• Aucun résultat trouvé

Renormalization group for an Ising system with competing interactions

N/A
N/A
Protected

Academic year: 2021

Partager "Renormalization group for an Ising system with competing interactions"

Copied!
8
0
0

Texte intégral

(1)

HAL Id: jpa-00210296

https://hal.archives-ouvertes.fr/jpa-00210296

Submitted on 1 Jan 1986

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.

Renormalization group for an Ising system with competing interactions

P.M.C. de Oliveira

To cite this version:

P.M.C. de Oliveira. Renormalization group for an Ising system with competing interactions. Journal

de Physique, 1986, 47 (7), pp.1107-1113. �10.1051/jphys:019860047070110700�. �jpa-00210296�

(2)

1107

Renormalization group for an Ising system with competing interactions (*)

P. M. C. de Oliveira

Instituto de Física, Universidade Federal Fluminense, Outeiro de São João Batista s/n Niterói, RJ 24210, Brasil (Reçu le 29 juin 1985, révisé le 14 octobre 1985, accepté le 19 mars 1986)

Résumé.

2014

Nous appliquons une approche de groupe de renormalisation dans l’espace direct à un hamiltonien

d’lsing sur réseau carré avec des interactions de proche (J1) et second (J2) voisins. Nous obtenons le diagramme

de phase complet. Celui-ci exhibe trois phases ordonnées distinctes : ferromagnétique (F), antiferromagnétique (AF)

et super-antiferromagnétique (SAF). Dans cette dernière phase, les interactions concurrentes J1 > 0 et J2 0 jouent un rôle crucial entraînant une dégénérescence d’ordre deux des états fondamentaux. Ce phénomène se

traduit dans notre approche par un attracteur avec bifurcation d’ordre deux pour cette phase. De plus la forme des

trajectoires de groupe de renormalisation près de la ligne critique suggèrent que la transition SAF - P n’est pas universelle.

Abstract

2014

We apply a Position Space Renormalization Group approach to the Ising Hamiltonian in a square lattice with first (J1) and second (J2) neighbour exchange interactions. The full phase diagram is obtained with

three distinct ordered phases : ferro (F), antiferro (AF) and superantiferromagnetic (SAF). In particular, in the last one, the competing interactions J1 and J2 0, play a crucial role yielding a two-fold space degeneracy of the ground states. This behaviour is reproduced in the present approach by a bifurcated two-fold attractor for this

phase. Moreover, the non-universal character of the SAF - P transition is inferred from the Renormalization Group trajectories near the critical line.

LE JOURNAL DE PHYSIQUE

J. Physique 47 ( 1986) 1107-1113 JUILLET 1986,

Classification

Physics Abstracts

05.70J

-

05.50

-

75.10H

1. Introduction.

The magnetic properties of the square lattice with first and second neighbour coupling have been investigated extensively [1-5] by Position Space Renormalization

Group (PSRG) techniques [1-4] and series expansions [5]. The magnetic models adopted are Ising [2, 5],

Potts [1], bond percolation [3] and site percolation [4].

Various interesting features of these systems have been studied : i) the possibility of extracting the magnetic

critical exponents without introducing any external

field, from the crossover behaviour at the tricritical

point localized between the ferro (F), antiferro (AF)

and paramagnetic (P) phases ( 1-4] (see Fig. 9 =below) ; ii) the universal behaviour at the F - P and AF - P critical frontiers [1-5]; iii) the existence of a super-

(*) Supported by CNPq of Brazil.

antiferromagnetic phase (SAF) for sufficiently large negative values of the second neighbour Ising coupling

constant JZ [2, 5] ; iv) the impossibility of sustaining

the AF ordered phase for finite temperatures without sufficiently large positive values of J2, for Potts models with 3 or more spin states [1, 6] ; v) the non-universal behaviour at the SAF - P critical frontiers [2, 5].

Another interesting feature is the two-fold space

degeneracy of the SAF ground states for the Ising case

(see section 4 of Ref. [1]).

The purpose of the present paper is to study the

SAF strange behaviour, using PSRG, by scaling finite

lattices with the appropriate symmetries of the SAF ground states.

Section 2 presents the method for calculating the

PSRG transformation, including a review of recent

progresses obtained by preserving geometrical sym- metries. Section 3 presents the results, and the conclu- sions about the SAF phase are in section 4.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019860047070110700

(3)

1108

2. The PSRG transformations.

In a cell-to-cell PSRG calculation, one obtains quali-

tative features and quantitative approximations for an

infinite lattice, from physical quantities exactly calcu-

lated for finite cells of the lattice (unlike the alternative PSRG procedure of decimating sites of an infinite lattice, the cell-to-cell PSRG does not generate higher

order interactions). First, the values of a particular physical quantity are exactly obtained for two different

sized cells, the smaller one at temperature T’ and the other at T. A scaling transformation T’(T) is then

obtained by imposing that this physical quantity be

invariant under scaling. The length scale value is known a priori as being the ratio between the length

sizes of the cells. The Renormalization Group Theory [7] is then applied to the scaling transformation T’(7)

to obtain critical temperatures and other critical properties of the system. The reliability of the results

depends on various points : i) the arbitrary choice of a

convenient invariant physical quantity to be calculated for both cells; ii) the geometrical properties of the

cells that must reproduce those of the entire lattice, and

those of the ground state configurations of the ordered phases under study; iii) the size of the cells. In general,

the invariant physical quantity depends on parameters others than the temperature. A crucial point is to

include in the PSRG transformation all the parameters being relevant for the physical system. The number of those parameters defines the number of different invariant physical quantities to be calculated for both cells. In this case, the PSRG transformation is no more a simple function T’(T), but it is a map in a muti- dimensional space (the space of parameters).

In the first attempt to use PSRG [8], good approxi-

mations for the exactly known critical values for planar Ising models were obtained. The physical quantity

chosen to be invariant under scaling was the sign of the majority of the spins in the cell (Ising spins can be

either up or down corresponding to dynamical

variable values + 1 or - 1).

The critical temperature of the square lattice Ising ferromagnet was first obtained [9] exactly from the

self duality property of the Ising model on this lattice.

An improvement over the early PSRG calculation can

be obtained by choosing as invariant physical quantity

the transmissivity (defined below) which besides the

free-energy preserves the self duality property of the family of cells shown in figure 1. In this manner, the

exact critical temperature of the square lattice Ising ferromagnet has been reproduced [10] as follows. The

B

Hamiltonian of the Ising model with only first neigh- bour exchange interaction J1 is

where a; = ± 1 is the dynamical variable at site i,

and (i, j) is a pair of nearest neighbour sites. The transmissivity of a cell like the one shown in figure 1 b

Fig. 1.

-

Family of self-dual cells appropriate for the square lattice Ising ferromagnet. The length sizes of the cells are

b

=

I (a), b

=

2 (b), b

=

3 (c), and b

=

4 (d) in units of the lattice parameter. The full circles are sites inside the cell,

and the open circles are sites of the neighbouring cells. The

number of sites in each cell is N

=

b2.

for example, is defined by taking the following boun- dary conditions : i) the two sites at the bottom of the cell are taken as the same spin; ii) the two sites above the cell (open circles) are also taken as the same

(other) spin; iii) these two collapsed sites are the

terminals. Figure 2 shows the graphs corresponding

to the cells of figure 1 after applying these boundary

conditions. Each graph has a pair of terminal sites

(open squares) that can be in two distinct states : paired (both up or both down) or unpaired (one up, the other down). The transmissivity of the cell is defined

as the difference between the probability of the termi-

nal sites being paired and the probability of their being unpaired. This quantity can be calculated from the partition function of the system, obtained from the energy (1) of each possible spin configuration. The

number of configurations to be computed is 2 (b2 -b+l)

for a b-sized cell; this number increases exponentially

with the number N = b2 of sites in the cell. For the

simplest b

=

1 case (Figs. 1 a and 2a) there are no

internal sites in the graph, and one obtains the trans-

missivity tl of a single bond

For the b

=

2 case (Figs. Ib and 2b), the transmissivity tl is given by equation (3), as a function of tl.

Fig. 2.

-

Family of graphs corresponding to the cells of

figure 1 after applying the boundary conditions described in the text Open squares correspond to terminal sites,

while full circles are internal sites. The horizontal bonds

from the cell to its right neighbour cell are omitted, because

they have no influence on the vertical transmissivity.

(4)

Equation (3) is the required PSRG transformation

T’(7) for a length scaling factor b/b’

=

2. In the physical region 0 tl 1, it presents two distinct behaviours under repeated iterations, depending on the

initial value of ti : the successive iterated values of tl

will converge to the low temperature attractor 11

=

1,

or, alternatively, to the high temperature attractor tl

=

0, if the initial value of 11 is above or below the

critical value tt = j2 - 1, respectively. This critical value corresponds to the exactly known critical temperature [9], because the graphs of figure 2 pre-

serve the self-duality of the square lattice Ising ferromagnet, for any value of b. The flow diagram of equation (3), shown in figure 3, corresponds to the

exact phase diagram of the square lattice Ising ferromagnet. Moreover, good approximations [10]

can be obtained for the correlation length critical exponent v (Table I). Note that the approximation is improved for increasing cell size. The value of the critical exponent measures the rate of decrease of the correlation length near the critical point, and for this

reason its approximated value is obtained from the rate of iterative deviation from the critical point T*,

i.e. the derivative of the PSRG transformation, as

shown in equation (4).

For the study of the critical properties of the Ising antiferromagnet, where the first neighbour interaction

J1 becomes negative in equation (1), the cells of

figure 1 (or the graphs of Fig. 2) are no longer appro-

priate. The reason for this failure is easily understood by analysing figure 4. The boundary condition of taking

all the bottom line of the cell in the same spin state

is correct for the ferromagnetic ground state (Fig. 4a),

but it does not work for the antiferromagnetic case (Fig. 4b). Consequently, equation (3) does not present

Fig. 3.

-

Flow diagram of equation (3). Full squares are the attractors, and the open circle is the critical point separating

two distinct phases : ferromagnetic (F), where the points

are attracted to tl = 1 by repeated iterations of equa- tion (3) ; and paramagnetic (P), where the points are attracted to ti =0.

Table I.

-

Approximations [10] for the correlation

length critical exponent v obtained from PSRG trans- formation. A cell of size b

=

2, 3, 4 (Fig. 1) is trans- formed by scaling into a cell of size b’

=

1.

Fig. 4.

-

Ground state configurations of the three possible

ordered phases for the first and second neighbour square lattice Ising model. Ferromagnetic F (a), Antiferromagnetic

AF (b) and Super-antiferromagnetic SAF (c) (c’) phases.

Decoupling of the lattice into two sublattices in the J1

=

0

limit (d). The SAF phase has two degenerate ground states (c) and (c’).

the ordered phase attractor at t1

= -

1, neither the correct critical point tf

=

1

-

J2. The same failure

due to the same geometrical constraint is also observed in the presence of second neighbour interactions

[1, 3, 4], as well as in the case of anisotropic (different

horizontal and vertical interactions) systems [11].

The anisotropic square lattice [12, 13], as well as the

inclusion of second neighbour interactions [1, 3, 4],

besides the simple antiferromagnetic case can be

treated by using the family of cells shown in figure 5, and the corresponding graphs of figure 6. The boun-

Fig. 5.

-

Family of self-dual cells appropriate for the

square lattice Ising antiferromagnet with length sizes

b = (a), b

=

2 (b), b

=

3 (c) and b

=

4 (d). The number

of sites in each cell is N

=

2 b2.

Fig. 6.

-

Family of graphs corresponding to the cells of

figure 5 after applying the boundary conditions. For sim-

plicity, only the b

=

1 (a) and b

=

2 (b) cases are shown.

(5)

1110

dary conditions for the simple antiferromagnetic case (J1 0) are : i) the sites at the bottom line of the cell

are taken as the same spin and ii) the sites at the upper line inside the cell (full circles) are also taken as the same

(other) spin. Comparing figures 4 and 5 (the lattice has suffered a 450 rotation), we note that these boundary conditions are now appropriate also to the AF ground

state. The PSRG transformation for the scaling of the

b

=

2 cell into the b

=

1 single bond is presented in equation (5).

The flow diagram of equation (5) is shown in

figure 7. In the physical interval - 1 K ti 5 1 there

are now three distinct attractors, each one correspond- ing to one of the three possible phases P, F and AF.

The critical values tt

=

± (VF2 - 1) correspond to

the exact critical temperature of both the F - P or AF - P phase transitions. The Ising antiferromagnet

can be mapped into the corresponding ferromagnet, by interchanging the paired and the unpaired energy

values J and - J of each interacting pair of spins.

As the number of the paired possible configurations (both up or both down) and the number of the unpaired

ones (one up the other down) are the same, the map-

ping is possible. In particular, the square lattice Ising antiferromagnet has also the self-duality property, and this symmetry is preserved by the graphs of figure 6, also for any value of b. The approximations

obtained for the correlation length critical exponent, using equation (4), are shown in table II.

The same cells can be used for the case of including

second neighbour interactions (J2 > 0) in the Hamil- tonian (1). In this case, the space of parameters is two- dimensional, with two independent bond transmissi-

vities ti and t2 for first and second neighbour pairs of sites, respectively. Two distinct invariant quantities

are now necessary for obtaining the PSRG two

Fig. 7.

-

Flow diagram of equation (5). By successive iterations, an initial value t1 > J2 - 1 (F phase) will

converge to the attractor tl

=

1. Another initial value tl 1 - J2 (AF phase) will converge to other attractor

t1

= -

1. Intermediate initial values 1

-

J2 tl J2 -.1 I

(P phase) will converge to the central attractor ti 1

=

0.

Table II.

-

Approximations for the correlation length

critical exponent v, using the cells of figure 5. The value for b

=

2 and 3 coincides with those of reference [13].,

dimensional transformation. The first is the same

transmissivity defined above where both terminals

correspond to sites inside the cell; the second is obtained [1, 3, 4] by taking the upper terminal as the bottom line of the upper neighbouring cell (open

circles in Fig. 5). The b

=

2 PSRG transformation obtained [1] in the (tl, t2) space reduces to equation (5)

in the t2 - 0 limit, and also reduces to equation (3)

in the tl 0 limit. The same reductions occur for any value of b. In particular, the tl --+ 0 limit corresponds

to the decoupling of the whole lattice into two square sublattices [1-5] (see Fig. 4d), and this crossover

behaviour allows the evaluation of the magnetic cri-

tical exponent without introducing a magnetic field

in equation (1) [1-4J. As can be seen in figure 4, the AF ground state corresponds to two interpenetrating

sublattices (a and fl in Fig. 4d). Each sublattice separa-

tely has a ferromagnetic (F) configuration (Fig. 1 a).

The cells of figure 5 are obtained by superimposing two interpenetrating equal cells of figure 1, one for each sublattice (details in Ref. [1], section IV). In this

manner, the cells of figure 5 and the corresponding boundary conditions preserve the geometrical sym- metries of the AF ground state as the cells of figure 1

have preserved early the symmetries of the F ground

state.

The symmetry of the SAF ground states (Figs. 4c

and c’) is not preserved by the cells of figures 1 or 5 and

their corresponding boundary conditions, and it is

necessary to construct new cells to treat the J2 0 problem. Each one of the degenerate SAF ground

states corresponds to two interpenetrating sublattices

a and p, each one in an AF configuration. Using the

same previous reasoning, the cells of figure 8 are

constructed [1] by superimposing two equal cells of figure 5. The boundary conditions to be adopted are taking the sites A as one terminal and the sites B as the other terminal in each cell : this gives the first invariant

quantity. Taking A and C as terminals, we get the other invariant quantity needed for mapping (tl, t2) into (t’, t2).

Fig. 8.

-

Family of cells appropriate for the study of the

SAF phase. They are constructed by superimposing two equal cells of figure (5), one for each sublattice a and (see Fig. 4d). The length sizes are b === 1 (a) and b

=

2 (b).

The continuous (dashed) bonds correspond to first (second)

neighbour interactions. The number of sites in each cell is

N

=

4 b2.

(6)

The important role of preserving the symmetries of

the various ground states is conveniently emphasized

in references [14].

An important remark concerns the degeneracy of

the ground states for F, AF and SAF phases (Figs. 4a, b,

c and c’). There is a trivial spin degeneracy common

to those four ground-states, determined by the opera- tion of reversing all spins : this is a global symmetry transformation performed in the space of spins. This kind of degeneracy cannot be reflected in the flow of a

PSRG transformation performed in the real space.

Besides this trivial one, however, the SAF phase presents another non-trivial degeneracy (Figs. 4c

and c’) that cannot be associated to a global symmetry in the space of spins, and must be related to a geome- trical symmetry transformation performed in real

space (a 90° rotation around an axis perpendicular to

the lattice). This kind of space degeneracy must be

reflected in the flow of a conveniently constructed PSRG transformation. Because any RG attractor

represents the ground-state(s) of the corresponding phase, space degeneracies must be reflected in the behaviour of these attractors.

3. Results.

The flow diagram of the PSRG transformation, ob-

tained by scaling figure 8b into 8a, is shown in figure 9.

There are four distinct attractors (full squares) P, F, AF and (SAFI, SAF2), the latter formed by a pair

of points, one being the image of the other by the

PSRG transformation. Each attractor has its own

basin of attraction (the region of initial points that

converge to it by successive iterations of the PSRG

transformation), corresponding to each one of the

Fig. 9.

-

Flow diagram of the PSRG transformation

(tl, t2) --+ (t,, t2). The cell of figure 8b is scaled into the cell of figure 8a.

four possible phases of the physical system. The four basins of attraction (four phases) are separated by the

critical frontiers F - P, AF - P, F - AF and SAF -

P. Hereafter, we present separately the analysis of the

critical behaviour on each frontier.

3.1 F - P LINE. 2013 Points located exactly on this line

are attracted by Al. The critical behaviour of the F - P phase transition, however, is determined by

how an initial point near the line turns away from

it by successive iterations of the PSRG transformation.

The flow trajectory of such a point is directed towards

point Al, nearly parallel to the line. Only after approaching point Al, the trajectory deviates from the line, regardless of where is its beginning (near the line). This kind of reasoning is the RG explanation of universality : the critical behaviour is the same

(determined by point Al) along all the line. The approximated value v

=

0.718 is obtained at point A1, using the b/b’

=

2 scaling factor. Certainly, this value

would be improved if the PSRG transformation could be calculated for greater cells.

Point T is a critical point separating three distinct

phases F, AF and P, and is located at the exact value

t2 = /2 - 1. In the RG language, it presents two

distinct critical behaviours : i) a strong rate of devia- tion from it along the horizontal direction; and ii) a

weak rate of deviation along the t2 axis where the a

and fl sublattices (Fig. 4d) are decoupled In this latter

case (ti

=

0) each cell of figure 8 reduces to two equal

cells of figure 5, one for each sublattice a and fl. In this

manner, the critical exponents obtained are those of table II. Moreover, it can be shown [1-4] that the other

critical behaviour at point T (along the horizontal) corresponds to another independent critical exponent : the inverse fractal dimension 1 /df of the incipient

infinite island of correlated sites along the lattice.

The magnetic susceptibility critical exponent, for example, can be then obtained from the scaling rela-

tion y

=

df v. The approximated values for df are

shown in table III.

3.2 AF - P LINE.

-

This line is the mirror image of

the preceding F - P line, and has the same proper- ties. This symmetry is also present in the real physical

system.

3. 3 F - AF LINE. 2013 This line also displays universal behaviour, because point B attracts all points exactly

located on the line and determines the critical beha- viour for points near it. The corresponding critical

exponent lld can be calculated along the t2

=

1 axis

Table III.

-

Approximations for the fractal dimension

df of the incipient infinite islands of correlated sites,

formed near the tricritical point T (Fig. 9), using the

family of cells of figure 8.

(7)

1112

for any value of b, as given by

This exponent measures the coupling of two infinite

islands of ferromagnetically correlated spins, one in

the a sublattice and the other in {3. The positive second neighbour exchange interaction J2 is strong enough

to allow the formation of both islands, and a very small first-neighbour exchange interaction J, couples

the islands together. The coupling can be ferro or antiferromagnetic for positive or negative values of J1, respectively. The exact value d

=

2 can be obtained from the thermodynamic limit (b ---> oo) taken in equation (6). This value shows that the infinite islands

are two dimensional objects (like the lattice itself) along all the F - AF line, except at point T where they become infinite with a fractal dimension df = 1.75.

3.4 SAF - P LINE.

-

It is well known [2, 5] that the

critical behaviour is non-universal along SAF - P line.

The flow direction along this line in figure 9, however indicates that the critical behaviour would be the same

along all the line, determined by point C (the same

values presented in Table II). In order to reproduce the

non universal behaviour, point C could not attract the

points exactly located on the line, as in figure 9. To

further investigate this point, we have determined the flow direction near point C along the SAF - P line, for greater cells. Namely, the horizontal derivatives of the PSRG transformation have been calculated for

increasing cell sizes (Table IV). For b

=

4, this value becomes negative (which merely means that the

flow becomes oscillatory around C), and greater than unity in absolute value, which means that point C

becomes fully repulsive.

4. Discussion of the SAF phase.

By preserving the fundamental geometrical features

of both the lattice and the four possible ground-states

of the system, the present PRSG gives a SAF phase

characterized by a bifurcated attractor (SAFI and

SAF2 in Fig. 9). The line denoted by numbers 1, 2, 3,

etc. in figure 9 is a typical PSRG trajectory for an

initial SAF point. The oscillatory repulsive behaviour

is already present in the neighbourhood of point D, for this small b

=

2 transformation. The two-fold sym- metry of the SAF attractor reproduces the two-fold space degeneracy of the SAF ground states. The oscil- latory repulsive behaviour is observed also in the neigh-

bourhood of point C for b

=

4 transformation

(see Table IV). Unfortunately, we are not able to compute the full b

=

4 phase diagram (there are

Table IV.

-

Derivative A of the PSRG transformation along the horizontal, at point C (Fig. 9).

257 ~ 1017 configurations to compute), but we can

infer its qualitative features from two facts : i) the

attractor is still the pair (SAFI, SAF2) located at the

same place (the ends of the critical line); ii) point C

becomes repulsive along the critical line. We then expect that an initial point exactly located on the

critical line will be directly attracted to (SAF1, SAF2) :

this is possible because the attractor is unusually

localized on the critical line (no other RG treatment presents such a behaviour, to the best of our knowledge). The critical behaviour along the line

would not be dominated by any special point on it :

initial points near the line will be repelled from it directly, without the usual approaching to a special point on the line, and the rate of deviation will become

a local property of the initial localization near the line.

In this manner, the non universal character of the SAF - P phase transition is explained by the present

PSRG transformation.

Two final comments concern the successful repro- duction of the two-fold space degeneracy of the SAF ground states. First, the correspondence between the attractor (SAFI, SAF2) and the two ground states

cannot be splitted, SAFI corresponding to figure 4c

and SAF2 to figure 4c’ for instance. The correspon- dence is necessarily between both pairs each one

considered as a whole entity. Second, the existence of

competing interactions when J2 0 is by itself responsible for the two-fold space degeneracy of the

SAF ground-states. If disorder is included in this system, one expects the degeneracy to be enhanced for

increasing values of some disorder control parameter.

In the RG language, one expects a cascade of Feigen-

baum bifurcations of the SAF attractor. For suffi-

ciently high disorder, the degeneracy becomes infinite,

the attractor becomes continuous, and the system becomes a spin glass [15]. We think the present approach can be applied to the study of systems with these features.

Acknowledgments.

The author acknowledges Dr. F. A. de Oliveira for a

critical reading of the manuscript, and Drs. S. L. A. de

Queiroz and R. R. dos Santos for comments before its

final form.

(8)

References

[1] OLIVEIRA, P. M., TSALLIS, C. and SCHWACHHEIM, G., Phys. Rev. B 29 (1984) 2755.

[2] NIENHUIS, B. and NAUENBERG, M., Phys. Rev. Lett. 33

(1974) 944 ;

VAN LEEUWEN, J. M. J., ibid 34 (1975) 1056 ;

SWENDSEN, R. H. and KRINSKY, S., ibid 43, 177 (1979).

[3] RIERA, R., OLIVEIRA, P. M., CHAVES, C. M. and DE

QUEIROZ, S. L. A., Phys. Rev. B 22 (1980) 3481.

[4] OLIVEIRA, P. M., Phys. Rev. Lett. 47 (1981) 1423.

[5] OITMAA, J., J. Phys. A 14 (1981) 1159.

[6] BAXTER, R. J., Proc. R. Soc. London A 383 ( 1982) 43.

[7] WILSON, K., Phys. Rev. B 4 (1971) 1374.

[8] NIEMEIJER, Th. and VAN LEEUWEN, J. M. J., Phys. Rev.

Lett. 31 (1973) 1412, and Physica (Utrecht) 71 (1974) 17..

[9] KRAMERS, H. A. and WANNIER, G. H., Phys. Rev. 60 (1941) 252.

[10] TSALLIS, C. and LEVY, S. V. F., Phys. Rev. Lett. 47 (1981) 950;

MARTIN, H. O. and TSALLIS, C., J. Phys. C 14 (1981) 5645 ; Z. Phys. B 44 (1981) 325.

[11] CHAVES, C. M., OLIVEIRA, P. M., RIERA, R. and DE QUEIROZ, S. L. A., Prog. Theor. Phys. 62 (1979) 1550 ;

NAKANISHI, H., REYNOLDS, P. J. and REDNER, S., J.

Phys. A 14 (1981) 855.

[12] OLIVEIRA, P. M., Phys. Rev. B 25 (1982) 2034.

[13] OLIVEIRA, P. M. and TSALLIS, C., J. Phys. A 15 (1982)

2865.

[14] JUNGLING, K., J. Phys. C 9 (1976) L1;

KRINSKY, S. and MUKAMEL, D., Phys. Rev. B 16 (1977) 2313.

[15] MCKAY, S. R., BERKER, A. N. and KIRKPATRICK, S.,

Phys. Rev. Lett. 48 (1982) 767.

Références

Documents relatifs

The case of stochastic evolutions (in particular subjecting the Dyson measures to an infinite-temperature evolution) should be fairly immediate, for short times the results of

But we believe that in these shorter series the influence of other singularities was still very strong, so that the evaluation of the amplitude of the confluent

Analysis of ising model critical exponents from high temperature series expansion..

Pour le modele d'Ising avec spin S = & on a trouv6 que la catkgorie de complexions qui sont des arbres (graphes finis connexes sans cycles) fait une contribution

The phase diagram presents three phases (namely the paramagnetic, the bulk ferromagnetic and the surface ferromagnetic ones) which join on a multicritical point

modulated structures observed in other systems (V-Ni-Si, Mo-Cr-Fe). According to the electron diffraction patterns, these modulated phases are precursor states of the cubic

In [15], the authors have defined an annealed Ising model on random graphs and proved limit theorems for the magnetization of this model on some random graphs including random

We wish to gain insight into the interface in the Ising model with the help of our previous results and the classical coupling of Edwards and Sokal (see chapter 1 of [9] for