f;f»f;of,lJ.tfl :bfl!J.Jl'IlJ.

fJJl JltUSJlSfl}~S
EPEBAHCKHF! (JJH3WIECKHF! _ HHCTHTYT

EXPM-613 ( 3) -83

G.K.SAVVIDY

CLASSICAL AND QUANTUM MECHANICS

OF NON-ABELIAN GAUGE FIELDS

l; f» t; of. fl. t 1983 EPEBAH

 EliV!-6 I3(3)-8 3

r.K.CABB!W!
KJIACCV!'JECKAff VI KBAHTOBAi! MEXAID!KA HEAEEl!ElltiX
KAJll!EPOBO'lllliX IIOJIEi'!

B pa6oTe Kccne~TCH Knaccz~ecKaH H KBaHTOBaH MexaHHKa .He-

a6eJieB Hx !OU!t6p oBO'IIIb !X noJieil RaJ< 6ea cnbHTa HHoro Hapyme !fKH,
TaR " co cnoHTaHHI>M HapymeHKeM CHMM9Tp!<H. PaccM
aTplma eTM WYH-
~a!JeHTaJibHaH ITO~CHCT9Ma KnaCC~eCKOfi
MexaHHKH HHra-M HJIJica
(KMilM ). lloi<aaa Ho, 'ITO WYH.lli iMeHTa Ju.i!M rro~CH
CTeMa KM'lM HBJIHe.Tc.<;;
K - CHCTeMofi KOJIMOropoBa H CJie~oBa ':9Jil>HO o6Jialr;
leT Cl!JibHI<MII
CTaTHC~ecKHMH cBOnCTBa~. HEl~neHH T~e HHTerp
opyeMH e ITOHCHC,
TeMH 1 K KOTOj)liM 6JIH3K a\B CM!iCJie KAM - T9Cp!<H)Knacc~eCKaH Mexa~
l!liRa iiHra-l> liWIJIC a-XHrrc a (RMilMX). 06cylll!laRJTM KB!IIITO BOMex allll'!eC "f
KH9 CBOllCTBa K - CHCT9M.

EpeBaH CKHn qH31'!~cKHn HHCTHTyT

EpeBaH I963
 E~i1-6I8(3)-S3

G.K.SAVVIDY

CLASSICAL AND QUANTUhl MECHANICS

OF NOII-ABELIAIT GAUGE FIELDS

Classical and quantum mechanics of nort-Abelia.n gauge

fields is investigated both with and with~ut spontaneous sym-
metry breaking. The jundamental subsystem (FS} of the Yang-
Mills classical mechanics (YMClf) is considered. It is shown to
be a Kolmogorov K-system and, hence, to have strong statisti-
cal properties. !ntegrable sYstems ·are jllso found, to which in
terms of KAM-theory the Yang-Mills-Hig gs classical mechanics
(YMHCM) is close. Quantum-mech$.nicnl properties of K-system
and their relatio~: to the problem of confinement are discussod.

Yerevan Physics Institute

Yerevo.n 1983
Y-EREVAN PHYSICS INSTITU TE

Ri?ll-<ii3 (3 )-<33

G.K.SAVVIDY

CLASSICAL AND QUANTUM MECHANICS

OF NOll-ABELIAN GAUGE FIELDS

Yerevan 1983
@ EpeeaHCKuiJ. tfluBu .. euuiJ ·uHcTuTyr, 1983r.
 To my fethe r' s memory

1. Introd uction

In refs. [1 ,2,3]
author s have inves tigate d cla.ssi'cP.:.
equati ons of motion of non-A belian gaug9 fields ,
prope rties of
both separa te soluti ons and the system as a whole
. In partic u-
~ar, they have consid ered the soluti
ons which in a c~rtain co-
o.
ordina te system depend ed on the time ( P,JI- (i.) ) onlY- It has
.
been shown that the system descri bed by such potent
i~ls
(AJU ( t) ) is reduce d to the. discre te nonlin ear mecha nical sys-
tem (YMCM) with the 'Hemi ltonia n(2]

t ~ Q. a ~e. [a. a a. a
HYM=.t "E (A;) •~; (A, fl,)- (fi,AJ
Q. Q.

) J
and coupli ng equati ons

n.. s c'"'c 11~ A~ ( 1 .2)

w;ttere i.,j. a.,~.c =·1,2,1 3 are ror the SU(2) group.
K. = 1,2,3,

nonlin ear plane waves P.JU

. (K·X)
Soluti ons of this system in an arbitr ary coordL nate
. system are
with a non-ze ro square of the
wave vecto r Ka=jJ?- ( 2 , 3].
The gener al pictur e of the variat iori of the color ampli -
tudes ~~~ in time is chara cteriz ed by altern ate rapid
'
3

\
 r ampl itude s and grow th
osci llati ons and decre ase in so~e colo
11 2
i:n o·theJ .'a; colo r ''bea ts ( 1.
syste m
The stron g :ixJ.a tabili ty of traje ctor ies of this
al cond ition s in
with ·resp ect to smal l varia tions at' initi
[2, 1] to a conc lusio n
phas e space he.s led the auth ors of ref.
ity of YMCM. Char acte-
on the stoch e.sti city and nani nteg rabil
set of perio dica l tra-
risti call y it has at leas t a coun table
jecto ries. l21.
In the last pape r of these auth ors
L1) analo gous prop er-
a spon taneo usly brok en
ties of clas sica l gaug e syste ms with
e syste m with isod?uble~
symm etry are inve stiga ted. Such a gaug
break ing of SU{2) group in the gaug e
A<>.• ~ 0 is desc ribed b) •
tho l!alni l toni an \:11

(1.))

(1.4)

is the vacuum expe ctati on valu e of the

scal ar
whor e 'l
field
( 1. 5)

\j1 shou ld coin cide with tfva. c at large

\XI
Since the field
then this impl ies for unifo rm field s
that 6 ~ f>a. : 0

4
In J
[1 a two-dimension al case was investigated when fl~ =. XJ

~~~ " ~ and all the other components nLl~~ are zero
9' • •
H~~(x'+~')• c_'L (x'•Y~)·~(xy)' ( 1. 6)

This system is characterized by one parameter

( 1. 7)

it being shown in (11 that at~~ 0.15 there occurs a phase

transition in the 'following_ sense: _at large values of
9f the
system is close to the integrable one, whereas at ~-< 5fc
the motion is stochastic, just as for 9T = 0 ( ~ F. 0, 1. = 0),
This phenomenon is well illustrated by figs. 1-3 ( g[ = 4.84,
1
5f2 = 0.35f Sfc. ~ 0.15) from ref. (1}, which present. cross
sections of phase trajectories in the four-dimension al space
(X, X, Y, Y ) with a plane ci. = ( y, y) at X (d.) > 0 •
The purpose of the present paper is the further investi-
gation of the gauge systems (1.1) and (1.J). Let's first note
that in refs. [1 ,2 ,3] such two- and three-dimensio nal subsys-.
terns of (1.1) and (1.3) are considered, when the coupling
equations (1.2) and (1.4) are id~ntically satisfied. It ts the
case when the matrix fl ~ is diagonal in indices Q. 1 i. • It
<t·.

has seemed important to consider the general case when all

'Q... !
the components of the matrix ~ A
are not zero and coupling
equations are taken into account. In a recent pawer by two
authors ll 1
7 new compact variables *
have been introduced,

• It is implied that their variation range lies on the com-

pact manifold, in this case on a sphere (see sec. 2).
where the general system looks much simpler S.nd has a natural
analogy with the sclid body mechanics. In the second section
the necessary information from ref. [17] is presented and at
the same time a method of describing YMCM by means of compact
variables is developed. The concept of FS (2.12) of YMCM in-
ducing ~tochastic properties of a complete system is intro-
duced.
In section 3 the YMCM is considered from the viewpoint
of ergodic theory. The need in such an investigation followed
from refs. [1 ,2] .. As has been mentioned, it has been proven
in these papers that trajectories of FS (2~12) have a strong
instability which leads to the fact that the phase trajectory
chaotically fills up the energy surface E const. It is in-
tuitively clear what is implied by chaotic or stochastic mo-
tiont but it is important to define these concepts more stric·~..­

ly.
In the ergodic theory, or otherwise, the theory of dyna-
mic systems (DS) the classification of motions as the increase
of chaotic-stochastic properties is obtained: they are ergo-
dic systems, systems with weak mixing, with mixing, with n-
fold mixing and, finally, K-aystems. It is important to estab-
lish which class of DS the Y:MCM belongs to. It is also
clear that if the arbitrary trajectory unifoxmly fills up the
hypersurface E ~ const, then there is no sense in searching
for concrete solutions of the system. InStead one should in-
vesti~te the global properties of DS as a whole. It turns
out that if it is establiShed which class of DS the YMCM be-
longs to, .then it will be possible to find out these global

b
 prop ertie s.
Sect ion 3 is pla..nned as follo ws.
Firs t, the nece sHar y
info rma tion from the. clas sifi cati
on of DS is pres ent ado Then
we cons ider the suff icie nt cond
ition s sati sfyi pg whic h one
may answ er the ques tion : whi.ch clas
s of DS the syste m with the
give n Ham ilton ian H belo ngs to ?
In pape rs by Hada mard , Hed-
lund , Hop.f, Kryl ov, Anos ov, ·Sin
ai et al. [9- 12] fair ly gene ral
crit eria for the abov e ques tion
were obta ined .
Usin g thes e suff icie nt condi~ions
and the conc ept of
stru ctur al stab ility in the narr
ow sens e (see sec. 3, . p ~ 2 4 )
one may prov e that the YMCM is equi
vale nt (hou eom orph ic) to
the bill iard s syste m of Kry lov- Sina
i with hyp erbo lic wall s
(fig . 4) and that it is a Kolm 6gor
ov K-sy stem [a]. Phy sica lly
the stro ng stat isti cal prop ertie
s of the YMCM are due to the
inst abil ity aris ing at the scat teri
ng of phas e traj ecto ry on
the conv ex insi de hype rbol ic boun
dary of fig. 4· In this cas~
the conv ex insi de boun dary play a
the scat teri r.g role of nega -
tive curv atur e (see App endi x B).
Afte r this fact bein g esta blis hed
, all the glob al propeJ>-
ties of K-sy stem s are assi gned . to
the YMCM. The YMCM poss esse s
expo nent ially com pres sing and exte
ndin g tran sver se . fibe ring s
equi vale nt to appr opri ate fibe
ring s of the hyp erbo lic bill i-
ards , posi tive entr opy of Kol~ogor
~v [a], the stro nge st stat is-
tica l prop ertie s, K-m ixing . Fina
lly, ther e is one prop erty
whic h we woul d like to espe cial ly
dwe ll on. In 1931 Koopman [7 J
obta ined a bril lian t resu lt indi
cati ng that to each clas sica l
DS ·corr espo nds a one- para metr ic
grou p of unit ary oper ator s
whos e spec tral prop ertie s char acte
rize the ·DS~ Si:rice it has
 prove11 b:y KolmofSOl'OV that K-aystc ms have u. countab
ly-
bc~S:n

!rlul-.'. t tJ.l<:$ Lehosc::..~f~ .<J_ptJC ~rum, i ~ turns out that the group of

1~.:-:·i t:,_ry operato rs correap owiing to the YMCM has a

un:l.form countablY··multipl~ Lebesgu e spectrw n.

'l'lws ~ ~he YMC.'M .( s uot only a noninte grable system but

are
n1so po::w8ss es the ::;tronge st statist ical proper ties which

del>cri-9;..·'·' 'lY.'! concept s of the ergodic theory~

}.n section 4 GH-4Ca ::.1:ys.tems vn th spontan eous sym.'Uetry

breakill g are investigt.~.ted~ It i.s shovm b·y introdu cing conve-

n:l..eut variabi es of the type angle-a ctio.n ·that at !Jf >> 5\ C.

the syst8m (1.6) is close to the integra ble system (in terms
of KAJ.'i.-theory) Vlith the Hamilto nian {4.20) whose phase picture
looks like 1. This result is of interes t because the per-
fig~

turbaU on Hnmil tertian H1 (4~3) consist s of five terms of the

Nune vrder~ ~ he fact that there is a close integra ble
1 subsys-

tem is obtaine d by averagi ng over rapid variabl es. 'rhe

phase
to
transit ion in the YMH system to the disorde r phase is shown
tions.
be due to the infinit e sequenc e of rapidly growing bifurca
Results of section s 3 and 4 are substan tially of theore-
tical-no npertur bative nature owing to the· follo~ing: if H0 is
the HPmilto nian of the integra ble system, and gr H1 the Ha-

miltoni an of perturb ation, and if by means of a canonic al.

transfo rmation (rq)- lP:
q') one ·manage s to reduce the Hamil-
toni an H., -+ 3f, "n 1 to , H'1 ( p,, q')
H'0 \ p, ) + gra
then during the time t . .,. :J\ we shall have an approxi mated
solutio n with an error of the order 1 /~ • That is the clas-
sical perturb ation theory describ es the system during a finite
tim~ interva l, therefore~ any stateme nt concern ing the natucre

8
 bativ e chara cter und canno t be obtain ed in any
finite order of
pertu rbatio n theor y • ~

In the concluai'l~ fifth sectio n qunntu m-rile chunic al pro-

perti es of the Y!;! mecha nics and their relat ion to
the probJ em
of confin ement are discu ssed.

2. Funda menta l Subsy stem of YMCM

As is shown in ref. [17} the Hanlil toni an ( 1.1)

has an ob-
vious mecha nical inter preta tion. Due to SO(J)
~SO(J) s~netry
are prese rved the moments

(2.1)

(2.2)
with (2.1) coinc iding with coupl ing equat ion
(1.2) at a cons-
tant exter nal densi ty of color charg e. If the
new varia bles 0,,
E , 0 2 are introd uced by means of the relat ion

p. ~ o, E Oa
T
(2.3)
where

~t)
0

and 0, and
E"
0,
(
y ( t)
.. '<+. )
) (2.4)

. are ortho gonal matri ces,

The creat ion of a conve rgent pertu.rbatio n theor

etc.) is a vivid exwnp le of the achiev y (KAAI-theory
the~ the Hamil -

theor y of DS. In the quantu m mecha nicsement s of the class ical

a simil ar resul t is
luckin g despi te the recen t progr ess.

9
t01u.an (1 .. 1) will b~ rewritten as (17]

(2,5a)

(2.5b)

where
a t. e. e a
I,= IJ • z' , I a• x • 1. , I3 x + y
& '
(2.6)
I a a = 2. X", I >3 = 2.X ~,
Icti.' 0 ~or c'/'0.
and 2L and GJa.. are the vectors of angular velocity dual
to appropriate antisynunetric tensors D7]
•T
2 •-0 a 0._. (2. 7)

?rejections of the momenta (2. 1-2) on the "movablei• set of co-

ordinates have the form n.a.= o~'N'; . rnl = Oatj Mj b7)
(2.8)
No.~Ia.wo..-Ia.i. Q,, Mc~I;Qt-Ita.Wa..
and obey the generalized Euler equations f 17]

Thus eqs .. (2.9-10) together with the relations (2.8),

 (2.6) comp letely deter mine the YMCM which
has three irui& pen-
dent singl e-va lued integ rals [ 17 ]

H~M >

:L'he follo wj.ng analo gy with the mech

anics of oolid bc1y 'iB
rath er usef ul. i1.s is know n the kine tic
energ y of th':! sr;l:i.d bo-
dy is
Tso!td~ : L I. Q.~
c. ~ L
and due to SOU ) symm etry the J atte r
pre~erves the squu. rv of
the tota l momentum Me
If the inert it! tenso r I io3 j nde-
'·
pend ent of time , I~= O, ·the energ y is'
p1·es erved a.s well ,
whic h perm its to comp letely integ
;rnte the Bule r eq_uati.ot~ [131.J
other vdse (It I. 0) Jt wo<tld be impo ssibJ .e, since ~ 7}
dT"oeiJ i
-
dt
=- T? I,. Q,a (2.11 e.)
'·
Simi larly in (2~5)
we have a kind of "guu ge" bod,y w:i.th
time- depe nden t iner tia moments (2.6)
whic h rota tes in usua. l
and inter nal apac~s (prec essio n of spin
and isosp in), tillU it
is esse ntial that due to nond ie.go
nal elen. :mts ,of the iner tia
tenso r I a. c there la a corr elati on b~tween these rota -
tiona .
We shal l call the syste m with the
Ham ilton ian

(2.12 )

the funda ment al subsy stem of the Y:r.:i:CM

. Then

(2.13 )

As o. whol e tbe pictu re iu thin: OllP. mny

imag ine the ynri -

11
 a ma ter ial poi nt per -
abl es X. 1 y, Zto be the coo rdi nat es of
ce lim ited
in a thr ee- dim ens ion al spa
form ing compound mo tion
e
by an equ ipo ten tial sur fac
(2. 14)

shown in fig . 4. If
At Ty,..., = 0 ( Qt=" <.JQ.= 0) it has the form
s
equ ipo ten tial sur fac e lie
Q L. Wa. -#. 0, the n TyM > 0 and t~e
1
e com plic at-
. 4 and has a somewhat mor
wit hin tha t shown in fig
ven ien t to
tim e-d epe nde nt. It is con
ed form and , bes ide s, is
· e M"' ~· ""d Na
. ng wuy. s ~ne
t he f ol 1 ow~
pre sen t t h e rot ati on in
M~ and Ni.
may pre sen t the vec tor s
(2. 11) are con ser ved , one
h rad ii M• and N?
as poi nts on sph ere s wit
when it
s des cri pti on of the YMCM
So we hav a a ver y obv iou
"ru n" ove r
ee poi nts , two of whi ch
is nec ess ary to foll ow thr
thi rd
crib e the rot ati on, a~d the
the sph ere sur fac e and des
(2. 14) .
by equ ipo ten tial sur fac e
11
run s 11 in the spa ce lim ited
rol e
nex t sec tio n, the bas ic
As it wil l be shown in the
M is pla yed
stic pro per ties of the YMC
in the for ma tion of sto cha
whi ch has a ne-
by the equ ipo ten tial
sur fac e of the FS (2.1 2)
int o the
sen se tha t it is con vex ed
gat ive cur vat ure , in the
nat es (X, Y,
the poi nt wit h the coo rdi
reg ion of the mo tion of
Kolmo-
M (2.1 2) its elf , it is a
~ ). As for the FS of the YMC
sta tis tic al
pos ses ses the stro nge st
gor ov K-s yste m ru1d, hen ce,
pro per ties .

12
 ). Yang-Mills Classical Mechanics as a Kolmogorov 1(-System

Consider the YMCM from the viewpoint of ergodic theory.

It has been sho\'m in refs. [1 ,2] that trajectories of the
FS (2.12) are s·trone;ly unstable with re.cpect' to small changes

in initial conditione n...""J.d thut this system' is nonintegrable.

Phase trajectories chaotically fill the energy surface H.fs =

canst. It is intuit.ively clear v1hat b1 implied by chaotic nr

stochastic motion, but it is .important to formulate them mo:::e

definitely. In ergodic theory[4 - 7 1 these concepto are definite-
ly formulated and the cluusificution of the !rtvtion ao the in-
crease of chaotic-statistical prope.ct:!.e~:; is obtu_i.ned.

The aim of this section is to show that "... ho F3 cf the

YMCM (2.12) possesses the strongest statisttcal Pl'Upert.ics
from all known ergodic theories, that it is a Kolnwgorov E-·Gy/3·"
~em, and to descfibc its elobal properties. Let's first pre-
sent the necessa.t·;; inf'orlltntion from ergodic theory [4-lJ"
The ergodic theory investigates the statL;tical propel ti.e(l
of the measure-preserving mupping Tt of the phase npuce 1'-'!

on itself. 'rhe set of ordinary di!'f~r<':!uf;isl eqw_.tt;.I,,Jllfci 0.1.. thB

Hamiltonian dynaJnics

(3. I .i

p;:i.ves such a muprd.ng Ti of the pha~Jc :.;_pace. ~''·1 on J.tseJ

f:i'l nee to efl.ch po:u;. :; ( p> 9) ~X<.:: .i'--1 "J..G d('t c cm_i..~lec"\ L<ll :_tp proprl~'· :, 2

"" 'J~he rigJ:c1. hand 8J.de o.f: ( .\.;) :\e)·tcrmJ..ncB Jn ti-1;0) ni1w1r; . :rrac-
~'-1 th£: Ve>ctor f'i.. e.l<t cai'!.r~U 1;ht.: vectm· fi,_:l~l .... J pilauc.: :ro l (
C i L·.'f ~

., '
point (p, q) "X~ E M and due to the Louiville theorem the
t -
initial phase volume i.s oreserved.
'l'he mappinr, Tt
forms a one·-parametric group calle-d
4 7
phase flow. An important resu;tt of the ergodic theory( - ] is
the existence of the time average (averuge along the trajec-
tory)
+
r p . t) od-
~
llm-:rjt(Tx cit"' f
0 (3.2)
i: .... Qoo ~- ' .

is the arbi i;rnry function in M u f(x) =.XA (x)

is the ~haracteristic function of the set fl_C M < xjl = ,
i:C' Xr:, A ru1d lo ;~ero if X€ .A ), t;hen from the Birkhoff

theoren c~.2) f•Jllows the existem~e of the mean number of hits

in any m~asurcd set ,,

/I
for almost every point ( X ) •

Let us introdvce the classification of various -types of

J'lovm in th<e phase space M which characterizes the flows
by, the increase of their statistical properties. These are er-
godic sy:Jte:ms, .sys:;eros with weak mixing, with mixing, ·with n-
f~ld mixing and~ fine.lly, K-systems.
The l>S is culled ergodic if all invariant sets of the
mapping T ·;; ure seta o:f zero measure or sets of. total measure

M
The concept of ergodicity, undecomposab ility of mapping
or metrical transitivity is a natural requirement that the gi-
ven mapping Tt mixe~ the points of M well enough. It is

clear that if there is an invariant set E>(Ttl'l~e>)

whose measure differs from 0 u.nd jJ. [ M) then the DS may be

considered acting independently on E) and its supplement

M \& , i.e. the mapping does not mix the points ~ and M \I)

14
and, hence, the flow iD not ergodic.

Let, for example, the Hamiltoni an system (3.1) have u con-

serving integrul cp (P, qJ, then the condition ¢ (· p, q) :: canst
picks out the invarhmt subset and, hence, the .flow Tt is not
ergoU.ic. Ergodici ty thereby means that the D~ ho.s no conse::v-
ing integrals .
In particula r, it .follows from ref. [1, 2 J thut the }';-;
(2. 12) is an ergodic system. It is its weakest statistic al pn'
perty. We shall later shaw that it has much stronger st<=;tiati-

CHl propertie s.

l<'or the arbi tro.ry mapping the space M (with an

accuracy up to the set of zero measure) may be represent ed as
a sum of the coWltine; number of invariant subsets, on each of
which Tt is already ergodic [G J. 'l'hus, every DC Tt maJ be
decompose d to ergodic component s. 1'hus, in the previous exam-
ple, already on the subset ¢ (p, ~) :: t
canst the flow T is
ergodic.

l!'or ergodic systems the Birkhoff theorem allows to prove

the equuli ty of time averace.s to space ones [4-7 J
+JHT"x)
t

*i~ d<t = j f(xl dl' o.3)

D 0 'M
Let t:::: ..X A the~l i t follows from {J.J) that

sf(
M
X) dy = J' [ A )
hence, for erGodic systems tho trajector y of almost every
point X E: M falls in any measm·abl c subset of the positive

measure and· stays in this set .for a time in me<:J.D proportio nal
to its measure. 1'he equu.lity (J.J) muy be considere d as uno-·

15
 follo ws that for ar-
ther defin ition of ergo dicit y. From (J.J)
bitra ry funct ions l(x) and ~ ( x) in M we have

dent only on the avera ge.

stica l prope r-
Let us detel~ne the DS with stron ger stati
-\.
ties. The flow T has a prop erty of weak mixin g, if for ar-
in M we have l - ]
4 7
bitra ry runct ions l and <a
t
tern: fM.-[HTI)ca\x)drJhxl df'Jcatx)d!'l'~ o (3.6)
t-o, Jo "' J M

and the prope rty of mixin g, if sim.p ly

(J.7)
ttm jf(Ti<)ca(x)<l,<l ~ Jt(x)dp htx) d!'
t1
t--eo M M

Let now again ~ =.X A , <J = .X e, then from (3. 7)

~i.m r [TtA f\ ~] =J' L. AJ·J" (. '(')}

(3.8)
.
t--
h~nce, durin g its motio n the arbit rary
set A
will all the

time trave rse the .set e::, , the measu re

of the part of A
\:. has falle n in B being asym ptoti cal-
that at the moment
ly pro;.. ortion al to that of B , i.e. fl.
is unifo rmly mixed

all over the phase space .

Let us cite un exam ple used by Gibbs expla ining these

16
 definitio ns. Add some vermouth into a vessel with gin (p.-copor-
tions 10% to 90%, respectiv ely). Ergodici ty m:<latls ;hat o.f·t;e.c
a fuirly long time there will be on the o.vere.ge Hr)~· of ver-
mouth itl an arbitrary purt of the vessel, thoue;h it is· not ex-
cluded that in definitel y far moments of tirne the liquid in an
arbitrary part o~ the vessel may be either too strong o:r· too·
cweet, whereas the mixing means that after o. fairly lent; tlme
the liquid in an arbitru.ry part of the vesoel ~contain

1~~ of vermouth.

It is easy to show that if the flow has u v1euk mix-

ing, then it is ergodic, if it is mixing, then it is both ergo-

dic and v1eakly mixing •.The o_pponi te is incor:tect : the:r·e are

ergodic systems without weak pixing, syster;s with weuk mixing,

~'i thout mixing and systems with nLi.xins.

The DS has an n-fold mixing if for any functions f.,,, ~ 1· · ·
... ~n in M we have[?]

0 j' p 2 t, t,•••t,) n ,
tim tVlt,(T> <) ... f"(T x d-"~1:\oJhcx)d.JU
1:, ... t.,-.__.. , .- M
a.nd .i.f f:::...XA"':l
1
, , •• fn~.J.A.n then

0
Aof\T /l,Jl ... r\Tt.~ ...An
t, t.,) r- J
lLrnjU
t 1... tn~oo
[
;
_
n!'cA.
n
\::1
The ordinary mixing (3. 7-8) is a 1 T.1Ul"tiplic i ty mix:i.nl.>
Let us finully .introduce the concept of K-flov1 wt.ich 'I'H:

sha_ll neeU later. K-systews were introduced by J(olmogoro v ;.n

r.('f. [sj wher~ they were culled quasir(~gulu.C'. K-uyGtcm.'.'l t1ave

still stronger stutist'ic al propcctic z -~han tho;:;e r,Jent J.oned
nbove. t!nf'oJ'tuno. t<)ly.,. l>: l~ i.mpos~Jible to d-:-fine t'he l\-flon i.!,

·, '?
terms of concep ts used above.
The flow Tt is cfilled K-flow , if there is such a measu-
M ( UC ~ M
x·o.ble split-u p ~o -=- { C ~ of the space
C~5.
: ~. n c~; '1' ) that
1. T'~. >--so ut t > 0

(3.10)

.~ t'
3.
--
1\ T >.

M v is
where C, is the split-u p to separa te points

the split-u p consis ting of M itself or empty set cp

vd... ~<>1. is the upper bound and 1\ ~ e{ the lower bound of the
"l"ol.
.:; • The 1nequa li ty betwee n splJ.t- ups is define d
o(
split-u ps
as follow s: ~~ tt is· not smalle r than ~ ), if each

elemen t of c~ ~ ~ is a combin ation of a number of elemen ts

of c~ €. rc . 'rhe upper and lower bounds are unders tood by

means of this orderin g relatiQ n.

The meanin g of this defini tion is the follow ing.
Let t1
the region on the plane covere d by the split-u p
~0
be
(see fis. 5). ~rhefirst r:.ondi tion of (3.10) implie s that :J.t the
t
mappin g 1"+.. there ro;ises a nevr split-u p T ~1) which is al-
that d'.lring
wB.ys smalle r. The second and third condit ions mean
the evolut ion of the DS f-rom - -=:> to + 0c::> split-u ps will

arise, whose interse ctions consis t simply of the points

M
lower bound
(they are the same as the upper bound) , und their
is eq_ual to M

18
 It is intuitiv ely clear from the very de.fini'ti on of (3.10)
'tho.t there occurs a very strong mixing. Indeed the K-flow has
/ a mixing of any mul 'tiplici ty n [6 J. At the same time there are
DS having mixings of any multipl icity, which ure not. K-.oystem s.
K-system s ha.ve very t;ood statisti cal p.i-opert ies (in a certain
sense the best). There is a fine example of a K-system in .Ap-
pendix A.
Let us nov1 assume that a DS is consider ed whose
Urunilton -
ia.n i.s given. Will it be an ereodic, mixing or K-system ? Arc
there sufficie nt conditio ns by means of which one can annwer
this question ? The importan ce of this question lies in the
fact that we wunt to define what ch1SS of lJS the YF>iCh belongs
to, i.e. whut mo.ximum ste.tistic o.l properti es lt ~us !
In the papers by Hadamard , Hedlund, Hopf, Krylov, Anosov,
Sinai eb al. (9-12] such sufficie nt conditio ns vrere obtrdncd .
~hese authorn have investig ated the behavior of geodel3ic fluwn
on Riemann ian manifold s of negative curvatur e .. They proved
that the geodesic flow on the manifold of u constant negative
curvatur e cmd on the surfaces of the variable of negative cur-
vuture has u strong instabil ity and that it is a K-flow. The-
refore, the negativi ty of the curvatur e of the manifold M in
the sufficie nt conditio n for DS; to be a K-systeff i.
On the oth~r
hand, it follows from the fl:laupert uis princi-
ple that ·trajecto ries of the system (3.1) are geodesic flows
of scm~ Riemann ian metri~s ['3]. If the Hamilton ian is written
as
1 1d S
H=y\ df
)a • u(9)

19
 a e (3.11)
ds a = dq' + . ' .. 1
q 9"
and the Riemannian metrics is gtven in the form

(3.12)

whc!·e E is the total energy, then the nego.ti.vi ty of the cur-

vo.ture in the region of configurn.t:Lon space U (9) < E will

be the sufficient conditi9n for the system (3o11.) to be a K-

.nystem~

Mi.~ N
..
Consider for simplicity the
!'!.
two~-dimensional

0 (a more general case is considered at the end of

l!'S (2.12) when

this section)

(J,1J)

The me·trics (3.12) has ·Lhe form

d/=7\(q,q,)ds"=[E-
The curvature in the region
r9.·9~1(d9~+ dq:)
U( ~) < E is
().14)

l (q~+q:)(E•~q',q.a)
R =- 2~ t>"~n..A = • > o
~ [E--fq~9;]~ '
(3.15)

82
a9.' +
and it is strictly posjtive. Thus, the sufficient condition is
not satisfied in this case. But this only comes to indic~te

that the phase trajectories are stable, if considered only in

the intervB.ls between two scatterings on the boundary U(q), =

20
 cons t (see i'ig 4a). As it will be shoom
late r, the inst abil i t~
det_e cted in [1 ,2] deve lops due to the
scat terin g on the equi -
pote ntia l boun dary U:; (q)-== cons t. For-
5 this purp ose spec ial cri-
teri a shou ld be used obta ined in pape
rs ment ione d abovJ9-l 2 ,3 2~
The inve stiga tion of geod esic flow s
on the ~lifold of ne-
ga.ti ve curv ature[9-12} has shoWh that
in the form ation of stat is-
tica l prop ertie s of K-sy stem s an impo rtant role is play ed
by th~ so-c alled expo nent ially comp
ressi ng and expa ndin g
tran sver se fibe rs [15]. In part icul ar,
if the curv ature is ne-
gati ve, then the exis tenc e of appr opri
ate fibe rs is prov en.
The neg~tivity of curv ature is a suff
icie nt but not a nece ssa-
ry cond ition . Ther~fore, the way to
find out when the DS is a
K-sy stem lies throu gh the cons truc
tion and inve stiga tion of
tran sver se fibe rs.
Anosov and Sina i [11 , 12 , 14]
succ eede d in cons truc ting all<.
inve stig atin g a wide clas s of DS with
truno verA e fib.e rs. They
are ~ syste ms of Anosov [11] and Slna i 1 s .sca
tteri ng 11 billi ·-
ards11 [1 5], whic h due to the exis tenc
e of fibe rs wiyh requ ired
prop ertie s are K-sy stem s. We must
shor. -_that the lilS (2.1 2)
has expo nent ially comp ressi ng and expa
ndin g tran nver se . fibe ra
Whic h are home omor phic to fibe rs of
hype rbol ic K-b illia rds of
Sina i and, henc e, the FS is a K-sy atem
.
In our case ther e is no need in a
s~rict defi nitio n of
fibe rs, it is enou gh to have an obvi
ous idea of them . Con si-
der the inte gral curv es of equa tions

near the solu tion X= y ::: 0 (see fig. 6).

21
 To describe the arbitrary trajectory near the solution
X::. y=0 One should foliow how its coordinates )( 1 y vary to-

gether with time in the plane perpendicular to' t It is

easy to see that solutions with initial values lying on the
X-u.xis form u frunily which at t- oo exponentially approxi-
mates to the solution X= S' '"' 0 , whereas solutions with ini-
tial values on the Y-axis behave in the snme way when the
time direction changes. The first family of solutions forms
ai" exponentially compressing transverse fiber, the second
an expandinb one. The l)S is a lJ system of Anosov, if the
phase curves near any datwa behave like integral· curves near
t~1.e solution X= y = 0 in the example considered above. For
such systems Anosov has constructed transverse fibers and
proved that they are K-systems.
In Sinai's papers [1 5] transverse fibers have been systema-
tically used to prove that the given DS is a K-system. In par-
tic~ar~ he has considered the motion of elasti9 balls in a

box with elastic w1d convex inside walls, a sccttering billi-

3.rds [3 2 J,
and by means of transverse fibers it has been pro-
ven to be a K-system. Unlike ':J systems of Anosov, Sinai's
scattering billiards has no uniform instability. However, the
instability arising at the scattering On the convex inside
boundary is already sufficient to fona strong statistical pro-
perties like those of K-systems. In this case the scattering
role oi' the negati vC curvature is played by the eonvex inside
boundary . •
In Appendix B an exam.ple is given used. iJy ;, TJ(' · •:,i -, •.•.)::t ..
rating this stutemen·t.

2?
 Let us come back to the Yang-M ills equatio ns. As we have SF~er,,
the curvatu re in the range l1s\9)< £ is positiv e, i.e, r;eode-
sic linea are stable there and ·the .system is not uuif0rm ly
un-
stable,, as was the case with lj' syster.1s • .Hut in spit\: e:f thts,
the equ~potential boundar y Ufs..:: E (see,fi e. 4)

ia convex iilside and has a negutiv e curvatw :e l As l~

the cw~e
with Sinai's billiar ds, just this circumst a.nc.:e provide s strong
statist ical propert ies of this system, K·-mixing. Be loY: it will
be proven that transve rse fibers of the FS (2.12) haVe the
same structu re as in the hyperbo lic K-billi ards of Sinai and,
hence, the FS (2.12) is a K-syste m.
To prove the equival ency (homeomorphism) of these systems
,
one should introdu ce the concept of the ·structu ral stabili ty
in the narrow sense. The concept of coa~seness or structu ral
stabili ty of DS was introdu ced by Androno v and Pontrya gin . [1a].
The dynamic al system Tt is called coarse, if for ~y system
St fairly close to Tt * (in the sense of C 1
metric s.f( f,~)
=SIt-~ I d..!" [19]) ·there exists a mutuall y single-
valued and continu ous transfo rmation (homeomorphism) clcse to
the identit y transfo rination {in tHe sense of C. 0 metrics
,P(~t'7)=supl {-'a J 19
( ]), transfor ming Tt system tra-
jectori es into those Of st system, the directi on Of the LlO-
tion along the traject ory being conserv ed (the velocit y of mo-
tion may be various , i.e. for example , periodi cal traject ories
* By the distanc e between DSs is implied the distanc e in the
metrics C" between the vector fields generat ing these sys-
tems (right hand parts in (3.1 ))-.

23
corresponding to each other 1nay have different motion periods)~
Ano:_lQV has proven the:~ any ~ system is coarse, hence~

so is ~f1e geode.:>ic l'low on the manifolds of negative cuiWa-

ture.
rhuSt ,,copite ~he complexity of the picture of phase
1

eurvcs! t~o.ch ci: ~!h.: ch in ·~xponentially unstable, at a transi-

tion to u c.lose ,rpetor f'lelrl defined by the right hand part of

(_J, 1) ,-,~1.1 ~hi.•J. p.i.ctux·c w_:.-~h cveryvehere dense phase curves and

cfw.nged. The li'S (2.'!2}, u::; we havE-, noted, is not a· Y system

w1d, po13sj.bly, i:-=" not eoa.rse. llO\H~Vet' 1.• nfl we shall show below,

:i.l. ts stable with X'f:Lspcc·<; to ~>pec:Lal pe:eturbutions and iR

\ile :-:Iwll call the oystem structurally stable in the

!'.f!:£2::g_'!L,"22.i::l.§.f.~ •.:.:' J.-~ :Ls s·~ru(x~urall:y stable with re~lpect to
1
(1, r·tar!n1t:: ~ms on1~y u.lont_, ::wme curve in the C topology u.nd
)WL ir~ the whoJ.·~ v.i.ci.n.:..ty. In fig. 7 the C.. -vicinity of the

<J:rsteFl rt in ttC\. C topOlogy iS represented by a circle,

and the G -vicinity along the curve by a fra€,1Ilent of the

cu:cve .i.H '>!i t-h ..i.n -Lhc circle.

Cons:Lder the following perturbation of the system (2.12)

'l'he parumetcr d.. defines the eurve in the space o~ vector

fields deterr.lining the DG. il..t d...= 2 the (), 17) coincides with

the ayste;;1 (2.12) (in fig. 7 we denote it by A), and :lt~:;:Oc>

(the point B in the space of vector fields in fig. 7) it is

Sinai's billiards with absolutely elaBtic walls depicted in
24
fig. 4.

The main &tatement of this section iS that at the change

of a( from two ·to infinity, ph~se trajectorie s s.re homeomcor-
phic to each other despite the large chant;~ of d.. 'We huve
made sure in this calculating periodical und nonperiodic al tra-
jectories of the system (3.17) with a computer and then com-
paring them to trajectorie s of the billiards ·system with d..::: 0o

constructed by mew1s of laws of elastic reflection from abso-

lutely elastic walls (fig. 4)

The above fact yields three significant conclusions :

firstly, the strong instability of the lMCM develops at the
scattering of the trajectory on the equipo~ential boundary of
negative curvature (fig. 4); secondly 1 the s.tructure
of exponentia lly compres•ing and expanding tran•verse !iberu
of the 1i'S ot: .the YlllCii (2.12) ia ai.mila.r 'io that o! Sinai 'f:
billiards system with hyperbolic walla of fig. 4, and hence,
finally, the FS or the YMCN (2.12) is a K~•yatem o£ Kolmogorov
and possesses atrong statistical properties.
The properties enumerated above are those of the global
characteris tics of DS. There ar~ two more global characteris -
tics of classical systems which are essential for the under-
standing of quantum mechanics of K- and YM-systems.
To each DS T
t
one may juxtapose the number h (T)
called the Kolroogorov entropy[s]. The quantity h is positive
and at various Tt may have values from 0 to+ 0o0

K-ay stems possess positive entropy[eJ. Sj.nce the FS of

the YL1C1i is a K-aystem, it appears t&at its entropy is strictly
positive.

h ( TJ 5 ) > 0 .
Secondly, to each DiJ Tt is juxtaposed a group of linear
operators Ut

Ut f (X) ~ f (T \)
where { t (x)} are the c.omplex functions in M , summed up
with the square ({ae. I..~(M))·
In 1931 Koopman has proven that if T -I: is a measure pre·
serving transformation, then the operators U t form a. one-
parametric group of ~ tary operators [ '
4 10 7
' ]. The properties
of the D:J manifested as spectral invariants of the group Ut
are called spectral properties. Ergodicity, weak mixing, mix-
ing and K-mixir1g are spectral properties. In the case of K-
flows one may calculate completely the spectrum of conjugated
group of unitary operators U-t • It is a countably-multiple
Lebese,'Ue spectrum[a]. This i~1plies, first, that the group Ut
acting in the Gilbert space L,(M) has no invariant su"<>spaces;
second, there is .such a countable set o! functions { ft (x)}
i.. 1. 2., .. . •. . . that u~ L - oo .: s < • ex:>

forms an orthogonal basis in La ( M) Thus, we manage to com-

pletely calculate the spectrum of the group of unitary opera-
t t
tors UJ'S conjugated with TJ'S (2.12),
Up to now we have considered two- und three-dimensional
FSs. Consider now a FS which arises in e;auge theories with
higher order symmetry group D. Let's define the N-dimensional

26
 4 l
- q !

'" "
spa ce U(q)<E. Acc ord ing to (4) the me
tric al ten sor i•

o...'J.d the 3Cn lur cnl'V I-ltU l'e is

_ \N- 1)
N
lt io see n !'rom thi s exp
.t•es sior l for cul:' vc,L Jt·u
IN) l.lt;•.L c.t N ~ b,
R.ss>- ~ . Thi s imp lieo th8. t wit hin (;}
the ru.. :t;<-"U.fs(q)<E tl'n -
joc tor ieo e.1·e at;u in ::::h.
1.ble, and -the i.ns ta::: lL.t
,y u.ri. sNJ due to
the sco .tte ring on the Cqu
.ipo tent iu.l bou ndrd 'Y U (N)
nec ;ati ve cur vat ure , sin 1 5c(9) : : . ft. of
ce the p,e is hom eOl! IDrp
hic to the nea t-
teri ng hyp erb olic bil liu
rds . 1Illlii!J fac t cnn not
be est abl ish ed
from ( J. 2 7) ,- sin ce it is
vnl itl bey ond the equ .Lp
(N) oten thtl bou n-
dur y UJ (q)-:::: E onl y, and doe sn't tak e
5 into acc oun t the
bou nda ry eff ect s whi ch
ar·e ess ent ial in thi s cas
e.
A ::1u rpri se ari sin g at
N > 6 is i;he eme rgen ce of thP-
ran ge wit h exp one ntia l
inz .tab ilit y def ine d by
'

sin ce the re
 -t
cs=J T
L_c~ t>O

s. Fig. 5
From (J. T?) follows that the range of exponential instability
is along the equipotential bol.Uldary and broadens inside the
(N)
range UJ < E with N -""- o.o ! This comes to once again
5
confirm the fact that the instability arises due to equipoten-
tial boundary.
In conl')lusion let us present the considera·t.J..ons indicat-
ing that statistical properties of the YMCM at the absence of
Q.
external sources ( 11. = 0 ) are caused just by the FS (2 .12).

~o do so let uc consider eq. (2.9) determining the motion of

rotational degrees of freedom, assuming at the same time that
the j.nertia moments (2~6), i.e. the variables X j Y 1 1.. are in·
dependent of time. Whereas it is easy to understand from
(2.11a) that T'IM 1'/ill be a conserving quantity

and that one can integrate· eqs. (2•9) like Euler equations for
a freely rotating solid body [13 1.
This is an obvious indica-
tion that rotational degrees of freedom in (2.9) do not induce
statistical properties of the Y.MCM, and that these properties
are caused by the FS (2.12).
In the next section classical gauge systems with sponta-
neously broken symmetry ~e consider~ and it -is shown that
when the value of the vacuum average of the scalar field tp
is sufficiently large, the system doesn't manifest its statis-
tical ~roperties and is in a kind of order phase (it is close
28
 to the integr ated system ), and that with the decrea
se in< tf'>
by way of an infin ite sequen ce of rapidl y growin
g bifurc ations
the system passes into the disord er phase .

4. Gauge System s with Sponta neous Symmetry Break

ing

In ref. (1] statis tical prope rties of classi cal

gauge
system s with sponta neous symme try breaki ng are inves
tigate d.
In the pr~sent sectio n these system s a.re EinaJ.ysed
theore tical
ly.
Let us first consid er the two-di mensi onal H&J.il tonian
( 1. 6:

(4.1)

.wher e Wa =~ • '1.~~ 2. Let's introd uce the variab les J~, lf~

q,=(2.:1</w) '/a Scnlf , Pt : ( 2. 'J ; w)'/a cos 'f' (4.2)

( i. = 11 2) where (4.1) will read

H = H.• H,, H,= (':l, +:Ja) c.J

• :J :J . •
H, = 1 ;.,.. [1-cOs2.lj',-cos 2.'fa• ~cos (2.<J', +2.'f,)+ ~cos(e<p,- e'f.2
(4.3)

As has been noted [ 1 }, the param eter

; ;
3T "::,
Gi
W
4
/9' a jJ 4 com-
pletel y chara cteriz es the system . At ~~......,.. o.c the
Hamil tonian
H is equal to Ho (harmo nic oscil lator) . Trajec to:des of
the system with the Hamil tonian H o are we:.:.l known
. The main
proble m is how they will behave at the presen ce
of the pertu r-
bation H1 , and t -- 0c:>
It is conve nient to consid er, follow ing Poinc are
w1d

29
 Birkho ff [2 0, 2 i}, not the trs.jec tories themse lves
in the phase
apace, but their attcce.a si ve :Lnte1·sec·~ions with two-di
mensio nal
eutt.int ; uu_dac e d.. • Let us determ ine the mappin g T
tr:.l.:.lt:-_i_:)l~J.1inc d... into itHelf ns follow s: fixing on cJ...
we let- the trajec tory out o:f it and follow

The new point will be TX

'l'.iLL:J io.ln.c:u ·e mappln[~ replac es the trajec tory by an infini te
obto:Ln cU ut succes sive applic ations of
T
n.
)\
,,
. ii.ll the essential properties· of the tra-
jcc ~Ol'J-' urc .cei'lec teli ill the pcoper ties of this set,
which we
pi1~11 cu11 '-'· pho.sc pict·:-1- re.

If the motion iiJ poricd ic, the crossin g occurs in

a f'inite
m..unbcr oJ' points ; if :L t is limited by the tol'e surfac
e, then
the points are lccate d on closed curves , and, finally
, e.t er-
godic motion the point chaoti cally wander s over
the plane d.. •
The phase piei;ur e of the system (4 • .3) with the Hamilt
onian
Ho in the plane cZ~(qa,
p._) at P1 (cl.)> 0 repres ents a
circle of imrnob ile points , unO. it is of intere st for
us how it
will chance :-~ t .:JT =f:. o.o
Let us pass from the variab les (4.2) to j•,Je.
1 ¢ 1, ¢._
J,,:J, •::J .. (j),_ t.p, + tf ..
'' 2.
(.4.4)

j._=J ,-J2 tp, lJl..

> ¢2= 2.
• Thisis ~ genera l constr uction allowi ng to compar e eS.ch
con-
tinuou s flow T 4:. (see sec.2) to the cascad e T 11 ,
where h. is
the intee;e r nUmber .
 then

Since H0
is independ ent of J e. the uotion proceeds at
one frequenc y CJ

J,. 0

When the motion iu the unpertur bed system H<> is describe d by

a smaller number of frequenc ies than the number of decrees of
freedom (the system is "shifted " by conse:r·vi nr; integral s)
then [13}

de.+. II1 ~_t-l~

1
" I/·~ 0 (4. 7)
',\ dJx d.Jm I,
This .l.s o. cuse of se~f-dcgenracy.

The unswer to tlLC qucs·:;:i.on ro:i:Jed at tl!-e be.:iinnin c of

the section on the e!".fP.c:t of the pel~ t.urb::!tio n
1 H i3 given
by the Kolmogor oc "tl. !Orum "22l
~ ... If t.bc perturb;-· ami rl,
Ho .i.s
not dcc;enoru:~e

·;;," H
-----
~ 0

then the majo:e:t ty of nonreson ant to r-es • arc sl·ightly de.for:ned .

--- --- ?
'rho f~ .. quenc:i.en on O.pill'Opr iute tares ~.re inco.t<"li.tcnsurabl£)
(see [::2}).

31
 Trajec tories of perturb ed motion beginn ing on these
deform ed
tares fill them everyw here densly and quasi- period ically.
These tares form a closed set nowher e dense (betwe en which
sli~s nre left fil~ed up With the remnan ts of disinte gratin g
re::wnance tares) , which has a positiv e measur e tendin g
·to that
of the whole phase space when the pertur bation H
1 -... 0 *.
If the Hamilt onian H is degene rate, as in our case
0
(4.7), then the pertur bation will have a differe nt result
. i 1he
problem s due to degene racy were solved in Arnold 's
papers [2 5, 26 1
Below the result s of pertur bation theory for degene rate
system s
are used.

4a. Self-D egener acy • ,Extrac tion of Age Part

The basi~ idea of classi cal perturb~tion theory for

dege-
nerate system s (26 consis ts in ~he ave:ra~ng of the
1 Hamil ton-
ian H, (4.5) over the fast variabl~ cp 1 and extrac ting
the age part from H [ 2 7]
1
a.sr. t.2.~
-H 1-- i3f j
1 H (
1 J1 1Je. 1'f', 'fa ) d .h
n. -" 't'1 =CJ (J,-
8 w ~Ja) [1 + 21 co~~"'"ta'] (4.9)
0
or, in the variab les ::1 1 , 'Ja, lp,> tRa

-·.·a JJ
H," 'ie~ .... [1+ ~ co~(2<Jl,-2\.f>._)] (4.10) .
. ~

The period ical part · H; is

• ThisaPenomenon was observ ed in various . numeri cal experi -

ments L2 3• 2 ~]. ·

32
 H,~ H-H,~
Ca.nonlcal equations with the Hamil toniun H 0 + H1

J, ~ 0 (4.12)

¢~~
2. dJ e
define the slow, 11 agen variation of· the para.me~ers j e. j cpa •
If the system with the Hamiltonian H" + H1 is integ-
rable, and in our case it is so, then at a proper selection of
new variables s1,5a I e,) Sa the age part of H will
be independent .of angular variables. The periodical part of H
leads merely tO an additional vibration of the perturbed tra-
jectory by the quasi-periodical motion with rapid ~1 ~ 1 /Jf
and ~low "Sc ........ (J frequencies

s,. o, s... o , 8, • s e. ~ ~·, 0

, (4.1))

The basic result obtained by · Arnold r25

~ '
261
is that nt
3. small enough perturbation H, for the majority of initial
condi tj.ons the perturbed motion is indeed quasi-periodical
and close to the above qua?i-periodical mr.tion (4.12-13) with
the Hamiltonian Ho-+ H, at all t

4b. Quasi-Peri.odical Motion with the Hamiltonian

The system (4.10), (4.12) with the Hamiltonian Ho+ H,

 .::-:uce Hc + H) depend<'' on the diff ere nce
of nng u.la r VH. riab les
'f.- 'f, onl y.
Let us def ine the phaDe _pic
ture of thls syst em in the
plan e o\= (9,, Pd at P, (ol) > 0 or in the var
iab les :1,,1,, 'f.,l/'0
(4.2 ) in the plan e lf'" 0 Let us sub stit ute :1 1 ::: 'j- 'j
.trom (4.1 4) into (4.1 0) 2

(4.1 5)
Thi s alg ebr aic equ atio n def
ine s the line s of inte rse ctio
n oi·
inv aria nt tare s wit h the
plan e d.. , sinc e one may :fin d from
as a fun ctio n of e lf
and then subr:rLi tute it in
(4.2 ). As a res ult the pha
se pic ture wil l loo k as show
n in
fig . 1.
I To cen ters of two clos ed curv
es corr esp ond stab le per io-
dic al traj ect ori es, and the
two poi nts ·of con tact are
met by
uns tab le per iod ica l tra jec
tor ies , sep ara tric es. i/e
i'in d the
per iod ic traj ect ori es from
the condit~on[ 8}
2

(4.1 6)
For tho stab le we hav e:

and for the uns tabl e

'
l.f -
1a
~ Q
lD
1
m-
It is see n fi:·om the se rela
tion s tha t i-t is the vic init
y of the
reuo nun cc tore of the syst
c.:.t wi tll the Htun il ton i au

34
 h~ (J,. J.) c.) ~ (4.19)

- W+ ~··
Q,-
J.
2.W"- '
i
that is substantially distorted: (at 'j 1 -:::.. 'Ja we have Q, = <2e)
the circle is replace by a zone containing crescent curves
grouping near periodical trajectories (see fig. 1).
E'inally, it is useful to write this integrable system in
the old variables (q"q;:..q 11 qa)

a •) 1 • ~ a
- 1 ( P,•wq,
H.•H,;r a
•e:(r..•wq,)• 8<t•w1 (p,+wq,')·
a a

(4.20)
•
(r: + w"q:h:w" [lP.Pa + w'q, q.)'- w•(p, k Pa q,)•] ·
Nhose Lagrangian is an infinite series in 1 jwa , and the quan-
tity

(4.21)
is the second integral. The periodical part of H. is

(4.22)

Thus at small perturbation the system (4.1) has a phase

pictu:r·e close to that of the integrable system (4.20) in
terms o:f the above KAM-theoryl22 1.
It should be now clarified what will happen to the phase
pictu.re when the perturbation increases ? The- KAM-theory

35

'
 gives no answer to this question.

4c. .Arnpli tude Instabi'li ty

The answer to this question is in principle knovm. It has

been shown in the third section that the classical gauge sys-
tem without spontaneo us symmetry breaking is a K-system o.nd
has strong statis_tic al propertie s because at 31'. = 0 ( '1. = 0)
- we have a K-system, and at 5i = ao at.. ..:.ntegrabl e system (4.20).
What happens in the intermedi ate region ? Is there any cri-
tical value 5f c at which most KAM-tores will be destroyed ?
There are no satisfacto ry answers to such questions yet
but it has been ascertain ed by means of numerical experimen ts
that isolated resonance s have no serious effect on the motion
1

but as soon as resonance zones are overlappe d, there appears

an instabili ty, and relativel y many trajector ies leave their

teres too [3 11.

One may estimate the critical value 5fc. at which ·the
overlappi ng of resonance zones occurs. To do so, one should,
_followin g Walker and Ford [za}, turn to the variables S,.Sa 9,9~
1
(4.1.3), where the Hamiltoni an Ho+ H, is diagonal

(4.23)
H; H.(S.,S~)•H,(s,s.)• H, (s,s.e,, e,)
Expanding R' in Fourier series in 6) 1 · and 92.
'
H _L
,- ll.,m
H•mns,s
( ,)ei(n G, + m Eie) (4.24)

we shall obtain new, the so-called secondary (h..~ m ) resonan-

ces, whose position on the plane d.. is estimable , since at

)6
each fixed ( ~ 1 nn ) the system is integrabl e. The or~g~n of
secondary resonance~ is well illustrate d in fig. 2. 3fc.. way
now be approxima tely estimated by means of the overlappi ng
condition of the primary 2-2 und these secondary resonance s.
'Fhe ''phase 11 transition detected and discussed in [1] is due t"
thia very bifurcat~on.

~onsider now the three-dim ensional cnse

(4.25)

which also is cha.xa.cter ized by one parameter .!Jf • Let' a int-

toduce the variables :::fi.J lfL (4.2)

a
(4.26)
H1 :-'L [ J' J. [1- co~ 2. 'f'c- co~2.'P. + ~ co::,(2'fl. +2 <f,)•k-co~(2'fl-2'f.li.
'2.(.)a\.<K. 1
c::. .,

The age part can be distinguis hed by introdl.lCi ng -the VOl'iublt-;S

~. by means of canonical transform ation
j .l. J 'f"L

(4.27)
vrhenc;e

(4.28)

J'oc ~he Htua.iltoni .an Ho the variable 4\ is rapid, and

are slow. Averaging · the perturbat ion H1
over the rapid variuble cP 1 we obtain

(4.29)

37
 In the system with the Hamiltonian Ho + H1 one may immedia-
tely find the second integral

(4.JO)

i Has it a third integral ? The answer to this question is un-

knovm to us.,

What will happen at various gr in the system {4.25} '?

As in the two-dimensional case, at we h~ve

i three noninteracting oscillators. With the decree.se in 3f

(i.e. with the increase in perturbation) the system (4.25)
comes closer to the system with the H~ltonian

g• '
h = (J,<le•J,)w • 2w•(J,·Je<leJ3<:J3J,)
which in ini.t~al coordina:t;es has the f'orm
3, e. i

h=a:f-
I " &) '3 'C'"" • 2 2 e 2 ~)
ft•w~, +8w•L.(P-•W9.)(F"•w
( • 2 a (4.)2)
,_. 1 iLK l K t

:Pinally, at some jf the system will be stochastic.

5. YJl and K-System QuantUm Mechanics

;·' r ~ • In previous sections propert~as of classical solutions

of' Y~ll&,equations have been studied. It has been shown.
that in the phase space ot the system the energe surface
H -= c'onst is chaotically fille4 with trajectories, and that
t~s flow is a K-flow close to that of' the billiards system
with hyperbolic walls (f'i.g. 4). lfb.>';lre!ore, the phase flow··
of the Yang-Mills ~lassical mechanics possesses all the pro-

18
 pert_ies of K-flow: mixing of uny multipl~.city, the positive
Kolmogorov entropy 1 the spectrwn of ·the unitary operator UT
conjugated to T'I'M , the counta.bly multiple Lebesgue one.
'l'he natural question arising now is what quuntu.:il-:necho.ni-

cel properties the system with the Hrunil tonian H'{M pos~;Jes-
ses if in the clt1.S.sical limit il - 0 it is a K-system'?
The significance of the answer to this question consists
in the following. In the field theory, ·e.e. in Q'2.1J, the electro-
t.1B:gnetic field is represented in the form of a set of harmoni-c
oGcillator:J •,vhose quo.nturn-mechru1:.cal p1·operties (as of an in-
tegrable _system) are well rcnovm, u.nd the interaction between
them iS taken in·to a.ccount by the perturbation theory • .:>uch an
approach excellently describes the experimental situation. In
QC:U the state of things is quite different .. -'l'he properties of
the ThiCM as ·of a K-system can'<;; be establi.W.1ed in a.ny finite

order of perturbation theory to a ho.rr,Jonlc oscillator • (See

sees. J and 4). 'l'hel'e..t'ore, to understand the !..;CD, it seems i1a-
portent to investigate the quant.um-mechurlical propel'ties o~ the
systems which in the classical lim.l t are K-sys~.. ems.

Historically the _problem of the quantizct::.on or noni:rte[;-

rable systews, i.e. syst';"mS which have no conservint; ir:te;;1·als
<n:cept for energy, goea be..cl-;; to the dmm of quanturr. :nechan.i.cs
o..fte.r the def:;.nition of the "Bohr<lor.uner:feld. ounntization
20 ' JJ' Jt~}. It hc.s been
l
. prover, by J.:'oi~carc the.. : the
ru1 es
p::.·oblcm o:!:' three bodies is r:ot inte~~r:.!.hle, ~~nd, t:~.crciore. the

• l'he classical oe.:~turbn.t.Lml ~heor.v :.s .J!.lplieti, no·::

. '
t un-.necnun .'. . co.1.[16} •

YJ
 Bohr-S ommerf eld quanti zation rules c~~ot be applie d to this
system [20, 33, 34], .

theory in the Hamil toni an gauge A

0 = 0
.
Consid er the station ary equatio n of Schrod inger in the
Y:M

with the couplin g equatio ns

a.
wp.ere Hi.= et C.t.jK a.
Gj K
• Let the fields A be again time-d epen-
dent only. Then (5.1-2 ) may be rev~itten as

where E.-
is the energy densit y. Let us call the system
(5.3-4 ) the Yiii quantum mechan ics. In the case when the
matric e
a.
of the fields fl t. is diagon al {see sec .. 1), we shall
have

Eq. (5.5) is a quantum analog y of the FD (2.12) . Since

the FS
is nonint egrabl e (see sees. 3 and 4) the SchrOd inger
equatio n
(5.5) cannot be solved by the separa tion of variab les.
The only
quantum nwaber is the energy , since there are no other
integ-
rals.

40
 The stationary o•mve functions of the quantum system ~e­

pend quasi-perio dically on time, therefore, there can be no

question of the properties of mixing and instabill ty like ·those
of classical systems. The .complexity of the classical system
trajectorie s turns now into the peculi~rities df the energy
level Spectrum and wave function of the correspondi ng quantum
system.
It now seems reasonable to define more exactly the ques-
tion raised at the beginning of the section in the.followi ng
way: what is the structure of the energy level spectnun and
wave functions of' quantized gauge systems, if in the classical
limit they are nonintegrab le and are K-systems ?
Consider f.irst integrable systems. The peculiarity of ener-
gy spectra of integrable systems with N degrees of' freedom is
the degenracy of their energy spectra, since in that. case th€
SchrOdinger equation is factorized into N one-dimensi onal
equations wi~h N various quantum numbers.
In contrast, nonintegrab le systems have no degeneracy ut
e.ll[35~3 6 ,J7 11 3]., The energy levels are at finite distance
from each other, and their relative arrangement is like that
artsing in the systems defined by the Hamiltonian matrice with
randomly distributed ele.ments [~S, 39 , 40] e There occurs the
so-called." level repulsion 11 expressed in the fact that the pro-
bability of finding two neighbourin g levels at a 'distance A
smaller than the mean value < D. > tends to zero .as

Af.l- 0
a-o
where ~ is the critical exponentq This implies the exist~nce

41"
of a strong correlation between near levels. For the tYpical--
integrable systems this quantity behaves as

exp{- 4} - 1
"- 0
These conclusions are confirmed at analyzing energy levels
of concrete K-systems: square billiards with a scattering ci~~­

cle of the :radius H in the middle[41 ] (see fig. 10), and the
billiards with a stadium-like boundary- two circie halves of
the radius ~ are connected by stre..... &ht lines of the length.
2
0.[4 ] (see fig. 11).
Both systems have integrable limits: the first at R = 0-·
'is an infinite potential well, the second at CL= 0 is a round
well of the r~dius ~ As is known, the energy spectrum in
this limit is strongly degenerate. At R and CL being dif-
ferent from zero they are K-systems, and the behaviour of their
energy levels is shovm in fig. 12 from ref. [41]. The degene-
racy is eliminated. Di~tributions of energy level spacing are
shown in fig. 13 [ 41 ' 42], whlch is consistent with (5.6).
Of special interest is the structure of wave functions.
In fig. 11 a typical wave function is presented. The nodal
curves are seen to be irregular in direction and their separa-
tion is rouehly regular verifying the prediction about the ran-
dam orientation of the Wave vector at different poritions.
Let us come back to the gauge system (5.5). ~et's apply
to it the regularization method applied to the FS (2.12), i.e.
introduce a spontaneous symmetry breaking. The Hamiltonian
will then have the form of (4.25), where p and
9 are the
operators.

42
 ~
In the integrable limit ( CJ - <>e ) the spectrum Of this
system coincides ·;iith the spcctrwn of the sum of three hamo-
nic oscillators. It is degenerate, as for -:;he billi&rcir:l syster,;JJ,
at R::. Q..:::. 0 • Viith the increase in perturbation (G.) <co
.
the potential energy gains channels along the axes x, y, z.
On eliminating the regularization <J':: 0(1.::: o) the channels
lengthen and tend to infinity.
Making use of the results obtl,l.ined at analyzine the bil-
liards systems [ 41 ' 42] , one may qualitatively describ~ the
gauge spectrum .. It consists of the nonzero energy gr01.md stat<:.
and perturbed states where degeneracy is lacking. 'l'herc occurs
a level repulsion (5.6) with the critical exponent }J, so far
unknown. The wave function relief is like the one depicted in

fig. 11. \li.th the decrease in cJ • t'he grOund state energy

tends to zero together with the .average tlistS:ncc · between le-

vels. This fact is easy to understand taking into ::..ccount that
there are no dimensional parameters at (J ·= 0 . ?Jo'!;e tho..t

the dimensional pararaeter may arise only when allow:i.lltS .I'or

space degrees o.t' freedom that we have neglected.

In conclusion consider the conhection of the di.Gcuosed

problems with cvnfiner;J.ent. In papers by Nielsen, Olesen, l'u.r::.-
si et a.l. [ 4 3- 4 5] a simple are,-ument ho.::; been propoGcd why one
has confinement in a random vacuuw. '.;.'he assumption that in
V!.l.cuum there exist random fields with nontrivial topology ho.s
been based on trio vucutu:J. instc.bili ty [ 4 3' 46] •

questions the U:!"'SVH"~·s to •sllich .s.hould. be c;iVcn by the eouplete

dynunic,s.l -::heor;, ol vucuum, r·emu:ir.ed un.solvcd. .. 'fhc rew<Jl-LG oo-
tuined a:t thP in•J,-,-:::1--i ··c•:io-.1 of u onc-dimf!nsinnal L,odel nf
:;o.ut;e f'iclC.s ( Y:..Imi) allows to hope that the assurnption of ran-
1

aomness ha.s a deep clyr:11..1.rnicf-<l origin ..

b. Conclusion

The basic purpose of this paper is to ilwt=;stigate ·~he

Y~o.ng-i,:ills classical equations in the case when the vector po-

1;er.tln.1 d.epends on the time only. 'l'he PS of the Yl:1Ch1 is sho·.m

to be a ?,olmogorov K-system and, hen"~, possess strong statis-

tic~J.l ,;irOj_>crties. l'he.sc trajectorie s of this system by the)..r

s~ructure are si:nila.r to the trajectorie s of the Krylov-Sina i

billiurds system • .L~ke eve1y X-system the :B'S of the YJ.1C1.1 has

a pot:;::. ·~::.vc Z:oll!logo:rov entx·opy, a mixing of any rnul tiplici ty ~

·:c_,_; "che .>_r.ect:':';;r.I of one-parru.ne tric group of unitary operators

con.jucated to the flaw TyM j_s n. countably-m ultiple Lebesgue

ono ..
i..!onsiG.ered ~.:.'e e.lso the r.:yste.':1s with cpontaneous synunetl'Y

ti:3 ;_.ystcr.. i.s close to the .Lnte(.::rc.ble one, ur.d the phB.se soncc;

is •i.11 tbe• r.L'1:!n cownoscd ,~.f ·~nvt.J.r.i_ur.t torcD. ','hen, Vlith the

,:!r;cre~>.'->e :!.n ~~ , oy ;.l.eC!.r.s of the infinite :Jequcmce of ro.piU.ly

p;.rowinr, bii'u.r'cat:i.on s the sys'ter:1 becol:len stochastj_c. ·dl ~:w~e

resuli~> :.:~re c.f u thcc.:.:·etJ.c:: ..lly-no:.:lpe!' ·turbative cha.ra<.:tcr and

ct:..n1.10t be obtu:;.nco iH u.ny fini tc order of cl(~Gsic:.:.l pe:tturbn-

tior: theor;ya
symmetry in the region Jf ._. 0o where one may still mnkc use
of conserving integrals, is close to t)'l.at of harmonic oscillu-
tor. With the decrease in !IT there occurs a relevelling o.nU

the degenerate spectrum transforms into the nondegenerute specM

trum of a quantum K-system, whose basic property is the level
repulsion, and their wave fucntions are isotropic ..
In conclusion the author would like to thank H.M..Asatryan
G.L.Balia.n, V.G.Gurzadyan, K.Zalevsky, S.G.Matinyan, B.A.Uo.la-
med, N.G.'l'er-Arutunian-Savvidy and P. 6vi ta.novic for helpful
discussions.

45
 Appendix ,\

In s0ction J the mapping of the phase space hl on itself

depending on the time Tt was cons:i.::lere d. It is natural to

' '
' consider also discrete mappinga which merely repeatedl y uct on

'I-1. 'l'hese mapp:i.ngs are automorph isms [?] . '?hroueh T n.X the

result of n-fold mapp:i..ng is denoted~

11 1
ln the proposed example[ - 3] as a phase spaCe we have
'taken the surface of the tore with geographi c coordinat es: the
longitude and the latitude lf>a • The square Q==; 4J,, 2 ~ 2..91
lf,
on the plane ~~ 1
~e serves as the tore map. One may as well

consider the whole plane ( lf 1

,<.fa ) without distingui shing the

!Points with th.e coordinat es multiple to 2. .f!"

tore automorph ism is determine d by the integer n~~ber
~he

\mirnodula r linear transform ation of the plane (lf,,fa) with the

matrix [ 11 -13]'

(~~)=C :)(~~) U•.1)

rhe action of the transf'orrn ations 1n. n.:;;. 0 on the plane con-
'
sists in u rapid, ·at an exponenti al speed, compressi on in one
direction o.nd' extension in another one. 'rhese di:rection s ore

determine d by eigenvect ors of the transform ation (A.1).

The propertie s of automorph ism (h.1) are well investiga ted
11
in ( - 1 3]. It has a countable number of cycles,·i s coarse
1

i(see the final paragraph s oi' sec. }) and possesses compressi ng

I
and e::.:tending fibers aq.d, hence, is a K-system.
'l'o make this exar.tple look more "physical " let us dro.w o.
square with a black blot in the centre. Under the effect oi' th•
transfor mation (A.1), which is not difficul t to construc t, the
blot, already after several first iteratio ns will tmiforml y
spread over the squ~re (fig. 8).

[!J -
T

l<'ig. 8

.;
i'his is a good mod.e1 of irrevers ible solution s of black ink
1

in a glass of transpar ent water, with the tranafon nation (A.1)

correspo nding to the mixing.

Appendix B

The 6xmple proposed by Arnold b6 ' 30 J

illustra tes hovr
the mixing (K-mi:xin g) appears in the Krylov-S ina.i billiard s
system, and how this property of being a K-system relates to
"the behaviou r of geodesic lines on surfaces of negative curva-

Consider s"imultan eously two systems. In the first the

point moves by the geodesic al on the surface of a triaxial el-
lipsoid of fig. 9a, in the second by the.surf ace of a pretzel
from fig .. 9b. The ellipsoi d has a positive curvatur e whose in-
tegral is 4:1\ , the pretzel has a negative curvatur e Vlhose in-

47
tegral is- 4 !IT , theref ore, the geodes ic flux on the pretze ]

eurfac e is a K-flux , and on the ellipso id

surfnc e it j_s quite

integr able [1J} and, theref ore, non-mi xing.

~llipsoid is
On the other htind, if the sm!:l.ll axis of' the
e U.."l.d the
reduce d to zero, the ellipso id turns into un ellips
billiar ds
gt.::ode sical flux in the lj_mit transfo rmn into the
fig. 10a, with
system in the region limite d by the ellips e of

a) Fig .. 9

the traject orie_s w~.thin the ellipse never being everyw here
·sy-s-~em along the
dense. Let us turn the pretze l in the second

with a hole (.::'ir:o 100)~ o.n.d the ~eodesic l'.low into UJ.e bii.:U_,.

(fig~ 'lOb) (the dotted

urds system o:r, the i,ore w:l.th a hole

;,)

su.1·t'n<~0 'J:f' [;ore-·;,

line denote s -tnc motion in an intern al
 red along the c.i.r-
negat ive curva ture of the pretz el is gathe
cumph erence edge durin g the flatte ning.
s, (2.12 )
Here v1e see a good analo gy with gauge system

\Yhose mo~ion is stabl e withi n the regio

n l) VM < E. o.nd

the bound ary forme d

the insta bility devel ops o.t scatt ering ·an
lly the equip otent ial surfa ce UyM = E of
tlte conve x insid e

. regio n of motio n.

t\
Pig. 1
 Fig. 3

J!ig. 4
 l!'ig. 6

B (ol = 0o)

A(rJ.=2)

Fig .. 7

52
 Fit> 11

I'(b)
f,Q
<A>= 1

Fig. 13

5J
 References

I. MaTIIHllH c.r., CaBBIUl;ll r.K., Tep-ApyTilHffl!-CaBBil;III< H.r. CTo·

xacTH~OCTb Knacc~eCKOft MexaHHKK HHra~ca K ee ycTpa~
HeHHe MexaHK3MOM Xnrrca. ITHcbMa B EOTP, I98I, T.34,c.6I3.
2. MaTI!HllH C.r., CaBBIUl;ll r.K:, Tep-ApyTIOHf!II-CaBBil;III< H.r.
Krracc~ecKM Mexalll!Ea llHra-MlilJJJ!ca. Hemme:ilm!e KOJieelrunm
~eTa. KSTP, 1981, T.60, c.830.
3. Eace!lH r.s •• MaTHHflll c.r •• CaBBE;Il- r.K. HeJII!Hell!me !IllOCKH~
BOJlllll B 6esmccoBol! Teopiil!! HHra-Mwmca. ITHcbMa B l[lJTP,
1979, T.29, c.641.
4. Xa.nMom II.P. Jlei<J1l'lll no spro,lii!:'IeCKOft Teopiil!!. M. :llHocTpa!llia>I
Jli!Tepa~. 1959.
5. Xomji 8, Spro,lii!:'IecKM Teopllf!. YMH ,IY, 1949, Bbtii.I, c.II3.
6. Po)[Jll!H B.A. floB!ill IIPOrpecc B Teoplm !IPeo6paaoBaBI!l! c I!HBa-
pl!aiiTHO!I Mepol!. YMH, XY, 1960, Bbtrr.4, c'.3.
7. Kopll\PeJil.,ll; li.II., CI!Haft ll.r., @oMI!H C.B. Spro)1>l'leCKM Teo-
pllf!. M.: Hay&a, 1980.
8. KoJIMOropoB A.H. HoB!ill MeT~ecRKI! I!HBaplii8HT TpaHSl!TKBI!llX
)1>!H~6CKBx CI!CT6M II aJlTOMOpqMSMOB !IpoCTPaHCTB Jle6era.
nAH CCCP, 1958, T.1I9, ~ 5, c.861.
9. Hedlund G.A. The Dynamics of Geodesic Flows.- BulleAmer.
Math. Soc., ·~939, vol.. 45, N.4, p.241.
10. Xomji 8 •. CmTHcTlll!a reo;I~es~ecRI!X JIIIIIIIll Ha MHOrco6paallf!X
0Tplll1!lT9JlhHol! Rjll<BH311ll. YMH, 1949, T.4, ~ 2, c 129.
n. AHOC,OB ,ll.B. reo;I~e911'<6CRI!6 llOTORI! Ha 3aMKeyT!lX PIIMaHOB!llt
MHoroo6paallf!X OTPIIDI'Tem.Holl K]lRBIISI!ll. M. : Hay&a, I967.

54
 I2. Al!ocoB J\.B., Clll!all !l.r. HeiWTOp!le rJI8,l(Kl!e spro,IOF
<ecKJie
cncTeMH, YMHr.XTII, 1967, BHrr.5, c.107.
13. ApHoJI».II B.l1. MaTeMaTmecKJie MeTO.I\H KJiaccm emwli
MexaHnKJI
M.: HayKa, 1979.
14. CI!Hal! H.r. BepOHTHOCTHRe ".!1811 B ·sprO.ll;l<'!ecRoll Teop11n
•
.!tAll 1961, T.I36, Ji 3, c.549.
15. CI!IJai1 H.r. J\nllaMmecRI!e ci!CTeM!l c ynpyrHMn OTJl8.llte
HIIHMII.
cpro;nnqecRI!e CBOliCTBa paccenBaiOiljiDC 6nJIMp.I\OB. YMH,
1970.
T.XXY, BlliT.2, c.141.
16. ApHOJII>.I\ B.l1. Ma.nHe SHaMeHaTeJ!ll n rrpo6JleM!l ycToll'!I
IBOCTn
.I\Bmt8HIIII B KJiaCCll'!8CKOJ! II He6eCHOii MexaHI!Re, YJ1Ifl,
1963,
T.XYID, Bll!I.6, c,92,
I?. Asatry an H.M., Savvid y G.K. Config uration al Manifo ld of
·Yang- Mills Classi cal Mechan ics ~PI-580(67)-82.

18. Al!.n;poHOB A.A., liOHTPf!I'liiH JI.C, I'py6He CllCTeMR.JlOKJI

.AH CCCP,
1937, T,14, ~ 5, c.247.
19. Illi!JIOB r .E. MaTeMaTll'!eCKiti! a!laJ!ll8 . Cne!lllllJ.U;HH!i
eypc. M.:
\linsmT rl!s, 1960
20. Poinca re !i. Lea method es nouve lles de la mecani que celest~.
I-III, Paris, 1892.
21. Bllplti'OifJ .ll)K,Jl, JlnHaMII'!eCRne cncTem , l;j:..J.[,: rocTex l!s;llaT, 194I;.
22. Ko.nMoropoB A.H. 0 coxpaHemm yCJioBHo-nepno,lliii'IecRI!X -
-
Hllli !Iplf M!IJIOM 113MeJ!eHiflf quHKI.\lll!! I'aMI!Jn.~OHa, JlAH, 1954, T,
98 ,
~ 4, c.527.
23. Fermi E., Pasta J., Ulam s. Studie s of Nonlin ear Problem
I. in: Collec ted \"forks of Enrico .Ii'ermi, Univer sity
of Chi-

55
 cago, vol. II
24. Henon M., Heiles c. The Applicability of the Third Integ_t

ral of Motion: Some Nwnerical Experiments .. - Astron .. J.,

1973, vol. 69, p,?J,
25. ApHom.;!( B,l!. 0 poli\IISHHII yCJIOBHo-rrepno;!(H'!BC!<oro )l;lllll<Sl!ID!
113 ceMei!cTBa rrepi!O;n;II'!SCK!!X ;!(BlllKei!Hi! , .IWl, I96 I, T. I38 ,
II! I, c.I3.
26. ApHOJib;!( B.l!. 0 RJiaccH~ecKoll Teopl!l! B03MymeHRil l! IIJ?OMeMe
yCTOll't!!BOCTI! IIJia!ISTHb!X Cl'lCTSM. J'!1{ , I962, T,I45, )o· 3,
c.4S7.
27, EorOJII060B H.H., MIITpoiiOJII,CJ!Hi! JO,A, AC!!MIITOTII'!SCKIIS MSTO,Il,ll
B TSO:pl'll'l HSJ!KHSl!!rux l<OJ!SOaHRil. M.: ilii!3MaTrl'l3, I958 •.
28. Walker G.. , Ford J. Phys.Rev., 1968, vol. 188, p.416.

29. JlaH.rlaY Jl.)l., .ll!!qmllu E.M. KlraccJ'lqecJ<aJI MeXaHKKa. M.: HayRa·

I965.
30, ApHOJib;!( B.l!. )lorrommTe.J!hliHe rJJa!lll Teopl'lll o6!1l<HOBeHH!IX
;n;mMiepeHIJ;EllJlJ>Hb!X ypaBHeHI!li.M.: l!aYJ<a, I978.
3I. Chirikov B.I. A Universal Instability of Many-dimensit al
Oscillator Systems.- Phys .. Rep .. , 1979, vol. 52, p.263.
32. KpWIOB H.C. Pa6oTH no o6ocHOBaHlllO cTaTl!CTII'ISCROl! qJII3l'lKI!.
113;1(-BO AH CCCP, 1950,
33, I!Jral!l< M•.Co6paHl'!e Hay'l!IIDC TIJY;!(OB. M.: Hay~<a, 1965, c.I07.
34, 3l!HIDTel!H A. C60p!!l'llt Hay'llllilC TIJY;!(OB. M.: HayRa, I966,T,3,
c.407.
35. Hund F. Ueber Zuordnungsfra gen, ins besondere uber die
Zuordnung von Multipletterme n zu Seriengrenzen .- Z.~hys.,

1929, vol. 42, p.601.

36. von Neumann J.,, Wigner E.P. On the Behaviou r of Eigenva-

lues in Adiabati c Processe s, in: Symmetry in the Solid

State, eds. R.S.Knox and A.Gold, Benjamin Inc. NsY.,

37. Teller~. J.l'hys.C hem., 1937, vol .. 41, p.109.

38. ~liJgJ.er Amer.Ma.the, 1951, vol. 53, p.)66; 1955,
E.P.

vol. 62, p.548; 1957, vo1.65, p.203; 1958, vol. 67, p.32?.
39, .llaH,nay JI.n., CMopo,ll;lm<iiWil A.M. JleKUJilll rro TeOJ1BK aToMHoro
li,l~Pa. M.: ~9MaTI'll:9, I966.
40. )l.al!coH !li. CTaTJIC'rll'tec:RaH Teoplilll aHepre'rll'teCKIIX yp01mell
CJIOEIIIIX ClllcTeM. M.: iiHocTJl8HIIM JIHTepaTYpa, 1963.
41. Berry M.V. Quantizi ng a Classica lly Ergodic System: Si-
nai's Billiard s and the KKR Method.- Ann. of Phys., 1961,
vol, 13,1, p.163.
42. McDonald S.W., Kaufman A.~. Spectrum an~ Eigenfun ctions
Phy~.Rev.
for e Hamilton ian with Stochas tic Trajecto ires.-
Lett., 1979, vol. 42, p.1189.
43. Nielsen HeB., Olesen P. Quark Confinem ent in a Random Co-
lor Magnetic Ether.- NBI-HE-7 9-45, 1979.
44. Parisi G. , Rapporte urs talk at Wisconsi c Conf. 1980.
45. Olesen P. Confinem ent and Random Fields.- BUTP-7-81.
46 .. Savvidy G.Ke Infrared Instabil ity ~f the Vacuum State .of
Ga~ge Theories ~d Asympto tic Freedom .- Phys.Le tt., 1977,
vol, 71B, p.13J.

The manuscr ipt was received 15 October 1982.

 r.K.CABBlWl
KJIACC'IECKAll I! KBAHTOBAii MRX1IHl!KA HEAEEJ!EBI:lX
KA.Jll!EPOBO'lH!lX rro!lEil
(aa ai!I".>ml!CKOM Jillm<e, nepeBO.ll JI.H.Bar;:r.acapm;a)
EpeBaHCKi'Il! ([ili3K'reCKi'Il! I'!HCTJ<TyT

Pe~~aRTop Jl.IT.MyxaaH
Tex.pe,llaKTOp A.C.A6paMRH
2a&a3 I06 Bili- 04334 Tl'!pBllt 299
!Ipenpi'!HT Elill Oopt<laT H3,llaHIDI 60x84/I6
!Io.!IITHcaHo K ne'ffiTJ< ~2/IY-b3 3.5 y'I.-H3)1.JI. U.53 K •
. vla)laHO OT,lleJIOM aay'!Ho-TeXHI'!'IeCKOll liHWCpMaiJ;I!II
EpeBaHcKoro @Hai!'Iemmro lii!CTHTyTa, EpereH 36, MapKap.<ma 2
HH,QeKC 3624