• Aucun résultat trouvé

Solution of self-consistent field equations by the recursion method

N/A
N/A
Protected

Academic year: 2021

Partager "Solution of self-consistent field equations by the recursion method"

Copied!
13
0
0

Texte intégral

(1)

HAL Id: jpa-00246835

https://hal.archives-ouvertes.fr/jpa-00246835

Submitted on 1 Jan 1993

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.

Solution of self-consistent field equations by the recursion method

J. Julien, D. Mayou

To cite this version:

J. Julien, D. Mayou. Solution of self-consistent field equations by the recursion method. Journal de

Physique I, EDP Sciences, 1993, 3 (8), pp.1861-1872. �10.1051/jp1:1993103�. �jpa-00246835�

(2)

Classification Ph_vsics Abstracts

71.20A 71.20C 71.25C

Solution of self-consistent field equations by the recursion method

J. P. Julien and D.

Mayou

Laboratoire d'Etudes des

propr16tds

Electroniques des Solides (*), CNRS, BP 166, 38042 Grenoble Cedex 9, France

(Receii>ed 18 January 1993, revised 8 March 1993, accepted 9 April 1993)

Rksumk. Nous montrons que les

Equations

de

l'approximation

du

potentiel

cohdrent (C.P.A.) peuvent due r6solues dons

l'espace

r6el par la mdthode de rdcursion. La fonction de Green est

reprdsentde par une fraction continue et la mdthode foumit une solution

rapide

et

prdcise.

Une premidre application est prdsentde.

Abstract. We show that the equations of the Coherent Potential Approximation (C.P.A.) can be solved in real space

using

the recursion method. The Green's function, from which the

density

of

states is calculated, is then represented by a continued fraction, and the method provides a

quick

and

precise

solution. A first

application

is

presented.

1. Introduction.

The Coherent Potential

Approximation [I]

is one of the most

widely

used mean field theories of disordered substitutional

alloys.

Even in

partially

ordered systems it is the basis for a

determination of interaction

potentials

that can lead to

prediction

of

phase stability [2].

This

approximation

leads to self-consistent

equations

for Green's function which are

usually

solved for each energy z. The aim of this paper is to demonstrate a way of

solving

the C-P-A-

equations by

a

completely

different

approach.

We show that the Green's function is that of an effective Hamiltonian H~~~ where the matrix elements are calculated step

by

step. Then the recursion method

[3]

is

applied

to that effective Hamiltonian and allows one to express the Green's function as a continued fraction. This method is

quick

and is also

interesting

since continued fractions have proven very useful to calculate densities of states or

integrated quantities

like energy or

charge

transfer.

(*) Associated with Universitd Joseph Fourier.

(3)

2.

Description

of the method.

For

simplicity

we consider a

binary alloy

with two components A and B

having

on site

(matricial) energies

e~ and e~. The

generalization

for more than two components will be

discussed elsewhere

[4].

The Hamiltonian for a

given configuration

is :

H=Ho+£E, (1)

Ho

does not

depend

on the

configuration (diagonal disorder)

and I denotes the atomic sites with E, = E~ or e, = e~

depending

on the

occupation

of site I

by

A or B atom. If p~ and p~ are the

probabilities

of

occupation by

an atom A or

B,

one has p~ + pB = I and one can

define the average energy

~

" PA ~A + PB ~B

(2)

In this paper we are interested in densities of states which are

calculated,

as

usual,

with the formalism of Green's functions. For a Hamiltonian H the Green's operator G

(z

is defined

by

:

G(z)

=

~

/

~

(3)

and the

partial density

of states

n~(E)

on a state

/$r)

is

given by

n~(E)

= Im

($r/G(E

+

ie)/$r) (4)

where Im

~f)

is the

imaginary

part of the

complex

number

f,

and

e is a

positive

real number

which tends to zero. More

generally

if we consider a

subspace

E

generated by

a set of

N orthonormal states

/$r,)

we define a

projector

on that

subspace P~ by

FE

=

( /~, ) l~,/ (5)

and the restriction of the Green's operator

G(z)

to that

subspace

E is :

G~(zi

=

P~G(ziP~. (6)

If E is the

subspace generated by

the orbitals of an A or B atom we note the restriction of the Green's operator to this

subspace G~(z)

or

GB(z).

Within the C.P.A. an atom A or B is embedded in an effective medium

consisting

of

identical atoms with on site

energies

> +

3(z) [1, 2].

Then the

self-consistency

condition states that the average Green's operator on an A or B atom in the effective medium is

equal

to the Green's operator G~~~(z

)

on an effective atom with on-site energy > +

3(z), placed

in the

same effective medium :

PA

~A(Z)

+ PB

~B(Z)

~

Geff(Z) (7)

G~(z)

=

~ ~

'~

~ ~~~

G~(z)

=

~ ~

~

~ ~~~

(8)

~~~~~~

~

z I

3(z) ~r(z)

~~~

(4)

«(z)

represents the

(matricial) self-energy

due to the

coupling

of an atom with the effective medium. The

self-energy 3(z),

which is

Herglotz [5],

can

always

be considered as the self-

energy due to the

coupling

to a chain C with the Hamiltonian in

figure

I where

A~,

B~ are square matrices with the same dimension as e~ and e~, If there is one orbital per site

A~, B~

are

just

numbers and

3(z)

is a scalar

quantity

which can be

expressed

as a continued

fraction

~2

3(z)

= °

(10)

B(

z-Aj

B~

~

z

-A~

z

-A~

Z(z)

= AI Az Aj C

B° Bi Bz 83

Fig.

I.

Representation

of the semi-infinite chain C. The

coupling

to the chain C is

equivalent

to

adding

the

self-energy

3(z). The circles represent

subspaces

of states whose dimension is

equal

to the number of orbitals of the atoms A or B. For each circle A~ is the on-site matricial energy of the

corresponding

subspace. B~ is the coupling between the two adjacent subspaces of the chain.

If there is more than one orbital per site

3(z

is a matricial

quantity

which can be

expressed

as a matricial continued fraction

3(z)

=

Bj Bo 3~(z )

=

Bj

~

B~

ii

)

Z-Ai-~i(Z) Z-An+i- n+i(Z)

As shown in

Appendix A,

this allows us to

replace

the usual effective Hamiltonian of the C.P.A. which is energy

dependent, by

a Hamiltonian H~~~ which is

independent

of energy, but is defined in a

higher

dimensional space since each atom with a

self-energy 3(z)

is now

coupled

to a chain C which simulates

exactly

the

self-energy 3(z).

Then the

self-energy

«(z)

of the central atom

(A

or

B)

in the energy

dependent

effective Hamiltonian is also that of the central atom in the energy

independent

Hamiltonian H~~~

(Fig. 2).

In this

representation

«

(z)

can be calculated

by

a classical recursion

procedure starting

from A or B. If there is one orbital per site

«(z)

is a scalar

quantity

which can be

expressed

as a continued fraction :

~2

«(z)

=

°

(12)

hi

z aj

b~

~

~ ~~

z a~

If there is more than one orbital per site «

(z

is a matricial

quantify

which can be

expressed

as a matricial continued fraction :

«(z)

=

hi

z ~~

'

«~~z~ ho ~rn(z)

=

hi

z ~~~

'

~r~~ ~~z~

bn (13)

where a~, b~ are square matrices with the same dimension as e~ and eB.

«(z)

can also be

interpreted

as the

self-energy

due to the

coupling

with any chain C'

(Fig. 3).

(5)

E+Z(z) E+Z(z) E+Z(z) E+Z(z)

ffi ---~ O ~ 0~~2a

EAorEB

83 83 83 83

82 Bz Bz Bz

B> Bi Bi Bi

Bo Bo Bo Bo

I ) I I

2b

8A Or 88

Fig. 2. a) Schematic

representation

of the energy dependent effective Hamiltonian of the C-P-A- The

central atom A or B has the on-site energy e, or F~. The other atoms have on-site energy

7 plus a

self-energy

3(z). The atoms are coupled

by

the Hamiltonian Ho which is independent of the atomic

configurations.

hi Schematic

representation

of the energy

independent

effective Hamiltonian

H~j

used in the recursion

procedure.

This Hamiltonian is obtained from the

previous

one

by attaching

a

chain C to each atom. This chain C simulates exactly the effect of the

self-energy

3(2 j.

°~~~* b~8

~~

~

~

~

~

ic,1

Fig.

3.

Representation

of the semi-infinite chain C'. The

coupling

to the chain C' is equivalent to the

self-energy

«(z). The circles represent

subspaces

of states whose dimension is

equal

to the number of orbitals of the atoms A or-B- For each circle a~ is the on-site matricial energy of the

corresponding

subspace. b~ is the

coupling

between the two

adjacent

subspaces of the chain.

Let us consider first the case of one orbital per atom. We note that if we

apply

the recursion method

starting

from an orbital

/$ro)

which is located on the central A or B atom, then the vector [fl~~) of the recursion method has non zero components

only

up to sites

(n

I

)

in the chain

representing 3(z) (I.e.

the

subspace

with on-site energy

A~_ j).

In order to calculate b~ and a~

~ i with :

~n "

I

~eff'i'n)

~n

i'n)

~n

i'n II

I

(14)

i'fi

+

)

"

) (lieff' i'n)

~fi

i'n) ~n i'n II (15)

~

a~

~ =

j*n

+

Heft 4'n

+

1)

~~

(6)

we need to know the matrix elements of the Hamiltonian of the chain C

only

up to

A~, B~_

i with pm n.

Thus,

even if all

A~

and B~ are

unknown,

which is the case at the

beginning

of the

calculation,

we can still calculate the first terms

ho

and aj. If there is more than

one orbital per atom we can use the vector recursion which is

presented by Haydock [3]. Again

it is easy to see that in order to calculate b~ and a~

~ we need to know the matrix elements of the Hamiltonian of the chain C

only

up to

A~,

B~ with p w n. This fact is essential in our

method as we show below.

In order to go further and calculate the

following

terms a~~

~ j, b~~

(with

n ~

0)

the next step is to use

equations (7), (8), (9),

and express

3(z

as a function of «

(z).

We get

(see Appendix B)

:

3(z)

= ~ A

(17)

Z PA E~ p~ EA tT (Z

A is a Hermitian matrix

/~

"

fi(EA ~B) (18)

From the continued fraction

representation

of

3(z)

and «(z) (see

Eq. (I

I) and

(13))

this relation is

equivalent

to

Bo

= A A

i = PA EB + PB EA

('9)

B~ = b~

A~

~ i = a~ for n ~ 0

(20)

Using equations (19)

and

(20)

and the terms

bo

and ai, which can

already

be calculated as

explained

above, we get

Bo,

A and B

j,

A~.

From the above remark about the extension of the

vectors

[w~)

we see that the recursion

procedure

allows us then to calculate

hi,

a~, and

b~,

a~. Then

using equation (20)

this

gives

us

B~,

A~ and

B~, A~.

The recursion

procedure

then allows us to calculate the next terms

b~,

a~, and

b~,

a~.

Clearly,

with this

procedure

we can calculate

progressively

all the terms of the continued fraction

expansion

of

3(z)

and

«(;),

from which we deduce the Green operators

G~(z)

or

GB(z).

In

appendix

C we estimate the extent of memory

M(n)

and the

computing

time

T(n

needed for the recursion

procedure

with H~~~ up to stage n. We find that these

quantities

are greater than the

corresponding quantities M~(n)

and

T~(n

for a Hamiltonian in real space I-e- without chains attached to the atoms of the average medium. For the scalar case we get in the limit of

large

n :

)~(

~~~

ii~)~l'

~~~~

This shows that for a classical value of n

= 20 the

procedure presented

here is

applicable

to many systems.

3.

Application

and discussion.

Here we

apply

the method described above to

study

the electronic structure of a disordered

alloy Ax

B

x within the

simple

s-band model on a cubic lattice without

off-diagonal

disorder.

In a

forthcoming publication,

we will show how the method can be

directly

extended to multi- orbital cases (matricial

self-energy).

The s orbitals on the cubic lattice have on-site energy e~

(respectively e~)

with

probability

p~

(respectively p~)

and are

coupled

to their six nearest

neighbours

via a

hopping integral

t.

(7)

In

figure

4, we present the coefficients of the continued fraction versus the number of recursion steps for various cases of parameters. The

gapless

case

(Fig. 4a)

can be

easily recognized

when the coefficients present

damped

oscillations and converge

rapidly

to their

single asymptotic

value. The continued fraction is terminated

by

the

simple square-root

terminator and the

corresponding density

of states is

pictured

in

figure

5a. When the

coefficients

present undamped

oscillations

(Fig. 4b),

we use the terminator

proposed by

Turchi et al.

[6]

to get the

density

of states, which presents a gap

(Fig. 5b).

As can be seen in this

simple

illustrative

example,

the method

provides

an efficient way for

solving

the CPA

equations

and presents several attractive features. In

particular,

in contrast to

usual

methods,

the Green's functions do not need to be calculated for

complex energies having

a small but finite

imaginary

part and then deconvoluted up to real axis.

Furthermore,

in many

situations,

one needs

only

to know

integrated quantities

like band

energies

or

charge

transfer

(in particular during

the calculation of the self-consistent

potential)

and it is well-known

[6]

that a very

good precision

for these

quantities

is obtained

by calculating exactly only

the first few coefficients of the continued fraction. Our method is

well-adapted

to these calculations.

Finally,

our method avoids

integration

in the

reciprocal

space which may present some

advantages.

We also note that

despite

its

simplicity,

the one-band-model treated here is of interest in several

approximate

treatments of realistic systems. Indeed it is

formally equivalent

to the

o 9

,~ ,

-l ~ ~

-i. i

-1 2

-1.4

-1.5 i~

f~ -1.6 +

b

~

~

3.3

°

u

~ ~

3.2 °

3,1

3

2

2.8

~

5 lo

n

a)

Fig.

oefficients are hi.

a)

For e, = - eB " 2 t pA " 0. I, pB "

p~ = o.9.

(8)

o

,~ ,

-O.5

~

~~

+ + + + + + + +

-l

-1.5

+ + +

+ +

+ +

+ + +

+

~

-2 +

+ +

~ + ~

~ ~ +

m ~

~ 3

~

~

+

'~n'

u

~

~

o

~

4.4

O ~

O O

O O

o o O

O

O ~

~

~ O

4.2

~ o

o o

3.8

3.6 o

~ o o o o o o o

o 3.4

5 lo 15 20 25 30 35

hi

Fig.

4 (continued).

unhybridized

multiband case and the method can be

straightforwardly applied

to the so-called

« canonical »

picture [7] developed

in the L.M.T.O. scheme. In that case, the effect of off-

diagonal

disorder can even be taken into account

by setting

the

hopping integral

t of

tie

reference medium

equal

to the average p~t~ +p~t~ of the

hopping integrals

t~ and t~ of pure materials and

solving equations (7), (8), (9)

where p~

(respectively, pB)

has to be

replaced by

p~

(t~/t ) (respectively,

pB

(tB/t [8].

In a

hybridized

bands model with

several

inequivalent

orbitals per site it is still

possible

to use a scalar version of the C.P.A. as done

by Papaconstantopoulos

et al.

[9].

Each orbital of type I has a scalar

self-energy 3~(z).

For a

given

I the relations

(7), (8)

and

(9)

are still valid with all

quantities

in it

corresponding

to the same I. Then «~

(z)

is calculated

by applying

the recursion method to an

effective medium where each orbital of type I is

coupled

to a chain C~

simulating

3, (?).

One has to

perform parallel

recursions

starting

from all

inequivalent

orbitals of type

j

and this

procedure

is

particularly

suited to

parallel

processor

computing.

To

conclude,

we

point

out that the present

approach

can be

generalized

when there are more than two components and that the calculation can be done within the

reciprocal

space. We also note that the recursion method can be

applied

to other self-consistent theories of disordered systems in the same

spirit [10].

In each case, the effective medium is

represented by

some

effective Hamiltonian

(the

chains C in this

paper)

where the matrix elements are calculated step

by

step

using

the recursion method. This new

approach

to mean field theories of

alloys

is based

on the use of the

configuration

sum space which is

presented

in a

previous

paper

[I I].

It is

(9)

O.14

fs~f O.12

o,i

O.08

O.06

O.04

O.02

g

°

,5b,

O.12

o,i

O.08

O.06

O.04

O.02

-lo -5 5 lo

ENERGY (E/t)

Fig. 5. al Total

density

of states reconstructed by the continued fraction without gap (same parameters

as Fig. 4a) hi Total

density

of states reconstructed

by

the continued fraction with gap (same parameters

as

Fig.

4b).

interesting

to note that this

configuration sum'"I)ace

is

equivalent

to the

augmented

space of

Mookerjee

and collaborators

[12]

in the case of substitutional disorder. However it is more

general

and allows one to treat the case of chemical order. These aspects will be

developed

in a

forthcoming

paper

[4].

Acknowledgments.

We are

grateful

to J.

Kudrnovsky

for fruitful discussions

concerning

the

application

of the C.P.A. method in the L.M.T.O. formalism.

Appendix

A.

Our aim is to show that the Green's function

G~(z)

or

GB(z)

of A or B atom in the effective medium can be calculated

by attaching

chains

C,

to each atom I in the medium. These chains

C,

simulate

exactly

the self energy

3(z ).

Thus we consider the system

represented

in

figure

2b.

We define the

projector P,

in the

subspace

of the orbitals of an atom I and also :

P

=

£P,. (Al)

(10)

We define the

projector Q,

in the

subspace

of all states of the chains C~ and also

Q

=

£ Qi (A2)

Thus

P +

Q

=

(A3)

Using

the Schur

formulae,

we get

PGP

=

~

(A4)

z P HP

PHQ QHP

z

QHQ

However the chain

C,

is

coupled only

to the atom I and thus

PHQ

=

iP, HQ, QHP

=

£ Q, HP, (A5)

Further more, as the chains are not

directly coupled,

it is

straightforward

to show that :

z

$HQ

"

(

z

Q,'HQ, ~~~~

Thus one has

P GP

=

~

(A7)

: P HP

£ P, 3, (z P,

with

~

~"

~'

~

z

Q, HQ, ~~' ~~~~

Thus

coupling

each atom I to a chain

C,

is identical to

adding

the

self-energy 3,

(=

(A8)

to the

atom I as shown

by expression

(A7).

Appendix

B.

We consider two matrices ~r~

and,<B

and two

probabilities

p~ and pB with p~ + pB =

I. Our aim is first to show that :

~~

+

~~

=

(Bl)

'~ '~ ~

pA IA + pB XB pA pB (XA .KB (X~ .KB)

U

with

U

= PA ~B + PB 'A

(~2)

or

equivalently

~~ ~~ ~ ~~ ~~ ~~

~~~~~ ~~~

~~~

~~~

~

PA PB

~~~~

>A X~

(11)

Indeed one has

~

+

~~

" ~I'A XB + PB XA " ~PAXB + PB XA

)

~ (~~)

A B A B B

thus :

~~+~~

=

U

= U

(85)

XA XB XA XB XB XA

We then write

PA xA + PB xB = U

)

~PA xA + PB .<B ~ ~PB xA + PA xB

I

°~A

xA + PB xB

(86)

Developing (86)

we get

PA XA + PB XB " PA PB XA

j

XA + PA PB XB

j

.~B +

P~

XA

j

XB +

P~

XB

)

XA

(~7)

Inverting (85),

we deduce

xA

I

xB = xB

I

xA

(88)

Developing ~p~

+

pB)~

= l and

using (88),

we get

Pi

xA

I

xB +

Pi

xB

I

xA = xA

I

xB 2 PA PB xA

I

xB

~B9)

applying (85), (88)

and

(89)

to

(87)

we get

(82)

or

equivalently (Bl).

In our case, we can choose

x~ = z e~ «

(z) (B lo)

xB = z eB «

(zl' (Bl1)

and we deduce

z

e)~

«

(z

) ~ z

e)~

«

(z)

z

~(z z(z) ~~~~~

with :

2

" PA EA + PB EB

(B13)

and :

~~~~

~

Z PB ~A

A

~B

~'(Z)

~

~~~~~

A

"

fi

(EA

EB). (B15)

(12)

Appendix

C.

It is

interesting

to compare the memory

place

and

computation

time needed for

calculating

one stage of the continued fraction in the effective medium and in real space

(I.e.

without any

coupling

of atoms to

chains).

Let us first define the index m of an orbital as the minimum number m for which

fl~~)

has a non-zero component on that orbital. It is easy to see that if an

orbital is located at the level n and is

coupled

from a chain to an atom with index

m, then the index of that orbital is m + n that is all fl~~) with p < m + n have zero component

on that orbital.

Quite generally,

if we want to calculate the recursion vectors up to level

m, we need to take into account

only

the orbitals with indices I mm- The number of these

orbitals can be estimated in the

following

way. Let us call

N~(m)

the number of orbitals with

index m. Thus in order to calculate

[fl~~)

in real space, we must take into account

ii~(m)

orbitals with

&~(m)

=

f N~~p). (Cl)

p=o

In the effective

medium,

the number of orbitals with indices less or

equal

to m is

& (m

=

~o jj NR~P)

x

(m

p + I

j (c2)

In the limit of

large

m,

N~(m)

grows like m square and we can

approximate

ii~~p)

as

ii~~p)

m

ii~~p)

x

~.

If

we take p

=

20 as a

typical

value we see that the memory 4

place

needed for the effective Hamiltonian is about 5 times that for the real space Hamiltonian which is still reasonable for many systems.

We now compare the number of

operations

needed at step p to calculate

a~

and

b~ in the effective medium and in

real-space (I.e.

without any

coupling

of atoms to

chains).

If [fl~~_

j),

[fl~~) and a~, b~ n w p I have

already

been

calculated,

the next steps are

I)

calculate

H[fl~~)

;

2)

calculate

a~

=

(fl~~[H[

fl~~)

3 calcul ate b~ fl~~~ i

)

= H fl~~)

a~

fl~~ b~ fl~~

)

from which

b~

and fl~~~

)

are

deduced ( [fl~~~

i)

is normalized

by definition).

One sees

easily

that the number of

operations

per orbital in steps 2 or 3 does not

depend

on

the Hamiltonian. On the contrary, the number of

operations

per orbital in step I is

proportional

to the number of orbitals

coupled

to one

given

orbital and this number is

larger

for orbitals in real space than for orbitals in the chains. Thus, if we call

Z~ (respectively Z~)

the number of

operations

per orbitals

(for

steps 1, 2 and

3)

in real space

(respectively

in the

chains),

we

always

have Z~ ~

Z~.

If the number of orbitals

coupled

to one

given

orbital is

sufficiently large

then step I is

large

and Z~ can be

only

a fraction of

Z~.

The ratio R between the numerical work for stage p of the recursion in the effective medium and in real space is thus

Z~

+ Z~

pm

Z~

R p

m m +

-.

(C3)

ZR ZR

4

If we take p

- 20 as a

typical

value, we see that R S 5 since Z~ <

Z~.

We conclude that in many systems the solution of C-P-A-

equations by

the recursion method can be done in a

reasonable

computing

time.

(13)

References

[Ii Soven P., Phys. Rev. B 156 (1967) 809

Velicky B.,

Kirkpatrick

S., Ehrenreich H., Phys. Ret,. B 175 (1968) 747.

[2] Ducastelle F., Gautier F., J.

Phys.

F 6 (1976) 2039 ;

Treglia

G., Ducastelle F., Gautier F., J. Phys. F 8 (1978) 1437.

[3]

Haydock

R., Solid State

Physics,

H. Ehrenreich, F. Seitz, D. Turnbull Eds., Vol. 35 (Academic Press, New York) (1980) p. 215.

[4] Mayou D., Julien J. P., to be published.

[5] Muller-Hartmann E., Solid State Commun. 12 (1973) 1269.

[6] Turchi P., Duscatelle F., Treglia G., J.

Phys.

C IS (1982) 2891;

Turchi P., Thesis (Paris, 1984)

unpublished.

[7]

Kudmovsky

J., Drchal V., Masek J.,

Phys.

Rev. B 35 (1987) 2487.

[8] Sigli C., Thesis (Columbia University, 1986).

[9]

Papaconstantopoulos

D. A., Pasturel A., Julien J. P., Cyrot-Lackmann F., Phys. Rev. B 40 (1989) 8844.

[10]

Mayou

D., Thesis (Grenoble, 1987)

unpublished.

[I Ii Mayou D., Pasturel A., Nguyen Manh D., J. Phys. C 19 (1986) 719.

[12]

Mookerjee

A., J. Phy.I. C 6 (1973) L 205 and J. Phys. C 6 (1973) 1340.

Références

Documents relatifs

The proof of theorem 4.1 actually shows that the heat kernel associated to the Lichnerowicz operator is positivity improving in case (M n , g 0 (t)) t≥0 is an expanding gradient

Aware of the grave consequences of substance abuse, the United Nations system, including the World Health Organization, has been deeply involved in many aspects of prevention,

In any 3-dimensional warped product space, the only possible embedded strictly convex surface with constant scalar curvature is the slice sphere provided that the embedded

First introduced by Faddeev and Kashaev [7, 9], the quantum dilogarithm G b (x) and its variants S b (x) and g b (x) play a crucial role in the study of positive representations

McCoy [17, Section 6] also showed that with sufficiently strong curvature ratio bounds on the initial hypersur- face, the convergence result can be established for α &gt; 1 for a

represents an irrational number.. Hence, large partial quotients yield good rational approximations by truncating the continued fraction expansion just before the given

This paper suggests the modification of the generalized recursion algorithm of the exact reachable sets com- putation for the linear discrete-time dynamic systems developed in the

Conclude to the existence and uniqueness of a global (in time) variational solution to the nonlinear McKean- Vlasov equation..