Abstract
We give in these notes a short presentation of both the main ideas underlying Wilson’s renormalization group (RG) and their concrete implementation under the form of what is now called the non-perturbative renormalization group (NPRG) or sometimes the functional renormalization group (which can be perturbative). Prior knowledge of perturbative field theory is not required for the understanding of the heart of the article.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Let us emphasize that apart from the field renormalization, the whole renormalization process is nothing but a reparametrization [6].
- 2.
In quantum field theory, Feynman diagrams represent the summation over probability amplitudes corresponding to all possible exchanges of virtual particles compatible with a given process at a given order. Note that these integrals are cut-off in the ultraviolet by \(\varLambda \) and in the infrared by the “mass” \(r_0\) (see Eq. (2.6)). In statistical mechanics, the mass is related to the correlation length \(\xi \) by \(r_0\sim \xi ^{-2}\) (at the mean-field approximation).
- 3.
It is extremely rare that renormalization group enables to solve exactly a model that was not already solved by another and simpler method.
- 4.
Note that this is true only for universal quantities. The critical temperatures for instance, which are non-universal, depend on microscopic details such as the shape of the lattice. We shall come back on this notion in the following.
- 5.
Let us already mention that if Wilson’s RG equations are truncated in a perturbation expansion, all the usual perturbative results are recovered as expected.
- 6.
Fixed point hamiltonian is meant in the usual sense. If \(H(\varvec{K}^*,S_i)\) is a fixed point hamiltonian this means that \(\varvec{K}^*\rightarrow \varvec{K}^*\) by summation over the \(\sigma _I^{\alpha \pm }\)’s, Eq. (2.19). The rescaling of the lattice spacing, which is equivalent to measuring all dimensionful quantities in terms of the running lattice spacing and not in terms of the (fixed) initial lattice spacing, is a necessary step to obtain fixed point hamiltonians.
- 7.
We shall see in the following that this rescaling induces a rescaling of all coupling constants as well as of the magnitude of the spin, the so-called field renormalization.
- 8.
If we wanted to compute correlation functions of the original spins, we would have first to couple the system to an arbitrary magnetic field (in order to be able to compute derivatives of \(Z\) with respect to the magnetic field \(B_i\)). This is a complicated task.
- 9.
Something is said to be self-similar if it looks everywhere the same. In our case, the self-similar character comes from the fact that the functional form of the RG flow does not depend on the initial couplings \(\varvec{K}^{(0)}\) since the same function \(\varvec{T}\) is used to transform \(\varvec{K}^{(0)}\) into \(\varvec{K}^{(p)}\) or \(\varvec{K}^{(r)}\) into \(\varvec{K}^{(p)}\). This results in the fact that the right hand side of Eq. (2.25) is independent of \(r\) since the left hand side is. This independence is completely similar to the independence of the bare theory on the renormalization scale in perturbative renormalization (or of the renormalized theory on the bare scale). This is what allows to derive the Callan-Symanzik RG equations in the perturbative context.
- 10.
- 11.
Once the continuum limit has been taken and continuous RG transformations are implemented this means that the effective hamiltonians involve all powers of the field and of its derivatives.
- 12.
Let us point out here a subtlety. This statement is not fully rigorous since the original spins are Ising spins whereas the block spins \(\mathcal{ S}_I\) are not. The correlation functions \(\langle S_i^I S_j^J\rangle \) and \(\langle \mathcal{ S}_I \mathcal{ S}_J \rangle \) are therefore not computed exactly in the same way since the summation over the configurations of \(S_i\) and of \(\mathcal{ S}_I\) do not run on the same values. In fact, after several blocking iterations, the spins that are summed over become almost continuous variables and the aforementioned difficulty disappears.
- 13.
It will also be convenient to rescale the spin-field by the appropriate power of \(a\): \(S_i\rightarrow \phi (\varvec{x})\) so that the gradient term \((\nabla \phi )^2\) comes in the hamiltonian of the field theory with a dimensionless pre-factor (chosen to be 1/2 for convenience): \(H~=~\int d^dx\,\left( \frac{1}{2}(\nabla \phi )^2 +U(\phi ) \right)\). We find from this equation that \([\phi (x)]=[x^{-\frac{d-2}{2}}]\) so that the rescaling involves a factor \(a^{-\frac{d-2}{2}}\). Note that the original variables \(S_i\) are dimensionless since \(S_i=\pm 1\). The function \(G^{(2)}\) in Eq. (2.39) is therefore also dimensionless.
- 14.
In perturbation theory, the fact that the \(\beta \) function of the marginal coupling is cut-off independent and thus scale independent is a consequence of the perturbative renormalizability of the model.
- 15.
It can happen that some eigenvalues are complex. In this case, the RG flow around the fixed point is spiral-like (focus) in the corresponding eigendirections.
- 16.
We first consider an \(N\)-dimensional gaussian integral and then take the limit \(N\rightarrow \infty \).
- 17.
The \(O(N)\) models are completely exceptional in this respect since they are the only ones for which it has been proven that the series of the \(\beta \)-function is Borel-summable in \(d=3\) (in the so-called zero momentum massive scheme). In all other cases, either this is not known or it is known that the series are not Borel-summable. For QED, this is not yet a problem because the smallness of the coupling constant, the fine structure constant, ensures up to now an apparent convergence of the perturbative results.
- 18.
These calculations did not correspond to a series expansion of the exact NPRG equation on \(V_k\). They enabled to retrieve the one-loop results easily but became very cumbersome beyond one-loop.
- 19.
- 20.
It is impossible to get easily any physical information from it except at “mean field-like” level: a functional integral has still to be performed.
- 21.
The Helmoltz free energy is \(F=-k_B T \log Z[B]\). It is a functional of the source \(B(x)\). The Gibbs free energy is obtained by a Legendre transform from \(F\) and is a functional of the magnetization \(M(x)\). It is the generating functional of the one-particle irreducible (1PI) correlation functions.
- 22.
Let us emphasize that at the mean-field approximation, the Gibbs free energy of the system is identical to the hamiltonian. Equation (2.111) is an exact version of this statement (remember that no fluctuation is taken into account at the mean-field level).
- 23.
The function \(Z_k(M)\) has, of course, nothing to do with the partition function \(Z_k[B]\) introduced in Eq. (2.113) although it is customary to use the same name for both functions.
- 24.
The computation of quantities like the total magnetization or the susceptibilities require only the knowledge of the spin-spin correlation function at zero momentum. The same thing holds for the correlation length.
- 25.
Let us notice that if the \(k\)-dependence of the couplings was neglected, this ansatz would exactly coincide with the ansatz chosen by Landau to study second order phase transitions. We know that it would lead to the mean field approximation. It is remarkable that keeping the scale dependence of the couplings and substituting precisely this ansatz into the RG equation of \(\varGamma _k\) is sufficient to capture almost all the qualitative features of the critical physics of the Ising and \(O(N)\) models in all dimensions (see the following).
- 26.
The derivation of these equations is outlined in Subsect. 2.2.7.
- 27.
- 28.
At vanishing external magnetic field, the magnetization is given by \(B=0=\partial U/\partial M\) and corresponds therefore to the location of the minimum of the potential.
- 29.
We shall see in the following that there is a subtlety here because of the convexity of the potential in the limit \(k\rightarrow 0\).
- 30.
For \(d\le 2\) it is necessary to take into account the field renormalization, that is the anomalous dimension, to obtain a coherent picture of the physics. This requires to go beyond the LPA, see Subsect. 2.2.7.
- 31.
The following relation is not general since if we choose a point on the critical surface, the corresponding potential is attracted towards \(\tilde{U}^*(\tilde{M})\). This approach is governed by the so-called critical exponent \(\omega \) corresponding to correction to scaling. Thus, in general, infinitesimally close to the fixed point, the evolution of the potential is given by the sum of two terms, one describing the approach to \(\tilde{U}^*\) on the critical surface and one describing the way the RG flow escapes the fixed point if one starts close but away from the critical surface [18].
- 32.
- 33.
At first sight, this statement could seem incorrect in particle physics if one considers a given diffusion process. For instance the reaction \(e^+ e^-\rightarrow \gamma \gamma \) seems to involve four bodies only. This is actually wrong since, as virtual states involved in the loop expansion, an arbitrarily large number of particles can be exchanged during this reaction. The full Fock space structure is thus necessary to describe any kind of diffusion in the quantum and relativistic framework. The same is true in statistical mechanics. The infinite number of degrees of freedom of the system is not the relevant point. A perfect gas for instance can well involve infinitely many degrees of freedom, we all know that the whole machinery of field theory is not necessary to study it. The important point is the number of degrees of freedom that effectively interact together, that is the value of the correlation length. As long as \(\xi \sim a\) field theory is not necessary since the system breaks down in small sub-systems of size \(\xi \) that are almost independent of each other. This is the reason why the law of large numbers is valid in this case and the fluctuations gaussian. Field theory is relevant only when \(\xi \gg a\) in which case, for length scales \(l\) such that \(a\ll l\ll \xi \) field theory is relevant.
- 34.
Remember that \(U_k\) corresponds to the zero momentum configuration of all \(\varGamma ^{(n)}_k\).
- 35.
Truncating the field dependence of the running potential \(U_k\) by keeping only the \(M^2\) and \(M^4\) terms at all scales \(k\) is nothing but a very crude approximation that eventually leads to neglecting all functions \(\varGamma ^{(n)}\) with \(n\ge 6\) in the effective action \(\varGamma [M]=\varGamma _{k=0}[M]\).
- 36.
All qualitative features that are explained below do not depend on this approximation.
- 37.
In general, the attractive submanifold is of dimension the number of (perturbatively) renormalizable couplings including the masses. In the critical case, the mass is vanishing by definition and the attractor is of dimension one for models for which only one coupling is renormalizable.
- 38.
One should remember that the space of coupling constants is infinite dimensional and that it is therefore non-trivial that the projection of \(L\) onto the \(g_4\)-axis be non vanishing. A randomly chosen vector in an infinite dimensional space has in general a vanishing projection onto a given direction.
- 39.
For the Ising model, different natural initial conditions could correspond for instance to different kinds of lattices or to different types of couplings among the spins: next nearest neighbor couplings, anisotropic couplings, etc.
- 40.
Let us mention here that within dimensional regularization, for instance, there is no explicit ultraviolet regularization scale. The only scale introduced in this regularization scheme is the scale, often called \(\mu \), necessary to preserve dimensional analysis when, for loop-integrals, \(\int d^4 q\) is replaced by \(\int d^dq\): \(\int d^4 q\rightarrow \mu ^{4-d}\int d^d q\). This scale is also used most of the time as the scale of the renormalization prescriptions that are either explicit, see Eq. (2.171), or implicit as in the \(\overline{\text{ MS}}\) scheme (the fact that they can be implicit does not change anything to our discussion). However, any regularization consists in modifying the short distance behavior of the theory because this is where the divergences come from. Thus the ultraviolet cut-off, although not explicit must be built from \(\mu \) and \(\epsilon =4-d\). From a comparison of the divergences obtained in dimensional regularization and in the cut-off regularization, it is easy to get a qualitative correspondence between the two schemes and thus an estimate of the UV cut-off scale of dimensional regularization. When a logarithmic divergence occurs in the cut-off regularization scheme, a pole in \(\epsilon \) occurs in dimensional regularization. Thus it is reasonable to imagine that \(\log \varLambda \sim 1/\epsilon \). To make this correspondence dimensionally valid we must have \(\log \varLambda /\mu \sim 1/\epsilon \). We therefore find that in dimensional regularization, the UV scale behaves as \(\varLambda \sim \mu \exp (1/\epsilon )\).
- 41.
Initializing this flow in the IR has several advantages. First, as we already mentioned, this is anyway obligatory once the infinite cut-off limit has been taken since, in this case, any reference to a microscopic model defined at an UV scale has been lost. Second, if the model under study is not derived from a more fundamental model at scale \(\varLambda \)—in which case the analytical form of \(\varGamma _\varLambda \) is not known—, its initialization at scale \(\varLambda \) would require in general infinitely many phenomenological input parameters since \(\varGamma _\varLambda \) is in general not polynomial. This is of course impossible and should be compared to what is done in the IR: only the values of the renormalizable couplings have to be fixed since the dimension of the submanifold \(L\) is the number of renormalizable terms. For the Ising model for instance, it is possible to initialize the flow at scale \(\varLambda \) since already at this scale it is an effective model derived from a more fundamental model (coming for instance from the Hubbard model). The same occurs for fluids for instance: microscopic models can be derived from other models that are more fundamental.
- 42.
Reciprocally, any model defined at an UV scale \(\varLambda \) on the critical surface behaves at large distance as if it was free (this is the so-called “triviality problem” of the \(\phi ^4\) theory in \(d=4\)). This is the reason why \(d=4\) is the upper critical dimension of the \(\mathbb{ Z} _2\)-invariant models.
- 43.
Once again we deal only with a massless theory. The presence of a mass does not change qualitatively the discussion below.
- 44.
One should remember that we are dealing here with dimensionless coupling constants. In a natural theory, we expect all dimensionful coupling constants, at scale \(\varLambda \) for instance, to be of order of \(\varLambda \). This precisely means that the dimensionless coupling constants must all be of order one. Of course, the continuum limit does not exist for a non-renormalizable model.
- 45.
In this sense, working with the dimensionless and renormalized quantities consists in going to a “co-moving frame” where the explicit \(k\)-dependence has been eliminated.
- 46.
Let us emphasize that this is not the case with the Wilson-Polchinski approach: even the one loop result in \(d=4-\epsilon \) is not reproduced at any finite order of the derivative expansion [62].
- 47.
Needless to say that this is completely out of reach of the perturbative expansions performed either from the \(\phi ^4\) theory (around \(d=4\)) or from the non-linear sigma model (around \(d=2\)).
- 48.
Although not proven, it is reasonable to assume that the smaller the anomalous dimension the better the convergence of the derivative expansion (\(\eta =0\) within the LPA). Thus \(\eta \) is perhaps the “small parameter” of the derivative expansion.
- 49.
References
Le Bellac, M.: Quantum and Statistical Field Theory. Clarendon Press, Oxford (1991)
Binney, J.J., Dowrick, N.J., Fisher, A.J., Newman, M.E.J.: The Theory of Critical Phenomena: An Introduction to the Renormalization Group. Oxford University Press, Oxford (1992)
Goldenfeld, N.: Lectures on Phase Transitions and the Renormalization Group. Addison-Wesley, Reading (1992)
Ryder, L.H.: Quantum Field Theory. Cambridge University Press, Cambridge (1985)
Zinn-Justin, J.: Quantum Field Theory and Critical Phenomena, 3rd edn. Oxford University Press, New York (1989)
Delamotte, B.: Am. J. Phys. 72, 170 (2004)
Kadanoff, L.P.: Physica 2, 263 (1966)
Wegner, F.J., Houghton, A.: Phys. Rev. A 8, 401 (1973)
Wilson, K.G., Kogut, J.: Phys. Rep., Phys. Lett. 12C, 75 (1974)
Wegner F.J.: The critical state, general aspects. In: Domb, C., Greene, M.S. (eds.) Phase Transitions and Critical Phenomena, vol. 6. Academic Press, New York (1976)
Weinberg, S.: Critical phenomena for field theorists. In: Zichichi, A. (ed.) Understanding the Fundamental Constituents of Matter. Plenum Press, New York (1978)
Polchinski, J.: Nucl. Phys. B 231, 269 (1984)
Shirkov, D.V., Kovalev, V.F.: Phys. Rep. 352, 219 (2001)
Fortuin, C.M., Kasteleyn, P.W.: Physica 57, 536 (1972)
Swendsen, R.H., Wang, J.S.: Phys. Rev. Lett. 58, 86 (1987)
Niemeijer, T., van Leuwen, J.: Renormalization: ising-like spin systems. In: Domb, C., Greene, M.S. (eds.) Phase Transitions and Critical Phenomena, vol. 6. Academic Press, New York (1975)
Golner, G.R., Riedel, E.K.: Phys. Rev. Lett. 34, 856 (1975)
Hasenfratz, A., Hasenfratz, P.: Nucl. Phys. B [FS] 270, 687 (1986)
Nicoll, J.F., Chang, T.S., Stanley, H.E.: Phys. Rev. Lett. 33, 540 (1974). err. ibid. 33, 1525 (1974)
Nicoll, J.F., Chang, T.S., Stanley, H.E.: Phys. Lett. A 57, 7 (1976)
Nicoll, J.F., Chang, T.S.: Phys. Lett. A 62, 287 (1977)
Ringwald, A., Wetterich, C.: Nucl. Phys. B 334, 506 (1990)
Wetterich, C.: Nucl. Phys. B 352, 529 (1991)
Tetradis, N., Wetterich, C.: Nucl. Phys. B 383, 197 (1992)
Wetterich, C.: Phys. Lett. B 301, 90 (1993)
Ellwanger, U.: Z. Phys. C 58, 619 (1993)
Ellwanger, U.: Z. Phys. C 62, 503 (1993)
Tetradis, N., Wetterich, C.: Nucl. Phys. B [FS] 422, 541 (1994)
Ellwanger, U.: Phys. Lett. B 335, 364 (1994)
Morris, T.R.: Int. J. Mod. Phys. A 9, 2411 (1994)
Morris, T.R.: Phys. Lett. B 329, 241 (1994)
Berges, J., Tetradis, N., Wetterich, C.: Phys. Rep. 363, 223 (2002)
Delamotte, B., Mouhanna, D., Tissier, M.: Phys. Rev. B 69, 134413 (2004)
Newman, K.E., Riedel, E.K.: Phys. Rev. B 30, 6615 (1984)
Newman, K.E., Riedel, E.K., Muto, S.: Phys. Rev. B 29, 302 (1984)
Bervillier, C.: Phys. Lett. A 331, 110 (2004)
Kunz, H., Zumbach, G.: J. Phys. A 26, 3121 (1993)
Zumbach, G.: Phys. Rev. Lett. 71, 2421 (1993)
Zumbach, G.: Nucl. Phys. B 413, 754 (1994)
Zumbach, G.: Phys. Lett. A 190, 225 (1994)
Kleinert, H., Schulte-Frohlinde, V.: Critical Properties of Phi 4-Theories. World Scientific, Singapore (2001)
Liao, S., Polonyi, J., Strickland, M.: Nucl. Phys. B 567, 493 (2000)
Litim, D.: J. High Energy Phys. 0111, 059 (2001)
Litim, D.F.: Nucl. Phys. B 631, 128 (2002)
Canet, L., Delamotte, B., Mouhanna, D., Vidal, J.: Phys. Rev. D 67, 065004 (2003)
Canet, L., Delamotte, B., Mouhanna, D., Vidal J.: Phys. Rev. B 68, 064421 (2003)
Tissier, M., Delamotte, B., Mouhanna, D.: Phys. Rev. Lett. 84, 5208 (2000)
Bagnuls, C., Bervillier, C.: Phs.Rev. B 41, 402 (1990)
Bagnuls, C., Bervillier, C.: Phys. Rep. 348, 91 (2001)
Bervillier, C.: J. Phys. Condens. Matter 17, S1929 (2005)
Bagnuls, C., Bervillier, C.: Int. J. Mod. Phys. A 16, 1825 (2001)
Blaizot, J.-P., Galain, R.M., Wschebor, N.: Phys. Rev. E 74, 051116 (2006)
Blaizot, J.-P., Galain, R.M., Wschebor, N.: Phys. Lett. B 632, 571 (2006)
Blaizot, J.-P., Galain, R.M., Wschebor, N.: Phys. Rev. E 051117 (2006)
Litim, D.: Mind the gap. Int. J. Mod. Phys. A 16, 2081 (2001)
Morris, T.R.: Phys. Lett. B 345, 139 (1995)
Gersdorff, G.V., Wetterich, C.: Phys. Rev. B 64, 054513 (2001)
Ballhausen, H., Berges, J., Wetterich, C.: Phys. Lett. B 582, 144 (2004)
Seide, S., Wetterich, C.: Nucl. Phys. B 562, 524 (1999)
Canet, L., Chaté, H., Delamotte, B.: Phys. Rev. Lett. 92, 255703 (2004)
Itzykson, C., Drouffe, J.M.: Statistical Field Theory. Cambridge University Press, Cambridge (1989)
Morris, T.R., Tighe, J.F.: J. High Energy Phys. 08, 007 (1999)
Gies, H.: hep-ph/0611146 (2006)
Pelissetto, A., Vicari, E.: Phys. Rep. 368, 549 (2002)
Calabrese, P., Parruccini, P., Pelissetto, A., Vicari, E.: Phys. Rev. B 70, 174439 (2004)
Canet, L., Delamotte, B., Deloubrière, O., Wschebor, N.: Phys. Rev. Lett. 92, 195703 (2004)
Canet, L., Chaté, H., Delamotte, B., Dornic, I., Munoz, M.A.: Phys. Rev. Lett. 95, 100601 (2005)
Tissier, M., Mouhanna, D., Vidal, J., Delamotte, B.: Phys. Rev. B 65, 140402 (2002)
Fisher, D.S.: Phys. Rev. Lett. 56, 1964 (1986)
Le Doussal, P., Wiese, K.J.: Phys. Rev. lett. 89, 125702 (2002)
Feldman, D.E.: Phys. Rev. Lett. 88, 177202 (2002)
Le Doussal, P., Wiese, K.J., Chauve, P.: Phys. Rev. E 69, 026112 (2004)
Tarjus, G., Tissier, M.: Phys. Rev. Lett. 93, 267008 (2004)
Tissier, M., Tarjus, G.: Phys. Rev. Lett. 96, 087202 (2006)
Strumia, A., Tetradis, N.: Nucl. Phys. B 542, 719 (1999)
Strumia, A., Tetradis, N.: JHEP, 9911, 023 (1999)
Diehl, S., Wetterich, C.: Phys. Rev. A 73, 033615 (2006)
Diehl, S., Gies, H., Pawlowski, J.M., Wetterich, C.: Phys. Rev. A 76, 021602 (2007)
Parola, A., Reatto, L.: Adv. Phys. 44, 211 (1995)
Caillol, J.-M.: Mol. Phys. 104, 1931 (2006)
Delamotte, B., Canet, L.: Condens. Matter Phys. 8, 163 (2005)
Acknowledgments
I want first to thank D. Mouhanna without whom the little group in Paris (with an antenna in Grenoble) working on the NPRG would have never existed. I owe my understanding of this subject to many discussions and common works with him. I also thank M. Tissier with whom I collaborated and from whom I have learned a lot. It is also a great pleasure to thank L. Canet and H. Chaté with whom I have collaborated and discussed many aspects of statistical mechanics and renormalization. I also thank I. Dornic and J. Vidal with whom I collaborated and G. Tarjus, J. Berges, C. Bervillier and C. Wetterich for many discussions about the NPRG. More recently, discussions with R. Mendez-Galain and N. Wschebor on their own version of the NPRG have greatly improved my understanding of this subject and I want to thank them for that. I also thank J-M. Caillol for several discussions. Finally, I thank Yu. Holovatch who encouraged me to write down my lecture notes, the students who helped me to improve them (P. Hosteins and G. Gurtner in particular) and A. Schwenk and J.-P. Blaizot who invited me in Trento to give a set of lectures on the NPRG.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix
Appendix
“Nobody ever promised you a rose garden.”
J. Polchinski
2.1.1 Definitions, Conventions
\(\bullet \) Integrals in \(x\) and \(q\) spaces
In real and Fourier spaces we define
\(\bullet \) Fourier transform
Depending on the context, we omit or not the tilde on the Fourier transform.
\(\bullet \)Definition of\(v_d\)
with
\(\bullet \)Functional derivatives
\(\bullet \)Correlation functions
\(\varGamma [M]\) is a functional of \(M(x)\). We define the 1PI correlation functions by
We also define the Fourier transform of \(\varGamma ^{(n)}\) by
and also
The relation between \(\tilde{\varGamma }^{(n)}\) and \(\bar{\varGamma }^{(n)}\) follows from Eq. (2.208):
\(\bullet \)Cut-off function in\(x\)and\(q\)spaces
One should be careful about the fact that \(R_k\) is sometimes considered as a function of \(q\) and sometimes as a function of \(q^2\). It can be convenient to define a cut-off function with two entries by
Then
\(\bullet \)\(k\)-dependent anomalous dimension
By definition:
\(\bullet \) Threshold functions \(l_n^{d}\)
where
and, by definition of \(s(y)\)
\(\bullet \) Threshold functions \(m_{n_1,n_2}^{d}\)
with
\(\bullet \) Universal value of \(l_n^{2n}(0,0)\) for \(n>0\)
For \(n>0\) and independently of the choice of cut-off function \(R_k\):
\(\bullet \) Derivative of \(l_n^{d}\)
\(\bullet \) \(\theta \)-cut-off
A convenient cut-off function \(R_k\) that allows to compute analytically some threshold functions is
With this cut-off we find
\(\bullet \) Threshold functions \(l_n^{d}\) and \(m_{2,2}^d\) with the \(\theta \)-cut-off
With the cut-off function, Eq. (2.224), the \(l_n^d\) threshold functions can be computed analytically
2.1.2 Proof of Eq. (2.106)
We define:
and we rewrite the exponent:
We now define
and change variables: \((y,z)\rightarrow (u,x)\). The jacobian is 1 and thus:
2.1.3 The Exact RG Equations
For the sake of simplicity, we consider a scalar theory (e.g. Ising). We have by definition
with, by definition of \(M(x)\):
When \(B(x)\) is taken \(k\)-independent (as in \(Z_k[B]\)) then \(M(x)\) computed from \(W_k\) is \(k\)-dependent. Reciprocally, if \(M(x)\) is taken fixed (as in \(\varGamma _k[M]\)), then \(B(x)\) computed from Eq. (2.236) becomes \(k\)-dependent.
2.1.3.1 RG Equation for \(W_k[B]\)
We therefore obtain for \(W_k\):
which is equivalent to the Polchinski equation.
2.1.3.2 RG Equation for \(\varGamma _k[M]\)
We first derive the reciprocal relation of Eq. (2.232). The Legendre transform is symmetric with respect to the two functions that are transformed. Here the Legendre transform of \(W_k\) is \(\varGamma _k+1/2 \int R_k M M\). Thus
and then
In the Polchinski equation (2.234), the \(k\)-derivative is taken at fixed \(B_x\). We must convert it to a derivative at fixed \(M\):
Acting on Eq. () with \({{\partial _k}_{\vert }}_{B}\), we obtain:
Substituting Eqs. (2.236, 2.234, 2.237) into this equation we finally obtain
The last step consists in rewriting the right hand side of this equation in terms of \(\varGamma _k\) only. We start from (2.232) and act on it with \(\delta /\delta M_z\):
Now, using (2.236), we obtain
We define
and thus
\(\varGamma ^{(2)}_k+\mathcal{ R}_k\) is therefore the inverse of \(W^{(2)}_k\) in the operator sense. Note that although we did not specify it, \(W^{(2)}_k\) is a functional of \(B(x)\) and \(\varGamma ^{(2)}_k\) a functional of \(M(x)\). Relation (2.243) is valid for arbitrary \(M\). The RG equation (2.239) can now be written in terms of \(\varGamma _k\) only:
In Fourier space this equation becomes:
2.1.3.3 RG Equation for the Effective Potential
The derivative expansion consists in expanding \(\varGamma _k\) as
where we have supposed that the theory is \(\mathbb{ Z} _2\) symmetric so that \(U_k, Z_k, \dots \) are functions of \(M^2\) only. To compute the flow of these functions it is necessary to define them from \(\varGamma _k\). The effective potential \(U_k\) coincides with \(\varGamma _k\) when it is evaluated for uniform field configurations \(M_{\text{ unif.}}\):
where \(\varOmega \) is the volume of the system. It is easy to derive an RG equation from this definition of \(U_k\) if we use the local potential approximation (LPA) that consists in truncating \(\varGamma _k\) as in (2.246) with \(Z_k(M)=1\):
By acting on Eq. (2.247) with \({\partial _k}\) we obtain:
Thus, we have to invert \( \varGamma ^{(2)}_k+R_k\) for a uniform field configuration and within the LPA. From now on, we omit the superscript LPA on \(\varGamma _k\). An elementary calculation leads to
Using \(\delta (q=0)=\varOmega (2\pi )^{-d}\) we find
It is convenient to re-express this equation in terms of
which is the \(\mathbb{ Z} _2\)-invariant.
where \(U_k^{\prime }(\rho )\) and \(U_k^{\prime \prime }(\rho )\) are derivatives of \(U_k\) with respect to \(\rho \).
To obtain the RG equation for the dimensionless potential we have to perform the change of variables of Eq. 2.154. We find
and
Inserting these relations together with Eq. (2.154) and with
in Eq. (2.253) leads to the RG equation on \(\tilde{U}_t\), Eq. (2.155).
Rights and permissions
Copyright information
© 2012 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Delamotte, B. (2012). An Introduction to the Nonperturbative Renormalization Group. In: Schwenk, A., Polonyi, J. (eds) Renormalization Group and Effective Field Theory Approaches to Many-Body Systems. Lecture Notes in Physics, vol 852. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-27320-9_2
Download citation
DOI: https://doi.org/10.1007/978-3-642-27320-9_2
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-27319-3
Online ISBN: 978-3-642-27320-9
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)