1: %\input EPSfig.macros
2: \input amstex
3: \input eqnumams
4: %\proofmodetrue
5: % \forwardreferencetrue
6: % \initialeqmacro
7: % \initialeqmacro
8: % If you want to forward reference, you also must put an empty file
9: % named labelfil.tex in the directory in which you are working.
10: \documentstyle{amsppt}
11: \overfullrule=0pt
12: \magnification =\magstep1
13: \baselineskip=18pt
14: \vcorrection{-.33truein}
15: \pageheight{9.0truein}
16: %\input EPSfig.macros
17: \input amssym.def
18: %%%%%%%%%%
19: %%%%%%%%%%
20: \def\BU{\bar U}
21: \def\BG{\bar G}
22: \def\tG{\tilde{\CalG}}
23: \def\ratio{\bar \alpha}
24: \def\Tpsi{{\tilde{\psi}}}
25: \def\tvar{{\tilde{\phi}}}
26: \def\bvarphi{{\bar{\varphi}}}
27: \def\bpsi{{\bar{\psi}}}
28: \def\blambda{{\bar{\lambda}}}
29: \def\bDelta{{\bar{\Delta}}}
30: \def\balpha{{\bar{\alpha}}}
31: \def\ratio{\bar \alpha}
32: \def\etaT{{\widetilde{\eta}}}
33: \def\Tmu{{\widetilde{\mu}}}
34: \def\tildev{{\widetilde{v}}}
35: \def\tildeN{{\widetilde{N}}}
36: \def\tildeM{{\widetilde{M}}}
37: \def\tildeW{{\widetilde{w}}}
38: \def\tildeC{{\widetilde{C}}}
39: \def\Wbar{{\bar{w}}}
40: \def\Ubar{{\bar{U}}}
41: \def\Sbar{{\bar{\Cal{S}}}}
42: \def\BbbR{\Bbb{R}}
43: \def\BbbC{\Bbb{C}}
44: \def\BbbD{\Bbb{D}}
45: \def\BbbA{\Bbb{A}}
46: \def\BbbB{\Bbb{B}}
47: \def\BbbJ{\Bbb{J}}
48: \def\BbbK{\Bbb{K}}
49: \def\BbbN{\Bbb{N}}
50: \def\CalO{\Cal{O}}
51: \def\CalA{\Cal{A}}
52: \def\CalD{\Cal{D}}
53: \def\CalG{\Cal{G}}
54: \def\CalH{\Cal{H}}
55: \def\CalL{\Cal{L}}
56: \def\CalR{\Cal{R}}
57: \def\CalS{\Cal{S}}
58: \def\CalT{\Cal{T}}
59: \def\CalU{\Cal{U}}
60: \def\CalV{\Cal{V}}
61: \def\CalW{\Cal{W}}
62: \def\bU{{\bar{u}}}
63: \def\bv{{\bar{v}}}
64: \def\R{\roman{Re\ }}
65: \def\I{\roman{Im\ }}
66: \def\sgn{\text{\rm sgn\ }}
67: \def\tr{\text{\rm Tr\ }}
68: \def\div{\text{\rm div\ }}
69: \def\det{\text{\rm det\ }}
70: \def\deff{\text{$:=$}}
71: \def\ess{\text{\rm ess\ }}
72: \def\Range{\text{\rm Range\ }}
73: \def\CoRange{\text{\rm CoRange\ }}
74: \def\Residue{\text{\rm Res}}
75: \def\Span{\text{\rm Span\ }}
76: \def\stab{{($\Cal{D}$)}}
77: \def\mstab{{($\Cal{D}_\xi$)}}
78: \def\factorial{!}
79: \def\dim{\text{\rm dim \ }}
80: \def\adj{\text{\rm adj\ }}
81: \def\diag{\,\text{\rm diag}\,}
82: \def\Re{\text{\rm Re\ }}
83: \def\errfn{\text{\rm errfn\ }}
84: \def\Im{\text{\rm Im }}
85: \def\deltdot{{\cdot \atop {\raise8pt\hbox{$\delta$}}}}
86: \def\myqed{\vrule height3pt depth2pt width3pt \bigskip}
87: \def\newsection{\centerline}
88: \def\cal{\Cal}
89: \def\exp{{\text{\rm exp\ }}}
90: \def\Loc{\text{\rm Loc\ }}
91: \def\real{{min}}
92: \def\bfO{{\Cal{O}}}
93:
94: \def\zigzag{\buildrel > \over <}
95: \def\ttalpha{\buildrel \approx \over \alpha}
96:
97: % definitions introduced by Corrado
98:
99: \def\Lop{L}
100: \def\spn{\text{span }}
101: \def\Fyx{{\Cal F}^{y\to x}\,}
102: \def\Fzx{{\Cal F}^{z\to x}\,}
103: \def\Fzy{{\Cal F}^{z\to y}\,}
104: \def\e{\varepsilon}
105:
106: %%%%%
107: \pagewidth{6truein}
108: \pageheight{9truein}
109: %%%%%
110: \topmatter
111: \title
112: {Stability of viscous shock profiles for dissipative symmetric
113: hyperbolic-parabolic systems}
114: \endtitle
115: \leftheadtext{Stability of viscous shock profiles}
116: \rightheadtext{Corrado Mascia and Kevin Zumbrun}
117: %
118: \thanks
119: The second author thanks
120: Instituto per le Applicazioni del Calcolo
121: ``M. Picone'' (CNR) and European TMR Project ``Hyperbolic Systems of
122: Conservation Laws'' for their hospitality and for making possible
123: the visit (April 10--May 10, 2000)
124: during which this work was carried out.
125: Research of the first author was supported in part by
126: European TMR Project ``Hyperbolic Systems of Conservation Laws''.
127: Research of the second author was supported
128: in part by the National Science Foundation under Grants No. DMS-9107990
129: and DMS-0070765.
130: \endthanks
131: %
132: \abstract
133: Combining pointwise Green's function bounds obtained
134: in a companion paper [MZ.2] with earlier, spectral stability
135: results obtained in [HuZ],
136: we establish nonlinear orbital stability
137: %We establish nonlinear orbital stability
138: %with respect to small $L^1\cap H^3$ perturbations
139: of small amplitude viscous shock profiles for the class of
140: dissipative symmetric
141: hyperbolic-parabolic systems identified by Kawashima [Kaw],
142: notably including compressible Navier--Stokes
143: equations and the equations of magnetohydrodynamics,
144: obtaining sharp rates of decay in
145: $L^p$ with respect to small $L^1\cap H^3$ perturbations,
146: $2\le p\le \infty$.
147: %For large amplitude profiles, and somewhat more general systems,
148: %we show that linearized $L^1\cap L^p\to L^p$ orbital
149: %stability is equivalent to strong spectral stability plus
150: %hyperbolic stability of the associated ideal shock.
151: Our analysis follows the approach introduced in
152: [MZ.1] to treat stability of relaxation profiles.
153: \endabstract
154: %
155: \author
156: {Corrado Mascia and Kevin Zumbrun}
157: \endauthor
158: %
159: \date{August 3, 2001}
160: %TODO: switch format when appropriate:
161: %\date{August 2, 2001; Revised: August 3, 2001}
162: \enddate
163: %
164: \address
165: Dipartimento di Matematica ``G. Castelnuovo'',
166: Universit\`a di Roma ``La Sapienza'',
167: P.le Aldo Moro, 2 - 00185 Roma (ITALY)
168: \endaddress
169: %
170: \email
171: mascia\@ mat.uniroma1.it
172: \endemail
173: %
174: \address
175: Department of Mathematics,
176: Indiana University,
177: Bloomington, IN 47405-4301
178: \endaddress
179: %
180: \email
181: kzumbrun\@indiana.edu
182: \endemail
183: %
184: \endtopmatter
185: %%%%
186: \document
187: \newsection {\bf Section 1. Introduction}
188: \sectionnumber=1
189: \theoremnumber=0
190: \equationnumber=0
191: \smallskip
192: \TagsOnLeft
193:
194: %
195: Consider the class of degenerate parabolic conservation laws
196: of {\it dissipative, symmetric hyperbolic--parabolic type} in
197: the sense of Kawashima [Kaw], i.e., systems
198: $$
199: G(\tilde U)_t + F(\tilde U)_x = (B(\tilde U) \tilde U_x)_x,
200: \eqnlbl{general}
201: $$
202: satisfying
203: %
204: \medskip
205: (A1)\quad
206: $dF$, $dG$, $B$ symmetric, $dG>0$, $B\ge 0$
207: (symmetric hyperbolic--parabolicity),
208: \medskip
209: \medskip
210: (A2)\quad
211: No eigenvector of $dF dG^{-1}$ lies in the kernel of $BdG^{-1}$ (dissipativity),
212: \medskip
213: \noindent and
214: \medskip
215: (A3) \quad
216: $
217: \tilde U=\pmatrix \tilde u \\ \tilde v\endpmatrix$,
218: \quad
219: $
220: B=
221: \pmatrix 0 & 0 \\ 0 & b \endpmatrix
222: $,
223: \quad
224: $b>0$
225: (block structure),
226: \medskip
227: \noindent
228: in some neighborhood $\Cal{U}$ of a particular base point $\tilde U_*$,
229: where $\tilde U\in \BbbR^n$,
230: $\tilde u\in \BbbR^{n-r}$, $\tilde v\in \BbbR^r$,
231: and $b\in \BbbR^{r\times r}$.
232: As discussed in [Kaw], this class of equations includes many physical
233: models arising in continuum mechanics: in particular,
234: compressible Navier--Stokes equations and
235: the equations of compressible magnetohydrodynamics (MHD).
236: We make also the regularity assumption
237: \medskip
238: (H0) \quad $F$, $G$, $B\in C^3$,
239: \medskip
240: \noindent needed for our later analysis.
241:
242: An interesting class of solutions of \eqnref{general} are
243: {\it viscous shock profiles}, or asymptotically constant
244: traveling wave solutions
245: $$
246: \tilde U=\bar U(x-st);
247: \quad
248: \lim_{z\to \pm \infty} \bar U(z)=\tilde U_\pm
249: \eqnlbl{profile}
250: $$
251: connecting endstates $\tilde U_\pm$ corresponding
252: to (discontinuous) shock waves of the associated hyperbolic system
253: $$
254: G(\tilde U)_t + F(\tilde U)_x = 0.
255: \eqnlbl{hyperbolic}
256: $$
257: Existence of viscous profiles has been treated in the large
258: for compressible Navier--Stokes by, e.g.,
259: %Boethe and Gilbarg [Bo,Gi]
260: %TODO: was it Weyl, Boethe, or ???? check mathrev..
261: Gilbarg [Gi]
262: and for MHD by Conley and Smoller [CS] (see [Fre.1] for a treatment
263: in the small amplitude case).
264: Most recently, Freist\"uhler [Fre.2] has carried out a general treatment
265: of existence in the small,
266: analogous to that of Majda and Pego in the strictly parabolic case [MP],
267: for arbitrary models \eqnref{general} satisfying (A1)--(A3).
268:
269: Let
270: $$
271: a_1(u)\le \dots \le a_n(u)
272: $$
273: denote the eigenvalues of $A_G:=dFdG^{-1}(\tilde U)$,
274: $r_j(\tilde U)$ and $l_j(\tilde U)$ a smooth
275: choice of associated right and left eigenvectors, $l_j\cdot r_k=\delta^j_k$,
276: and assume
277: %at the base point $\tilde U_*$ that:
278: within neighborhood $\Cal{U}$ of base point $\tilde U_*$ that:
279:
280: \medskip
281: (H1) \quad The $p$th characteristic
282: field is of multiplicity one, i.e. $a_p(\tilde U)$ is a simple
283: eigenvalue of $A_G(\tilde U)$.
284:
285: (H2) \quad The $p$th characteristic field is genuinely
286: nonlinear, i.e. $\nabla a_p\cdot r_p (\tilde U)\ne 0$.
287: \medskip
288: (H3)\quad
289: The eigenvalues of $dF_{11}dG_{11}^{-1}$
290: (necessarily real and semisimple) are: (i) of constant
291: multiplicity; and, (ii) different from $a_p$.
292: \medskip
293: \noindent
294: Then, we have:
295:
296: \proclaim{Proposition \thmlbl{freist1} ([Fre.2])}
297: Let (A1)--(A3), and (H1)--(H3) hold, and $F$, $G$, $B\in C^2$
298: in \eqnref{general} (implied in particular by (H0)).
299: Then, for left and right
300: states $\tilde U_\pm$ lying within a sufficently small neighborhood
301: $\CalV\subset \Cal{U}$ of $\tilde U_*$, and speeds $s$ lying within
302: a sufficiently small neighborhood of $a_p(\tilde U_*)$,
303: there exists a viscous profile \eqnref{profile} that is
304: ``local'' in the sense that the image of $\bar u(\cdot)$ lies
305: entirely within $\Cal{V}$ if and only if the triple $(\tilde U_-,\tilde U_+,s)$
306: satisfies both the Rankine--Hugoniot relations:
307: $$
308: s[G]=[F],
309: \tag{RH}
310: $$
311: and the Lax characteristic conditions for a $p$-shock:
312: $$
313: a_p(\tilde U_-)>s>a_p(\tilde U_+); \quad \sgn (a_j(\tilde U_-)-s)=\sgn (a_j(\tilde U_+)-s)\ne 0 \text{ for }
314: j\ne p.
315: \tag{L}
316: $$
317: (Note: The structure theorem of Lax [La,Sm] implies that (RH), always a
318: {\it necessary} condition for existence of profiles, holds for
319: $\tilde U_\pm\in \Cal{V}$ only if $s$ lies near some $a_j(\tilde U_*)$;
320: thus, the restriction on speed $s$ is only the assumption that the
321: triple $(\tilde U_-,\tilde U_+,s)$ be associated with the $p$th and not some
322: other characteristic field).
323: \endproclaim
324:
325: {\bf Remark \thmlbl{ideal}.}
326: The generically satisfied conditions
327: (H1) and (H2) are implied by strict hyperbolicity
328: and genuine nonlinearity, respectively, of the associated
329: hyperbolic system \eqnref{hyperbolic}.
330: In particular, they hold always for the equations of compressible
331: gas dynamics, under the assumption of an ideal gas [Sm].
332: Condition (H3)(ii) corresponds to the requirement that ``hyperbolic,''
333: or unsmoothed modes in the solution be noncharacteristic,
334: at least as regards the principal characteristic speed $a_p$.
335: This technical condition is needed in order that
336: the traveling wave ODE be of nondegenerate type, and is a standard
337: assumption in the theory.
338: See assumption (4) of [Fre.2], assumption ($\tilde{\text{\rm H}}$1)(ii) of [Z.3],
339: Appendix A.2, or assumption (H5) of [SZ] for restatements in
340: various different contexts.
341: Condition (H3)(i) is an additional technical assumption
342: that was used in the detailed Green's function analysis carried out
343: in [MZ.2] for slightly more general systems; we suspect that it can
344: be dropped in the present, symmetrizable case.
345: Conditions (H3)(i)--(ii) are likewise satisfied for compressible
346: ideal gas dynamics, for which $dF_{11}dG_{11}^{-1}$ is $1\times 1$
347: and identically equal to particle velocity ($u$ in Eulerian
348: coordinates, $0$ in Lagrangian coordinates).
349: \medskip
350:
351: Stability of viscous profiles has been examined for compressible
352: Navier--Stokes equations in [MN,KMN,L.2], vith various partial results
353: concerning special (mainly zero-mass) initial data;
354: the results of [KMN,L.2] are restricted to small amplitude profiles, while
355: the results of [MN] in the case of an isentropic $\gamma$-law gas
356: apply to profiles of amplitude $\alpha(\gamma)$, with
357: $\alpha\to \infty$ as $\gamma \to 1$.
358: Yet, as pointed out recently in the fluid-dynamical survey [Te], the basic
359: problem of stability of gas-dynamical shocks in its full generality
360: remains open even for small amplitude waves,
361: a significant gap in the theory of compressible gas dynamics.
362: More recently, Humpherys and Zumbrun [HuZ] have established {\it strong
363: spectral stability} of general, small amplitude shock waves of Kawashima
364: class systems, of the type constructed by Freist\"uhler, generalizing
365: a corresponding result of Goodman [Go.1--2] in the strictly parabolic
366: case (see Section 2, below).
367: As discussed in [ZH,HuZ], strong spectral stability is roughly equivalent
368: to, but slightly weaker than stability with respect to zero-mass perturbations.
369:
370: %Applying the spectral machinery developed in [ZH,MZ.1],
371: In [MZ.2], applying the general machinery developed in [ZH,MZ.1],
372: we have shown, for a slightly more general class of systems and for
373: profiles of arbitrary amplitude and type,
374: that strong spectral stability, plus hyperbolic stability of the corresponding
375: ideal shock (always satisfied for weak Lax shocks satisfying (H1)--(H3)),
376: are necessary and sufficient conditions for linearized orbital stability,
377: and, moreover, yield extremely detailed pointwise bounds on the Green's
378: function of the linearized evolution equations, analogous to those
379: obtained for relaxation profiles in [MZ.1].
380: In combination with the result of [HuZ], these results imply
381: $L^1\cap L^p\to L^p$ {\it linearized orbital stability} of small-amplitude
382: shock profiles of Kawashima class systems, with sharp rates of decay
383: for all $1\le p\le \infty$.
384:
385: The purpose of the present paper, extending the work of [HuZ,MZ.2],
386: is to establish $L^1\cap H^3\to L^p$ {\it nonlinear orbital stability}
387: as solutions of \eqnref{general} of
388: small amplitude profiles of general Kawashima class systems,
389: with sharp rates of decay for all $1\le p\le \infty$.
390: More precisely, we shall establish:
391:
392: \proclaim{Theorem \thmlbl{nonlin}}
393: Let there hold (A1)--(A3) and (H0)--(H3) for
394: a general relaxation model \eqnref{general},
395: with $\tilde U_*$ and $\Cal{U}$ as above.
396: Then, for $\Cal{V}\subset \Cal{U}$ sufficiently small,
397: %NOTE: possibly smaller than in Freist. result..
398: the viscous profiles $\Bar U$ described in Proposition
399: \thmref{freist1} are nonlinearly orbitally stable
400: from $L^1\cap H^3$ to $L^p$, for all $p\ge 2$.
401: %
402: More precisely, for initial perturbations $U_0:=\tilde U_0-\bar U$
403: such that $|U_0|_{L^1\cap H^3}\le \zeta_0$,
404: $\zeta_0$ sufficiently small,
405: the solution $\tilde U=(\tilde u,\tilde v) (x,t)$ of \eqnref{general} with
406: initial data $\tilde U_0$ satisfies
407: $$
408: |\tilde U(x,t)-\bar U(x-\delta(t))|_{L^p}\le C \zeta_0
409: (1+t)^{-\frac{1}{2}(1-1/p)}
410: \eqnlbl{3.31}
411: $$
412: and
413: $$
414: |\left(\tilde v(x,t)-\bar v(x-\delta(t))\right)_x|_{L^p}\le C \zeta_0
415: (1+t)^{-\frac{1}{2}(1-1/p)},
416: \eqnlbl{deriv3.31}
417: $$
418: for all $2\le p\le \infty$, for some $\delta(t)$ satisfying
419: $$
420: |\dot \delta (t)|\le C \zeta_0 (1+t)^{-\frac{1}{2}}
421: \eqnlbl{3.32}
422: $$
423: and
424: $$
425: |\delta(t)|\le C \zeta_0.
426: \eqnlbl{3.33}
427: $$
428: \endproclaim
429:
430: \noindent In particular,
431: this implies nonlinear stability
432: of all small amplitude gas-dynamical profiles under the
433: assumptions of an ideal gas;
434: see Remark \thmref{ideal} above.
435:
436: Our analysis in this paper, and in the companion paper [MZ.2],
437: follows closely to that used in [MZ.1] to
438: treat stability of relaxation profiles, making use of
439: structural similarities between the two problems
440: pointed out in [Ze.2,Z.3];
441: in turn, the argument of [MZ.1] makes substantial use of the
442: pointwise semigroup machinery introduced in [ZH,Z.2] to
443: treat the parabolic case.
444: As are general features of the pointwise semigroup approach,
445: %we obtain in the course of this analysis extremely detailed
446: we obtain through this program extremely detailed
447: pointwise bounds on the Green's function (more properly,
448: distribution) of the linearized operator about the wave,
449: sufficiently strong that the nonlinear stability analysis
450: becomes comparatively simple.
451: Moreover, these bounds (though not the nonlinear stability argument
452: in which they are used, see Remark 1.5 just below)
453: depend only on a generalized spectral stability
454: (i.e., Evans function) condition and not on the amplitude or
455: type (Lax, under-, or overcompressive) of the wave.
456:
457: {\bf Remark \thmlbl{ngnl}.} The assumption of genuine nonlinearity (H2)
458: is not needed either for the existence or the stability result,
459: but is made only to simplify the discussion.
460: Though we stated above only the restriction
461: to the genuinely nonlinear case,
462: existence was in fact treated for the general (nongenuinely nonlinear)
463: case in [Fre.2], substituting for the Lax entropy condition (L)
464: the strict Liu entropy inequality (E) of [L.3].
465: Likewise, as described in [HuZ], the result of spectral stability
466: may be extended to the general case by substituting for the
467: ``Goodman-type'' weighted energy estimates of [HuZ], Section 5,
468: the variation introduced by Fries [Fri.1--2]
469: to treat the nongenuinely nonlinear case for strictly parabolic viscosities.
470: Accordingly, the result of Theorem \thmref{nonlin} holds also in this
471: case.
472: %
473: \medskip
474: {\bf Remark \thmlbl{gammalaw}.} The restriction to small-amplitude
475: shocks arises only through the energy estimates used to close the
476: nonlinear iteration argument.
477: In particular, it should be possible for the isentropic, $\gamma$-law
478: gas case to treat amplitudes of the same order $\alpha(\gamma)$ treated in [MN].
479: A fundamental open problem is to remove the amplitude
480: restriction altogether, as done in the
481: strictly parabolic case [ZH,Z.2] and in the case of discrete kinetic
482: relaxation models [MZ.1], replacing it with
483: the generalized spectral condition of the linearized theory.
484: \medskip
485:
486: {\bf Remark \thmlbl{nec}.}
487: Useful necessary conditions for viscous stability have been obtained
488: in [Z.3] for arbitrary amplitude profiles of the more general class
489: of models considered in [MZ.2], using the stability index of [GZ,BSZ].
490: Strengthened versions of the one-dimensional inviscid stability criteria
491: of Erpenbeck--Majda [Er,M.1--3], these
492: readily yield examples of unstable large-amplitude profiles,
493: similarly as in the strictly parabolic case (see,
494: e.g., [GZ,FreZ,ZS,Z.3]).
495: This shows that the spectral stability requirement is not vacuous
496: in the large amplitude case.
497:
498: \medskip
499: {\bf Plan of the paper.}
500: In Section 2 we cite the spectral stability result of [HuZ],
501: and in Section 3 the pointwise Green's function bounds of [MZ.2].
502: As an immediate corollary, we establish in Section 4 the
503: linearized orbital stability of general shock profiles satisfying
504: the necessary conditions of spectral and hyperbolic stability:
505: in particular, of small amplitude profiles of Kawashima class systems.
506: Finally, we establish in Section 5 the main result of
507: nonlinear stability of weak profiles of Kawashima class systems,
508: by a modified version of the argument of [MZ.1].
509:
510: \bigskip
511: {\bf Note}:
512: Liu and Zeng have informed us [LZe.2] that they also have obtained
513: nonlinear stability of weak Navier--Stokes profiles,
514: by a different argument based on the approximate
515: Green's function approach of [L.2].
516: This method is inherently limited to weak shocks of classical, Lax type,
517: both at the linear and nonlinear level.
518: \bigskip
519:
520: %\newpage
521: %TODO: remove if space too big...
522: \newsection {\bf Section 2. Spectral stability.}
523: \sectionnumber=2
524: \theoremnumber=0
525: \equationnumber=0
526: \smallskip
527: \TagsOnLeft
528: \bigskip
529:
530: Take without loss of generality $s=0$, so that
531: $\tilde U=\bar U(x)$ becomes a stationary solution.
532: Then, the linearized equations of \eqnref{general} about
533: $\bar U$ take the form
534: $$
535: U_t =LU:= -(A^0)^{-1}(AU)_x + (A^0)^{-1}(BU_x)_x,
536: \eqnlbl{linearized}
537: $$
538: where
539: $$
540: B:= B(\bar U),
541: \quad A^0:=dG(\bar U),
542: \quad Av:= dF(\bar U)v
543: - (dB v)\bar U_x.
544: ,
545: \eqnlbl{AB}
546: $$
547:
548: {\bf Definition \thmlbl{stable}.} We call the profile $\bar U(\cdot)$
549: {\it strongly spectrally stable} if the linearized operator
550: $L$ about the wave has no spectrum in the closed unstable
551: complex half-plane $\{\lambda: \, \text{\rm Re }\lambda \ge 0\}$
552: except at the origin, $\lambda=0$.
553: (Recall, [Sat], that $\lambda=0$ is always in the spectrum of $L$,
554: since $L\bar U_x=0$ by direct calculation/differentiation of
555: the traveling wave ODE).
556: \medskip
557:
558: The spectral stability of small-amplitude viscous profiles of
559: the slightly more more general class of dissipative, {\it symmetrizable}
560: degenerate parabolic systems has been investigated in [HuZ].
561: To apply these results, we have only to note that \eqnref{general},
562: expressed with respect to variable $\tilde G:=G(\tilde U)$, becomes
563: $$
564: \tilde G_t + F(\tilde U(\tilde G))_x= (B(\tilde U(\tilde G))\tilde U(\tilde G)_x)_x,
565: \eqnlbl{symmetrizable1}
566: $$
567: or, in quasilinear form,
568: $$
569: \tilde G_t + F_{\tilde U}G_{\tilde U}^{-1}\tilde G_x = (B G_{\tilde U}^{-1} \tilde G_x)_x.
570: \eqnlbl{symmetrizable}
571: $$
572: System \eqnref{symmetrizable} is evidently symmetrizable
573: (by the symmetric positive definite $G_{\tilde U}^{-1}$), and
574: {\it dissipative} (since the dissipativity condition is independent
575: of coordinate system), with the additional {\it block structure}
576: property that the left kernel of the new viscosity matrix
577: $B G_{\tilde U}^{-1}$ is constant.
578: This is almost the class of equations considered in [HuZ], the
579: difference being that there the block structure assumption was
580: that the viscosity matrix have constant right instead of left
581: kernel. This discrepancy is unimportant in the analysis, since,
582: at the linearized level, one case may be converted to the
583: other by the (linear) change of coordinates
584: $A^0 V:=U$, corresponding to similarity
585: transform $M\to (A^{0})^{-1} M A^0$.
586: Alternatively, in the particular case of our interest, we may simply
587: carry out energy estimates in the natural coordinates of \eqnref{linearized};
588: for related calculations, see Section 4.1, below.
589:
590: Thus, the results of [HuZ] apply to the somewhat larger class
591: of dissipative symmetrizable systems of form
592: \eqnref{symmetrizable1}--\eqnref{symmetrizable1},
593: with the block structure condition that {\it either} the left or
594: right kernel of the viscosity matrix be constant, and we may
595: conclude, in particular:
596:
597:
598: \proclaim{Theorem \thmlbl{HuZ}[HuZ]}
599: Let (A1)--(A3) and (H0)--(H3) hold,
600: and let $\bar U(x-st)$ be a viscous shock solution of \eqnref{general}
601: such that the profile $\{\bar u(z)\}$ lies entirely within a sufficently
602: small neighborhood $\CalV\subset \Cal{U}$ of $\tilde U_*$, and the speed
603: $s$ lies within a sufficiently small neighborhood of $a_p(\tilde U_*)$.
604: Then, $\bar U$ is strongly spectrally stable, in the sense of Definition
605: \thmref{stable} above.
606: \endproclaim
607: \bigskip
608:
609:
610: \newsection {\bf Section 3. Pointwise Green's function bounds.}
611: \sectionnumber=3
612: \theoremnumber=0
613: \equationnumber=0
614: \smallskip
615: \TagsOnLeft
616: \bigskip
617:
618: In a companion paper to this one [MZ.1], we have investigated linearized
619: stability and behavior of viscous profiles of systems of the general
620: form \eqnref{symmetrizable}, not necessarily symmetrizable, satisfying
621: the standard set of conditions identified in [Z.3].
622: As remarked in [Z.3,MZ.1], these hold always for symmetric systems \eqnref{general}
623: satisfying (A1)--(A3), (H0)--(H3), and the small-amplitude profiles described in
624: Proposition \thmref{freist1};
625: however, they may also hold in much greater generality, in particular for
626: shocks of large amplitude, or nonclassical type.
627: For simplicity of exposition, we shall restrict our discussion here to the
628: present case of interest; for the general case, see [MZ.1].
629:
630:
631: The first main result of [MZ.2], generalizing the corresponding
632: results established for viscous, strictly parabolic, shocks in [ZH],
633: and for relaxation shocks in [MZ.1], is:
634:
635: \proclaim{Theorem \thmlbl{D}[MZ.2]}
636: Under assumptions (A1)--(A3), (H0)--(H3),
637: small amplitude shock profiles are $L^1\cap L^p\to L^p$
638: linearly orbitally stable for $p>1$ if and only if
639: they are strongly spectrally stable, with sharp decay bounds
640: $$
641: |U(\cdot, t)+\delta(t)\bar U'(\cdot)|_{L^p}\le
642: C(1+t)^{-\frac{1}{2}(1-1/p)}(|U_0|_{L^1}+ |U_0|_{L^p})
643: \eqnlbl{lindecaybound1}
644: $$
645: for initial data $U_0\in L^1\cap L^p$, some choice of $\delta$.
646: \endproclaim
647:
648: Theorem \thmref{D} is obtained (see, e.g., Corollary 4.2, below for
649: one direction) as a consequence of
650: detailed, pointwise bounds on the Green's function
651: (more properly speaking, distribution)
652: $\CalG(x,t;y)$ of the linearized evolution equations
653: $$
654: G_t= L_G G
655: :=- (A_GG)_x + (B_G G_x)_x,
656: \eqnlbl{linsymmetrizable}
657: $$
658: $$
659: %\aligned
660: B_G:= B (A^0)^{-1}(\bar U(x)),
661: %TODO: check that A^0 defined above... ref it?
662: \quad
663: A_Gv:= dF (A^0)^{-1} v - dB(\bar U(x))v \bar U'(x),
664: %\endaligned
665: \eqnlbl{Gcoeffs}
666: $$
667: corresponding to \eqnref{symmetrizable}, i.e., expressed
668: with respect to the conservative variable $G$.
669: We now describe these bounds, for use in the following sections.
670:
671: Let $a_j^\pm$, $j=1,\dots (n)$ denote the eigenvalues of
672: $A_G(\pm\infty)$, and $l_j^\pm$ and $r_j^\pm$
673: associated left and right eigenvectors, respectively,
674: normalized so that $l_j^{\pm t} r_k^{\pm}=\delta^j_k$.
675: %
676: In case $A_G^\pm$ is strictly hyperbolic, these are uniquely defined.
677: In the general case, we require further that $l_j^\pm$, $r_j^\pm$
678: be left and right eigenvectors also of $P_j^{\pm } B_G^\pm P_j^\pm$,
679: $P_j^\pm:=R_j^\pm L_j^{\pm t}$, where
680: $L_j^\pm$ and $R_j^\pm$ denote $m_j^\pm\times m_j^\pm$
681: left and right eigenblocks associated
682: with the $m_j^\pm$-fold eigenvalue $a_j^\pm$, normalized so that
683: $L_j^\pm R_j^\pm= I_{m_j^\pm}$.
684: (Note: The matrix $P_j^{\pm } B_G^\pm P_j^\pm\sim
685: L_j^{\pm t} B_G^\pm R_j^\pm$
686: is necessarily diagonalizable,
687: by simultaneous symmetrizability of $A_G$, $B_G$).
688:
689: Eigenvalues $a_j(x)$, and eigenvectors $l_j$, $r_j$
690: correspond to large-time convection rates and modes of propagation
691: of the degenerate model \eqnref{symmetrizable}.
692: Likewise, let $a^{*}_j(x)$, $j=1,\dots,(n-r)$
693: denote the eigenvalues of
694: %$[A_{G}]_{11}(x)\in \BbbR^{(n-r)\times(n-r)}$,
695: $$
696: \aligned
697: A^*_G&:= A_{G,11}- A_{G,12} B_{G,22}^{-1}B_{G,21}\\
698: &=
699: A_{11}(A^0_{11})^{-1} ,
700: %\in \BbbR^{(n-r)\times(n-r)},
701: \endaligned
702: \eqnlbl{A*}
703: $$
704: and $l^*_j(x)$, $r^*_j(x)\in \BbbR^{n-r}$
705: associated left and right eigenvectors,
706: normalized so that $l^{*t}_jr_j\equiv \delta^j_k$.
707: More generally, for an $m_j^*$-fold eigenvalue, we choose
708: $(n-r)\times m_j^* $ blocks $L_j^*$ and $R_j^*$ of eigenvectors
709: satisfying the {\it dynamical normalization}
710: $$
711: L_j^{*t}\partial_x R_j^{*}\equiv 0,
712: $$
713: along with the usual static normalization
714: $L^{*t}_jR_j\equiv \delta^j_kI_{m_j^*}$; as shown in Lemma 4.9, [MZ.1],
715: this may always be achieved with bounded $L_j^*$, $R_j^*$.
716: Associated with $L_j^*$, $R_j^*$, define extended, $n\times m_j^*$ blocks
717: $$
718: \CalL_j^*:=\pmatrix L_j^* \\ 0\endpmatrix,
719: \quad
720: \CalR_j^*:=
721: \pmatrix R_j^*\\
722: -B_{G,22}^{-1}B_{G,21} R_j^*\endpmatrix.
723: \eqnlbl{CalLR}
724: $$
725: %
726: %bridge here: these corr to...
727: Eigenvalues $a_j^*$ and eigenmodes $\CalL_j^*$, $\CalR_j^*$
728: correspond, respectively, to short-time hyperbolic characteristic speeds
729: and modes of propagation for the reduced, hyperbolic part of degenerate
730: system \eqnref{symmetrizable}.
731: Note that our discussion in the introduction of condition (H3) is
732: validated by the second equality in \eqnref{A*}, a
733: nontrivial consequence of symmetry/block structure in the
734: original system \eqnref{general}.
735:
736: Define time-asymptotic, {\it scalar diffusion rates}
737: $$
738: \beta_j^{\pm}:= \left(l_j^{t} B_G r_j\right)_\pm, \quad j=1,\dots, n,
739: \eqnlbl{beta}
740: $$
741: and local, $m_j\times m_j$ {\it dissipation coefficients}
742: $$
743: \eta_j^*(x):= -l_j^{*t} D^* r_j^* (x), \quad j=1,\dots,J\le n-r,
744: \eqnlbl{eta}
745: $$
746: where
747: $$
748: %NOTES:
749: %{D^*}:= f_v\Big[g_u-g_vq^{-1}_v q_u+ f^*_u q^{-1}_vq_u\Big],
750: {D^*}:= A_{G,12} \Big[A_{G,21}-A_{G,22}
751: B_{G,22}^{-1} B_{G,21}+ A^*_G
752: B_{G,22}^{-1} B_{G,21} \Big],
753: \eqnlbl{D*}
754: $$
755: is an effective dissipation precisely analogous to the
756: effective diffusion predicted by formal, Chapman--Enskog expansion
757: in the (dual) relaxation case.
758: As described in Appendix A2 of [MZ.2], these quantities arise in a natural
759: way, through Taylor expansion of the (frozen-coefficient) Fourier symbol
760: $$
761: -i\xi A_G(x)-\xi^2 B_G(x)
762: \eqnlbl{frozenx}
763: $$
764: of the linearized operator $L_G$ about $\xi=0$ and $\xi=\infty$, respectively.
765:
766: The important quantities $\eta_j^*$, $\beta_j$ were identified by
767: Zeng [Ze.1,LZe.1] in her study by Fourier transform techniques
768: of decay to {\it constant}
769: (necessarily equilibrium) {\it solutions}
770: $(\bar u, \bar v) \equiv (u_\pm,v_\pm)$
771: of relaxation systems, corresponding at the linearized level
772: to the study of the limiting equations
773: $$
774: U_t=L_{G}^\pm U:= -A_G^{\pm} U_x+ B_G^\pm U_{xx}
775: \eqnlbl{limiting}
776: $$
777: as $x\to\pm \infty$ of the linearized evolution equations
778: \eqnref{linsymmetrizable}.
779: As a consequence of dissipativity, (A2), we have (see, e.g., [Kaw,LZe.1])
780: that
781: $$
782: \beta_j^{\pm}>0, \quad
783: j=1,\dots,n
784: \eqnlbl{goodbeta}
785: $$
786: and
787: $$
788: \R \sigma(\eta_j^*(x))>0, \quad
789: j=1,\dots,J\le n-r.
790: \eqnlbl{goodeta}
791: $$
792:
793: \proclaim{Proposition \thmlbl{greenbounds}[MZ.2]}
794: For weak shock profiles, under assumptions (A1)--(A3), (H0)--(H3),
795: the Green's function $\CalG(x,t;y)$ associated with
796: the linearized evolution equations \eqnref{linsymmetrizable}
797: may in the Lax or overcompressive case be decomposed as
798: $$
799: \CalG(x,t;y)= H + E+ S + R,
800: \eqnlbl{ourdecomp}
801: $$
802: where, for $y\le 0$:
803: $$
804: \aligned
805: H(x,t;y)&:=
806: \sum_{j=1}^{J}
807: \CalR_j^*(x) \zeta_j^*(y,t)
808: \delta_{x-\bar a_j^* t}(-y)
809: \CalL_j^{*t}(y)\\
810: &=
811: \sum_{j=1}^{J} \CalR_j^*(x) \Cal{O}(e^{-\eta_0 t})
812: \delta_{x-\bar a_j^* t}(-y)
813: \CalL_j^{*t}(y),
814: \endaligned
815: \eqnlbl{multH}
816: $$
817: $$
818: \aligned
819: E&(x,t;y):=
820: %\\ &
821: \sum_{a_k^- > 0}
822: [c^{0}_{k,-}]
823: \bar G'(x)
824: l_k^{-t}
825: \left(\errfn\left(\frac{y+a_k^{-}t}{\sqrt{4\beta_k^{-}t}}\right)
826: -\errfn \left(\frac{y-a_k^{-}t}{\sqrt{4\beta_k^{-}t}}\right)\right),
827: \endaligned
828: \eqnlbl{E}
829: $$
830: and
831: $$
832: \aligned
833: S(x,t;y)&:=
834: \chi_{\{t\ge 1\}}
835: \sum_{a_k^{-}<0}r_k^{-} {l_k^{-}}^t
836: (4\pi \beta_k^-t)^{-1/2} e^{-(x-y-a_k^{-}t)^2 / 4\beta_k^{-}t}
837: \\
838: &+
839: \chi_{\{t\ge 1\}}
840: \sum_{a_k^{-} > 0} r_k^{-} {l_k^{-}}^t
841: (4\pi \beta_k^{-}t)^{-1/2} e^{-(x-y-a_k^{-}t)^2 / 4\beta_k^{-}t}
842: \left({e^x \over e^x+e^{-x}}\right)\\
843: &+
844: \chi_{\{t\ge 1\}}
845: \sum_{a_k^{-} > 0, \, a_j^{-} < 0}
846: [c^{j,-}_{k,-}]r_j^{-} {l_k^{-}}^t
847: (4\pi \bar\beta_{jk}^{-} t)^{-1/2} e^{-(x-z_{jk}^{-})^2 /
848: 4\bar\beta_{jk}^{-} t}
849: \left({e^{ -x} \over e^x+e^{-x}}\right),\\
850: &+
851: \chi_{\{t\ge 1\}}
852: \sum_{a_k^{-} > 0, \, a_j^{+} > 0}
853: [c^{j,+}_{k,-}]r_j^{+} {l_k^{-}}^t
854: (4\pi \bar\beta_{jk}^{+} t)^{-1/2} e^{-(x-z_{jk}^{+})^2 /
855: 4\bar\beta_{jk}^{+} t}
856: \left({e^{ x} \over e^x+e^{-x}}\right)\\
857: \endaligned
858: \eqnlbl{S}
859: $$
860: denote hyperbolic, excited, and scattering terms, respectively,
861: and $R$ denotes a faster decaying residual, satisfying:
862: $$
863: \aligned
864: R(x,t;y)&=
865: \Cal{O}(e^{-\eta(|x-y|+t)})\\
866: &+\sum_{k=1}^n
867: \Cal{O} \left( (t+1)^{-1/2} e^{-\eta x^+}
868: +e^{-\eta|x|} \right)
869: t^{-1/2}e^{-(x-y-a_k^{-} t)^2/Mt} \\
870: &+
871: \sum_{a_k^{-} > 0, \, a_j^{-} < 0}
872: \chi_{\{ |a_k^{-} t|\ge |y| \}}
873: \Cal{O} ((t+1)^{-1/2} t^{-1/2})
874: e^{-(x-a_j^{-}(t-|y/a_k^{-}|))^2/Mt}
875: e^{-\eta x^+}, \\
876: %SYMMETRIC
877: &+
878: \sum_{a_k^{-} > 0, \, a_j^{+}> 0}
879: \chi_{\{ |a_k^{-} t|\ge |y| \}}
880: \Cal{O} ((t+1)^{-1/2} t^{-1/2})
881: e^{-(x-a_j^{+} (t-|y/a_k^{-}|))^2/Mt}
882: e^{-\eta x^-}, \\
883: \endaligned
884: \eqnlbl{Rbounds}
885: $$
886: $$
887: \aligned
888: R_y(x,t;y)&=
889: \sum_{j=1}^J \Cal{O}(e^{-\eta t})\delta_{x-\bar a_j^* t}(-y)
890: +
891: \Cal{O}(e^{-\eta(|x-y|+t)})\\
892: &+\sum_{k=1}^n
893: \Cal{O} \left( (t+1)^{-1/2} e^{-\eta x^+}
894: +e^{-\eta|x|} \right)
895: %TIMES?
896: t^{-1}
897: e^{-(x-y-a_k^{-} t)^2/Mt} \\
898: &+
899: \sum_{a_k^{-} > 0, \, a_j^{-} < 0}
900: \chi_{\{ |a_k^{-} t|\ge |y| \}}
901: \Cal{O} ((t+1)^{-1/2} t^{-1})
902: e^{-(x-a_j^{-}(t-|y/a_k^{-}|))^2/Mt}
903: e^{-\eta x^+} \\
904: %SYMMETRIC
905: &+
906: \sum_{a_k^{-} > 0, \, a_j^{+} > 0}
907: \chi_{\{ |a_k^{-} t|\ge |y| \}}
908: \Cal{O} ((t+1)^{-1/2} t^{-1})
909: e^{-(x-a_j^{+}(t-|y/a_k^{-}|))^2/Mt}
910: e^{-\eta x^-}, \\
911: \endaligned
912: \eqnlbl{Rybounds}
913: $$
914: $$
915: \aligned
916: R_x(x,t;y)&=
917: \sum_{j=1}^J \Cal{O}(e^{-\eta t})\delta_{x-\bar a_j^* t}(-y)
918: +
919: \Cal{O}(e^{-\eta(|x-y|+t)})\\
920: &+\sum_{k=1}^n
921: \Cal {O} \left( (t+1)^{-1} e^{-\eta x^+}
922: +e^{-\eta|x|} \right)
923: %TIMES?
924: t^{-1} (t+1)^{1/2}
925: e^{-(x-y-a_k^{-} t)^2/Mt} \\
926: &+
927: \sum_{a_k^{-} > 0, \, a_j^{-} < 0}
928: \chi_{\{ |a_k^{-} t|\ge |y| \}}
929: \Cal{O}(t+1)^{-1/2} t^{-1})
930: e^{-(x-a_j^{-}(t-|y/a_k^-|))^2/Mt}
931: e^{-\eta x^+} \\
932: %SYMMETRIC
933: &+
934: \sum_{a_k^{-} > 0, \, a_j^{+} > 0}
935: \chi_{\{ |a_k^{-} t|\ge |y| \}}
936: \Cal{O}(t+1)^{-1/2} t^{-1})
937: e^{-(x-a_j^{+}(t-|y/a_k^{-}|))^2/Mt}
938: e^{-\eta x^-}, \\
939: \endaligned
940: \eqnlbl{Rxbounds}
941: $$
942: and
943: $$
944: \aligned
945: R_{xy}(x,t;y)&=
946: (\partial/\partial y)\big(\sum_{j=1}^J
947: \Cal{O}(e^{-\eta t})\delta_{x-\bar a_j^* t}(-y) \big)\\
948: & +
949: \sum_{j=1}^J \Cal{O}(e^{-\eta t})\delta_{x-\bar a_j^* t}(-y)
950: +
951: \Cal{O}(e^{-\eta(|x-y|+t)})\\
952: &+\sum_{k=1}^n
953: \Cal {O} \left( (t+1)^{-3/2} e^{-\eta x^+}
954: +e^{-\eta|x|} \right)
955: %TIMES?
956: t^{-3/2} (t+1)
957: %(t+1)^{-1/2}
958: e^{-(x-y-a_k^{-} t)^2/Mt} \\
959: &+
960: \sum_{a_k^{-} > 0, \, a_j^{-} < 0}
961: \chi_{\{ |a_k^{-} t|\ge |y| \}}
962: \Cal{O}(t+1)^{-1/2} t^{-3/2})
963: e^{-(x-a_j^{-}(t-|y/a_k^-|))^2/Mt}
964: e^{-\eta x^+} \\
965: %SYMMETRIC
966: &+
967: \sum_{a_k^{-} > 0, \, a_j^{+} > 0}
968: \chi_{\{ |a_k^{-} t|\ge |y| \}}
969: \Cal{O}(t+1)^{-1/2} t^{-3/2})
970: e^{-(x-a_j^{+}(t-|y/a_k^{-}|))^2/Mt}
971: e^{-\eta x^-}, \\
972: \endaligned
973: \eqnlbl{Rxybounds}
974: $$
975: for some $\eta$, $M>0$, where $x^\pm$ denotes the positive/negative
976: part of $x$, and indicator function $\chi_{\{ |a_k^{-}t|\ge |y| \}}$ is
977: one for $|a_k^{-}t|\ge |y|$ and zero otherwise.
978: %Deleted, irrelevant:
979: %and indicator function $\chi_{\{ t\ge 1 \}}$ is
980: %one for $t\ge 1$ and zero otherwise.
981: Symmetric bounds hold for $y\ge0$.
982: %
983:
984: Here, the averaged convection rates
985: $ \bar a_j^*(y,t)$ in \eqnref{multH} and \eqnref{Rxbounds}--\eqnref{Rxybounds}
986: denote the time-averages over $[0,t]$
987: of $a_j^*(x)$ along characteristic paths $z_j^*=z_j^*(y,t)$ defined by
988: $$
989: dz_j^*/dt= a_j^*(z_j^*), \quad z_j^*(0)=y,
990: \eqnlbl{char}
991: $$
992: and the dissipation matrix
993: $\zeta_j^*=\zeta_j^*(y,t)\in \BbbR^{m_j\times m_j}$
994: is defined by the {\it dissipative flow}
995: $$
996: d\zeta_j^*/dt= -\eta_j^*(z_j^*)\zeta_j^*, \quad \zeta_j^*(y)=I_{m_j}.
997: \eqnlbl{diss}
998: $$
999: %
1000: Similarly, in \eqnref{S},
1001: $$
1002: z_{jk}^{\pm(y,t)}:=a_j^{\pm}\left(t-\frac{|y|}{|a_k^{-}|}\right)
1003: \eqnlbl{zjk}
1004: $$
1005: and
1006: %
1007: $$
1008: \bar \beta^{\pm}_{jk}(x,t;y):=
1009: \frac{|x^\pm|}{|a_j^{\pm} t|} \beta_j^{\pm}
1010: +
1011: \frac{|y|}{|a_k^{-} t|}
1012: \left( \frac{a_j^\pm}{a_k^{-}}\right)^2 \beta_k^{-},
1013: \eqnlbl{barbeta}
1014: $$
1015: represent, respectively, approximate scattered characteristic
1016: paths and the time-averaged diffusion rates along those paths.
1017: In all equations, $a_j$, $a_j^{*\pm}$, $l_j$, $\CalL_j^{*\pm }$,
1018: $r_j$, $\CalR_j^{*\pm}$,
1019: $\beta_j^{\pm}$ and $\eta_j^*$ are as defined just above,
1020: and scattering coefficients $[c_{k,-}^{j,i}]$, $i=-,0,+$, are constants,
1021: uniquely determined by
1022: $$
1023: \sum_{a_j^{-} < 0} [c_{k, \, -}^{j, \, -}]r_j^{-} +
1024: \sum_{a_j^{+} > 0} [c_{k, \, -}^{j, \, +}]r_j^{+} +
1025: [c_{k,-}^{0}] \big (G(+\infty)-G(-\infty) \big)
1026: = r_k^{-}
1027: \eqnlbl{scattering}
1028: $$
1029: for each $k=1,\dots n$, and satisfying
1030: $$
1031: \sum_{a_k^->0} [c_{k,-}^{0}] l_k^{-}
1032: %(y)
1033: =
1034: \sum_{a_k^+<0} [c_{k,+}^{0}] l_k^{+}
1035: =\pi,
1036: %(y)
1037: \eqnlbl{pi}
1038: $$
1039: where the constant vector $\pi$
1040: is the left zero effective eigenfunction of $L_G$ associated
1041: with the right eigenfunction $\bar G'$.
1042:
1043: \endproclaim
1044:
1045: Proposition \thmref{greenbounds}, the variable-coefficient
1046: generalization of the constant-coefficient results of
1047: [Ze.1,LZe.1], was established in [MZ.2] by Laplace transform
1048: (i.e., semigroup) techniques generalizing the Fourier transform
1049: approach of [Ze.1--2,LZe];
1050: for discussion/geometric interpretation, see [Z.2,MZ.1--2].
1051: In our stability analysis, we will use only a small part of
1052: the detailed information given in the proposition, namely
1053: $L^p\to L^q$ estimates on the time-decaying portion
1054: $H+S+R$ of the Green's function $\CalG$ (see Lemma 4.1, below).
1055: However, the stationary portion $E$ of the Green's function
1056: must be estimated accurately for an efficient stability analysis.
1057:
1058: \proclaim{Lemma \thmlbl{goodproj}}
1059: Under the assumptions of Proposition \thmref{greenbounds}, there
1060: holds, additionally,
1061: $$
1062: \Pi_2 (A^0(x))^{-1}H(x,t;y)\equiv 0,
1063: \eqnlbl{good1}
1064: $$
1065: where $\Pi_2:=\pmatrix 0&0\\
1066: 0&I_r\endpmatrix$
1067: denotes projection onto the final $r$ coordinates of $G$.
1068: \endproclaim
1069:
1070: {\bf Proof.}
1071: Equivalently, we must show that $\Pi_2 (A^0)^{-1}\CalR_j^*\equiv 0$
1072: for all $1\le j\le J$.
1073: This is most easily verified by the intrinsic property
1074: of $\CalL_j^*$ and $\CalR_j^*$ (readily seen from our formulae)
1075: that they lie,
1076: respectively, in the left and right kernel of $B_G$.
1077: For, this gives
1078: $$
1079: \aligned
1080: 0\equiv B_G \CalR_j^*
1081: &:= \pmatrix 0 & 0\\0& b\endpmatrix
1082: (A^0)^{-1}\CalR_j\\
1083: &=
1084: b \Pi_2 (A^0)^{-1}\CalR_j^*,
1085: \endaligned
1086: \eqnlbl{calc}
1087: $$
1088: yielding the result by invertibility of $b$, (see condition (A3)).
1089: \myqed
1090:
1091: Lemma \thmref{goodproj} quantifies the observation that, in $U=(u,v)^t$
1092: coordinates, the ``parabolic'' variable $v$ experiences smoothing
1093: under the evolution of \eqnref{linearized}, whereas the
1094: ``hyperbolic'' coordinate $v$ does not.
1095: (Recall that $U=(A^0)^{-1}G$).
1096:
1097: \bigskip
1098:
1099: \newsection {\bf Section 4. Linearized stability.}
1100: \sectionnumber=4
1101: \theoremnumber=0
1102: \equationnumber=0
1103: \smallskip
1104: \TagsOnLeft
1105: \bigskip
1106:
1107: We now show, for the Lax case under consideration,
1108: that linearized orbital stability
1109: follows immediately from the pointwise bounds
1110: of Proposition \thmref{greenbounds}, thus partially recovering
1111: the result of [MZ.2] that was stated in Proposition \thmref{D}.
1112: This analysis motivates the nonlinear argument to follow
1113: in Section 5.
1114:
1115: We carry out our analysis with respect to the
1116: conservative variable $G=A^0(x)U$, i.e. with respect to
1117: linearized equations \eqnref{linsymmetrizable}.
1118: Similarly as in [Z.2,MZ.1--2],
1119: define the {\it linear instantaneous projection}:
1120: $$
1121: \aligned
1122: \varphi(x,t)&:=
1123: \int_{-\infty}^{+\infty} E(x,t;y)G_0(y)\, dy
1124: \\
1125: &=:-\delta(t) \bar G'(x),\\
1126: \endaligned
1127: \eqnlbl{linproj}
1128: $$
1129: where $G_0$ denotes the initial data for \eqnref{linearized},
1130: and $\bar G=G(\bar U(x))$ as usual.
1131: The amplitude $\delta$ may be expressed, alternatively, as
1132: $$
1133: \delta(t)=
1134: -\int_{-\infty}^{+\infty} e(y,t)G_0(y)\, dy,
1135: $$
1136: where
1137: $$
1138: E(x,t;y)=:\bar G'(x)
1139: e(y,t),
1140: \eqnlbl{eE}
1141: $$
1142: i.e.,
1143: $$
1144: e(y,t):=
1145: \sum_{a_k^{-}}
1146: \left(\errfn\left(\frac{y+a_k^{-}t}{\sqrt{4\beta_k^{-}t}}\right)
1147: -\errfn \left(\frac{y-a_k^{-}t}{\sqrt{4\beta_k^{-}t}}\right)\right)
1148: l_k^{-}
1149: \eqnlbl{e}
1150: $$
1151: for $y\le 0$, and symmetrically for $y\ge 0$.
1152:
1153: Then, the solution $G$ of \eqnref{linearized} satisfies
1154: $$
1155: G(x,t)-\varphi(x,t)=
1156: \int_{-\infty}^{+\infty}(H + \tG)(x,t;0)G_0(y)\, dy,
1157: \eqnlbl{convolution}
1158: $$
1159: where
1160: $$
1161: \tG:=S+R
1162: \eqnlbl{tG}
1163: $$
1164: is the regular part and $H$ the singular part of
1165: the time-decaying portion of the Green's function $\CalG$.
1166:
1167: \proclaim{Lemma \thmlbl{2.05}}
1168: For weak shock profiles of \eqnref{general},
1169: under assumptions (A1)--(A3), (H0)--(H3), there hold:
1170: $$
1171: |\int_{-\infty}^{+\infty} \tG(\cdot,t;y)f(y)dy|_{L^p}
1172: %\le C \min \{|f|_{L^p}, (1+t)^{-\frac{1}{2}(1-1/p)} |f|_{L^1} \},
1173: \le C (1+t)^{-\frac{1}{2}(1-1/r)} |f|_{L^q},
1174: \eqnlbl{tGbounds}
1175: $$
1176: $$
1177: |\int_{-\infty}^{+\infty} \tG_y(\cdot,t;y)f(y)dy|_{L^p}
1178: %\le C \min \{(1+t)^{-1/2}|f|_{L^p}, (1+t)^{-\frac{1}{2}(1-1/p)-1/2} |f|_{L^1} \},
1179: \le C (1+t)^{-\frac{1}{2}(1-1/r)-1/2} |f|_{L^q}
1180: + Ce^{-\eta t} |f|_{L^p},
1181: \eqnlbl{tGybounds}
1182: $$
1183: $$
1184: |\int_{-\infty}^{+\infty}
1185: \tG_x(\cdot,t;y)
1186: f(y)dy|_{L^p}
1187: %\le C \min \{|f|_{L^p}, (1+t)^{-\frac{1}{2}(1-1/p)} |f|_{L^1} \},
1188: \le C (1+t)^{-\frac{1}{2}(1-1/r)} |f|_{L^q}
1189: + Ce^{-\eta t} |f|_{L^p},
1190: \eqnlbl{tGxbounds}
1191: $$
1192: $$
1193: |\int_{-\infty}^{+\infty}
1194: \tG_{xy}(\cdot,t;y)
1195: f(y)dy|_{L^p}
1196: %\le C \min \{(1+t)^{-1/2}|f|_{L^p}, (1+t)^{-\frac{1}{2}(1-1/p)-1/2} |f|_{L^1} \},
1197: \le C (1+t)^{-\frac{1}{2}(1-1/r)-1/2} |f|_{L^q}
1198: + Ce^{-\eta t} |f|_{W^{1,p}},
1199: \eqnlbl{tGxybounds}
1200: $$
1201: and
1202: $$
1203: |\int_{-\infty}^{+\infty} H(\cdot,t;y)f(y)dy|_{L^p}
1204: \le Ce^{-\eta t} |f|_{L^p},
1205: \eqnlbl{Hbounds}
1206: $$
1207: for all $t\ge 0$, some $C$, $\eta>0$, for any
1208: $1\le q\le p$
1209: (equivalently, $1\le r\le p$)
1210: and $f\in L^q \cap W^{1,p}$, where $1/r+1/q=1+1/p$.
1211: \endproclaim
1212:
1213: {\bf Proof.}
1214: Bounds \eqnref{tGbounds}--\eqnref{tGxybounds} follow by
1215: the Hausdorff-Young inequality together
1216: %TODO: check spelling of Hausdorff...
1217: %is it just Hausdorf?
1218: with bounds \eqnref{S} and \eqnref{Rbounds}--\eqnref{Rxybounds},
1219: precisely as in [Z.2,MZ.1--2].
1220: Bound \eqnref{Hbounds} follows by direct computation and
1221: the fact that particle paths $ z_j(y,t)$
1222: satisfy uniform bounds
1223: $$
1224: 1/C \le |(\partial/\partial y)z_j| < C,
1225: $$
1226: for all $y$, $t$, by the fact that characteristic speeds $a_j(x)$ converge
1227: exponentially as $x\to \pm \infty$ to constant states.
1228: \myqed
1229:
1230: \proclaim{Corollary \thmlbl{suff}}
1231: Let $\bar U$ be a weak shock profile of \eqnref{general},
1232: under assumptions (A1)--(A3), (H0)--(H3).
1233: Then, strong spectral stability implies
1234: $L^1\cap L^p\to L^p$ linearized orbital stability, for any $p>1$.
1235: More precisely,
1236: for initial data $U_0\in L^1\cap L^p$, the solution $U=(u,v)^t(x,t)$
1237: of \eqnref{linearized} satisfies the linear decay bounds
1238: $$
1239: |U(\cdot, t)+\delta(t)\bar U'(\cdot)|_{L^p}\le
1240: C(1+t)^{-\frac{1}{2}(1-1/p)}(|U_0|_{L^1}+ |U_0|_{L^p}).
1241: \eqnlbl{lindecaybound}
1242: $$
1243: Moreover, provided $U_0\in W^{1,p}$, there hold also the derivative bounds
1244: $$
1245: |\big(v(\cdot, t)+\delta(t)\bar v'(\cdot)\big)_x|_{L^p}\le
1246: C(1+t)^{-\frac{1}{2}(1-1/p)}(|U_0|_{L^1}+ |U_0|_{W^{1,p}}).
1247: \eqnlbl{dlindecaybound}
1248: $$
1249: \endproclaim
1250:
1251: {\bf Proof.}
1252: For \eqnref{lindecaybound},
1253: it is equivalent to show that,
1254: for initial data $G_0\in L^1\cap L^p$, the solution $G(x,t)$
1255: of \eqnref{linsymmetrizable} satisfies
1256: $$
1257: |G(\cdot, t)-\varphi(\cdot, t)|_{L^p}\le
1258: C(1+t)^{-\frac{1}{2}(1-1/p)}(|G_0|_{L^1}+ |G_0|_{W^{1,p}}).
1259: \eqnlbl{Gdecaybound}
1260: $$
1261: But, this follows immediately from \eqnref{convolution}
1262: and bounds \eqnref{tGbounds} and \eqnref{Hbounds}, with $q=p$.
1263: Likewise, for \eqnref{dlindecaybound},
1264: it is equivalent to show that,
1265: for initial data $G_0\in L^1\cap W^{1,p}$, the solution $G(x,t)$
1266: of \eqnref{linsymmetrizable} satisfies
1267: $$
1268: |\left(\Pi_2(A^0)^{-1}\big(G(\cdot, t)-\varphi(\cdot, t)\big)\right)_x|_{L^p}\le
1269: C(1+t)^{-\frac{1}{2}(1-1/p)}(|G_0|_{L^1}+ |G_0|_{W^{1,p}}).
1270: \eqnlbl{Gdecaybound}
1271: $$
1272: This follows, similarly, from the derivative analog
1273: $$
1274: \left(\Pi_2 (A^0)^{-1}
1275: \big(G(x,t)-\varphi(x,t) \big) \right)_x=
1276: \int_{-\infty}^{+\infty}
1277: \left(\Pi_2 (A^0)^{-1}(H + \tG)\right)_x(x,t;0)G_0(y)\, dy
1278: \eqnlbl{dconvolution}
1279: $$
1280: of \eqnref{convolution},
1281: together with \eqnref{good1} and bounds \eqnref{tGbounds}, \eqnref{tGxbounds},
1282: with $q=p$.
1283: \myqed
1284: \bigskip
1285:
1286: \newsection {\bf Section 5. Nonlinear stability.}
1287: \sectionnumber=5
1288: \theoremnumber=0
1289: \equationnumber=0
1290: \smallskip
1291: \TagsOnLeft
1292: \bigskip
1293:
1294: Finally, we establish our main result
1295: of nonlinear orbital stability with respect to perturbations
1296: $ U_0\in L^1\cap H^3 $
1297: of weak viscous profiles (necessarily Lax type, by Proposition \thmref{freist1})
1298: of dissipative, symmetric hyperbolic--parabolic systems
1299: of type \eqnref{general},
1300: of strength $\varepsilon>0$ sufficiently small with respect to the parameters of
1301: the system in question.
1302: We follow the basic iteration scheme of [MZ.1,Z.2];
1303: for precursors of this scheme, see [Go.2,K.1--2,LZ.1--2,ZH,HZ.1--2].
1304:
1305: Define the nonlinear perturbation
1306: $$
1307: U(x,t):= \tilde U (x+\delta(t),t)-
1308: \bar U(x),
1309: \eqnlbl{pert}
1310: $$
1311: where $\delta(t)$ (estimating shock location) is to be determined
1312: later; for definiteness, fix $\delta(0)=0$.
1313: Substituting \eqnref{pert} into \eqnref{general}, we obtain
1314: $$
1315: G(\tilde U)_t + F(\tilde U)_x - (B(\tilde U)\tilde U_x)_x=
1316: \dot \delta G(\tilde U)_x,
1317: \eqnlbl{tildeU}
1318: $$
1319: and thereby the basic perturbation equation
1320: $$
1321: \left(G(\tilde U)-G(\bar U)\right)_t
1322: +\left(F(\tilde U)-F(\bar U)\right)_x
1323: -\left(B(\tilde U)\tilde U_x-B(\bar U)\bar U_x\right)_x
1324: =\dot \delta(t)G(\tilde U)_x,
1325: \eqnlbl{basicpert}
1326: $$
1327: where $\tilde U$ now denotes $\tilde U(x+\delta(t),t)$ and
1328: $\bar U$ denotes $\bar U(x)$.
1329:
1330: \medskip
1331: {\bf 5.1. Energy estimates.}
1332: We begin by establishing the following basic energy estimate, by which we
1333: will eventually close our nonlinear iteration argument:
1334:
1335: \proclaim{Proposition \thmlbl{energy}}
1336: Let $U_0\in H^3$, and suppose that, for $0\le t\le T$,
1337: both the supremum of $ |\dot\delta|$ and the $H^2\cap W^{2,\infty}$
1338: norm of the solution
1339: $U=(u,v)^t$ of \eqnref{pert}--\eqnref{basicpert} remain
1340: bounded by a sufficiently
1341: small constant $\zeta>0$,
1342: for a small-amplitude shock profile as described in Proposition
1343: \thmref{freist1} of a system \eqnref{general} satisfying
1344: (A1)--(A3), (H0)--(H3),
1345: with $\varepsilon :=|\tilde U_+ - \tilde U_-|$ sufficiently small.
1346: Then, there hold the bounds:
1347: $$
1348: |U|_{H^3}(t), \, \int_0^t \big(|U_x|_{H^2}+|v_x|_{H^3}^2\big)(s) ds \le
1349: C\int_0^t \left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1350: + \dot \delta^2 \right)(s)ds,
1351: \eqnlbl{ebounds}
1352: $$
1353: for all $0\le t\le T$.
1354: \endproclaim
1355:
1356: %TODO: some transition here?
1357: {\bf Remark \thmlbl{nearoptimal}.}
1358: Note, with the expected decay rates $|U|_{L^\infty}$,
1359: $|\dot \delta(t)|\sim C(1+t)^{-1/2}$,
1360: and $|U|_{L^2}\sim C(1+t^{-1/4})$,
1361: that the righthand side of \eqnref{ebounds}
1362: becomes order $\log (1+t)$, very nearly recovering
1363: the order one bound available in the constant-coefficient
1364: case [Kaw].
1365: \medskip
1366:
1367: \proclaim{Lemma \thmlbl{skew} ([SK])}
1368: Assuming (A1), condition (A2)
1369: is equivalent to either of:
1370: \medskip
1371: (K1) \quad
1372: There exists a smooth skew-symmetric matrix $K(u)$ such that
1373: $$
1374: \text{\rm Re }\left( K(A^0)^{-1}A + B \right)(u) \ge \theta>0.
1375: \eqnlbl{skew}
1376: $$
1377: $A^0$, $A$, $B$ as in \eqnref{AB}.
1378:
1379: \medskip
1380: (K2) \quad
1381: For some $\theta>0$, there holds
1382: $$
1383: \text{\rm Re }\sigma(-i\xi (A^0)^{-1}A(u)-|\xi|^2 (A^0)^{-1}B(u))
1384: \le -\theta |\xi|^2/(1+|\xi|^2),
1385: \eqnlbl{symbol}
1386: $$
1387: for all $\xi\in \BbbR$.
1388: \endproclaim
1389:
1390: {\bf Proof.}
1391: These and other useful equivalent formulations are established in [SK]. \myqed
1392:
1393: \proclaim{Lemma \thmlbl{pbounds}}
1394: Let (A1)--(A3) and (H1)--(H3) hold,
1395: and let $\bar U(x-st)$ be a viscous shock solution such that
1396: the profile $\{\bar U(z)\}$ lies entirely within a sufficently
1397: small neighborhood $\CalV\subset \CalU$ of $\tilde U_*$, and the speed
1398: $s$ lies within a sufficiently small neighborhood of $a_p(\tilde U_*)$:
1399: i.e., a profile as described in Proposition \thmref{freist1}.
1400: Then, letting $\varepsilon :=|\tilde U_+ - \tilde U_-|$ denote shock strength,
1401: we have
1402: for $q=0,\dots 4$ the uniform bounds:
1403: $$
1404: \aligned
1405: |\partial_x^q \bar U(x)| &\le C\varepsilon^{q+1}e^{-\theta\varepsilon|x|},\\
1406: |\partial_x^q \bar U|_{L^p} &\le C\varepsilon^{q+1-1/p},\\
1407: \endaligned
1408: \eqnlbl{weakbounds}
1409: $$
1410: for some $C$, $\theta>0$.
1411: %
1412: \endproclaim
1413:
1414: {\bf Proof.}
1415: Though the bounds \eqnref{weakbounds} are not explicitly stated
1416: in [Fre.2], they follow immediately from the detailed description
1417: of center manifold dynamics obtained in the proof, exactly as
1418: in the strictly parabolic case [MP].
1419: \myqed
1420:
1421:
1422: {\bf Proof of Proposition \thmref{energy}}.
1423: Writing \eqnref{pert} in quasilinear form
1424: $$
1425: \left(\tilde A^0\tilde U_t-\bar A^0 \bar U_t \right)
1426: +\left(\tilde A\tilde U_x-\bar A\bar U_x\right)
1427: -\left(\tilde B\tilde U_x-\bar B\bar U_x\right)_x
1428: =\dot \delta(t)\tilde A^0\tilde U_x,
1429: \eqnlbl{basicpert}
1430: $$
1431: where
1432: $$
1433: \tilde A^0:=A^0(\tilde U), \bar A^0:=A^0(\bar U);
1434: \quad \tilde A:=A(\tilde U), \bar A:=A(\bar U);
1435: \quad \tilde B:=B(\tilde U), \bar B:=B(\bar U),
1436: \eqnlbl{tildebar}
1437: $$
1438: using the quadratic Leibnitz relation
1439: $$
1440: A_2 U_2- A_1 U_1= A_2(U_2-U_1) + (A_2-A_1)U_1,
1441: \eqnlbl{nLeib}
1442: $$
1443: and recalling the block structure assumption (A3),
1444: we obtain the alternative perturbation equation:
1445: $$
1446: \tilde A^0 U_t +\tilde A U_x -(\tilde B U_x )_x =
1447: M_1(U)\bar U_x
1448: + (M_2(U) \bar U_x)_x
1449: +\dot \delta(t)\tilde A^0 U_x
1450: +\dot \delta(t)\tilde A^0\bar U_x,
1451: \eqnlbl{Leibnitz}
1452: $$
1453: where
1454: $$
1455: M_1(U):=\tilde A-\bar A
1456: =
1457: \left( \int_0^1 dA(\bar U + \theta U )\, d\theta \right) U,
1458: \eqnlbl{M1}
1459: $$
1460: and
1461: $$
1462: \aligned
1463: M_2(U)&:= \tilde B-\bar B\\
1464: &=
1465: \pmatrix
1466: 0&0\\
1467: 0&
1468: ( \int_0^1 db(\bar U + \theta U )\, d\theta ) U
1469: \endpmatrix,
1470: \endaligned
1471: \eqnlbl{M2}
1472: $$
1473: with
1474: $$
1475: \tilde b:=b(\tilde U), \, \bar b:=b(\bar U).
1476: \eqnlbl{btilde}
1477: $$
1478:
1479: We now carry out a series of successively higher order energy estimates of the
1480: the type formalized by Kawashima [Ka.1].
1481: The origin of this approach goes back
1482: to [K,MN] in the context of gas dynamics; see, e.g., [HoZ] for further
1483: discussion/references.
1484:
1485: Let $\tilde K$ denote the skew-symmetric matrix
1486: described in Lemma \thmref{skew}
1487: associated with $\tilde A^0$, $\tilde A$, $\tilde B$.
1488: Then, regarding $\tilde A^0$, $\tilde K$, we have the bounds
1489: $$
1490: \aligned
1491: \tilde A^0_x&=
1492: dA^0(\tilde U) \tilde U_x, \quad
1493: \tilde K_x=
1494: dK(\tilde U) \tilde U_x, \quad
1495: \tilde A_x=
1496: dA(\tilde U) \tilde U_x, \quad
1497: \tilde B_x=
1498: dB(\tilde U) \tilde U_x,
1499: \\
1500: \tilde A^0_t&=
1501: dA^0(\tilde U) \tilde U_t , \quad
1502: \tilde K_t=
1503: dK(\tilde U) \tilde U_t, \quad
1504: \tilde A_t=
1505: dA(\tilde U) \tilde U_t, \quad
1506: \tilde B_x=
1507: dB(\tilde U) \tilde U_t, \\
1508: \endaligned
1509: \eqnlbl{AKbounds}
1510: $$
1511: and (from defining equations \eqnref{pert}--\eqnref{tildeU}):
1512: $$
1513: |\tilde U_x|=|U_x + \bar U_x|\le |U_x|+ |\bar U_x|
1514: \eqnlbl{xbound}
1515: $$
1516: and
1517: $$
1518: \aligned
1519: |\tilde U_t|&
1520: \le C( |\tilde U_x| + |\tilde v_{xx}| + |\dot \delta||\tilde U_x|)\\
1521: &
1522: \le C( |U_x| + |\bar U_x| + |v_{xx}|+ |\bar v_{xx}| + |\dot \delta||U_x|
1523: +|\dot\delta| |\bar U_x|)\\
1524: &
1525: \le C( |U_x| + |\bar U_x| + |v_{xx}|+ |\bar v_{xx}| ).\\
1526: \endaligned
1527: \eqnlbl{tbound}
1528: $$
1529: Thus, in particular
1530: $$
1531: |\dot\delta|, \,
1532: |\tilde A^0_x|,\,
1533: |\tilde K_x|,\,
1534: |\tilde A_x|,\,
1535: |\tilde B_x| ,\,
1536: |\tilde A^0_t|,\,
1537: |\tilde K_t|,\,
1538: |\tilde A_t|,\,
1539: |\tilde B_t|
1540: \le C(\zeta+\varepsilon).
1541: \eqnlbl{smallness}
1542: $$
1543:
1544: \medskip
1545:
1546: {\it $H^1$ estimate.}
1547: We first perform a standard, ``Friedrichs-type''
1548: estimate for symmetrizable hyperbolic systems.
1549: Taking the $L^2$ inner product of $U$ against \eqnref{Leibnitz},
1550: we obtain after rearrangement/integration by parts, and several
1551: applications of Young's inequality, the energy estimate
1552: $$
1553: \aligned
1554: \frac{1}{2}\langle U_{},\tilde A^0 U_{}\rangle_t
1555: &=
1556: \langle U_{},\tilde A^0 U_{t}\rangle
1557: +\frac{1}{2}\langle U_{},\tilde A^0_t U_{}\rangle
1558: \\
1559: &=
1560: -\langle U,\tilde A U_x\rangle
1561: +
1562: \langle U,(\tilde B U_x )_x \rangle
1563: +\langle U, M_1(U)\bar U_x \rangle
1564: + \langle U, (M_2(U) \bar U_x)_x\rangle\\
1565: &\qquad+\dot \delta(t)\langle U,\tilde A^0 U_x\rangle
1566: +\dot \delta(t)\langle U,\tilde A^0\bar U_x\rangle
1567: +\frac{1}{2}\langle U_{},\tilde A^0_t U_{}\rangle\\
1568: &=
1569: \frac{1}{2}\langle U,\tilde A_x U\rangle
1570: -
1571: \langle U_x,\tilde B U_x \rangle
1572: +\langle U, M_1(U)\bar U_x \rangle
1573: - \langle U_x, M_2(U) \bar U_x\rangle\\
1574: &\qquad-\frac{1}{2}\dot \delta(t)\langle U,\tilde A^0_x U\rangle
1575: +\dot \delta(t)\langle U,\tilde A^0\bar U_x\rangle
1576: +\frac{1}{2}\langle U_{},\tilde A^0_t U_{}\rangle\\
1577: &\le -
1578: %\frac{1}{2}
1579: \langle U_{x},\tilde BU_{x}\rangle \\
1580: &\qquad
1581: + C\int \left( (|U_x|+|\bar U_x|+ |v_{xx}|)|U|^2
1582: + |v_x||U||U_x| + |\dot \delta||U||\bar U_x| \right) \\
1583: &\le -
1584: %\frac{1}{2}
1585: \langle U_{x},\tilde BU_{x}\rangle \\
1586: &\qquad
1587: + C\int \left( (|U_x|^2+|U|^2+ |v_x|^2+ |v_{xx}|^2)(|U| + |\bar U_x|)
1588: + |\dot \delta|^2|\bar U_x| \right) \\
1589: &\le -
1590: \langle U_{x},\tilde BU_{x}\rangle \\
1591: &\qquad
1592: + C\left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1593: + |\dot \delta|^2 \right)
1594: + C(\varepsilon +\zeta )\left( |U_x|_{L^2}^2 + |v_{xx}|_{L^2}^2\right),
1595: \endaligned
1596: \eqnlbl{sym}
1597: $$
1598: where $\varepsilon$, $\zeta>0$ is as in the statement of the Proposition.
1599: Here, we have freely used the weak shock assumption and
1600: consequent bounds \eqnref{weakbounds}, as well as \eqnref{AKbounds}.
1601: %
1602: (Note: we have also used in a crucial way the block-diagonal
1603: form of $M_2$ in estimating
1604: $|\langle U_x, M_2(U) \bar U_x\rangle|\le
1605: C\int |v_x||U||\bar U_x|$ in the first inequality).
1606: %
1607:
1608: Likewise, differentiating \eqnref{Leibnitz}, taking the $L^2$
1609: inner product of $U_x$ against the resulting equation,
1610: and substituting the result into
1611: $$
1612: \frac{1}{2}\langle U_{x},\tilde A^0 U_{x}\rangle_t
1613: =
1614: \langle U_{x},(\tilde A^0 U_{t})_x\rangle
1615: -\langle U_{x},\tilde A^0_x U_{t}\rangle
1616: +\frac{1}{2}\langle U_{x},\tilde A^0_t U_{x}\rangle,
1617: \eqnlbl{one}
1618: $$
1619: we obtain after an integration by parts:
1620: $$
1621: \aligned
1622: \frac{1}{2}\langle U_{x},\tilde A^0 U_{x}\rangle_t
1623: &=
1624: -\langle U_x,(\tilde A U_x)\rangle
1625: +
1626: \langle U,(\tilde B U_x )_x \rangle
1627: +\langle U, M_1(U)\bar U_x \rangle
1628: + \langle U, (M_2(U) \bar U_x)_x\rangle\\
1629: &\qquad+\dot \delta(t)\langle U,\tilde A^0 U_x\rangle
1630: +\dot \delta(t)\langle U,\tilde A^0\bar U_x\rangle
1631: +\frac{1}{2}\langle U_{},\tilde A^0_t U_{}\rangle\\
1632: &=
1633: -\frac{1}{2}\langle U_x,\tilde A_x U_x\rangle
1634: - \langle U_{xx},\tilde B U_{xx} \rangle
1635: - \langle U_{xx},\tilde B_x U_{x} \rangle
1636: +\langle U_x, (M_1(U)\bar U_x)_x \rangle \\
1637: &\qquad - \langle U_{xx}, (M_2(U) \bar U_x)_x\rangle
1638: +\frac{1}{2}\dot \delta(t)\langle U_x,\tilde A^0_x U_x\rangle
1639: +\dot \delta(t)\langle U_x,\tilde A^0\bar U_{xx}\rangle\\
1640: &\qquad +\dot \delta(t)\langle U_x,\tilde A^0_x\bar U_x\rangle
1641: -\langle U_{x},\tilde A^0_x U_{t}\rangle
1642: +\frac{1}{2}\langle U_{x},\tilde A^0_t U_{x}\rangle,\\
1643: \endaligned
1644: \eqnlbl{nextsym0}
1645: $$
1646: which by \eqnref{smallness}, plus various applications of Young's
1647: inequality yields the next-order energy estimate:
1648: $$
1649: \aligned
1650: \frac{1}{2}\langle U_{x},\tilde A^0 U_{x}\rangle_t
1651: &\le -
1652: \langle U_{xx},\tilde BU_{xx}\rangle \\
1653: &\qquad
1654: + C\int\left(
1655: (|U|^2+|\dot\delta|^2)(|\bar U_{xx}|+ |\bar U_x|)
1656: + (\varepsilon +\zeta )( |U_x|^2 + |U_x||v_{xx}| ) \right),\\
1657: &\le -
1658: \langle U_{xx},\tilde BU_{xx}\rangle \\
1659: &\qquad
1660: + C\left((|U|_{L^\infty}^2+
1661: |\dot \delta|^2 \right)
1662: + C(\varepsilon +\zeta )\left( |U_x|_{L^2}^2 + |v_{xx}|_{L^2}^2\right)\\
1663: &\le - \frac{1}{2}
1664: \langle U_{xx},\tilde BU_{xx}\rangle \\
1665: &\qquad
1666: + C\left((|U|_{L^\infty}^2+
1667: |\dot \delta|^2 \right)
1668: + C(\varepsilon +\zeta )|U_x|_{L^2}^2.
1669: \endaligned
1670: \eqnlbl{nextsym}
1671: $$
1672:
1673: Next, we perform a nonstandard, ``Kawashima-type'' derivative estimate.
1674: Taking the $L^2$ inner product of $U_{x}$
1675: against $\tilde K(\tilde A^0)^{-1} $ times \eqnref{Leibnitz},
1676: and noting that (integrating by parts, and using skew-symmetry of $\tilde K$)
1677: $$
1678: \aligned
1679: \frac{1}{2}\langle U_{x},\tilde KU_{} \rangle_t
1680: &=
1681: \frac{1}{2}\langle U_{x},\tilde KU_{t} \rangle+
1682: \frac{1}{2}\langle U_{xt},\tilde KU_{} \rangle
1683: +\frac{1}{2}\langle U_{x},\tilde K_tU_{} \rangle\\
1684: &=
1685: \frac{1}{2}\langle U_{x},\tilde KU_{t} \rangle
1686: -\frac{1}{2}\langle U_{t},\tilde KU_{x} \rangle\\
1687: &\quad
1688: -\frac{1}{2}\langle U_{t},\tilde K_x U_{x} \rangle
1689: +\frac{1}{2}\langle U_{x},\tilde K_tU_{} \rangle\\
1690: &=
1691: \langle U_{x},\tilde KU_{t} \rangle
1692: +\frac{1}{2}\langle U_{},\tilde K_x U_{t} \rangle
1693: +\frac{1}{2}\langle U_{x},\tilde K_tU_{} \rangle ,\\
1694: \endaligned
1695: \eqnlbl{parts}
1696: $$
1697: we obtain by calculations similar to the above
1698: the auxiliary energy estimate:
1699: $$
1700: \aligned
1701: \frac{1}{2}\langle U_{x},\tilde KU_{} \rangle_t &\le
1702: -\langle U_{x}, \tilde K(\tilde A^0)^{-1}\tilde A U_{x}\rangle\\
1703: &\qquad
1704: + C(\zeta+\varepsilon) |U_x|_{L^2}^2 +
1705: C\zeta^{-1}|v_{xx}|_{L^2}^2 + C(|U|_{L^\infty}^2+ |\dot\delta(t)|^2).
1706: \endaligned
1707: \eqnlbl{est2}
1708: $$
1709: Adding \eqnref{sym}, \eqnref{est2}, and
1710: \eqnref{nextsym} times a suitably large constant $M>0$,
1711: and recalling \eqnref{skew},
1712: we obtain, finally:
1713: $$
1714: \aligned
1715: \frac{1}{2}
1716: \Big(
1717: \langle U, \tilde A^0 U_{}\rangle
1718: + &\langle U_{x},\tilde KU_{} \rangle
1719: +M\langle U_x, \tilde A^0 U_{x}\rangle \Big)_t \\
1720: &\le
1721: -\theta (|U_{x}|_{L^2}^2
1722: +|v_{xx}|_{L^2}^2)
1723: +C \left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1724: + |\dot \delta|^2 \right),\\
1725: \endaligned
1726: \eqnlbl{estfinal}
1727: $$
1728: $\theta>0$,
1729: for any $\zeta$, $\varepsilon $ sufficiently small, and $M$, $C>0$ sufficiently large.
1730:
1731: \medskip
1732: {\it Higher order estimates.}
1733: Performing the same procedure on the once- and twice-differentiated
1734: versions of equation \eqnref{Leibnitz}, we obtain,
1735: likewise, the $H^q$ estimates,
1736: $q=2, \, 3$, of:
1737: $$
1738: \aligned
1739: \frac{1}{2}
1740: \Big(
1741: \langle \partial_x^{q-1}U_{},&\tilde A^0\partial_x^{q-1}U_{}\rangle
1742: + \langle \partial_x^{q-1}U_{x},\tilde K\partial_x^{q-1}U_{} \rangle
1743: +M\langle \partial_x^{q}U,\tilde A^0\partial_x^{q}U\rangle \Big)_t \\
1744: &\le
1745: -\theta (|\partial_x^q U|_{L^2}^2
1746: +|\partial_x^{q+1}v|_{L^2}^2)\\
1747: &\qquad +(\varepsilon + \zeta) |U_{x}|_{H^{q-2}}^2
1748: +C \left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1749: + |\dot \delta|^2 \right).\\
1750: \endaligned
1751: \eqnlbl{esthigher}
1752: $$
1753: We omit the calculations, which are entirely similar to those
1754: carried out already.
1755: \medskip
1756:
1757: {\it Final estimate.}
1758: Summing our $H^q$ estimates from $q=1$ to $3$,
1759: and telescoping the sum of the righthand sides,
1760: %with weights $C_q>0$ decreasing sufficiently rapidly with $q$,
1761: we thus obtain, for $\varepsilon$, $\zeta$ sufficiently small:
1762: $$
1763: \aligned
1764: \sum_{q=1}^3
1765: %C_q
1766: \big( \langle &\tilde A^0 \partial_x^{q-1}U,\partial_x^{q-1}U\rangle
1767: + \frac{1}{2}\langle \partial_x^{q}U,\tilde K\partial_x^{q-1}U
1768: \rangle
1769: +M\langle \tilde A^0 \partial_x^{q}U,\partial_x^{q}U\rangle \big)_t \\
1770: &\le
1771: -\theta (|U_x|_{H^2}+|v_x|_{H^3}^2)
1772: +C \left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1773: + |\dot \delta|^2 \right),\\
1774: \endaligned
1775: \eqnlbl{estsummed}
1776: $$
1777: or,
1778: integrating from $0$ to $t$:
1779: $$
1780: \aligned
1781: \sum_{q=1}^3
1782: %C_q
1783: \big( \langle \tilde A^0 \partial_x^{q-1}U,\partial_x^{q-1}U\rangle
1784: + &\frac{1}{2}\langle \partial_x^{q}U,\tilde K\partial_x^{q-1}U
1785: \rangle
1786: +M\langle \tilde A^0 \partial_x^{q}U,\partial_x^{q}U\rangle \big)|^t_0 \\
1787: &\le
1788: -\int_0^t \theta (|U_x|_{H^2}+|v_x|_{H^3}^2)(s)ds\\
1789: &\quad+
1790: C\int_0^t \left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1791: + |\dot \delta|^2 \right)(s)ds.
1792: \endaligned
1793: \eqnlbl{intcalc0}
1794: $$
1795:
1796: Noting that, for $M$ sufficiently large, we have for each $q$,
1797: by Young's inequality, and positive definiteness of $\tilde A^0$:
1798: $$
1799: \big(
1800: \langle \partial_x^{q-1}U,\tilde A^0 \partial_x^{q-1}U\rangle
1801: + \frac{1}{2}\langle \partial_x^{q}U,\tilde K\partial_x^{q-1}U
1802: \rangle
1803: +M\langle \partial_x^{q}U,\tilde A^0 \partial_x^{q}U\rangle \big)(t)\ge
1804: \theta(|\partial_x^{q-1} U|_{L^2}^2
1805: +|\partial_x^{q} U|_{L^2}^2),
1806: \eqnlbl{absorb}
1807: $$
1808: for some $\theta>0$, we may rearrange \eqnref{intcalc0} to obtain
1809: our ultimate goal:
1810: $$
1811: \aligned
1812: |U|_{H^3}^2(t)
1813: +&\int_0^t \theta (|U_x|_{H^2}+|v_x|_{H^3}^2)(s)ds\\
1814: &\le C|U|_{H^3}^2(0)+
1815: C\int_0^t \left(|U|_{L^\infty}(|U|_{L^\infty}+ |U|_{L^2}^2)
1816: + |\dot \delta|^2 \right)(s)ds,
1817: \endaligned
1818: \eqnlbl{goal}
1819: $$
1820: from which the result immediately follows.
1821: \myqed
1822:
1823: \medskip
1824:
1825: {\bf 5.2. Nonlinear iteration.}
1826: We now carry out the nonlinear iteration, following [Z.2,MZ.1].
1827: For this stage of the argument, it will be convenient to work
1828: again with the conservative variable
1829: $$
1830: G:=G(\tilde U)-G(\bar U),
1831: \eqnlbl{Gpert}
1832: $$
1833: writing \eqnref{pert} in the more standard form:
1834: $$
1835: G_t-LG=N(G,G_x)_x
1836: +\dot \delta (t)(\BG_x + G_x),
1837: \eqnlbl{29}
1838: $$
1839: $\bar G:= G(\bar U)$, where
1840: $$
1841: \aligned
1842: N(G,G_x)&=\Cal{O}(|G|^2 + |G||v_x|),\\
1843: N(G,G_x)_x&=\Cal{O}(|G|^2 + |G||v_x|
1844: + |G_x||v_x| + |G||v_{xx}|),\\
1845: %NOTE: just as in [MZ.1], we have to worry about "coeff." deriv. ~ (|\bar G_x|
1846: %+ |G_x|), only bounded, no better...
1847: \endaligned
1848: \eqnlbl{29.1}
1849: $$
1850: so long as $|G|$, $|G_x|$ remain bounded.
1851: Here, $v$ denotes the second coordinate of the alternative
1852: perturbation variable $U=(u,v)^t$ defined in \eqnref{pert}.
1853: %
1854:
1855: By Duhamel's principle, and the fact that
1856: $$
1857: \int^\infty_{-\infty}\CalG(x,t;y)\BG_x(y)dy=e^{Lt}\BG_x(x)=\BG_x(x),
1858: \eqnlbl{stationary}
1859: $$
1860: we have, formally,
1861: $$
1862: \aligned
1863: & G(x,t)=\int^\infty_{-\infty}\CalG(x,t;y)G_0(y)dy\\
1864: %
1865: &+\int^t_0 \int^\infty_{-\infty} \CalG_y(x,t-s;y)
1866: (N(G,G_x)+
1867: \dot \delta G ) (y,s)dyds\\
1868: %
1869: &+ \delta (t)\BG_x.\\
1870: \endaligned
1871: \eqnlbl{30}
1872: $$
1873:
1874: Defining, by analogy with the linear case,
1875: the {\it nonlinear instantaneous projection}:
1876: $$
1877: \aligned
1878: \varphi(x,t)
1879: &:= -\delta(t)\BG_x\\
1880: %
1881: &:= \int^\infty_{-\infty}{E}(x,t;y)G_0(y) dy\\
1882: %
1883: &-\int^t_0 \int^\infty_{-\infty}E_y(x,t-s;y)(N(G,G_x)+
1884: \dot \delta G)(y,s)dy,\\
1885: \endaligned
1886: \eqnlbl{proj}
1887: $$
1888: or equivalently, the {\it instantaneous shock location}:
1889: $$
1890: \aligned
1891: \delta (t)
1892: &=-\int^\infty_{-\infty}e(y,t) G_0(y)dy\\
1893: %
1894: &+\int^t_0\int^{+\infty}_{-\infty} e_{y}(y,t-s)(N(G,G_x)+
1895: \dot \delta G)(y,s) dy ds,\\
1896: \endaligned
1897: \eqnlbl{31}
1898: $$
1899: where $E$, $e$ are defined as in \eqnref{E}, \eqnref{e},
1900: and recalling \eqnref{tG}, we thus
1901: obtain the {\it reduced equations}:
1902: $$
1903: \aligned
1904: G(x,t)
1905: &=\int^\infty_{-\infty} (H+\tG)(x,t;y)G_0(y)dy\\
1906: %
1907: &+\int^t_0
1908: \int^\infty_{-\infty}H(x,t-s;y)(N(G,G_x)_x
1909: +\dot \delta G_x)(y,s)dy \, ds\\
1910: &-\int^t_0
1911: \int^\infty_{-\infty}\tG_y(x,t-s;y)(N(G,G_x)+
1912: \dot \delta G)(y,s) dy \, ds,\\
1913: \endaligned
1914: \eqnlbl{32}
1915: $$
1916: and, differentiating \eqnref{31} with respect to $t$,
1917: $$
1918: \aligned
1919: \dot \delta (t)
1920: &=-\int^\infty_{-\infty}e_t(y,t) G_0(y)dy\\
1921: %
1922: &+\int^t_0\int^{+\infty}_{-\infty} e_{yt}(y,t-s)(N(G,G_x)+
1923: \dot \delta G)(y,s) dy ds.\\
1924: \endaligned
1925: \eqnlbl{33}
1926: $$
1927:
1928: {\it Note:}
1929: In deriving \eqnref{33}, we have used the fact
1930: that $e_y (y,s)\rightharpoondown 0$ as $s \to 0$, as the
1931: difference of approaching heat kernels, in evaluating the
1932: boundary term
1933: $$
1934: \int^{+\infty}_{-\infty} e_y (y,0)(N(G,G_x)+
1935: \dot \delta G)(y,t)dy=0.
1936: \eqnlbl{35}
1937: $$
1938: (Indeed, $|e_y(\cdot, s)|_{L^1}\to 0$, see Remark 2.6, below).
1939: \medskip
1940:
1941: The defining relation $ \delta (t)\bar u_x:= -\varphi$ in \eqnref{proj}
1942: can be motivated heuristically by
1943: $$
1944: \eqalign{
1945: \tilde G(x,t)-\varphi(x,t) \sim G&= \pmatrix u\\v\endpmatrix (x+\delta(t),t)
1946: - \pmatrix \bar u\\ \bar v\endpmatrix(x) \cr
1947: &\sim \tilde G(x,t) + \delta (t)\bar G_x(x),}
1948: $$
1949: where $\tilde G$ denotes the solution of the linearized
1950: perturbation equations, and $\bar G$ the background profile.
1951: Alternatively, it can be thought
1952: of as the requirement that the instantaneous projection
1953: of the shifted (nonlinear) perturbation variable $G$ be zero, [HZ.1--2].
1954:
1955: \proclaim{Lemma \thmlbl{3} [Z.2]} The kernel ${e}$ satisfies
1956: $$
1957: |{e}_y (\cdot, t)|_{L^p}, |{e}_t(\cdot, t)|_{L^p}
1958: \le C t^{-\frac{1}{2}(1-1/p)},
1959: \eqnlbl{36}
1960: $$
1961: $$
1962: |{e}_{ty}(\cdot, t)|_{L^p}
1963: \le C t^{-\frac{1}{2}(1-1/p)-1/2},
1964: \eqnlbl{37}
1965: $$
1966: for all $t>0$. Moreover, for $y\le 0$ we have the pointwise bounds
1967: $$
1968: |{e}_y (y,t)|, |{e}_t (y,t)| \le Ct^{-\frac{1}{2}} e^{-\frac{(y+a_-t)^2}{Mt}},
1969: \eqnlbl{38}
1970: $$
1971: $$
1972: |{e}_{ty} (y,t)| \le C t^{-1} e^{-\frac{(y+a_-t)^2}{Mt}},
1973: \eqnlbl{39}
1974: $$
1975: for $M>0$ sufficiently large (i.e. $>4b_\pm$),
1976: and symmetrically for $y\ge 0$.
1977: \endproclaim
1978:
1979: {\bf Proof.}
1980: For definiteness, take $y \le 0$.
1981: Then, \eqnref{e} gives
1982: $$
1983: {e}_y(y,t)=
1984: \left(\frac{1}{u_+-u_-}\right)
1985: \left(K(y+a_-t,t)-K(y-a_-t,t)\right),
1986: \eqnlbl{41}
1987: $$
1988: $$
1989: {e}_t(y,t)=
1990: \left(\frac{1}{u_+-u_-}\right)
1991: \left( (K+K_y)(y+a_-t,t)-(K+K_y)(y-a_-t,t)\right),
1992: \eqnlbl{41}
1993: $$
1994: $$
1995: {e}_{ty}(y,t)=
1996: \left(\frac{1}{u_+-u_-}\right)
1997: \left( (K_y+K_{yy})(y+a_-t,t)-(K_y+K_{yy})(y-a_-t,t)\right),
1998: \eqnlbl{42}
1999: $$
2000: where
2001: $$
2002: K(y,t):= \frac{e^{-y^2/4b_-t}}{\sqrt{4\pi b_-t}}
2003: \eqnlbl{43}
2004: $$
2005: denotes an appropriate heat kernel. The pointwise bounds
2006: \eqnref{38}--\eqnref{39} follow immediately for $t\ge 1$
2007: by properties of the heat kernel, in turn yielding \eqnref{36}--\eqnref{37}
2008: in this case. The bounds for small time $t\le 1$ follow from estimates
2009: $$
2010: \aligned
2011: |K_y (y+a_-t,t)-K_y (y-a_-t,t)|
2012: &=|\int^{y-a_-t}_{y+a_-t}K_{yy}(z,t)dz| \\
2013: %
2014: &\le Ct^{-3/2}\int^{y-a_-t}_{y+a_-t} e^{\frac{-z^2}{Mt}} dz\\
2015: &\le Ct^{-1/2}e^{-\frac{(y+a_-t)^2}{Mt}},\\
2016: \endaligned
2017: \eqnlbl{44}
2018: $$
2019: and, similarly,
2020: $$
2021: \aligned
2022: |K_{yy}(y+a_-t,t)-K_{yy}(y-a_-,t)|
2023: &= |\int^{a_-t}_{-a_-t}K_{yyy}(z,t)dz|\\
2024: %
2025: &\le Ct^{-2}\int^{y-a_-t}_{y+a_-t} e^{\frac{-z^2}{Mt}} dz,\\
2026: &\le Ct^{-1}e^{-\frac{(y+a_-t)^2}{Mt}}.\\
2027: \endaligned
2028: \eqnlbl{45}
2029: $$
2030: The bounds for $|{e}_y|$ are again immediate.
2031: Note that we have taken crucial account of cancellation in
2032: the small time estimates of $e_t$, $e_{ty}$.
2033: \qed
2034:
2035: {\bf Remark \thmlbl{2.6}:}
2036: For $t\le 1$, a calculation analogous to that of
2037: \eqnref{44} yields
2038: $
2039: |e_y(y,t)|\le
2040: Ce^{-\frac{(y+a_-t)^2}{Mt}},
2041: $
2042: and thus $|e(\cdot,s)|_{L^1}\to 0$ as $s\to 0$.
2043:
2044: \medskip
2045:
2046: With these preparations, we are ready to carry out our analysis:
2047: \medskip
2048:
2049: {\bf Proof of Theorem \thmref{nonlin}.}
2050: %NOTE: very similar to general symm. case in [MZ.1],
2051: %i.e., just cut and paste and add a derivative in the argument...
2052: Define
2053: $$
2054: \aligned
2055: \zeta(t)
2056: &:= \sup_{0\le s \le t,\, 2\le p\le \infty}
2057: \Big[ \, \big(|U(\cdot, s)|_{L^p}+ |v_x(\cdot, s)|_{L^p}\big)
2058: (1+s)^{\frac{1}{2}(1-1/p)}\\
2059: &\qquad
2060: + |\dot \delta (s)|(1+s)^{1/2}
2061: + \big( |\delta (s)|
2062: + |U(\cdot, s)|_{H^2}
2063: + |U(\cdot, s)|_{W^{2,\infty}}\big)
2064: \, \Big].\\
2065: \endaligned
2066: \eqnlbl{zeta2}
2067: $$
2068: We shall establish:
2069:
2070: {\it Claim.} For all $t\ge 0$ for which a solution exists with
2071: $\zeta$ uniformly bounded by some fixed, sufficiently small constant,
2072: there holds
2073: %
2074: $$
2075: \zeta(t) \leq C_2(\zeta_0 + \zeta(t)^2).
2076: \eqnlbl{claim}
2077: $$
2078: \medskip
2079: {}From this result, it follows by continuous induction that,
2080: provided $\zeta_0 < 1/4C_2$,
2081: there holds
2082: $$
2083: \zeta(t) \leq 2C_2\zeta_0
2084: \eqnlbl{bd}
2085: $$
2086: for all $t\geq 0$ such that $\zeta$ remains small.
2087: By standard short-time theory/continuation, we find that
2088: the solution (unique, in this regularity class) in fact remains in
2089: $H^2$ for all $t\ge 0$,
2090: with bound \eqnref{bd}, at once yielding existence and the claimed
2091: sharp $L^p$ bounds, $2\le p\le \infty$.
2092: %
2093: Thus, it remains only to establish the claim above.
2094: \medskip
2095:
2096: {\it Proof of Claim.}
2097: We must show that each of the quantities $|U|_{L^p}(1+s)^{\frac{1}{2}(1-1/p)}$,
2098: $|v_x|_{L^p} (1+s)^{\frac{1}{2}(1-1/p)}$,
2099: $|\dot \delta|(1+s)^{1/2}$, $|\delta|$, $|U|_{H^2}$,
2100: and $|U|_{W^{2,\infty}}$ are separately bounded by
2101: $$
2102: C(\zeta_0 + \zeta(t)^2),
2103: \eqnlbl{sep}
2104: $$
2105: for some $C>0$, all $0\le s\le t$, so long as $\zeta$ remains
2106: sufficiently small.
2107:
2108: Provided that we can establish the others, the
2109: final two bounds follow easily from the energy estimates of
2110: Proposition \thmref{energy}.
2111: For, by bounds \eqnref{sep} on $|\dot\delta|$ and $|U|_{L^\infty}$,
2112: and the assumption that $|U|_{H^2\cap W^{2,\infty}} \le \zeta$
2113: remains small, we obtain from \eqnref{ebounds} that
2114: $$
2115: |U|_{H^3}(s)\le
2116: C_3(\zeta_0 + \zeta(t)^2)\log (1+S),
2117: \eqnlbl{logbound}
2118: $$
2119: for $0\le s\le t$.
2120: Interpolating with assumed bound
2121: $$
2122: |U|_{L^2}(s)\le
2123: C_3(\zeta_0 + \zeta(t)^2)(1+s)^{-1/4},
2124: $$
2125: we obtain
2126: $$
2127: \aligned
2128: |U|_{H^2}(s)&\le
2129: C_3(\zeta_0 + \zeta(t)^2)|U|_{H^3}^{2/6}|U|_{L^2}^{1/6}\\
2130: &\le
2131: C_3(\zeta_0 + \zeta(t)^2)(1+t)^{-1/25},
2132: \endaligned
2133: \eqnlbl{H2bound}
2134: $$
2135: and, by Sobolev estimate,
2136: $$
2137: |U|_{W^{2,\infty}}\le
2138: |U|_{H^2}^{1/2} |U|_{H^3}^{1/2}
2139: \le
2140: C_3(\zeta_0 + \zeta(t)^2)(1+t)^{-1/51},
2141: \eqnlbl{Wkpbound}
2142: $$
2143: both better than claimed.
2144:
2145: Thus, in order to establish the result,
2146: we have only to establish the remaining bounds,
2147: on $|U|_{L^p}$, $|v_x|_{L^p}$,
2148: $|\dot \delta|$, and $|\delta|$.
2149: These will be carried out using the Green's function
2150: estimates of Lemma \thmref{2.05}.
2151: Accordingly, we first convert the problem to conservative,
2152: $G$ coordinates, via:
2153:
2154: {\it Observation \thmlbl{Gequiv}.}
2155: It is sufficient to establish corresponding bounds on
2156: %NOTE: (in order)
2157: $|G|_{L^p}$, $|v_{G,x}|_{L^p}$,
2158: $|\dot \delta|$, and $|\delta|$, where
2159: $
2160: v_G(x,t):=A^0(x)^{-1}G(x,t).
2161: %\eqnlbl{vGdef}
2162: $
2163: \medskip
2164:
2165: {\it Proof.}
2166: We have
2167: $$
2168: \aligned
2169: U&=G^{-1}(\tilde G) - G^{-1}(\bar G)
2170: = (A^0)^{-1}_{\text{ave}}(x,t)G\\
2171: &:=\left(\int_0^1 (A^0)^{-1}\big(\bar U(x) + \theta U(x,t)\big)d\theta
2172: \right) \, G,
2173: \endaligned
2174: \eqnlbl{roe}
2175: $$
2176: where
2177: $$
2178: |(A^0)^{-1}_{\text{ave}}(x,t) -A^0(x)^{-1}|
2179: \le
2180: C|U|
2181: \le
2182: C|G|
2183: \eqnlbl{linfty}
2184: $$
2185: and
2186: $$
2187: |\left((A^0)^{-1}_{\text{ave}}(x,t) -A^0(x)^{-1}\right)_x|
2188: \le
2189: C(|\bar U_x|+ |U_x|)
2190: \le
2191: \zeta,
2192: \eqnlbl{Winfty}
2193: $$
2194: whence
2195: $$
2196: |U|_{L^p}\le C|G|_{L^p}
2197: $$
2198: and
2199: $$
2200: |U_x - ((A^0)^{-1}G)_x|_{L^p}\le
2201: C(|G|_{L^p}|G_x|_{L^\infty} + \zeta |G|_{L^p})
2202: \le
2203: C\zeta |G|_{L^p},
2204: $$
2205: from which the result easily follows.
2206: \myqed
2207:
2208: By \eqnref{32}--\eqnref{33}, we have
2209: $$
2210: \aligned
2211: |G|_{L^p}(t)&\le
2212: |\int^\infty_{-\infty} (H+\tG)(x,t;y)G_0(y)dy|_{L^p}\\
2213: %
2214: &+|\int^t_0
2215: \int^\infty_{-\infty}H(x,t-s;y)(N(G,G_x)_x
2216: +\dot \delta G_x)(y,s)dy \, ds |_{L^p}\\
2217: &+|\int^t_0
2218: \int^\infty_{-\infty}\tG_y(x,t-s;y)(N(G,G_x)+
2219: \dot \delta G)(y,s) dy \, ds|_{L^p}\\
2220: &=: I_a + I_b + I_c ,\\
2221: \endaligned
2222: \eqnlbl{reduced}
2223: $$
2224: $$
2225: \aligned
2226: |v_{G,x}|_{L^p}(t)
2227: &:=
2228: |\big((A^0)^{-1}G\big)_x|_{L^p}(t)
2229: \le
2230: |\int^\infty_{-\infty} \left((A^0)^{-1}\tG\right)_x(x,t;y)G_0(y)dy|_{L^p}\\
2231: %
2232: &+|\int^t_0
2233: \int^\infty_{-\infty}\big(
2234: (A^0)^{-1} \tG_y\big)_x(x,t-s;y)(N(G,G_x)+
2235: \dot \delta G)(y,s) dy \, ds|_{L^p}\\
2236: &=: II_a + II_b ,\\
2237: \endaligned
2238: \eqnlbl{vreduced}
2239: $$
2240: $$
2241: \aligned
2242: |\dot \delta| (t)
2243: &\le |\int^\infty_{-\infty}e_t(y,t) G_0(y)dy|\\
2244: %
2245: &+|\int^t_0\int^{+\infty}_{-\infty} e_{yt}(y,t-s)(N(G,G_x)+
2246: \dot \delta G)(y,s) dy ds|\\
2247: &=: III_a + III_b,\\
2248: \endaligned
2249: \eqnlbl{deltadot}
2250: $$
2251: and
2252: $$
2253: \aligned
2254: |\delta| (t)
2255: &\le |\int^\infty_{-\infty}e(y,t) G_0(y)dy|\\
2256: %
2257: &+|\int^t_0\int^{+\infty}_{-\infty} e_{y}(y,t-s)(N(G,G_x)+
2258: \dot \delta G)(y,s) dy ds|\\
2259: &=: IV_a + IV_b.\\
2260: \endaligned
2261: \eqnlbl{delta}
2262: $$
2263:
2264: We estimate each term in turn, following the approach of [Z.2,MZ.1].
2265: The linear term $I_a$ satisfies bound
2266: $$
2267: I_a \le C\zeta_0 (1+t)^{-\frac{1}{2}(1-1/p)},
2268: \eqnlbl{Ia}
2269: $$
2270: as already shown in
2271: the proof of Corollary \thmref{suff}.
2272: Likewise, applying the bounds of Lemma \thmref{2.05} together
2273: with \eqnref{29.1} and definition \eqnref{zeta2}, we have
2274: $$
2275: \aligned
2276: I_b&=
2277: |\int^t_0 \int^\infty_{-\infty}H(x,t-s;y)(N(G,G_x)_x
2278: +\dot \delta G_x)(y,s)dy \, ds |_{L^p}\\
2279: &\le
2280: C\int_0^t e^{-\eta (t-s)}(|G|_{L^\infty}+|v_{G,x}|_{L^\infty}
2281: + |\dot \delta|)|G|_{W^{2,p}}(s) ds\\
2282: &\le
2283: C\zeta(t)^2
2284: \int_0^t e^{-\eta (t-s)}(1+s)^{-1/2}ds\\
2285: &\le
2286: C\zeta(t)^2 (1+t)^{-1/2},\\
2287: \endaligned
2288: \eqnlbl{Ib}
2289: $$
2290: and (taking $q=2$ in \eqnref{tGybounds})
2291: $$
2292: \aligned
2293: I_c&=
2294: |\int^t_0 \int^\infty_{-\infty}\tG_y(x,t-s;y)(N(G,G_x)+
2295: \dot \delta G)(y,s) dy \, ds|_{L^p}\\
2296: &\le
2297: C\int_{0}^t e^{-\eta(t-s)}(|G|_{L^\infty}+|v_{G,x}|_{L^\infty}
2298: + |\dot \delta|)|G|_{L^p}(s) ds\\
2299: &\quad+
2300: C\int_0^{t} (t-s)^{-3/4+1/2p}
2301: (|G|_{L^\infty}+|v_{G,x}|_{L^\infty}
2302: + |\dot \delta|)|G|_{L^2}(s) ds\\
2303: &\le
2304: C\zeta(t)^2
2305: \int_{0}^t e^{-\eta(t-s)}
2306: (1+s)^{-\frac{1}{2}(1-1/p)-1/2}ds\\
2307: &\quad +
2308: C\zeta(t)^2
2309: \int_0^{t} (t-s)^{-3/4+1/2p}
2310: (1+s)^{-3/4}ds\\
2311: &\le
2312: C\zeta(t)^2 (1+t)^{-\frac{1}{2}(1-1/p)}.\\
2313: \endaligned
2314: \eqnlbl{Ic}
2315: $$
2316: Summing bounds \eqnref{Ia}--\eqnref{Ic}, we obtain \eqnref{sep},
2317: as claimed, for $2\le p\le \infty$.
2318: The desired bounds on $II_a$ and $II_b$ follow by an identical
2319: calculation, once we notice that, in Lemma \thmref{2.05},
2320: $\left((A^0)^{-1}\tG\right)$ and
2321: $\left((A^0)^{-1}\tG\right)_y$
2322: satisfy the same $L^q\to L^p$ bounds as do $(H+\tG)$ and $(H+\tG)_y$,
2323: respectively.
2324:
2325: Similarly, applying the bounds of Lemma \thmref{3} together
2326: with definition \eqnref{zeta2}, we find that
2327: $$
2328: \aligned
2329: III_a&=
2330: \big|\int^\infty_{-\infty}e_t(y,t) G_0(y)dy\big|\\
2331: &\le
2332: |e_t(y,t)|_{L^\infty}(t) |G_0|_{L^1}\\
2333: &\le
2334: C\zeta_0 (1+t)^{-1/2}
2335: \endaligned
2336: \eqnlbl{IIIa}
2337: $$
2338: and
2339: $$
2340: \aligned
2341: III_b&=
2342: |\int^t_0\int^{+\infty}_{-\infty} e_{yt}(y,t-s)(N(G,G_x)+
2343: \dot \delta G)(y,s) dy ds|\\
2344: &\le
2345: \int^t_0
2346: |e_{yt}|_{L^2}(t-s)
2347: (|G|_{L^\infty}+|v_{G,x}|_{L^\infty} + |\dot \delta|)
2348: |G|_{L^2}(s) ds\\
2349: &\le
2350: C\zeta(t)^2
2351: \int^t_0
2352: (t-s)^{-3/4} (1+s)^{-3/4} ds\\
2353: &\le
2354: C\zeta(t)^2 (1+t)^{-1/2},
2355: \endaligned
2356: \eqnlbl{IIIb}
2357: $$
2358: while
2359: $$
2360: \aligned
2361: IV_a&=
2362: \big|\int^\infty_{-\infty}e(y,t) G_0(y)dy\big|\\
2363: &\le
2364: |e(y,t)|_{L^\infty}(t) |G_0|_{L^1}\\
2365: &\le
2366: C\zeta_0
2367: \endaligned
2368: \eqnlbl{IVa}
2369: $$
2370: and
2371: $$
2372: \aligned
2373: IV_b&=
2374: \big|\int^t_0\int^{+\infty}_{-\infty} e_{y}(y,t-s)
2375: \dot \delta G(y,s) dy ds\big|\\
2376: &\le
2377: \int^t_0
2378: |e_{y}|_{L^2}(t-s)
2379: (|G|_{L^\infty}+|v_{G,x}|_{L^\infty} + |\dot \delta|)
2380: |G|_{L^2}(s) ds\\
2381: &\le
2382: C\zeta(t)^2
2383: \int^t_0
2384: (t-s)^{-1/4} (1+s)^{-3/4} ds\\
2385: &\le
2386: C\zeta(t)^2.
2387: \endaligned
2388: \eqnlbl{IVb}
2389: $$
2390: Summing \eqnref{IIIa}--\eqnref{IIIb} and
2391: \eqnref{IVa}--\eqnref{IVb},
2392: we obtain \eqnref{sep} as claimed.
2393:
2394: This completes the proof of the claim, and the result.
2395: \myqed
2396:
2397: \bigskip
2398:
2399: \Refs
2400: \medskip\noindent
2401: %[AGJ] J. Alexander, R. Gardner and C.K.R.T. Jones,
2402: %{\it A topological invariant arising in the analysis of
2403: %traveling waves}, J. Reine Angew. Math. 410 (1990) 167--212.
2404: %\medskip\noindent
2405: %[BE] A.A. Barmin and S.A. Egorushkin,
2406: %{\it Stability of shock waves,}
2407: %Adv. Mech. 15 (1992) No. 1--2, 3--37.
2408: %\medskip\noindent
2409: %[BSZ] S. Benzoni--Gavage, D. Serre, and K. Zumbrun,
2410: %{\it Alternate Evans functions and viscous shock waves,}
2411: %to appear, SIAM J. Math. Anal.
2412: %\medskip\noindent
2413: %\medskip\noindent
2414: %[Be] H.A. Bethe,
2415: %{\it On the theory of shock waves for an arbitrary equation of state,}
2416: %Office of Scientific Research and Development, Rept. 545,
2417: %Serial No. NDRC-B-237, Office Sci. Res.
2418: %Develop., U.S. Army Ballistic Research Laboratory,
2419: %Aberdeen Proving Ground, MD (1942), in:
2420: %{\it Classic papers in shock compression science,} Springer, New York (1982) 421--492.
2421: %\medskip\noindent
2422: %\medskip\noindent
2423: %[Br.1] L. Q. Brin, {\it Numerical testing of the stability of viscous
2424: %shock waves}, Ph.D. dissertation, Indiana University, May 1998.
2425: %\medskip\noindent
2426: %[Br.2] L. Q. Brin,
2427: %{\it Numerical testing of the stability of viscous shock waves,}
2428: %to appear, Math. Comp.
2429: %\medskip\noindent
2430: %[Br] L. Brin,
2431: %{\it Numerical testing of the stability of viscous shock waves},
2432: %Doctoral thesis, Indiana University (1998).
2433: %\medskip\noindent
2434: %\medskip\noindent
2435: %[CL] E.A. Coddington and N. Levinson, {\it Theory of ordinary
2436: %differential equations,} McGraw-Hill Book Company, Inc.,
2437: %New York-Toronto-London (1955) xii+429 pp.
2438: \medskip\noindent
2439: [CS] C. Conley and J. Smoller,
2440: {\it On the structure of magnetohydrodynamic shock waves,}
2441: Comm. Pure Appl. Math. 28 (1974),
2442: %no.?
2443: 367--375.
2444: %\medskip\noindent
2445: %[Co] W. A. Coppel,
2446: %{\it Stability and asymptotic behavior of differential equations},
2447: %D.C. Heath and Co., Boston, MA (1965) viii+166 pp.
2448: %\medskip\noindent
2449: %[CF] R. Courant and K.O. Friedrichs,
2450: %{\it Supersonic flow and shock waves,}
2451: %Springer--Verlag, New York (1976) xvi+464 pp.
2452: %\medskip\noindent
2453: %[CH.1] R. Courant and D. Hilbert,
2454: %{\it Methods of mathematical physics. Vol. I,}
2455: %Interscience Publishers, Inc., New York, N.Y., 1953. xv+561 pp.
2456: %\medskip\noindent
2457: %[CH.2] R. Courant and D. Hilbert,
2458: %{\it Methods of mathematical physics. Vol. II:
2459: %Partial differential equations,}
2460: %(Vol. II by R. Courant.) Interscience Publishers (a division
2461: %of John Wiley \& Sons), New York-Lon don 1962 xxii+830 pp.
2462: \medskip\noindent
2463: [Er] J. J. Erpenbeck,
2464: {\it Stability of step shocks,} Phys. Fluids 5 (1962) no. 10, 1181--1187.
2465: %\medskip\noindent
2466: %[E.1] J.W. Evans,
2467: %{\it Nerve axon equations: I. Linear approximations,}
2468: %Ind. Univ. Math. J. 21 (1972) 877--885.
2469: %\medskip\noindent
2470: %[E.2] J.W. Evans,
2471: %{\it Nerve axon equations: II. Stability at rest,}
2472: %Ind. Univ. Math. J. 22 (1972) 75--90.
2473: %\medskip\noindent
2474: %[E.3] J.W. Evans,
2475: %{\it Nerve axon equations: III. Stability of the nerve impulse,}
2476: %Ind. Univ. Math. J. 22 (1972) 577--593.
2477: %\medskip\noindent
2478: %[E.4] J.W. Evans,
2479: %{\it Nerve axon equations: IV. The stable and the unstable impulse,}
2480: %Ind. Univ. Math. J. 24 (1975) 1169--1190.
2481: %\medskip\noindent
2482: %[Ev] L.C. Evans,
2483: %{\it Partial differential equations,} Graduate Studies in Mathematics,
2484: %19. American Mathematical Society, Providence, RI,
2485: %1998. xviii+662 pp. ISBN: 0-8218-0772-2.
2486: \medskip\noindent
2487: [Fre.1] H. Freist\"uhler,
2488: {\it Small amplitude intermediate magnetohydrodynamic shock waves,}
2489: Phys. Scripta T74 (1998) 26--29.
2490: \medskip\noindent
2491: [Fre.2] H. Freist\"uhler,
2492: {\it Profiles for small Laxian shock waves in Kawashima
2493: type systems of conservation laws,} Preprint (2000).
2494: \medskip\noindent
2495: [FreZ] H. Freist\"uhler and K. Zumbrun,
2496: {\it Examples of unstable viscous shock waves,}
2497: unpublished note, Institut f\"ur Mathematik, RWTH Aachen, February 1998.
2498: %\medskip\noindent
2499: %[Fr] A. Friedman,
2500: %{\it Partial differential equations of parabolic type},
2501: %Prentice-Hall, Englewood Cliffs, NY (1964), Reprint Ed. (1983).
2502: \medskip\noindent
2503: [Fri.1] C. Fries,
2504: {\it Stability of viscous shock waves
2505: associated with non-convex modes,}
2506: Arch. Ration. Mech. Anal. 152 (2000), no. 2, 141--186.
2507: \medskip\noindent
2508: [Fri.2] C. Fries,
2509: {\it Nonlinear asymptotic stability of general
2510: small-amplitude viscous Laxian shock waves,}
2511: J. Differential Equations 146 (1998), no. 1, 185--202.
2512: %\medskip\noindent
2513: %[GJ.1] R. Gardner and C.K.R.T. Jones,
2514: %{\it A stability index for steady state solutions of
2515: %boundary value problems for parabolic systems},
2516: %J. Diff. Eqs. 91 (1991), no. 2, 181--203.
2517: %\medskip\noindent
2518: %[GJ.2] R. Gardner and C.K.R.T. Jones,
2519: %{\it Traveling waves of a perturbed diffusion equation
2520: %arising in a phase field model},
2521: %Ind. Univ. Math. J. 38 (1989), no. 4, 1197--1222.
2522: \medskip\noindent
2523: [GZ] R. Gardner and K. Zumbrun,
2524: {\it The Gap Lemma and geometric criteria for instability
2525: of viscous shock profiles},
2526: Comm. Pure Appl. Math. 51 (1998), no. 7, 797--855.
2527: \medskip\noindent
2528: %[Ge] Germaine, MHD profiles, TODO. (check Fre.2).
2529: %\medskip\noindent
2530: [Gi] D. Gilbarg,
2531: {\it The existence and limit behavior of
2532: the one-dimensional shock layer,} Amer. J. Math. 73, (1951). 256--274.
2533: \medskip\noindent
2534: [Go.1] J. Goodman, {\it Nonlinear asymptotic stability of viscous
2535: shock profiles for conservation laws,}
2536: Arch. Rational Mech. Anal. 95 (1986), no. 4, 325--344.
2537: \medskip\noindent
2538: [Go.2] J. Goodman,
2539: {\it Remarks on the stability of viscous shock waves}, in:
2540: Viscous profiles and numerical methods for shock waves
2541: (Raleigh, NC, 1990), 66--72, SIAM, Philadelphia, PA, (1991).
2542: %\medskip\noindent
2543: %[GH] J. Guckenheimer and P. Holmes,
2544: %{\it Nonlinear oscillations, dynamical systems, and bifurcations of vector fields},
2545: %(Revised and corrected reprint of the 1983 original),
2546: %Springer--Verlag, New York (1990), xvi+459 pp.
2547: %\medskip\noindent
2548: %[HK] J. Hale and H. Kocak,
2549: %TODO: accent under c...
2550: %{\it Dynamics and bifurcations,}
2551: %Texts in Applied Mathematics, 3. Springer-Verlag, New York, 1991.
2552: %xiv+568 pp. ISBN: 0-387-97141-6.
2553: %\medskip\noindent
2554: %[He] D. Henry,
2555: %{\it Geometric theory of semilinear parabolic equations},
2556: %Lecture Notes in Mathematics, Springer--Verlag, Berlin (1981),
2557: %iv + 348 pp.
2558: %\medskip\noindent
2559: %[HoZ.3] D. Hoff and K. Zumbrun,
2560: %{\it Asymptotic behavior of multi-dimensional scalar viscous shock fronts,}
2561: %to appear, Indiana Univ. Math. J. (2000).
2562: %\medskip\noindent
2563: \medskip\noindent
2564: [HoZ] D. Hoff and K. Zumbrun,
2565: {\it Multi-dimensional diffusion waves for the Navier-Stokes equations of
2566: compressible flow,} Indiana Univ. Math. J. 44 (1995), no. 2, 603--676.
2567: %\medskip\noindent
2568: %[HoZ.2] D. Hoff and K. Zumbrun,
2569: %{\it Pointwise Green's function bounds for multi-dimensional
2570: %scalar viscous shock fronts,} preprint (2000).
2571: %\medskip\noindent
2572: %[H.1] P. Howard,
2573: %{\it Pointwise estimates on the Green's function
2574: %for a scalar linear convection-diffusion equation,}
2575: %J. Differential Equations 155 (1999), no. 2, 327--367.
2576: %\medskip\noindent
2577: %[H.2] P. Howard,
2578: %{\it Pointwise methods for stability of a scalar conservation law,}
2579: %Doctoral thesis (1998).
2580: %\medskip\noindent
2581: %[H.3] P. Howard,
2582: %{\it Pointwise Green's function approach to
2583: %stability for scalar conservation laws,}
2584: %Comm. Pure Appl. Math. 52 (1999), no. 10, 1295--1313.
2585: \medskip\noindent
2586: [HZ.1] P. Howard and K. Zumbrun,
2587: {\it A tracking mechanism for one-dimensional
2588: viscous shock waves,} preprint (1999).
2589: \medskip\noindent
2590: [HZ.2] P. Howard and K. Zumbrun,
2591: {\it Pointwise estimates for dispersive-diffusive shock waves,}
2592: to appear, Arch. Rational Mech. Anal.
2593: \medskip\noindent
2594: [HuZ] J. Humpherys and K. Zumbrun,
2595: {\it Spectral stability of small amplitude
2596: shock profiles for dissipative symmetric hyperbolic--parabolic systems,}
2597: to appear ZAMP (2001).
2598: %\medskip\noindent
2599: %[JX] S. Jin and Z. Xin,
2600: %{\it The relaxation schemes for systems of conservation laws
2601: %in arbitrary space dimensions,}
2602: %Comm. Pure Appl. Math. 48 (1995), no. 3, 235--276.
2603: %\medskip\noindent
2604: %[J] C.K.R.T. Jones,
2605: %{\it Stability of the travelling wave solution of the FitzHugh--Nagumo system},
2606: %Trans. Amer. Math. Soc. 286 (1984), no. 2, 431--469.
2607: %\medskip\noindent
2608: %[JGK] C. K. R. T. Jones, R. A. Gardner, and T. Kapitula,
2609: %{\it Stability of travelling waves
2610: %for non-convex scalar viscous conservation laws},
2611: %Comm. Pure Appl. Math. 46 (1993) 505--526.
2612: \medskip\noindent
2613: [Ka] Ya. Kanel,
2614: {\it On a model system of equations of of one-dimensional
2615: gas motion,} Diff. Eqns. 4 (1968) 374--380.
2616: \medskip\noindent
2617: [K.1] T. Kapitula,
2618: {\it Stability of weak shocks in $\lambda$--$\omega$ systems},
2619: Indiana Univ. Math. J. 40 (1991), no. 4, 1193--12.
2620: \medskip\noindent
2621: [K.2] T. Kapitula,
2622: {\it On the stability of travelling waves in weighted $L^\infty$ spaces},
2623: J. Diff. Eqs. 112 (1994), no. 1, 179--215.
2624: %\medskip\noindent
2625: %[KS] T. Kapitula and B. Sandstede,
2626: %{\it Stability of bright solitary-wave solutions
2627: %to perturbed nonlinear Schrödinger equations,} Phys. D 124
2628: %(1998), no. 1-3, 58--103.
2629: \medskip\noindent
2630: [Kaw] S. Kawashima,
2631: {\it Systems of a hyperbolic--parabolic composite type,
2632: with applications to the equations of magnetohydrodynamics},
2633: thesis, Kyoto University (1983).
2634: \medskip\noindent
2635: [KM] S. Kawashima and A. Matsumura,
2636: {\it Asymptotic stability of traveling wave solutions of systems
2637: for one-dimensional gas motion,} Comm.
2638: Math. Phys. 101 (1985), no. 1, 97--127.
2639: \medskip\noindent
2640: [KMN] S. Kawashima, A. Matsumura, and K. Nishihara,
2641: {\it Asymptotic behavior of solutions for the equations of a
2642: viscous heat-conductive gas,}
2643: Proc. Japan Acad. Ser. A Math. Sci. 62 (1986), no. 7, 249--252.
2644: %\medskip\noindent
2645: %[KSy] S. Kawashima and Y. Shizuta,
2646: %{\it On the normal form of the symmetric hyperbolic-parabolic
2647: %systems associated with the conservation laws},
2648: %Tohoku Math. J. 40 (1988) 449--464.
2649: %\medskip\noindent=
2650: %NOTE: reference found by Denis for boundary layer paper,
2651: %doesn't fit discussion here.
2652: %[Kat] T. Kato,
2653: %{\it Perturbation theory for linear operators},
2654: %Springer--Verlag, Berlin Heidelberg (1985).
2655: %\medskip\noindent
2656: %[KK.1]
2657: %G. Kreiss and H.O. Kreiss,
2658: %{\it Stability of systems of viscous conservation laws,}
2659: %Comm. Pure Appl. Math. 51 (1998), no. 11-12, 1397--1424.
2660: %\medskip\noindent
2661: %[LU] O.A. Ladyzenskaja and N.N. Ural'tseva,
2662: %{\it On linear and quasij-linear equations
2663: %and systems of elliptic and parabolic types,} 1963 Outlines Joint Sympos.
2664: %Partial Differential Equations (Novosibirsk, 1963) 146--150 Acad. Sci. USSR
2665: %Siberian Branch, Moscow.
2666: %\medskip\noindent
2667: %[LSU] O.A. Ladyzenskaja, V.A. Solonnikov, and N.N. Ural'tseva,
2668: %{\it Linear and quasi-linear equations of parabolic type,}
2669: %Translations of Math. Monographs 23, AMS, Providence, RI (1968).
2670: \medskip\noindent
2671: [La] P.D. Lax,
2672: {\it Hyperbolic systems of conservation laws and the mathematical theory of shock waves},
2673: Conference Board of the Mathematical Sciences Regional Conference
2674: Series in Applied Mathematics, No. 11. Society for Industrial and Applied Mathematics,
2675: Philadelphia, Pa., 1973. v+48 pp.
2676: %\medskip\noindent
2677: %[LP] Lax, Peter D.; Phillips, Ralph S. Scattering theory. Second
2678: %edition. With appendices by Cathleen S. Morawetz and Georg Schmidt.
2679: %Pure and Applied Mathematics, 26. Academic Press, Inc.,
2680: %Boston, MA, 1989. xii+309 pp. ISBN: 0-12-440051-5.
2681: %\medskip\noindent
2682: %[Le] E.E. Levi, {\it Sulle equazioni lineari totalmente ellittiche
2683: %alle derivate parziali,}
2684: %Rend. Circ. Mat. Palermo 24 (1907) 275--317.
2685: %%\medskip\noindent
2686: %[L.1] T.-P. Liu,
2687: %{\it Nonlinear stability of shock waves for viscous conservation laws},
2688: %Mem. Amer. Math. Soc. 56 (1985), no. 328, v+108 pp.
2689: \medskip\noindent
2690: [L.1] T.-P. Liu,
2691: {\it Pointwise convergence to shock waves for viscous conservation laws},
2692: Comm. Pure Appl. Math. 50 (1997), no. 11, 1113--1182.
2693: \medskip\noindent
2694: [L.2] T.-P. Liu,
2695: {\it Shock waves for compressible Navier--Stokes equations are stable,}
2696: Comm. Pure Appl. Math. 39 (1986), no. 5, 565--594.
2697: \medskip\noindent
2698: [L.3] T.-P. Liu,
2699: {\it The Riemann problem for general systems of conservation laws,}
2700: J. Diff. Eqs. 18 (1975) 218--234.
2701: \medskip\noindent
2702: [LZ.1] T.P. Liu and K. Zumbrun,
2703: {\it Nonlinear stability of an undercompressive shock for complex
2704: Burgers equation,} Comm. Math. Phys. 168 (1995), no. 1, 163--186.
2705: \medskip\noindent
2706: [LZ.2] T.P. Liu and K. Zumbrun,
2707: {\it On nonlinear stability of general undercompressive viscous shock waves,}
2708: Comm. Math. Phys. 174 (1995), no. 2, 319--345.
2709: \medskip\noindent
2710: [LZe.1] T.-P. Liu and Y. Zeng,
2711: {\it Large time behavior of solutions for general
2712: quasilinear hyperbolic--parabolic systems of conservation laws},
2713: AMS memoirs 599 (1997).
2714: \medskip\noindent
2715: [LZe.2] T.-P. Liu and Y. Zeng,
2716: {\it Private communication} (manuscript in preparation).
2717: \medskip\noindent
2718: [M.1], A. Majda,
2719: {\it The stability of multi-dimensional shock fronts -- a
2720: new problem for linear hyperbolic equations,}
2721: Mem. Amer. Math. Soc. 275 (1983).
2722: \medskip\noindent
2723: [M.2], A. Majda,
2724: {\it The existence of multi-dimensional shock fronts,}
2725: Mem. Amer. Math. Soc. 281 (1983).
2726: \medskip\noindent
2727: [M.3] A. Majda,
2728: {\it Compressible fluid flow and systems of conservation laws in several
2729: space variables,} Springer-Verlag, New York (1984), viii+ 159 pp.
2730: \medskip\noindent
2731: [MP] A. Majda and R. Pego,
2732: {\it Stable viscosity matrices for systems of conservation laws},
2733: J. Diff. Eqs. 56 (1985) 229--262.
2734: \medskip\noindent
2735: [MN] A. Matsumura and K. Nishihara,
2736: {\it On the stability of travelling wave solutions of a one-dimensional
2737: model system for compressible viscous gas,}
2738: Japan J. Appl. Math. 2 (1985), no. 1, 17--25.
2739: %\medskip\noindent
2740: %[MeP] R. Menikoff and B. Plohr,
2741: %{\it The Riemann problem for fluid flow of real materials,}
2742: %Rev. Modern Phys. 61 (1989), no. 1, 75--130.
2743: %\medskip\noindent
2744: \medskip\noindent
2745: [MZ.1] C. Mascia and K. Zumbrun,
2746: {\it Pointwise Green's function bounds and stability of relaxation
2747: shocks,} preprint (2001).
2748: \medskip\noindent
2749: [MZ.2] C. Mascia and K. Zumbrun,
2750: {\it Pointwise Green's function bounds for shock profiles
2751: with degenerate viscosity,} in preparation.
2752: %[Pa] A. Pazy, {\it Semigroups of linear operators and applications
2753: %to partial differential equations,} Applied Mathematical Sciences, 44,
2754: %Springer-Verlag, New York-Berlin, (1983) viii+279 pp. ISBN: 0-387-90845-5.
2755: %\medskip\noindent
2756: %[PW] R. L. Pego and M.I. Weinstein,
2757: %{\it Eigenvalues, and instabilities of solitary waves},
2758: %Philos. Trans. Roy. Soc. London Ser. A 340 (1992), no. 1656, 47--94.
2759: %\medskip\noindent
2760: %\medskip\noindent
2761: %[RS] M. Reed and B. Simon,
2762: %{\it Methods of modern mathematical physics. I-IV},
2763: %Academic Press, Inc. [Harcourt Brace Jovanovich,
2764: %Publishers], New York--London, 1980.
2765: %\medskip\noindent
2766: %[Sat] D. Sattinger,
2767: %{\it On the stability of waves of nonlinear parabolic systems},
2768: %Adv. Math. 22 (1976) 312--355.
2769: \medskip\noindent
2770: [SZ] D. Serre and K. Zumbrun,
2771: {\it Boundary layer stability in real-vanishing viscosity limit,}
2772: to appear, Comm. Math. Phys. (2001).
2773: \medskip\noindent
2774: [SK] Y Shizuta and S. Kawashima,
2775: {\it Systems of equations of hyperbolic--parabolic
2776: type with applications to the discrete Boltzmann equation,}
2777: Hokkaido Math. J. 14 (1984) 435--457.
2778: \medskip\noindent
2779: [Sm] J. Smoller,
2780: {\it Shock waves and reaction--diffusion equations,}
2781: Second edition, Grundlehren der Mathematischen Wissenschaften
2782: [Fundamental Principles of Mathematical Sciences], 258. Springer-Verlag,
2783: New York, 1994. xxiv+632 pp. ISBN: 0-387-94259-9.
2784: %\medskip\noindent
2785: %[St] J.C. Strikwerda, {\it Finite difference schemes
2786: %and partial differential equations,}
2787: %(Chapman and Hall, New York (1989) xii+ 386 pp.
2788: %\medskip\noindent
2789: %\medskip\noindent
2790: %[SX.1] A. Szepessy and Z. Xin,
2791: %{\it Nonlinear stability of viscous shock waves,}
2792: %Arch. Rat. Mech. Anal. 122 (1993) 53--103.
2793: %\medskip\noindent
2794: %[Ti] E.C. Titschmarsh,
2795: %{\it Eigenfunction expansion associated with second-order
2796: %differential equations I-II,} Oxford (1946), Revised Ed. (1958).
2797: %\medskip\noindent
2798: \medskip\noindent
2799: [Te] R. Temam,
2800: {\it Some developments on Navier-Stokes equations in the
2801: second half of the 20th century,} Development of mathematics
2802: 1950--2000, 1049--1106, Birkhäuser, Basel, 2000.
2803: %\medskip\noindent
2804: %[Y] K. Yosida, {\it Functional analysis,}
2805: %Reprint of the sixth (1980) edition, Classics in Mathematics,
2806: %Springer-Verlag, Berlin, 1995, xii+501 pp. ISBN: 3-540-58654-7.
2807: \medskip\noindent
2808: [Ze.1] Y. Zeng,
2809: {\it $L^1$ asymptotic behavior of compressible, isentropic,
2810: viscous $1$-d flow,}
2811: Comm. Pure Appl. Math. 47 (1994) 1053--1092.
2812: \medskip\noindent
2813: [ZH] K. Zumbrun and P. Howard,
2814: {\it Pointwise semigroup methods and stability of viscous shock waves,}
2815: Indiana Mathematics Journal V47 (1998) no. 4, 741--871.
2816: \medskip\noindent
2817: [Z.1] K. Zumbrun, {\it Stability of viscous shock waves},
2818: Lecture Notes, Indiana University (1998).
2819: \medskip\noindent
2820: [Z.2] K. Zumbrun, {\it Refined Wave--tracking and Nonlinear
2821: Stability of Viscous Lax Shocks}, to appear, Math. Anal. Appl. (2001);
2822: preprint (1999).
2823: \medskip\noindent
2824: [Z.3] K. Zumbrun, {\it Multidimensional stability of
2825: planar viscous shock waves}, TMR Summer School Lectures:
2826: Kochel am See, May, 1999, to appear,
2827: Birkhauser's Series: Progress in Nonlinear Differential
2828: Equations and their Applications (2001), 207 pp.
2829: \medskip\noindent
2830: [Z.4] K. Zumbrun, {\it Stability of general undercompressive shocks
2831: of viscous conservation laws,}
2832: in preparation.
2833: %TODO: put back in, with appropriate change in ab/intro
2834: %\medskip\noindent
2835: %[Z.6] K. Zumbrun, {\it Stability of shock
2836: %profiles for compressible Navier--Stokes equations,}
2837: %in preparation.
2838: \medskip\noindent
2839: [ZS] K. Zumbrun and D. Serre,
2840: {\it Viscous and inviscid stability of multidimensional
2841: planar shock fronts,} Indiana Univ. Math. J. 48 (1999), no. 3,
2842: 937--992.
2843: \endRefs
2844: %\vfil\eject
2845: \enddocument
2846: