Sie sind auf Seite 1von 590

Lecture Notes in Control and Information Sciences

Editor: M. Thoma

258

Springer

London

Berlin

Heidelberg

New York

Barcelona

Hong Kong

Milan

Paris

Singapore

Tokyo

Alberto Isidori, Franqoise Lamnabhi-Lagarrigue,

andWitoldRespondek(Eds)

Nonlinear Control in the Year 2000

Volume 1

With 86 Figures

~

Springer

Series

Advisory

Board

A.

Bensoussan • M.J. Grimble • P. Kokotovic • A.B. Kurzhanski •

H.

Kwakernaak • J.L. Massey • M. Morari

Editors

Alberto Isidori, Professor

Dipartimento di Informatica e Sistemistica, Universitt~di Roma "La Sapienza",

00184 Rome Italy

Franqoise Lamnabhi-Lagarrigue,

Laboratoire des Signaux et Systems, CNRS SUPELEC,

91192 Gif-sur-Yvette, France

Witold Respondek, Professor Laboratoire de Math6matique

76131 Mont Saint Aignan, France

Docteur D'6tat

et Informatique, INSA de Rouen,

ISBN 1-85233-363-4 Springer-Verlag London Berlin Heidelberg

British Library Cataloguing in Publication Data

A catalog record for this book is availablefrom the British Library

Library of Congress Cataloging-in-PublicationData Nonlinear control in the year 2000 / Alberto Isidori, Franqoise Larrmabhi-Lagarrigue, and Witold Respondek.

p. cm. -- (Lecture notes in control and information sciences, ISSN 0170-8643 ; 258-259)

Includes bibliographical references.

ISBN 1-85233-363-4 (v. 1 : acid-free paper) -- ISBN 1-85233-364-2 (v.2 : acid-free paper) 1. Nonlinear control theory. I. Isidori, Alberto. II. Lamnabhi-Lagarrigue, F. (Franqoise), 1953- III. Respondek, W. IV. Series. QA402.35.3N66 2001

003.5--dc21

00-045600

Apart from an), fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms

of licences issued by the Copyright Licensing Agency.

those terms should be sent to the publishers.

© Springer-Verlag London Limited 2001

Printed in Great Britain

Enquiries concerning reproduction outside

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence

of a specific statement, that such names are exempt from the relevant laws and regulations and

therefore free for general use.

The publisher makes no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept an}, legal responsibility or liability for any errors or omissions that may be made.

Typesetting: Camera ready by editors Printed and bound at the Athenaeum Press Ltd., Gateshead, Tyne & Wear 69/3830-543210 Printed on acid-free paper SPIN 10768731

Preface

These two volumes contain papers based on talks delivered at the 2nd Work-

shop of the Nonlinear Control Network (http://www.supelec.fr/lss/~lCN),

held in Paris, June 5-9, 2000. The Authors of the presented papers, as well as the Editors of these two volumes, hope that "Nonlinear Control in the Year 2000" is not only one more book containing proceedings of a workshop. Two main reasons justifying our hope are that, firstly, the end of the century is a natural moment to think about past developments in nonlinear control and about its perspectives in the twenty-first century; and, secondly, we be- lieve that nonlinear control has reached an age of maturity which enables the community to sketch a state-of-the-art of the field. We hope that papers contained in the two volumes serve to fulfill these goals: many of them have their roots and their origins in nonlinear control theories which have been developed in past three decades and which, by now, form a basis of canonical results in the field. Such papers form a bridge between the actual theory and its future developments. Many other papers contained in the book present completely new ideas and suggest new directions and, in this sense, they are directed towards the future of the nonlinear control.

We would like to emphasize one peculiarity of our field: nonlinear control is an example of a theory situated at a crossroad between mathematics and engineering science. Due to this position, nonlinear control has its roots in both fields and, as we deeply believe, can bring new ideas and new results for both domains. The book reflects very well this "double character" of nonlinear control theory: the reader will find in it results which cover a wide variety of problems: starting from pure mathematics, through its applications to nonlinear feedback design, all way to recent industrial advances.

Eight papers contained in the book are based on invited talks delivered at the Workshop by:

Alessandro Astolfi,

John Baras,

Christopher Byrnes,

Bronistaw Jakubczyk,

Anders Rantzer,

Kurt Schlacher,

Eduardo Sontag,

Hector Sussmann.

Altogether the book contains 80 papers and therefore it is impossible to mention all discussed topics: to give a flavour of the presented material let us mention a few of them. For many theoretical papers a common factor is optimal control, for example sub-Riemannian geometry (in particular fun- damental results on (non)subanalyticity of small balls and of the distance function), the use of generalized differentials in generalized Maximum Princi- ple, singular and constrained optimal control problems. Another subdomain of nonlinear control attracting many contributions to the book is stability and asymptotic behavior. In this area, stemming from traditional Lyapunov

VI

techniques, the methods based on the concept of input-to-state stability have established a leading role in analysis and design, and new ideas have emerged, such as those aiming at the analysis of non-equilibrium steady-state behav- iors, or at the evaluation of asymptotic convergence for almost all initial conditions via dual Lyapunov analysis. Applications of these ideas in non- linear control are widespread: stabilization of nonlinear systems, trajectory tracking, adaptive control. Other subdomains of nonlinear control attract- ing a lot of attention, and represented in the book, are that of observability and observers for nonlinear systems, sliding mode control, theory of nonlin- ear feedback (invariants, classification, normal forms, flatness and dynamic feedback), recursive design (backstepping and feedforwarding). The papers present in the two volumes cover various aspects of all just mentioned topics. Moreover, the book contains also papers discussing main results of a plenary Session devoted to industrial applications.

We wish to thank all invited speakers and all contributors to the 2nd Non- linear Control Network Workshop for making this conference an outstanding intellectual celebration of the area of Nonlinear Control at the turn of the century. The Editors are grateful to all the chairpersons:

Dirk Aeyels, Alfonso BaSos, Fritz Colonius, Halina Frankowska, Bronistaw Jakubczyk, Jean L~vine, Frdddric Mazenc, Philippe Mullhaupt, Elena Panteley, Pierre Rouchon, Arjan van der Schaft, Fatima Silva Leite, Alan Zinober.

Andrei Agrachev, Georges Bastin, Emmanuel Delaleau, Jean-Paul Gauthier, Philippe Jouan, Antonio Loria, Gdrard Montseny, Henk Nijmeijer, Laurent Praly, Joachim Rudolph, Jacquelien Scherpen, Hebertt Sira-Ramirez,

Andrea Bacciotti, Antonio Bicchi, Michel Fliess, Henri Huijberts, loan Landau, Riccardo Marino, Claude Moog, Romeo Ortega, Anders Rantzer, Andrey Sarychev, Rodolphe Sepulchre, Fabian Wirth,

They have excellently played their role during the presentations and also have helped us in reviewing the contributed papers accepted for publica- tion in this book. We would like to thank the TMR Program for the financial support which, in particular, helped numerous young researchers in attending the Workshop. We express our thanks to Christiane Bernard (Mathematics and Information Sciences Project Officer at the European Community) and Radhakisan Baheti (Program Director for Control at NSF) for their partic- ipation at the Session on Founding to Academic Research on Control. We also thank the CNRS staff as well as PhD students and Postdocs at L2S in Gif-sur-Yvette for their help in the organization of the Workshop.

Frangoise Lamnabhi-Lagarrigue

Gif-sur-Yvette

Alberto Isidori

Rome

Witold Respondek

Rouen

Contents

Volume

1

Subanalyticity

of distance

and

 

spheres

in S-R

geometry

1

Andrei Agrachev, Jean-Paul Gauthier

 

Principal

invariants

of Jacobi

curves

 

9

Andrei Agrachev, Igor Zelenko

 

The

De

Casteljau

algorithm

on

SE(3)

23

Claudio Altafini

 

Trajectory tracking by cascaded backstepping control for a second-order nonholonomic mechanical system

35

Nnaedozie P.I. Aneke, Henk Nijmeijer, Abraham G. de Jager

Static output feedback stabilization: from linear to nonlinear and back

49

Alessandro Astolfi,

Patrizio Colaner~

 

Semi-linear diffusive representations for nonlinear fractional differential systems

73

Jacques Audounet, Denis Matignon,

Gdrard Montseny

 

Controllability properties of a class of control systems on Lie groups

83

Victor Ayala, Luiz A. B. San Martin

 

Stability analysis to parametric uncertainty extension to the multivariable case

93

Miguel Ayala Botto, Ton van den Boom, Josg Sd da Costa

 

External

stability and

continuous

Liapunov

functions

103

Andrea Bacciotti

 

Optimal

control

with

harmonic

rejection

of induction

machine

113

Iyad Balloul, Maze, Alamir

VIII

Nonlinear

QFT

synthesis

based

on harmonic

balance

and

multiplier

theory

 

123

Alfonso BaSos, Antonio Barreiro, Francisco Gordillo, Javier Aracil

 

Group

invariance

and

symmetries

in

nonlinear

control

and

 

estimation

 

137

John S. Baras

 

A

globalization procedure

for locally stabilizing controllers

 

171

Jochen Behrens, Fabian Wirth

 

Optimal

control and

implicit Hamiltonian

systems

 

I85

Guido Blankenstein,

Arian van der Schaft

 

Robust

absolute

stability of delay systems

 

207

Pierre-Alexandre Bliman

 

Stratification

du

secteur

anormal

dans

 

la sphere

de

Martinet

 

de petit

rayon

 

239

Bernard Bonnard, Emmanuel Trdlat

 

Nonequilibrium

theory

for nonlinear

systems

 

253

Christopher I. Byrnes

 

A

regularization

of Zubov's

equation

for

robust

domains

of

attraction

 

277

Fabio Camilli, Lars Griine, Fabian Wirth

 

A

remark

on Ryan's

generalization of Brockett's

condition

to

discontinuous

stabilizability

 

291

Francesca Ceragioli

 

Applications

of Lie algebroids in mechanics

and

control

theory 299

Jesds Clemente-Gallardo

 

Observer

design for locally observable analytic systems:

 

convergence and

separation

property

 

315

Fabio Conticelli, Antonio Bicchi

 

An

Hcc-suboptlmal fault

detection

filter for bilinear

systems

331

Claudio De Persis, Alberto Isidori

IX

Adaptive control of feedback linearizable systems orthogonal approximation functions

by

341

Domitilla Del Vecchio, Riccardo Marino, Patrizio Tomei

Sampled-data low-gain integral control of linear systems with actuator and sensor nonlinearities

355

Thomas Fliegner, Hartmut Logemann, Eugene P. Ryan

State feedbacks without asymptotic observers and generalized PID regulators

367

Michel Fliess, Richard Marquez', Emmanuel Delaleau

Eigenstructure of nonlinear Hankel

operators

385

Kenji Fujimoto, Jacquelien M.A. Scherpen

Distributed

architecture for teleoperatlon over the

internet

399

Denis Gillet, Christophe Salzmann, Pierre Huguenin

Singular L-Q problems and the Dirac-Bergmann theory of constraints

409

Manuel Guerra

Robust tracking of multl-varlable linear systems under parametric uncertainty

423

Veit Hagenmeyer

Flatness-based control of the

separately

excited

DC

drive

439

Veit Hagenmeyer, Philipp Kohlrausch,

Emmanuel Delaleau

State

detection and

stability for uncertain

dynamical systems

453

Mohamed Ali Hammami

Controllability properties of numerical eigenvalue algorithms

. 467

Uwe Helmke, Fabian Wirth

On the

diseretizatlon of sliding-mode-like controllers

481

Guido Herrmann, Sarah K. Spurgeon,

Christopher Edwards

Nonlinear adaptive state space control for a class of nonlinear

systems with unknown

parameters

497

Christian Hintz, Martin Rau, Dierk Schr6der

×

An

observer

view

on

synchronization

509

Henri J.C. Huijberts, Henk Nijmeijer

 

Regularity

of the

sub-Riemannian

distance

and

cut

locus

521

Sdbastien Jacquet

Industrial

sensorless control

of induction

motors

535

Fabrice Jadot, Philippe Martin, Pierre Rouchon

 

Feedback

invariants

and

critical trajectories;

Hamiltonian

formalism

for feedback

equivalence

 

545

Bronistaw Jakubczyk

 

Paths

in sub-Riemannian

geometry

 

569

Frgddric Jean

Observability

of C°°-systems for

Loo-single-inputs

 

575

Philippe Jouan

Robust

control of a

synchronous

power

generator

 

583

Matei Kelemen, Aired Francis Okou, Ouassima Akhrif, Louis-A. Dessaint

Volume 2

×l

Control of a reduced size model of US navy crane using only motor position sensors

1

Bdlint Kiss, Jean Ldvme, Philippe Mullhaupt

Algorithms

for identification

of continuous

time

nonlinear

systems:

a passivity

approach

13

Ioan D. Landau, B. D. O. Anderson,

F. De Bruyne

Flatness-based boundary control of a nonlinear parabolic equation modelling a tubular reactor

45

Alan F. Lynch, Joachim Rudolph

 

Dynamic feedback transformations of controllable linear time- varying systems

55

Franfois Malrait, Philippe Martin, Pierre Rouchon

Asymptotic controllability implies continuous-discrete time feedback stabilizability

63

Nicolas Marchand, Mazen Alamir

 

Stabilisation of nonlinear systems by discontinuous dynamic state feedback

81

Nicolas Marchand, Mazen Alamir, Iyad Balloul

On the

stabilization

of a class of uncertain

systems

by bounded

control

95

Lorenzo Marconi, Alberto Isidori

 

Adaptive

nonlinear

control

of synchronous

generators

107

Riccardo Marino, Gilney Datum, Franfoise Lamnabhi-Lagarrigue

Nonlinear

observers

of time

derivatives

from

noisy

measure-

ments

of periodic

signals

123

Riccardo Marino, Giovanni L. Santosuosso

Hamiltonian representation of distributed parameter systems with boundary energy flow

137

Bernhard M. Maschke, Arjan van der Schaft

×11

Differentiable Lyapunov function and

center manifold theory

.

143

Frdddric Mazenc

Controlling self-slmilar traffic and

shaping

techniques

149

Radl J. Mondrag6n C, David K. Arrowsmith, Jonathan Pitts

 

Diffusive representation for pseudo-differentlally damped nonlinear systems

 

163

Gdrard Montseny,

Jacques Audounet,

Denis Matignon

 

Euler's discretization and dynamic equivalence of Nonlinear Control Systems

 

183

Ewa Pawtuszewicz, Zbigniew Bartosiewicz

 

Singular systems in dimension 3: cuspidal case and tangent elliptic fiat case

 

193

Mich~le Pelletier

Flatness of nonlinear control systems and exterior differential systems

 

205

Paulo S~rgio Pereira da Silva

 

Motion planning for heavy chain systems

 

229

Nicolas Petit, Pierre Rouchon

Control of an industrial

polymerization reactor

using flatness . 237

Nicolas Petit, Pierre Rouchon, Jean-Michel Boueilh, Frgddric Gugrin, Philippe Pinvidic

Controllability of nonlinear multidimensional control systems . 245

Jean-Franfois Pommaret

Stabilization of a series DC motor

by dynamic output

Richard Pothin,

Claude H. Moog, X. Xia

feedback257

Stabilization

of nonlinear systems via Forwarding

rnod{LaV }

265

Laurent Praly, Romeo Ortega, Georgia I(aliora

 

A robust globally asymptotically stabilizing Feedback:

 

The example

of the

Artstein's circles

279

Christophe Prieur

Robust stabilization for the nonlinear benchmark problem (TORA) using neural nets and evolution strategies

Cesdreo Raim(mdez

Xlll

301

On

convexity in stabilization of nonlinear

systems

 

315

Anders Rantzer

 

Extended

Goursat

normal

form:

a geometric

characterization

. 323

Witold Respondek,

William Pasillas-Lgpine

 

Trajectory tracking for 7r-flat nonlinear delay systems with a motor example

339

Joachim Rudolph, Hugues Mounier

 

Neuro-genetic robust regulation design for nonlinear parameter dependent systems

 

353

Giovanni L. Santosuosso

 

systems

via high-order

 

Stability criteria for time-periodic averaging techniques

Andrey Saryehev

 

365

Control

of nonlinear

descriptor

systems,

a

computer

algebra

based approach

 

379

Kurt Schlachev , Andreas If ugi

 

Vibrational

control of singularly perturbed

systems

 

397

Klaus R. Schneider

 

Recent

advances

in output

regulation of nonlinear

systems

409

Andrea Serrani, Alberto Isidori, Cristopher I. Byrnes, Lorenzo Marconi

Sliding mode

control of the PPR

mobile robot

with

a flexible

joint

421

Hebertt Sira-Ram(rez

 

The

ISS philosophy as a unifying framework

for stability-like

 

behavior

 

443

Eduardo D. Sontag

 

Control

design of a crane for offshore lifting operations

469

Michael P. Spathopoulos, Dimosthenis Fragopoulos

×IV

Set-valued differentials and the maximum principle of optimal control

Hdctor J. Sussmann

Transforming

a single-input nonlinear

system

to

a strict

487

feedforward form

via feedback

 

527

Issa Amadou

Tall,

Witold Respondek

 

Extended

active-passive decomposition

of chaotic systems

543

Ranjan

Vepa

 

On

canonical decomposition

of nonlinear

dynamic

systems

555

Alexey N. Zhirabok

 

New

developments

in dynamical

adaptive

 

backstepping

control

 

565

Alan S.I. Zinober,

Julie C. Scarratt, Russell E. Mills,

 

All Jafari Koshkouei

 

List of participants

at

the

2nd

NCN

Workshop

 

623

Subanalyticity

Sub-Riemannian

of Distance

and

Geometry

Spheres

in

Andrei Agrachev 1 and Jean-Paul Gauthier 2

1 S.I.S.S.A., Via Beirut 2-4

34013 Trieste, Italy

and Steklov Mathematical Institute, ul. Gubkina 8 117966 Moscow, Russia

agrachev@sissa, it

Laboratoire d'analyse appliqu4e et optimisation Universit de Bourgogne D4partement de Math4matiques, B.P. 47870

21078 Dijon, France

gauthierQu-bourgogne,fr

1

Introduction

 

Let

M

be

a

C ~

Riemannian manifold, dimM

=

n.

A distribution on

M

is

a smooth linear subbundle A of the tangent bundle TM. The number

k

= dim Aq is the rank of the distribution. The restriction of the Riemannian

structure to A is a sub-Riemannian

We fix a point q0 6 M and study only admissible paths starting from this

q0. Sections of the linear bundle

point, i.e. meeting the initial condition q(0) =

structure.

A are smooth vector fields; we set

A={XEVecM:X(q)

6 Aq,

q e

M},

the space of sections of A. Iterated

flag

Lie brackets of the fields in /~ define a

Aqo C A2 qo Q

"'" C

Am qo

"'" C TqM

in the following way:

A m -: span{[X1, [X2, [.

qo

9

"1

Xm]

9

9

.](q0) : Xi

9 z~,

i

-= 1

~

9

"

"~

rn}.

A distribution A is bracket generating at q0 if A m :

If

of

Tqo M

for some m

>

0

qo

A is bracket generating, and if q0 and ql are close, the Riemannian length

the shortest path, connecting q0 to ql, is the sub-Riemannian distance or

Carnot--Caratheodory

distance between q0 and ql-

In

a given initial point

between q0 and q. p is a continuous function defined on a neighborhood of

q0. Moreover, p is H61der-continuous with

aq' = TqoM.

the

remainder of the

paper

we assume that

denote

by p(q)

A is bracket

generating at

q0. We

the

sub-Riemannian distance

the

HSlder exponent

~,1

where

2

Andrei Agrachev and Jean-Paul Gauthier

We study mainly the case of real-analytic M and A. The germ at q0 of a Riemannian distance is the square root of an analytic germ. This is not true for a sub-Riemannian distance function p. Moreover, p is never smooth in a punctured neighborhood of q0 (i.e. in a neighborhood without the pole q0)- It may happen that p is not even subanalytic. Main results of the paper concern subanalyticity properties of p in the case of a generic real-analytic A.

We prove that, generically, the germ of p at q0 is subanalytic if n < (k-1)kq-1

(Th. 5)and is not subanalytic ifn _> (k-1)(~ + ~ + l) (Th. 8). The bails

p-1 ([0, r]) of small enough radius are subanalytic if n > k > 3 (Th. 7). This statement about the balls is valid not only generically, but up to a set of distributions of codimension co.

In particular, if k > 3, n > (k-l)(~5 ~ + ~ + 1), then (generically!)the balls

p- 1([0, r]) are subanalytic but p is not! The results are just stated herein, and the complete proofs can be found in [16].

This paper is a new step in a rather long research line, see [1,5,6,9,10,12,13,15]. The main tools are the nilpotent approximation, Morse-type indices of geodesics, both in the normal and abnormal cases, and transversality tech- niques.

2

Nilpotentization,

endpoint

mapping

and

geodesics

Nilpotentization or nilpotent approximation is a fundamental operation in the geometric control theory and sub-Riemannian geometry; this is a real nonholonomic analog of the usual linearization (see [2,3,7,8]). Here we refer specially to [7].

We are working in a small neighborhood Oqo of q0 E M, where we fix an

orthonormal frame X1,

under consideration. Admissible paths are thus solutions of the Cauchy prob-

lem;

,Xk

E Vec Oqo of the sub-Riemannian structure

k

 

(t = Zui(t)Xi(q),

q E Oqo,

q(O) = qo,

 

(1)

 

i=1

where u = (ul(-), following equation :

,uk('))

E Lk[0, 1].

The nilpotentization

of (1)

has

the

k

=

x

=

0,

(i)

i=1

where X

:

Oqo

-+

lRn is

an

"adapted

coordinate

map",

and

)(i(x)

is

the

"nilpotentization" of Xi, i = l, 9 9 k. In adapted coordinates :

l<i<l,

x(qo)=O,

x, lr,0

(

qo)=~kl@

~k,

Subanalyticity of Distance and Spheres in S-R Geometry

3

dim(A~0/A~ol ) = k,, i =

1,

,

I.

We introduce a dilation dt : ll~n -+ 1W~, t E JR, by the formula:

Below Ilull =

~t(~1,~2,

(i,0

u~(t) dt

Ur

=

,

~)

= (t~l,t~2,

,t%).

is the

norm in Lk[0, 1]. Let

{u C L~[0, 1]:

Ilull -- r},

(2)

be the sphere of radius r in L~ [0, 1]. We consider the endpoint mapping f :

u ~ q(1). It is a well-defined smooth mapping of a neighborhood of the origin of Lk2[0, 1] into M. Clearly, p(q) = min{llull : u a L~[0, 1], f(u) = q}.

We define also

the mapping ]:

L~[0, 1] --+ ~n

by the

rule

]

: u(.) ~-~ x(1),

where x(.) = x(.; u) is the

solution of (i).

 

Proposition

1.

Let

X

=

(X1,

,Xt),

Xj

:

Oqo

--+ ~kj,

j

=

1,

,l

.

Then

the following identities holdfor any u(.) ~ L~[0, 1], c ~ ]~:

](u(.))

=

ui(t)Xix1 (qo) dt,

\0

i=1

f

f

O<_tl<

E

<t~<_lij=l

~,,(t~)

,

~,,(t~)2,

](eu(.)) = ~](u(-)),

where ~

is the dilation

o

(2).

o2,,x~(qo)dt~

dt~

;

We set f~(u)) = ~X(f(eu)). Then f~ is a smooth mapping from a neigh-

borhood of 0 in L~[O, 1] to 1R'~. Moreover, any bounded subset

contained in the domain of f~ for e small enough.

Theorem

of

L~ [0, 1] is

in

gence of the mappings and all their derivatives on the balls in Lk2[0, 1].

Recall that p(q) = min{[]u[[ : f(u) = q, u E L~[0, 1]} is the sub-Riemannian distance function. We set:

topology of the uniform conver-

1.

f~

~

]

as e

---4

0

the C ~

1

 

p~(x) = min{llull : f~(u) = x,

u

c

L~[O, 1]} = ~p(x-l(,~x))

and:

 

~(x) = min{[]u[[ : ](u)

=

x,

u E L~[O, 1]}.

Thus/~ is the sub-Riemannian distance for the nilpotentization of the original system.

4

Andrei Agrachev and Jean-Paul Gauthier

Lemma

KC~

~.

1.

Theorem

2.

The family of functions p~ Ig is equicontinuous for any compact

p, ----+~ uniformly on compact subsets of]~ n as ~

~ O.

The following proposition is a modification of a result by Jacquet

Proposition

[13].

2.

Let ~t~ =

{u E U,

: 3c~ E (0, 1] s.t.

au

is minimal for (1)}.

C

Then Mr

is a compact subset of the Hilbert sphere U,

and ](Mr

\ .Mr)

is a minimizing control for

system (i).

Subriemannian minimizers (geodesics) can be of two types: normal or ab- normal. Normal geodesics are projections re(t) on M of trajectories on T*M of the Hamiltonian vector field "-~ associated with the Hamiltonian

where hi is the Hamiltonian

h,

~-1 (r); in particular,

any element of Mr

\

Mr

k

89E

h(r

=

h~(r

Vq E M,

r

E TqM,

i=1

lift of Xi. Abnormal geodesics are solutions of:

k

= ~-~ui(t)hi(r

i=1

hi(r

=

0.

(3)

There is the following fundamental result:

Proposition 3.

(Goh-Legendre condition) If u(.) is a strictly abnormal min-

imizer,

it does exist

a r

satisfying (3) with:

 

{hi,hj}(C(t))

= O

Vi,j E {1,

,k},

(4)

 

k

k

~'~{hi,{hj,~-~uz(t)h,}}vivj

 

<0

V(vx,

,vk)

E]R k,

(5)

i,j=l

s=l

for almost

all

t

~

[0,1],

where {a,b}

=

ab

is the Poisson

bracket of the

Hamiltonians a, b.

[]

Remark. Identity (4) is called the Goh condition while Inequality (5) is the generalized Legendre condition. It is easy to see that both conditions are ac- tually intrinsic: Identity (4) doesn't depend on the choice of the orthonormal frame X1, Xk since hi(r i = 1, k, vanish anyway. Inequality (5) doesn't depend on the choice of the orthonormal frame provided that (4) is satisfied.

it

We say that u(.) is a Goh control if (4) is satisfied for an appropriate r is a Goh-Legendre control if both (4) and (5) are satisfied.

We come to the first subanalyticity result, which is a consequence of theorem 2 and of general facts about subanalytic sets:

Theorem

,

,

3.

If the germ of p at qo is subanalytic,

then ~ is subanalytic.

Subanalyticity of Distance and Spheres in S-R Geometry

5

3

Subanalyticity

and

nilpotentization

In this section, we assume that everything is analytic.

Theorem 4. If the nilpotent system (2] does not admit nonzero

Goh-

Legendre abnormal controls, then the germ of p at qo is subanalytic.

[]

The system (2) is said to be medium fat if:

TqoM = A~o + span{[X,

[Xi, Xj]](qo)

: i, j

=

1,

.

.

. , k }

for any X E z~, X(qo) r 0 (see [5]). Medium fat systems do not admit nontrivial Goh controls. It follows directly from the definitions that a system

is medium fat if and only if its nilpotentization is. We come to the following:

Corollary 1.

subanalytic.

If the system

(2)

is medium fat,

then

the germ

of p

at

qo

is

It

is proved in

[5] that

generic germs of distributions

are medium fat for

n

<

(k -

1)k + 1. This gives the following general result.

sub-

Riemannian distance function associated with a generic germ of a rank k

distribution on an n-dimensional real-analytic Riemannian analytic.

is sub-

Theorem

5.

Assume

that

n

<

(k

-

1)k +

1.

Then

the

germ

of the

manifold

4

Exclusivity

of

Goh

Controls

for

Rank

>

2

Distributions

First we'll make precise the term exclusivity. Rank k distributions on M are smooth sections of the "Grassmannization" HkTM of the tangent bundle TM. The space of sections is endowed with the C ~ Whitney topology and

is denoted by HkTM. Smooth families of distributions parametrized by the

manifold N are sections of the bundle pNHkTM over N x M induced by the standard projection pg : N x M --~ M. Let ,4 C HkTM be a set of distributions. We say that .4 has codimension c~ in HkTM if the subset:

 

{D E pNHkTM

: D

x•

~

A,

VX e N},

is

everywhere dense in pNHkTM,

VN.

We will also use a real-analytic version of the definition, just given. The only difference with the smooth case is that the manifolds and the sections are assumed to be real-analytic, while the topology remains the same Whitney topology.

6

Andrei Agrachev and Jean-Paul Gauthier

Theorem 6. For any k > 3, the distributions admitting nonzero Goh con- trols form a subset of codimension oo in the space of all smooth rank k dis- tributions on M.

It was proved in [1, Cot.4] that the small sub-Riemannian balls are subana-

lytic for any real-analytic sub-Riemannian structure without nontrivial Goh controls. Combining this fact with Theorem 6 we obtain the folowing result. Recall that all over the paper we keep the notation p(q), q E M, for the sub- Riemannian distance between q and the fixed point q0. The sub-Riemannian distance is defined by a given distribution ,4 on the Riemannian manifold M.

Theorem 7. Suppose that

subset ,4 of codimension oo in the space of rank k real-analytic distributions

on M such that the relation ,5 ~ .A implies the subanalyticity of the sub- Riemannian balls p-t ([0, r]) for all r, small enough.

M

is real-analytic and k

>_ 3.

There exists

a

5

Nilpotent

Systems

The system:

=

k

i=1

E

=

0,

is

adapted coordinates.

called

nilpotent if it coincides with its own nilpotentization

expressed

(6)

in

In other

words,

JR" is presented

as

a

direct

sum

~"

=

I~ k~ 0"- 9

 

]Rk',

kt

=

k,

so

that

any vector x

E

~n

takes

the

form x

=

(xt,

,xt),

xi

=

(xil,

,xik,)

E

]Rk~, i

=

1,

,1.

The

vector

fields

Y/,

i

=

1,

,k,

are

polynomial and quasi-homogeneous.

More precisely, they

are homogeneous

of

weight -1

with respect to the dilation:

 

6t:(xl,x2,

,xt)~(txt,t~x~,

,ttx~),

t

E ~;

Jr.Y/= tY/, i = 1,

We keep the notation ]

where x(.; u) is the solution of (6),

~: ~n _+ ]R+ for the sub-Riemannian distance, t5(x) = min{llul] : ](u) = x}.

is

called "the flat Martinet system". We will use the special notation pm: ~n --r JR+ for the sub-Riemannian distance in this case, which plays an important role below.

A special case of the system

: L~ [0, 1] --~ ]Rn for the endpoint mapping u ~-+ x(1; u),

u

(6) with

=

n

(ut(-),

=

l

=

3,

,

kl

uk(.)),

=

2,

and

ks

=

the notation

0,

k3

=

1,

Proposition 4. Assume that k = 2, k3 ~ O. Then there exists a polynomial

submersion ~: ]~n ~ IR3 such

that (pm))-t ([0, r]) = ~ (fi-l([O, r])), Vr _> O.

Subanalyticity of Distance and Spheres in S-R Geometry

7

Corollary

balls ~([0, r]), r > 0, are not subanalytic.

2.

Under the

conditions

of proposition

~

the

sub-Riemannian

Now consider nilpotent distributions of rank greater than 2, i.e. k = kl > 2. We restrict ourselves to the case of maximal possible ks, k3. It means

ks = min{n

-

k,

k(k 2-

1) }

k3 = min{n

k(k2+

1),

(k +

1)k(k3

1)}.

Remark.

the maximal possible growth vector and, in particular, the maximal possible

have

Generic germs of distributions

and

their

nilpotentizations

ks, k3.

Proposition

5.

Assume thatn

> (k-1)

(~

+

~

+

1)

and k~,k3 are max-

imal possible.

Then there exists a polynomial submersion 9 : ]Rn --~ ~3 such

that (pm))-i (r) = ,~) (/~-l(r)),

Vr >

0.

Corollary 3. Under the conditions of Proposition 5, the sub-Riemannian balls ~([0, r]), r > 0, are not sub-analytic.

Let now A be an arbitrary (not necessarily nilpotent) germ of a bracket generating distribution at q0 E M, and let p be the germ of the associated sub-Riemannian distance function. Combining corollaries 2, 3, and Theorem 3 we obtain the following:

Theorem

8.

Assume that either k = 2 and A 3

qo

~

dimM>

(k-l)

+-~-+1

A s

qo

)

Or

and the segment (k,dim Aq0z , &m"

p is not subanalytic. analytic.

Then

In particular, generic germs are such that p is not sub-

Aq0)3

of the growth vector is maximal.

Finally, combining Theorem surprising result.

8 with Theorem

7 we come to the following

asso-

ciated with a generic germ of real-analytic distribution of rank k >_ 3, on a

n-dimensional manifold, n > (k-l) (~ + ~ + 1). Then the balls p-l([0, r])

are subanalytic for all small enough r, but the function p is not subanalytic/

Corollary

4.

Let p be a germ of sub-Riemannian

distance function

8

Andrei Agrachev and Jean-Paul Gauthier

References

1. A. A. Agrachev, Compactness for sub-Riemannian length-minimizers and sub- analyticity. Rend. Semin. Mat. Torino, 1998, v.56

2. A. A. Agrachev, R. V. Gamkrelidze, A. V. Sarychev, Local invariants of smooth control systems. Acta Applicandae Mathematicae, 1989, v.14, 191-237.

3. A. A. Agrachev, A. V. Sarychev, Filtrations of a Lie algebra of vector fields and nilpotent approximation of control systems. Dokl. Akad. Nauk SSSR, 1987, v.295, 777-781; English transl, in Soviet Math. Dokl., 1988, v.36, 104-108.

4. A. A. Agrachev, A. V. Sarychev, Abnormal sub-Riemannian geodesics: Morse index and rigidity. Armales de l'Institut Henri Poincar~--Analyse non lin~aire, 1996, v.13, 635-690.

5. A. A. Agrachev, A. V. Sarychev, Sub-Riemannian metrics: minimality of ab- normal geodesics versus subanalyticity. J. ESAIM: Control, Optimisation and Calculus of Variations, 1999, v.4, 377-403.

6. A. A. Agrachev, B. Bonnard, M. Chyba, I. Kupka, Sub-Riemannian sphere in Martinet flat case. J. ESAIM: Control, Optimisation and Calculus of Variations, 1997, v.2, 377-448.

7. A. Bellai'che, The tangent space in sub-Riemannian geometry. In the book:

"Sub-Riemannian geometry", Birkhs 1996, 1-78.

8. R. M. Bianchini, G. Stefani, Graded approximations and controllability along a trajectory. SIAM J. Control Optim., 1990, v.28, 903-924.

9. B. Bormard, M. Chyba, M~thodes ggomgtriques et analytique pour gtudier l'application exponentiele, la sphere et le front d'onde en ggometrie SR dans le cas Martinet. J. ESAIM: Control, Optimisation and Calculus of Variations, submitted.

10. B. Bormard, G. Launay, E. Tr~lat, The transcendence we need to compute the sphere and the wave front in Martinet SR-geometry. Proceed. Int. Confer. Ded- icated to Pontryagin, Moscow, Sept.'98, to appear.

11. J.-P. Gauthier, I. Kupka, Observability for systems with more outputs than inputs and asymptotic observers. Mathem. Zeitshrift, 1996, v.223, 47-78.

12. Zhong Ge, Horizontal path space and Carnot-Caratheodory metric, Pacific J. Mathem., 1993, v.161,255-286.

13. S. Jacquet, Subanalyticity of the sub-Riemannian distance. 3. Dynamical and Control Systems, 1999, v.5

14. P. K. Rashevsky, About connecting two points of a completely nonholonomic space by admissible curve. Uch. Zapiski Ped. Inst. Libknechta, 1938, n.2, 83-94.

15. H. J. Sussmann, Optimal control and piecewise analyticity of the distance func- tion. In: A. Ioffe, S. Reich, Eds., Pitman Research Notes in Mathematics, Long- man Publishers, 1992, 298-310.

16. A. Agrachev, J-P. Gauthier, On Subanaliticity of Carnot-Caratheodory Dis- tances. March 2000. Submited to Annales de l'Institut Henri PoincarC

Principal Invariants of Jacobi Curves

Andrei Agrachev 1 and

Igor ZelenkJ

1 S.I.S.S.A. Via Beirut 2-4 34013 Trieste, Italy and Steldov Mathematical Institute ul. Gubkina 8 117966 Moscow, Russia

agrachev@sissa, it

2 Department of Mathematics, TechnJon-lsrael Institute of Technology Haifa 32000, Israel z igor%t echunix, t echnion, ac. il

Abstract. Jacobi curves are far going generalizations of the spaces of "Jacobi fields" along Riemannian geodesics. Actually, Jacobi curves are curves in the La- grange Grassmannians. Differential geometry of these curves provides basic feed- back or gauge invariants for a wide class of smooth control systems and geometric structures. In the present paper we mainly discuss two principal invariants: the generalized Ricci curvature, which is an invariant of the parametrized curve in the Lagrange Grassmannian providing the curve with a natural projective structure, and a fundamental form, which is a 4-order differential on the curve. This paper is a continuation of the works [1, 2], where Jacobi curves were defined, although it can be read independently.

1

Introduction

Suppose

dle to

transversal

face in TqM.

M

M.

q

is a smooth manifold and

Let

H

be

to TqM,

Let

~

and

~r --

a codimension

Vq E

be

d~

the

M;

then

E

M;

these

a

7r : T*M

--+ M

is the cotangent bun-

such

that

H

is

hypersur-

A o lr., then

are transversal

on T'M;

=

define

a

1 submanifold

Hq

=

H O TqM

in T*M

is a smooth

T~M,

in

H

~

and

canonical

standard

Liouville form on

line distribution

symplectic structure

O'IH is a corank 1 closed 2-form. The kernels of (qlH)x, A E H

to 7q'M,

E T'M,

be the

kernels form

characteristic 1-foliation C of H. Leaves of this foliation are characteristic curves of alH.

Suppose 7 is a segment of a characteristic curve and ON is a neighborhood of 3' such that N = OT/(CIo,) is a well-defined smooth manifold. The quo- tient manifold N is in fact a symplectic manifold endowed with a symplectic structure ~ induced by alH. Let r : O N --+ N be the canonical factoriza- tion; then r O Or), q E M, are Lagrangian submanifolds in N. Let

10

A. Agrachev, I. Zelenko

L(TTN ) be the

L(T~N) = {A C T~N : A L = A}, where D z = {e E T~N : ~(e,D) = 0},

VD C T.~N. Jacobi curve is the mapping

Lagrange Grassmannian of the symplectic space TTN , i.e.

A ~-~ r

A 9 7,

from 7 to L(T~N).

Jacobi curves are curves in the Lagrange Grassmannians. They are invariants

of the hypersurface H in the cotangent bundle. In particular, any differential

invariant of the curves in the Lagrange Grassmannian by the action of the

linear Symplectic Group produces a well-defined function on H.

Set W = T~N and note that the tangent space TAL(W) to the Lagrange Grassmannian at the point A can be naturally identified with the space of quadratic forms on the linear space A C W. Namely, take a curve A(t) 9 L(W) with A(0) = A. Given some vector l 9 A, take a curve l(.) in W such that l(t) 9 A(t) for all t and l(0) = I. Define the quadratic form qA(.)(l) &(dl(0),l). Using the fact that the spaces A(t) are Lagrangian, i.e. A(t) z = A(t), it is easy to see that the form qA(.)(l) depends only on dA(0). So, we have the map from TAL(W) to the space of quadratic forms on A. A simple counting of dimension shows that this mapping is a bijection.

=

Tangent vectors to the Jacobi curve "17 at a point JT(A),

A 9 7, are equivalent (under linear substitutions of variables in the correspon-

dent quadratic forms) to the "second fundamental form" of the hypersurface H~(x) C T~(x)M at the point A.

Proposition 1.

In particular, the velocity of J'r at A is a sign-definite quadratic form if and

only if the hypersurface nr(x) is strongly convex at A.

A similar construction can be done for a submanifold of codimension 2 in

T*M. In the codimension 2 case characteristic curves do not fill the whole submanifold; they are concentrated in the characteristic variety consisting of

the points, where the restriction of a to the submanifold is degenerate.

We are mainly interested in submanifolds that are dual objects to smooth control systems. Here we call a smooth control system any submanifold V C TM, transversal to fibers. Let Vq = V NTqM; The "dual" normal variety H 1 and abnormal variety H ~ are defined as follows:

HX=

H ~ =

[J {AETqM:3vE

qEM

U

qEM

e T;M

\

O:

Vq, (A,v)= 1, (A,T~Vq)=O},

=

=

0}.

These varieties are not, in general, smooth manifolds; they may have sin- gularities, which we do not discuss here. Anyway, one can obtain a lot of information on the original system just studying smooth parts of H 1, H ~

Principal lnvariants of Jacobi Curves

11

Characteristic curves of ~1~ (~1~) are associated with normal Cabnormal) extremals of the control system V. The corresponding Jacobi curves admit a

purely variational construction in terms of the original control system and in

a very general setting

One of the varieties H 1, H ~ can be empty. In particular, if Vq = OWq, where Wq is a convex set and 0 E intWq, then H ~ = I~. Moreover, in this case the Liouville form never vanishes on the tangent lines to the characteristic curves

by

of CrIH~, and any characteristic curve 7 has a canonical parametrization the rule C~,'Y) = 1.

If

previous case, H i --- Oand ~ vanishes on the tangent lines to the characteristic

curves of aIHo. The characteristic curves are actually unparametrized.

If Vq are compact, then

mension > 2 in all nontrivial cases.

The rank of the "second fundamental form" of the submanifolds H~ and Hq~

(singularities included), see [1, 2, 3].

subsets

Vq C TqM are conical, aVq --

Vq, Va

>

1 in

0, then, in contrast

T'M,

to the

H 1 has codimension

while H ~ has codi-

of TqM

at

any point

is no greater

than

dimVq.

Indeed,

let

A E

H~;

then

(T Vq)•

=

1, for some v

Vq. We have

+

+Rv)

C

So A belongs

to

an

affine subspace

of dimension

n

-

dim Vq -

1, which

is

contained in He.

For

A E

H~

3v

E Vq such that

A E (TvVq) •

CA, v)

-

O.

Then the afline subspace A + (T,~Vq + ~v) • is contained

Suppose that H 1 has codimension 1 in T*M and 7 is a characteristic curve

of aiH~. Then the

greater than dim V~(x) (see proposition 1). The same is true for the Jacobi

curves associated with characteristic curves of a]H0, if H ~ has codimension

2.

Dimension of Vq is the number of inputs or control parameters in the control system. Less inputs means more "nonholonomic constraints" on the system.

It happens that the rank of velocity of any Jacobi curve generated by the

velocity of the Jacobi curve A ~-~ Jr(A), A E 7, has rank no

in Hq~

system never exceeds the number of inputs.

2

Derivative

Curve

Let A be a Lagrangian subspace of W, i.e. A E L(W). For any w E A, the linear form #(., w) vanishes on A and thus defines a linear form on W/A. The nondegeneracy of ~ implies that the relation w ~-~ or(-, w), w E A, induces a canonical isomorphism A ~- (W/A)* and, by the conjugation, A* ~ W/A.

We set Ar = {F E L(W) : F n A = 0}, an open

L(W). Let Sym2(A) be the space of self-adjoint linear mappings from A* to A; this notation reflects the fact that Sym 2CA) is the space of quadratic forms

on A* that is the symmetric square of A. A~ possesses a canonical structure of an affine space over the linear space SyruP(A) = Sym2((W/A)*). Indeed, for any A E Ar and coset (w+A) E W/A, the intersection AA(w+A) of the

everywhere dense subset

of

12

A. Agrachev, I. Zelenko

linear subspace A and the affine subspace w + A in W consists of exactly one point. To a pair F, A E Am there corresponds a mapping (F- A) : W/A --+ A,

where

(F-

A)(w+

A) d~_fFA

(w+A)

-

An

(w +A).

It is easy to check that the identification W/A "-- A* makes (F - A) a self-

acljoint mapping from A* to A. Moreover, given A E A"~, the correspondence

F ~

of the affine space are obviously satisfied.

A ~" one obtains a canonical identification A _~ W/A = A*. In

particular, (1" -

last linear mapping we will use the notation

mapping has a much more straightforward description. Namely, the relations

W =

(A, F, A) : A --+ A. In fact, this

A) E Sym ~(A) turns into the mapping from A to A. For the

Sym 2(A) and the axioms

(F

-

A)

is

a

one-to-one mapping of Am onto

Fixing A E

A

$

A,

F

n A --- 0, imply that

T' is the graph of a linear mapping from

A

to A. Actually, it

is the

graph

of the

mapping

(A, F, A).

In

particular,

ker(A, F, A) =

A n

F.

If A N F

=

0, then

(m, F, A)

=

(A,/', A) -1.

Let us give coordinate assume that

representations

of the introduced objects. We may

W=~"e~t"

= {(x,y): x,y

~ ~m},

~'((xl,yl), (x~,y2)) = (xl,y2) -

(x~,yl),

A

=

IR" (~O,

,4 =

0~

]~'~.

a by S -1 .

The coordinates in A induce the identification of Sym~A with the space of symmetric m • m matrices. Ar is an affine subspace over Sym~A; we fix ,4 as

the matrix S, while the operator

Then any F symmetric m •

E ,4"~ takes

the form

F

--

{(x, Sx) (A, F, ,4)

:

x

E ]~n}, where

S

is

m matrix. The operator

: A -~ ,4 is represented

by the matrix

(A, F, A) is represented

the origin in this affine subspace and thus obtain a coordinatization of A~ by symmetric m • m matrices. In particular, the "point" F = {(x, Sx) : x E ]Rn}

in A ~" is represented

A subspace

det(S - So) r O. Let us pick coordinates {x} in F0 and fix ,4 as the origin in

the affine space F0~. In the induced coordinatization of F0~ the "point" F is -

represented

Let t ~-~ A(t) be a smooth curve in L(W). We say that the curve A(.) is ample

at v if Sk > 0 such that for any representative

by the matrix S -1.

{(x, S0x)

:

x

E ]Rn}

So) -1.

F0

=

is transversal

to

F

if and

only

if

by the matrix (S

A~ (.) of the

k-jet of A(.) at

r,

St

such that

Akr(t)n A(r)

=

0. The curve A(.) is called

ample if it is ample

at

any point.

We have given an intrinsic definition

of an

ample curve.

In coordinates

it

takes the following form: the curve t

~-~ {(x, St•

: x

E ~n}

is ample at

v

if

and only if the function t ~ det(St -

St)

has a root of finite order at v.

 

Assume that A(.) is ample at r. Then A(t) e A(r) ~ for all t from a punctured neighborhood of v. We obtain the curve t ~-+ A(t) E A(r) ~ in the affine

space A(r) ~" with the pole at r. Fixing vector function with values in Syrn 2(A)

and with the pole at v. Such a vector

an "origin" in A(r) ~" we make A(-) a

Principal Invariants of Jacobi Curves

13

function admits the expansion in the Laurent series at v. Obviously, only free term in the Laurent expansion depends on the choice of the "origin" we did

to identify the affine space with the linear one. More precisely, the addition of

a vector to the "origin" results in the addition of the same vector to the free term in the Laurent expansion. In other words, for the Laurent expansion of

a curve in an affine space, the free term of the expansion is a point of this

afiine space while all other terms are elements of the corresponding linear space. In particular,

A(t) ~ Ao(r)

+

~

(t -

r)iQi(r),

(1)

i#o

where Ao(r) E A(r) ~, Qi(r) e Sym2A(r).

Assume that the curve A(.) is ample. Then Ao(r) E A(r) ~ is defined for all r. The curve r ,-+ A0(v) is called the derivative curve of A(.).

Another characterization of Ao(r) can be done in terms of the curves t ~-~ (A, A(t), A(r)) in the linear space Horn(A, A(r)), A E A(r):". These curves have poles at r. The Laurent expansion at t = r of the vector function

t

,-+ (A, A(t), A(r))

has zero free term

if and

only if A

=

Ao(r).

The coordinate version of the series

matrix-valued function t ~-+ (St - x E I~n}.

(2.1)

at

ST) -1

is the

Laurent

expansion

of the

t

=

r, where A(t) = {(x,Stx)

:

3

Curvature

operator

and

regular

curves.

Using derivative curve one can construct an operator invariant of the curve A(t) at any its point. Namely, take velocities A(t) and A0(t) of A(t) and its derivative curve Ao(t). Note that A(t) is linear operator from A(t) to A(t)* and Ji0(t) is linear operator from Ao(t) to Ao(t)*. Since the form ~ defines the canonical isomorphism between Ao(t) and A(t)*, the following operator

R(t) : A(t) -+ A(t) can be defined:

R(t) = -Ao(t)

o A(t)

(2)

This operator is called curvature operator of A at t. Note that in the case of Riemannian geometry the operator R(t) is similar to the so-called Ricci operator v --+ RV(~/(t), v);/(t), which appears in the classical Jacobi equa- tion XT~(t)XT-~(t)V+ RV(~/(t), V)Zy(t) -" 0 for Jacobi vector fields V along the geodesic 7(t) (here R v is curvature tensor of Levi-Civita connection ~7), see "-"

The curvature operator can be effectively used in the case of so-called regular curves. The curve A(t) in Lagrange Grassmannian is called regular, if the

[1]. This is the reason for the sign

in (2).

14

A. Agrachev, I. Zelenko

quadratic form Ji(t) is nondegenerated for all t. Suppose that the curve A(-) is regular and has a coordinate representation A(t) = {(x, Stx) : x 9 ~n}, ST = 0. Then the function t ~-+ S~-1 has a simple pole at t =- v and one can get the following formula for the curvature operator (see [1]):

R(t) =

((2s;),-i s,,,')

-

((2s;),-i

,,)2

(3)

Note that the right-hand side of (3) is a matrix analog of so-called Schwarz derivative or Schwarzian . Let us recall that the differential operator:

1

"

d

~o"

"

~o"

.2.

1 ~(3)

1

"WIt" 2

acting on scalar function ~ is called Schwarzian. The operator ~ is character- ized by the following remarkable property: General solution of the equation S~ = p w.r.t ~ is a MSbius transformation (with constant coefficients) of some particular solution of this equation. The matrix analog of this operator has similar property, concerning "matrix MSbius transformation" of the type (AS + B)(CS + D)-1. It implies that in the regular case the curvature opera- tor R(t) determines the curve completely up to a symplectic transformation.

4

Expansion

of the

cross-ratio

and

Ricci

curvature.

For the nonregular curve A(t) = {(x, Stx) : x E I~n}, the

function t

,-+

(St - ST) -1 has a pole of order greater than

1

at

r

and it

is much more

difficult to compute its Laurent expansion. In particular, as we will see later in the nonregular case the curvature operator does not determine the curve up to a symplectic transformation. However, using the notion of cross-ratio it is possible to construct numerical invariants for a very nonrestrictive class of curves.

Suppose that Ao, A1, A2, and A3 are Lagrangian subspaces of W and Ao and

A1 N A2 = A3 CIAo = 0. We have

A3 in

the Lagrange Grassmannian is, by definition, the following linear operator in

A2:

Ao. The cross-ratio

(Ao, A1, A2) : Ao --+ A2, (A2, A3, Ao) : A2 --+

of four "points"

Ao, A1, A2, and

IAo,A1,A2,A3J

L

J

[Ao, AI,A2,A3]= (Ao,A1,A2)(A2, Aa, Ao).

(5)

This notion is a "matrix" analog of the classical cross-ratio of four points in the projective line. Indeed, let Ai = {(x, Six) : x E ~n}, then, in coordinates {x}, the cross-ratio takes the form:

[Ao, At,A2,A3]= ($2 - S1)-I(S1-

So)(So -

$3)-1 ($3-

$2)

(6)

Principal Invaxiants of Jacobi Curves

15

By construction, all coefficients of the characteristic polynomial of

A0, A1, As, AaI are

invariants of four subspaces A0, A1, A2, and Aa.

Now we are going to show how to use the cross-ratio in order to construct invariants of the curve A(t) in the Lagrange Grassmannian. Let, as before,

x E ~n} be the coordinate representation of a germ of the

t ,-~ {(x, Stx):

curve A(.).

Assumption

1 For all parameters

tl

the functions

have

at t = tl

zero of the same finite order k.

t

--~ det(St

-

Stl)

By the above the function (to, tl, t2, t3) --r det [A(t0), a(tl), A(t2), A(t3)] is symplectic invariant of the curve A(t). Using this fact, let us try to find sym- plectic invariants of A(t) that are functions of t. For this it is very convenient to introduce the following function

det[A(to)'A(tl)'A(t2)'A(t3)].

G(to,tx,tz,tz)

=

In

[to,tl,t2,t3]k

,

(7)

where Lt~

=

(tl_t~)(ta_to)

to, tl, t2,and tz. The function G(t0, tl, t2, t3) is also a symplectic invariant of A(t). It can be easily expanded in formal Taylor series at any "diagonal" point (t,t, t, t) and the coefficients of this expansion are invariants of the germ of A(.) at t.

Indeed, by Assumption 1, we have:

det(S, o-St1)

= (t0--tl)kx(t0,tl),

X(t,t)

~s 0

(8)

for any t. It follows that X(to, tl) is a symmetric function (changing the order in (8) we obtain that X can be symmetric or antisymmetric, but the last case

is impossible by the fact that X(t, t) # 0 ). Define also the following function

f(to,tl)

= lnlX(to,tl)[

(9)

This function is also symmetric, so it can be expanded in the formal Taylor series at the point (t, t) in the following way:

f(to,tl)

~

Z

i,j=O

ai,j(t)(to -

t)i(tl -

t)j,

ai,/(t) = aj,i(t)

(10)

One can easily obtain the following lemma, using the fact that

G(to,tl,t~,ta)

= f(to,tl)

-

f(tl,t2)

+ f(t2,t3)

-

f(ta,to)

16

A. Agrachev, I. Zelenko

Lemma

1.

For

Taylor expansion

any

t

the function

6(to,tl,t2,t3)

at the point

(t , t , t , t ) :

O0

has

the following formal

G(to,tl,t2,t3)

~

~

aih(t)(~

-

~)(~

-

~),

(11)

 

i,j=t

whereat = tt -t,

I = 0,1,2,3.

From (11) it follows that all coefficients his(t), i,j > 1, are symplectic in- variants of the curve A(t).

Definition

1.

The first appearing coefficient a1,1 (t) is called Ricci curvature

of A(t).

by a direct

computation, one can get the following relation between p(r) and curvature operator R(r) for the regular curve: p(r) = 89 R(r). Actually, this relation justifies the name Ricci curvature for the invariant p(t). In some cases we are interested in symplectic invariants of unparametrized curves in Lagrange Grassmannian (i.e., of one-dimensional submanifolds in Lagrange Grassmannian). For example, so-called abnormal extremals of vec- tor distributions and consequently their 3acobi curves a priori have no special parametrizations.

Now we want to show how, using the Ricci curvature, one can define a canon- ical projective structure on the unparametrized curve A(.). For this let us check how the Ricci curvature is transformed by a reparametrization of the

In the sequel

the Ricci curvature

is denoted

by p(t). Note that,

curve A(t).

Let

the function playing for A(7) the same role as the function ~ for A(t). Then

from (7) it follows that

t

=

~(r)

be a reparametrization

and let A(r)

=

A(~(r)).

Denote

by

(70,

7"1,7"2,7-3) = ~(to, tl,t2,t3) +k

In ([~o(ro), ~o(rl), ~o(r~), ~O(T3)]) ,(12)

where ti = ~(7i), i = 0, 1,2, 3. By direct computation it can be shown that the

has the following Taylor

function

expansion up to the order two at the point

(T0, vl, v2, v3) ~-~ In ([~(r0),~(r,),~(r~),~(r~)]~

k

[~o,~,~,~l

] (v, 7, 7, v):

In ([qo(70),~(71),~(Tz), ~(73)]'~ = ~(7)(r/o

\

[70,

T2,

/

-

r/2)(r/t -

y3) +.--

(13)

where S~ is Schwarzian defined by (4) and T/i = T/-

T,

i

=

0, 1,2,

3.

Suppose for simplicity that in the original parameter t the Ricci curvature p(t) = 0 and denote by/5(v) the Ricci curvature of the curve A(7). Then from (11), (12), and (13) it follows easily that:

(14)

fi(v) = k~(r).

Principal Invariants of Jacobi Curves

17

Conversely, if the Ricci curvature p(t) of the curve A(t) is not identically zero we can find at least locally (i.e., in a neighbourhood of given point) a reparametrization t = ~o(r) such that fi(r) = 0 (from (14) it follows that in

this case ~(r)

The set of all parametrization of A(.) with Ricci curvature identically equal to zero defines a projective structure on A(.) (any two parametrization from this set are transformed one to another by MSbius transformation). We call it the canonical projective structure of the curve A(.). The parameters of the canonical projective structure will be called projective parameters.

has to satisfy the equation S(~o-1)(t) = P (~2)

5

Fundamental

form

of

the

unparametrized

curve.

The Ricci curvature p(.) is the first coefficient in the Taylor expansion of the function G at the point (t, t, t, t). The analysis of the next terms of this expan- sion gives the way to find other invariants of the curve A(-) that do not de- pend on p(t). In this section we show how to find candidates for the "second" invariant of A(-) and then we construct a special form on unparametrized curve A(.) (namely, the differential of order four on A(.) ), which we call the fundamental form of the curve A(.).

First

analyzing the expansion (10) one can easily obtain the fol-

lowing lemma

note that

Lemma

holds

2.

Let oti,j(t)

be as in expansion

(10).

Then

the following

relation

a~j(t)

=- (i + 1)ai+l,j(t)

+ (j + 1)(~ij+l(t)

(15)

In particular, from (15) it follows easily that

=

1,

(t),

2,2(t)

=

1,,

(t)

-

3

(16)

These relations imply that the function a3,1 (t) is a candidate for the second invariant (as well as the function a2,~(t)). Now let t be a projective parameter on A(.). Then by definition p(t) -- O, and by (16) o/2j(t) ~ 0 and ot2,2(t) = -30~3,t(t). This together with (11) and the fact that a3,1(t) = crl,3(t) implies that the function g(to,t~, t2,t3) has the following Taylor expansion up to the order four at the point (t, t, t, t):

G(to, tl, t2, t3) = cr3,1(t)p4 (~0, ~1, ~2, ~3) +--.,

where ~i

nomial of degree four (more precisely, P4(~0, (1, ~, ~3) -

:

ti

-

t,

i

:

3

O, 1,2, 3, and/94 (~0, ~l, ~2, ~3) is

-

-

(17)

a homogeneous poly-

(~3 _ ~3)(~1

_

~3) +

18

A. Agrachev, I. Zelenko

and

Let 7- be another projective parameter on A(.) (i.e., t = ~(v) =

denote by ds,l(r) the function that plays the same role for the curve A(~(v)) as the function a3,1(t) for A(t). Then from (12), (17), and the fact that the cross-ratio is preserved by Mhbius transformations it follows that

=r+d at+b)

4 = as,l(t)(dt)'

(18)

It means that the form ~3,1(t)(dt) 4 does not depend on the choice of the projective parameter t. We will call this form a fundamental form of the curve A(.).

If t is an arbitrary (not necessarily projective) parameter on the curve A(-),

then the fundamental form in this parameter has to be of the form A(t)(dt) 4, where A(t) is a smooth function (the "density" of the fundamental form). For projective parameter A(t) = az, l(t). For arbitrary parameter it can be shown, using (11), (12), that A(t) = az,l(t) - 1 p(t) 2 -

If A(t) does not change sign, then the canonical length element IA(t)1 88 is

defined on A(-). The corresponding parameter 7- (i.e., length with respect to this length element) is called a normal parameter (in particular, it implies that abnormal extremals of vector distribution may have canonical (normal) parametrization). Calculating the mcci curvature Pn(7-) of A(.) in the normal parameter, we obtain a functional invariant of the unparametrized curve. We will call it projective curvature of the unparametrized curve A(.). If t = ~(v)

is the transition function between a projective parameter t and the normal

parameter v, then by (14) it follows that pn(7-) = k S~(7-). Note that all constructions of this section can be done for the curve in the Grassmannian G(m, 2m) ( the set of all m-dimensional subspaces in the 2m- dimensional linear space) instead of Lagrange Grassmannian by the action of the group GL(2m) instead of Symplectic Group.

6

The

method

of

moving

frame.

In this section we consider nonregular curves having two functional invariants and prove that the above defined invariants p and A constitute a complete system of symplectic invariants for these curves (completeness of the sys- tem of invariants means that the system defines the curve uniquely up to a symplectic transformation)

Assume that dimension of the symplectic space W is four and consider ample curves A(t) in L(W) such that for any t the velocity Ji(t) is a quadratic form of rank 1. Without loss of generality we can assume that Ji(t) is nonnegative definite for any t. Let us fix some parameter r and let {(x, Stx): x e A n} be

a coordinate representation of A(t) such that Sr = 0. Since the curve A(t) is

ample, the curve St I has a pole at r. The velocity ~S;a-1 : A(r)* --+A(r) is

a

well defined self-adjoint operator. Moreover, by our assumptions, ~S t is

d

-1

Principal Invariants of Jacobi Curves

19

a nonpositive self-adjoint operator of rank 1. So for t ~ 7- there exists unique,

up to the sign, vector w(t, r) E A(r) such

that for any

v E m(r)"

 

d _,

,)

=

(19)

It

is clear

that

the

curve

t

--4 w(t, 7-) also

has

the

pole at

7-. Suppose

that

the order

of the pole is equal

to I. Denote

by

u(t, v)

the

normalized curve

t --+u(t, 7-) = (t-

v)tw(t, 7-) and define the following vectors in A(v): ej(r)

---

Otj-'Tj~'-U~%

0j-a

t4 7-)]t=r.

It is not hard to show that

to I if and only if I is the minimal positive integer such that the vectors el (7-)

and et(r) are linear independent (in particular, el(r) and e~(r) are linear independent if and only if I = 2). It implies easily that the set of points 7-, where the vectors el(r) and e2(r) are linear dependent, is a set of isolated points in IR.

the order

of pole

of t --+ w(t, r)

at

t

=

r

is equal

4 such that for

anyt the velocity A(t) is a quadratic form of rank 1 andet(t), e2(t) are linear independent.

Assumption

2

A(t)

is a curve

in L(W)

with dim W =

By the above it is easy to see that if A(.) satisfies the previous assumption,

4. So, the invariants p(-) and A(-)

are defined for A(.). Note that the curve .4(.) can be described by the curve

then it satisfies

Assumption

1 with

k

=

t --~ w(t, r)

of

the vectors on the plane, i.e. A(.) can be described by two

functions. The natural question is whether (p(.), A(.)) is a complete system

of symplectic invariants of A(-).

Since vector w(t, r) is defined up to the sign, the vector el(r) is also defined up to the sign. So, for any 7- one can take (el(r), e2(r)) or (-el(r),-e2(r)) as the canonical bases on the plane A(r). Recall that by constructions of the section 2 for the curve A(-) the derivative curve A0(') is defined and for any 7- the subspaces A(r) and A0(v) are transversal. So, in addition to the vectors el(r), e2(r) on the plane A(r), one can choose two vectors fl(r) and f2(r) on

the plane Ao(r) such that four vectors (el(r), e2(t), fl(r), f2(r)) constitute

]

symplectic basis (or Darboux basis) of W (it means that cr(fi(r),ej(r)) =

8i,j). So, the curve .4(.) defines

and one can derive the structural equation for this frame:

a moving frame

(el(r),e2(r),fl(r),f2(r))

20

A. Agrachev, I. Zelenko

Proposition

2.

The frame (el (r), e2(r), fi (r), f2(r))

structural equation:

l ea =

e2=

3e2

1

~pel + 4f2

l i,

35

= -(nA-

1_2_

2

---i~P el -

7

!

9

~pe2 -

p")el -

3fi

5p'e

-

88

satisfies the following

(20)

Note that the coefficients in the equation (20) depend only on p and A and any symplectic basis can be taken as an initial condition of (20). It implies the following:

Theorem 1. The curve A(.) satisfying Assumption 2 is determined by its invariants (p(-), A(.)) uniquely up to the symplectic transformation of W.

Remark 1. It can be shown by a direct calculation that the curvature op- erator R(r) : A(r) -+ A(r) of the curve A(-) satisfying Assumption 2 has

the f~176

the basis (ei(r)'e2(r)): R(r) = (~

88

,i.e.,R

depends only on p. This means that in contrast to the regular case, the curva- ture operator does not determine the curve A(-) uniquely up to a symplectic transformation.

Theorem

1 implies the following result on unparametrized

curves:

Theorem 2. Assume that the curve A(.) satisfies Assumption 2 for some

A(t)(dt) 4 does not vanish. Then

the sign of A(t) and the projective curvature Pn(') determine A(.) uniquely up to a symplectic transformation of W and a reparametrization.

parametrization and its fundamental form

7

Flat

curves.

The following definition is natural.

Definition

O, A(t) -

O.

2.

The curve A(t), satisfying Assumption 2, is called fiat if p(t) =_

As a consequence of Theorem 1, expansion (11), and structural equation (20) one can obtain the following characterization of the fiat curve:

Principal Invariants of Jacobi Curves

21

Theorem

3.

The curve A(t), satisfying Assumption 2, is flat if and only if

one of the following condition holds:

I) all coe~cients Qi(t) with i > 0 in the Laurent expansion (1) are equal to

zero;

2) the derivative curve An(t) is constant, i.e., Ji0(t) = 0;

3) for

any to, tl, t2, t3

det ([A(t0), A(tl), A(t2), A(t3)])

=

[to, tl, t2, t3]4.

(21)