields associated with differential operators \(\mathcal {D}_i\) and \(\bar{\mathcal {D}}_i\) that allow us to circumvent coordinate singularities near the origin and to obtain high order estimates effectively. They are obtained by allowing \(\frac{1}{r}\) in addition to \(D_r\) whenever \(D_r\) appears in the chains of \(\mathcal {D}_i\) and \(\bar{\mathcal {D}}_i\). In other words,
for \(j\geqq 0\) and set \(\mathcal {P}_0=\{1\} \). Likewise, we define
for \(j\geqq 0\) and set \(\bar{\mathcal {P}}_0=\{1\} \). The properties of \(\mathcal {P}\) and \(\bar{\mathcal {P}}\) are presented in detail in “Appendix A”.
In what follows, we derive the bounds of \(\bar{\mathcal {P}}\) of various quantities involving \(\phi _{\mathrm{app}}\), \(\phi \), \(\phi +\Lambda \phi \) and so on that will be useful for the high-order energy estimates.
4.3 Pointwise Bounds on \(\phi _{\mathrm{app}}\)
Recall \(\phi _{\mathrm{app}}\) in (1.57).
Lemma 4.8
The following bounds hold true:
Proof
Let \(V\in \bar{\mathcal {P}}_i\) be given. By Lemma A.7 we have
where we have used Proposition 2.8 in the second line.
Recall that \(\Lambda \phi _{\mathrm{app}}= M_g \partial _\tau \phi _{\mathrm{app}}+ r\partial _r\phi _{\mathrm{app}}\). By Lemma A.7, definition (1.44) of \(M_g \), and the property (1.21) we obtain
where we have used the same argument as in the proof of (4.45) to obtain the second summand in the last bound above. \(\square \)
A simple consequence of Lemma 4.8 is the following corollary:
Corollary 4.9
The following bounds hold true:
where we recall that \(\delta ^*\) is given by (2.24).
Lemma 4.10
For any \(1\le i\le N\) we have
Proof
Proof of (4.51). By a simple calculation \( \Lambda ^2=M_g ^2\partial _{\tau \tau } + 2 r M_g \partial _{ r\tau }+ M_g \partial _\tau M_g \partial _\tau + r\partial _rM_g \partial _\tau + (r\partial _r)^2\). By the product rule \(V(M_g ^2\partial _{\tau \tau }\phi _{\mathrm{app}})\) can be written as a linear combination of expression of the form
Any such expression is bounded by \( r^{2n-i} \tau ^{\frac{2}{3}-2} = \tau ^{\frac{2}{3}} \left( \frac{r^n}{\tau }\right) ^2 r^{-i}\). A similar argument shows that \(\left| V ( r M_g \partial _{ r\tau }\phi _{\mathrm{app}})\right| \lesssim \varepsilon \tau ^{\frac{2}{3}+\delta }\frac{ r^n}{\tau }p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) r^{-i}\), \(\left| V (M_g \partial _\tau M_g \partial _\tau \phi _{\mathrm{app}})\right| \lesssim \tau ^{\frac{2}{3}}\frac{ r^{2n}}{\tau } r^{-i}\), \(\left| V ( r\partial _rM_g \partial _\tau \phi _{\mathrm{app}})\right| \lesssim \tau ^{\frac{2}{3}}\frac{ r^n}{\tau } r^{-i}\), \(\left| V ((r\partial _r)^2\phi _{\mathrm{app}})\right| \lesssim \varepsilon \tau ^{\frac{2}{3}+\delta } p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) r^{-i}\). Summing the above bounds we obtain (4.51). \(\square \)
Proof of (4.52)
Note that \(\Lambda \left( 3\phi _{\mathrm{app}}^2+2\phi _{\mathrm{app}}D \phi _{\mathrm{app}}\right) = 6\phi _{\mathrm{app}}\Lambda \phi _{\mathrm{app}}+2(\Lambda \phi _{\mathrm{app}})^2+2\phi _{\mathrm{app}}\Lambda ^2\phi _{\mathrm{app}}\). Using the product rule, bounds (4.45), (4.46), and (4.51) we obtain (4.52). \(\square \)
Proof of (4.53)
The proof is similar to (4.52). From (1.13) we have \( \Lambda \mathscr {J}[\phi _{\mathrm{app}}]=3\phi _{\mathrm{app}}^2\Lambda \phi _{\mathrm{app}}+2\phi _{\mathrm{app}}(\Lambda \phi _{\mathrm{app}})^2+\phi _{\mathrm{app}}^2 \Lambda ^2\phi _{\mathrm{app}}\). Now the statement follows from the product rule and bounds (4.45), (4.46), and (4.51). \(\square \)
Proof of (4.54)
We must use the chain rule. We note that \(V (\mathscr {J}[\phi _{\mathrm{app}}]^a)\) can be expressed as a linear combination of expressions of the form
We may use (4.45) and (4.50) to conclude that \(\left| W\mathscr {J}[\phi _{\mathrm{app}}]\right| \lesssim \tau ^2 q_1\left( \frac{r^n}{\tau }\right) r^{-j}\) for any \(W\in \bar{\mathcal {P}}_j\). Since \(\tau ^2 q_1\left( \frac{r^n}{\tau }\right) \lesssim \mathscr {J}[\phi _{\mathrm{app}}] \lesssim \tau ^2q_1\left( \frac{r^n}{\tau }\right) \), we can bound the above expression by \(\tau ^{2a}q_a\left( \frac{r^n}{\tau }\right) r^{-i}\). \(\square \)
Proof of (4.55)–(4.57)
The proof follows by the product rule (A.405) and (4.53), (4.54), (4.45), (4.46). \(\square \)
Proof of (4.58)
Recalling (1.47) it is easy to check that
We now apply the product rule (A.405) and bounds (4.54), (4.53), (4.45) and the estimate \(|w'|\lesssim r^{n-1}\) to conclude
since \(\tau \le 1\). Replacing \(r^{-(i+2)}\) by \(\tau ^{-\frac{i+2}{n}} \left( \frac{r^n}{\tau }\right) ^{-\frac{i+2}{n}}\) above, we obtain the claim, where in particular we use \(\gamma >1\). \(\square \)
4.4 Preparatory Bounds
Recall \(\phi = \phi _{\mathrm{app}}+ \tau ^m \frac{H}{r}\).
Lemma 4.11
For any \(1\leqq i\leqq N\), we have
Proof
Bound (4.60) follows directly follows from
Further more
where we have used the identities \(r\partial _r\left( \frac{H}{r}\right) = \partial _r H - \frac{H}{r}\) and \(\bar{\mathcal {D}}_i \partial _r H = \bar{\mathcal {D}}_i \left( D_r H-\frac{2}{r} H\right) = \mathcal {D}_{i+1}H - 2\bar{D}_i \left( \frac{H}{r}\right) \). For \((*)\), we first note that
For the first term, we use (A.401) to rewrite
Therefore we deduce that
It is easy to see that
Putting together the above bounds we obtain (4.61). \(\square \)
The same conclusions hold in Lemma 4.11 when we replace \(\bar{\mathcal {D}}_i\) by any \(V\in \bar{ \mathcal {P}}_i\).
Lemma 4.12
(High-order \(\phi \)-bounds). The following \(L^\infty \)-bounds hold:
The following \(L^2\)-bounds hold:
Proof
Note that from (4.60) and (4.49),
Therefore from (C.432), (C.435) and (C.436) we deduce (4.67)–(4.69). Bounds (4.70)–(4.71) follow from (C.429) and (C.430), where we use the bound
\(\square \)
Lemma 4.13
(High-order \(\phi +\Lambda \phi \)-bounds). The following \(L^\infty \)-bounds hold:
The following \(L^2\)-bounds hold:
Proof
From (4.61) and (4.46), we note that
Therefore, bounds (4.73)–(4.75) follow from (C.432)–(C.436). Bounds (4.76)–(4.77) follow from (C.429) and (C.430), respectively. \(\square \)
Finally, the key collection of a priori bounds is provided by the following lemma, and will be used repeatedly in our energy estimates in Section 5.
Lemma 4.14
Let \(a,b,c\in \mathbb {R}\), \(b<0\), \(c\le -b\), be given. For any \(i\in \{0,1\}\) we have
If \(2\le i\le N-1\) then
If \(2\le i\le N-3\) then
Finally, if \(i=N\) we have
Proof
By definition of \(\mathscr {J}[\phi ]\) we have
Applying the product and the chain rule, for any \(V\in \bar{\mathcal {P}}_i \), \(i\geqq 1\), \(V\left( \phi ^a\mathscr {J}[\phi ]^{b} \right) \) can be written as a linear combination of
where \(0\leqq p\leqq i\). In order to estimate \(V_j\phi \) and \(W_j (\phi +\Lambda \phi )\), it suffices to estimate \(\bar{\mathcal {D}}_i \phi \) and \(\bar{\mathcal {D}}_i (\phi + \Lambda \phi )\).
Let \(k_* = \max \{i_j, a_\ell \}\) in (4.84). Without loss of generality, we may assume that indices appearing in (4.84) are non-decreasing: \(i_1\leqq \cdots \leqq i_{j_m}\) and \(a_1\leqq \cdots \leqq a_{\ell _m}\). Then \(k_*=\max \{i_{j_m}, a_{\ell _m}\}\). \(\square \)
Proof of (4.79)
Bound (4.79) is obvious from (4.83) if \(i=0\). If \(i=1\) then the claim follows from (4.67) and (4.73). \(\square \)
Proof of (4.80)
If \(k_*=1\), by using (4.67) and (4.73), the expression in (4.84) is bounded by
and therefore, the worst bound occurs at \(p=k\) and the last line is bounded by
where we note that that \(\Vert w^{\alpha -N+2i+2}\Vert _{L^\infty } \lesssim 1\) since \(i\ge 2\) and \(N= \lfloor \alpha \rfloor +6\). Suppose that \(2\leqq k_*\leqq N-1\).
We first consider \(k_*=a_{\ell _m}\geqq i_{j_m}\). Let \(j_{m_0}+1\) be the first index for which \(i_{j_{m_0}+1}\geqq 2 \) so that \(i_j=1\) for \(j\leqq j_{m_0}\). In this case, we rearrange the w-weight in (4.84) as follows:
The goal is to estimate the last term \(w^{\frac{\alpha + 2a_{\ell _m} + 2 -N}{2}} \frac{ W_{\ell _m} (\phi +\Lambda \phi ) }{\phi + \Lambda \phi }\) in \(L^2\)-norm and all the remaining ones in \(L^\infty \). Note that
Therefore, the exponent of the first w in the second line is non-negative and therefore, that factor is bounded. Now all \(i_j\)’s and \(a_\ell \)’s except \(a_{\ell _m}\) cannot be bigger than \(N-3\), otherwise, it would contradict the definition of \(k_*\). Thus we can apply (4.68) and (4.74) to the first line above. Moreover, since \(2\leqq a_{\ell _m}\leqq N-1\), we can apply the weighted \(L^2\)-embedding (4.76) to the \(W_{\ell _m}\) term in the second line of the right-hand side of (4.86). By (4.79) \(\Vert q_c\left( \frac{r^n}{\tau }\right) \phi ^{a+2b} (\phi + \Lambda \phi )^{b} \Vert _{L^\infty } \le \tau ^{\frac{2}{3} a+b}\) since \(b+c\le 0\) by our assumptions. This gives the bound
The case \(k_*=i_{j_m} > a_\ell \) can be treated in the same fashion where we use (4.70) instead of (4.76). \(\square \)
Proof of (4.81)
In this case, since \(k_*\le N-3\) and as above we first consider the case \(k_*= a_{\ell _m}\). We then have
Note that \(\sum _{j={j_{m_0}} +1}^{j_m}(i_j-2)+\sum _{\ell =1}^{\ell _{m}}(a_\ell -1) \leqq i\) and therefore, the exponent of the first w in the next-to-last line above is non-negative as before. Using (4.67)–(4.68), (4.73)–(4.74), and the bound \(\Vert q_c\left( \frac{r^n}{\tau }\right) \phi ^{a+2b} (\phi + \Lambda \phi )^{b} \Vert _{L^\infty } \le \tau ^{\frac{2}{3} a+b},\) we can bound all the remaining factors to finally obtain (4.81). \(\square \)
Proof of (4.82)
When \(i=N\) and \(k_*\le N-1\) we may use the already proven (4.80) to infer that the \(\Vert \cdot \Vert _{\alpha +N+2}\)-norm of (4.84) is bounded by the right-hand side of (4.82). It now remains to discuss the case \(k_*=N\) in which case either 1) \(j_m=1\) and \(i_{j_m}=N\) (\(a_{\ell _m}=0\)) or 2) \(\ell _m=1\) and \(a_{\ell _m}=N\) (\(i_{j_m}=0\)). When \(a_{\ell _m}=N\), the expression (4.84) reads as
and therefore, by (4.77) and (4.79), we deduce
as claimed. When \(i_{j_m}=N\), the corresponding estimate reads as
where we have used (4.71). \(\square \)
We conclude the section with several a priori estimates that will be important for the energy estimates in Section 5.
Lemma 4.15
Recall \(g^{00}\) defined in (3.27). The following bounds hold:
Proof
It suffices to prove the bounds for \(Vg^{00}\) as the corresponding bound for \(V(\frac{1}{g^{00}})\) is a simple consequence of the chain rule (A.407) and the bound (4.35). From (3.27) and (3.19) it follows that for any \(V\in \bar{\mathcal {P}}_i\) with \(i\ge 1\) we have
In particular, if \(\ell _2\le i-1\) we may estimate
Using Lemma 4.14 now with \(c=2-\frac{2}{n} (<\gamma +1)\), estimates (4.90)–(4.92) follow easily. If \(\ell _2=i\) and additionally \(i\le N-1\) we may still run the same argument. If however \(\ell _2=N\), we lose \(\sqrt{\varepsilon }\) in (4.93) due to (4.82). \(\square \)
Since by (3.28) for any \(V\in \bar{\mathcal {P}}_i\) with \(i\ge 1\) we have
and by an analogous argument we have the following lemma:
Lemma 4.16
Recall \(g^{01}\) defined in (3.28). The following bounds hold:
5 Energy Estimates
To facilitate our proof and carry out the energy estimates, for the remainder of this section we assume that H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\), the a priori assumptions (4.13) hold, and the following (rough) bootstrap condition is true:
5.1 Estimates for \(\mathscr {L}_{\mathrm{low}}\)-terms
The goal of this section is the following proposition.
Proposition 5.1
Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) and the bootstrap assumption (5.1) hold. Then for any \(( \tau , r)\in [\kappa ,T]\times [0,1]\) the following bound holds:
5.1.1 Decomposition of \( \mathscr {L}_{\mathrm{low}} H \)
We rewrite the linear operator \(\mathscr {L}_{\mathrm{low}}\) in the form
where
Lemma 5.2
(Estimates for \(\mathscr {L}_{\mathrm{low}}^1\)). Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) and the bootstrap assumption (5.1) hold. Then
Proof
By the product rule (A.405)
We now use (4.58) and the \(L^2\)-embeddings (C.429) if \(\ell _2\ge 3\) and otherwise (C.431) to conclude
where we have used the bound \(w^{\alpha +i} \lesssim w^{\alpha +2i-N}\lesssim w^{\alpha +2\ell _2-N}\).
We now focus on the second term in the first line of (5.4).
By the product rule (A.405)
since \(\left| A_1\left( \frac{w}{g^2 r^2g^{00}}\right) \right| \lesssim r^{-2-\ell _1}\). Consider first Case I. \(\ell _3\le i-1\). By (4.52) the third line of (5.10) is bounded by
We now distinguish two cases.
Case I-1. \(\ell _3\ge \ell _4\). If \(\ell _3\le 1\) by Lemma 4.14 and (C.431) we then have
Therefore, since \(w^{\alpha +i}\lesssim 1\), by (2.23)–(2.24),
If \(2\le \ell _3\le i-1\), then in case \(\ell _4\ge 3\),
Recall that the total derivative number N is defined in (2.22). Since \(N+i-2\ell _3-2\ell _4+2 \ge N+i-2i+2= N-i+2\ge 0\) the \(L^\infty \)-norm of \(w^{\frac{N+i-2\ell _3-2\ell _4+2}{2}}\) is bounded. Moreover, by Lemma 4.14\(\Vert q_2\left( \frac{r^n}{\tau }\right) w^{\frac{\alpha -N+2\ell _3+2}{2}} A_3\left( \frac{\phi ^2}{\mathscr {J}[\phi ]^{\gamma +1}}\right) \Vert _{L^2 } \lesssim \tau ^{-\frac{2}{3}-2\gamma -\frac{\ell _3}{n}}\) and by (C.433) \(\Vert w^{\ell _4-2} A_4 H \Vert _{L^\infty }\lesssim \tau ^{\frac{1}{2}(\frac{14}{3}-\gamma )}(D^N)^{\frac{1}{2}}\). Plugging this into (5.13) we obtain the upper bound \( \varepsilon \tau ^{\delta ^*}(D^N)^{\frac{1}{2}}\) just like in (5.12). If on the other hand \(\ell _4\le 2\), we replace the \(L^\infty \)-bound of \(w^{\ell _4-2} A_4 H\) by an \(L^\infty \)-bound on \(A_4H\) provided by (C.431). This allows us to estimate the first line of (5.12) by
where we have used \(N+i-2\ell _3-2\ge N+i-2(i-1)-2=0\), Lemma 4.14, and (C.431).
Case I-2. \(\ell _3<\ell _4\). If \(\ell _4\le 2\) then we are in the regime that has already been discussed above. Assume \(\ell _4\ge 3\). If \(\ell _3\ge 2\) we use (4.81) and (C.429) to obtain
We have used the inequality \(i-2(\ell _3+\ell _4)+N\ge N-i\ge 0\). The case \(\ell _3\le 2\) is handled similarly, with (4.79) instead of (4.81).
Case II. \(\ell _3=i\). In this case we need to bound
with \(V\in \bar{\mathcal {P}}_i\). If \(i\in \{0,1\}\) we can use (4.79) and if \(2\le i\le N-1\) we may use (4.80) to bound \(\Vert V\left( \frac{\phi ^2}{\mathscr {J}[\phi ]^{\gamma +1}}\right) \Vert _{L^2 }\) and \(\Vert w^{\frac{\alpha -N+2i+2}{2}}V\left( \frac{\phi ^2}{\mathscr {J}[\phi ]^{\gamma +1}}\right) \Vert _{L^2 }\) respectively. The remaining terms are estimated in \(L^\infty \) and we conclude that the expression in (5.15) is bounded by \( \varepsilon \tau ^{\delta ^*} (D^N)^{\frac{1}{2}}\) just like above. If however \(i = N\) we must use (4.82). It then follows that the expression in (5.15) is bounded by
The 3rd-7th term in (5.4) are estimated analogously. Note that the terms \(\partial _\tau H\) and \(\frac{H}{\tau }\) and similarly \(\frac{\partial _\tau H}{\tau }\) and \(\frac{H}{\tau ^2}\) are on equal footing from the energy stand point or more precisely
where we recall the definitions (3.32)–(3.33) of \(E^N\) and \(D^N\). In particular, the estimates for the 3rd, 5th, and the 7th term in (5.4) are very similar and we sketch the details for the 7th (next-to-last) term. By the product rule (A.405) we have
A case-by-case analysis analogous to the one above, Lemma 4.14, and Lemmas C.3–C.5 yield
The same bound, with \(\frac{H_\tau }{\tau }\) replaced by \(\frac{H}{\tau ^2}\) follows analogously.
The 4-th and the 6-th term on the right-hand side of (5.4) are easier to bound. In the 6-th term the factor \(w'\) gives a regularising power of r near the center \(r=0\) due to the bound \(|\partial _r^a w'|\le r^{n-a-1}\) (which in turn follows from (1.19)). Similarly, the presence of \( \Lambda \mathscr {J}[\phi _{\mathrm{app}}]\) in the 4-th term, by virtue of (4.53) affords a power of \(\left( \frac{r^n}{\tau }\right) \) in our estimates, which again counteracts any potential singularities coming from the negative powers of r near \(r=0\). Routine application of Lemmas C.3–C.5 and Lemma 4.14 yields the desired bound.
To estimate the last line in (5.4) we first observe that
Therefore
We can therefore break up the last line of (5.4) into a sum of terms that are of similar structure as the ones showing up above, and thus the estimate follows analogously and thus obtain the same bound as in (5.16). \(\square \)
Lemma 5.3
(Estimates for \(\mathscr {L}_{\mathrm{low}}^2\)). Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) and the bootstrap assumption (5.1) hold. Then
Proof
We focus on the first and the most complicated term on the right-hand side of (5.5). Recall that \( r\partial _r\left( \frac{H}{r}\right) = D_r H- 3\frac{H}{r}\). By analogy to (5.10) we have
Case I-1. \(\ell _3=i\). In this case \(\ell _1=\ell _2=\ell _4=0\) and we note that
where we we have used (4.46). Therefore we bound the \(\Vert \cdot \Vert _{\alpha +i}\) norm of the last line of (5.20) by
If \(i=N\) by (4.82) we have
Since \(\frac{1}{2}(\gamma -\frac{2}{3})-\frac{1}{n}+\frac{2}{3}+\frac{1}{2}(\frac{11}{3}-\gamma )-2\gamma = \delta ^*+\frac{1}{n}-\frac{1}{2}\) (recall (2.24)), and \(\left( p_{1,0}\left( \frac{r^n}{\tau }\right) + p_{\lambda ,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) \right) \lesssim 1\), it follows that (5.22) is bounded by
as needed. If \(2\le i \le N-1\) we use (4.80) instead and if \(i=1\) we use (4.79) instead, to bound (5.22) by \(\varepsilon \tau ^{\delta ^*-\frac{1}{2}}(E^N)^{\frac{1}{2}}\).
Case I-2. \(\ell _4=i\). In this case \(\ell _1=\ell _2=\ell _3=0\) and \(A_4=\bar{\mathcal {D}}_i\). Using (5.21) and (4.79) we can bound the last line of (5.20) by
If \(i=N\) we have \(\bar{\mathcal {D}}_ND_r = \mathcal {D}_{N+1}\). Therefore, by (C.430), \(w^{\alpha +N+2}\lesssim w^{\alpha +N+1}\), and \(q_{-\frac{\gamma +1}{2}}\left( \frac{r^n}{\tau }\right) \lesssim 1\),
On the other hand, using (A.403), we also have
Plugging the last bounds into the last line of (5.24) and recalling (2.23) we bound it by
If \(2\le i \le N-1\) we use (C.429) instead of (C.430) above and obtain the upper bound \(\varepsilon \tau ^{\delta ^*-\frac{1}{2}}(E^N)^{\frac{1}{2}}\). Similarly, if \(i\le 2\) we may use (C.431) instead.
Case II. \(\ell _3,\ell _4\le i-1\). Recalling (4.46) and the bound \(|A_{1}(\frac{w}{g^{2}rg^{00}})|\lesssim r^{-1-\ell _{1}}\), by (5.21) we have
Case II-1. \(\ell _3\le \ell _4\le i-1\). If \(\ell _4\le 1\) and therefore \(\ell _3\le 1\), we can estimate the \(\Vert \cdot \Vert _{\alpha +i}\)-norm of the last line of (5.26) using (4.79) and (C.432) by
If \(2\le \ell _4\le i-1\), assume first that \(\ell _3\ge 2\). We rely on (C.429) and (4.81) to bound the \(\Vert \cdot \Vert _{\alpha +i}\)-norm of the last line of (5.26) by
where we have used the bound \(i+N-2(\ell _3+\ell _4+1)\ge 0\), which is true if \(\ell _3+\ell _4\le i-1\), to bound \( \Vert w^{\frac{i+N-2(\ell _3+\ell _4+1)}{2}}\Vert _{\infty }\) by a constant. If on the other hand \(\ell _3+\ell _4=i\), then \(\ell _1=0\) and therefore we have an additional power of w in our estimate which by the same idea as above allows us to obtain the bound (5.28).
If \(\ell _3\le 1\) we then use (4.79) instead of (4.81) and deduce the same bound analogously.
Case II-2. \(\ell _4\le \ell _3\le i-1\). This case is handled analogously to the case II-1 above and relies on a similar case distinction (\(\ell _4\ge 2\) and \(\ell _4\le 1\)) as well as Lemma 4.14 and estimates (C.431), (C.429).
This completes the bound of the first term on the right-hand side of (5.5). The estimates for the remaining 2 terms proceed analogously. Note that we use (4.55) crucially to estimate the third term on the right-hand side of (5.5). \(\square \)
5.2 High Order Commutator Estimates
The goal of this section is to establish the following proposition:
Proposition 5.4
Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) holds. Then
Lemma 5.5
(The commutator estimates). Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) holds. Then
where we remind the reader that the coefficients \(\zeta _{ij}\), \(i=1,\ldots , N\), \(j=0,\ldots i-1\) are defined in Lemma B.1.
Proof
Proof of (5.30). By (B.418) we have the formula
Since \(\left| \partial _r\left( \frac{g^{01}}{g^{00}} \right) \right| \lesssim \varepsilon \tau ^{\delta ^*-1}\) by Lemma 4.16, the bound \(w^{\alpha +i}\lesssim w^{\alpha +2i-N}\), and definitions (2.23)–(2.24) of \(\delta \) and \(\delta ^*\), we have
In order to bound the second term on the right-hand side of (5.33) we distinguish several cases by analogy to Lemma 5.2.
Case I: \(\ell _1\le \ell _2\). If \(\ell _2\le 2\) and therefore \(\ell _1\le 2\), we can use (C.431) and (4.95) to obtain
If \(3\le \ell _2\le N\) we again distinguish 2 cases. If \(\ell _1\ge 2\) we can use (C.429), (4.97), and (2.23)–(2.24) to obtain
If \(\ell _1=1\) we then use (4.95) instead of (4.97) and obtain the same conclusion.
Case II: \(\ell _1\ge \ell _2\). In this case we proceed analogously and rely crucially on Lemmas 4.16 and estimates (C.431)–(C.433). The only nonstandard situation occurs when \(\ell _1=N\). In that case \(\ell _2=0\) and we must use the bound (4.98) together with (C.431). We then obtain
To estimate the last term on the right-hand side of (5.33) we note that for any \(A_2\in \bar{\mathcal {P}}_{\ell _2}\), we have \(A_2D_r \in \mathcal {P}_{\ell _2+1}\) and since \(\ell _2\le i-2\) we are in the regime treated above. This concludes the proof of the bound for \(\Vert \left[ \mathcal {D}_i, \frac{g^{01}}{g^{00}} \partial _r\right] \partial _\tau H \Vert _{\alpha +i}\). The remaining 2 terms on the left-hand side of (5.30) are estimated analogously and their proofs rely crucially on Lemmas 4.15 and 4.16. The second term is less singular with respect to \(\tau \) and the presence of the \(g^{01}\) does not change the structure of the estimates due to Lemma 4.16. The third term contains the factor \(\frac{H}{\tau ^2}\) which, from the point of view of the energy, scales just like \(\frac{H_\tau }{\tau }\) and thus the structure of the estimates is similar to the above. \(\square \)
Proof of (5.31)
From (4.1) we have
By the commutator formula (B.420) we have
The second sum on the right-hand side of (5.39) can be estimated analogously to the estimates for (5.30) above, using (5.38). Thereby we observe that the total number of derivatives in the operator \(A_2D_r L_\alpha \) is at most i, since \(\ell _2\le i-3\). We next focus on the first term on the right-hand side of (5.39). Since \({\bar{\mathcal {D}}}_{i-2} D_r = \mathcal {D}_{i-1}\), using (5.38) we can write it as
By the product rule (A.405) we can isolate the top-order term
We now use (4.79), (C.430) and conclude, in the case \(i=N\)
where the estimate (4.79) has been used in the third line. When using (4.79), we first recall (3.19) and use the product rule to write
We note that by Lemma 4.15 and (1.19) we have \(\left| \frac{1}{g^2 g^{00}} \right| + \left| \partial _r\left( \frac{1}{g^2 g^{00}}\right) \right| \lesssim 1\) and therefore (4.79) yields the third line above. The remaining below-top-order terms can be estimated analogously to (5.30) to finally obtain (5.31). \(\square \)
Proof of (5.32)
Since \(\left| \partial _r^kw\right| \lesssim r^{n-k}\) for any \(k\in \{1,\ldots ,n\}\) it follows from (B.416) \(|\zeta _{ij}|\lesssim r^{n-j-2}\). Therefore by (4.79) for any \(j\le i\) we have
Therefore for any \(j\ge 3\) we have
where we have used (C.429) in the second line. If \(j\le 2\) we use (C.431) instead of (C.429) and obtain the same bound. \(\square \)
5.3 High-Order Estimates for \(\mathscr {M}[H]\)
We first recall \(K_a[\theta ]\), \(a\in \mathbb {R}\) in (3.3):
and also
Lemma 5.6
We have the following bound:
Remark 5.7
\( \tau ^{\frac{1}{2}(\frac{5}{3}-\gamma )} (E^N)^\frac{1}{2} \) can be replaced by \( \tau ^{\frac{1}{2}(\frac{8}{3}-\gamma )} (D^N)^\frac{1}{2} \) in the above bounds.
Proof
First we recall that (3.3) implies
which together, with (C.432) and \(\theta =\tau ^m\frac{H}{r}\), yields
or equivalently,
The representation (5.43) then gives (5.44) or equivalently
Next we evaluate \(\bar{\mathcal {D}}_j K_a [\theta ]\). We start with \(a=1\). By applying the product rule to (5.47) and using (C.432), (C.433) and (C.435), (4.45), (4.46), we deduce that
For general \(a\in \mathbb {R}\), let us write down the expression for \(\bar{\mathcal {D}}_j K_a [\theta ]\). For \(j=1\), using (3.11), we have
For \(j\geqq 2\), by applying the product rule and chain rule, we deduce that
which can be proved based on the induction argument on j. Therefore, we deduce (5.45). Also, by (C.429) we have (5.46). \(\square \)
Before we proceed with the estimates, we examine the structure of \(\mathscr {M}[H]\). Recall (4.8) and the formula (3.19) \(c[\phi ] = \frac{\phi ^4}{g^2\mathscr {J}[\phi ]^{\gamma +1}}\). Then
where we have used (4.1) and written \(L_\alpha H = - (1+\alpha ) w' D_r H - w\partial _r D_r H\). Our source of concern is rectangular bracket above, as it contains top-order terms with two derivatives falling on H (either through \(\partial _r(\mathscr {J}[\phi ]^{-\gamma -1})\) or \(\partial _r\mathscr {N}_0[H]\)) and seemingly insufficient number of w-powers to allow us to bound them through our w-weighted norms. This situation is a typical manifestation of the vacuum singularity at the outer boundary. Our key insight is that, due to special algebraic structure of the equation, the terms involving two spatial derivatives of H without the corresponding multiple of w will be cancelled. In the following lemma, we present the rearrangement of \(\mathscr {M}[H]\) that elucidates such an important cancelation.
Lemma 5.8
(Cancellation lemma):
-
(i)
\(\mathscr {M}[H]\) can be rewritten into the following form
$$\begin{aligned} \mathscr {M}[H] =&\varepsilon \gamma (\gamma +1) (1+\alpha ) w' \frac{\phi ^4 }{g^2 g^{00} } \mathscr {J}[\phi ]^{-\gamma -2}\partial _r \mathscr {J}[\phi _{\mathrm{app}}] D_r H \nonumber \\&- \varepsilon \gamma (1+\alpha )w' \partial _r \left( \frac{\phi ^4}{g^2 g^{00}} \right) \mathscr {J}[\phi ]^{-\gamma -1} D_r H\nonumber \\&\qquad - \varepsilon \gamma w \partial _r\left( \frac{c[\phi ]}{g^{00} }\right) \partial _r D_r H \nonumber \\&+ \varepsilon \frac{1}{\tau ^m g^{00}} \mathfrak {K}_4[\frac{\tau ^m H}{r}] + \varepsilon \left( \partial _r\left( \frac{1}{g^{00}}\right) + \frac{1}{g^{00}} \frac{2}{r} \right) \mathscr {N}_0[H], \end{aligned}$$(5.55)where
$$\begin{aligned} \mathfrak {K}_4[\theta ]&=-\gamma (1+\alpha ) w' \frac{\phi ^2}{g^2} K_{-\gamma -1}[\theta ]\big (\partial _r K_1[\theta ] -\phi ^2[ r\partial _r^2 \theta + 4\partial _r\theta ]\big ) \nonumber \\&\quad +(1+\alpha ) w' \frac{\phi ^2}{g^2} \big \{ -\gamma K_{-\gamma -1}[\theta ]\partial _r \mathscr {J}[\phi _{\mathrm{app}}] + \gamma \partial _r( \mathscr {J}[\phi _{\mathrm{app}}]^{-\gamma -1}) K_1[\theta ] \nonumber \\&\quad -\gamma (\gamma +1) K_{-\gamma -2}[\theta ] \partial _r \mathscr {J}[\phi _{\mathrm{app}}] \phi ^2[ r\partial _r\theta + 3\theta ] +2\gamma K_{-\gamma -1}[\theta ]\phi \partial _r \phi [ r\partial _r\theta + 3\theta ] \big \} \nonumber \\&\quad + (1+\alpha ) \partial _r \left( w' \frac{\phi ^2}{g^2} \right) \left( K_{-\gamma }[\theta ] + \gamma \mathscr {J}[\phi _{\mathrm{app}}]^{-\gamma -1} K_1[\theta ] +\gamma K_{-\gamma -1}[\theta ]\phi ^2[ r\partial _r\theta + 3\theta ] \right) \end{aligned}$$(5.56)and \(\mathscr {N}_0[H]=\frac{r}{\tau ^m} \mathfrak {K}_3[\frac{\tau ^m H}{r}]\) where \(\mathfrak {K}_3\) is defined by (3.17).
-
(ii)
Each expression in the right-hand side of (5.55) contains at most two spatial derivatives. If two spatial derivatives of H appear in the expression, they always contain a factor of w. In particular, the last bracket of the first line of \( \mathfrak {K}_4[\theta ]\) in (5.56) can be rewritten without any second spatial derivatives of H:
$$\begin{aligned}&\partial _r K_1[\theta ] -\phi ^2[ r\partial _r^2 \theta + 4\partial _r\theta ] \nonumber \\&\quad = \phi ^2 M_g \partial _r \partial _\tau \theta + \partial _r(\phi ^2M_g ) \partial _\tau \theta + 2\phi ( \Lambda \phi _{\mathrm{app}}+ r\partial _r \phi )\partial _r\theta \nonumber \\&\qquad +\left[ \partial _r (3\phi _{\mathrm{app}}^2 + 2\phi _{\mathrm{app}}\Lambda \phi _{\mathrm{app}}) +\partial _r(3\phi _{\mathrm{app}}+ \Lambda \phi _{\mathrm{app}})\theta \right] \theta \big \}. \end{aligned}$$(5.57)
Proof
To verity (5.55), we will first rewrite the rectangular bracket in (5.54). By (3.22) \(\mathscr {N}_0[H]=\frac{r}{\tau ^m} \mathfrak {K}_3[\frac{\tau ^m H}{r}]\) where \(\mathfrak {K}_3\) is defined by (3.17). Then
By using \( \partial _{r}K_{-\gamma }[\theta ]=-\gamma \mathscr {J}[\phi ]^{-\gamma -1}\partial _{r} K_{1}[\theta ]-\gamma K_{-\gamma -1}[\theta ]\partial _{r}\mathscr {J}[\phi _{\mathrm{app}}]\), we have
which in turn implies
where we have used \( r\partial _r\left( \frac{H}{r}\right) + 3\frac{H}{r} = D_r H\) and \(\theta = \frac{\tau ^m H}{r}\).
For the first term in the rectangular bracket in (5.54), we note
Together with (5.58), the rectangular bracket in (5.54) gives rise to the first line and the first term of the third line of (5.55). The following line of (5.54) corresponds to the second line of (5.55) where we have used \(L_\alpha H = - (1+\alpha ) w' D_r H - w\partial _r D_r H\).
Finally we will count the number of spatial derivatives and the weight w. First of all, it is clear that all the terms appearing in (5.55) contain at most two spatial derivatives of H. For instance, the first term in (5.55) does not contain the second derivatives of H. In the second line, both terms contain the second derivatives of H and they have a factor of w. Note that \(\partial _r g^{00}\) has a term involving two derivatives (see (3.27)) but that comes with w. The same counting applies to the rest. The only expression that is not obvious at first sight is the first line of (5.56) because we do see the two spatial derivatives of H without the weight w. It turns out that those second derivatives disappear after cancelation. A direct computation using (5.47) yields the identity (5.57). \(\square \)
Lemma 5.9
Let H be a solution to (3.26) on a time interval \([\kappa ,T]\) for some \(T\le 1\) and assume that the a priori assumptions (4.13) holds. Then
Proof
We note that the terms in the first two lines of (5.55) have the similar structure as the terms resulting from \(D_r (\frac{1}{g^{00}} \mathscr {L}_{\mathrm{low}}^2 H)\) in terms of the highest order derivative count and the weight w count. For instance, the first line of (5.55) is comparable to the case when the derivative falls into w of the last term of (5.5). The difference is whether the coefficients are set by \(\phi _{\mathrm{app}}\), \(\mathscr {J}[\phi _{\mathrm{app}}]\) or \(\phi \), \(\mathscr {J}[\phi ]\), but the coefficients enjoy similar bounds due to Lemmas 4.84.10 for \(\phi _{\mathrm{app}}\), Lemma 4.14 for \(\phi \) and our a priori assumption (5.1). We therefore have
where \(\mathscr {M}_{12}\) denotes the first two lines of (5.55).
We focus on the last line of (5.55) and present the detail for the bound on \(\varepsilon \tau ^{-m}\bar{\Vert }\bar{\mathcal {D}}_{i-1}(\frac{1}{g^{00}} \mathfrak {K}_4[\frac{\tau ^m H}{r}])\Vert _{\alpha +i}\). We restrict our attention first to the following term coming from the first line of (5.56):
As shown in the previous lemma, the identity (5.57) assures that \(\partial _r K_1[\theta ] - \phi ^2[ r\partial _r^2 \theta + 4\partial _r\theta ]\) contains at most one spatial derivative of H and therefore no issues associated with the w-weights near the boundary will occur.
We proceed with \(\bar{\mathcal {D}}_{i-1} (\star )\) for \(1\leqq i\leqq N\). By the product rule, \(\bar{\mathcal {D}}_{i-1} (\star )\) can be written as a linear combination of the following form:
where \(A_1\in \bar{\mathcal {P}}_{\ell _1}\), \(A_2\in \bar{\mathcal {P}}_{\ell _2}\), \(A_3\in \bar{\mathcal {P}}_{\ell _3}\), \(\ell _1+\ell _2+\ell _3=i-1\). As before, we divide into several cases. If \(\ell _k\leqq 2\) for all \(k=1,2,3\), all the indices are low and we just use \(L^\infty \) bounds (4.67), (4.90), (4.92), (5.45). In the following, we assume that at least one index is greater than 2.
\(\underline{\hbox {Case I: } \ell _3\geqq \max \{\ell _1,\ell _2\}}\). In this case, \(3\leqq \ell _3\leqq i-1\). Since \(\ell _1,\ell _2 \leqq \frac{N-1}{3}\leqq N-4\), and we apply \(L^\infty \) bounds for \(A_1\) and \(A_2\) factors and \(L^2\) bounds for \(A_3\) factor. In particular, by assuming \(\ell _1\geqq 1\) (the case of \(\ell _1=0\) follows similarly), we arrange the w weights as follows:
By the product rule and by using (4.67), (4.68), (4.90), (4.92), we deduce that
and, by further using (5.45), that
We have derived so far that
We claim that
Note that from (5.57) we may rewrite \(r\{ \partial _r K_1[\theta ] - \phi ^2[ r\partial _r^2 \theta + 4\partial _r\theta ] \}\) as
Apply \(A_3\) to the above. We focus on the first term which can be written as
for \(A_{31}\in {\bar{\mathcal {P}}}_{\ell _{31}}\) and \(A_{32}\in {\bar{ \mathcal {P}}}_{\ell _{32}}\) where \(\ell _{31}+\ell _{32} = \ell _3 \leqq i-1\). As previously done, depending on the size of \(\ell _{31}, \ell _{32}\), we may use \(L^\infty \) and \(L^2\) bounds. We verify the claim (5.64) when \(\ell _{31}=0\) and \(\ell _{32}=\ell _3\). Note that
for \(B\in \mathcal {P}_{\ell _3+1}\). Now we have \(w^{\alpha +i-2(\ell _1+\ell _2)}= w^{\alpha + 2 (\ell _3+1) -N} w^{N+i - 2(\ell _1+\ell _2+\ell _3+1)}\leqq w^{\alpha + 2 (\ell _3+1) -N}\) since \(\ell _1+\ell _2+\ell _3 +1=i\) and \(i\leqq N\) and hence by the definition of \(E^N\) and \(L^2\) embedding, we obtain
which gives (5.64). Other terms can be estimated similarly.
Therefore we deduce that
\(\underline{\hbox {Case II: } \ell _2\geqq \max \{\ell _1,\ell _3\}}\). In this case, \(3\leqq \ell _2\leqq i-1\) and \(\ell _1,\ell _3 \leqq \frac{i-1}{3}\). We apply \(L^\infty \) bounds for \(A_1\) and \(A_3\) factors and \(L^2\) bounds for \(A_2\) factor. We arrange the w weights as follows:
We have the same bound for \(A_1\) factor as in (5.62). For \(A_3\) factor, from (5.65), we deduce that
It suffices to estimate
Using (5.53) for \(a=-\gamma -1\), we have the expression
Following the case-by-case analysis as before and using (5.44), (5.45), (5.46), Lemma 4.14 and (4.54), we deduce that
Therefore we obtain the same bound as Case I
where we have used (4.91).
\(\underline{\hbox {Case III: } \ell _1\geqq \max \{\ell _2,\ell _3\}}\). In this case, \(3\leqq \ell _1\leqq i-1\) and \(\ell _2,\ell _3 \leqq \frac{i-1}{3}\). We apply \(L^\infty \) bounds for \(A_2\) and \(A_3\) factors and \(L^2\) bounds for \(A_1\) factor. For \(L^2\) bounds for \(A_1(\frac{1}{g^{00}})\), we use Lemma 4.15. The proof follows in the same fashion and we get the same bound as in the previous cases.
All the other terms in (5.56) are estimated analogously and we have the following bound:
The last term in (5.55) can be estimated similarly by using Lemma 4.15, (5.58), and the previous estimates on \(\mathfrak {K}_4[\theta ]\):
This finishes the proof Lemma. \(\square \)
5.4 Nonlinear Estimates
Before we formulate the main estimate in Proposition 5.11, we collect several identities that can be regarded as a special form of the product rule that connects the algebraic structure of the nonlinearity to the algebraic properties of the vector field class \(\mathcal {P}\).
Lemma 5.10
For any \(i\in \{0,\ldots ,N\}\) there hold the identities
where \(a_i^{A_1A_2},b_i^{A_1A_2A_3}, c_k^{iBC}\) are some universal real constants. Note that the operators \(A_j\), \(j=1,2,3\) are at most of order i.
Proof
Proof of (5.73). The proof is based on the induction on i. Note that
Let \(i=1\). Then
Since both \(\partial _r \left( \frac{\cdot }{r} \right) \) and \(\partial _r D_r \) belong to \( \mathcal {P}_{2}\), the claim is true for \(i=1\). Now suppose the claim is true for all \(i\leqq \ell \) and let
If \(\ell \) is even,
Note that each term in the summation belongs to \(\mathcal {P}_j\) for some j. Therefore, the claim is true for \(i=\ell +1\). If \(\ell \) is odd, we can rearrange terms as follows:
which shows that the claim is true for \(i=\ell +1\). The proofs of (5.74)–(5.75) are similar. \(\square \)
Proposition 5.11
(Estimates for the nonlinear term). Let H be a solution of (3.21). Then for any \(i\in \{0,1,\ldots ,N\}\) the following bound holds:
Since \(\phi =\phi _{\mathrm{app}}+\theta \), we have by simple algebra
From (3.25) we may write \(\mathscr {N}[H]\) in the form
Using (3.16), the first term on the right-hand side of (5.78) takes the form
We denote the first and second terms of the right-hand side of (5.79) by \(\mathscr {N}_1[H]\) and \(\mathscr {N}_2[H]\). We first present the estimation of \(\mathscr {N}_1[H]\) and \(\mathscr {N}_2[H]\).
5.4.1 Estimates for \(\mathscr {N}_1[H]=2\gamma \tau ^m w\frac{\phi ^3 }{g^2\mathscr {J}[\phi ]^{\gamma +1} } \frac{1}{r} \left( r\partial _r \left( \frac{H}{r}\right) \right) ^2\)
Lemma 5.12
For each \(1\leqq i\leqq N-1\), we have the following:
For \(i=N\),
Proof
Using the product rule (A.405) we have
where we recall the notation \(\mathscr {G} =\frac{1}{r} \left( r\partial _r \left( \frac{H}{r}\right) \right) ^2\) from the proof of Lemma 5.10.
\(\underline{\hbox {Case I: } \ell =0}\). First we have \(C\mathscr {G}=\mathscr {G}\) and \(\mathscr {G}=\partial _r (\frac{H}{r}) (D_r H -3\frac{H}{r})\). By using \(L^\infty \) bound (C.431), we have
\(\underline{\hbox {Case I-1: } \ell =0 \hbox { and } 0\le k\le 1}\). By (4.79) \(|A\left( \frac{w}{g^2}\right) B\left( \frac{\phi ^3 }{\mathscr {J}[\phi ]^{\gamma +1}}\right) |\lesssim \tau ^{-2\gamma }q_{-\gamma -1}\left( \frac{ r^n}{\tau }\right) \). Therefore, recalling (2.24) and the above definition of \(I^{iABC}\) we obtain
s
\(\underline{\hbox {Case I-2: } \ell =0, k\geqq 2}\). If \(j=0\) and \(k=i\le N-1\) we use (4.80) to conclude
If \(j=0\) and \(k=i=N\) we use (4.82) instead of (4.80) and thanks to an additional power of w, this leads to
If \(j\ge 1\) we have \(k\le N-1\) and \(\left| A\left( \frac{w}{g^2}\right) \right| \lesssim r^{n-j} \lesssim \tau ^{1-\frac{j}{n}}q_{1-\frac{j}{n}}\left( \frac{r^n}{\tau }\right) .\) Using the bound analogous to (5.84) we conclude
\(\underline{\hbox {Case II: } \ell \geqq 1}\). In this case, we will make use of the representation obtained in (5.73): for \(C\in \mathcal {P}_\ell \), we write it as
Let \(\ell _*=\max \{ q+1, \ell -q+2 \}\). Without loss of generality, we may assume that \(\ell _*= \ell -q+2\) so that \(C_2\in \mathcal {P}_{\ell _*}\) and \(C_1\in \mathcal {P}_{\ell -\ell _*+3}\). Note that \(\frac{\ell +3}{2}\leqq \ell _*\leqq \ell +1\) and \(1\leqq q=\ell -\ell _*+2\leqq \frac{\ell +1}{2}\).
\(\underline{\hbox {Case II-1: } \ell \geqq 1, j=0 \hbox { and } k=0}\). We first consider \(\frac{N-\alpha }{2}\leqq \ell _*\leqq N\). In this case, by (4.79) and thanks to an additional power of w,
where we note that since \(p+\ell _*=\ell +2\), \(\frac{ N+i -2\ell _*+2 }{2}-q+1 = \frac{N+i - 2\ell }{2} \geqq 0\), and thus \(\Vert w^{\frac{ N+i -2\ell _*+2 }{2}-q+1} \Vert _{L^\infty }\lesssim 1\).
Furthermore, since \(\ell -\ell _*+1=q-1 \leqq \frac{\ell -1}{2} \leqq N-4\) due to \(N=\lfloor \alpha \rfloor +6\geqq 9\), \(\tau ^{\frac{1}{2} (\gamma -\frac{11}{3})} w^{\frac{ N+i -2\ell _*+2 }{2}-q+1} w^{q-1}C_1H\) is bounded by \((E^N)^{\frac{1}{2}}\) via (C.433). Finally, we used the \(L^2\) embedding (C.429) to bound \(\Vert w^{\frac{\alpha +2\ell _*-N}{2}} C_2H \Vert _{L^2}\).
Now suppose \(\ell _*=N+1\) (\(\ell =N\) and \(q=1\)). We then have
We estimate the \(L^2\)-norm of the above expression by estimating the first line in \(L^\infty \) norm and the second line in the \(L^2\)-norm. Recalling (2.22), by (C.430), (4.79), (C.431) we obtain
The only remaining case is when \(\ell _*<\frac{N-\alpha }{2}\), namely \(\ell _*=2\) and \(\ell =1\). In this case, we can just use the \(L^\infty \) bound (C.433) to derive the same bound as in (5.87).
\(\underline{\hbox {Case II-2: } \ell \geqq 1, j=0 \hbox { and } k\geqq 1}\). In this case, \(2\leqq \ell _*\leqq i\) and \(k\leqq N-1\) since \(\ell +k=i\le N\). If \(\ell _*\ge k\) we have
where we have used (4.81) and the embeddings (C.429) and (C.435). Moreover, \(\Vert w^{\frac{ N+i -2k-2q+2 -2 \ell _*}{2}}\Vert _{L^\infty }\lesssim 1\) since \(N+i -2k-2q+4 -2 \ell _*= N-i\ge 0\).
If \(\ell _*< k\), as in Case I-2, we estimate \(w^{\frac{\alpha +N-2k}{2}}B\left( \frac{\phi ^3}{\mathscr {J}[\phi ]^{\gamma +1}}\right) \) in the \(L^2\)-norm and the appropriately weighted terms \(C_1H\) and \(C_2H\) in the \(L^\infty \)-norm and obtain the same bound as in (5.91).
\(\underline{\hbox {Case II-3: } \ell \geqq 1 \hbox { and } j\geqq 1}\). In this case, we have \(2\leqq \ell _*\leqq i+1-j\), \(k\leqq i-j-1\) and \(\ell +k= i-j\). If \(k=0\) we proceed as in Case II-1:
where we have used \(N+i -2\ell _*- 2q+2= N+i -2\ell -2 \geqq 0\) because \(\ell = i-j \leqq i-1\). If \(k \geqq 1\). We proceed as in Case II-2. We distinguish the two cases \(\ell _*\ge k\) and \(\ell _*<k\). Proceeding analogously to Case II-2, relying on the embeddings (C.429) and (C.435), and Lemma 4.14 we conclude
\(\square \)
5.4.2 Estimates for \(\mathscr {N}_2[H]= 2\gamma \tau ^m w\frac{\phi ^3 }{g^2\mathscr {J}[\phi ]^{\gamma +1}} \frac{M_g }{ r^2} \left( \partial _\tau H + \frac{m}{\tau } H \right) r\partial _r \left( \frac{H}{r}\right) \)
Lemma 5.13
For any \(i\in \{0,1,\ldots ,N\}\) the following bound holds:
Proof
By the product rule (A.405) and the identity \( r\partial _r\left( \frac{H}{r}\right) =D_r H - 3\frac{H}{r}\) we have
Case I. \(\ell _3\le i-1\). Each factor in the last line of (5.95) can be estimated by
We now distinguish several cases.
Case I-1. \(\ell _3=\max \{\ell _2,\ell _3,\ell _4\}\). Assume first \(\ell _2\le \ell _4\le \ell _3\) and \(\ell _2\ge 2\). In this case the \(\Vert \cdot \Vert _{\alpha +i}\) norm of (5.96) is bounded by
where we have used (4.81) to bound \(\Vert w^{\ell _2} q_1\left( \frac{r^n}{\tau }\right) A_2\left( \frac{\phi ^3 }{\mathscr {J}[\phi ]^{\gamma +1}}\right) \Vert _{\infty }\), (C.433) to bound \( \Vert w^{\ell _4-2}(A_4H_\tau -\frac{m}{\tau }A_4H)\Vert _{\infty } \) and (C.429) to bound \(\Vert w^{\frac{\alpha +2\ell _3-N}{2}}A_3\left( D_r H - 3\frac{H}{r}\right) \Vert _{L^2}\). Note that we have used the bounds \(\tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert \mathcal {D}_j H_\tau \Vert _{\alpha +j}+\tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert \mathcal {D}_j H\Vert _{\alpha +j} \lesssim (E^N)^{\frac{1}{2}}\), \(j\in \{0,1,\ldots , N\}.\) The case \(\ell _4\le \ell _2\le \ell _3\) is handled analogously.
If \(\ell _2\le 1 \) and \(\ell _3\ge 3\) we then use (4.79) instead of (4.81) above and obtain the same bound. If \(\ell _2\le 1\) and \(\ell _3\le 2\) we then use (4.79) and (C.431) instead of (C.433) in the aboive argument and obtain the same upper bound.
Case I-2. \(\ell _4= \max \{\ell _2,\ell _3,\ell _4\}\) or \(\ell _2 =\max \{\ell _2,\ell _3,\ell _4\}\). These cases can be treated similarly, with a similar case distinction, and with help of Lemma 4.14, (C.431)–(C.433) and (C.429).
Case II. \(\ell _3=i\). If \(i\le N-1\) we proceed as in Case I. Assume now \(\ell _3=N\). Since in this case \(A_3 = \bar{\mathcal {D}}_N\) the last line of (5.95) takes the form
We take special notice of the additional power of w available in this case. Since \(\left| \frac{M_g }{g^2 r^2}\right| \lesssim \tau ^{1-\frac{2}{n}}p_{1,-\frac{2}{n}}\left( \frac{r^n}{\tau }\right) q_1\left( \frac{r^n}{\tau }\right) \) we can estimate the \(\Vert \cdot \Vert _{\alpha +N}\)-norm of the above expression by
where we have used (4.79), (C.430), the bound \(\tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert \mathcal {D}_j H_\tau \Vert _{\alpha +j}+\tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert \mathcal {D}_j H\Vert _{\alpha +j} \lesssim (E^N)^{\frac{1}{2}}\), \(j\in \{0,1,\ldots , N\},\) and (A.403). The proof follows from (5.97) and (5.99). \(\square \)
The third, fourth, and fifth lines on the right-hand side of (5.79) are easily bounded by the same ideas as above, where we systematically use the product rule (A.405), Lemmas 4.14 and 4.10.
We only highlight the potential difficulties and how they can be overcome. In the 3rd term on the right-hand side of (5.79) there is nothing dangerous; we may write it as \(\varepsilon \gamma \tau ^{m} w\frac{\phi ^2}{\mathscr {J}[\phi ]^{\gamma +1}}\frac{1}{g^1} \Lambda (3\phi _{\mathrm{app}}+ \Lambda \phi _{\mathrm{app}})H \frac{H}{r} \) and then estimate its \(\mathcal {D}_i\) derivative using the case-by-case analysis analogous to the above, the product rule (A.405), Lemma 4.14, and the bounds (4.46), (4.51). Note that the last two estimates afford the presence of a power of \(\frac{r^n}{\tau }\) in our bounds, which in turn has the regularising effect of diminishing any potential singularities due to negative powers of r at \(r=0\).
The 4-th term on the right-hand side of (5.79) looks potentially dangerous due to the presence of the negative powers of r in \(\frac{H^2}{r^2}\) and \(\frac{H^3}{r^3}\). However, by (4.53) the bounds on \(V \Lambda \mathscr {J}[\phi _{\mathrm{app}}]\), \(V\in \bar{\mathcal {P}}_i\), will afford a presence of a power of of \(\frac{r^n}{\tau }\), thus averting all difficulties with potential singularities at \(r=0\).
Finally, the 5th term on the right-hand side of (5.79) contains \(w'\) explicitly, and since \(|\partial _r^\ell w'|\lesssim r^{n-\ell -1}\) in the vicinity of \(r=0\) we have the above mentioned regularising effect. The estimates are then routinely performed using the the product rule (A.405) and Lemma 4.14. The outcome is
To estimate the last line of (5.79) the crucial insight is that
which follows from (5.43) with \(a=-\gamma -1\). Therefore the left-hand side above is in fact quadratic in \(K_1[\theta ]\). We now estimate the high-order derivatives of the above left-hand side using the product rule (A.405), Lemma 5.6, and Remark 5.7. By analogy to the proof of Lemma 5.9, we obtain
On the other hand, when \(i=N\), \(\mathcal {D}_NK_1[\theta ]\) contains a top-order term \(\mathcal {D}_{N+1}H\) in which case we have to use (C.430) with loss of \(\sqrt{\varepsilon }\):
We next discuss the rest of terms in (5.78). Using the product rule (A.405) and (5.74) we have
With this decomposition, Lemma 4.14 (applied with \(b=0\)), Lemma 4.8, and the case-by-case analysis analogous to the proof of Lemma 5.13 we obtain the bound
where we have used \(\frac{N+2}{n}= 2(\frac{4}{3}-\gamma )-\delta ^*\) at the last step. Similarly,
In order to estimate \( \varepsilon \tau ^m \frac{P[\phi _{\mathrm{app}}] H^2}{\phi _{\mathrm{app}}^2 r}\) (the last term on the right-hand side of (5.78)) Using the product rule (A.405)
For \(A_3 \left( \frac{H^2}{r}\right) \), we may use (5.74) to further decompose it into a linear combination of \(A_{31}H A_{32} H\) where \(A_{31}, A_{32} \in \mathcal {P}_{i_1},\mathcal {P}_{i_2} \), \(i_1+i_2=\ell _3+1\), \(i_1,i_2\le \ell _3\). We next recall (4.59). Applying (4.58) and the case-by-case analysis analogous to the proof of Lemma 5.13 we obtain the bound
Proof of Proposition 5.11
Bound (5.76) follows from (5.79), Lemmas 5.12–5.13, bounds (5.100)–(5.102), (5.104)–(5.105), and (5.107). \(\square \)
5.5 Source Term Estimates
Recall the definition (2.2) and formula (2.21).
Lemma 5.14
(Source term estimates). For any \(i\in \{0,1,\ldots N\}\) the following bounds hold:
Proof
Proof of (5.108). Recall that \(R^\varepsilon _{M,2}\) is defined through (2.7). A detailed look at the Taylor expansion of the function \(R^\varepsilon _{M,\nu }\) reveals that for any \(D\in \mathbb {N}\) there exist constants \(c_{\alpha _1,\ldots \alpha _M}^j\), \(j\in \{1,\ldots , D\}\) and a smooth function \(r_{M,\nu }^{D,\varepsilon }\) such that
where the remainder term \(r^{D,\varepsilon }_{M,\nu }(x_1,\ldots x_m)\) has the property that all mixed derivatives \(\partial _{x_1}^{\alpha _1}\ldots \partial _{x_M}^{\alpha _m}r^{D,\varepsilon }_{M,\nu }\) vanish at \(\mathbf{0}\) if \(\sum _{i=1}^Mi \alpha _i\le M+D\). Using the chain rule, (5.111), and the bound
the bound (5.108) follows immediately. \(\square \)
Proof of (5.109)
Recall that \(R_P^\varepsilon \) is defined through (2.17). To estimate the first term on the right-hand side of (2.17) we use the following crude bound:
By (2.61), Proposition 2.8 we can bound the last line above by
To estimate the second term on the right-hand side of (2.17) we first note that
When we apply \(\mathcal {D}_i\) to the first term on the right-hand side of (5.113) we use the product rule (A.405) to break down the resulting expression into a linear combination of terms of the form
with \(A_1\in \mathcal {P}_{\ell _1}\) and \(A_{2,3}\in \bar{\mathcal {P}}_{\ell _2,\ell _3}\) with \(\ell _1+\ell _2+\ell _3 = i\). By Proposition 2.8 we have
Combining the previous line with (4.53) and (4.54) we can bound the absolute value of (5.114) by
To bound the last term on the right-hand side of (2.17) we can use the refined expansion (5.111) to obtain the bound
where we have used (2.41), (2.39), (2.8), and (2.13), and the bound \(|R_\mathscr {J}|\lesssim \tau ^{M\delta }\). By an analogous argument we have the bound \(\left| (r\partial _r)^\ell h_M\right| \lesssim \tau ^{(M+1)\delta }\). Using the last two bounds, the product rule, Proposition 2.8, and by analogy to the above, we obtain
Since \(\tau ^{\frac{4}{3} - 2\gamma +(M+1)\delta -\frac{i+1}{n}}=\tau ^{-\frac{4}{3}+(M+1)\delta -\frac{i-1}{n}}\), the claim follows from (5.112), (5.115), and (5.116). \(\square \)
Proof of (5.109)
Since \(\mathscr {S}(\phi _{\mathrm{app}}) = r\tau ^{-m} S(\phi _{\mathrm{app}})\), from (2.6) and (5.108)–(5.109) we obtain
where we have used the bound \(\tau ^{-\frac{i-1}{n}}\le \tau ^{-\frac{i}{n}}\), for \(\tau \in (0,1]\). \(\square \)
As a corollary, we obtain the following bound for the source terms:
Proposition 5.15
(Source term estimates). Let H be a solution of (3.21). Then for any \(i\in \{0,\ldots , N\}\) the following bound holds:
Proof
By the previous lemma and the Cauchy–Schwarz inequality we obtain
Since \(i\le N\) and \(\delta ^*= \delta - \frac{N}{n}\), we can estimate the above expression by a multiple of
\(\square \)
Remark 5.16
In order for the \(\tau \)-power to be integrable on [0, T] we need to impose \(2(1-\frac{2}{n})+ (2M-2)\delta + 2\delta ^*-3\gamma -2m>-1\) which is equivalent to \((M-1)\delta +\delta ^*>\frac{3}{2}(\gamma -1)+m+\frac{2}{n}\). Since \(\delta ^*>\frac{2}{n}\) by (2.27) and \(0<\gamma -1<\frac{1}{3}\), a sufficient condition for the previous estimate is for M to be sufficiently large so that
5.6 Proof of Theorem 1.13
We are ready to estimate \( \int \nolimits _\kappa ^\tau \mathcal {R}_i\,\mathrm{d}\tau ' \) where \(\mathcal {R}_i\) is given in (4.12). The only missing estimate is the last term of (4.12). By (4.23), (4.24), and (4.37), we have \(|\left( \frac{c[\phi ]}{c[\phi _0]g^{00}}\right) _\tau |\lesssim \varepsilon \tau ^{\delta -1} \) and hence we obtain the following estimate of the last term of \(\mathcal {R}_i\) in (4.12):
Combining Propositions 5.1, 5.4, Lemma 5.9, Propositions 5.11, 5.15, and (5.120), we obtain the bound
We note that the last line of (5.121) Let \(\bar{\delta }: = \min \{\delta ^*, \frac{\delta }{2}\}>0\). With the choices
we have \(m-\frac{1}{2}+\frac{5}{2}\left( \frac{4}{3}-\gamma \right) \ge 0\), \(m+\delta ^*-\frac{3}{2}-\frac{3}{2}(\frac{4}{3}-\gamma )\ge 0\), \(2(1-\frac{2}{n})+ (2M-2)\delta + 2\delta ^*-3\gamma -2m\ge 0\) (for the last bound we use (2.27) which implies \(\delta ^*>\frac{2}{n}\)). Consequently, bound (5.121) together with the a priori assumption \(E^N\le 1\) implies
where we have used the bound \(2|ab|\le a^2+b^2\) to go from the first to the second estimate and the a priori bound \(E^N\lesssim 1\).
We now integrate the energy identity (4.11) over the time interval \([\kappa ,\tau ]\), \(\tau \le 1\), and obtain by virtue of Proposition 4.7 conclude that there exists a universal constant \(C_0>2\) such that
The positivity of \(\delta \) and \(\bar{\delta }=\min \{\delta ^*, \frac{\delta }{2}\}\) guarantees that the last time integral on the right-most side of (5.124) is finite and bounded independently of the constant \(\kappa \). As a consequence of (5.124) and Proposition 4.7 we conclude
Since by the local well-posedness theorem Proposition D.1, the map \(\tau \mapsto S_\kappa ^N(\tau )\) is continuous, a standard continuity argument applied to (5.125) implies that there exist \(0<\sigma _*,\varepsilon _*<1\) such that for any \(0<\sigma <\sigma _*\) the following is true: for any choice of initial data \((H,H_\tau )\big |_{\tau =\kappa }\) satisfying
and any \(0<\varepsilon <\varepsilon _*\) the solution exists on the interval \([\kappa ,1]\) and satisfies the uniform-in-\(\kappa \) bound
Justification of the a priori assumptions (4.13) and (5.1). The size restrictions \(0<\varepsilon <\varepsilon _*\), \(0<\sigma <\sigma _*\) for \(\varepsilon _*,\sigma _*\) sufficiently small are necessary to ensure that the a priori assumptions (4.13) and (5.1) can be consistently recovered from the standard continuity argument. First let \((\ell _1,\ell _2)\ne (0,2)\). The embedding inequality (C.432) immediately gives
for \( 0\leqq \ell _1+\ell _2\leqq 2, \ (\ell _1,\ell _2)\ne (0,2)\). Now for \((\ell _1,\ell _2)= (0,2)\), it suffices to derive the bound for \(\Vert \tau ^2\partial _\tau ^2(\frac{H}{r})\Vert _\infty \). Since \((H,\partial _\tau H)\) is a classical solution to (3.26), we may use the equation directly:
Using (5.127), (4.35), (4.38) it is easy to see that the first three terms of the right-hand side are bounded by \(\tau ^{\frac{1}{2}(\frac{11}{3}-\gamma )} (E^N)^\frac{1}{2}\). For the fourth term, by (4.30), \( |\tau ^2 c[\phi ] |\lesssim \tau ^{\delta +\frac{2}{n}}q_{-\gamma -1} (\frac{r^n}{\tau })\) and moreover by (C.426), we have \(\Vert w \frac{ \mathcal {D}_2 H }{r} \Vert _\infty \lesssim \tau ^{\frac{1}{2}(\frac{11}{3}-\gamma )} (E^N)^\frac{1}{2} \). Hence, the fourth term is also bounded by \(\tau ^{\frac{1}{2}(\frac{11}{3}-\gamma )} (E^N)^\frac{1}{2}\). Similarly, by using the source estimates and estimating \(L^\infty \) norm of \(\frac{\mathscr {N}_0[H]}{r}\), \(\frac{ \mathscr {L}_{\mathrm{low}} H}{r}\) and \(\frac{\mathscr {N}[H]}{r}\), we deduce that the second line is also bounded by \(\tau ^{\frac{1}{2}(\frac{11}{3}-\gamma )} (E^N)^\frac{1}{2}\) and \(\varepsilon ^{M+1}\). Therefore, we obtain
It is now clear that there exists a universal C so that if we choose \(\sigma '=C(\varepsilon _*+\sigma _*)\) the bound (4.13) is consistent and can be justified by a classical continuity argument. The same comment applies to (5.1).
6 Compactness as \(\kappa \rightarrow 0\) and Proof of the Theorem 1.6
Let \(B^k\) be the Hilbert space generated by the norm
namely \(B^k=\overline{C_c(0,1)}^{\Vert \cdot \Vert _{B^k}}\). From the theory of weighted Sobolev spaces [27, 28], we deduce that \(B^k\) is compactly embedded into \(B^{k-1}\) for \(k\ge 1\).
Let a family of given initial data \((H_0^\kappa ,H_1^\kappa )\) satisfy the uniform bound
In particular, this gives the uniform bound of \(\Vert (\kappa ^{\frac{1}{2}(\gamma -\frac{11}{3})} H_0^\kappa , \kappa ^{\frac{1}{2}(\gamma -\frac{5}{3})} H_1^\kappa )\Vert _{B^N\times B^N} <\sqrt{2}\sigma \). By compact embedding of \(B^N\) into \(B^{N-1}\), there exists a sequence of \(\{\kappa _j \}_{j=1}^\infty \) such that \(\kappa _j \rightarrow 0\) and \(({\kappa _j}^{\frac{1}{2}(\gamma -\frac{11}{3})} H_0^{\kappa _j}, {\kappa _j}^{\frac{1}{2}(\gamma -\frac{5}{3})} H_1^{\kappa _j})\) converge in \(B^{N-1}\times B^{N-1}\). Fix such a sequence \(\kappa _j\) and initial data \(H_0^{\kappa _j}\) and \(H_1^{\kappa _j}\).
Now let \((H^{\kappa _j}, \partial _\tau H^{\kappa _j})\) be the solution to the initial value problem (3.21) with initial data \(H_0^{\kappa _j}\) and \(H_1^{\kappa _j}\) given by Theorem 1.13. Consider its well-defined trace at time \(\tau =1\) (that is \(t=0\) in the original coordinates). Since \(S_{\kappa _j}^N(\tau =1)<C_0\left( \sigma ^2+\varepsilon ^{2M+1}\right) \) for all \(\kappa _j\), in particular we have the uniform bound
where we have used the crude bound \(\varepsilon ^{2M+1}\le \varepsilon ^2\). Therefore, there exists a subsequence of \(\kappa _j\), denoted by \(\kappa _j\) again and \((H_0,H_1)\in B^{N}\times B^{N}\) so that
We now consider the solution of (3.21) with the final value \((H_0,H_1)\) at time \(\tau =1\). By the local well-posedness theory obtained similarly as in Proposition D.1, there exists a unique solution \((H, \partial _\tau H)\) to (3.21) on a maximal interval of existence (T, 1] for some \(T<1\).
We claim that \(T=0\). To see this, assume the opposite, that is \(0<T<1\). Then for each \(\kappa _j\in (0,\frac{T}{2}]\), consider the sequence of solutions \((H^{\kappa _j}, \partial _\tau H^{\kappa _j} ) \) to (3.21) with given initial condition \(H_0^{\kappa _j}\) and \(H_1^{\kappa _j}\). Then, on the interval \([\frac{T}{2},1]\) the sequence \((H^{\kappa _j}, \partial _\tau H^{\kappa _j})\) satisfies the uniform-in-j bound (5.126). In particular, as \(j\rightarrow \infty \), possibly along a subsequence, \((H^{\kappa _j}, \partial _\tau H^{\kappa _j})\) converges to some \((\bar{H},\partial _\tau \bar{H})\) in \(C^0\left( [\frac{T}{2},1], B^{N-1}\times B^{N-1}\right) \) and the resulting limit \((\bar{H},\partial _\tau \bar{H})\) is a classical solution of (3.21) on \([\frac{T}{2},1]\). Since the final condition at \(\tau =1\) has to coincide for \((\bar{H},\partial _\tau \bar{H})\) and \((H,\partial _\tau H)\), by the uniqueness part of the local well-posedness theorem, \(\bar{H} \) and H coincide on \([\frac{T}{2},1]\) which contradicts the assumption that (T, 1] is the maximal interval of existence for H.
Therefore we have established the existence of a classical solution
to (1.46) on the space-time domain \((\tau ,r)\in (0,1]\times [0,1]\). In particular, the leading order behavior of \(\phi \) is driven by the dust solution \(\phi _0=\tau ^{\frac{2}{3}}\) and we have
Claims (1.37)–(1.38) follow easily by going back to the (s, r)-coordinate system, which in turn give (1.39)–(1.41). This completes the proof of Theorem 1.6.
Data at \(s=0\). Note that the initial conditions (1.25) that correspond to the obtained collapsing solution are now given by
In particular, by the smallness of weighted norms of H and Hardy–Sobolev embeddings, we conclude
We may now express the initial density \(\tilde{\rho }_0\) and the initial velocity vector field \(\tilde{\mathbf{u}}_0\) (at time \(s=0\)) in Eulerian variables. Let \(Y=\chi _0(r)y=\phi (1,r)y\), \(\bar{\chi }_0(R)=\chi _0(r)\), \(R=|Y|=r\chi _0(r)\). By (1.11)
By (1.24) we have
From (1.6) we then conclude
Since by (1.59) \(w^\alpha (r) = 1- c r^n + o_{r\rightarrow 0}(r^n)\) for some \(c>0\) in the vicinity of \(r=0\), we conclude that we have the expansion
This formula in view of (6.2) gives a quantified sense in which the initial density is flat about the origin.
The Eulerian description of collapsing solutions. Let \(0<\tau \leqq 1\) be fixed. Note that \(\mathscr {J}[\phi ]>0\) and the Eulerian density is given by
where we have written \(\tilde{\varrho }(\tau , \phi (\tau ,r) r) = \tilde{\rho } (\frac{1-\tau }{g(r)}, \phi (\tau ,r) r) = \tilde{\rho } (s, \chi (s,r) r) \). Let \(\tilde{R}:= \phi (\tau ,r) r\). Then since \(\mathscr {J}[\phi ]>0\), there exists the inverse mapping \( r =\tilde{r}(\tau , \tilde{R})\) such that \( \tilde{r} (\tau , \phi (\tau ,r) r) = r \) for all \(r\in [0,1]\). We may rewrite the Eulerian density
where we have written \(\phi (\tau , \tilde{R})=\phi (\tau , r)\) for \(\tilde{R} = \phi (\tau ,r)r\). By our construction, \(\lim _{\tau \rightarrow 0^+}\frac{\phi }{\phi _0}=1 \) and \(\lim _{\tau \rightarrow 0^+}\frac{\mathscr {J}[\phi ] }{\mathscr {J}[\phi _0]}=1\) for all \(\tilde{R} \in [0,\phi (\tau ,1)]\). Therefore, we deduce that
The right-hand side is nothing but the density driven by the dust profile (1.34) written in \(\tau \) coordinate. Switching back to the (s, r)-coordinate system, this is precisely in agreement with (1.39) and highlights the role of the dust profile in our collapse.
Notes
We note that a related idea of finding an approximate solution has been used recently in the semilinear and energy subcritical wave equation [10].
References
Binney, J., Tremaine, S.: Galactic Dynamics. Princeton University Press, Princeton 2008
Blottiau, P., Bouquet, S., Chièze, J.P.: An asymptotic self-similar solution for the gravitational collapse. Astron. Astrophys. 207, 24–36, 1988
Bogoyavlensky, O.I.: Methods in the Qualitative Theory of Dynamical Systems in Astrophysics and Gas Dynamics. Springer, Berlin 1985
Borisov, A.V., Kilin, A.A., Mamaev, I.S.: The Hamiltonian dynamics of self-gravitating liquid and gas ellipsoids. Regul. Chaotic Dyn. 14(2), 179–217, 2008
Bouquet, S., Feix, M.R., Fijalkow, E., Munier, A.: Density bifurcation in a homogeneous isotropic collapsing star. Astrophys. J. 293, 494–503, 1985
Brenner, M.P., Witelski, T.P.: On spherically symmetric gravitational collapse. J. Stat. Phys. 93(3–4), 863–899, 1998
Calvez, V., Carrillo, J.A., Hoffmann, F.: Equilibria of homogeneous functionals in the fair-competition regime. Nonlinear Anal. TMA 159, 85–128, 2017
Calvez, V., Carrillo, J.A., Hoffmann, F.: The geometry of diffusing and self-attracting particles in a one-dimensional fair-competition regime, vol. 2186. Lecture Notes in MathematicsCIME Foundation Subseries, Springer 2018
Carrillo, J.A., Wróblewska-Kaminska, A., Zatorska, E.: On long-time asymptotics for viscous hydrodynamic models of collective behavior with damping and nonlocal interactions. Math. Models Methods Appl. Sci. 29(1), 31–63, 2019
Cazenave, T., Martel, Y., Zhao, L.: Solutions blowing up on any given compact set for the energy subcritical wave equation. J. Differ. Equ. 268(2), 680–706, 2020
Chandrasekhar, S.: An Introduction to the Study of Stellar Structures. Dover, London 1967
Cheng, B., Tadmor, E.: An improved local blow-up condition for Euler–Poisson equations with attractive forcing. Physica D 238, 2062–2066, 2009
Chiodaroli, E., De Lellis, C., Kreml, O.: Global ill-posedness of the isentropic system of gas dynamics. Comm.un Pure Appl. Math. 68(7), 1157–1190, 2015
Christodoulou, D.: Violation of cosmic censorship in the gravitational collapse of a dust cloud. Commun. Math. Phys. 93, 171–195, 1984
Christodoulou, D.: The Formation of Shocks in 3-Dimensional Fluids. EMS Monographs in Mathematics. EMS Publishing House, London 2007
Christodoulou, D., Miao, S.: Compressible Flow and Euler’s Equations, Surveys in Modern Mathematics, vol. 9, pp. 1–602. International Press, New York 2014
Coutand, D., Shkoller, S.: Well-posedness in smooth function spaces for the moving boundary three-dimensional compressible Euler equations in physical vacuum. Arch. Ration. Mech. Anal. 206(2), 515–616, 2012
Dafermos, M., Holzegel, G., Rodnianski, I.: The linear stability of the Schwarzschild solution to gravitational perturbations. Acta. Math. 222(1), 1–214, 2019
Dafermos, M., Rodnianski, I., Shlapentokh-Rothman, Y.: Decay for solutions of the wave equation on Kerr exterior spacetimes III: the full subextremal case \(|a|<M\). Ann. Math. 183, 787–913, 2016
De Lellis, C., Székelyhidi Jr., L.: High dimensionality and h-principle in PDE. Bull. Am. Math. Soc. 54(2), 247–282, 2017
Deng, Y., Liu, T.P., Yang, T., Yao, Z.: Solutions of Euler–Poisson equations for gaseous stars. Arch. Ration. Mech. Anal. 164(3), 261–285, 2002
Deng, Y., Xiang, J., Yang, T.: Blowup phenomena of solutions to Euler–Poisson equations. J. Math. Anal. Appl. 286, 295–306, 2003
Dyson, F.J.: Dynamics of a spinning gas cloud. J. Math. Mech. 18(1), 91–101, 1968
Fu, C.-C., Lin, S.-S.: On the critical mass of the collapse of a gaseous star in spherically symmetric and isentropic motion. Jpn. J. Indust. Appl. Math. 15(3), 461–469, 1998
Goldreich, P., Weber, S.: Homologously collapsing stellar cores. Astrophys. J. 238, 991–997, 1980
Gu, X., Lei, Z.: Local Well-posedness of the three dimensional compressible Euler–Poisson equations with physical vacuum. J. Math. Pures Appl. 105(5), 662–723, 2016
Gurka, P., Opic, B.: Continuous and compact imbeddings of weighted Sobolev spaces. I. Czechoslov. Math. J. 38, 730–744, 1988
Gurka, P., Opic, B.: Continuous and compact imbeddings of weighted Sobolev spaces. II. Czechoslov. Math. J. 39, 78–94, 1989
Hadžić, M., Jang, J.: Nonlinear stability of expanding star solutions in the radially-symmetric mass-critical Euler–Poisson system. Commun. Pure Appl. Math. 71(5), 827–891, 2018
Hadžić, M., Jang, J.: Expanding large global solutions of the equations of compressible fluid mechanics. Invent. Math. 214(3), 1205–1266, 2018
Hadžić, M., Jang, J.: A class of global solutions to the Euler-Poisson system. Comm. Math. Phys. 370(2), 475–505, 2019
Holzegel, G., Luk, J., Speck, J., Wong, W.: Stable shock formation for nearly simple outgoing plane symmetric waves. Ann. PDE 2(2), 1–198, 2016
Jang, J.: Nonlinear instability in gravitational Euler–Poisson system for \(\gamma =6/5\). Arch. Ration. Mech. Anal. 188, 265–307, 2008
Jang, J.: Nonlinear instability theory of Lane–Emden stars. Commun. Pure Appl. Math. 67(9), 1418–1465, 2014
Jang, J.: Time periodic approximations of the Euler–Poisson system near Lane–Emden stars. Anal. PDE 9(5), 1043–1078, 2016
Jang, J., Masmoudi, N.: Well-posedness for compressible Euler equations with physical vacuum singularity. Commun. Pure Appl. Math. 62, 1327–1385, 2009
Jang, J., Masmoudi, N.: Well-posedness of compressible Euler equations in a physical vacuum. Commun. Pure Appl. Math. 68(1), 61–111, 2015
Johnson, W.P.: The Curious history of Faa di Bruno’s formula. Am. Math. Monthly 109, 217–234, 2002
Larson, R.B.: Numerical calculations of the dynamics of a collapsing protostar. Monthly Not. R. Astron. Soc. 145, 271–295, 1969
Luk, J., Speck, J.: Shock formation in solutions to the 2D compressible Euler equations in the presence of non-zero vorticity. Invent. Math. 214(1), 1–169, 2018
Luo, T., Smoller, J.: Existence and nonlinear stability of rotating star solutions of the compressible Euler–Poisson equations. Arch. Ration. Mech. Anal. 191(3), 447–496, 2009
Luo, T., Xin, Z., Zeng, H.: Well-posedness for the motion of physical vacuum of the three-dimensional compressible Euler equations with or without self-gravitation. Arch. Ration. Mech. Anal. 213(3), 763–831, 2014
Makino, T.: Blowing up solutions of the Euler–Poisson equation for the evolution of gaseous stars. Transp. Theory Stat. Phys. 21, 615–624, 1992
Makino, T., Perthame, B.: Sur les Solutions á Symétrie Sphérique de l’Equation d’Euler-Poisson pour l’Evolution d’Etoiles Gazeuses. Jpn. J. Appl. Math. 7, 165–170, 1990
Oppenheimer, J.R., Snyder, H.: On continued gravitational contraction. Phys. Rev. 56, 455–459, 1939
Ovsiannikov, L.V.: New solution of hydrodynamic equations. Dokl. Akad. Nauk SSSR 3(1), 47–49, 1956
Penston, M.V.: Dynamics of self-gravitating gaseous spheres-III. Analytical results in the free-fall of isothermal cases. Monthly Not. Roy. Astron. Soc. 144, 425–448, 1969
Rein, G.: Non-linear stability of gaseous stars. Arch. Ration. Mech. Anal. 168(2), 115–130, 2003
Ringeval, C., Bouquet, S.: Dynamical stability for the gravitational evolution of a homogeneous polytrope. Astron. Astrophys. 355, 564–572, 2000
Shkoller, S., Sideris, T.C.: Global existence of near-affine solutions to the compressible Euler equations. Arch. Ration. Mech. Anal. 234(1), 115–180, 2019
Shu, F.H.: Self-similar collapse of isothermal spheres and star formation. Astrophys. J. Part 1 214, 488–497, 1977
Sideris, T.C.: Formation of singularities in three-dimensional compressible fluids. Commun. Math. Phys. 101, 475–485, 1985
Sideris, T.C.: The lifespan of 3D compressible flow. Ecole Polytechnique. Exposé No. V, Centre de Mathematiques 1991
Sideris, T.C.: Spreading of the free boundary of an ideal fluid in a vacuum. J. Differ. Equ. 257(1), 1–14, 2014
Sideris, T.C.: Global existence and asymptotic behavior of affine motion of 3D ideal fluids surrounded by vacuum. Arch. Ration. Mech. Anal. 225(1), 141–176, 2017
Speck, J.: Shock formation in small-data solutions to 3D quasilinear wave equations. Math. Surv. Monogr. (AMS) 1–515, 2016, 2016
Speck, J.: A summary of some new results on the formation of shocks in the presence of vorticity. In: Nonlinear analysis in geometry and applied mathematics. Harv. Univ. Cent. Math. Sci. Appl. Ser. Math., 1, Int. Press, Somerville, MA, pp. 133–157, 2017
Taylor, M.: The nonlinear stability of the Schwarzschild family of black holes (joint with M. Dafermos, G. Holzegel, I. Rodnianski). Oberwolfach Reports, Workshop ID 1832, Mathematical General Relativity 2018
Tohline, J.E.: Hydrodynamic collapse. Fundam. Cosmic Phys. 8, 1–82, 1982
Tolman, R.C.: Effect of inhomogeneity on cosmological models. Proc. Natl. Acad. Sci. USA 20, 169–176, 1934
Yahil, A.: Self-similar stellar collapse. Astrophys. J. 265, 1047–1055, 1983
Zel’dovich, Y.B., Novikov, I. D., Relativistic Astrophysics Vol. 1: Stars and Relativity. Chicago University Press, Chicago, 1971
Acknowledgements
Y. Guo is indebted to D. Christodoulou for bringing his attention to the study of the Newtonian gravitational collapse in 1996. He acknowledges the support of the NSF Grant DMS-1810868. M. Hadž ić acknowledges the support of the EPSRC Grant EP/N016777/1. J. Jang acknowledges the support of the NSF Grants DMS-1608494 and DMS-2009458, and the Simons Fellowship (grant # 616364).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by N. Masmoudi
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
A Vector Field Classes \(\mathcal {P}\) and \(\bar{\mathcal {P}}\)
In what follows, we present key lemmas that highlight the importance of \(\mathcal {P}_i\) and \(\bar{\mathcal {P}}_i\).
The first key is that the original \(\mathcal {D}_i\)’s control those admissible vector fields in \(L^2\) sense so that the members of \(\mathcal {P}_i\) and \(\bar{\mathcal {P}}_i\) can be freely used in the energy estimates. A clue is in the divergence structure of \(D_r\) that grants an extra control of \(\frac{1}{r}\). More precisely, we have
Lemma A.1
Let \(i\in \mathbb {N}\) be given. Then we have the following identity:
Moreover, we have the following estimate:
where \(\chi \geqq 0\) is a smooth cutoff function satisfying \(\chi =1\) on \([0,\frac{1}{2}]\), \(\chi =0\) on \([\frac{3}{4},1]\), and \(\chi '\leqq 0\).
Proof
We first establish (A.401). The proof is based on the induction on i. First observe
and hence (A.401) holds for \(i=1\). Suppose (A.401) is valid for all \(i\leqq \ell \). If \(\ell \) is even, \(\mathcal {D}_{\ell +1} = D_r \mathcal {D}_\ell \), \(\bar{\mathcal {D}}_\ell = D_r \bar{\mathcal {D}}_{\ell -1}\), and \( r\partial _r \bar{\mathcal {D}}_{\ell -1} \left( \frac{X}{r}\right) + (\ell +2) \bar{\mathcal {D}}_{\ell -1} \left( \frac{X}{r}\right) = r \bar{\mathcal {D}}_\ell \left( \frac{X}{r}\right) + \ell \bar{\mathcal {D}}_{\ell -1} \left( \frac{X}{r}\right) \). Then by the induction hypothesis, we deduce
which verifies (A.401) for \(i=\ell +1\). If \(\ell \) is odd, \(\mathcal {D}_{\ell +1} = \partial _r \mathcal {D}_\ell \), \(\bar{\mathcal {D}}_\ell = \partial _r \bar{\mathcal {D}}_{\ell -1}\), and \( r\partial _r \bar{\mathcal {D}}_{\ell -1} \left( \frac{X}{r}\right) + (\ell +2) \bar{\mathcal {D}}_{\ell -1} \left( \frac{X}{r}\right) = r \bar{\mathcal {D}}_\ell \left( \frac{X}{r}\right) + (\ell +2 )\bar{\mathcal {D}}_{\ell -1} \left( \frac{X}{r}\right) \). Then
which also verifies (A.401) for \(i=\ell +1\). This finishes the proof of (A.401).
We will prove (A.402) only since the other can be shown similarly. To this end, we compute the square integral of \(\mathcal {D}_i X\) by using (A.401). Then
Since \(\chi '\leqq 0\), the result follows for \(i> 1\). For \(i=1\), we need to show that \(\frac{X}{r}\) is bounded by \(\mathcal {D}_1 X\). Observe that
from which (A.402) follows also for \(i=1\). \(\square \)
Remark A.2
We note that the cut-off \(\chi \) in the above proof can be replaced by any nonnegative \(C^1\)-function supported on [0, 1] such that \(\chi '\le 0\). In particular, choosing \(\chi (r) = q_{-\gamma -1}\left( \frac{r^n}{\tau }\right) w^{\alpha +i+1}\) as a corollary we obtain
An important consequence of Lemma A.1 is the following:
Lemma A.3
Suppose \(\mathcal {D}_i X\) is bounded in \(L^2([0,\frac{3}{4}], r^2\mathrm{d} r)\). Then we have the estimate
where \(\chi \) is the same cutoff function in Lemma A.1.
Proof
If there is no \(\frac{1}{r}\) in a given \(\mathfrak {D}_i\), we are done since it is the same as \(\mathcal {D}_i\). Suppose \(\frac{1}{r}\) appears in \(\mathfrak {D}_i\). Then we may write \(\mathfrak {D}_i X = \bar{\mathcal {D}}_{i-j-1} \frac{1}{r}Y_j X\) for some \(Y_j\in \mathcal {P}_j\). Apply Lemma A.1 to get the \(L^2\) bound for \(\mathfrak {D}_i X\) by \(\mathcal {D}_{i-j} Y_j X\). If \(Y_j\) does not have \(\frac{1}{r}\), we are done. If it does, then we repeat the same argument by writing \(\mathcal {D}_{i-j} Y_j = \bar{\mathcal {D}}_{i-j-k-1} \frac{1}{r}Z\) for some \(Z \in \mathcal {P}_k\) and applying Lemma A.1 etc. The repetition ends in at most \(\lfloor \frac{i+1}{2}\rfloor \) steps. \(\square \)
Another appealing feature of \(\mathcal {P}_i\) and \(\bar{\mathcal {P}}_i\) is that they give rise to an algebraic structure via the following Leibniz rule:
Lemma A.4
(Product rule) Let \(i\in \mathbb {N}\) be given.
-
(a)
For any \(A\in \mathcal {P}_i\) the following identity holds:
$$\begin{aligned} A \left( f g \right) = \sum _{k=0}^i \sum _{ B \in \mathcal {P}_k \atop C\in \bar{\mathcal {P}}_{i-k}} c^{ABC}_{k} \ Bf\, Cg, \end{aligned}$$(A.404)for some real-valued constants \(c^{ABC}_{k}\).
-
(b)
For any \(A\in \bar{\mathcal {P}}_i\) the following identity holds:
$$\begin{aligned} A \left( f g \right) = \sum _{k=0}^i \sum _{ B \in \bar{\mathcal {P}}_k \atop C\in \bar{\mathcal {P}}_{i-k}} {\bar{c}}^{ABC}_{k} \ Bf\, Cg, \end{aligned}$$(A.405)for some real-valued constants \({\bar{c}}^{ABC}_{k}\).
Proof
The proof is based on the induction on i. We start with (A.405). Let \(i=1\). Then \(\mathcal {P}_1 =\{D_r, \frac{1}{r}\}\). Since we have
and
(A.405) holds for \(i=1\). Suppose (A.405) is true for all \(i\leqq \ell \). We will show that it is true for \(i=\ell +1\).
First let \(\ell \) be even. Then \(A\in \mathcal {P}_{\ell +1}\) is either \(A=D_r A'\) or \(A=\frac{1}{r}A'\) for some \(A' \in \mathcal {P}_{\ell }\). By induction hypothesis, it suffices to show that \(D_r (B' f C' g )\) and \(\frac{1}{r}(B' f C' g )\) for \(B'\in \mathcal {P}_{k'}\) and \(C'\in \bar{\mathcal {P}}_{\ell -k'}\) can be written as a linear combination of (Bf)(Cg) for some \(B\in \mathcal {P}_k\) and \(C\in \bar{\mathcal {P}}_{\ell +1-k}\). If \(k'\) is even, we write \(D_r (B' f C' g ) = (D_r B' f) (C' g) + (B' f) (\partial _r C' g )\) and \(\frac{1}{r}(B' f C' g ) = (\frac{1}{r}B' f ) C' g\). Since \(D_r B' \in \mathcal {P}_{k'+1}\), \(\frac{1}{r}B'\in \mathcal {P}_{k'+1}\) and \(\partial _r C' \in \bar{\mathcal {P}}_{\ell +1-(k'+1)}\), both expressions are in the desirable form. If \(k'\) is odd, we may write \(D_r (B' f C' g ) = (\partial _r B' f) (C' g) + (B' f) (D_r C' g )\) and \(\frac{1}{r}(B' f C' g ) = (B' f ) (\frac{1}{r}C' g)\) so that they are in the desirable form.
Now let \(\ell \) be odd. Then \(A=\partial _r A'\) where \(A'\in \mathcal {P}_\ell \). We will show that \(\partial _r(B' f C' g) \) for \(B'\in \mathcal {P}_{k'}\) and \(C'\in \bar{\mathcal {P}}_{\ell -k'}\) can be written into a desirable form. As before, we consider \(k'\) even, odd separately. If \(k'\) is even, we write
Note that \(D_r B' \in \mathcal {P}_{k'+1}\) and \(D_r C', \ \frac{1}{r} C' \in \bar{\mathcal {P}}_{\ell +1-k'}\), and hence it has the desirable form. If \(k'\) is odd, \( \partial _r( B'f C' g) = (\partial _r B' f) C' g +B'f (\partial _r C' g)\). The result follows since \(\partial _r B' \in \mathcal {P}_{k'+1}\) and \(\partial _r C' \in \bar{\mathcal {P}}_{\ell +1-k'}\).
(A.406) is an easy consequence of (A.405), because any \(A\in \bar{\mathcal {P}}_i\) can be written as \(A = A ' \partial _r\) for some \(A' \in \mathcal {P}_{i-1}\). Write \(A(fg) = A' ((\partial _r f) g) + A'((\partial _r g) f)\) and apply (A.405) with \(A'\) to obtain the desired result. \(\square \)
Lemma A.5
(Chain rule). Let \(a\in \mathbb {R}, i\in \mathbb {N}\) be given and fix a vectorfield \(W\in \bar{\mathcal {P}}_i\). Then for any sufficiently smooth f the following identity holds:
for some real constants \(c_{k,i_1,\ldots ,i_k}\).
Proof
The proof relies on an induction argument. Let \(i=1\). Then \(W=\partial _r\) and \(\partial _r (f^a) = a f^{a-1} \partial _r f \) which verifies (A.407). Suppose (A.407) is true for all \(i\leqq \ell \). Then we will show that it is true for \(i=\ell +1\). First let \(\ell \) be even. Then \(W\in \bar{\mathcal {P}}_{\ell +1}\) can be written as \(W=\partial _r W'\) for some \(W'\in \bar{\mathcal {P}}_{\ell }\). By induction hypothesis,
The first sum of the right-hand side is a linear combination of \(f^{a-k'} \prod _{j=1}^{k'} W'_j f\) where \(2\leqq k'\leqq \ell +1\), \(W_j =W'_j\) for \(j\leqq k'-1\), \(W_{k'}= \partial _r \) and \(W_j\in \bar{\mathcal {P}}_{i_j}\) with \(i_1+ \cdots i_{k'}=\ell +1 \). Hence it is in the desirable form. For the second sum, no increase in the number of the product occurs, but the number of derivatives increases by one. Note that \(\partial _r\Big ( \prod _{j=1}^k W'_j f \Big ) =\sum _{m=1}^k \prod _{j=1}^k W^m_j f \) where \(W^m_j = W'_j\) for \(j\ne m\) and \(W^m_j= \partial _r W'_j\) for \(j=m\). Now if the corresponding \(i_m\) is even, \(\partial _r W'_m \in \bar{\mathcal {P}}_{i_m+1}\). If \(i_m\) is odd, we write \(\partial _r W'_m = D_r W'_m - \frac{2}{r} W'_m\) such that each operation belongs to \(\bar{\mathcal {P}}_{i_m+1}\). In both cases, we have \(i_1+ \cdots i_k =\ell +1\) and hence (A.407) is valid for \(i=\ell +1\). Next let \(\ell \) be odd. Then \(W\in \bar{\mathcal {P}}_{\ell +1}\) can be written as either \(W=D_r W'\) or \(W=\frac{1}{r} w'\) for some \(W'\in \bar{\mathcal {P}}_{\ell }\). We consider \(W=\frac{1}{r} w'\) only since the other case follows similarly by combining the previous cases. Write
Now we claim that \(\frac{1}{r}\prod _{j=1}^k W'_j f = \prod _{j=1}^k W_j f \) for some \(W_j \in \bar{\mathcal {P}}_{i_j}\) with \(i_1+ \cdots i_k =\ell +1\). To this end, we first observe because \(\ell \) is odd, there exists at least one index m, \(1\leqq m\leqq k\) whose corresponding \(i_m\) is odd. Now let \(W_j = W'_j\) for \(j\ne m\) and \(W_m=\frac{1}{r} w'_m\). Then it is easy to see that all properties are satisfied so that the expression has its desirable form required by (A.407) for \(i=\ell +1\). \(\square \)
The next lemma implies that \(\partial _r^i X\) can be expressed as a linear combination of admissible operations belonging to \(\mathcal {P}_i\). Note that the other way around is not true in general: for instance, \(D_r \) can’t be expressed in terms of \(\partial _r \) only. Hence, our energy built upon \(\mathcal {D}_i\)’s controls \(\partial _r^i \)’s as well as \(\mathcal {P}_i\)’s.
Lemma A.6
Let \(i\in \mathbb {N}\) be given. Then we have
where \(c_{i}^{A}\)’s are constants.
Proof
The proof is based on the induction on i. Let \(i=1\). Then we have
Since \(D_r\) and \(\frac{1}{r}\) belong to \(\mathcal {P}_1\), (A.408) holds. Suppose (A.408) is true for all \(i\leqq \ell \). We will show that (A.408) holds for \(i=\ell +1\). By induction hypothesis,
Now if \(\ell \) is odd, \( \partial _r A\) is an admissible vector field belonging to \(\mathcal {P}_{\ell +1}\), and hence we are done. If \(\ell \) is even, then we write
Then \(D_r A\) and \(\frac{1}{r} A\) are both admissible vector fields belonging to \(\mathcal {P}_{\ell +1}\) and thus (A.408) holds for \(i=\ell +1\). \(\square \)
The same conclusion holds in Lemma A.6 when we replace \(A\in \mathcal {P}_i\) by \(A\in \bar{ \mathcal {P}}_i\).
We also write a few useful identities relating high-order \(\mathcal {D}\), \( r\partial _r\), and \(\partial _r\) derivatives.
Lemma A.7
-
(i)
For any \(j\in \mathbb {Z}_{\ge 0}\) there exist constants \(c_{k}\), \(k\in \{0,\ldots ,j\}\) such that
$$\begin{aligned} (r\partial _r)^j = \sum _{k=0}^j c_{k} r^{k}\partial _r^k. \end{aligned}$$(A.408) -
(ii)
For any \(j\in \mathbb {Z}_{\ge 0}\) there exist constants \(\bar{c}_{k}\), \(k\in \{0,\ldots ,j\}\) such that
$$\begin{aligned} \partial _r^j = r^{-j}\sum _{k=0}^j \bar{c}_{k} (r\partial _r)^k. \end{aligned}$$(A.409) -
(iii)
For any \(j\in \mathbb {Z}_{\ge 0}\) there exist constants \(\tilde{c}_{k}\), \(k\in \{0,\ldots ,j\}\) such that
$$\begin{aligned} \mathcal {D}_j = r^{-j}\sum _{k=0}^j \tilde{c}_{k} (r\partial _r)^k. \end{aligned}$$(A.410)The same conclusion holds when we replace \(\mathcal {D}_j\) by \(A\in \mathcal {P}_j\).
-
(iv)
For any \(j\in \mathbb {Z}_{>0}\) there exist constants \(\hat{c}_{k}\), \(k\in \{0,\ldots ,j\}\) such that
$$\begin{aligned} \bar{\mathcal {D}}_j = r^{-j}\sum _{k=1}^j \hat{c}_{k} (r\partial _r)^k. \end{aligned}$$(A.411)The same conclusion holds when we replace \(\bar{\mathcal {D}}_j\) by \(A\in \bar{\mathcal {P}}_j\).
Proof
The above statements follow easily by induction. \(\square \)
B High-Order Commutators
From definitions (4.1)–(4.2) and the product rule
it is easy to check that the following commutation rules hold:
More generally, we have the following commutation rules for \(\mathcal {D}_i L_\alpha \):
Lemma B.1
For any \(i\in \mathbb {Z}_{>0}\) there exist constants \(c_{ijk}\), \(j\in \mathbb {Z}_{\ge 0}\), \(k\in \mathbb {Z}_{>0}\) such that
where
Proof
The proof is based on the induction on i. First let \(i=1\). From (B.413), we have
and hence (B.415) holds with \(p_{10}= -(1+\alpha ) \partial _ r^2 w-2(1+\alpha )\frac{\partial _r w}{r}\). Suppose (B.415) is valid for all \(i\leqq \ell \). It suffices to show that (B.415) holds for \(i=\ell +1\). If \(\ell \) is even, by (B.415) and (B.413), we have
The last three terms can be rearranged as \(\sum _{j=0}^\ell \zeta _{1+\ell j} \mathcal {D}_{1+\ell -j} X\) where
Note that \(\zeta _{1+\ell j}\) takes the form given in (B.416) because of the induction assumption. Therefore, (B.415) holds for \(i=\ell +1\). If \(\ell \) is odd, we use (B.414) in place of (B.413) to derive the same conclusion. This finishes the proof. \(\square \)
Next we present the commutator identities useful for derivation of high order equations as well as for high order estimates.
Lemma B.2
Let \(i\in \mathbb {Z}_{>0}\) be given and let e, X be sufficiently smooth functions. For given differential operators A and B, let
denote the usual commutator. Then for any \(1\le k \le i\) and any \(A\in \bar{\mathcal {P}}_k\), \(B\in \mathcal {P}_{i-k}\) there exists a constant\(c_k^{iAB}\in \mathbb {R}\), and similarly for any \(A\in \bar{\mathcal {P}}_k\), \(B\in \bar{\mathcal {P}_{i-k}}\) there exists a constant \(\bar{c}_k^{iAB}\) such that following identities hold:
Proof
Proof of (B.418). The proof is based on the induction on i. Let \(i=1\). Then by using the identity \(D_r \partial _r X= \partial _r D_r X + \frac{2}{ r^2}X\),
which yields (B.418) for \(i=1\). Suppose (B.418) is valid for all \(i\leqq \ell \). It suffices to show (B.418) for \(i=\ell +1\). We will verify it when \(\ell \) is odd. The other case (\(\ell \) is even) will follow similarly. By using the induction assumption,
We note that each expression in the above summations belongs to either summation in (B.418) for \(i=\ell +1\). Proof of (B.419) and (B.420) follows analogously. \(\square \)
C Hardy–Sobolev Embedding
Let \(\chi , \; \psi \geqq 0\) be smooth cutoff functions satisfying \(\chi =1\) on \([0,\frac{1}{2}]\), \(\chi =0\) on \([\frac{3}{4},1]\) and \(\psi =1\) on \([\frac{1}{2},1]\), \(\psi =0\) on \([0,\frac{1}{4}]\), satisfying, in addition, that
Lemma C.1
(Localized Hardy inequalities). Let \(\chi ,\psi \) be the above defined cut-off functions and let \(u: B_1(0)\rightarrow \mathbb {R}\) be a given smooth radially symmetric function, where \(B_1(0)=\{x, \big | |x|\le 1\}\) is the unit ball in \(\mathbb {R}^3\). Then
-
1.
$$\begin{aligned} \int \nolimits _0^\frac{3}{4} |u|^2 \chi ^2\,\mathrm{d} r \lesssim \int \nolimits _\frac{1}{2}^\frac{3}{4} |u|^2 r^2\,\mathrm{d} r + \int \nolimits _0^\frac{3}{4} |A u|^2 r^2 \chi ^2\,\mathrm{d} r, \end{aligned}$$(B.420)
where \(A=\mathcal {D}_1 = \partial _r+\frac{2}{r}\) or \(A = \bar{\mathcal {D}}_1 = \partial _r\).
-
2.
Let \(a>1\) be given. Then
$$\begin{aligned} \int \nolimits _\frac{1}{4}^1 w^{a-2} |u|^2\psi ^2\,\mathrm{d} r \lesssim \int \nolimits _\frac{1}{4}^\frac{1}{2} w^a |u|^2\,\mathrm{d} r + \int \nolimits _\frac{1}{4}^1 w^a |A u|^2\psi ^2\,\mathrm{d} r, \end{aligned}$$(C.421)where \(A=\mathcal {D}_1 = \partial _r\) or \(A = \bar{\mathcal {D}}_1 = \partial _r+\frac{2}{r}\).
Proof
The proof is based on the standard Hardy inequality and the cutoff function argument. For \(A=\bar{\mathcal {D}}_1\), see [34]. The case \(A=\mathcal {D}_1=\partial _r\) follows from
The very same bound implies that
which immediately yields (C.421) with \(A=D_r\). The localized Hardy inequality near the boundary follows similarly. We note that \(w\sim 1-r\) in the vicinity of the boundary \(r=1\). \(\square \)
As a consequence of the above lemma, we have that for any smooth \(u:B_1(0)\rightarrow \mathbb {R}\) and any \(m\in \mathbb {Z}_{>0}\),
where either \(A=D_r\) and \(A_i = \mathcal {D}_i\) for all \(i=0,1,\ldots ,m\), or \(A=\partial _r\) and \(A_i = \bar{\mathcal {D}}_i\) for all \(i=0,1,\ldots ,m\). and the same estimate holds with \(D_r\) replaced by \(\partial _r\) and \(\mathcal {D}_i\) by \(\bar{\mathcal {D}}_i\). See Lemma 3.3 of [35] for the proof. Note the term \(\int \nolimits _\frac{1}{2}^\frac{3}{4} |u|^2 r^2\,\mathrm{d} r\) in (C.421) has been absorbed into the second summation in (C.424). We remark that away from the origin, both \(\sum _{i=0}^{m} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |{\mathcal {D}}_i u|^2\,\mathrm{d} r\) and \(\sum _{i=0}^{m} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |\bar{\mathcal {D}}^i u|^2\,\mathrm{d} r\) are equivalent to \(\sum _{i=0}^{m} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |\partial _r^i u|^2\,\mathrm{d} r\) for any \(m\in \mathbb {Z}_{>0}\).
The next result concerns the \(L^\infty \) bound.
Lemma C.2
Under the same assumptions as in Lemma C.1 and any \(m\in \mathbb {Z}_{>0}\), we have
where either \(B_i=\mathcal {D}_i\) or \(B_i = \bar{\mathcal {D}}_i\), \(i=0,1,\ldots ,m+1\). Moreover,
Proof
The proof follows from the Sobolev embedding inequality
While the summed norms generated by \(\mathcal {D}_i\) or \(\bar{\mathcal {D}}_i\) or \(\partial _r^i\) are all equivalent away from the origin, the ordered derivatives near the origin require some attention. We start with (C.425). First, we will verify (C.425) in the case \(B_i = \bar{\mathcal {D}}_i\), \(i=0,1,\ldots ,m+1\). By the above Sobolev embedding and (C.424),
where we have applied (C.424) to \(\Vert u\Vert _{L^1}\) with the choice \(A= \partial _r\) and \(A_i =\bar{\mathcal {D}}_i\) for \(i=0,\ldots ,m\) and then to \(\Vert \partial _r u\Vert _{L^1}\) with the choice \(A=D_r\) and \(A_i =\mathcal {D}_i\) for \(i=0,\ldots ,m\). For the last line, we simply remark that away from the origin, both \(\sum _{i=0}^{m} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |{\mathcal {D}}_i u|^2\,\mathrm{d} r\) and \(\sum _{i=0}^{m} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |\bar{\mathcal {D}}^i u|^2\,\mathrm{d} r\) are equivalent to \(\sum _{i=0}^{m} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |\partial _r^i u|^2\,\mathrm{d} r\) for any \(m\in \mathbb {Z}_{>0}\). This proves (C.425) when \(B_i = \bar{\mathcal {D}}_i\) for \(i=0,\ldots ,m+1\).
In the case \(B_i = \mathcal {D}_i\) for \(i=0,\ldots ,m+1\), we apply (C.424) to \(\Vert u\Vert _{L^1}\) with \(A= D_r\) and \(A_i = \mathcal {D}_i\) for \(i=0,\ldots ,m\) and to \(\Vert \partial _r u\Vert _{L^1}\) with \(A=\partial _r\) and \(A_i = \bar{\mathcal {D}}_i\) for \(i=0,\ldots ,m\), to obtain
Now it suffices to show that \( \int \nolimits _0^\frac{3}{4} | \partial _{r}^2 u |^2 r^2 \chi ^2\,\mathrm{d} r\) is bounded by \( \sum _{i=1}^{2} \int \nolimits _0^\frac{3}{4} |{\mathcal {D}}_i u |^2 r^2\chi ^2\,\mathrm{d} r + \sum _{i=0}^{m+1} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |{\mathcal {D}}_i u|^2\,\mathrm{d} r \). To this end, we first note that
Therefore,
where we have used \(\chi '\le 0\) in the last estimate. This yields (C.425) for \(B = D_r\) and \(B_i=\mathcal {D}_i\) for \(i=0,1,\ldots ,m+1\).
Next we will prove (C.426). First we have
where we have used the bound \(\sum _{i=0}^{m+1} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |{\mathcal {D}}_i (\frac{ u}{r} )|^2\,\mathrm{d} r \lesssim \sum _{i=0}^{m+1} \int \nolimits _{\frac{1}{4}}^1 w^{\alpha -\lfloor \alpha \rfloor + 2m} |{\mathcal {D}}_i u|^2\,\mathrm{d} r\) since \(\frac{1}{4}\le r\le 1\). By applying (C.421), we see that
For (a), we apply (C.421) to obtain
Note that by using (C.427)
which yields
For (b), we apply (C.421) again to obtain
By (C.427) and also using \(\partial _ r^2 D_r u = r \partial _r^3 (\frac{u}{r}) + 5 \partial _ r^2 (\frac{u}{r}) \),
where we have used (C.423) at the last equality. This in turn yields that
This finishes the proof of (C.426). \(\square \)
The same argument gives the following bound for \(\Vert ru\Vert _\infty \):
where we apply (C.421) just once near the origin, since \(\Vert ru\Vert _\infty \lesssim \Vert ru \Vert _{L^1} + \Vert \partial _r (r u)\Vert _{L^1}\), to derive the first sum.
We now recall the Hilbert space \(B^N\) with the norm
In what follows, we will derive the weighted \(L^2\) and \(L^\infty \) embedding inequalities for functions in \(B^N\) based on Lemma C.1, Lemma C.2.
Lemma C.3
(\(L^2\) weighted embeddings). Let \((H,\partial _\tau H)\in B^N\times B^N\) be given. Then we have
-
(i)
For any \( \frac{N-\alpha }{2} \leqq k\leqq N\)
$$\begin{aligned}&\tau ^{\gamma -\frac{5}{3}} \int w^{\alpha +2k-N} |\mathcal {D}_{k} \partial _\tau H |^2 r^2\,\mathrm{d} r+\tau ^{\gamma -\frac{11}{3}} \int w^{\alpha +2k-N} |\mathcal {D}_{k} H |^2 r^2\,\mathrm{d} r \lesssim E^N \nonumber \\&\tau ^{\gamma -\frac{8}{3}} \int w^{\alpha +2k-N} |\mathcal {D}_{k} \partial _\tau H |^2 r^2\,\mathrm{d} r+\tau ^{\gamma -\frac{14}{3}} \int w^{\alpha +2k-N} |\mathcal {D}_{k} H |^2 r^2\,\mathrm{d} r \lesssim D^N; \end{aligned}$$(C.428) -
(ii)
We further assume that \(\sum _{j=0}^N\int \frac{ w^{\alpha +j+1}}{(\tau +\frac{2}{3}M_g )^{1+\gamma }} |\mathcal {D}_{j+1} H |^2 r^2 \mathrm{d} r <\infty \). Then for any \( \frac{N-\alpha -1}{2} \leqq k\leqq N\)
$$\begin{aligned} \varepsilon \int \frac{ w^{\alpha +2k+1-N}}{(\tau +\frac{2}{3}M_g )^{1+\gamma }} |\mathcal {D}_{k+1} H |^2 r^2\,\mathrm{d} r \lesssim E^N. \end{aligned}$$(C.429)
Proof
We start with (C.429). Divide each integral appearing in the left-hand sides into two \(\int = \int \nolimits _0^\frac{1}{2} + \int \nolimits _{\frac{1}{2}}^1\). Then w is strictly positive for \(r\in [0,\frac{1}{2}]\) and hence \(w^{\alpha +2k-N}\lesssim w^{\alpha +k}\) for \(r\in [0,\frac{1}{2}]\), the first pieces are trivially bounded by the right-hand sides. Now for the second pieces, we apply (C.422) repeatedly \((N-k)\) times starting with \(a-2= \alpha + 2k -N \geqq 0\) to deduce the result.
Likewise, for (C.430), we divide the integral into two pieces. Then the integral restricted to \([0,\frac{1}{2}]\) is bounded by \(E^N\) because \(w^{\alpha +2k+1-N}\lesssim w^{\alpha +k+1}\) for \(r\in [0,\frac{1}{2}]\). For the integral from \(\frac{1}{2}\) to 1, we first observe that \((\tau +\frac{2}{3}M_g ) \) for \(r\in [\frac{1}{2},1]\) is bounded from below and above by positive constants and hence by applying (C.422) repeatedly \((N-k)\) times, we deduce the desired bound. \(\square \)
Lemma C.4
(\(L^\infty \) embedding). Let \((H,\partial _\tau H)\in B^N\times B^N\) be given. Then we have
-
(i)
For any \(k\in \mathbb {Z}_{\geqq 0}\) such that \(k\leqq \frac{N-\lfloor \alpha \rfloor -2}{2}\),
$$\begin{aligned} \tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert \mathcal {D}_k \partial _\tau H \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert \mathcal {D}_k H \Vert _\infty&\lesssim (E^N)^\frac{1}{2} \nonumber \\ \tau ^{\frac{1}{2}(\gamma -\frac{8}{3})}\Vert \mathcal {D}_k \partial _\tau H \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{14}{3})}\Vert \mathcal {D}_k H \Vert _\infty&\lesssim (D^N)^\frac{1}{2}; \end{aligned}$$(C.430) -
(ii)
Similarily, for any \(k\in \mathbb {Z}_{\geqq 0}\) such that \(k\leqq \frac{N-\lfloor \alpha \rfloor -2}{2}\),
$$\begin{aligned} \tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert \bar{\mathcal {D}}_k \partial _\tau (\frac{H}{r}) \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert \bar{\mathcal {D}}_k (\frac{H}{r}) \Vert _\infty&\lesssim (E^N)^\frac{1}{2} \nonumber \\ \tau ^{\frac{1}{2}(\gamma -\frac{8}{3})}\Vert \bar{\mathcal {D}}_k \partial _\tau (\frac{H}{r}) \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{14}{3})}\Vert \bar{\mathcal {D}}_k (\frac{H}{r}) \Vert _\infty&\lesssim (D^N)^\frac{1}{2}. \end{aligned}$$(C.431)
Proof
We start with (C.431). We present the detail for \(\Vert \mathcal {D}_k \partial H\Vert _\infty \) and other cases follow in the same way. By using (C.425) with \(u=\mathcal {D}_k H\) and \(m=\lfloor \alpha \rfloor + k +1\), we see that
where we take \(B_i =\mathcal {D}_i\) for k even and \(B_i=\bar{\mathcal {D}}_i\) for k odd. The first sum is trivially bounded by \( \tau ^{-(\gamma -\frac{11}{3})} (E^N)\) or \( \tau ^{-(\gamma -\frac{14}{3})} (D^N) \) since \(1\lesssim w^{\alpha +i+k}\) for \(r\in [0,\frac{3}{4}]\). For the second sum, since \(w^{\alpha -\lfloor \alpha \rfloor + 2 ( \lfloor \alpha \rfloor + k +1) } \lesssim w^{\alpha + i +k}\) for \(0\leqq i \leqq \lfloor \alpha \rfloor + k +2 \) and also the total number of derivatives appearing in the sum \( \lfloor \alpha \rfloor + k +2 + k\leqq N\), it is bounded by \( \tau ^{-(\gamma -\frac{11}{3})} (E^N) \) or \( \tau ^{-(\gamma -\frac{14}{3})} (D^N) \).
For (C.432), we apply (C.425) with \(m=\lfloor \alpha \rfloor + k +1\) with \(u=\bar{\mathcal {D}}_k (\frac{H}{r})\) and \(m=\lfloor \alpha \rfloor + k +1\),
where we take \(B_i =\bar{\mathcal {D}}_i\) for k even and \(B_i=\mathcal {D}_i\) for k odd. Now for the first sum, note that \(B_i \bar{\mathcal {D}}_k \frac{1}{r} \in {\mathcal {P}}_{i+k+1}\). Therefore, by Lemma A.3, it is bounded by \(\sum _{j=0}^{k+3} \int \nolimits _0^\frac{3}{4} | \mathcal {D}_j H |^2 r^2\,\mathrm{d} r \) and thus by \( \tau ^{-(\gamma -\frac{11}{3})} (E^N)\) or \( \tau ^{-(\gamma -\frac{14}{3})} (D^N) \) since \(k+3\leqq N\). Now for the second sum, note that when \(r\in [\frac{1}{4},1]\), in contrast to the first sum, \(\frac{1}{r}\) does not act as a derivative, in other words, \(|B_i (\bar{\mathcal {D}}_k (\frac{H}{r}) ) |^2\) is bounded by \(\sum _{j=0}^{i+k} |\mathcal {D}_j H|^2\). Therefore, by the same reasoning as in the previous case, we obtain the result. \(\square \)
Lemma C.5
(\(L^\infty \) \(w-\)weighted embedding). Let \((H,\partial _\tau H)\in B^N\times B^N\) be given. Then we have
-
(i)
For any \(k\in \mathbb {Z}_{\geqq 0}\) such that \(k+4 \leqq N\),
$$\begin{aligned} \tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert w^k \mathcal {D}_{k+2} \partial _\tau H \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert w^k \mathcal {D}_{k+2} H \Vert _\infty&\lesssim (E^N)^\frac{1}{2} \nonumber \\ \tau ^{\frac{1}{2}(\gamma -\frac{8}{3})}\Vert w^k \mathcal {D}_{p+2} \partial _\tau H \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{14}{3})}\Vert w^k \mathcal {D}_{k+2} H \Vert _\infty&\lesssim (D^N)^\frac{1}{2} \end{aligned}$$(C.432)for \(k=N-3\),
$$\begin{aligned} \tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert r \,w^{N-3} \mathcal {D}_{N-1} \partial _\tau H \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert r\, w^{N-3} \mathcal {D}_{N-1} H \Vert _\infty&\lesssim (E^N)^\frac{1}{2} \nonumber \\ \tau ^{\frac{1}{2}(\gamma -\frac{8}{3})}\Vert r \,w^{N-3} \mathcal {D}_{N-1} \partial _\tau H \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{14}{3})}\Vert r\, w^{N-3} \mathcal {D}_{N-1} H \Vert _\infty&\lesssim (D^N)^\frac{1}{2}; \end{aligned}$$(C.433) -
(ii)
Similarly, for any \(k\in \mathbb {Z}_{\geqq 0}\) such that \(k+5 \leqq N\),
$$\begin{aligned} \tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert w^k \bar{\mathcal {D}}_{k+2} \partial _\tau (\frac{H}{r}) \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert w^k \bar{\mathcal {D}}_{k+2} (\frac{ H}{r}) \Vert _\infty&\lesssim (E^N)^\frac{1}{2} \nonumber \\ \tau ^{\frac{1}{2}(\gamma -\frac{8}{3})}\Vert w^k \bar{\mathcal {D}}_{p+2} \partial _\tau (\frac{H}{r}) \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{14}{3})}\Vert w^k \bar{\mathcal {D}}_{k+2} (\frac{ H}{r}) \Vert _\infty&\lesssim (D^N)^\frac{1}{2} \end{aligned}$$(C.434)for \(k=N-4\),
$$\begin{aligned} \tau ^{\frac{1}{2}(\gamma -\frac{5}{3})}\Vert r \,w^{N-4} \bar{\mathcal {D}}_{N-2} \partial _\tau (\frac{H}{r}) \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{11}{3})}\Vert r\, w^{N-4} \bar{\mathcal {D}}_{N-2} (\frac{ H }{r})\Vert _\infty&\lesssim (E^N)^\frac{1}{2} \nonumber \\ \tau ^{\frac{1}{2}(\gamma -\frac{8}{3})}\Vert r \,w^{N-4} \bar{\mathcal {D}}_{N-2} \partial _\tau (\frac{H}{r}) \Vert _\infty + \tau ^{\frac{1}{2}(\gamma -\frac{14}{3})}\Vert r\, w^{N-4} \bar{\mathcal {D}}_{N-2} (\frac{ H }{r})\Vert _\infty&\lesssim (D^N)^\frac{1}{2}. \end{aligned}$$(C.435)
Proof
The proof follows in the same spirit as in the previous lemma. For \(w^k \mathcal {D}_{k+2} H\) in (C.433), we first apply (C.425) with \(u = w^k \mathcal {D}_{k+2} H\) and \(m= 3+\lfloor \alpha \rfloor -k\) to get
where we take \(B_i =\mathcal {D}_i\) for k even and \(B_i=\bar{\mathcal {D}}_i\) for k odd. Then it is easy to see that the first sum is bounded by our energy since \(k+4\leqq N\). For the second sum, by using the product rule and smoothness of w, we first note that \(|B_i \mathcal {D}_k (w^k\mathcal {D}_{k+2} H) |^2 \lesssim \sum _{j=0}^{i} w^{2k-2j} | W_j H |^2 \) where \(W_j\in \mathcal {P}_{k+2+i-j}\). Therefore, by further using Lemma A.3, the second sum is bounded by
Note, however, that \(w^{\alpha + \lfloor \alpha \rfloor + 6 -2j }= w^{\alpha +k + 2+i -j + (4+ \lfloor \alpha \rfloor -k -j) } \lesssim w^{\alpha +k + 2+i -j }\) because \(j\leqq i \leqq 4+ \lfloor \alpha \rfloor -k\). Furthermore, the total number of derivatives appearing is \(k+2+ i \leqq 6+\lfloor \alpha \rfloor =N \). Hence we obtain the desired bound. Other cases in (C.433) and (C.435) follow in the same way. (C.434) and (C.436) can be obtained similarly by applying (C.428) instead of (C.425). \(\square \)
D Local-in-Time Well-Posedness
Let \(\kappa >0\) be a sufficient small fixed number. In this section, we discuss the existence of H solving (3.21) in \([\kappa , T]\) with \(S_\kappa ^N(\tau )<\infty \) for all \(\tau \in [\kappa , T]\) for some time \(\kappa <T\leqq 1\).
Proposition D.1
Let \(1<\gamma <\frac{4}{3}\), assume that the physical vacuum condition (1.5) is satisfied, and let \(N=\lfloor \alpha \rfloor +6\). If \((H_0^\kappa ,H_1^\kappa )\) satisfy \(S_\kappa ^N(H_0^\kappa ,H_1^\kappa )\le \sigma ^2\), there exists a time \(T=T(\sigma )>\kappa \) and a unique solution \(\tau \rightarrow (H(\tau ,\cdot ))\) of the initial value problem (3.21) such that the map \([\kappa ,T]\ni \tau \mapsto S_\kappa ^N(\tau ) \) is continuous and the solution satisfies the bound
where the constant \(\tilde{C}\) depends only on \(\varepsilon \) and \(\sigma \).
Sketch of the Proof of Proposition D.1
The proof of Proposition D.1 follows by the well-posedness proof for the compressible Euler system of Jang and Masmoudi [36, 37]. The argument of [36, 37] will render the existence theory based on an appropriate approximate scheme and a priori bounds. To apply the result of [36, 37], we will design the approximate scheme for \(\mathfrak {H}:= D_r H\) and H from \(\mathfrak {H}\). We construct \(j^{\mathrm{th}}\) approximations \((\mathfrak {H}_j,\partial _\tau \mathfrak {H}_j)\) and \((H_{j}, \partial _\tau H_{j})\) as follows. The first approximation for \(j=1\), we use the initial data: let \((\mathfrak {H}_{1}, \partial _\tau \mathfrak {H}_1 )=(D_rH_0^\kappa , D_r H_1^\kappa )\) and \((H_1, \partial _\tau H_1) = (H_0^\kappa , H_1^\kappa )\) where \(S_\kappa ^N(H_0^\kappa ,H_1^\kappa )\le \sigma ^2\). Inductively we obtain the approximate solutions \((j+1)^{\mathrm{th}}\) approximations as follows: For \(j\geqq 1\), let \((\mathfrak {H}_{j+1}, \partial _\tau \mathfrak {H}_{j+1})\) be the solution to the initial value problem for the following linear PDE:
Note that the schemes mimic the Equation (4.6) for \(i=1\). The subscript j implies that the coefficients appearing in the expression are evaluated by using \(H_j\), \(\partial _\tau H_j\). With the bounds \(S^N_\kappa (H_{j}, \partial _\tau H_j)<\infty \) depending only on \(\varepsilon \) and \(\sigma \), the existence of \((\mathfrak {H}_{j+1}, \partial _\tau \mathfrak {H}_{j+1})\) follows from the duality argument in [36, 37]. By defining \(H_{j+1}\) by
and based on a priori estimates, we also deduce \(S^N_\kappa (H_{j+1}, \partial _\tau H_{j+1})<\infty \), whose bound depends only on \(\varepsilon \) and \(\sigma \). As \(j\rightarrow \infty \), after extracting a subsequence, we obtain the limit \((H,\partial _\tau H)\) of \((H_{j}, \partial _\tau H_{j})\) that solves (3.21) in \([\kappa , T]\) for some \(T=T(\sigma )>\kappa \) with \(S^N_\kappa (H, \partial _\tau H)<\infty \). \(\square \)
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Guo, Y., Hadžić, M. & Jang, J. Continued Gravitational Collapse for Newtonian Stars. Arch Rational Mech Anal 239, 431–552 (2021). https://doi.org/10.1007/s00205-020-01580-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00205-020-01580-w