• Aucun résultat trouvé

Transport of quantum states of periodically driven systems

N/A
N/A
Protected

Academic year: 2021

Partager "Transport of quantum states of periodically driven systems"

Copied!
15
0
0

Texte intégral

(1)

HAL Id: jpa-00212403

https://hal.archives-ouvertes.fr/jpa-00212403

Submitted on 1 Jan 1990

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Transport of quantum states of periodically driven

systems

H.P. Breuer, K. Dietz, M. Holthaus

To cite this version:

(2)

Transport

of

quantum

states

of

periodically

driven

systems

H. P.

Breuer,

K. Dietz and M. Holthaus

Physikalisches

Institut, Universität Bonn, Nussallee 12, D-5300 Bonn 1, F.R.G.

(Reçu

le 17 août 1989,

accepté

sous

forme définitive

le 23 novembre

1989)

Résumé. 2014 On traite des

holonomies quantiques

sur des surfaces de

quasi-énergie

de

systèmes

soumis à une excitation

périodique,

et on établit leur

topologie

globale

non triviale. On montre

que cette dernière est causée par des transitions

diabatiques

entre niveaux au passage

d’anti-croisements serrés. On donne brièvement

quelques

conséquences expérimentales

concernant le

transport

adiabatique

et les transitions de Landau-Zener entre états de

Floquet.

Abstract. 2014 We discuss the

transport of quantum states on

quasi-energy

surfaces of

periodically

driven systems and establish their non-trivial structure. The latter is shown to be caused

by

diabatic transitions at lines of narrow avoided

crossings.

Some

experimental

consequences

pertaining

to adiabatic transport and Landau-Zener transitions among

Floquet

states are

briefly

sketched. Classification

Physics

Abstracts 03.65 - 32.80 - 42.50

1. Introduction.

Periodically

driven

quantum systems

are of very broad

physical

interest and are

experimen-tally

readily

realised

by

studying

laser/maser - matter interactions.

Among

the many facets of related

phenomena

we shall address theoretical

problems

and their

experimental

repercussions

which

specifically pertain

to the

regime

of

strong

laser/maser -

fields ;

strong

field - matter interaction cannot be described

by perturbation theory, non-perturbative

methods have to be

applied.

The

particular problem

we shall treat in this note is the

transport

of

quantum

states

[1]

by

variation of external

parameters,

i. e. their adiabatic - diabatic motion.

In order to

get

a clear-cut

physical picture

let us consider a

strong

single

- mode laser field

interacting

with a

quantum system

(atom,

molecule or,

generally speaking,

any

mesoscopic

system)

which is described

by

the

following

set of Hamiltonians

where

(3)

we

explicitly

indicated the

parameter

dependence

of H :

À --+-

strength

of the laser/maser field

to laser/maser

frequency

e1

cp =

arg

-

- phase

controlling

the transition from

linearly

to

circularly

polarised plane

e2

wave

fields ;

Ho

describes the static

system,

the

dipole approximation

is not assumed at this state of

argument,

D is the matter current

density.

Introducing

now the

région

A in which the external

parameters À,

(J), cp can be varied

(in

a

given

experiment)

and

we can formulate our

problem

as follows : a time-variation in an interval of

length

Ta

of external

parameters

is

represented

as a curve r in

parameter

space

the

problem

of

quantum

transport

is now to

study

the behaviour of states, i.e. of solutions of the

Schrôdinger equation

under motions on the curve r.

A

particularly

transparent

description

emerges if we use

Floquet

states

[2, 3]

as a basis

[4].

We write the

general

solution of the

Schrôdinger equation

as a

superposition

of

Floquet

states

where the

« stationary »

states

Ua (t)

obey

Of course, ua and the

quasi-energies Ea

do

depend

on the

parameters

R

adiabatic motion of the state

gi

is characterised

by

the

requirement

that the

probabilities

1 aa 12

in

(1.5)

remain constant for a

sufficiently

slow

change

of

parameters

along

T and are thus

independent

of R.

Speaking

in terms of

experiments,

parameter

variations on curves T are well known : the

(4)

curves

necessarily

embedded in

higher

dimensional

parameter

regions

appear if in addition

to a

varying

field

strength

the

frequency cv

and/or the

phase

cp are modulated. Closed

curves T

are of

particular

interest if at least two of the three

parameters

vary in time such that a

non-vanishing

surface is enclosed. The ad libitum realisation of such curves is

doubtlessly

an

experimental challenge,

we

believe,

however,

and

hope

to convince the reader that new

physics

can be uncovered in

experiments

of this

type.

The

requirement

of

periodicity

for the «

stationary

» states ua in the

eigenvalue problem

(1.6)

entails energy conservation modulo W , i. e.

and the

quasi-energy

spectrum

is

arranged

in Brillouin zones. We assume a discrete

spectrum

of the static Hamiltonian

and

anticipate

a choice

for {e j}

such that

we then have

Nonetheless,

the

quasi-energy

spectrum {Ea}

is,

in

general,

dense. For our context this fact has as a consequence that the adiabatic

theorem

cannot be proven in its

original

form for

periodically

driven

quantum systems :

an essential

part

in its

proof

is the

requirement

that the state to be

transported adiabatically

be

separated

in energy from all other states of the

system

[5].

On the other

hand,

simple

model calculations

[6, 7]

have

clearly

demonstrated the

validity

of adiabatic

concepts ;

more realistic calculations

[8, 9]

pertaining

to the

problem

of microwave ionisation of excited

hydrogen

atoms

[10]

have shown the

importance

of adiabatic evolution for the

interpretation

of

experimentally

observed features. Numerical evidence of this kind and new

experimental

support

[11]

for the

concept

of diabatic - adiabatic

transitions at avoided level

crossings

may be taken as sound

guide

- lines in the

following

argumentation.

We start

by dividing

the

quasi-energy

spectrum

into sets of

specific

symmetry

of states, sets

which we denote

by { ê a } sym

The

point

is here that levels

belonging

to the same

symmetry

class do not cross

[12]

in the

generic

case, avoided

crossings

(AC’s)

appear as characteristic

patterns.

A

typical

feature of

periodically

driven

systems

is that AC’s emerge, in

general,

densely

distributed : the

set (s/) sym 8>

JL contains dense

patterns

of AC’s.

Not all of these AC’s are

dynamically

relevant for the

description

of the

parameter

motion

(1.3).

In a recent

publication

[13]

we have described a

simple

selection

procedure : given

an

(5)

with

5 e = level distance in a Brillouin zone

OT = time which the

system

spends

in the

vicinity

of the AC

(more

precisely speaking

OT

=

15R 1/1 Ji

1

where

1 5R

1

is the

parameter

distance within which the level distance

changes by

a factor of

J2).

The levels connected

by

these AC’s

determine a finite dimensional

subspace

of states

{ua }

eff

which,

in a very

good

approximation,

effectively

suffices to describe the diabatic-adiabatic motion of

quantum

states in

experiments

performed

with the very

set-up

considered :

where the aa do

depend,

in

general,

on time

during

the motion

(1.3)

on the

parameter

curve

r.

We are thus led to the discussiôn of a finite set of smooth effective

quasi-energy

surfaces

It so

happens

that

quasi-energy

surfaces in this set

get

very close :

they

almost touch

along

curves

of,

in

general,

finite

length.

A cross section of this

approach

is a very narrow

AC,

very narrow in the scale set

by Ta

defined in

(1.3)

and used for the construction of

{ Ua }

eff’

Dynamically,

these narrow AC’s are

ignored

while

moving

on T in the sense that the

system jumps

across from one

quasi-energy

surface to the

adjacent

one with

probability

very

close to one.

Geometrically speaking,

on the other

hand,

this line of narrow AC’s

plays

the role of a

twist,

the two surfaces should be visualised as manifolds cut

along

the line of narrow

AC’s with cross-wise identified

borders,

the cross-wise structure

being

due to the

purely

diabatic nature of transitions at these narrow AC’s. To

get

this

point unequivocally

clear :

having

set the time-scale

Ta by specifying experimental

conditions we are led to a

coarse-graining

[13]

of the

quasi-energy

spectrum

and thus to the construction of an effective set of

states

{ua} cff’

narrow AC’s

imply

diabatic transitions

and, hence,

cross-identification of

quasi-energy

surfaces the

geometric

purport

of which we express

by

taking

the two surfaces

diabatically

connected as two branches of one

geometrical object

connected in the twisted

manner

just

described. Needless to say, this is

again

an

effective, however,

consequent

construction

translating

into

geometry

the

dynamic

notions of diabatic-adiabatic and

purely

adiabatic transitions in

periodically

driven

quantum systems.

Clearly,

for a

sufficiently large

parameter

region

M the set

{ua}

cff

will,

in the

generic

case,

appear as one

geometrical entity

in which all the states ua are connected to all the other states

by

the above described cut ; the

assignment

of

quantum

numbers for its various sheets u,,,, is a matter of convention fixed

by

the choice of

parameters

R.

In the

following,

we shall describe the

precise

construction of such an « universal

covering

»

in some detail and

discuss,

in

particular,

connections for adiabatic motion and their

implications

for

generalised

Berry phases

[14].

Before

going

into the details we should state

very

clearly

the

logic underlying

our construction :

starting

with the

heuristically

determined

effective set of

dynamically

relevant

states {ua} cff

which suffices to describe the

dynamics

of

(6)

structures

responsible

for diabatic - adiabatic motion. It should become clear to the reader that even basic

experiments

on this

type

of

questions

need such a framework for an

adequate

theoretical

explanation.

The

following

two sections will first introduce the necessary mathematical

concepts

and then discuss their

physical implications.

2. Formal considérations on diabatic-adiabatic motion.

Motion on a curve r in parameter space

of a

system

driven

by

a

single

mode laser/maser field is

govemed by

two time-scales : the

laser/maser

period

which

is,

of course, in

general, T dependent,

and the time

Ta

required

to cover the curve

T. We treat these scales

separately by

first

defining

the

eigenvalue problem

(1.6)

explicitly

as

e.g. for a non-relativistic n-electron

system,

3C R

acts on functions ua

(rl,

..., rn,

t).

The

« slow » motion on r is

integrated

[8]

by

means of

w

(t ) :

= e

(t, t )

is then a solution of the

Schrôdinger equation

(1.4)

including

the

parameter

variation on r.

For every

R (T )

the set of effective

states {ua} cff

spans a finite dimensional

complex

vector

space

z

The fundamental mathematical structure

[15]

underlying

the diabatic - adiabatic motion of

our

system

is the N-dimensional

(N

= dim

FReff)’

complex

vector bundle B over the base space

M

The

projection

defines fibres

Because of the very construction

of { ua} eff’

to a

high degree

of accuracy, motion on

r with the fixed scale

Ta

never drives the state of the

system

out

of 9 ;

motion

from,

say,

(7)

We write a solution

et> a

(T )

of

(2.4) (we

do not indicate the

t-dependence

explicitly)

obeying

the « initial » condition

as

and

stipulate

that

and

Inserting

into

(2.4)

and

demanding

that the

projection

of the

right-hand

side onto the fibre

F,,R4’) vanish

we obtain

the

equation

for the

transport

matrix

Uf3a (T)

thus

defining

the gauge field

A

on the bundle t which effectuates

parallel

transport

of

quantum

states for

parameter

motion on r.

Explicitly

we find

where

is the scalar

product

in the Hilbert space

underlying

the

eigenvalue problem (2.3),

a(J)

and dots indicate differentiation with

respect

to (J) and T

respectively

which,

when

acting

on

periodic

functions with

period

T = 2 TI’ / (J) (T),

acts

only

on the Fourier

components,

the

shift of the Fourier basis

being already explicitly

taken into account in

(2.13).

Introducing

the scaled variable s =

T /Ta

we see that the first two terms in

(2.13)

are

proportional

to

Ta

whereas the last term is

independent

of the scale

Ta

(if

the curve

r is

kept

fixed)

and defines a hermitian connection 1-form

hermiticity

is immediate because of

« uo 1 uy >

=

S f3’Y ; .ae

can be taken as a

Lie-algebra

valued 1-form :

(8)

A concurrent

gauge-structure

now emerges

[16]. Changing

the basis in the fibre

Feff

induces a transformation of the connection 1-form

The

corresponding

curvature 2-form

reads

explicitly

We now consider closed curves

T,

i.e.

and write the formal solution of

(2.12)

as

where we

decomposed

(2.13)

and used the

symbol

3’ to indicate the

T ordering

to define the

exponential

integrals

in

(2.21).

The holonomies

U(l’) ;

i.e. the

images

of all closed curves in

parameter

space

mapped

into

the set of solutions of our

transport

equation (2.12)

or,

rather,

the

holonomy

group

closed curve

through :

contain information on the

topological

structure of the

bundle t ;

the

non-triviality

of

g,

for

instance,

is related to its curvature.

From

(2.21)

we have

Assuming

the existence of a surface S on which A is

smoothly

defined such that j’ = aS we find from Stokes’ theorem for the

geometric phase

factor

(9)

To recall a familiar

example,

let r =

S

1 and

denote the upper and lower

half-sphere

connected to it. We then have

and, hence,

The

quantisation

[15]

of the first Chern class

again

appears as a

consistency

condition on the

holonomy

U(T ).

We now come to the essential

point

of our

discussion,

namely

to establish a relation

between the time-scale of the motion on T and the form of

energy-surfaces

in the

corresponding parametrisation

R =

(À, (J), cp) (1.2).

The situation

qualitatively

shown in

figure

1 is

relatively uninteresting.

The

quasi-energy

surfaces are

separated by

finite,

approximately

constant gaps which are bounded from below such that for

sufficiently large

Ta

the adiabatic limit is attained. The

non-diagonal

elements of the

gauge-field A

are

exponentially

small such that the

equations

for the

transport

matrix

Ul3a (T)

decouple :

Fig.

1. - A sketch of

quasi-energy

surfaces tat’ ..., taN

plotted

as functions of parameters

(10)

and

where

Initial states

propagate

independently

with their « own »

Berry phase

[14],

the

holonomy

group is

simply

the direct sum of N

copies

of

U ( 1 )

A richer case is

qualitatively

shown in

figure

2. The two

quasi-energy

surfaces can be visualised as

being

constructed

by moving

a double cone

along

some line of finite

length

and

then

cutting

out a small

region containing

this

line,

smoothing

the cut, of course. The

typical

feature is now that our

system

e.g. moved

along

the closed

loop F depicted

in

figure

2

Fig.

2. -

Quasi-energy

surfaces Eal,

Ea2 plotted

as functions of k, co

showing

narrow

(S1 )

and broad

(S2 )

avoided

crossings

in cross-sections

Si

and

S2.

The line

separating

the two « roofs » has to be

visualised as

separating

them

by

a line of narrow AC’s as indicated in the cross-section

Sl.

For time scales such that transitions induced

by

e.g.

moving

the system in

Si

are

(almost)

completely

diabatic,

opposite

surfaces of these roofs are identified and a square-root like Riemannian surface appears

giving

an effective

description

of our system. A closed parameter curve r

passing through

a narrow and a

(11)

encounters two very different AC’s : a very narrow and a broad one, the

corresponding

cross-sections

Si

and

S2

are likewise shown in

figure

2.

We now demonstrate the relation between the time

Ta

needed to run

along

r and the

geometry

of the bundle 6.

i)

Certainly,

there is a scale

Ta

such that the

system

passes

Si

as well as

S2

adiabatically

and

the

geometry

of 6 is characterised

by

(2.30),

the situation

being

identical to the one shown in

figure

1.

ii)

If we now increase the

speed

in

traversing

r,

i.e. decrease

Ta,

we shall reach the

point

where diabatic-adiabatic

(Landau-Zener)

transitions occur at

S1.

The

transport

matrix obtains

non-diagonal

elements ;

the

holonomy

group has a non-abelian structure.

iii)

A further decrease of

Ta

leads to

(almost)

pure diabatic transitions at

Si

and adiabatic transitions at

S2.

In the latter case the

system,

starting

on one

quasi

energy

surface,

jumps

on

the other surface and remains there until the

starting point

on the

loop

T is reached

again :

in

traversing

the

loop

T the

system

has

changed

its

quantum

state

The energy surfaces Eal and Ea2 are thus

melted,

under the

premiss

that time-scales be such

that the line of narrow AC’s

along

the double-rim

displayed

in

figure

2 cannot be resolved and

purely

diabatic transitions take

place,

into one surface which

has,

topologically

speaking,

the

structure of a

square-root

Riemannian

surface,

the outermost

point

in

figure

2

being

the

branch-point.

Hence,

the diabolical

point

of a double-cone structure can turn into a square-root branch

pÓÎ11t

when the

system

under consideration is

periodically

driven with a

sufficiently large

amplitude.

The

integration

of e.g. the form

(2.15),

therefore,

does not

encounter any

singularity

and is well defined.

Geometrically

speaking,

iii)

means that states

corresponding

to

parameters

on the line in u1t

tracing

out the narrow AC’s have to be cross-wise identified in order to model the diabatic

jumps

thus

introducing

non-trivial

topological

structure into the bundle 6. To see this in some

detail,

let us

identify

the fibre

F,,R40)

with

C2

The cases

(i)

and

(üi)

with

(2.32)

then

correspond

to the holonomies

and

respectively.

Diabatic

jumps

reflect themselves in the

geometry

of 6

locally

as

topological

structures of

square-root

type

Riemannian surfaces.

In

general

terms, the

point

to be noted here is that it is the effective adiabatic theorem

which introduces non-trivial structures in the set of holonomies. We see that the bundle 6 in

our

example

has,

roughly

speaking,

the structure of a Môbius band : to construct the

identity

element the

appropriate

loop

r has to be followed twice. The observation that the notion of

an effective adiabatic theorem contains a time

scale,

either of intrinsic nature or

given by

the

experimental

apparatus,

leads to the

important physically

observable fact that the

connectivity

of

quasi-energy

surfaces,

tested

by

holonomies

U( ),

is not an a

priori

notion but rather

(12)

The

physical

realisation and

experimental

implications

of such

geometrical

features of 8 will

be discussed in the

following

section.

3. Conclusions.

The

non-perturbative

character of

strong

laser/maser - matter interaction is most

effectively

taken into account

by treating

its

dynamics

in a space of

Floquet

states ; in order to illustrate their

importance

for

quantum transport

phenomena

we discuss a

simple example

in a

separate

appendix.

Considered as a function of external

parameters

the

quasi-energy

spectrum

has a

complicated

structure even if the static

spectrum

is discrete : avoided

crossings

are

densely

distributed over any

region

M of

parameters

which we took as field

strength, frequency

and

polarisation angle

in the case of

single-mode

laser/maser interaction.

The observation that avoided

crossings

appear in very distinct

scales,

extracted from model calculations

(see,

e.g.

figure

1 in

[13]),

led us to the construction of effective sets of

Floquet

states { ua }

eff which describe the

dynamics

of an atom, molecule etc. for a

parameter

motion on curves F in it.

Physical examples

are interactions with

pulsed

laser/maser

beams,

switch-on

processes,

atomic and

molecular

beams in cavities etc. which introduce a time-scale

Ta.

This scale set, states

in {ua} eff

are, at a fixed laser/maser

frequency,

connected

only

via

avoided

crossings producing

Landau-Zener transitions. Seen in the 3-dimensional

region

f1,

states in

{ua}

eff are,

however,

also connected

by

lines of narrow avoided

crossings

where our

system

jumps diabatically

between

adjacent

states on the time-scale

Ta.

The set of states

{ua} eff’

hence,

seems to be connected at

square-root type

branch-points

the

square-root

character

resulting

from the cross-wise identification of states at narrow avoided

crossings.

The

typical

situation is

depicted

in

figure

2 where the diabatic transition in

Si

leads to the

cross-identification ;

starting

at

parameters

R(0)

with a state

ua1R(0)

a modulated laser/maser

pulse

of

length Ta

will lead us

along

the curve r to the state

UR(T.)with

R ( Ta )

=

R(0) :

we do

not end up in the same state, the Ua, a E

Xeff

are multivalued functions of the extemal

parameters,

it is

only

an even number of

pulses

that will lead us back into the same state.

We took care of this

non-uniqueness

of

Floquet

states

by

constructing

the bundle t defined in

(2.6)

to

(2.9)

where the

assignment

of

quantum

numbers to the states u a is a matter of choice of

coordinates,

quantum

numbers

being

a

path-dependent

concept.

The set

{ua}

eff of effective states is thus assembled into one structure, the bundle 6 which

represents

a

« universal

covering »

of the

parameter

space M. On such a

bundle,

adiabatic motion is a

well-defined

procedure,

multi-valuedness is translated into non-trivial holonomies on 8. On the other

hand,

this bundle construction is also necessary for the discussion of adiabatic and Landau-Zener

dynamics : quasi-energy

surfaces

separated by

avoided

crossings leading

to Landau-Zener transitions in one small

parameter

region

will be connected

by

narrow

avoided

crossings

in a

parameter

region sufficiently enlarged,

the latter

leading

to

purely

diabatic transitions

and, hence,

introducing

non-trivial

geometry.

There is a

simple

line of

reasoning

which

helps

to visualise the occurence of structures like the one shown in

figure

2 : choose w such that for

given integer n

and static

eigenvalues

E J O) ’ E k (0)

and, hence,

representatives £a.

J and

£ak

ion

a Brillouin zone are

exactly

degenerate

at

À = 0. For

non-vanishing À

these levels will

repel

each other and the

resulting

AC will be

(13)

Ta

is the

corresponding dynamical

mechanism.

Increasing

the field

strength À

leads to an

increase of

5 e,

for 6 E dT = h Landau-Zener transitions become

important

and the «

branch-point »

in

figure

2 is reached. We checked this scenario in model calculations similar to the

ones

reported

in

[6, 7].

In

particular,

we considered a

one-dimensional,

periodically

driven

electron in a box and

reproduced quantitatively

the structure which we

displayed

in

figure

2 as a

qualitative,

paradigmatic

example

relevant for the construction of a non-trivial

holonomy

(see (2.33)).

We

repeatedly emphasised

that the

position

of this

« branch-point » depends

on

Ta :

in the

vicinity

of each such

point

the wave function

splits

into a

superposition

of the two wave functions of the AC-levels.

Thus,

purely

diabatic and adiabatic motions are

occurring

in

parameter

regions

M with holes around « branch

points

», the holes

being

chosen such that

parameters

leading

to Landau-Zener transitions are removed from M.

An

experimental

verification of elemental structures as the one

depicted schematically

in

figure

2 has to cope with the

following boundary

conditions : first of

all,

the time scale

Ta

faces an upper limit which is either due to

spontaneous

emission or to the continuous

part

in atomic and molecular

spectra

leading

to a finite life-time of

Floquet

states. The

question

at

stake, is,

of course, whether or not the time-scale

Ta

can be chosen

large enough

for effective adiabatic motion to be dominant.

Furthermore,

one has to search for

systems

where

technically

feasible

loops

r in

parameter

space do not enclose too many branch

points

and AC’s in order not to obscure the basic

square-root type

connectedness in

{ua} eff

and the

corresponding

ramifications in the effective space of states caused

by

Landau-Zener transitions.

This does not mean that the construct of a bundle 6 of

physical

states is

meaningless

for the

description

of,

say, microwave ionisation of

hydrogen

atoms

where,

in certain

parameter

regions

4t,

the reaction is dominated

by

a

large

number of

close,

overlapping

AC’s. On the

contrary,

it is our belief that even a statistical treatment of such

agglomerations

of AC’s has to

take into account the non-trivial

topological

structure of 8 : an

understanding

of such

experiments

in more detail will

require

a

deciphering

of these

global geometrical properties

of

quasi-energy

surfaces.

In

preceeding

publications

[8, 9]

we have shown that a

simple purely

quantum

mechanical

frame for the

description

of microwave ionisation of excited

hydrogen

atoms and related

phenomena

is constituted

by invoking

Landau-Zener transitions in an effective set of

Floquet

states :

typical

scales for the

parameters

of the

corresponding

AC’s and their distributions

over

parameter

regions

M are

responsible

for

typical

features

of,

for

instance,

ionisation

curves. In this paper, we extended these notions

pointing

out non-trivial

geometrical

structures caused

by

diabatic transitions which

prevail

in

significant

sets of external

parameters.

It is

doubtlessly

an

interesting challenge

to hunt for

experimental

verification. On the theoretical

side,

an

understanding

of such a rich but

elementally

quantum

mechanical

scenario in terms of a semi-classical

approach

to

classically

complicated,

chaotic

phenomena

is

an

outstanding

problem.

Remains the

question

of the continuum. As we have remarked in

(1.10)

to

(1.13),

a discrete

spectrum

of

Ho

leads,

in the

generic

case, to a dense

quasi-energy

spectrum ;

a continuous

part

of the

spectrum

of

Ho

to a

quasi-energy

spectrum

comprising

the whole real axis. The

provocative question

arises of how any

computer

simulation of the

dynamics

of

periodically

driven

quantum systems

should ever

distinguish

the two cases ! It is

again

a

question

of

time-scales and

coarse-graining

and

leaving

out

dynamically

irrelevant states which can be

applied

(14)

Acknowledgment.

This work was done in

part

at the Harvard-Smithsonian

Observatory,

one of us

(K. D.)

would

like to thank Alex

Dalgarno,

Kate

Kirby

and

George

Victor for their kind

hospitality.

Fruitful

discussions with B.

Zygelman

are

equally acknowledged.

Appendix.

To illustrate the fact that

Floquet

states

provide

the

appropriate

basis for a

description

of

adiabatic

transport

in

periodically

driven

systems

we

give

an

explicit example

for an abelian

phase

factor. Let us consider a

quantum system

interacting

with a

linearily

polarized

radiation

mode described

by

the

Floquet

Hamiltonian

with variable

phase

5. It is easy to see that the

quasi-energies

Sa are

independent

of

8 and that the

general

solution of

is

given by

with

8-independent

Fourier modes

uam.

We now vary 6 from 0 to 2 ir and obtain from

(2.29)

the

Berry-phase

Since

we

finally

arrive at

As an observable

quantity

this

phase

factor

does,

for £a -

c

+ mw, not

depend

on m, i.e. on the second index in

(1.13).

Furthermore,

the role of

Floquet

states for adiabatic

transport

is reflected in the

explicit

occurence of the

Floquet index E a .

It is instructive to compare the

expression

(A6)

with the result of Bamett et al.

[17].

Considering

a two level

system

with level

spacing

Wo and

using

the

rotating

wave

approximation

these authors obtain the

following

expression

for the

phase

where

(15)

To compare

(A7)

with out formula

(A6)

we have to calculate the

Floquet

indices for the same

system

and obtain

Therefore we have

and we see

(A7)

to be a

special

case of the

general

formula

(A6)

which is neither restricted to a two level

system

nor to the

rotating

wave

approximation.

We remark that

(A6)

could also be of

experimental

interest in order to

distinguish

an

avoided

crossing

in the E -

w-plane

from a real

crossing by

means of a pure

phase

variation

since at an avoided

crossing

we have

2 ’TT dEa/dW = 0

mod 2 ?r.

References

[1]

AVRON J. E., RAVEH A., ZUR B., Rev. Mod.

Phys.

60

(1988)

873.

[2]

ZEL’DOVICH Ya. B., Sov.

Phys.

JETP 24

(1967)

1006.

[3]

FAINSHTEIN A. G., MANAKOV, N. L., RAPOPORT L. P., J.

Phys.

B 11

(1978)

2561.

[4]

OKUNIEWICZ J. M., J. Math.

Phys.

15

(1974)

1587.

[5]

BORN M., FOCK V., Beweis des Adiabatensatzes, Z.

Phys.

51

(1928)

165 ; AVRON J. E., SEILER R., YAFFE L. G., Comm. Math.

Phys.

110

(1987)

33.

[6]

BREUER H. P., DIETZ K., HOLTHAUS M., Z.

Phys.

D 8

(1988)

349.

[7]

BREUER H. P., DIETZ K., HOLTHAUS M., Z.

Phys.

D 10

(1988)

13.

[8]

BREUER H. P., HOLTHAUS M., Z.

Phys.

D 11

(1989)

1.

[9]

BREUER H. P., DIETZ K., HOLTHAUS M., J.

Phys.

B 22

(1989)

3187.

[10]

BAYFIELD J. E., SOKOL D. W.,

Phys.

Rev. Lett. 61

(1988)

2007 ;

GALVEZ E. J., SAUER B. E., MOORMAN L. , KOCH P. M., RICHARDS D.,

Phys.

Rev. Lett. 61

(1988)

2011.

[11]

VAN DER WATER W., VAN LEEUWEN K. A. H., YOAKUM S., GALVEZ E. J., MOORMAN L.,

BERGEMAN T., SAUER B. E. and KOCH P. M.,

Phys.

Rev. Lett. 63

(1989)

762.

[12]

VON NEUMANN J., WIGNER E.,

Phys.

Z. 30

(1929)

467.

[13]

BREUER H. P., DIETZ K., HOLTHAUS M., Bonn

Preprint

AM-89-01

(to

appear in Nuovo Cimento

B).

[14]

BERRY M. V., Proc. R. Soc. Lond. A 392

(1984)

45.

[15]

SIMON B.,

Phys.

Rev. Lett. 51

(1983)

2167.

[16]

WILCZEK F., ZEE A.,

Phys.

Rev. Lett. 52

(1984) ;

MEAD C. A., TRUHLAR D. G., J. Chem.

Phys.

70

(1979)

2284 ;

MEAD C. A.,

Phys.

Rev. Lett. 59

(1987)

161 ;

ZYGELMAN B.,

Phys.

Lett. A 125

(1987)

476 ; LI H. Z.,

Phys.

Rev. Lett. 58

(1987)

539.

[17]

BARNETT S. M., ELLINAS D., DUPERTUIS M. A., J. Mod.

Opt.

35

(1988)

565 ;

Références

Documents relatifs

Given that most of the widely used Internet protocols such as SMTP, NNTP and FTP are all based on commands which are given via ASCII strings, there seems no reason why

This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS

A caching proxy mesh where a request may travel different paths, depending on the state of the mesh and associated caches, can break HTTP content serialization,

Justine Culioli 33 years old, who is a graduate of the Ecole Centrale de Paris, had her first experience in Project Management of various key projects at Bouygues Construction.. She

This report is about recent progress on semi-classical localization of eigenfunctions for quantum systems whose classical limit is hyperbolic (Anosov systems); the main example is

Bell’s inequality has been derived in different ways [1-10], using the fol- lowing assumptions: locality, determinism, hidden variables and coun- terfactual definiteness.. From

From left to right: original relief (gradient of the flower image of Fig. 7), effect of the viscous transformation of size 30 (mercury model: ˜ T ), effect of the viscous closing

In this article, we show that homogeneous nucleation of solid helium does not occur up to 163 ± 20 bar, at least on a time scale of order 100 ns, the typical experimental time in