1:
2:
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: % Abbreviations
5: \newcommand{\be}{\begin{equation}}
6: \newcommand{\ee}{\end{equation}}
7: % End of abbreviations
8: \renewcommand{\theequation}{\thesection .\arabic{equation}}
9: \newcommand{\sign}{\mathop{\rm sign}\nolimits}
10: \newcommand{\tr}{\mathop{\rm tr}\nolimits}
11: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
12:
13:
14:
15: \documentstyle[aps,epsf,multicol]{revtex}
16: \begin{document}
17: \draft
18: \widetext
19: \title{The $\theta$ vacuum reveals itself as the fundamental theory of the
20: quantum Hall effect ~~~~~~~~~~~~~~~~II. The Coulomb interaction}
21: \author{A.M.M. Pruisken$^{1,2}$, M.A. Baranov$^{2,3}$ and
22: I.S. Burmistrov$^{4}$ }
23:
24: \address{$^{1}$ Indian Institute of Science, Bangalore 560012, India}
25: \address{$^{2}$ Institute for Theoretical Physics, Valckenierstraat 65, 1018XE Amsterdam, The Netherlands}
26: \address{$^{3}$ RRC Kurchatov Institute, Kuchatov Square 1, 123182 Moscow, Russia}
27: \address{$^{4}$L.D. Landau Institute for Theoretical Physics, Kosygina str. 2, 117940 Moscow, Russia }
28:
29: \maketitle
30:
31:
32: \begin{abstract}
33: \noindent{Within the Grassmannian $U(2N) / U(N) \times U(N)$ non-linear
34: $\sigma $ model representation of localization one can study the low energy
35: dynamics of both the free and interacting electron gas. We study the
36: cross-over between these two fundamentally different physical problems.
37: We show how the topological arguments for the exact quantization of the Hall conductance
38: are extended to include the Coulomb interaction problem. We discuss dynamical scaling
39: and make contact with the theory of variable range hopping.}
40: \end{abstract}
41:
42: \pacs{PACSnumbers 71.10Pm, 72.10.-d, 73.43.-f}
43:
44: \begin{multicols}{2}
45: \narrowtext
46:
47: Over the last few years, much effort has been devoted
48: to the problem of localization and interaction effects
49: in the quantum Hall regime~\cite{EuroLett,paperI,paperII,paperIII}. By now it is well understood that the
50: Coulomb interaction problem falls in a non-Fermi liquid
51: universality class of transport problems with a novel
52: symmetry, named $\cal F$ {\em invariance}~\cite{paperI}. Although the
53: results for scaling are in many ways similar to those obtained for
54: the free electron gas~\cite{freepart},
55: it is important to bear in mind that the Coulomb interaction problem
56: is a far richer one.
57: Unlike the free particle problem, for example, the field theory for
58: interacting particles provides the platform for a unification of
59: the {\em fractional quantum Hall regime}
60: and the quantum theory of {\em metals}~\cite{paperI,paperIII,fracedge}.
61: The principal features of this unification are encapsulated in a scaling
62: diagram for the {\em longitudinal} and {\em Hall conductances} $\sigma_{xx}$
63: and $\sigma_{xy}$ respectivily (Fig. \ref{mappedRG}). The
64: \begin{figure}
65: \begin{center}
66: \setlength{\unitlength}{1mm}
67: \begin{picture}(70,70)(0,0)
68: \put(0,0)
69: {\epsfxsize=70mm{\epsffile{fig3.eps}}}
70: \end{picture}
71: \caption{Unified scaling diagram for the quantum Hall effect
72: in the $\sigma_{xx}$, $\sigma_{xy}$ conductivity plane.
73: The arrows indicate the direction towards the infrared.}
74: \label{mappedRG}
75: \end{center}
76: \end{figure}
77: \noindent{Finkelstein} approach to localization and
78: interaction phenomena~\cite{Finkelstein,italians}, the topological concept
79: of an instanton vacuum~\cite{freepart}
80: as well as the Chern Simons statistical gauge fields~\cite{cf} are all essential
81: in composing this diagram.\\
82: \indent{The} main objective of this Letter is to embark on the most fundamental
83: aspect of the theory, the {\em observability} and {\em precision}
84: of the quantum Hall effect. This experimental phenomenon is represented
85: in Fig. \ref{mappedRG} by the infrared stable fixed points located at
86: $\sigma_{xx} =0$ and $\sigma_{xy} = k$ (integer effect) as well as
87: $\sigma_{xy} = k/(2k+1)$ (Jain series). These fixed points, however,
88: define the strong coupling phase of the unifying action where
89: analytic work is generally impossible. In spite of ample experimental
90: evidence for its existence,
91: the {\em robust quantization} of the Hall conductance has yet
92: to be established as a fundamental but previously unrecognized feature
93: of the topological $\theta$ vacuum concept.\\
94: \indent{In} what follows we shall benefit from the advancements reported in the preceding
95: Letter. In particular, since the Finkelstein theory is formally defined
96: as a $\sigma$ model
97: on the Grassmann manifold ${U(2N)}/{U(N) \times U(N)}$
98: with $ N$ equal to $N_r$ (number of {\em replica's})
99: times $N_m$ (number of {\em Matsubara frequencies}),
100: we can now use our general knowledge on the strong coupling behavior
101: of the theory and probe, for the first time, the quantum Hall phases
102: in the interacting electron gas.\\
103: \indent{To} achieve our goals we first shall outline
104: some of the recent advancements in the field.
105: It is important to emphasize that the {\em complete} effective action for
106: interacting particles now exists~\cite{paperI}.
107: This action includes the coupling to external potentials and/or Chern Simon
108: gauge fields.
109: This leads to a detailed understanding
110: of the electrodynamic $U(1)$ gauge invariance and provides invaluable
111: information on the renormalization of the theory that was not available
112: before.%\\
113: \indent{Secondly,} it is necessary to have a more detailed understanding of
114: how the subtleties of interaction effects can be understood as a field theory.
115: For this purpose we report new results on the Grassmannian non-linear $\sigma $ model
116: with $N_r =0$ and varying $N_m$. These show explicitly how the cross-over takes
117: place between a theory of {\em free particles} at finite values of $N_m$ and a {\em many body}
118: theory that is generally obtained in the limit $N_m \rightarrow \infty$ only.
119: Armed with these insights we next point out how the Coulomb interaction problem, at
120: zero temperature ($T$), displays the general topological features and
121: $\theta$ dependence that were discovered in the preceding Letter.\\
122: \indent{As} a third and final step toward the strong coupling phase we
123: discuss the subject of dynamical scaling.
124: As a unique product of our effective action procedure we obtain a distinctly different
125: behavior at finite $T$, depending on the specific regime of the interacting electron gas
126: that one is interested in. We establish, at the same time, the contact with the theory for
127: {\em variable range hopping}~\cite{hopping}.
128:
129: \vskip 1mm
130: \noindent{$\bullet$ } {\em The action.~~~}
131: Following Finkelstein~\cite{Finkelstein}, the effective quantum theory for
132: disordered (spin polarized or spinless) electrons is given in terms
133: of a generalized $\sigma$ model involving the unitary matrix field variables
134: ${Q}_{n,m}^{\alpha,\beta} (\vec{x} )$
135: which obey $Q^2 = {\bf 1}$. Here,
136: $\alpha, \beta$ represent the replica indices, $n,m$ are the indices of the Matsubara frequencies
137: $\nu_{k} = \pi T(2k+1)$ with $k=n,m$. In terms of ordinary unitary rotations
138: ${\cal V}_{n,m}^{\alpha,\beta}$ one can write
139:
140: \be
141: Q={\cal V}^{-1} \Lambda {\cal V},~~~\Lambda =\Lambda_{n,m}^{\alpha,\beta} = \sign (\omega_n )
142: \delta_{n,m}^{\alpha,\beta},
143: \label{Qmatrix}
144: \ee
145: indicating that the $Q$ describes a Goldstone manifold
146: of a broken symmetry between positive and negative frequencies.
147: A $U(1)$ gauge transformation in frequency space is
148: represented by a unitairy matrix ${\cal W}_{n,m}^{\alpha,\beta}$
149:
150: \be
151: {\cal W} = \exp \{i \sum_{n , \alpha} \phi^\alpha (\omega_n ) {I}_n^\alpha \} .
152: \label{Wmatrix}
153: \ee
154: Here,
155: $[{I}_k^\gamma ]_{n,m}^{\alpha,\beta} = \delta^{\alpha\gamma} \delta^{\beta\gamma}\delta_{n,m+k}$
156: denote the $U(1)$ generators.
157: In finite frequency space with a cut-off ($N_m$), the $I$ matrices no longer span a $U(1)$
158: algebra. To define the $U(1)$ gauge invariance in a truncated frequency
159: space we have developed a set of rules
160: ($\cal F$ {\em algebra}~\cite{paperI}). These involve
161: one more (frequency) matrix,
162: $\eta_{n,m}^{\alpha,\beta} = n \delta_{n,m}^{\alpha,\beta}$, that is used to represent
163: $\omega_n$.
164: The effective action for electrons in a static magnetic field $B$ and coupled to external
165: potentials and/or Chern Simons fields $a_{\mu}^{\alpha}(\vec x , \omega_n)$
166: with $\omega_n = 2 \pi T n \neq 0$,
167: can now be written as~\cite{paperI}
168:
169: \be
170: S_{eff} = S_\sigma +S_F +S_U + S_0 .
171: \label{Seff}
172: \ee
173: Here, $S_{\sigma}$ is the free electron piece~\cite{freepart}
174: \be
175: S_{\sigma} = \frac{\sigma_{xx}^0}{8} \int {\rm tr} [\vec D,Q]^2
176: +\frac{\sigma_{xy}^0}{8} \int {\rm tr} \epsilon_{ij}
177: Q[D_i,Q][D_j,Q] ,
178: \label{Ssigma}
179: \ee
180: where
181: $D_j = \nabla_j -i \sum_{n\alpha} a_j^{\alpha} (\vec{x} , \omega_n ) {I}_n^{\alpha} $
182: is the covariant derivative.
183: Next, the two pieces $S_F$ and $S_U$ are linear in $T$ and represent
184: interaction terms. $S_F$ is gauge invariant and contains the
185: {\em singlet interaction} term~\cite{Finkelstein}
186:
187: \be
188: S_{ F} = z_0 \pi T %\frac{\pi}{\beta}
189: \int \left[ {\sum_{\alpha n}} {\rm tr} {I}_n^\alpha Q
190: {\rm tr} {I}_{-n}^\alpha Q +4 {\rm tr} \eta Q - 6 {\rm tr} \eta \Lambda
191: \right].
192: \label{Sf}
193: \ee
194: The (Coulomb) term $S_U$ contains the scalar potential
195: $$
196: S_{U} = -\pi T {\sum_{\alpha n}}\int_{x} \int_{x'} \left[
197: \tr {I}_{-n}^\alpha Q(\vec{x})-\frac{1}{\pi T} {\tilde a}_0^\alpha (\vec{x}, \omega_{-n} )
198: \right]
199: \times
200: $$
201: \be
202: \;\;\;\;\;\;\;\;\;\;\; U^{-1} (\vec{x} -\vec{x'} )
203: \left[
204: \tr {I}_{n}^\alpha Q(\vec{x'})-\frac{1}{\pi T} {\tilde a}_0^\alpha (\vec{x'} , \omega_n )
205: \right] .
206: \label{Su}
207: \ee
208: $S_0$ contains the magnetic field
209: ${b}^{\alpha}=\nabla_x a^\alpha_y - \nabla_y a^\alpha_x$
210: \be
211: S_0 = -\frac{\rho_B^2 }{2\rho T}
212: \int_x \sum_{\alpha n} b^{\alpha}(\vec{x} , \omega_n)
213: b^{\alpha}(\vec{x} , \omega_{-n}) .
214: \label{S0}
215: \ee
216: We have defined (dropping the replica index $\alpha$ on $a_{\mu}$)
217: \be
218: \tilde{a}_0 =a_0 - i \rho_B b / \rho \;\;\; ; \;\; U(q)
219: = \rho^{-1} + U_0 (q).
220: \label{aU}
221: \ee
222: Here, the density of states
223: $\rho = {\partial n}/{\partial \mu}$ and the quantity $\rho_B = {\partial n}/{\partial B} $
224: are thermodynamic quantities.
225: The statement of gauge invariance now means that the theory is invariant under the following
226: transformation
227: \be
228: Q \rightarrow {\cal W}^{-1} Q {\cal W} \;\;\; ; \;\;\; a_\mu \rightarrow a_\mu + \partial_\mu \phi .
229: \label{gauge}
230: \ee
231: Using Eq. (\ref{gauge}) it is easy to see that the action is invariant under spatially
232: independent gauge transformations $\phi =\phi (\omega_n)$ provided the interaction
233: potential $U_0$
234: has an infinite range. This global invariance, termed $\cal F$ {\em invariance},
235: is an exact symmetry of the Coulomb interaction problem which in two spatial
236: dimensions is represented by $U_0^{-1} (q) = {\Gamma |q|}$.
237:
238: \vskip 1mm
239: \noindent{$\bullet$} {\em Static versus dynamic response.~~~}
240: Our introduction of external potentials (statistical gauge fields)
241: $a_\mu$ can be exploited immediately to elucidate fundamental aspects of
242: the quantum transport problem in strong $B$.
243: For this purpose we consider $S_{eff} (a_\mu)$ obtained after elimination
244: of the $Q$ fields.
245: Defining the particle density $n = T { \delta S_{eff} }/{\delta a_{0} }$ we obtain,
246: at a tree level, the continuity equation
247: $$
248: \omega_m (n+i\sigma _{xy}^{0} b ) =
249: \vec{\nabla} \cdot \left[ \sigma _{xx}^{0} ( \vec{e} + \vec{\nabla} U_{0} n )
250: - D_{xx}^0 \vec{\nabla} ( n + i \rho_B b ) \right] .
251: $$
252: Here, $D_{xx}^0 = \sigma_{xx}^0/\rho $ denotes the diffusion constant and
253: $\vec{e}$, $b$ are the external electric and magnetic fields respectively.
254: This result is familiar from the
255: theory of metals~\cite{NozPines} where the quantity $\rho_B$ is usually neglected.
256: Notice that in the static limit $\omega_n \rightarrow 0$ both quantities $\sigma_{ij}^0$
257: drop out and the equation now contains the thermodynamic
258: quantities $\rho, \rho_B$ and $U_0$ only. Since the fields
259: $a_\mu (\vec{x} , \omega_n =0)$ are completely
260: decoupled from the $Q$ field variables, the static response
261: is actually determined by a different, underlying theory.
262: This means that
263: $\rho, \rho_B$, $U_0$ and hence $S_U$ and $S_0$
264: should not have any quantum corrections in general, either perturbatively or
265: non-perturbatively. This observation can be used as a
266: general physical constraint that must be imposed on the quantum theory.
267: The only quantities that are allowed to have quantum corrections are the transport
268: parameters $\sigma_{xx}^0$, $\sigma_{xy}^0$, and the singlet interaction amplitude $z_0$.
269:
270: As an important check on the statements of gauge invariance and renormalization,
271: we have evaluated the quantum theory in $2+\epsilon$ spatial dimensions to order
272: $\epsilon^2$.
273: The results of the computation, along with an extensive analysis of dynamical scaling,
274: have been reported in~\cite{twoloop}.
275:
276: \vskip 1mm
277: \noindent{$\bullet$} $\cal F$ {\em invariance.~~~} The
278: renormalizability can be addressed more formally, by making contact with the theory of
279: ordinary non-linear sigma models~\cite{sigma}.
280: For this purpose we drop the external potentials from
281: the action and recall that for finite size matrices $Q$, operators like $S_F$
282: play the role of infrared regulators that
283: do not affect the singularity structure of the theory at short distances.
284: We know in particular that the theory is renormalizable in two
285: dimensions.
286: Besides the coupling constant or $\sigma_{xx}^0$,
287: one additional renormalization constant is needed for the operators
288: linear in the $Q$ matrix field and two more are generally needed for the operators
289: bilinear in the $Q$ (i.e. the symmetric and anti-symmetric representation respectivily)~\cite{sigma1}.
290: These general statements apply to the Finkelstein action as well
291: since the latter only demands that the number of Matsubara frequencies $N_m$ is taken to infinity
292: (along with $N_r \rightarrow 0$). To completly undust this point we have computed the
293: cross over functions for the theory where the quantity $U^{-1} ({\vec x} - {\vec x}' )$ in $S_U$,
294: Eq.(\ref{Su}), is replaced by its most relevant part
295:
296: \be
297: U^{-1} ({\vec x} - {\vec x}' ) \rightarrow z_0 (1-c_0 ) \delta ({\vec x} - {\vec x}' ) .
298: \label{zc}
299: \ee
300: Notice that $0 < c_0 < 1$ represents the
301: finite range interaction case. The extreme cases $c_0 = 0$ and $1$ describe the free electron gas
302: and the Coulomb system respectivily. $\cal F$ invariance is retained for $c_0 =1$ only
303: and broken otherwise.
304:
305: The following renormalization group functions have been obtained for the parameters $z$,
306: $c$ and the dimensionless resistance $g=\mu^\epsilon /\pi\sigma_{xx}$ in $2+\epsilon$
307: spatial dimensions ($\mu$ denotes an arbitrary momentum scale)
308:
309: \begin{eqnarray}
310: \frac{dg}{d\ln\mu} & = & \epsilon g - 2 g^2 ( f + \frac{1-c}{c} \ln (1-cf) )
311: \nonumber \\
312: \frac{d\ln z}{d\ln\mu} & = & g c f,\;\;\;\;\;\;\;
313: \frac{d c}{d\ln\mu} = g c( 1- cf).
314: \label{RGf}
315: \end{eqnarray}
316: Here, $f = M^2/(\mu^2 + M^2)$ is a $\mu$-dependent function with
317: $M^2 = 8\pi z_0 T N_m /\sigma_{xx}^0$ which depends on the cut-off $N_m$.
318:
319: For $f=0$ ($\mu >> M$ or short distances) we obtain the well known
320: results for free particles~\cite{freepart,sigma}, i.e.
321: $dg/d\ln\mu$ has no one-loop contribution,
322: $z$ has no quantum corrections in general and the result for $c$
323: coincides with the renormalization of symmetric operators, bilinear in $Q$.
324:
325: For $f=1$ ($\mu << M$ or large distances), we obtain
326: the peculiar Finkelstein
327: results of the interacting electron gas~\cite{paperII,italians}.
328: The symmetry breaking parameter $c$ now affects the renormalization
329: of all the other parameters. The concept of $\cal F$ invariance
330: ($c = 1$) manifests itself as
331: a new (non-Fermi liquid) fixed point in the theory. The problem
332: with $0 < c < 1$
333: lies in the domain of attraction
334: of the Fermi liquid line $c =0$ which is stable in the infrared.
335:
336:
337: Notice, however, that the $\cal F$ invariant fixed point $c = 1$
338: only exists if the mass $M$ in the theory
339: remains finite at zero $T$. This clearly shows that, in order for
340: $\cal F$ invariance
341: to represent an exact symmetry of the problem, $N_m$ must be infinite.
342: The time $\tau$ plays the role of an extra, non-trivial
343: dimension and this dramatically complicates the problem of plateau
344: transitions in the quantum Hall regime. The Coulomb interaction problem,
345: unlike the free electron theory, is given
346: as a $2+1$ dimensional field theory, thus invalidating any attempt
347: toward exact solutions of the experimentally observed critical
348: indices~\cite{experiments}.
349:
350: \vskip 1mm
351: \noindent{$\bullet$} {\em The qHe.~~~}
352: Next, we turn to the most interesting aspect of the theory, the
353: $\sigma_{xy}^0$ term ($\theta$ term), which is invisible in perturbative
354: expansions. However, we may proceed along the same lines as pointed out
355: in the previous Letter and separate, in the theory for $T=0$,
356: the {\em bulk} quantities from the {\em edge} quantities~\cite{largeN}.
357: Specifying to the Coulomb interaction problem ($c=1$) in two spatial dimensions,
358: we next make use of the principle
359: of $\cal F$ invariance and formulate an effective action for the edge. In the notation
360: introduced before we now have
361: \begin{eqnarray}
362: S_{eff} (q) & = & S_{bulk} (q) + 2\pi i k(\nu) C(q) ,
363: \nonumber \\
364: e^{-S_{bulk} (q)} & = & \int_{\partial V} D[Q_0 ] e^{-{\tilde S}_{\sigma}
365: ( t^{-1} Q_0 t) - S_F (t^{-1} Q_0 t)} .
366: \label{Sb}
367: \end{eqnarray}
368: Here ${\tilde S}_{\sigma}$ is the same as $S_\sigma$ with $\sigma_{xy}^0$ replaced
369: by its {\em unquantized} bulk piece $\theta(\nu)$. Recall that the
370: functional integral is performed with a fixed value $Q_0 = \Lambda$ at the edge.
371: It is important to notice that the interaction piece $S_F$ cannot be left out
372: since it affects, following Eq. (\ref{RGf}), the renormalization of the theory at $T=0$.
373:
374: The definition of $S_{eff} (t)$ is precisely the same as the
375: {\em background field} methodology adapted to the Coulomb interaction
376: problem~\cite{paperII,EuroLett}. The result is of the form
377:
378: \be
379: {S_{bulk} (q)} = F(\theta) + {\tilde S}_{\sigma} ' (q) + S_F ' (q) ,
380: \label{Sb2}
381: \ee
382: where the primes indicate that the parameters $\sigma_{xx}^0$, $\theta(\nu)$ and $z_0$ are
383: replaced by renormalized ones, $\sigma'_{xx} $, $\theta' $ and $z'$
384: respectivily, which are defined for system size $L$.
385:
386: This leads to the most important statement of this Letter which says
387: that, provided a $mass$ is generated for {\em bulk} excitations, the
388: renormalized theory $\sigma'_{xx} =\sigma_{xx} (L)$, $\theta' =\theta (L)$ and $z'=z (L)$
389: should vanish for large enough $L$,
390: i.e. the bulk of the system is insensitive to changes in the boundary
391: conditions except for corrections exponentially small in $L$.
392: Under these circumstances $S_{eff}[q]$ reduces to the
393: action of {\em massless chiral edge excitations}~\cite{paperIII,fracedge}.
394: The integer $k(\nu)$ equals
395: the number of edge modes and is now identified as the
396: {\em quantized Hall conductance}.
397:
398: These results describe the strong coupling
399: ``integer quantum Hall'' fixed points (Fig. \ref{mappedRG}) that were previously conjectured on
400: phenomenological grounds. From the weak coupling side, a detailed analysis of Eqs. (\ref{Sb}), (\ref{Sb2})
401: leads to the following expressions~\cite{EuroLett} for the renormalization group functions
402: $d\sigma_{xx}/d\ln\mu = \beta_\sigma (\sigma_{xx} , \theta)$,
403: $d\theta/d\ln\mu = \beta_\theta (\sigma_{xx} , \theta)$ and
404: $d\ln z/d\ln\mu = \gamma_z (\sigma_{xx} , \theta)$,
405:
406: \begin{eqnarray}
407: \beta_\sigma & = & \beta_\sigma^0 (\sigma_{xx} )
408: + D_0 \sigma_{xx}^2 e^{-2\pi\sigma_{xx}} \cos \theta
409: \nonumber \\
410: \beta_\theta & = & ~~~~~~~~~~~~~~D_0 \sigma_{xx}^2 e^{-2\pi\sigma_{xx}} \sin \theta
411: \nonumber \\
412: \gamma_z & = & \gamma_z^0 (\sigma_{xx} )
413: + D_0 \sigma_{xx} e^{-2\pi\sigma_{xx}} \cos \theta,
414: \label{RGi}
415: \end{eqnarray}
416: Here, $D_0$ is a positive constant determined by the instanton determinant and
417: $\beta_\sigma^0 $ and $\gamma_z^0 $ are the perturbative results
418: that recently have been extended to two-loop order ($A \approx 1.64$)~\cite{paperII,twoloop}
419: \be
420: \beta_\sigma^0 (\sigma_{xx} ) = \frac{2}{\pi} + \frac{4A}{\sigma_{xx}};~~
421: \gamma_z^0 (\sigma_{xx} ) = \frac{1}{\pi\sigma_{xx}} + \frac{18+\pi^2}{6\pi^2\sigma_{xx}^2} .
422: \label{2loop}
423: \ee
424: In summary, there is now fundamental support, both from the weak and strong coupling side,
425: for the scaling diagram of the integral quantum Hall effect~\cite{EuroLett}.
426: \vskip 1mm
427: %
428: \noindent{$\bullet $} {\em Finite T.~~~} At finite $T$ the infrared of the system is
429: controlled by the interaction terms $S_F$ and $S_U$. In this case one must go back
430: to the original theory (Eqs. (\ref{Seff})-(\ref{S0})) and obtain the transport parameters from linear
431: response in the field $a_\mu$~\cite{paperII}.
432: Specifying to the $a_0 = 0$ gauge as well as $\vec{\nabla} \cdot \vec{a} =
433: \vec{\nabla} \times \vec{a} = 0$ we can write
434: \begin{eqnarray}
435: S_{eff} (a_j ) = T\sum_{n>0} \int_x \omega_n \left[ \sigma'_{xx} \delta_{ij} +
436: \sigma'_{xy} \epsilon_{ij} \right] a_i (\omega_n ) a_j (-\omega_n )
437: \nonumber
438: \end{eqnarray}
439: where the expressions for $\sigma'_{ij}$ are known as the {\em Kubo formulae}~\cite{paperII}.
440: We stress that these expressions are
441: exactly the same as those obtained from the background field proceedure,
442: $\sigma'_{xx} =\sigma_{xx}(L)$
443: and $\sigma'_{xy} = k(\nu) + \theta (L)/2\pi$ (Eqs \ref{Sb2}, \ref{RGi}),
444: provided $S_{eff} (a_j ) $ is evaluated at $T=0 $ and with $Q=\Lambda$ at the
445: edge~\cite{paperI,paperII}
446:
447: The scaling results at finite $T$ generally depend on the
448: specific regime and/or microscopics of the disordered electron gas that one is
449: interested in. Here we consider the most interesting cases where $\theta (\nu) \sim \pm \pi $
450: and $\theta (\nu) \sim 0$ respectivily. The first case is realized when
451: the Fermi level passes through the center of the Landau
452: band where the electron gas is quantum critical and the transition takes place between
453: adjacent quantum Hall plateaus~\cite{freepart}.
454: Provided the bare parameter $\sigma_{xx}^0$
455: of the theory is close to the critical fixed point $\sigma_{xx}^* $ at $\sigma_{xy} =1/2$
456: (Fig. \ref{mappedRG})
457: the following universal scaling law is observed~\cite{freepart}
458:
459:
460: \begin{equation}
461: \sigma'_{xx} = \sigma_{xx} (X) ~~~ \sigma'_{xy} = k(\nu) + {\theta (X)}/{2\pi}
462: \label{ss}
463: \end{equation}
464: where $X = (zT)^{-\kappa } \Delta \nu$. Here $\Delta \nu$ is the filling fraction $\nu$ of the
465: Landau levels relative to the critical value $\nu^*$ which is half-integer.
466: The correlation (localization) length $\xi $ of the electron gas
467: diverges algebraically $\xi \propto |\Delta\nu |^{-1/y_\theta}$. The critical indices $\kappa$
468: and $y_\theta$ are a major objective of experimental research~\cite{experiments} and
469: the results have been discussed extensively and at many places~\cite{EuroLett,paperIII,twoloop,hopping}.
470:
471: Next we consider $\theta(\nu) \sim 0$ which is entirely different. This happens when
472: the Fermi energy is located at the tail end of the Landau bands corresponding to the
473: center of the quantum Hall plateau.
474: The bare parameter $\sigma_{xx}^0$ of the theory is now close to zero~\cite{paperIII}.
475: This means that the $T$ dependence is determined by the strong coupling
476: asymptotics of the renormalization ($\theta ,~\sigma_{xx} \rightarrow 0$). Notice
477: that the $\gamma_z$ function (Eqs. \ref{RGi}, \ref{2loop}) indicates that the singlet interaction term
478: $S_F$ eventually renders irrelevant as compared to
479: the Coulomb term $S_U$ (with $U^{-1} (q) =\Gamma |q|$) which, as we mentioned before,
480: is not affected by the quantum theory. One now expects $S_U$ to become the dominant
481: infrared regulator such that the scaling variable $X$ in Eq. (\ref{ss}) is now given by
482: $X = T\Gamma \xi$.
483:
484: This asymptotic limit of the theory
485: can be identified as the Effros - Shklovskii regime of
486: {\em variable range hopping} for which the following result is known
487: $\sigma'_{xx} = \sigma_{xx} (T\Gamma \xi ) = \exp \{ -2/ \sqrt{T\Gamma \xi } \}$~\cite{hopping}.
488: We therefore conclude that the dynamics of the electron gas is generally described
489: by distinctly different physical processes and controled by completely different fixed points
490: in the theory.
491:
492: The research was supported in part by FOM and INTAS
493: (Grant 99-1070).
494:
495:
496:
497: \begin{thebibliography}{99}
498:
499: \bibitem{EuroLett} A.M.M. Pruisken and M.A. Baranov, Europhys. Lett.
500: {\bf 31} 543 (1995) and references therein.
501:
502: \bibitem{paperI} A.M.M. Pruisken, M.A. Baranov and B. \v{S}kori\'{c},
503: Phys. Rev. {\bf B60}, 16807 (1999)
504:
505: \bibitem{paperII} M.A. Baranov, A.M.M. Pruisken and B. \v{S}kori\'{c},
506: Phys. Rev. {\bf B60}, 16821 (1999)
507:
508: \bibitem{paperIII} A.M.M. Pruisken, B. \v{S}kori\'{c} and M. Baranov,
509: Phys. Rev. {\bf B60}, 16838 (1999)
510:
511: \bibitem{freepart} A.M.M. Pruisken, Phys. Rev. Lett. {\bf 61}, 1297 (1988)
512:
513: \bibitem{fracedge} B. \v{S}kori\'{c} and A.M.M. Pruisken,
514: Nucl. Phys. B {\bf 559}, 637 (1999)
515:
516: \bibitem{Finkelstein} A.M. Finkelstein, Pis'ma Zh. Eksp. Teor. Fiz. {\bf 37},
517: 436 (1983)
518: [JETP Lett. {\bf 37}, 517 (1983)];
519: Zh. Eksp. Teor. Fiz. {\bf 86}, 367 (1984) [Sov. Phys. JETP {\bf 59},
520: 212 (1984)];
521: Physica B {\bf 197}, 636 (1994)
522:
523: \bibitem{italians} C. Castellani et al, Phys. {\bf B61}, 1297 (1988)
524:
525: \bibitem{cf} See e.g.
526: O. Heinonen, ed., {\it Composite Fermions} (World Scientific 1998)
527:
528: \bibitem{hopping} See e.g. D.G. Polyakov and B.I. Shklovskii,
529: Phys. Rev. Lett. {\bf 61}, 1294 (1988). Their idea of `one parameter scaling', however,
530: is in conflict with our results.
531:
532: \bibitem{NozPines} D. Pines and P. Nozieres, {\sl The Theory of Quantum
533: Liquids}, vol. I (W.A. Benjamin Inc. 1966)
534:
535: \bibitem{twoloop} M.A. Baranov, I.S. Burmistrov, and A.M.M. Pruisken,
536: cond-mat/0106446
537:
538: \bibitem{sigma} E. Brezin, S. Hikami and J. Zinn Justin,
539: Nucl. Phys. B {\bf 165}, 528 (1980)
540:
541: \bibitem{sigma1} A.M.M. Pruisken, Phys. Rev. {\bf B31}, 416 (1985)
542:
543: \bibitem{experiments}
544: H.P. Wei {\em et al.}, Phys. Rev. Lett. {\bf 61}, 1294 (1988);
545: R.T.F. van Schaijk {\em et al.}, Phys. Rev. Lett. {\bf 84},1567 (2000);
546: D. de Lang {\em et al.}, cond-mat/0106375;
547: A.M.M. Pruisken {\em et al.}, cond-mat/0109043
548:
549:
550:
551: \bibitem{largeN} A.M.M. Pruisken, M.A. Baranov, and M. Voropaev,
552: cond-mat/0101003
553:
554: \end{thebibliography}
555: \end{multicols}
556: \end{document}
557: