• Aucun résultat trouvé

Dimensional renormalizations of polymer theory

N/A
N/A
Protected

Academic year: 2021

Partager "Dimensional renormalizations of polymer theory"

Copied!
12
0
0

Texte intégral

(1)

HAL Id: jpa-00210237

https://hal.archives-ouvertes.fr/jpa-00210237

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.

Dimensional renormalizations of polymer theory

Bertrand Duplantier

To cite this version:

Bertrand Duplantier. Dimensional renormalizations of polymer theory. Journal de Physique, 1986, 47

(4), pp.569-579. �10.1051/jphys:01986004704056900�. �jpa-00210237�

(2)

Dimensional renormalizations of polymer theory

B. Duplantier

Service de Physique Théorique, CEN-Saclay, 91191 Gif-sur-Yvette Cedex, France (Reçu le 25 octobre 1985, accepte le 20 dicembre 1985)

Résumé. - Nous montrons comment on peut effectuer systématiquement, à tous les ordres, deux renormalisations dimensionnelles différentes (ou soustractions minimales), directement en théorie des polymères dans le cas critique (pour d

=

4 - 03B5, 03B5 ~ 0). L’un des schémas de renormalisation dimensionnelle coincide avec celui de la théorie des

champs et a été récemment justifié par M. Benhamou et G. Mahoux. L’autre, le schéma z de renormalisation dimen-

sionnelle, est nouveau et le paramètre z de Yamakawa de la théorie des polymères y est renormalisé de manière minimale. Les relations existant entre ces renormalisations et la « renormalisation directe » de J. des Cloizeaux

sont explicitées. Les indices critiques sont recalculés au second ordre en 03B5.

Abstract.

-

We show how to perform systematically to all orders, two different dimensional renormalizations (or minimal subtractions) directly in polymer theory, in the critical case (for d

=

4 - 03B5, 03B5 ~ 0). One of the dimensional renormalization schemes is that of field theory and has been justified recently by M. Benhamou and G. Mahoux.

The other one, the z-dimensional renormalization scheme is new and renormalizes minimally the Yamakawa parameter z of polymer theory. We explicit the relation of these renormalizations to the « direct renormalization » of J. des Cloizeaux. We recalculate the critical indices to order 03B52.

Classification Physics Abstracts

05.20

-

05.40 - 11.10 - 61.40K - 64.70

1. Introduction.

The most useful model for polymer theory is the

continuous chain model, introduced by S. F.

Edwards [1]. There exist various renormalization methods for treating it, i.e. field theoretic ones using

the n - 0 analogy due to de Gennes or more direct

ones using only polymer diagrams [2]. (See also [3]

and references therein.) The direct renormalization method of J. des Cloizeaux [2] is in fact a physical

one, since one considers there physical quantities

which fix the scale of the Kuhnian chain. The study by L. Schafer and T. A. Witten of the renormalization of polymer theory was closer to field theory [4]. These

authors used the so-called massless scheme of field

theory and transferred it to polymers, obtaining,

within their scheme, a multiplicative renormalization of the polymer theory, including polydispersity.

However, this scheme is not easily applied directly

in polymer theory since it relies mostly on field theory (through the use of vertex functions and not of poly-

mer partition functions). On the other hand, there is in field theory a dimensional renormalization method, popularized by t’Hooft and Veltman [5], which has

certain minimal subtraction properties [5]. Our aim

is to present here how to do rigorously various dimen-

sional renormalizations directly in polymer theory.

The t’Hooft-Veltman renormalization already appear- ed in works about polymer systems [3], but without general proofs. The idea of using the minimal renor-

malization prescription of t’Hooft and Veltman, for proving the finiteness of the des Cloizeaux scheme, is

due to M. Benhamou, who remarked that the Laplace-

de Gennes transform of polymer partition functions

« commutes » with dimensional renormalization fac- tors (1,2). This shows [6] that in polymer theory finite perturbation expansions can be obtained by multi- plicative renormalization of the partition functions (the direct renormalization method [2]).

We consider here the continuous model for polymer

chains in a good solvent [1], i.e. the two-parameter model, in a space dimension d 4. We present two different dimensional renormalization schemes (I and II) for this polymer theory, showing their existence

from dimensional renormalization in field theory.

(’) J. des Cloizeaux, private communication.

(2) In their different (massless) scheme Schafer and Wit-

ten [4] used also the commutation of the corresponding

renormalization factors with generalizations of the Laplace

transform.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01986004704056900

(3)

One of these schemes (II) is exactly the field theoretic

one and recovers the results used in reference [3] and justified in reference [6]. The other one is a dimen-

sional renormalization scheme proper to polymer,

since it renormalizes minimally the well-known dimen- sionless z-parameter of polymer theory. We perform explicit calculations with these dimensional renorma-

lizations directly in polymer theory. For completeness,

the des Cloizeaux’ method is considered and proven to give finite results to all orders. We calculate the finite renormalizations relating his method to the dimensional ones.

-

In the next article, we consider also the case of polymer chains near the tricritical theta point, and study the direct and dimensional renormalizations of

the associated three-parameter model [7]. Let us finally mention that we have also established in ano-

ther work [8] an explicit and convergent integral representation for the dimensionally renormalized

polymer partition functions. It permits to calculate directly (i.e. without renormalization factors) the

finite renormalized perturbation expansion of the polymer partition functions, in a particularly simple

way.

2. Polymer theory.

We consider here the continuous model for polymer chains, with the probability density for a configuration rj(sj) of N chains j

=

1, ..., N

The Sj are the Brownian areas of the N chains. They

can be written in terms of a unique Brownian area S :

In the following we shall consider the general poly- disperse case, but the examples will be treated in the

simpler monodisperse case

The interaction parameter b is related to the dimen- sionless z parameter by [2]

with d

=

4 - s. We work in dimensional regulariza-

tion : all the quantities we consider are analytical

continuations to d

=

4 - 8 of the same quantities

which are defined for d 2 [9]. In particular, one

defines the set of connected regularized partition

functions of N polymer chains, in Fourier space [2] by :

where Po represents the probability density for

Brownian chains (b

=

0). The connected total parti-

tion function of the N chains is then given by

By dimensional analysis, one has immediately [2] for Fl

identical chains :

where 3N (z) is a function of z alone, also depending on

the dimension d, and which is singular when d -+ 4.

Now we can define two different dimensional renormalization schemes, for polymer theory :

In the first one, we set (considering the monodis-

perse case for simplicity)

where Xz, Xb, Xi have the singular series expansion :

A is a pure number, entirely arbitrary (3). The fXn,j are

entirely independent of /L Substituting (I) into the

(3) It pl4ys the role of the well known t’Hooft’s mass p in

minimal prescription.

(4)

partition functions make their double expansions in

terms of ZR and s, and expressed as functions of

S’, £R, finite when s -+ 0. The particular form of equation (6) is the defmition of the dimensional renormalization or minimal subtraction scheme [5].

The renormalization constants have no fmite parts in 8 but only poles.

The relation between the regularized partition

functions (3) and the renormalized ones is then (for

the monodisperse case)

where iT[ has a finite limits for s -+ 0, when expressed

in terms of zR (and not z).

The second dimensional renormalization scheme we use is (for the polydisperse case)

where the renormalization constants Zb, Zm, Z have

also the minimal form (6). As we shall see, these

renormalization constants Zb, Zm, Z are exactly those of field theory in the limit of a number of components

n

=

0. In this scheme, the relation between regularized

and renormalized partition function will be

where the fl’ R are other renormalized partition func- tions, depending on { Sj,R } bR, and it, and regular in s, when s - 0.

The two schemes (I) and (II) are not the same. The

difference is in the renormalizations of b or z

=

(2 n)-d/2 bS 2 - d/2. That of b is the standard one in field theory. That of z is more convenient for polymer theory, since the variable z is a natural expansion

variable. This is the reason why we introduce it.

Incidentally, let us note that the convergent integral representation established by us for the dimensionally

renormalized polymer partition functions [8] corres- ponds to the scheme II of this article. This integral representation has nevertheless a quite simple expres- sion in terms of the z-parameter [8].

To prove the existence of these two schemes, we shall use field theory. We shall prove the existence of the scheme II first, and then of the other one I, which is

a little more subtle.

3. Field theory and proof of existence.

As it is well known [10], there is a correspondence

between the continuous polymer theory just described,

and the O(nN) field theory (n -+ 0) given by the

Hamiltonian

There are as many fields (p,, and masses mj, as there are polymer chains. If one defines in Fourier space

and the connected Green functions by

one has the exact correspondence

(The contour is taken on the right hand side of the singularities of the integrand.)

Now, we know that there exists [5] a (unique) dimen-

sional renormalization or minimal subtraction scheme

(for n - 0) :

where Zb, Z., Z have the same minimal polar struc-

ture as in (6). Substituting (11) makes the renormalized Green’s functions finite, when expressed in terms of bR, Mj2 ,R. The parameter p is the arbitrary mass scale [5],

which makes bR dimensionless. Two important facts

must be noted. First, the renormalization of the mass

is mass-independent as emphasized by previous

authors [5]. Second, in principle, for n # 0, there is

a mixing of the renormalizations of the masses [5] :

(5)

However, all the non-diagonal terms of the Zmi, j

matrix involve internal loops and vanish in the n - 0

limit, yielding a great simplification. Moreover all the

diagonal terms are equal (4). Incidentally, we note that

the same essential remark appears in reference [4] in

the different context of the massless scheme, and that

an argument similar to that of this section is used there.

The multiplicative renormalization of the Green’s functions reads then :

Now, by this change of variables (11), we know that gR has a finite limit when E - 0, once it is expressed in

terms of Mj2 ,R and bR only. Inserting (11), (12) into equation (10) and performing a change of variables

in the masses, it is easy to rewrite ff as

and where

The mj,R have become in (14) simple integration variables, The contour q’ still lies on the right-hand

side of the singularities of the integrand. gR being

finite for 8 -+ 0, 1l’ R is a regular function of s.

There are other quantities of interest in polymer theory, like the form factor of a single polymer chain [13]

This form factor is related to the Green’s functions of the field theory with insertions of cp2. One has exactly :

where the partition function with two insertions of q, - q,3 (2) [13] reads [11] :

g (2) is the Green function with two qJ2 insertions, defined by [ 11 ]

where m2(r) is allowed to depend on the position in

space.

Then, in the same way as before, we use the dimen- sional renormalization (11), (12) and (14) for obtaining immediately

where iTj2> is defined by

Now F(2) tends towards a finite limit when s --> 0.

Finally, owing to equations (13, 16), we find

(4) In terms of the usual renormalization factors Z, Z(2), Z(4) of the field p, of p2, and of the 4-point vertex function,

we have in the notation of Brezin et al. [12] Z,

=

Z (4)lz 2,

zm

=

Z(2/Z.

Thus h(q) is finite when expressed in terms of SR, bR as already used in [13].

We have thus recovered the renormalizability

to all orders of the theory in the scheme II, as also shown in reference [6]. We may also note at this stage that Schafer and Witten [4] established similar results in their different massless scheme. Since the latter and the dimensional one are related in field theory by

a finite (however complicated) renormalization, the

scheme II in polymer theory is, of course, also related

to that of reference [4] by a finite renormalization. Let

(6)

us now turn to the first renormalization scheme I.

From II, we have

Then we face immediately a difficulty. Zb, Zm are

minimal renormalization factors having the struc-

ture (6), with no constant terms. But Zb Z m e/2 is not

minimal, due to the occurrence of the 0(s) power. Thus from this it is not obvious that the minimal renorma-

lization constant Xz does exist. We are then led to the

following question : does it exist an 8-regular change

of variable b (zR,,6) which yields (2 7r)-d/2 bR Zb x Zm E/2(bR, 8)

=

ZR X.(ZRI 8), where Xz is minimal ?

In this case we set A

=

Jl2 SR.

A second requirement is that the other renormaliza- tion factors defined by

stay minimal. We show in the Appendix A that this is

actually possible and even that the change of variable bR(zR) is completely independent of s.

It follows then that the minimal renormalization I exists for the z variable, as II exists for the b variable.

Q.E.D.

In the following, we consider the des Cloizeaux’s scheme [2]. We also perform explicit calculations of these dimensional renormalizations, up to two-loop

order.

4. Link with the direct renormalization method.

We shall now consider physical quantities important

in polymer theory [2]. There are : the partition function

of one isolated chain

the Kuhnian size of the isolated polymer chain : R’ = I r(S) - r(O) 112 >= dX 2, where

the dimensionless second virial coefficient g [2] :

and finally the square radius of gyration

The previous analysis shows that all these quantities

are finite when expressed in terms of zR, SR, or bR, SR, using the dimensional renormalizations I or II. Let us now consider the « direct renormalization method » introduced by J. des Cloizeaux [2]. It states that the

new renormalized partition functions 3R defined in

that scheme by

are all finite when empressed in terms of the physical quantities X 2, g instead of S, z, when d -+ 4- or c - 0+. This can be proven to be true to all orders in e

and g, as a direct consequence of the multiplicative renormalizability described above. Using (7) in the

z-dimensional scheme, we find immediately :

Using (13) gives alternatively :

Thus 9 R " is a finite function when expressed in terms of S’, zR, h or in terms of SR, bR, M. Incidentally, let us

note that this is also true for the form factor h(q) (16).

Therefore for proving the « direct renormalization method » to work to all orders, it suffices to note, as stated above, that the changes of variables

or

are regular with respect to s, i.e. they contain no singularities when s ---> 0. This substitution is studied in

Appendix B. Of course, A or It cannot appear in the

physical quantities, which do not depend on them and

this requires that the variables { S’, zR, À. } or { SR, bR, p ) recombine for giving finally two variables { X 2, g }

in the physical quantities.

Let us note finally that, since the form factor h(q)

can be written as a regular function of bR, SR as in (17),

it has also a finite expression in terms of g, qZ X 2, staying well defined when s ---> 0. As a consequence,

h d.. I

.

6 R 2

..

I f the dimensional ratio N -

.2 G is a universal func- [R"

tion of g, finite order by order in g and s. These facts, intuitively obvious, were already checked in previous

works [13,14]. In Appendix C, we show in addition the

finiteness of the renormalization functions

(7)

and of the Wilson function ,

I

of the direct renormalization method.

5. Explicit calculations.

Let us now apply the general minimal schemes I and II introduced before, by performing calculations directly

in polymer theory, at the two-loop order. We shall in particular determine the renormalization factors Xz, Xo, xi and the renormalized zR, or equivalently Zbl Zm, Z and bR. For this we consider the four quantities (18-21) which have been previously computed directly

in polymer theory [2, 14]. One has [2]

and [14]

here

These expansions are very interesting since we

shall be able to calculate for instance zR, 3Co, 3Ci from

the three first ones, using the minimal renormalization I, and check that, for instance, they make N finite when expressed with zR. Using the form (I) and equa- tion (6) we find the simple results

from g :

from X 2 :

These relations go with the useful finite expressions (5) :

Inserting z(zR, 8) into N yields, as it must, the finite series expansion

If we use instead the dimensional renormalization scheme of the field theory II, and equation (6), and

insist on making finite the same quantities (22), (23), (24), we find naturally a new answer :

where

and for the area S :

and for the partition function :

These expressions go with fmite series expansions (6) :

which slightly differ from (29).

So, as expected, the two renormalization schemes

equations (26)-(28) and equations (31)-(33) differ at

the two-loop order. Incidentally we note that the expressions 31)-(32) appear in reference [15] which

uses the minimal subtraction scheme II, but without

general proof.

(5) For the last two equations of (29), we have not retained the A dependent terms, setting A

=

1.

(6) For the last two equations of (34), we have not retained

the p dependent terms, which amounts to set 2 1tJ.l2 SR

=

1.

(8)

Let us now calculate the finite renormalization functions which appear in the two schemes. The Wilson function is defined by

for the scheme (I), and by

in the field theoretic one (II).

Using (26) we find

with a fixed point value

The second scheme gives, using (31)

with a fixed point value

We note that both fixed point values (36), (38) yield,

when inserted respectively in (29) and (34), the same

fixed point value of the second virial coefficient of Kuhnian polymer chains :

in agreement with reference [2].

We also check that the Wilson function (37) of the

scheme II has exactly the form, as it must, given by the

Russian school [16] in the minimal renormalization scheme, for n

=

0.

Both Wilson functions W’(zR) or W(bR) have also

the fundamental property of being of the form

where W4 is independent of s. This well-known form characterizes the dimensional renormalization.

Let us check for instance the validity of our z-dimen-

sional renormalization scheme I by computing the

critical indices. We define the functions

From equations (26)-(28) one gets the finite series expansions

which, at the fixed point zQ (36), have the values

in agreement with the well-known s-expansions for

n

=

0 [2,12]. In the same way, we fmd for the universal ratio N (30)

in agreement with reference [14].

We are also in position to compare our method to

the « direct renormalization method » of reference [2]

(see Section 4). Consider the physical renormalization

constants :

and the second virial coefficient g (Eq. (20)). They

differ from those of the dimensional renormalizations

by a fmite renormalization. For instance, one has

immediately, from equations (27)-(29)

while g is given by (29).

The « direct renormalization method » is interest-

ing since it works directly with physical quantities.

The draw-back is related to the fact that it is not a

minimal renormalization. All the fmite renormaliza- tion functions for instance are more complicated. For instance, the associated Wilson function reads [2] :

with a fixed point value

as in (39). -

In conclusion, we have at our disposal three different

renormalization schemes for polymer theory. One

is the « direct renormalization » method of des

Cloizeaux [2] which uses physical quantities but is not

(9)

minimal. One may also use the z-dimensional renor-

malization proposed here which is minimal, and particularly convenient for polymer theory since it

uses the well-known z parameter, cherished by polymer physicists [17], but to renormalize it into zR.

It introduces « non physical » quantities S’, zR which

can be related to the physical quantities X 2, g,

However, from the point of view of the theorist, I

think it is much more fundamental with respect to the

structure of the theory of the Kuhnian chain. All the renormalization functions are then indeed very simple.

The last one, the field theoretic one, uses the minimal renormalization of the n

=

0 field theory. It has the

same advantage as the z-scheme for the theorist, having included in itself the heart of the renormaliza- tion structure. But it is also « not physical ». In poly-

mer theory, it will lead to slightly more complicated

calculations. It has an advantage : one may use the field theory calculations, as we have shown in another

work [8].

Finally, it would be very interesting to try to find

«physical quantities » corresponding in polymer theory to the minimal zR, Xb, Xi. It would be also very

interesting, since now one can work directly in poly-

mer theory with dimensional renormalization, to try to find the hidden renormalized Edward’s type model it corresponds to.

Acknowledgments.

I thank F. David for very interesting discussions.

Appendix A.

PROOF OF EXISTENCE OF THE MINIMAL Z-SCHEME.

-

The renormalization equations

are equivalent to the integral representations

where by definition :

In (A. 2) the lower value of the integration variable

is not precised (it can’t be zero), but this is natural since it corresponds to the arbitrariness of it.

Now, the minimal polar form of the renormalization

constants Zb, Zm (A. 2), (A. 3)

is equivalent to the fact that W has the well-known

particular form

where W4(t) is the value for d

=

4, which is comple- tely independent of B. ym(t) itself is also independent

of s. This can be checked easily on equations (A. 2), (A. 3). Now we have by definition

which can be rewritten, owing to (A .1)

From now on we drop the factor (2n)-2 which is

reabsorbed into the defmition of b : (2 n)-2 b - b,

and we also forget about the (2 nJ.l2 SR)E/2 = À. e term,

of which the renormalization factors are necessarily independent. As we already said, the factor Zb Zm2/2

does not have the minimal structure (A. 5), with respect to bR, or, equivalently, Fb Zm E/2 does not have

the form (A. 2), (A. 6).

So we search for a change of variable

regular in E, which makes Fb Zme/2 minimal with

respect to the new variable zR.

We have, owing to (A. 2), (A. 3)

or changing of variable t(t’) ; bR(zR) :

With respect to the variable t’ or zR, the scheme will be minimal if we can rewrite F. (A. 8) in the form

where W’(t’) has the particular form (A. 6)

(10)

So we identify (A. 8), (A. 9), with the help of (A. 10)

It is sufficient to set

or by integrating

Since y.(t)

=

0(t) for t small, the change of variable

.

(A. 11) has a regular series expansion in powers of t,

t’ = t + O( t 2). Thus we see that the change of variable

which is entirely independent of ~, gives a minimal

structure to Fz, and we can set

where xz has the property (A. 5). It is easy to check that the renormalization factors

which were minimal with respect to bR, are also left

minimal with respect to zR, in the change (A 12). So (A 12) gives a minimal subtraction scheme with respect to z. Q.E.D. We note that the present analysis applies identically for proving, in field theory, the

existence of a minimal subtraction scheme in the variable bm-E, instead of in b, a fact which apparently

has not been noticed before.

Appendix B.

REGULARITY OF THE SUBSTITUTION { ZR, SR, À. } - { g, X2}. - We shall prove this regularity in the

z-dimensional scheme. The same can be proven in the b-dimensional scheme in the same way.

Going back to equation (19) and using the funda-

mental equation (7), we have

By dimensional analysis, this yields

where Fo is a regular function of s.

On the other hand, the dimensionless second virial

coefficient g defined in (20) reads, owing to (7) and to

the preceding equation :

By dimensional analysis, this can certainly be written

as

where F2 is, again, a regular function of e. This achieves the proof of the regularity of the substitution

{ ZR, Si, Â. } --+ { g, X 2 }.

°

In the same way, one shows that for the substitution

{bR, SR, A}_{g,X2}, there exist two functions Fo, F2 such that

which are regular with respect to 8.

Appendix C.

RENORMALIZATION FUNCTIONS AND CRITICAL INDICES.

-

We shall here consider the renormalization equa- tion (13)

and draw consequences from it, which have been assumed and used in reference [2]. We define the

renormalization functions

For { k }

=

{ 0 }, the F are the partition functions

and thus the aN are effective indices. The mass

is trivially kept fixed since ff does not depend on ,u.

But this proves useful in the following. We use the

fundamental equation (13) to get

But bR(b, M, s) does not depend on S. So trivially

(11)

and thus, since SR

=

Z. S, one has also

So we immediately see that UN is finite in s, when

expressed in terms of renormalized quantities SR, bR.

Moreover, since we know that the substitution

{ SR, bR, ,u } - ( X 2, g } is regular, we see that the 6N

are finite functions (in s) of X and g. Q.E.D. They are

also fmite functions of (SR, zR) or (SR, bR), or (X, zR)

or (X, bR).

Let us also consider the Wilson function introduced in reference [2]

and prove it is a finite function of g and E. We write it trivially and use again the identity

Then, according to the analysis of Appendix B we have g

=

F2(bR, ,u2 SR, s) and thus

Thus W" is a finite function (in s) of bR, ,u2 SR’ It is

therefore also finite when expressed in terms of g,

J.l2 X 2, but, since it does not depend on p, it is a finite function of g alone, Q.E.D.

We can also relate easily these effective indices (IN

to the usual critical indices of field theory and critical phenomena. Let us consider again (C .1) and apply

on both sides of it. Since trivially :

one gets, for the special case { k } = { 0}, and using

dimensional analysis, the Callan-Symanzyk like equa- tion for the connected partition function TN :

where the renormalization functions W, Ym, ’1 are defined in the usual way [12] by

In terms of the usual critical indices [12] one has the-

refore [12]

Using (C. 5), we obtain from (C. 7)

Now, at the infrared fixed point bR of the field theory

one has

and, in this limit :

o o

Using (C .9) we find

Thus we find, since by definition

1-

that the connected partition function of N chains

behaves for large S like

an interesting equation which had not been explicitely

shown nor written before. For N

=

1, we recover the known behaviour 1[’1 1’-1 S Y -1 but for N > 1, we see that the other independent index v appears in the

asymptotic behaviour of9% .

One can show in a similar way that the effective index corresponding to the swelling factor xo

=

X 2/S ;

i.e.

is finite to all orders in terms of bR, SR or g, X 2, and,

at the fixed point bR, takes the known value

Thus

Let us note that the two scaling behaviours

YN _ Svd(N-I)+N(y-1), and >o - S2v-1, were neces-

sary for the direct renormalization method of refe-

rence [2] to be valid. Indeed, if one defmes the connected

dimensionless virial coeficients 9N by

(12)

one sees, according to (C .13) (C .14), that these g N

have a finite limit when S -+ oo. This was the basic

assumption made in reference [2] concerning g2 --- g, since the Wilson function W"(g) = (g) S a TS 70 b b

9

was

supposed to vanish for S -+ oo, i.e. at the infrared

fixed point Q.E.D. So one can, in a more general way, define the set of Wilson functions WN

=

S a .

They all will tend to zero for large S, and could be used equivalently to determine the infrared fixed

point for S -+ oo.

References

[1] EDWARDS, S. F., Proc. Phys. Soc. 85 (1969) 613.

[2] DES CLOIZEAUX, J., J. Physique 42 (1981) 635.

[3] OONO, Y., Statistical Physics of Polymer Solutions,

Adv. Chem. Phys. 61 (1985) and references therein.

See in particular :

OONO, Y., OHTA, T., FREED, K. F., J. Chem. Phys. 74 (1981) 6458.

KHOLODENKO, A. L., FREED, K. F., J. Chem. Phys. 78 (1983) 7390.

[4] SCHAFER, L., WITTEN, T. A., J. Physique 41 (1980) 459.

[5] ’T HOOFT, G. and VELTMAN, M., Nucl. Phys. B 44 (1972) 189.

’T HOOFT, G., Nucl. Phys. B 61 (1973) 455, see also:

COLLINS, J. C., MACFARLANE, A. J., Phys. Rev. D 10 (1974) 1201.

COLLINS, J. C., Phys. Rev. D 10 (1974) 1213.

COLLINS, J. C., Nucl. Phys. B 80 (1974) 341.

BREITENLOHNER, P., MAISON, D., Commun. Math.

Phys. 52 (1977) 11.

SPEER, E. R., Commun. Math. Phys. 37 (1974) 83.

[6] BENHAMOU, M., MAHOUX, G. (1985) Saclay preprint SPhT/85/157.

[7] DE GENNES, P.-G., J. Physique Lett. 36 (1975) L-55.

DUPLANTIER, B., J. Physique 42 (1982) 991.

[8] DUPLANTIER, B., Europhys. Lett. 1 (3) (1986) 99.

[9] BERGERE, M., DAVID, F., J. Math. Phys. 20 (1979)

1244.

[10] DE GENNES, P.-G., Phys. Lett. 38A (1972) 339.

DES CLOIZEAUX, J. Physique 36 (1975) 281.

[11] DUPLANTIER, B., C. R. Heb. Séan. Acad. Sci. 290B

(1980) 199. See also :

SCHAFER, L., WITTEN, T. A., J. Chem. Phys. 66 (1977)

2121.

[12] BREZIN, E., LE GUILLOU, J. C., ZINN-JUSTIN, J., in Phase Transitions and Critical Phenomena, Vol. 6, Editors C. Domb, M. S. Green (Academic Press)

1976.

[13] DES CLOIZEAUX, J., DUPLANTIER, B., J. Physique Lett.

46 (1985) L-457. For former calculations of the form factor, see

WITTEN, T. A., SCHAFER, L., J. Chem. Phys. 74 (1981) 2582 ;

OHTA, T., OONO, Y., FREED, K. F., Phys. Rev. A (1982)

2801.

[14] BENHAMOU, M., MAHOUX, G., J. Physique Lett. 46 (1985) L-689.

[15] KHOLODENKO, A. L., FREED, K. F., Opus cit.

[16] VLADIMIROV, A. A., KAZAKOV, D. I., TARASOV, O. V., Th. Eksp. Teor. Fiz. 77 (1979) 1035.

[17] YAMAKAWA, H., Modern Theory of Polymer Solutions

(Harper and Row) 1971.

Références

Documents relatifs

The aim of the present article is to present this direct renormalization method for semi-attractive chains, and to prove that it ensures renormalization to all

2014 The polymer excluded volume exponent 03BD is calculated using a direct position space renormalization group approach without reference to the zero-component

and obtained, via a general loop expansion, the physical tricritical laws for polymer solutions. We have calculated the square radius :ll2 of a chain in. solution,

Study of the surface tension of polymer solutions: theory and experiments in theta solvent conditions... Study of the surface tension of polymer

A version of field theory was used in the first formulation of polymer problems in a field theoretic formulation by Edwards and Freed [7, 8] but this original work

In strongly interacting systems, such as excluded volume melts or polymer blends and block copolymer melts ail interactions lead to extra forces which affect trie dynamics of a

These Green’s functions will once again be defined by diagrams.. Lagrangian formalism with external field. 3). Thus, by construction, these quantities

single polymer molecule [1].) In the poor solvent, the form of the overall density-density correlation function was first given by Edwards [6] and in section