1: % versione rispedita a greg e ugo con correzioni (14/2)

2: 

3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

5: \documentclass{article}

6: \textwidth 169mm

7: \textheight 225mm

8: \topmargin -5mm

9: \oddsidemargin -5mm

10: 

11: %\usepackage{graphicx}

12: \usepackage{amsmath}

13: \usepackage{amssymb}

14: \usepackage{color}

15: 

16: 

17: 

18: 

19: 

20: \usepackage{latexsym}

21: %\usepackage[scanall]{psfrag}

22: \usepackage[english]{babel}

23: \usepackage[dvips]{graphicx}

24: \usepackage{verbatim}

25: \setlength{\parskip}{0cm}

26: %\include{diagrams}

27: 

28: 

29: \newcommand{\fintersect}{forwardly intersect}

30: \newcommand{\bad}{unbounded}

31: 

32: 

33: \newcommand{\Supp}{\mathit{Supp}}

34: \newcommand{\dbt}{(X+Y)/2}

35: %\newcommand{\Xt}{\widetilde X}

36: %\newcommand{\Yt}{\widetilde Y}

37: \newcommand{\Xt}{X'}

38: \newcommand{\Yt}{Y'}

39: \newcommand{\TT}{V_p}

40: 

41: \newcommand{\bbibitem}{\bibitem}

42: %\newcommand{\bbibitem}[1]{{\bibitem{#1} {\it [#1]} }}      

43: 

44: \newcommand{\llabel}[1]{{\label{#1}}}

45: %\newcommand{\llabel}[1]{{\mbox{{~~~\it[#1]}}}{\label{#1}}}

46: 

47: %\newcommand{\ffoot}[1]{}

48: \newcommand{\ffoot}[1]{\footnote{\noindent#1}}

49: 

50: \renewcommand{\r}[1]{(\ref{#1})}

51: 

52: 

53: %\newcommand{\ex}[1]{

54: %

55: %\vspace{1cm}{\tt{\tiny #1}}    } 

56: \newcommand{\ex}[1]{} 

57: 

58: %\input{psfig.tex}

59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% R DI REALE ETC..........

60: \font\tenmsb=msbm10

61: \font\sevenmsb=msbm7

62: \font\fivemsb=msbm5

63: \newfam\msbfam

64: \textfont\msbfam=\tenmsb

65: \scriptfont\msbfam=\sevenmsb

66: \scriptscriptfont\msbfam=\fivemsb

67: \def\Bbb#1{{\fam\msbfam\relax#1}}

68: 

69: 

70: 

71: 

72: %************************************************************************

73: %*********************** MIEI COMANDI  GENERALI 

74: %************************************************************************

75: \newcommand{\bi}{\begin{itemize}}

76: \newcommand{\ei}{\end{itemize}}

77: \newcommand{\bd}{\begin{description}}

78: \newcommand{\ed}{\end{description}}

79: \renewcommand{\i}{\item}

80: 

81: \newcommand{\bqn}{\begin{eqnarray}}

82: \newcommand{\eqn}{\end{eqnarray}}

83: \newcommand{\eqnn}{\nonumber\end{eqnarray}}

84: \newcommand{\eqnl}[1]{\llabel{#1}\end{eqnarray}}

85: 

86: \newcommand{\nn}{\nonumber}

87: \newcommand{\ba}[1]{\begin{array}{#1}}

88: \newcommand{\ea}{\end{array}}

89: 

90: \newcommand{\mr}{\Bbb{R}}

91: \newcommand{\R}%{\Bbb{R}}

92: {\mathbb{R}}

93: \newcommand{\mc}{\Bbb{C}}

94: \newcommand{\N}{\mathbb{N}}

95: 

96: 

97: \newcommand{\fine}{\end{document}}

98: 

99: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

100: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

101: % NEW THEOREM

102: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

103: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

104: \def \trait (#1) (#2) (#3){\vrule width #1pt height #2pt depth #3pt}

105: \def \qed{\hfill

106:         \trait (0.1) (6) (0)

107:         \trait (6) (0.1) (0)

108:         \kern-6pt   

109:         \trait (6) (6) (-5.9)

110:         \trait (0.1) (6) (0)

111: \medskip}

112: \def \qedmio{\hfill

113:              \trait (8) (8) (-0.1)

114:              \medskip}

115: \def \quadp{{\Huge $\qedmio$}}

116: \def \quadv{{\Huge $\qed$}}

117: 

118: \newtheorem{Theorem}{\bf Theorem}

119: \newtheorem{ml}[Theorem]{\bf Lemma}

120: \newtheorem{mo}{\bf \underline{{\sl Observation}}}

121: \newtheorem{mcc}[Theorem]{\bf Corollary}

122: \newtheorem{Definition}[Theorem]{\bf Definition}

123: \newtheorem{mpr}[Theorem]{\bf Proposition}

124: \newtheorem{mproperty}[Theorem]{\bf Property}

125: \newtheorem{corollary}[Theorem]{Corollary}

126: \newtheorem{mrem}{\bf \underline{{\sl Remark}}}

127: \newtheorem{rmk}[Theorem]{Remark}

128: 

129: \newcommand{\bt}{\begin{Theorem}}

130: \newcommand{\et}{\end{Theorem}}

131: \newcommand{\bl}{\begin{ml}}

132: \newcommand{\el}{\end{ml}}

133: \newcommand{\bo}{\noindent\begin{mo}\rm}

134: \newcommand{\eo}{\end{mo}}

135: \newcommand{\bp}{\begin{mpr}}

136: \newcommand{\ep}{\end{mpr}}

137: \newcommand{\bc}{\begin{mcc}}

138: \newcommand{\ec}{\end{mcc}}

139: \newcommand{\bdeff}{\begin{Definition}}

140: \newcommand{\edeff}{\end{Definition}}

141: \newcommand{\bproperty}{\begin{mproperty}}

142: \newcommand{\eproperty}{\end{mproperty}}

143: \newcommand{\brem}{\begin{mrem}\rm}

144: \newcommand{\erem}{\end{mrem}}

145: 

146: 

147: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

148: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

149: % FIGURE

150: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

151: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

152: \newcommand{\bfi}{\begin{figure}}

153: \newcommand{\efi}{\end{figure}}

154: 

155: 

156: 

157: 

158: 

159: \newcommand{\capt}{

160: \refstepcounter{figure}

161: \begin{center}Figure~\thefigure\end{center}

162: }

163: 

164: 

165: 

166: 

167: %\newcommand{\captl}[1]{

168: %\refstepcounter{figure}

169: %\begin{center}Figure~\thefigure\end{center}\label{#1}

170: %}

171: 

172: %\newcommand{\captl}[1]{

173: %\stepcounter{figure}

174: %\addtocounter{figure}{-1}

175: %{

176: %\color{white} 

177: %\caption{\label{#1}

178: %{\color{black} Figure \arabic{figure}}

179: %}}

180: %}

181: 

182: 

183: 

184: \newcommand{\ppotR}[3]

185: {

186: 

187: \begin{figure}\begin{center}

188: ~\includegraphics[width=#3truecm]{#1.eps}\\

189: \caption{#2}

190: \llabel{#1}

191: \end{center}

192: \end{figure}

193: \noindent$\!\!$}

194: 

195: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

196: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

197: %               MIEI COMANDI SPECIALI

198: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

199: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

200: 

201: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

202: %%%%%%%%%%% LETTERE GRECHE %%%%%%%%%%%%%%%%%%%%%%

203: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

204: \newcommand{\Lam}{\Lambda}

205: \newcommand{\lam}{\lambda}

206: \newcommand{\lb}{\lambda}

207: \newcommand{\g}{\gamma}

208: \newcommand{\al}{\alpha}

209: \newcommand{\eps}{\varepsilon}

210: \newcommand{\de}{\delta}

211: \newcommand{\om}{\omega}

212: \newcommand{\Om}{\Omega}

213: 

214: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

215: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

216: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

217: \newcommand{\pro}{$({\cal P})$\ }

218: \newcommand{\con}{{\cal C}}

219: \newcommand{\E}{{\cal E}}

220: \newcommand{\F}{{\cal F}}

221: \newcommand{\I}{{\cal I}}

222: \renewcommand{\O}{{\cal O}}

223: \newcommand{\A}{{\cal A}}

224: \renewcommand{\H}{{\cal H}}

225: \newcommand{\X}{{\cal X}}

226: \newcommand{\Y}{{\cal Y}}

227: \newcommand{\Z}{{\cal Z}}

228: \newcommand{\K}{{\cal K}}

229: \newcommand{\U}{{\cal U}}

230: \newcommand{\T}{{\cal T}}

231: 

232: 

233: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

234: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

235: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

236: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

237: %ultime cose

238: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

239: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

240: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

241: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

242: 

243: \newcommand{\cc}{ constant control }

244: 

245: \newcommand{\gm}{{\g^m_{x_0}}}

246: \newcommand{\gM}{\g^M_{x_0}}

247: \newcommand{\qq}{{\cal Z}}

248: \newcommand{\dea}{\delta_A}

249: \newcommand{\deb}{\delta_B}

250: \newcommand{\oma}{\omega_A}

251: \newcommand{\omb}{\omega_B}

252: \newcommand{\roa}{\rho_A}

253: \newcommand{\rob}{\rho_B}

254: \newcommand{\D}{{\cal D}}

255: \renewcommand{\l}{\lim_{t\to\infty}}  

256: \newcommand{\ph}{\varphi}

257: \renewcommand{\c}{{\cos(\ph)}}

258: \newcommand{\s}{{\sin(\ph)}}

259: 

260: 

261: 

262: \newcommand{\lu}{\lam_1}

263: \newcommand{\ld}{\lam_2}

264: \newcommand{\lt}{\lam_3}

265: \renewcommand{\lq}{\lam_4}

266: \newcommand{\rocc}{\mbox{\Large $\rho$}_{CC}}

267: \newcommand{\rorc}{\mbox{\Large $\rho$}_{RC}}

268: \newcommand{\rorr}{\mbox{\Large $\rho$}_{RR}}

269: 

270: 

271: 

272: 

273: 

274: 

275: 

276: \newcommand{\rr}{{\bf (RR)\ }}

277: \renewcommand{\cc}{{\bf (CC)\ }}

278: \newcommand{\rc}{{\bf (RC)\ }}

279: 

280: \newcommand{\V}{\mbox{{\bf V}}}

281: \newcommand{\proof}{{\bf Proof. }}

282: 

283: 

284: %\newcommand{\ltimes}{\psfig{file=ltimes.eps,width=.2truecm,angle=0}} 

285: 

286: 

287: 

288: 

289: %\newcommand{\ddd}{{\cal A} \!\!\!\!{\cal A}}

290: %\newcommand{\Uu}{\mathcal{U}}

291: 

292: \newcommand{\aaa}{\mbox{{\bf A}}}

293: \newcommand{\Aa}{{\cal A} \!\!\!\!{\cal A}}

294: 

295: \newcommand{\Dd}{\mathcal{D}}

296: 

297: 

298: \newcommand{\Cc}{\mathcal{C}}

299: \newcommand{\lunga}{\longrightarrow}

300: \newcommand{\impl}{\Longrightarrow}

301: \newcommand{\ra}{\rho_A}

302: \newcommand{\rb}{\rho_B}

303: \newcommand{\la}{\lambda}

304: 

305: \newcommand{\ca}{\mathcal{K}}

306: \newcommand{\del}{\delta}

307: 

308: \newcommand{\vett}[2]{\Big(\begin{array}{c} #1 \\ #2 \end{array}\Big)}

309: 

310: \newcommand{\AL}{\mbox{\Large $\al$}}

311: 

312: \newcommand{\GUES}{{\bf GUES}}

313: \newcommand{\US}{{\bf (US)}}

314: 

315: \newcommand{\brs}{\begin{eqnarray*}}

316: \newcommand{\ers}{\end{eqnarray*}}

317: \newcommand{\br}{\begin{eqnarray}}

318: \newcommand{\er}{\end{eqnarray}}

319: 

320: \newcommand{\csi}{\Xi}

321: 

322: 

323: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%MACROSCHIFUSEDIMARIO

324: \newcommand{\rw}{\rightarrow}

325: \newcommand{\Ga}{\Gamma}

326: \newcommand{\ga}{\gamma}

327: \newcommand{\lp}{\left(}

328: \newcommand{\rp}{\right)}

329: \newcommand{\mt}{\mapsto}

330: \def\EOP{\ \hfill \rule{0.5em}{0.5em} }

331: \newcommand{\be}{\begin{equation}}

332: \newcommand{\ee}{\end{equation}}

333: \newcommand{\n}{\noindent}

334: \newcommand{\coco}{{\cal G}}

335: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

336: 

337: \newcommand{\C}{\mathbb{C}}

338: 

339: \newcommand{\B}{{\cal B}}

340: \newcommand{\bB}{{\overline{\B}}}

341: \newcommand{\Bbq}{\overline{\B(q)}}

342: \newcommand{\Dbq}{\overline{{\cal D}(q)}}

343: \newcommand{\bD}{{\overline{\cal D}}}

344: 

345: 

346: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5

347: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5

348: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5

349: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5

350: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5

351: 

352: \begin{document}

353: \begin{center} \noindent

354: {\LARGE{\sl{\bf Stability of Planar Nonlinear Switched Systems}}}

355: \end{center}

356: 

357: \vskip 1cm

358: \begin{center}

359: Ugo Boscain,

360: 

361: {\footnotesize SISSA-ISAS,

362: Via Beirut 2-4, 34014 Trieste, Italy. E-mail: {\tt boscain@sissa.it}}

363: 

364: Gr\'egoire Charlot

365: 

366: {\footnotesize Universit\'e Montpellier II, Math\'ematiques, CC051, 34095 Montpellier Cedex 5, 

367: France. E-mail: {\tt cha@math.univ-montp2.fr }}

368: 

369: 

370: Mario Sigalotti

371: 

372: {\footnotesize  INRIA, 2004 route des lucioles, 06902 Sophia Antipolis, France. E-mail: {\tt Mario.Sigalotti@sophia.inria.fr}}

373: 

374: \end{center}

375: 

376: \vspace{1cm}

377: 

378: 

379: 

380: 

381: \begin{quotation}\noindent  {\bf\em Abstract --- %--

382: %\begin{abstract}

383: We consider the time-dependent nonlinear system

384: $\dot q(t)=u(t)X(q(t))+(1-u(t))Y(q(t))$, where $q\in\R^2$, $X$ and

385: $Y$ are two %$C^\infty$

386: smooth vector fields, globally  asymptotically

387: stable at the origin and

388:   $u:[0,\infty)\to\{0,1\}$ is an arbitrary measurable function.

389: Analysing the topology of the set where $X$ and $Y$ are parallel,

390: we give some sufficient and some necessary conditions for global

391: asymptotic stability, uniform with respect to $u(.)$. Such conditions

392: can be verified without any integration or construction of a Lyapunov

393: function, and they are

394: robust under small perturbations of the vector fields.

395: }\end{quotation}

396: 

397: Keywords --- %-- 

398: Global asymptotic stability, planar switched systems, nonlinear. 

399: 

400: 

401: 

402: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

403: \section{Introduction}

404: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

405: A  {\it switched system} is a family of continuous-time dynamical

406: systems endowed with a 

407: rule that determines, at every time, which dynamical

408: system is responsible for the time evolution. More precisely let $\{f_u\,|\;u\in U\}$

409: be a

410: (possibly infinite) set of smooth vector fields  on

411: a manifold $M$, and consider, as $u$ varies in $U$, the family of dynamical systems

412: \bqn\label{-1}

413: \dot q=f_u(q)\,,~~~~q\in M\,.

414: \eqn

415: A non-autonomous dynamical system is obtained by assigning 

416: a so-called {\it switching

417: function} $u(.):[0,\infty)\to U$. 

418: 

419: In this paper, the switching function models the behavior of a parameter 

420: which cannot be predicted a priori. It

421: %is given from outside and 

422: represents some phenomena (e.g., a

423: disturbance) that it is not possible to control or include in the

424: dynamical system model.

425: %In what follows, we write $u$ to denote a fixed element of $U$, while  $u(.)$ stands for a switching function. 

426: 

427: A typical problem related to switched systems is to obtain, out of a 

428: property which is shared by all the autonomous dynamical systems governed

429: by the vector fields  $f_u$, some, maybe weaker, property for the time-dependent system

430: associated with an arbitrary switching function $u(.)$. For a discussion

431: on various issues related to switched systems we refer the reader to

432: \cite{liberzon-book,survey}.

433: 

434: 

435: In this paper, we consider a two-dimensional

436: nonlinear switched system of the type

437: \bqn

438: \dot q=u\,X(q)+(1-u)\,Y(q)\,,~~~~q\in\R^2\,,~~~~u\in\{0,1\}\,,

439: \label{system}\label{nonlinear}

440: \eqn

441: where the two vector fields 

442: $X$ and $Y$ are smooth (say, $\con^\infty$) on $\R^2$.

443: In order to define proper non-autonomous systems, we require

444: the switching functions to be measurable.

445: 

446: Assume that $X(0)=Y(0)=0$ and that the two dynamical

447: systems $\dot q=X(q)$ and $\dot q=Y(q)$ are \underline{globally 

448: asymptotically

449: stable} at the origin.

450: Our main aim is to study under which conditions on $X$

451: and $Y$ the origin is globally asymptotically stable

452: for the system \r{nonlinear}, uniformly with respect to the switching functions

453: (GUAS for short). For the precise formulation of this and other stability properties, 

454: see Definition \ref{stabilities}. 

455: 

456: In order to study the stability of \r{system} it is natural to consider 

457: its convexification, i.e., the case in which $u$ varies in the whole interval $[0,1]$. It  turns 

458: out that the stability properties of the two systems are equivalent (see Section 

459: \ref{s-convex}).\\\\

460: %%%%%%%%%%%%%%%%%%%%%%%%%

461: The linear version of the system introduced above, 

462: namely,  

463: \bqn\label{1}

464: \dot q=u\,A\,q+(1-u)\,B\,q\,,~~~~q\in\R^2\,,~~~~u\in\{0,1\}\,,

465: \eqn

466: where  the $2\times2$ real matrices

467: $A$ and $B$ have eigenvalues with 

468: strictly negative real part, 

469:  was studied in \cite{SIAM} (see also \cite{lyapunov-paolo}). More

470: precisely, the results in \cite{SIAM} establish

471: a necessary and sufficient condition for GUAS 

472: in terms of three relevant parameters, two depending 

473: on the eigenvalues of $A$ and $B$ respectively, and the third one 

474: (namely, the cross ratio of the four eigenvectors of $A$ and $B$ in

475: the projective line $\C P^1$) accounting for the interrelations

476: among the two systems.

477: The precise necessary and sufficient condition ensuring GUAS 

478: of 

479: \r{1} is quite

480: technical and can be found in \cite{SIAM} (see also \cite{lyapunov-paolo}).

481: Notice that, in the linear case, 

482: %asymptotic stability is equivalent to GUAS,

483: %which, in turns, 

484: GUAS

485: is equivalent to

486: the more often quoted GUES property, i.e., global exponential 

487: stability, uniform with respect to the switching rule (see, for example, 

488: \cite{angeli} and references therein). 

489: For related results on linear switched systems, see \cite{agr,blanchini,daw,hesp,lyapunov-paolo}. 

490: 

491: 

492: \bigskip

493: 

494: For nonlinear systems, the problem of characterizing GUAS completely, 

495: without assuming the explicit knowledge of the %families of

496: integral curves of $X$ and $Y$, 

497: is hopeless. 

498: 

499: The problem, however, admits some partial solution. 

500: The purpose of this paper is to provide some sufficient and some necessary 

501: conditions for stability which are robust (with respect to 

502: small perturbations of the vector fields) and easily verifiable, directly on 

503: the vector fields $X$ and $Y$, without requiring any integration or 

504: construction of a Lyapunov function. 

505: 

506: Denote by $\qq$ the set on which $X$ and $Y$ are parallel.

507: One of our main results is that, if $\qq$ reduces to the singleton 

508: $\{0\}$, then \r{nonlinear} is GUAS (Theorem \ref{t-mai-paralleli}). 

509: The proofs works by showing that an admissible trajectory starting from a point $p\in\R^2$ is forced 

510: to stay in a compact region bounded by the integral curves of $X$ and $Y$ from $p$. The fact that $X$ and $Y$ are linearly independent  outside the origin plays as a sort of drift

511: which guarantees that the only possible accumulation point of an admissible trajectory is the origin. 

512: 

513: 

514: When $\qq$ is just compact, we prove that \r{system} is at least

515: bounded (see Theorem \ref{t-solo-compatte}). Roughly speaking, this means that its trajectories do not 

516: escape to infinity. The idea of the proof is that, if we modify $X$ and $Y$ only in a compact

517: region of the plane, then the boundedness properties of the system are left unchanged. 

518: Taking advantage of the result obtained in Theorem~\ref{t-mai-paralleli}, we manage to 

519: prove the boundedness of \r{system} by reducing, using compact perturbations, $\qq$ to $\{0\}$, 

520: while preserving the global asymptotic stability of $X$ and $Y$.

521: 

522: Other conditions can be formulated taking into account  

523: the relative position of $X$ and $Y$ along $\qq$. Assume that 

524: $\qq\setminus\{0\}$ contains at least one point $q_0$. Since  

525: both $X(q_0)$ and $Y(q_0)$ are different from zero, 

526: the property of pointing in the same or in the opposite versus can be

527: stated unambiguously.

528: If  $X(q_0)$ and $Y(q_0)$

529: have opposite versus, then there exists a switching function, for the 

530: convexified system, whose output

531: is the constant trajectory which stays in $q_0$. As a consequence, the 

532: system \r{system} is not GUAS. 

533: 

534: 

535: Additional results can be obtained under the assumption that the

536: pair of vector fields $(X,Y)$ is generic. 

537: (For the notion of genericity appropriate to our aims, see Section \ref{s-basic-facts}.) 

538: In particular, the genericity assumption can be used to guarantee 

539: that 

540: %, out from an arbitrarily small ball containing the origin, 

541: $\qq\setminus\{0\}$ is  an embedded one-dimensional 

542: submanifold of the plane.

543: Clearly, $\qq$ needs not to be connected. 

544: If the connected component of $\qq$ containing the origin reduces to $\{0\}$ 

545: and on all other components $X$ and $Y$ point in the same versus, transversally 

546: to $\qq$, then 

547: \r{system} is GUAS. This result is formulated in Theorem \ref{t-mai-tangent}, 

548: which follows the pattern of proof of Theorem~\ref{t-mai-paralleli}.

549: 

550: Conversely, Theorem \ref{t-inv-non-comp} states that, if one 

551: connected component of $\qq\setminus\{0\}$ is unbounded and such that 

552: $X$ and $Y$ have opposite versus on it, 

553: then  \r{system}  admits a trajectory going to infinity.  

554: Intuitively, this happens because %, near to a point of $\qq\setminus\{0\}$,

555: the orientation of $(X(p),Y(p))$ changes while $p$ crosses $\qq\setminus\{0\}$. If $X(p)$

556: is not tangent to $\qq$ at $p$ and $X(p)$ points in the opposite direction with respect to $Y(p)$, 

557: then one can embed $\qq$, locally near $p$, in a foliation %of a neighborhood of $p$ 

558: made of admissible trajectories of \r{system}, whose 

559: running direction 

560: is reversed while crossing $\qq$ (see Figure~\ref{foli}).

561: \bfi[h]

562: \begin{center}

563: \input{foli.pstex_t}

564: \caption{A local foliation embedding $\qq$}\label{foli}

565: \end{center}

566: \efi

567: Since, generically, the points where $X$ is tangent to $\qq$ are isolated, it turns out that 

568: there exists

569: an admissible trajectory which tracks globally the unbounded connected component of $\qq\setminus\{0\}$

570: on which $X$ and $Y$ have opposite versus.

571: 

572: 

573: \bigskip

574: The paper is organized as follows.

575: In Section \ref{s-basic-facts}, we recall the main definitions of stability in which we are 

576: interested, we introduce the convexified system, and we describe

577: the topological structure of the  set $\qq$. 

578: The main results are stated in Section \ref{s-main-results}, where their robustness is also discussed.

579: The proofs are given in Sections~\ref{mp}, \ref{mt}, \ref{sc}, and \ref{inc}.

580: 

581: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

582: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

583: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

584: \section{Basic definitions and facts}\label{s-basic-facts}

585: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

586: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

587: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

588: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

589: \subsection{Definitions of stability}

590: Fix $n,m\in \N$ and consider the switched system

591: \bqn\label{ss-system}

592: \dot q=f_u(q)\,,~~~~q\in \R^n\,,~~~~u\in U\subset\R^m\,,

593: \eqn

594: where $U$ is a measurable subset of  $\R^m$ and $(q,u)\mapsto f_u(q)$ is 

595: the restriction on  $\R^n\times U$ of a $\con^\infty$ function from $\R^n\times\R^m$ to 

596: $\R^n$. Assume that $f_u(0)=0$ for every $u\in U$. \\

597: \noindent

598: For every $\de>0$, denote by $B_\de\subset \R^n$ the ball of radius $\de$, centered at the 

599: origin.

600: %

601: Set

602: $$\U=\{u:[0,\infty)\rw U\,|\;u(.)\ \mbox{measurable}\}\,.$$

603: 

604: For every $u(.)$ in $\U$ and every $p\in \R^n$, 

605: denote by $t\mt \ga(p,u(.),t)$ the solution of 

606: \r{ss-system} such that $\ga(p,u(.),0)=p$. 

607: Notice that, in general, $t\mt \ga(p,u(.),t)$ needs not to be 

608: defined for every $t\geq 0$, since the non-autonomous vector field 

609: $f_{u(t)}$ may not be complete. 

610: Denote by $\T(p,u(.))$ the maximal element of $(0,+\infty]$ such that 

611: $t\mt \ga(p,u(.),t)$ is defined on 

612: $[0,\T(p,u(.)))$, and let 

613: $$\Supp(\ga(p,u(.),.))=\ga(p,u(.),[0,\T(p,u(.))))\,.$$

614: If $\Supp(\ga(p,u(.),.))$ is bounded, then $\T(p,u(.))=

615: +\infty$.

616: 

617: Given 

618: $p\in \R^n$, the {\it accessible set from $p$}, 

619: denoted by ${\A(p)}$, is 

620: %the union of all supports of admissible trajectories starting form $p$, i.e.,

621: defined as

622: $$\A(p)=\cup_{u(.)\in\U}\Supp(\ga(p,u(.),.))\,.$$ 

623: 

624: \n Several notions of stability for the switched system \r{ss-system} can be 

625: introduced.

626: \begin{Definition}\label{stabilities}

627: We say that \r{ss-system} is

628: \bi

629: \i {\bf \bad} 

630:    if there exist $p\in\R^n$ and $u(.)\in\U$ such that

631:  $\ga(p,u(.),t)$ goes to infinity as $t$ tends to $\T(p,u(.))$;

632: 

633: \i {\bf bounded} 

634:    if, for every $K_1\subset \R^n$ compact, there exists $K_2\subset \R^n$ 

635: compact such that $\ga(p,u(.),t)\in K_2$ for every $u(.)\in\U$, $t\geq 0$ and $p\in K_1$;

636: 

637: %\i {\bf stable}  at the origin

638: 

639: 

640: \i {\bf uniformly stable}  at the origin 

641: if, for every %small enough 

642: $\de>0$, there exists $\eps>0$ such that $\A(p)\subset B_\de$ for every 

643: $p\in B_\eps$;

644: 

645: 

646: \i {\bf locally attractive}  at the origin if there exists $\de>0$

647: such that, for every $u(.)\in\U$ and every $p\in B_\de$,  $\ga(p,u(.),t)$ %\rw0$ 

648: converges to the origin 

649: as $t$ goes to infinity; 

650: 

651: 

652: \i {\bf globally attractive}  at the origin if, 

653: for every $u(.)\in\U$ and every $p\in\R^n$, $\ga(p,u(.),t)$ %\rw0$ 

654: converges to the origin 

655: as $t$ goes to infinity; 

656: 

657: \i {\bf globally uniformly attractive}  at the origin if, 

658: for every $\de_1, \de_2>0$, there exists $T>0$ such that  $\ga(p,u(.),T)\in B_{\de_1}$ for every 

659: $u(.)\in\U$ and every $p\in B_{\de_2}$;

660: 

661: 

662: 

663: \i {\bf globally uniformly stable (GUS)}  at the origin 

664:    if it is bounded and uniformly stable at the origin;

665: 

666: \i {\bf locally asymptotically stable (LAS)}  at the origin if it is uniformly stable and locally attractive at the origin;

667: 

668: \i {\bf globally asymptotically stable (GAS)}  at the origin if it is 

669: uniformly stable and globally attractive at the origin;

670: 

671: \i {\bf globally uniformly asymptotically stable (GUAS)}  at the origin if it is uniformly stable and globally uniformly  attractive at the origin.

672: \ei

673: \end{Definition}

674: %

675: 

676: It has been showed by Angeli, Ingalls, Sontag, and Wang \cite{AISW} that, 

677: when $U$ is compact, the notions of  GAS and GUAS are equivalent. This is 

678: the case for system \r{system}. Moreover, it is well known that, in 

679: the case in which all the vector fields $f_u$ are linear,  local and 

680: global properties are equivalent.

681: 

682: \subsection{The convexified system}

683: \label{s-convex}

684: In this paper, we focus on the planar switched system 

685: \be\llabel{nons-system}

686: \dot q=u\,X(q)+(1-u)\,Y(q)\,,~~~~q\in\R^2\,,~~~~u\in\{0,1\}\,, 

687: \ee

688: where $X$ and $Y$ denote two 

689: vector fields on $\R^2$, of class $\con^\infty$, such

690: that $X(0)=Y(0)=0$. 

691: We assume moreover that $X$ and $Y$ are \underline{globally asymptotically 

692: stable} at the origin. 

693: Notice, in particular, that $X$ and $Y$ are forward complete.

694: 

695: \bigskip

696: 

697: A classical tool in stability analysis is the convexification of the set of admissible velocities. 

698: Such transformation does not change the closure of the accessible sets. Moreover,  it was proved in 

699: \cite{12} (see also \cite[Proposition 7.2]{AISW}) that,

700: for every $p'\in\R^2$, every switching function $u':[0,\infty)\to[0,1]$,

701: and every positive continuous function $r$ defined on $[0,\T(p',u'(.)))$, there exist

702: $u(.)\in \U$

703: and $p\in\R^2$ such that

704: \[

705: \|\ga(p,u(.),t)-\ga(p',u'(.),t)\|\leq r(t)

706: \]

707: for every $t\in[0,\T(p',u'(.)))$.

708: As a consequence  each of the notions

709: introduced in Definition~\ref{stabilities}

710: holds for (\ref{nons-system}) if and only if it holds for the same system

711: where

712: $U=\{0,1\}$ is

713: replaced by $[0,1]$.

714: 

715: In the following, to simplify proofs, we deal with the convexified system

716: \be\llabel{s-system}

717: \dot q=u\,X(q)+(1-u)\,Y(q)\,,~~~~q\in\R^2\,,~~~~u\in[0,1]\,. 

718: \ee

719: 

720: \smallskip

721: \noindent

722: {\bf Notations.} 

723: When $u(.)$ is constantly equal 

724: to zero (respectively, one), we write 

725: $\ga_Y(p,t)$ (respectively, $\ga_X(p,t)$) for $\ga(p,u(.),t)$. 

726: Given $p,p'\in\R^2$ and $u(.),u'(.)$ in $\U$, we say that $\ga(p,u(.),.)$ 

727: and $\ga(p',u'(.),.)$ {\it \fintersect} if 

728: %the two supports 

729: $\Supp(\ga(p,u(.),.))$ and 

730: $\Supp(\ga(p',u'(.),.))$ have nonempty intersection.

731: 

732: 

733: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

734: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

735: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

736: \subsection{The collinearity set of $X$ and $Y$}

737: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

738: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

739: \n A key object in order to detect stability properties 

740: of \r{s-system} turns out to be the set $\qq$ on which $X$ and $Y$ are 

741: parallel. We have that $\qq=Q^{-1}(0)$, where  

742: \bqn\label{decca}

743: Q(p)=\mathit{det}(X(p),Y(p))\,,\ \ \ \ p\in\R^2\,.

744: \eqn

745: 

746: 

747: 

748: In \cite{SIAM}, the stability of the linear switched system \r{1} was studied by  

749: associating with every point of $\R^2$ 

750: a suitably defined ``worst'' trajectory

751: passing through it, whose construction was based upon $\qq$.

752: The global asymptotic stability of the linear switched system \r{1} 

753: was then proved to be equivalent to the convergence to the origin 

754: of every such worst trajectory.

755: We recall that in the linear case, % \r{1}, 

756: excepted for some degenerate

757: situations, $\qq$ is either equal to $\{0\}$  or 

758: % it

759: is made of  

760: two straight lines passing through

761: the origin. 

762: 

763: In the nonlinear case, the

764: situation is more complex.

765: Let us represent %the set 

766: $\qq$ as

767: % the union of

768: %a subset connected to the origin and all the other connected components.

769: \bqn\label{dec}

770: \qq=\{0\}\cup\bigcup_{\Gamma\in \coco}\Gamma\,, %~~j\in J\subset\R \mbox{ and } k=0,1,

771: \eqn%l{dec}

772: where $\coco$ is the set of all connected components of $\qq\setminus\{0\}$. Notice that $\coco$ needs not, in general, to be %finite or even 

773: countable.

774: %$\Gamma^0$ being the subset of $\qq$ connected to the origin, while the $\Gamma_j^1$ are the other connected components of $\qq$. 

775: With a slight abuse of notation, we will refer to the elements of 

776: $\coco$ as to the {\it components} of $\qq$.

777: \bdeff

778: Let $\Gamma$ be a component of $\qq$ and fix $p\in\Gamma$. We say that 

779: $\Gamma$ is {\it direct} (respectively, {\it inverse}) if $X(p)$ and $Y(p)$ 

780: have the same (respectively, opposite) direction. 

781: \edeff

782: \begin{rmk}

783: The definition is independent  of the choice of $p$, since neither $X$ nor $Y$ 

784: vanish along $\Gamma$.

785: \end{rmk}

786: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

787: An example of how $\qq$ can look like is represented in Figure~\ref{qq}.

788: 

789: \bfi[h]

790: \begin{center}

791: \input{qq.pstex_t}

792: \caption{The set  $\qq$}\label{qq}

793: \end{center}

794: \efi

795: 

796: \bigskip

797: 

798: Some of the results of this paper are obtained assuming that the set $\qq$ has suitable 

799: regularity properties, which are generic in the sense defined below. 

800: 

801: A base for the {\it Withney topology} on $\con^\infty(\R^2,\R^2)$ 

802: (the set of %$C^\infty$ 

803: smooth vector fields on $\R^2$) can be defined, using the multi-index notation, as the 

804: family of sets of the type

805: \[

806: {\cal V}(k,f,r)=\left\{g\in 

807: \con^\infty(\R^2,\R^2)\,\left|\;\left\|\frac{\partial^{|I|}(f-g)}{\partial 

808: x^I}(x)\right\|<r(x), \forall x\in\R^2,|I|\leq k\right.\right\}\,,\]

809: where $k$ is a nonnegative integer, $f$ belongs to  $\con^\infty(\R^2,\R^2)$, and $r$ is 

810: a positive continuous function defined on $\R^2$.

811: Denote by GAS$(\R^2)$ the set of smooth vector fields on $\R^2$ which are globally 

812: asymptotically stable at the origin, and endow it with the topology induced by Withney's one. 

813: A {\it generic property} for (\ref{s-system}) is a property which holds for 

814: an open dense subset of GAS$(\R^2)\times$GAS$(\R^2)$, endowed with the product topology of GAS$(\R^2)$.

815: 

816: 

817: \bl \label{genio}

818: For a generic pair of vector fields $(X,Y)$, 

819: $\qq\setminus\{0\}$ is an embedded one-dimensional submanifold of $\R^2$. 

820: Moreover, $Q(p)$ changes sign while $p$ crosses $\qq\setminus\{0\}$.

821: \el

822: %

823: The lemma is a standard result in genericity theory. It follows from the fact that the condition

824: \bd

825: \i[%~~~

826: (G1)] If $p\neq0$ and $Q(p)=0$, then $\nabla Q(p)\neq0$,

827: \ed

828: is generic (see, for instance, \cite{generiko}). When $\qq\setminus\{0\}$ is a manifold, we say 

829: that $p\in \qq\setminus\{0\}$ is a \underline{tangency point} if $X(p)$ is tangent to $\qq$. Under 

830: condition {\bf (G1)}, $p\in \qq\setminus\{0\}$ is a tangency point if and only if  $\nabla Q(p)$ and $X(p)$ (equivalently, $Y(p)$) are 

831: orthogonal.

832: 

833: Some of our results are obtained under additional generic conditions. 

834: One of these, namely,

835: \bd

836: \i[%~~~

837: (G2)] The Hessian matrix of $Q$ at the origin is non-degenerate,

838: \ed

839: ensures that $\qq$, in a neighborhood of the origin, is given either by $\{0\}$ or by the union 

840: of two transversal one-dimensional manifolds intersecting at the origin.

841: 

842: Under the generic conditions {\bf (G1)} and  {\bf (G2)}, the connected component of 

843: $\qq$ containing the origin looks like one of Figure~\ref{f-origin}.

844: 

845: A third generic condition which we will sometimes assume to hold is 

846: \bd

847: \i[%~~~

848: (G3)] If $p\neq0$, $Q(p)=0$, and $\nabla Q(p)$ is orthogonal to $X(p)$, 

849: then the second derivative of $Q$ at $p$ along $X$ (equivalently, $Y$) is different from zero,

850: \ed

851: which, together with {\bf (G1)}, guarantees that the tangency points on $\qq$ are isolated.

852: 

853: 

854: \bfi[h]

855: \begin{center}

856: \input{f-origin.pstex_t}

857: \caption{The connected component of $\qq$ containing the origin}\label{f-origin}

858: \end{center}

859: \efi

860: 

861: 

862: 

863: 

864: 

865: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

866: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

867: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

868: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

869: \section{Statement of the results}\label{s-main-results}

870: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

871: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

872: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

873: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

874: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

875: 

876: We organize our results in sufficient 

877: and necessary conditions with respect to the stability properties. 

878: 

879: Notice that all such conditions are easily verified 

880: without any integration or construction of a Lyapunov

881: function. Moreover, they are  robust under small perturbations of the vector fields, as explained 

882: in Section \ref{s-robustness}. Let us recall that 

883: $X$ and $Y$ are assumed to be globally asymptotically stable at the origin and that 

884: all the results given 

885: %

886: %

887: below,

888: although stated for the case $u\in[0,1]$, are also valid 

889: for the system where $u$ varies in $\{0,1\}$. 

890: 

891: 

892: 

893: %in this section 

894: %do not depend

895: %on the choice $u\in\{0,1\}$ or $u\in[0,1]$.

896: 

897: Before stating our main theorems, observe that classical results on 

898: linearization imply the following.

899: \bp

900: \label{nec-cond}  

901: Assume that the eigenvalues of $A=\nabla X|_{p=0}$ and $B=\nabla Y|_{p=0}$ have 

902: strictly negative real part.

903: Then \r{s-system} is  LAS if and only if \r{1} is GUAS. 

904: \ep

905: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

906: \subsection{Sufficient conditions}

907: The following theorem gives a simple sufficient condition for GUAS, 

908: which generalizes the analogous one already known 

909: for the linear system \r{1} (see \cite{SIAM,lyapunov-paolo}).

910: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

911: \bt %{\bf (Sufficient condition)}

912: \label{t-mai-paralleli}

913: Assume that 

914: $\qq=\{0\}$. Then the switched system \r{s-system} is GUAS 

915: at the origin.

916: \et

917: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

918: Under the generic assumptions 

919: {\bf (G1)} and  {\bf (G2)}, Theorem \ref{t-mai-paralleli} can be generalized as follows. 

920: \bt\label{t-mai-tangent}

921: Assume that the  generic conditions {\bf (G1)} and {\bf (G2)} hold. 

922: Assume, moreover, that  the origin is isolated in $\qq$ and that there is no  tangency point in 

923: $\qq\setminus\{0\}$. Then the 

924: switched system \r{s-system} is GUAS.

925: \et

926: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

927: When $\qq$ is bounded, although different from $\{0\}$, 

928: some weaker version of Theorem \ref{t-mai-paralleli} still holds.

929: \bt

930: \label{t-solo-compatte}

931: Assume that $\qq$ is compact. Then the switched system \r{s-system} is bounded.

932: \et

933: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

934: As a direct consequence of Proposition 

935: \ref{nec-cond} and Theorem \ref{t-solo-compatte}, we have the following 

936: sufficient condition for GUS.

937: \begin{corollary}\llabel{c1}

938: Let $\qq$ be compact, and the linearized 

939: switched system be non-degenerate and GUAS. %asymptotically stable. 

940: Then 

941: the switched system \r{s-system} is GUS. 

942: \end{corollary}

943: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

944: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

945: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

946: \subsection{Necessary conditions}

947: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

948: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

949: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

950: The following proposition expresses  the straightforward remark that

951: the inverse  components of $\qq$  constitute obstructions to the 

952: stability of \r{s-system}. 

953: The reason is clear: if $\Gamma$ is inverse and $p$ belongs to $\Gamma$, 

954: then a constant switching function $u(.)$ exists 

955: such  that $\ga(p,u(.),t)=p$ for every $t\geq 0$.

956: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

957: \bp 

958: \label{p-mai-inverse}

959: If $\qq$ has an inverse component, then the switched system \r{s-system} is not globally attractive.

960: \ep

961: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

962: The following theorem gives a necessary condition for boundedness, under generic conditions.

963: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

964: \bt

965: \label{t-inv-non-comp} 

966: Assume that the  generic conditions {\bf (G1)} and {\bf (G3)} hold. 

967: If 

968: $\qq$ contains an unbounded inverse component, then the switched

969: system \r{s-system} is \bad.

970: \et

971: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

972: \subsection{Robustness}\label{s-robustness}

973: We say that a property satisfied by  $(X,Y)$ is \underline{robust} if it still holds for small 

974: perturbations of the pair $(X,Y)$, that is, if it holds for all the elements of 

975:  a neighborhood of $(X,Y)$ in GAS$(\R^2)\times$GAS$(\R^2)$.

976: Such notion of robustness is also known as {\it structural stability}, an expression which we prefer to avoid, in order to prevent confusion with the many 

977: definitions of stability already introduced for \r{s-system}.

978: 

979: 

980: Under the generic conditions {\bf (G1)} and {\bf (G2)}, 

981: one can easily verify that the 

982: topology of the set $\qq$ does not change 

983: for small perturbations 

984: of $X$ and $Y$. Moreover, fixed one 

985: component 

986: $\Gamma$ of $\qq$,  the fact that $\Gamma$ is direct or 

987: inverse is robust. Similarly, if $\Gamma$ is a component of $\qq$, which has not the origin in its 

988: closure,  the absence of tangency points along $\Gamma$ is robust.

989: As a consequence, the conditions formulated by the theorems above are robust. More precisely: 

990: %, in the following sense.

991: %\bt

992: %Under the generic assumptions {\bf (G1)} and {\bf (G2)}, if 

993: %Proposition~\ref{p-mai-inverse} or any of the  

994: %Theorems \ref{t-mai-paralleli}, \ref{t-mai-tangent}, \ref{t-solo-compatte} 

995: %applies to the pair $(X,Y)$, then it applies in a neighborhood of $(X,Y)$. 

996: %%

997: %The same holds for Theorem~\ref{t-inv-non-comp} (respectively, Corollary \ref{c1}) under the additional generic hypothesis {\bf (G3)} (respectively, that the linearized system is non-degenerate).

998: %\et

999: 

1000: \bt

1001: Under generic assumptions, if any of Theorems \ref{t-mai-paralleli}, \ref{t-mai-tangent}, \ref{t-solo-compatte}, \ref{t-inv-non-comp},  Corollary \ref{c1}, or

1002: Proposition \ref{p-mai-inverse}   

1003: applies to the pair $(X,Y)$, then it applies in a neighborhood of $(X,Y)$ in GAS$(\R^2)\times$GAS$(\R^2)$. 

1004: \et

1005: 

1006: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1007: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1008: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1009: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1010: \section{Proof of Theorem \ref{t-mai-paralleli}}\label{mp}

1011: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1012: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1013: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1014: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1015: 

1016: 

1017: 

1018: 

1019: Assume that $\qq=\{0\}$. We already recalled in Section \ref{s-basic-facts} that GAS and GUAS are two equivalent notions. 

1020: The main step of the proof consists in showing that (\ref{s-system}) is globally attractive.

1021: The uniform stability will be obtained as a byproduct of

1022: the adopted demonstration technique.

1023: 

1024: 

1025: 

1026: Fix $q\in \R^2\setminus\{0\}$. 

1027: We first prove that ${\A(q)}$

1028: is bounded. 

1029: Then we  show that, for every $u(.)$ in $\U$,  

1030:  the only possible accumulation point of $\ga(q,u(.),t)$ is

1031: the origin. 

1032: These two facts imply that $\ga(q,u(.),t)$ %\rw0$ 

1033: converges to the origin

1034: as $t$ goes to infinity.

1035: 

1036: 

1037: 

1038: \subsection{Boundedness of ${\A(q)}$}\label{barbecue}

1039: 

1040: \noindent We distinguish two cases.

1041: 

1042: \noindent

1043: %\underline

1044: {\bf \underline{First case:} $\gamma_X(q,.)$ and $\gamma_Y(q,.)$ do not \fintersect}. Then, we

1045: can define a closed, simple, piecewise smooth curve, by

1046: $$

1047: \gamma_{X,Y}(q,t)=\left\{\begin{array}{lr}\!\!\gamma_X(q,\tan(t\pi)) & \mbox{if }\ t\in\left[0,\frac{1}{2}\right],\\[1mm]

1048: \!\!\gamma_Y(q,\tan((1-t)\pi)) & \mbox{if }\ 

1049: t\in\left[\frac{1}{2},1\right],

1050: \end{array}\right.

1051: $$

1052: where $\gamma_X(q,\tan(\pi/2))$ and $\gamma_Y(q,\tan(\pi/2))$ are 

1053: identified with the origin. 

1054: The support of $\gamma_{X,Y}(q,.)$ separates $\R^2$ in two sets, one being

1055: bounded. Let us call $\B(q)$ the interior of the bounded set and 

1056: $\D(q)$ the interior of the unbounded one. 

1057: %We have that $\R^2=\B(q)\cup\D(q)\cup\gamma_{X,Y}(q,[0,1])$.

1058: 

1059: 

1060: \bl

1061: \label{goo}

1062: $\A(q)$ is contained in

1063: $\Bbq=\B(q)\cup\gamma_{X,Y}(q,[0,1])$.

1064: \el

1065: 

1066: \noindent\proof 

1067: Consider the vector field $\dbt$. At the point $q$, it points either inside or outside $\B(q)$. Then, as it becomes clear through a local rectification of $\dbt$, 

1068: the same 

1069: holds true at all points of  $\gamma_{X,Y}(q,[0,1])$ sufficiently close to $q$.

1070: Moreover, 

1071: since the orientation defined by $(X,Y)$ does not vary on 

1072: $\R^2\setminus\{0\}$ and coincides with the

1073: ones induced by $(X,(X+Y)/2)$ and $((X+Y)/2,Y)$, 

1074: then $\dbt$ is pointing  constantly  either inside or outside $\B(q)$, all along 

1075: $\gamma_{X,Y}(q,[0,1])\setminus\{0\}$.

1076: 

1077: 

1078: 

1079: Let us assume that $\dbt$ points inside $\B(q)$. Then $\Bbq$ is 

1080: invariant for the flow of all the vector fields of the type $u\,X+(1-u)\,Y$, with $u\in[0,1]$, that is, it is

1081: invariant for the dynamics of \r{s-system}.

1082: Hence, $\A(q)$ is contained in $\Bbq$.

1083: 

1084: Assume now, by contradiction, that $\dbt$ points outside $\B(q)$.

1085: The same reasoning as above shows that $\A(q)$ is contained in $\Dbq$. 

1086: Define, for every $t\geq 0$ and every $\tau\in \R$,

1087: $$

1088: \gamma^{X,Y,t}(q,\tau)=\left\{\begin{array}{lr}

1089: \!\!\gamma_X(q,-\tau) & \mbox{if }\ \tau<-t\,, \\

1090: \!\!\gamma_Y(\gamma_X(q,t),\tau+t) & \mbox{if }\ \tau>-t\,.

1091: \end{array}\right.

1092: $$

1093: The support of $\gamma^{X,Y,t}$ is  given by the union of the integral curves of $X$ and $Y$ 

1094: connecting $\gamma_X(q,t)$ and the origin 

1095: (see Figure~\ref{gxyt}). For every $t\geq0$, 

1096: we can identify $\gamma^{X,Y,t}$ with a closed curve passing through the origin. 

1097: 

1098: \bfi[h]

1099: \begin{center}

1100: \input{inde.pstex_t}

1101: \caption{The curves $\gamma^{X,Y,t}$}\label{gxyt}

1102: \end{center}

1103: \efi

1104: 

1105: Fix a point $q'$ in $\B(q)$. By hypothesis, no $\gamma^{X,Y,t}$ passes through $q'$.

1106: Notice that the index of

1107: $\gamma^{X,Y,0}$ with respect to $q'$ is equal to one, since the 

1108: support of $\gamma^{X,Y,0}$ coincides with the boundary of $\B(q)$. 

1109: The stability of $Y$ at the origin implies that the index with respect to $q'$ of the curve

1110: $\gamma^{X,Y,t}$ depends continuously of $t$, that is, it is constant on $[0,\infty)$. 

1111: Hence, for every $t\in [0,\infty)$,

1112: \[\max_{\tau\in\R}\|\gamma^{X,Y,t}(q,\tau)\|> \|q'\|>0\,.\]

1113: On the other hand, when $t$ goes to infinity, $\gamma_X(q,t)$ converges to the origin and

1114: \bqn

1115: \sup_{\tau<-t}\|\gamma^{X,Y,t}(q,\tau)\|=\sup_{\tau>t}\|\gamma_X(q,\tau)\|\stackrel{t\rw\infty}{-\!\!\!-\!\!\!-\!\!\!\!\!\longrightarrow} 

1116: 0\,.\eqnn

1117: Therefore, there exist $p\in\R^2$, arbitrarily close to the origin, 

1118: such that 

1119: the curve $s\mt \gamma_Y(p,s)$, $s>0$, exits the ball $B_{\|q'\|}$, which contradicts the %asymptotic 

1120: stability of $Y$ at the origin. \EOP

1121: 

1122: \medskip

1123: 

1124: \noindent{\bf \underline{Second case:} $\gamma_X(q,.)$ and $\gamma_Y(q,.)$ do \fintersect}. Let $t$ be the first positive time

1125: such that the point $\gamma_Y(q,t)$ is equal to $\gamma_X(q,\tau)$ for some $\tau>0$. Define, for every $s\in [0,\tau+t]$,

1126: $$

1127: \gamma_{X,Y}(q,s)=\left\{\begin{array}{lcr}\!\!\gamma_X(q,s) & \mbox{if }&

1128: s\in[0,\tau]\,,\\[1mm]

1129: \!\!\gamma_Y(q,t+\tau-s) & \mbox{if }&

1130: s\in[\tau,\tau+t]\,.

1131: \end{array}\right.

1132: $$

1133: The curve $\gamma_{X,Y}(q,.)$ is simple and closed, and separates $\R^2$ in two open sets $\B(q)$ and $\D(q)$, $\B(q)$ being

1134: bounded. 

1135: 

1136: \bl

1137: \label{hoo}

1138: $\A(q)$ is contained in

1139: $\Bbq=\B(q)\cup\gamma_{X,Y}(q,[0,\tau+t])$.

1140: \el

1141: \proof

1142: Assume that

1143: $\dbt$ points inside $\B(q)$ at $q$. Hence, the same is

1144: true for all points of %the support of

1145: $\gamma_{X,Y}(q,[0,\tau+t])$ sufficiently close to $q$. 

1146: Since $\qq=\{0\}$, the property extends to 

1147: the entire curve $\gamma_{X,Y}(q,[0,\tau+t])$, except 

1148: possibly at the point $\gamma_{X,Y}(q,\tau)$. 

1149: %And it will propagate all along the curve

1150: %$\gamma_{X,Y}(q,.)$, except maybe at $\gamma_{X,Y}(q,\tau)$. 

1151: %

1152: The same reasoning %as above, 

1153: can be applied 

1154: %to $(-X-Y)/2$ 

1155: at $\gamma_{X,Y}(q,\tau)$, showing that $\dbt$ points either inside or outside $\B(q)$ at all points

1156: of the type $\gamma_{X,Y}(q,s)$, with $s$ close to $\tau$. The only non-contradictory 

1157: possibility is that

1158:  $\dbt$ points inside $\B(q)$ all along $\gamma_{X,Y}(q,[0,\tau+t])$.

1159: Hence, $\Bbq$ is invariant under the flow of each vector field $u\,X+(1-u)\,Y$, $u\in[0,1]$, so that %Therefore, 

1160: $\A(q)$ is contained in $\Bbq$. 

1161: 

1162: 

1163: 

1164: Assume, by contradiction, that $\dbt$ points 

1165: %strictly 

1166: inside $\D(q)$.

1167: The same reasoning as above shows that $\A(q)$ is contained in $\Dbq$.

1168: In particular, the origin belongs to $\Dbq$. 

1169: On the other hand,  the fact that $\dbt$ %strictly 

1170: points outside $\B(q)$ all along

1171: $\partial \B(q)$ implies that it has a zero inside $\B(q)$. Which is impossible, unless $\B(q)$ contains the origin. \EOP

1172: 

1173: \medskip

1174: 

1175: We proved that, in both cases, the set $\A(q)$ is bounded.

1176: The precise description of $\A(q)$ is given by the following lemma, where the definition of $\B(q)$

1177: depends on whether $\ga_X(q,.)$  and $\ga_Y(q,.)$  \fintersect\ or not.

1178: 

1179: \bl

1180: \label{gooh}

1181:  $\A(q)=\Bbq\setminus\{0\}$.

1182: \el

1183: 

1184: 

1185: \noindent\proof

1186: %

1187: First notice that the origin does not belong to $\A(q)$, being a steady point for both $X$ 

1188: and $Y$. The inclusion of $\A(q)$ in $\Bbq\setminus\{0\}$ is thus a consequence of Lemma~\ref{goo} 

1189: and Lemma~\ref{hoo}. 

1190: 

1191: As for the opposite inclusion, notice that $\partial\B(q)\setminus\{0\}$ is, by construction, made of integral curves of $X$ and $Y$ starting from $q$. Therefore, $\partial\B(q)\setminus\{0\}\subset \A(q)$.

1192: 

1193: Fix now $p\in\B(q)\setminus\{0\}$. We are left to prove that $p\in \A(q)$. 

1194: %

1195: Define

1196: $$C=\{\gamma_{X}(p,\tau)\,|\;\tau\leq 0\}\,,$$

1197: and let $V$ be a neighborhood  of the origin such that $p\not\in V$. 

1198: 

1199: Due to the  stability of $X$ and the boundedness of $\B(q)$, there exists $T>0$ such that $\ga_X(\Bbq,T)\subset V$. Since $\ga_X(\gamma_{X}(p,-T),T)=p\not\in V$, then $C$ is not contained in $\Bbq$. Therefore, there exists $\tau<0$ such that $\gamma_{X}(p,\tau)\in\partial \B(q)$. Notice that $\gamma_{X}(p,\tau)$ is different from the origin, since otherwise we would have $p=0$. 

1200: %Therefore 

1201: Finally, $\gamma_{X}(p,\tau)\in \A(q)$, which implies that $p=\ga_X(\gamma_{X}(p,\tau),|\tau|)$ belongs to $\A(q)$.  

1202: \EOP

1203: 

1204: 

1205: 

1206: \subsection{Global attractivity}

1207: 

1208: In the previous section, we showed that the accessible set from every  point is bounded. Hence, 

1209: the global attractivity of (\ref{s-system}) is proved if we ensure that

1210: no admissible curve %of (\ref{s-system}) 

1211: has an accumulation point different from the origin.

1212: 

1213: Let us show that, for every point $p\neq0$, there exist 

1214: $\eps>0$ and a neighborhood $V_p$ of $p$ such that every admissible curve $t\mapsto \ga(q,u(.),t)$ 

1215: entering $V_p$ 

1216: at time $\tau$ leaves $V_p$ before time $\tau+\eps$ and never comes back to $V_p$ after 

1217: time $\tau+\eps$.

1218: 

1219: %\medskip

1220: 

1221: Since $X$ and $Y$ are not parallel at $p$, we can choose a coordinate system $(x,y)$ such that 

1222: $X(p)=(1,-1)$ 

1223: and $Y(p)=(1,1)$. We denote $p=(p_x,p_y)$, $X(x,y)=(X_1(x,y),X_2(x,y))$, $Y(x,y)=(Y_1(x,y),Y_2(x,y))$.

1224: The fields $X$ and $Y$ being continuous, there exists $\al>0$ such that, if $(x,y)\in B_\infty(\al)=

1225: \{(a,b)\;|\; |a-p_x|<\al, \; \; |b-p_y|<\al \}$, then $X_1(x,y)$, $Y_1(x,y)$, $-X_2(x,y)$, and  

1226: $Y_2(x,y)$ are in 

1227: $[{1}/{2},{3}/{2}]$.

1228: 

1229: %\medskip

1230: 

1231: Let $p'=(p_x-\frac{\al}{10},p_y)$ and consider $\ga_X(p',.)=(\ga_X^1(p',.),\ga_X^2(p',.))$. Its first 

1232: coordinate $\ga_X^1(p',.)$ is increasing and its derivative takes values in 

1233: $[{1}/{2},{3}/{2}]$. 

1234: %

1235: The same is true for $-\ga_X^2(p',.)$. Hence $\ga_X(p',.)$ does not leave the set 

1236: $B_\infty(\al)$ before time ${2\al}/{3}$. Since $\ga_X^1(p',2\al/5)$ is larger than

1237: $p_x+\frac{\al}{10}$ and 

1238: $\ga_X^2(p',2\al/5)$ is in $[p_y-\frac{3\al}{10},p_y-\frac{\al}{10}]$, then the curve $\ga_X(p',.)$

1239: intersects the segment $S_p=B_\infty(\al)\cap\{(x,y)|x=p_x+\frac{\al}{10}\}$ 

1240: in a time $\tau_X$ smaller than ${2\al}/{5}$. 

1241: 

1242: The same occurs for $\ga_Y(p',.)$. Denote by $\tau_Y$ its intersection time with $S_p$. 

1243: 

1244: 

1245: Choose as $\TT$ the bounded set whose boundary 

1246: is given by the union of

1247: $\ga_X(p',[0,\tau_X])$, $\ga_Y(p',[0,\tau_Y])$, and

1248: the segment $[\ga_X(p',\tau_X),\ga_Y(p',\tau_Y)]=\{\lb\,\ga_X(p',\tau_X)+(1-\lb)\,\ga_Y(p',\tau_Y)\,|\;0\leq \lb\leq 1\}$ (see Figure~\ref{TTT}).

1249: 

1250: \bfi[h]

1251: \begin{center}

1252: \input{stab2.pstex_t}

1253: \caption{The set $\TT$}\label{TTT}

1254: \end{center}

1255: \efi

1256: 

1257: The following %Lemma \ref{pT} 

1258: lemma states that $\TT$ satisfies the required properties. % stated at the beginning of the section. 

1259: As a consequence, $p$ cannot be the 

1260: accumulation point of any admissible curve.

1261: 

1262: 

1263: \bl \label{pT}

1264: We have the following:

1265: (i) $\TT$ is a neighborhood of $p$; (ii) every admissible curve entering $\TT$ leaves $\TT$ in a time smaller than ${2\al}/{5}$ through  the segment $[\ga_X(p',\tau_X),\ga_Y(p',\tau_Y)]$; (iii) once an admissible curve leaves $\TT$, it enters $\A(p')\setminus \TT$ and never leaves it.

1266: %\begin{enumerate}

1267: %\item $p\in \mbox{int}(\TT)$;

1268: %\item Any admissible curve entering in $\TT$ leaves $\TT$ in time less than $\frac{2\al}{5}$ 

1269: %along the segment $[\ga_X(p',\tau_X),\ga_Y(p',\tau_Y)]$.

1270: %\item Once the curve leaves $\TT$, it enters $A(p')\setminus \TT$ and never leaves it.

1271: %\end{enumerate}

1272: \el

1273: 

1274: 

1275: 

1276: \noindent\proof

1277: The first point follows by the construction of $\TT$. As for ({\it ii}),

1278: notice that all the points of $\TT$ have first coordinate in 

1279: $[p_x-\frac{\al}{10},p_x+\frac{\al}{10}]$. Since the first coordinate 

1280: of $X$ and $Y$ is larger than ${1}/{2}$, then every admissible curve entering $\TT$ 

1281: leaves it in a time smaller than ${2\al}/{5}$. 

1282: Moreover, since along $\ga_X(p',.)$ and $\ga_Y(p',.)$ the admissible velocities of \r{s-system} point inside $\TT$, then an admissible curve can leave $\TT$ only through  the segment $[\ga_X(p',\tau_X),\ga_Y(p',\tau_Y)]$.

1283: Finally, ({\it iii}) follows from the remark that $\A(p')\setminus \TT$ is invariant 

1284: for the dynamics, since the admissible velocities of \r{s-system} 

1285: point inside $\A(p')\setminus \TT$ all along its boundary. \EOP

1286: 

1287: 

1288: 

1289: 

1290: %\item Clear.

1291: %\item Because all the points of $\TT$ have first coordinates in 

1292: %$[p_x-\frac{\al}{10},p_x+\frac{\al}{10}]$ and because the first coordinates of $X$ and $Y$ are 

1293: %larger than $\frac{1}{2}$, then any curve entering $\TT$ leaves it in time less than $\frac{2\al}{5}$.

1294: %Moreover because along $\ga_X(p',.)$ and $\ga_Y(p',.)$ the dynamics points inside $\TT$, the curve can leave $\TT$ 

1295: %only along the segment $[\ga_X(p',\tau_X),\ga_Y(p',\tau_Y)]$. Hence it enters $\A(p')\setminus \TT$.

1296: %\item The dynamics points inside the set $\A(p')\setminus \TT$ hence, once an admissible curve enters 

1297: %$\A(p')\setminus \TT$, it cannot leave it. Hence, an admissible curve entering $\TT$ leaves $\TT$ by entering  

1298: %$\A(p')\setminus \TT$, in time less than $\frac{2\al}{5}$, and never comes back to $\TT$. \EOP

1299: %\end{enumerate}

1300: 

1301: 

1302: 

1303: 

1304: \subsection{Conclusion of the proof of Theorem \ref{t-mai-paralleli}}

1305: %

1306: We are left to prove that \r{s-system} is uniformly stable. To this extent, fix $\de>0$.

1307: Since both $X$ and $Y$ are stable at the origin, then

1308: there exists

1309: $\eps>0$ such that every integral curve of $X$ or $Y$ starting in $B_\eps$

1310: is contained in $B_\de$. 

1311: Hence, for every $q\in B_\eps$,

1312: the boundary of $\A(q)$ is contained in  $B_\de$.

1313: %in $B_\eps$, 

1314: %$$\partial \A(q)\subset \gamma_X(q,[0,\infty))\cup \gamma_Y(q,[0,\infty))\cup\{0\}\subset B_\de\,.$$

1315: Therefore,  $\A(q)$, being bounded, 

1316: is itself contained in $B_\de$. \EOP

1317: 

1318: 

1319: 

1320: \begin{rmk}

1321: The proof of Theorem~\ref{t-mai-paralleli} naturally extends  to the following case: if $V$ is an open and

1322: simply connected subset of $\R^2$, if $X$ and $Y$  point inside $V$ along its 

1323: boundary, and if $\qq\cap V=\{0\}$, 

1324: then 

1325: \r{s-system}

1326: is uniformly asymptotically stable on $V$. 

1327: \end{rmk}

1328: 

1329: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1330: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1331: \section{Proof of Theorem \ref{t-mai-tangent}}\label{mt}

1332: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1333: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1334: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1335: 

1336: 

1337: 

1338: 

1339: The proof follows the main steps as the one of 

1340: Theorem \ref{t-mai-paralleli}. The idea is again to fix a point $q\in\R^2$, to characterize the boundary 

1341: of its accessible set $\A(q)$, to prove that such set is bounded, 

1342: and, finally, to show that no admissible curve has an accumulation point different from the origin.

1343: 

1344: In order to describe the boundary of ${\A(q)}$, we need some extra construction.

1345: Notice that every component $\Ga$ of $\qq$ separates the plane in two 

1346: parts. 

1347: Since $\Ga$ contains no tangency points, then one of such two regions must be invariant for $X$, and the same argument holds for $Y$ as well.

1348: Necessarily, the invariant region is the one containing the origin, which is attractive both for $X$ and $Y$. In particular, $\Ga$ is direct

1349: %$X$ and $Y$ are pointing in the same direction along $\Ga$.

1350: and every admissible curve %$\ga$ 

1351: crosses $\Ga$ at most once.

1352: Associate with every point $q\in \R^2$ the number $n(q)$ 

1353: of components of $\qq$ that the curve $\ga_X(q,.)$ crosses 

1354: at strictly positive times, 

1355: before converging to the origin (see Figure~\ref{nq}). 

1356: %As a convention, if $q$ lies on a component $\Ga$ of $\qq$, 

1357: %then $\Ga$ is not computed in $n(q)$.

1358: Since the curve $\ga_X(q,(0,\infty))$ is bounded and crosses 

1359: each component of $\qq$ at most  once, then $n(q)$ 

1360: is finite. 

1361: \bfi[h]

1362: \begin{center}

1363: \input{nq.pstex_t}

1364: \capt\label{nq}

1365: \end{center}

1366: \efi

1367: 

1368: %\vspace{-20mm}

1369: 

1370: For every $i\leq n(q)$, let us denote by $\Ga_i$ the $i$-th component of $\qq$ crossed by 

1371: $\ga_X(q,.)$. We claim that $\ga_Y(q,.)$ crosses exactly the same components as $\ga_X(q,.)$, 

1372: in the same order. 

1373: Otherwise, as one can easily check, $X$ and $Y$ would not both be GAS at the origin (the reason is that

1374: the components of $\qq$ separate the plane and can be crossed by

1375: an admissible curve at most once).

1376: 

1377: Let us define two admissible curves, starting from $q$, 

1378: that can be used to characterize the boundary of $\A(q)$, 

1379: in analogy with what has been done in the proof of Theorem \ref{t-mai-paralleli}. 

1380: The first of such curves follows the flow of $X$ until it reaches $\Ga_1$, 

1381: then follows the flow of $Y$ 

1382: until it crosses $\Ga_2$, and so on.

1383: The second one follows alternatively the flows of $X$ and $Y$ in the other way round, 

1384: starting with $Y$ and switching to $X$ as it meets $\Ga_1$. 

1385: Such two curves converge to the origin, since $n(q)$ is finite. 

1386: %One can prove this by induction on $n(q)$. 

1387: As in the proof of Theorem~\ref{t-mai-paralleli}, we can distinguish two cases, depending on whether the two curves intersect or not (see Figure~\ref{acc3}). 

1388: \bfi[h]

1389: \begin{center}

1390: \input{acc3.pstex_t}

1391: \capt\label{acc3}

1392: %\caption{}

1393: \end{center}

1394: \efi

1395: The arguments of Section~\ref{mp} can be adapted

1396: in order to prove the boundedness of \r{s-system} and the absence of accumulation points different from the origin. The details are left to the reader.\EOP

1397: 

1398: 

1399: 

1400: 

1401: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1402: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1403: \section{Proof of Theorem \ref{t-solo-compatte}}\label{sc}

1404: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1405: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1406: 

1407: 

1408: 

1409: 

1410: Consider a system of coordinates $(x,y)$ on $\R^2$ which preserves the origin and renders 

1411: $X$ radial outside a ball $B_{R_0}$, 

1412: $R_0>0$. (Such system can be defined using the level sets of a smooth Lyapunov 

1413: function for $X$, see \cite{g2g}.)

1414: Taking possibly a larger $R_0$, we can assume that $X$ and $Y$ are never collinear in $\R^2\setminus B_{R_0}$.

1415: 

1416: 

1417: For every $R>0$, 

1418: %let $B_R$ be the ball of center $R$ centered at the origin, and 

1419: let

1420: $$\Om_R=\cup_{p\in B_R} \A(p)\,.$$

1421: Our aim is to prove that each $\Om_R$ is bounded. 

1422: 

1423: Fix $R>R_0+1$.

1424: If $(X,Y)$ is replaced with a pair of vector fields $(\Xt,\Yt)$ which coincides with $(X,Y)$ outside $B_{R_0+1}$, then the set $\Om_R$, constructed as above,

1425: does not change.

1426: The idea is to choose $\Xt$ and $\Yt$ in such a way that they are never parallel outside the origin and still GAS. The boundedness of $\Om_R$ follows then from Theorem \ref{t-mai-paralleli}.

1427: 

1428: Set

1429: \brs

1430: X_0(x,y)&=&-x\partial_x-y\partial_y\,,\\

1431: Y_0(x,y)&=&y\partial_x-x\partial_y+\lambda X_0\,,\ \ \ \ \ \ \ \ \ \lb>0\,,

1432: \ers

1433: and notice that $X$ and $X_0$ are collinear outside $B_{R_0}$.

1434: Notice, moreover, that, if $\lambda$ is large enough, then the angle between $X_0$ and $Y_0$ is smaller than the minimum of the angles between $X$ and $Y$ in $B_{R_0+1}\setminus B_{R_0}$ (see Figure~\ref{rad2}). Fix such a $\lambda$.

1435: 

1436: 

1437: 

1438: 

1439: \bfi[h]

1440: \begin{center}

1441: \input{radial2.pstex_t}

1442: \capt\label{rad2}

1443: \end{center}

1444: \efi

1445: 

1446: The function $Q$ has constant sign on $\R^2\setminus B_{R_0}$. Without lost of generality, we can assume that it is positive. 

1447: Fix a smooth function $\phi:[0,+\infty)\rw[0,1]$ such that $\phi(r)=0$ if $r\leq R_0$ and $\phi(r)=1$ if $r\geq R_0+1$.

1448: Define

1449: \brs

1450: \Xt(x,y)&=&\lp 1-\phi\lp\sqrt{x^2+y^2}\rp\rp X_0(x,y)+\phi\lp\sqrt{x^2+y^2}\rp X(x,y)\,,\\

1451: \Yt(x,y)&=&\lp 1-\phi\lp\sqrt{x^2+y^2}\rp\rp Y_0(x,y)+\phi\lp\sqrt{x^2+y^2}\rp Y(x,y)\,.

1452: \ers

1453: 

1454: By construction, $(\Xt,\Yt)$ coincides with $(X,Y)$ outside $B_{R_0+1}$ and $det(\Xt,\Yt)$ is strictly positive on $\R^2\setminus\{0\}$. 

1455: We are left to check the global asymptotic stability of $\Yt$,  

1456: the one of $\Xt$ being evident. 

1457: This can be done by using a comparison argument 

1458: between the integral curves of

1459: $Y$ and $\Yt$. % on $B_{R_0+1}\setminus B_{R_0}$.

1460: Indeed, since the angle between $\Xt$ and $\Yt$ is smaller than 

1461: %or equal to 

1462: the angle between 

1463: $\Xt$ and $Y$ in $B_{R_0+1}\setminus B_{R_0}$, then 

1464: the integral curve of $\Yt$ starting from a point $q\in\R^2\setminus B_{R_0}$ joins 

1465: $B_{R_0}$ in finite time, 

1466: with a smaller total variation in the angular component that the integral curve of $Y$ starting from the same point $q$.

1467: \EOP

1468: 

1469: \begin{rmk}

1470: The proof given above applies, without modifications, to the more general case where 

1471: the points at which 

1472: $X$ and $Y$ are 

1473: globally asymptotically stable 

1474: are allowed to be different.

1475: \end{rmk}

1476: 

1477: 

1478: \begin{rmk}\label{gus}

1479: %\noindent {\bf Remark.} 

1480: The conclusion of Theorem~\ref{t-solo-compatte} would not hold 

1481: under the weaker hypothesis that $X$ and $Y$ are GUS, instead of GAS. 

1482: %

1483: A counterexample can be given as follows: Let $\varphi:[0,1]\rw\R$ be a smooth function such that $0<\varphi(t)<\pi/2$ for every $t\in(0,1)$ and $\varphi^{(k)}(0)=\varphi^{(k)}(1)=0$ for every $k\geq 0$. Denote by $(r,\theta)$ the radial coordinates on $\R^2$. 

1484: Define, using the radial representation of vectors in $\R^2$, 

1485: \[X(r,\theta)=\left\{\begin{array}{ll}

1486:                                   \lp r,\theta+\frac\pi2+\varphi(r)\rp&\mbox{if }\ r\in[0,1]\,,\\[.8ex]

1487:                                   \lp r,\theta+\frac\pi2-\varphi(r-[r])\rp&\mbox{if \ }r>1\,,

1488: \end{array}

1489: \right.\]

1490: and

1491: \[Y(r,\theta)=\left\{\begin{array}{ll}

1492:                                   \lp r,\theta-\frac\pi2-\varphi(2r)\rp&\mbox{if }\ r\in\left[0,\frac12\right]\,,\\[.8ex]

1493:                                   \lp r,\theta-\frac\pi2+\varphi\lp r+\frac12-\left[r+\frac12\right]\rp\rp&\mbox{if }\ r>\frac12\,,

1494: \end{array}

1495: \right.\]

1496: where $[r]$ denotes the integer part of $r$. 

1497: Then, for every $r\geq 1$, $X(r,\theta)$ and $Y(r,\theta)$ are linearly independent, since

1498: %\[0<

1499: the difference between their angular components is given by

1500: \[0<\pi-\varphi(r-[r])-\varphi\lp r+\frac12-\left[r+\frac12\right]\rp<\pi\,.\]

1501: Hence, $\qq$ is compact. On the other hand, 

1502: %a strategy which sends $\gamma(p,u(.),t)$ to infinity is given by 

1503: the feedback strategy

1504: \[u(t)=\left\{\ba{ll}0&\mbox{if }\ r-[r]\in\left[\frac14,\frac34\right),\\

1505:                      1&\mbox{otherwise}

1506: \ea\right.\]

1507: is such that, for every $p\in\R^2\setminus B_{3/4}$,

1508: $\|\gamma(p,u(.),t)\|$ tends to infinity 

1509: as $t$ tends to $\T(p,u(.))=+\infty$. 

1510: 

1511: 

1512: 

1513: Notice that the example can be easily modified in such a way that $\qq$ not only is compact, but actually shrinks to $\{0\}$.

1514: It suffices to take $X(r,\theta)=(r,\theta+\psi_X(r))$ and

1515: $Y(r,\theta)=(r,\theta+\psi_Y(r))$, where the graphs of $\psi_X$

1516: and $\psi_Y$ are as in Figure~\ref{psis}.

1517: \bfi[h]

1518: \begin{center}

1519: \input{psis.pstex_t}

1520: \capt\label{psis}

1521: \end{center}

1522: \efi

1523: 

1524: 

1525: \end{rmk}

1526: 

1527: 

1528: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1529: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1531: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1532: \section{Proof of Theorem \ref{t-inv-non-comp}}\label{inc}

1533: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1534: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1535: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1536: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1537: 

1538: Let $\Gamma$ be an inverse unbounded component of $\qq$ and assume that {\bf (G1)} and {\bf (G3)} hold.

1539: %Due to the assumption {\bf (G1)}, 

1540: Due to Lemma~\ref{genio}, $\Gamma$  is a one-dimensional submanifold of $\R^2$, which can be parameterized 

1541: by an injective and smooth map $c:\R\rw \R^2$. 

1542: 

1543: 

1544: 

1545: 

1546: Fix a point $p=(p_x,p_y)=c(\tau)$ on $\Ga$. %Without loss of generality, $p_x=0$. 

1547: According to the results by Davydov (see \cite[Theorem 2.2%, p.~37

1548: ]{dav}),

1549: % (see also \cite[Theorem 4.1]{dav}), 

1550: %under the generic conditions {\bf (G1)} and {\bf (G3)}, 

1551: up to a change of coordinates (which, in particular, sets $p_x=0$), the vector fields $X$ and $Y$ can be represented locally 

1552: by one of the following three normal forms

1553: \begin{enumerate}

1554: \item $X(x,y)=(1,x),\ Y(x,y)=(-1,x)$;\label{passing}

1555: \item $X(x,y)=(1,y-p_y-x^2),\ Y(x,y)=(-1,y-p_y-x^2)$;\label{tu1}

1556: \item $X(x,y)=(-1,x^2-y+p_y),\ Y(x,y)=(1,x^2-y+p_y)$.\label{tu2}

1557: \end{enumerate}

1558: 

1559: \bfi[h]

1560: \begin{center}

1561: \input{forme-normali-finali.pstex_t}

1562: \capt\label{FN}

1563: \end{center}

1564: \efi

1565: \indent Notice that the type \ref{passing} corresponds to the situation in which $X$ and $Y$ are transversal to $\Ga$ at $p$, while \ref{tu1} and \ref{tu2} are the normal forms 

1566: for the case in which $X$ and $Y$ are tangent to $\Ga$ at $p$.

1567: 

1568: Recall that $p$ is said to have the {\it small time local transitivity}   property  (STLT, for short) %for  \r{s-system} (seen as a control system) 

1569: if, for every $T>0$ and every neighborhood $V$ of $p$, there exists a neighborhood $W$ of $p$ such that every two points in $W$ are accessible from each other within time $T$ by an admissible trajectory contained in $V$.

1570: It has been proved in  \cite[Theorem 3.1%, p.~43

1571: ]{dav} that, under the assumption that the system admits a local representation in normal form, 

1572: $p$ has the STLT property 

1573: if and only if it

1574: is of the type \ref{passing}. 

1575:  In particular, if $p$ is of the type \ref{passing}, then there exist $t(p),T(p)>0$ such that, for every $r,s\in (-t(p),t(p))$ there exists an admissible trajectory which steers $c(\tau+r)$ to $c(\tau+s)$ within time $T(p)$. 

1576: 

1577: 

1578: 

1579: 

1580: Assume now that $p$ is a point of the type \ref{tu1} or \ref{tu2}. %, that is, that $X(p)$ is tangent to $\Ga$.  

1581: The curve $\Ga$ stays (locally) on one side of the affine line 

1582: $$p+\mbox{span}(X(p))=\{(x,p_y)\,|\;x\in\R\}\,,$$ 

1583: which is 

1584: the affine tangent space to $\Ga$ at $p$. 

1585: Up to a reversion in the parameterization of $\Ga$, 

1586: we can assume that, for every $t$ in a right neighborhood of $\tau$, $X(c(t))$ points into the locally convex part of the plane bounded by $\Ga$ 

1587: (see Figure~\ref{FN}).

1588: %

1589: %\begin{center}

1590: %\input{papa.pstex_t}\\

1591: %Figure 5.

1592: %\end{center}

1593: %

1594: %

1595: It can be easily verified that the two branches of $\Ga\setminus\{p\}$ are connected by 

1596: integral curves of $X$ and $Y$ arbitrarily close to $p$, in the following sense: 

1597: for every $t>0$ small enough,

1598: %neighborhood $U$ of $p$ 

1599: there exist $\theta,T>0$  such that, for every $r\in(0,\theta)$, both curves 

1600: $s\mt \ga_X(c(\tau+r),s)$

1601: and

1602: $s\mt \ga_Y(c(\tau-r),s)$

1603: intersect $\Ga$ in a positive time smaller than $T$, and the intersection points are of the type $c(\tau+\rho)$, with $0<|\rho|\leq t$.

1604: We can conclude, using the STLT property at points of $\Ga\setminus\{p\}$ close to $p$,  

1605:  that there exists $t(p)>0$ such that,  for every $\mu\in(0,1)$, 

1606: %there exists $T(p,\mu)>0$ such that, 

1607: every two points 

1608: of 

1609: $$\Sigma=\{c(\tau+r)|\;\mu\,t(p)<|r|<t(p)\}$$

1610: can be joined by an admissible trajectory 

1611: of time-length bounded by a uniform $T(p,\mu)>0$.

1612: 

1613: Therefore, given any pair of points $p_i=c(\tau_i),p_f=c(\tau_f)$ on $\Ga$ of type \ref{passing}, there exists an admissible trajectory going from $p_i$ to $p_f$ of time-length smaller that 

1614: $T(c(\tau_1),\mu_1)+\cdots+T(c(\tau_k),\mu_k)$, where 

1615: $$(\tau_1-t(c(\tau_1)),\tau_1+t(c(\tau_1))),\ldots,(\tau_k-t(c(\tau_k)),\tau_k+t(c(\tau_k)))$$

1616: is a covering of the compact segment of $\R$ bounded by $\tau_i$ and 

1617: $\tau_f$, $\mu_1,\ldots,\mu_k\in (0,1)$ are properly chosen 

1618:  and $T(p,\mu)=T(p)$ if $p$ is of type \ref{passing}.

1619: In particular, system (\ref{s-system}) admits  trajectories going to infinity.\EOP

1620: 

1621: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1622: \begin{rmk}

1623: In the non-generic case the statement of Theorem \ref{t-inv-non-comp} is false. A 

1624: counterexample can be 

1625: found even in the linear case. Indeed, consider the vector fields

1626: \bqn

1627: &&X(q)=A\,q,\ \ \ \ \ \ \ \mbox{ where}~~~ 

1628: A=\left(\ba{cc} -1/20& -1/E\\E&-1/20\ea\right),~~~

1629: E=-\frac{201}{200}-\frac{\sqrt{401}}{200}\,,\nn\\

1630: &&Y(q)=B\,q,\ \ \ \ \ \ \ \mbox{ where}~~~ 

1631: B=\left(\ba{cc} -1/20& -1\\1&-1/20\ea\right).

1632: \eqn

1633: The integral curves of $X$ are ``elliptical spirals'', 

1634: %($E$ is the eccentricity), 

1635: while the integral curves of $Y$ are ``circular spirals''. 

1636: The integral curves of $X$ and $Y$ rotate around the origin in 

1637: opposite sense (since $E<0$).

1638: One can easily check that, in this case, the set $\qq$ is a single 

1639: straight line of equation

1640: \bqn

1641: y=-\frac{20}{\sqrt{401}-1}x\,,

1642: \eqn

1643: and its two components are inverse (see Figure~\ref{ex-l}). 

1644: 

1645: It can be checked by hand %has been proved in \cite{SIAM} 

1646: that 

1647: the switched system defined by $X$ and $Y$ is GUS, although not GUAS 

1648: (see also \cite{SIAM}, Theorem 2.3, case (CC.3)). 

1649: %(The result is somehow 

1650: %intuitive, since by switching among the two vector fields, one cannot go 

1651: %far from the origin.)

1652: 

1653: \bfi[h]

1654: \begin{center}

1655: \input{ex-lin.pstex_t}

1656: \capt\label{ex-l}

1657: \end{center}

1658: \efi

1659: 

1660: \end{rmk}

1661: 

1662: 

1663: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1664: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1665: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1666: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1667: 

1668: 

1669: 

1670: 

1671: 

1672: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1673: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1674: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1675: 

1676: 

1677: 

1678: \newcommand{\auth}[1]{\textsc{#1}}

1679: \newcommand{\tit}[1]{\textit{#1}}

1680: \newcommand{\jou}[1]{\textrm{#1}}

1681: \newcommand{\bff}[1]{{\bf {#1}}}

1682: \newcommand{\bbf}[1]{{\bf {#1}}}

1683: \begin{thebibliography}{10}

1684: 

1685: 

1686: 

1687: \bibitem{generiko} \auth{R. Abraham and J. Robbin}, \tit{Transversal mappings and flows}, W. A. 

1688: Benjamin, Inc., New York-Amsterdam, 1967.

1689: 

1690: 

1691: \bbibitem{agr} \auth{A.~A.~Agrachev and D.~Liberzon}, \tit{Lie-algebraic stability

1692: criteria for

1693: switched systems},  

1694: \jou{SIAM J. Control Optim.}, 

1695: %\jou{SIAM Journal on Control and Optimization}, 

1696: %Vol. 40, no. 1, pp.~253--269.

1697: \bff{40} (2001), pp.~253--269.

1698: 

1699: \bbibitem{angeli}  \auth{D.~Angeli}, \tit{A note on stability of arbitrarily switched

1700: homogeneous systems}, to appear on \jou{Systems Control Lett.}

1701: 

1702: 

1703: \bbibitem{AISW} \auth{D.~Angeli, B.~Ingalls, E.~D.~Sontag, and Y.~Wang}, \tit{Uniform global asymptotic stability of differential inclusions},  \jou{J. Dynam. Control Systems},  \bff{10}  (2004),  

1704: %no. 3, 

1705: pp.~391--412.

1706: 

1707: 

1708: \bibitem{blanchini} \auth{F. Blanchini and S. Miani},

1709: \tit{A new class of universal Lyapunov functions for the control of

1710: uncertain linear systems}, IEEE Transactions on Automatic Control, \bff{44} (1999),

1711: %no. 4, 

1712: pp.~641--647.

1713: 

1714: 

1715: \bbibitem{SIAM}  \auth{U.~Boscain}, \tit{Stability of planar switched systems: the

1716: linear single input  case},   

1717: \jou{SIAM J. Control Optim.}, \bff{41} (2002), pp.~89--112.

1718: %{ SIAM Journal on Control and Optimization}, Vol. {41}. no.1, pp. 89-112. 

1719: 

1720: 

1721: 

1722: 

1723: \bibitem{dav} \auth{A.~A.~Davydov}, \tit{Qualitative theory of control systems}, 

1724: \jou{Transl. Math. Monogr.}, Amer. Math. Soc., Providence, 1994.

1725: %\jou{Translations of Mathematical Monographs}, American Mathematical Society, 1994.

1726: 

1727: 

1728: \bibitem{daw} \auth{W.~P.~Dayawansa and C.~F.~ Martin}, \tit{A converse Lyapunov

1729: theorem for a class of dynamical systems which undergo switching},

1730: \jou{IEEE Trans. Automat. Control}, \bff{44} (1999), pp.~751--760.

1731: %\jou{IEEE Transactions on Automatic Control}, 

1732: 

1733: 

1734: 

1735: \bbibitem{g2g} \auth{L.~Gr\"une, E.~D.~Sontag, and F.~R.~Wirth}, \tit{Asymptotic stability equals exponential stability, and ISS equals finite energy gain---if you twist your eyes}, \jou{Systems Control Lett.}, \bff{38} (1999), %no. 2, 

1736: pp.~127--134.

1737: 

1738: 

1739: \bbibitem{12} \auth{B.~Ingalls, E.~D.~Sontag, and Y.~Wang}, \tit{An infinite-time relaxation theorem for differential inclusions},  \jou{Proc. Amer. Math. Soc.},  \bff{131}  (2003),  

1740: %no. 2, 

1741: pp.~487--499.

1742: 

1743: 

1744: 

1745: \bibitem{liberzon-book}  \auth{D. Liberzon}, \tit{Switching in systems and

1746: control}, Volume in series  Systems \& Control: Foundations \& Applications,

1747: Birkh\"auser, Boston, 2003.

1748: 

1749: \bibitem{hesp} \auth{D.~Liberzon,  J.~P.~Hespanha, and A.~S.~Morse}, \tit{Stability of switched systems: a Lie-algebraic condition}, \jou{Systems Control Lett.}, \bff{37} (1999), pp.~117--122.

1750: %, no. 3, pp. 117-122, Jun 1999. 

1751: 

1752: 

1753: 

1754: \bbibitem{survey} \auth{D.~Liberzon and A.~S.~Morse}, \tit{Basic problems in stability and design of switched systems}, \jou{IEEE Control Syst. Mag.}, %IEEE Control Systems Magazine

1755:  \bff{19} (1999),  pp.~59--70.

1756: %, no. 5, pp. 59-70, Oct. 1999.

1757: 

1758: \bbibitem{lyapunov-paolo} \auth{P.~Mason, U.~Boscain, and  Y.~Chitour}, 

1759: \tit{Common polynomial 

1760: Lyapunov 

1761: functions for linear switched systems}, Preprint SISSA 52/2004/M, 

1762: {\tt http://arxiv.org/abs/math.OC/0403209}. 

1763: 

1764: 

1765: 

1766: 

1767: 

1768: \end{thebibliography}

1769: 

1770: \end{document}

1771: 

1772: % LocalWords:  Ugo Boscain SISSA ISAS Trieste Charlot Sigalotti INRIA Antipolis

1773: % LocalWords:  convexification XXX GUAS GUES submanifold Angeli Ingalls Sontag

1774: % LocalWords:  linearization DAVYDOV Lyapunov diffeomorphism STLT affine pp

1775: % LocalWords:  Davydov Dayawansa Liberzon Hespanha Narendra AMMUINA convexified

1776: % LocalWords:  genericity Withney Withney's attractivity Montpellier Cedex

1777: % LocalWords:  lucioles

1778: