cond-mat0503651/aps.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%
3: %%
4: %%   QMC study for degenerate Hubbard model
5: %%
6: %%     A. Koga, N. Kawakami, T.M. Rice and M. Sigrist
7: %%
8: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
9: %\documentclass[a4paper,aps,twocolumn,showpacs]{revtex4}     
10: \documentclass[prb,twocolumn,showpacs]{revtex4}   % Physical Review B
11: %\documentclass[prb,preprint]{revtex4}   % Physical Review B             
12: 
13: 
14: \usepackage{graphicx}
15: 
16: \begin{document}
17: 
18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
19: \title
20: {
21: Spin, charge and orbital fluctuations in a multi-orbital Mott insulator
22: }
23: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
24: 
25: 
26: \author
27: {                                                       
28: Akihisa Koga and Norio Kawakami
29: }
30: 
31: 
32: 
33: 
34: \affiliation
35: {
36: Department of Applied Physics, Osaka University, 
37: Suita, Osaka 565-0871, Japan
38: }
39: 
40: \author
41: {                                                       
42: T.M. Rice and Manfred Sigrist
43: }
44: 
45: \affiliation
46: {
47: Theoretische Physik, 
48: ETH-H\"onggerberg, 8093 Z\"urich, Switzerland
49: }
50: \date{\today}
51: 
52: \begin{abstract}
53: The two-orbital degenerate Hubbard model with distinct hopping 
54: integrals is studied 
55: by combining dynamical mean-field theory with quantum Monte Carlo simulations.
56: The role of orbital fluctuations for  
57: the nature of the Mott transition is elucidated by examining the temperature dependence
58: of  spin, charge and 
59: orbital susceptibilities as well as the one-particle spectral function.
60: We also consider the effect of the hybridization between the two
61: orbitals, which is important particularly close to  the Mott 
62: transition points. The introduction of the hybridization induces 
63: orbital fluctuations,
64: resulting in the formation of a Kondo-like heavy-fermion behavior, similarly to
65: $f$ electron systems, but involving electrons in bands of comparable width. 
66: %%Some implications for the compounds $\rm (Ca, Sr)_2RuO_4$ and 
67: %%$\rm La_{n+1}Ni_nO_{3n+1}$ are also discussed.
68: \end{abstract}
69: 
70: \pacs{71.10.Fd, 71.30.+h}% 
71: 
72: \maketitle
73: 
74: 
75: 
76: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
77: \section{Introduction}\label{intro}
78: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
79: Strongly correlated electron systems with multi-orbital bands pose a
80: variety of intriguing problems.  One of the recently debated topics 
81: is the orbital-selective Mott transition (OSMT) in
82: highly correlated $d$-electron systems.
83: \cite{Anisimov,Fang,KogaLett,Sigrist,Liebsch,sces} 
84: It is a fundamental issue of multi-orbital systems whether 
85: Mott-transitions would take place in sequence or simultaneously for
86: all bands, if correlation would gradually be turned on. 
87: There are however also specific materials which have been discussed in this
88: context such as the calcium-doped single layer strontium ruthenate
89: $\rm Ca_{2-x}Sr_{x}RuO_4$\cite{Nakatsuji} and 
90: the ternary nickel oxide $\rm La_{n+1}Ni_nO_{3n+1}$,\cite{LaNiO,Kobayashi}
91: where the chemical substitution (or the change in the temperature) 
92: may trigger the OSMT in the $t_{2g}$ ($e_g$) orbitals
93: in the former (latter) case.
94: 
95: The extensive studies on the Mott transition in the multi-orbital systems
96:  clarified that  the competition between 
97: the intra- and inter-orbital interactions as well as the 
98: Hund coupling plays a key role to determine  the nature 
99: of the Mott transition.\cite{KogaLett}  It was found that under
100: special conditions in a two-band system the Mott-transitions
101: may merge to a single one, but would split for a generic form of the model. 
102: In particular, the presence of Hund coupling seems to be essential
103: to observe distinct transitions.  These conclusions 
104: were drawn from the analysis of the quasi-particle weights computed 
105: at zero temperature. In order to characterize the transitions the behavior of
106: the spin, charge and orbital fluctuations provides additional valuable
107: information. A systematic study of the temperature dependence of
108: certain susceptibilities will give us the necessary insight to analyze
109: in particular the electronic degrees of freedom which are localized 
110: through the Mott transition.
111: 
112: The above discussions on the Mott transition are restricted so far 
113: to systems for which the bands  do not 
114: hybridize, but are coupled to each other
115: only through electron-electron interactions.
116: However, the  hybridization between the bands may be important in some 
117: compounds.\cite{Kusunose}  In particular, this effect could give rise to a 
118: qualitative change in the phase diagram,  
119: when there occurs the OSMT, for which  
120: the intermediate phase appears with one orbital localized and 
121: the other itinerant. One thus naively wonders whether
122: Kondo-like  heavy fermion states would  be induced  by the hybridization 
123: between the orbitals. 
124: In fact, certain observed features can possibly be attributed to Kondo like behavior 
125: in the compound 
126: $\rm Ca_{2-x}Sr_xRuO_4$ $(0.2<x<0.5)$,\cite{Nakatsuji}
127: where the hybridization between orbitals
128: is induced by the tilting of RuO$_6$ octahedra.\cite{tilting} 
129: It is surprising that this behavior emerges from
130: electrons which originate from bands of comparable width.
131: These interesting observations naturally motivate us to
132: explore the effect of hybridization in more detail. 
133: 
134: In this paper, we study a two-orbital Hubbard model
135: with the distinct hopping integrals by combining dynamical mean field theory 
136: (DMFT)\cite{Metzner,Muller,Georges,Pruschke} 
137: with quantum Monte Carlo (QMC) simulations.\cite{Hirsch,Sakai}
138: We examine the spin, charge and orbital fluctuations which give insight into
139: the electronic properties in the regime of the OSMT.
140: We further consider the effect of hybridization, which may be 
141: important in real materials, and show that heavy-fermion-like behavior
142: emerges upon introduction of the hybridization.
143: The paper is organized as follows.
144: In \S\ref{sec2}, we introduce the model Hamiltonian for the two-orbital 
145: system and briefly explain the framework of DMFT. 
146: We discuss how the spin and orbital fluctuations affect 
147: the metal-insulator transition in \S \ref{sec3}.
148: A brief summary is given in the last section.
149: 
150: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
151: \section{Model and Method}\label{sec2}
152: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
153: 
154: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
155: %%\subsection{Two-orbital Hubbard model}
156: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
157: 
158: %%To discuss the nature of metal-insulator transitions in
159: %%the multi-orbital system,
160: We consider the following two-orbital Hubbard Hamiltonian,
161: %%%%%%%%%%%%%%%%%%%%%%%%%%
162: \begin{eqnarray}
163: H&=&\sum_{\stackrel{<i,j>}{\alpha\beta\sigma}}
164: t_{ij}^{(\alpha\beta)} c_{i\alpha\sigma}^\dag c_{j\beta\sigma}
165: %\nonumber\\
166: %&+&
167: +U\sum_{i\alpha}n_{i\alpha\uparrow}n_{i\alpha\downarrow}\nonumber\\
168: &+&\left(U'-J\right)\sum_{i\sigma}n_{i1\sigma}n_{i2\sigma}
169: +U'\sum_{i\sigma}n_{i1\sigma}n_{i2\bar{\sigma}}\nonumber\\
170: &-&J\sum_{i}\left[ c_{i1\uparrow}^\dag c_{i1\downarrow}
171: c_{i2\downarrow}^\dag c_{i2\uparrow}+c_{i1\downarrow}^\dag c_{i1\uparrow}
172: c_{i2\uparrow}^\dag c_{i2\downarrow}\right] \nonumber\\
173: &-&J\sum_{i}\left[ c_{i1\uparrow}^\dag c_{i1\downarrow}^\dag
174: c_{i2\uparrow} c_{i2\downarrow}+ c_{i2\uparrow}^\dag c_{i2\downarrow}^\dag
175: c_{i1\uparrow} c_{i1\downarrow} \right]\label{eq:model}
176: \label{Hamilt}
177: \end{eqnarray}
178: %%%%%%%%%%%%%%%%%%%%%%%%
179: where $c_{i\alpha\sigma}^\dag (c_{i\alpha\sigma})$ 
180: creates (annihilates) an electron 
181: with  spin $\sigma(=\uparrow, \downarrow)$ and orbital
182: index $\alpha(=1, 2)$ at the $i$th site and 
183: $n_{i\alpha\sigma}=c_{i\alpha\sigma}^\dag c_{i\alpha\sigma}$. 
184: $U$ ($U'$) represents the intraband (interband) Coulomb interaction and
185: $J$ the Hund coupling.
186: For electron hopping, we introduce
187: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
188: \begin{eqnarray}
189: t_{ij}^{(\alpha\beta)}=t_{ij}^{(\alpha)}\delta_{\alpha\beta}+V\delta_{ij},
190: %%%%-\mu\delta_{\alpha\beta}\delta_{ij},
191: \end{eqnarray}
192: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
193: with  the orbital-dependent nearest-neighbor hopping $t_{ij}^{(\alpha)}$ 
194: and the hybridization $V$ between two orbitals.
195: %%%$\mu$ the chemical potential.
196: By this generalized model, we can study several different models 
197: in the same  framework. For $V=0$, 
198: the system is reduced to the multi-orbital Hubbard model 
199: with the same $(t_{ij}^{(\alpha)}=t_{ij})$
200: %\cite{Koga,Momoi,OnoED,multi} 
201: or distinct orbitals.\cite{KogaLett,Liebsch}
202: On the other hand, for $t_{ij}^{(2)}=0$, the system is reduced to 
203: a correlated electron system coupled to localized electrons, 
204: such as the periodic Anderson model ($J=0$) for 
205: heavy-fermion systems\cite{Coleman,Rice,Yamada,Kuramoto,Kim}
206:  or the double exchange model
207: ($J>0$) for some transition metal oxides.
208: \cite{Zener,Anderson,Kubo,Furukawa}
209: For general choices of the parameters, we 
210: expect a variety of 
211: characteristic  properties inherent in these limiting models to appear naturally.
212: %%Such a system may appear with the spontaneous reduction 
213: %%of effective dimensions, 
214: %%{\it eg}. the cubic vanadates $\rm YVO_3$ and $\rm LaVO_3$.
215: 
216: %%%%%%%%%%%%%%%
217: %%\subsection{Dynamical mean field theory}
218: %%%%%%%%%%%%%%%
219: 
220: To investigate the above degenerate Hubbard model,
221: we make use of DMFT, \cite{Metzner,Muller,Georges,Pruschke}
222: which has successfully been applied to various electron systems such as 
223: the single band Hubbard model, 
224: \cite{Caffarel,2site,single1,OSakai,single2,single3,single4,BullaNRG}
225: the multi-orbital Hubbard model, 
226: \cite{2band1,2band2,Koga,Momoi,OnoED,Sakai,multi,sces,KogaLett,Liebsch}
227: the periodic Anderson model. \cite{PAM,Mutou,Saso,Sato,Ohashi,Medici}
228: In DMFT, the lattice model is mapped to
229:  an effective impurity  model, 
230: where local electron correlations are taken into account precisely. 
231: The lattice Green function is then obtained via self-consistent
232: conditions imposed on the impurity problem.
233: %The treatment is exact in  $d\rightarrow\infty$ dimensions, and
234: %even in three dimensions,  DMFT has successfully explained 
235: %interesting physics such as the Mott metal-insulator transition.
236: 
237: In  DMFT for the multi-orbital model,
238: the Green function in the lattice system is given as,
239: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
240: \begin{eqnarray}
241: {\bf G}\left(k, z\right)^{-1}={\bf G}_0\left(k, z\right)^{-1}
242: -{\bf \Sigma}\left(z \right),
243: \end{eqnarray}
244: %%%%%%%%%%%%%%%%%%%%%%%%%
245: with
246: %%%%
247: \begin{equation}
248: {\bf G}_0\left( k, z\right)^{-1}=\left(
249: \begin{array}{cc}                 
250: z+\mu-\epsilon_1( k) & -V\\
251: -V & z+\mu-\epsilon_2( k)
252: \end{array}
253: \right),
254: \end{equation}
255: and
256: \begin{equation}
257: {\bf \Sigma}\left(z\right)=\left(
258: \begin{array}{cc}
259: \Sigma_{11}(z) & \Sigma_{12}(z) \\
260: \Sigma_{21}(z) & \Sigma_{22}(z) 
261: \end{array}
262: \right),
263: \end{equation}
264: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
265: where $\mu$ is the chemical potential, and $\epsilon_\alpha (k)$
266: is the bare dispersion relation 
267: for the $\alpha$-th orbital. 
268: %%Here, we have used the fact that the self-energy is independent 
269: %%of the momentum in the $d\rightarrow\infty$ dimensions.
270: In terms of the density of states $\rho (x)$ rescaled by the 
271: band width $D_\alpha$,
272: the local Green function is expressed as,
273: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
274: \begin{eqnarray}
275: G_{11}(z)&=&\int dx \frac{\rho(x)}{\xi_1\left(z,x\right)-
276: \frac{\displaystyle v(z)^2}{\displaystyle \xi_2\left(z, x\right)}},
277: \nonumber\\
278: G_{12}(z)&=&\int dx \frac{v(z)}
279: {\xi_1\left(z, x\right)\xi_2\left(z, x\right)-v(z)^2},
280: \nonumber\\
281: G_{22}(z)&=&\int dx \frac{\rho(x)}{\xi_2\left(z, x\right)-
282: \frac{\displaystyle v(z)^2}{\displaystyle \xi_1\left(z, x\right)}},
283: \end{eqnarray}
284: where
285: \begin{eqnarray}
286:  \xi_1\left(z, x\right)&=&z+\mu-\Sigma_{11}-D_1 x,\nonumber\\
287:  \xi_2\left(z, x\right)&=&z+\mu-\Sigma_{22}-D_2 x,\nonumber\\
288:  v\left(z\right)&=&V+\Sigma_{12}\left(z\right).
289: \end{eqnarray}
290: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
291: In the following, we use the semicircular density of states 
292: $\rho (x)=2/\pi \sqrt{1-x^2}$.
293: %%which corresponds to an infinite-coordination Bethe lattice.
294: 
295: %%To take into account local electron correlations precisely, 
296: %%we solve the impurity problem. 
297: %%The effective bath so-called "cavity Green function" ${\bf g}_0 (z)$ is 
298: %%given by the local Green function and the self-energy as,
299: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
300: %%\begin{eqnarray}
301: %%{\bf g}_0 \left(z\right)^{-1}&=&{\bf G} \left(z\right)^{-1}
302: %%+{\bf \Sigma} \left(z\right).
303: %%\end{eqnarray}
304: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
305: 
306: There are  various numerical methods 
307: to solve the effective impurity problem.
308: Note that the explicit model Hamiltonian for the impurity system 
309: is not obtained straightforwardly in our case,
310: since the lattice Green function has a frequency-dependent term in the 
311: non-diagonal element when the system has the hybridization $V$ and 
312: the finite band width in both orbitals. 
313: Therefore, it is not necessarily most efficient  to 
314: apply the exact diagonalization\cite{Caffarel} or
315: the two-site DMFT\cite{2site} methods
316: as impurity solvers, because these methods require the knowledge 
317: of the explicit form of the Hamiltonian. 
318: Furthermore, self-consistent perturbation theories such as
319: the iterative perturbation method and the non-crossing approximation
320: are not appropriate to discuss  orbital fluctuations
321: in the vicinity of the critical point.
322: In the present study, we make use of QMC
323: to treat the impurity model at finite temperatures.\cite{Hirsch}
324: In this connection, we note here that the Hund coupling plays
325:  a key role in controlling the nature of the Mott transition
326: in the multi-orbital system.\cite{sces}
327: Therefore, it is important to carefully analyze the effect of the Hund coupling 
328: in the framework of QMC.
329: To this end, we use the algorithm proposed by Sakai et al.,\cite{Sakai} 
330: where  the Hund coupling is represented in terms of
331: discrete auxiliary fields.
332: When we solve the effective impurity model by means of QMC method,
333: we use the Trotter time slices $\Delta \tau = (TL)^{-1} \le 1/6$,
334: where $T$ is the temperature and $L$ is the Trotter number.
335: 
336: In the following, we fix the band widths as $(D_1, D_2)=(1.0, 2.0)$ 
337: and the chemical potential as $\mu=-U/2-U'+J/4$ 
338: to discuss the metal-insulator transitions at half-filling.
339: 
340: 
341: 
342: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
343: \section{Results}\label{sec3}
344: \subsection{Non-hybridizing bands}
345: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
346: 
347: Before presenting the results computed at finite temperatures,
348: we briefly summarize the nature of the zero-temperature 
349: phase diagram for $V=0$ obtained by DMFT 
350: together with the exact diagonalization,\cite{KogaLett}  which 
351: is shown in  Fig. \ref{fig:zero-phase}.
352: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
353: \begin{figure}[htb]
354: \begin{center}
355: \includegraphics[width=7cm]{phase-V=0.eps}
356: \end{center}       
357: %\vskip -35mm
358: \caption{
359: Phase diagram for the two-orbital Hubbard model
360: with  $D_1=1$ and $D_2=2$. Note that the condition of rotational
361: symmetry,
362: $U=U'+2J$, is imposed (only the region of $U\ge U'$ is relevant).
363:  In the phase (i) (phase (ii)), 
364: both bands are  metallic (insulating), whereas
365: in the phase (iii)  
366: the metallic state coexists with the Mott insulating state.
367: Two lines along $U=U'$ with $J=0$  and
368: $U'/U=3/4$ with $J/U=1/8$ are shown, for which  
369: thermodynamic properties at finite temperatures are 
370: examined  in the text.
371: }
372: \label{fig:zero-phase}
373: \end{figure}
374: %%%%%%%%%%%%%%%%%%%%%%%%%%
375: There are three distinct phases depending on the strength of
376: the interactions.
377: It is seen that the metallic phase (i) remains stable up to  
378: large Coulomb interaction $U$ along the line
379:  $U \sim U'$ (small $J$), where
380: the Mott transitions merge to a single transition. 
381: Away from the symmetric limit, i.e.  $ U > U' $ 
382: with $ 2J = U - U' $,
383:  we find two separate Mott transitions in general.
384: In between the intermediate metallic phase (iii) appears
385: with one band localized and the other itinerant. 
386: 
387: We now analyze the temperature dependence of the charge, spin and 
388: orbital fluctuations 
389:  by combining DMFT with QMC simulations. We still restrict here
390: to the case of non-hybridized bands $(V=0)$.
391: Two typical sets of the parameters are considered, which satisfy
392: the conditions $(U'/U, J/U)=(3/4, 1/8)$ and $(1,0)$.  
393: As seen from Fig. \ref{fig:zero-phase},
394: the Mott transitions occur at two different critical points
395: $U_{c1}\sim 3$ and $U_{c2}\sim 4$ in the former case, while 
396: in the latter case they are merged to a single Mott transition 
397: at the critical point $U_c\sim 7$ for zero temperature.
398: The charge (c), spin (s) and orbital (o) susceptibilities are defined as 
399: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400: \begin{eqnarray}
401: \chi_\gamma &=& \int_0^\beta 
402: d\tau \chi_\gamma(\tau),
403: \end{eqnarray}
404: %%%%%%%%%%%%%%%%%
405: with $\gamma=c, s, o$, and 
406: %%%%%%%%%%%%
407: \begin{eqnarray}
408: \chi_c(\tau-\tau')&=& \langle
409: T|\left[n(\tau)-2\right]\left[n(\tau')-2\right]\rangle,\nonumber\\
410: \chi_s(\tau-\tau')&=& \langle
411: T|\left[n_\uparrow(\tau)-n_\downarrow(\tau)\right]
412: \left[n_\uparrow(\tau')-n_\downarrow(\tau')\right]\rangle,\nonumber\\
413: \chi_o(\tau-\tau')&=& \langle
414: T|\left[n_1(\tau)-n_2(\tau)\right]
415: \left[n_1(\tau')-n_2(\tau')\right]\rangle,
416: \end{eqnarray}
417: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
418: where $T$ is the time-ordering operator, 
419: $n(\tau)=\sum_{\alpha\sigma} n_{\alpha\sigma}(\tau)$, 
420: $n_\alpha(\tau)=\sum_{\sigma} n_{\alpha\sigma}(\tau)$,
421: $n_\sigma(\tau)=\sum_{\alpha} n_{\alpha\sigma}(\tau)$, and
422: $\tau$ is an imaginary time.
423: 
424: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
425: \begin{figure}[htb]
426: \begin{center}
427: \includegraphics[width=7cm]{chi-o+.eps}
428: \end{center}       
429: %\vskip -35mm
430: \caption{
431: Orbital susceptibility as a function of the temperature $T$ for $V=0$. 
432: Open (solid) symbols represent the results in the case $U=U'$ and $J=0$ 
433: ($U'/U=3/4$ and $J/U=1/8$) and dashed lines those for the 
434: non-interacting case.
435: }
436: \label{fig:chi-o}
437: \end{figure}
438: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
439: We first turn to the orbital fluctuations. The temperature-dependent 
440: orbital susceptibility is shown in Fig. \ref{fig:chi-o}.
441: In the non-interacting system,
442: the orbital susceptibility increases with decreasing temperature,
443: %and shows Pauli-paramagnetic behavior at low temperatures.
444: and reaches a constant value at zero temperature.
445: If we now turn on the interactions (fixing  
446: the ratios $U'/U=3/4$ and $J/U=1/8$), the orbital susceptibility is suppressed
447: at low temperatures.  
448: This implies that electrons tend to localized in each band 
449: independently such that onsite fluctuations are unfavorable.  Eventually 
450: for $U \ge U_{c1} \sim 3$, one of the orbitals is entirely localized, 
451: so that orbital fluctuations are suppressed completely, giving 
452: $\chi_o=0$ at $T=0$.
453: 
454: On the other hand, very different behavior can be seen 
455: along the line $U'=U$ in Fig. \ref{fig:zero-phase}.
456: In this case, the
457: orbital susceptibility is increased with growing interactions
458: even at low temperatures. 
459: Interpreting this result in the context of the  phase diagram in Fig. \ref{fig:zero-phase},
460: we can say that the enhanced orbital fluctuations are 
461: relevant for stabilizing the metallic phase in the
462: strong correlation regime.
463: While such behavior is naturally expected
464: for models with
465: two equivalent orbitals, it appears even in systems with
466: nonequivalent bands.\cite{Koga}
467: %We wish to note again that this non-trivial behavior appears 
468: %even in the system with distinct hopping integrals.
469: 
470: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
471: \begin{figure}[htb]
472: \begin{center}
473: \includegraphics[width=8cm]{chi-c+.eps}
474: \end{center}       
475: %\vskip -35mm
476: \caption{
477: Charge susceptibility as a function of the temperature $T$ for $V=0$.
478: Dashed lines represent the results for the non-interacting case.}
479: \label{fig:chi-c}
480: \end{figure}
481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
482: To examine whether the system
483: shows metallic or insulating properties at finite temperatures,
484: we calculate the charge susceptibility (compressibility).
485: The obtained results are shown in Fig. \ref{fig:chi-c}.
486: In the case $U'/U=3/4$ and $J/U=1/8$,  the system with $U=3$
487: is located near the critical point between the metallic phase 
488: (i) and the intermediate phase (iii).
489: With decreasing temperature the 
490: charge susceptibility  decreases  below $T \sim 1$.
491: The appearance of a pseudogap  feature in an intermediate
492: temperature range gives rise to a depletion of density of
493: states at the Fermi energy for both bands.
494: Upon further lowering of the temperature
495: the charge susceptibility converges to a finite 
496: value, since the system still remains in a 
497: metallic phase, at least for one of the two orbitals.
498: For $U=4$, which corresponds to the boundary between 
499: the phases  (ii) and (iii), the charge susceptibility at low 
500: temperatures is almost zero, suggesting that the system has become
501: completely insulating corresponding to phase (ii). 
502: In contrast for $U'=U$ we observe in an intermediate range of
503: $U $ that with lowering temperature a decrease of the charge susceptibility
504: is followed by an eventual increase at lowest temperatures (Fig. \ref{fig:chi-c}).
505: Comparing this with Fig. \ref{fig:chi-o}, we see that the enhanced orbital fluctuations 
506: indeed have a tendency to stabilize  the metallic state.
507: 
508: 
509: %%Now that we have confirmed that charge and orbital susceptibilities
510: %%exhibit characteristic properties consistent with the zero-temperature
511: %%phase diagram, 
512: We now  move to the spin susceptibility. 
513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
514: \begin{figure}[htb]
515: \begin{center}
516: \includegraphics[width=8cm]{chit-s+.eps}
517: \end{center}       
518: %\vskip -35mm
519: \caption{
520: The effective Curie constant
521:  $\chi_s T$ as a function of the temperatures $T$ 
522: for $U'/U=3/4$ and 
523: $J/U=1/8$. Inset shows the results in the case $U'=U$ and $J=0$.
524: Dashed lines represent the results for the non-interacting case.
525: }
526: \label{fig:chit-s}
527: \end{figure}
528: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
529: In Fig. \ref{fig:chit-s},  we plot 
530: the effective Curie constant  $\chi_s T$  as a function 
531: of the temperature.
532: We first look at  the case of $U'/U=3/4$ and $J/U=1/8$.
533: At high temperatures, all the spin configurations are equally 
534: populated, so that the effective Curie constant takes the value
535: $1/2$ for each orbital in our units, yielding $\chi_s T\sim 1$.
536: When electron correlations are weak
537: ($U=1$), the system is still in the metallic phase,
538: so that the Pauli paramagnetic behavior with a constant 
539:  $\chi_s$ emerges, leading to $\chi_s T \rightarrow 0$  
540: as $T \rightarrow 0$. It is seen that the increase of the interactions
541: enhances the spin susceptibility at low temperatures, as a result of
542: the progressive trend to localize the electrons. 
543: %%%%%%%%%%%%%%%%%%%%%%%%%
544: %We remind here
545: %that the effective Curie constant takes $\chi_sT=2$,
546: %if free spins are developed independently in each orbital.
547: %%%%%%%%%%%%%%%%%%%%%%%%%
548: The effective Curie constant is $\chi_sT=2$ when a free spin is realized
549: in each orbital.
550: It is seen that the Curie constant increases beyond the value of 
551: 2 with the increase of the interactions ($U=3, 4$).
552: This means that ferromagnetic correlations due to 
553: the Hund coupling appear here. 
554: 
555: When $U'=U$ (inset of Fig. \ref{fig:chit-s}), both  spin
556: and orbital fluctuations are enhanced in the 
557: presence of the interactions. 
558: Accordingly, both  spin and orbital susceptibilities
559: increase at low temperatures, forming  heavy-fermion states
560: as far as the system stays in the metallic phase 
561: (see also Fig. \ref{fig:chi-o}).  Note that for $U=6$, at which 
562: the system is  close to the  Mott
563: transition point, the spin susceptibility is enhanced with the effective
564: Curie constant $\chi_sT \sim 4/3$ down to very low temperatures,
565: as seen in the inset of Fig. \ref{fig:chit-s}.
566: The value of 4/3 immediately follows if one takes into account 
567: two additional  configurations of doubly-occupied orbital besides
568:  four magnetic configurations, which are all degenerate
569:  at the metal-insulator 
570: transition point. Although not clearly observed
571: in the temperature range shown, $\chi_sT$ should vanish at
572: zero temperature for $U=U'=6$, since the system is still in 
573: the metallic phase, as seen from Fig. \ref{fig:zero-phase}.
574: 
575: 
576: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
577: \begin{figure}
578: \begin{center}
579: \includegraphics[width=8cm]{dos-V=0.eps}
580: \end{center}       
581: %\vskip -35mm
582: \caption{Density of states for the degenerate Hubbard model $(D_1,
583:  D_2)=(1.0, 2.0)$. The data are for the temperatures $T=2, 1,
584:  1/2$ and $1/6$ from the top to the bottom.
585: }
586: \label{fig:dos0}
587: \end{figure}
588: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
589: 
590: To see the above characteristic properties more clearly,
591:  we show 
592: the density of states for each orbital in Fig. \ref{fig:dos0},
593: which is computed by the maximum entropy method.\cite{MEM1,MEM2,MEM3}
594: When the interactions increase along the line  $U'/U=3/4$ and $J/U=1/8$,
595:  the OSMT should occur. Such tendency indeed 
596: appears at low temperatures in Fig. \ref{fig:dos0}(a).
597: Although both orbitals stay in metallic states down to
598: low temperatures ($T=1/6$) for $U=1$,  the OSMT
599: seems to occur for $U=2$; namely one of the bands develops the Mott Hubbard
600: gap, while the other band still remains metallic.
601: At a first glance, this result is slightly different from
602:  the ground-state phase diagram 
603: shown in Fig. \ref{fig:zero-phase}, where the system is in 
604: the phase (i) even at  $U=2$.  
605: However, this deviation is naturally understood
606: if we take into account the fact that for $U=2$, the 
607: narrower band is already in a highly correlated  
608:  metallic state, so that the sharp quasi-particle peak immediately 
609: disappears as the temperature increases beyond 
610: the small characteristic energy scale. This explains the behavior 
611: observed in the density of states at $T=1/6$.  For $U=3$, both 
612: bands are insulating at $T=1/6$ (the system
613: is almost on the boundary between the phases (ii) and (iii) at 
614: zero temperature).
615: 
616: In the case $U'=U$, as expected we encounter the qualitatively different 
617: behavior shown in Fig. \ref{fig:dos0}.
618:  In this case, 
619: both bands gradually develop quasi-particle peaks
620:  as the interactions increase, and 
621:  they still remain metallic even at $U=U'=3$.
622:   As mentioned above, all these features which are in contrast 
623: to  the situation for $U' \neq U$, 
624: are caused by the special symmetry for  $U=U'$, which gives rise to
625: equally enhanced spin and orbital fluctuations. 
626: 
627: 
628: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
629: \subsection{Hybridization between distinct orbitals}
630: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
631: 
632: We have so far treated the degenerate Hubbard model, in which 
633: two types of orbitals do not mix with each other.
634: In our treatment with DMFT, the Mott insulating phase (ii) as well as 
635: the intermediate phase (iii) may be unstable 
636: against certain perturbations. There may be several possible 
637: mechanisms that stabilize such insulating phases.
638: One of the mechanisms,
639: which may play an important role in real materials,
640: is the hybridization between
641: the two distinct orbitals. We address the effect in this section.
642: 
643: This hybridization effect is relevant in some real materials.
644: For instance, 
645: in the compound $\rm Ca_{2-x}Sr_x Ru O_4$, \cite{Nakatsuji}
646: the hybridization between $\{\alpha, \beta\}$ and $\gamma$ orbitals 
647: is induced by the tilting of RuO$_6$ octahedra in the 
648: region of  $\rm Ca$-doping $0.2<x<0.5$,\cite{tilting}. This leads to 
649: Kondo-lattice like effective model and 
650: may be connected
651: with the reported heavy fermion behavior, 
652: \cite{Nakatsuji} similar to some $f$-electron systems.
653: This interesting aspect motivates us to study  the mixing effect
654: between the localized and itinerant electrons in the 
655: intermediate phase (iii).  Moreover
656: the compound $\rm La_{n+1}Ni_nO_{3n+1}$\cite{LaNiO} possesses
657: hybridization between $d_{3z^2-r^2}$ and $d_{x^2-y^2}$ orbitals
658: in the $e_g$ subshell. The OSMT may lead to the metallic but the 
659: less-conducting state is realized
660: below the critical temperature $T_c=550K$.\cite{Kobayashi}
661: Consequently we would like also to explore how the hybridization 
662: of different-type $d$-bands affects 
663: electronic properties especially around
664: the OSMT.
665: 
666: We study the general case with
667: $U'\neq U$ and $J\neq 0$ in the presence of the hybridization
668: $V$. In Fig. \ref{fig:dos-V}, the 
669:  density of states calculated by the maximum entropy
670: method is shown for different choices of  $V$.
671: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
672: \begin{figure}
673: \begin{center}
674: \includegraphics[width=8cm]{dos-V.eps}
675: \end{center}       
676: %\vskip -35mm
677: \caption{Solid (dashed) lines represent the density of states 
678: for the orbital $\alpha=1$ ($\alpha=2$)
679: when $(D_1, D_2)=(1.0, 2.0)$ at $T=1/6$
680: with  the  fixed  parameters of $U'/U=3/4$ and $J/U=1/8$.
681: The data are plotted for $V=0.0, 0.25, 0.5, 0.75, 1.0, 1.25$ 
682: and $1.5$ from top to bottom.
683: }
684: \label{fig:dos-V}
685: \end{figure}
686: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
687: We start with the weak coupling case, $U=1$, where
688: the metallic states are realized in both orbitals at $V=0$.
689: Although the introduction of small $V$ does not alter
690: the nature of the ground state, 
691: further increase of $V$ splits
692: the density of states ($V=1.5$), signaling the formation of the 
693: band insulator: namely all kinds of elementary 
694: excitations possess the gap.
695: In contrast, we encounter different behavior when electron
696: interactions are increased up to $U=2$ and 3. In these parameters,
697:  the system at $V=0$ shows the intermediate
698: or Mott-insulating properties at $T=1/6$. It is seen that
699: the density of states around the Fermi level increases
700: as $V$ increases.  For $U=2$, the intermediate state
701: is first changed to a metallic state, where the quasi-particle 
702: peaks appear in both orbitals ($V=0.75,1.0$). 
703: For fairly large $V$, both bands fall into the 
704: renormalized band  insulator ($V=1.5$).
705: Similarly, for $U=3$, the hybridization first drives the Mott-insulating
706: state to an intermediate one, as is clearly seen at $V=0.75$, which 
707: is followed by two successive transitions as is the case for $U=2$.
708: 
709: 
710: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
711: \begin{figure}
712: \begin{center}
713: \includegraphics[width=8cm]{chiall-v.eps}
714: \end{center}       
715: %\vskip -35mm
716: \caption{Charge, spin and orbital susceptibilities 
717: as a function of the hybridization $V$ at the temperature $T=1/6$.
718: }
719: \label{fig:chiall-v}
720: \end{figure}
721: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
722: 
723: The above characteristic properties also emerge in the
724: charge, spin and orbital susceptibilities
725: at low temperature, as shown in Fig. \ref{fig:chiall-v}.
726: For weak interactions ($U=1$), the charge 
727: susceptibility $\chi_c$ monotonically decreases with the increase of $V$.
728: When electron correlations become strong, 
729: the non-monotonic behavior appears in $\chi_c$:
730: the charge fluctuations, which are suppressed at $V=0$,
731: are somewhat recovered by the hybridization, 
732: which leads to metallic behavior. 
733: For large $V$, $\chi_c$ is again  suppressed
734: since the system turns into a band insulator.
735: We can see that the orbital susceptibility exhibits 
736: non-monotonic behavior similar to the charge susceptibility,
737:  the origin of which is essentially the
738: same as in $\chi_c$; the orbital fluctuations suppressed 
739: at $V=0$ are recovered by $V$, and then 
740: the formation of  the band insulator causes 
741: the gradual decrease of $\chi_o$.
742: In contrast,  the spin susceptibility  monotonically 
743: decreases with the increase of  $V$ irrespective of 
744: the strength of the interactions.  As discussed for
745: $V=0$, the effective spin is enhanced by ferromagnetic 
746: fluctuations due to the Hund coupling in the insulating 
747: and intermediate phases. Upon introducing the hybridization in 
748: these phases, the ferromagnetic fluctuations are 
749: suppressed, leading to the monotonic decrease of the 
750: effective Curie constant.
751: 
752:  From the above observations, we can say that the introduction of 
753: appropriate hybridization induces 
754: heavy-fermion metallic behavior.
755: In fact, this tendency can be observed more clearly in an
756: extreme choice of the bandwidths, $(D_1, D_2)=(1.0, 10.0)$,
757:  shown in Fig. \ref{fig:ex}. 
758: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
759: \begin{figure}
760: \begin{center}
761: \includegraphics[width=8cm]{ex.eps}
762: \end{center}      
763: %\vskip -35mm
764: \caption{(a) Effective Curie constant as a function of the temperature 
765: and (b) density of states in the narrower band $(\alpha=1)$  
766: at $T=1/4$  for
767: an extreme choice of the bandwidths, $(D_1,D_2)=(1.0, 10.0)$. 
768: The density of states for the wider band is not shown here.
769: The other parameters are $U=4.0, U'=3.0$ and $J=0.5$.
770: }
771: \label{fig:ex}
772: \end{figure}
773: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
774: %%when electrons in wider band are hardly renormalized in the 
775: %%intermediate phase $(U_{c1}<U<U_{c2})$.
776: At $V=0.0$, the system is in the intermediate phase, so that
777: the completely localized states [Fig. \ref{fig:ex} (b)] appear
778:  in the narrower band in the background of 
779: the nearly free bands. This double structure in the system gives rise to
780: two peaks in the temperature-dependent effective Curie constant,
781: as shown in Fig. \ref{fig:ex} (a).
782: Since the completely localized state plays a role of the
783: $f$-state in the Anderson lattice model,
784: \cite{Kusunose} a ''heavy-fermion'' peak appears
785: at the Fermi energy in the 
786: presence of $V$, which is essentially the 
787: same as  that observed in Fig. \ref{fig:dos-V}.
788: 
789: %
790: %%It is found that double structures characterizing the OSMT
791: %%still appear at the finite temperatures when the hybridization
792: %%is introduced, which may explain the existence of the
793: %% OSMT at $T_c\sim 550K$
794: %%in the compound $\rm La_{n+1}Ni_nO_{3n+1}$\cite{Kobayashi}
795: %%with the hybridization in $e_g$ orbitals.
796: %Note that this type of magnetic behavior is not changed by a 
797: %%small hybridization, implying that the orbital-selective 
798: %%Mott transition appears even at finite temperatures.
799: %%Therefore, we think that the effect of hybridization on the 
800: %%correlated metal may be related to heavy-fermion behavior observed 
801: %%experimentally in $\rm Ca_{1-x}Sr_{x}RuO_4$ ($0.2<x<0.5$).\cite{Nakatsuji}
802: 
803: Finally, some comments are in order on the 
804: phase diagram at zero temperature.
805: In our approach, it is not easy to deal with the system 
806: at very low temperatures, since  QMC simulations  suffer
807:  from minus sign problems.
808: Nevertheless, we may give some qualitative arguments on the 
809: expected phase diagram at zero temperature.
810: As discussed above, the metallic phase (i) is not so sensitive to 
811: $V$ as far as it is small. This is also the case for the completely 
812: insulating phase (ii). In contrast, a more subtle situation appears
813: in the intermediate phase (iii).  As mentioned above,
814: the intermediate phase exhibits Kondo-like heavy fermion
815: behavior at low temperatures in the presence of $V$. 
816: Recall, however,  that we are now concerned 
817: with the half-filled band. Therefore, this Kondo-like metallic phase 
818: should acquire a Kondo-insulating gap due to commensurability
819: at zero temperature.  We would thus say that the intermediate
820: phase (iii) is changed into the Kondo-insulator with a tiny
821: excitation gap in the presence of $V$ at zero temperature. Accordingly, 
822:  the  sharp transition between the phases (ii) and (iii) at $V=0$
823: may  be smeared and changed to crossover behavior.
824: These considerations lead us to 
825: a schematic description of phase diagram for the two-orbital model
826: with mixing between the 
827: distinct orbitals, as shown in Fig. \ref{fig:phase}.
828: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
829: \begin{figure}
830: \begin{center}
831: \includegraphics[width=8cm]{schematic+.eps}
832: \end{center}       
833: %\vskip -35mm
834: \caption{
835: Schematic phase diagram for the two-orbital  Hubbard model 
836: with finite hybridization between two orbitals.
837: Solid lines represent the phase boundaries between the metallic and 
838: insulating phases. 
839: Dashed line indicates the crossover between the Mott insulator and 
840: the Kondo insulator.
841: }
842: \label{fig:phase}
843: \end{figure}
844: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
845: On the line of $V=0$, the OSMT,
846: which may occur in general choices of the parameters, separates 
847: the phase at $V=0$ into three regions.
848: The metallic phase for small $U$ is simply driven to 
849: the band-insulator (iv) beyond  a certain critical value of hybridization.
850: The intermediate phase  at $V=0$ is 
851:  changed to the Kondo-insulator 
852: in the presence of any finite $V$. This insulating state 
853: first undergoes a phase transition to the metallic phase, and
854: eventually enters the band-insulator as $V$ increases.
855: The completely Mott insulating phase first shows a crossover to
856: the Kondo insulator, which is further driven to the metallic phase 
857: and then to the band-insulating phase.
858: Note that at finite temperatures above the Kondo-insulating gap,
859: we can observe  a Kondo-type heavy
860: fermion behavior in the intermediate phase  with finite $V$.
861: 
862: 
863: %Before closing this section, 
864: %we wish to mention the recent experimental results for 
865: %the angle-resolved photoelectron spectroscopy (ARPES),\cite{ARPES}
866: %where three distinct Fermi surfaces were observed in the compound 
867: %$\rm Ca_{1.5}Sr_{0.5}RuO_4$.
868: %This result may be qualitatively explained by the fact that 
869: %the metallic state is induced by the hybridization.
870: %In addition, this result is consistent with the Kondo-like behavior 
871: %observed,
872: %which is similar to the typical $f$-electron system $\rm CeRu_2Si_2$.
873: %\cite{private}
874: %However, two-dimensional structure, which was not taken into account 
875: %in our framework, is also important in the compounds,
876: %which is under consideration.
877: 
878: 
879: %%%%%%%%%%%%%%%%%%%%
880: \section{Summary}
881: %%%%%%%%%%%%%%%%%%%
882: 
883: We have investigated the degenerate Hubbard model with distinct 
884: hopping integrals by combining DMFT with 
885: QMC simulations. By examining the spin, charge 
886: and orbital susceptibilities calculated at finite temperatures,
887: we have clarified that equally enhanced spin and 
888: orbital fluctuations play a vital role
889: on stabilizing the metallic states in the multi-orbital systems.
890: This remarkable effect is responsible for whether the system
891: undergoes a single Mott transition or OSMTs.
892: Also,  we have discussed how the phase diagram at
893: finite temperatures slightly
894: deviates from the ground-state one because of smearing 
895: effect of the narrow quasi-particle peak.
896: 
897: We have further explored the effect of the hybridization between 
898: the distinct orbitals, and have found that it plays a
899: crucial role especially around  the OSMT.
900: The introduction of the hybridization in  the intermediate phase 
901:  enhances the charge and 
902: orbital fluctuations,  inducing the metallic phase 
903: with a sharp quasi-particle peak.
904: Accordingly,  Kondo-like heavy fermion states show up 
905: at finite temperatures, which eventually drop in
906: the Kondo insulating phase for our half-filled bands.
907: We have also pointed out that the hybridization effect  smears the sharp 
908: OSMT at zero temperature, and changes it to a
909: crossover behavior.  Nevertheless, we can still observe 
910: the OSMT at finite temperatures.
911: 
912: In this paper, we have used QMC as an impurity solver
913: in DMFT, which is not powerful enough to treat  properties at 
914: very low temperatures. Therefore, it is desirable to 
915: exploit a complementary approach to study such low-temperature 
916: properties more precisely, although we have arrived at a reasonable 
917: phase diagram at zero temperature.  Various remaining open
918: problems could not be addressed in the present study.
919:  One of the most important issues to explore is
920:  magnetism of the system, which has not been seen here, 
921: since we have restricted our attention to the 
922: paramagnetic phase. This problem is under 
923: consideration.
924: 
925: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
926: \section{Acknowledgments}
927: We would like to thank K. Ishida, S. Sakai, S. Nakatsuji and Y. Maeno 
928: for useful discussions.
929: This work was partly supported by a Grant-in-Aid from the Ministry 
930: of Education, Science, Sports and Culture of Japan, 
931: the Swiss National Science Foundation and the Centre of Theoretical Studies
932: at ETH Z\"urich.
933: A part of computations was done at the Supercomputer Center at the 
934: Institute for Solid State Physics, University of Tokyo
935: and Yukawa Institute Computer Facility.
936: 
937: 
938: 
939: 
940: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
941: %                        REFERENCES                                 %
942: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
943: %
944: \begin{thebibliography}{99}
945: 
946: \bibitem{Anisimov}
947: V.I. Anisimov, I.A. Nekrasov, D.E. Kondakov, T.M. Rice and M. Sigrist,
948: Eur. Phys. J. B {\bf 25}, 191 (2002).
949: 
950: \bibitem{Fang}
951: Z. Fang and K. Terakura, Phys. Rev. B {\bf 64}, 020509(R) (2001);
952: Z. Fang, N. Nagaosa and K. Terakura, %unpublished.
953: Phys. Rev. B {\bf 69}, 045116 (2004).
954: 
955: \bibitem{KogaLett}
956: A. Koga, N. Kawakami, T.M. Rice, and M. Sigrist, 
957: Phys. Rev. Lett. {\bf 92}, 216402 (2004).
958: 
959: \bibitem{Liebsch}
960: A. Liebsch, Europhys. Lett., {\bf 63}, 97 (2003); 
961: Phys. Rev. Lett., {\bf 91}, 226401 (2003);
962: cond-mat/0405410.
963: 
964: \bibitem{Sigrist}
965: M. Sigrist and M. Troyer, Eur. J. Phys. B, {\bf 39}, 207 (2004)
966: 
967: \bibitem{sces}
968: A. Koga, N. Kawakami, T.M. Rice, and M. Sigrist, cond-mat/0406457.
969: 
970: 
971: %\bibitem{PT} Y. Maeno, T.M. Rice and M. Sigrist, Physics Today, 
972: %{\bf 54}, 42 (2001).
973: 
974: %\bibitem{RMP}
975: %A.P. Mackenzie and Y. Maeno, Rev. Mod. Phys. {\bf 75}, 657 (2003).
976: 
977: \bibitem{Nakatsuji}
978: S. Nakatsuji et al., Phys. Rev. Lett. 90, 137202 (2003);
979: S. Nakatsuji and Y. Maeno, Phys. Rev. Lett. {\bf 84}, 2666 (2000).
980: 
981: \bibitem{LaNiO}
982: K. Sreedhar, {\it et al.}, J. Solid State Comm. {\bf 110} (1994) 208;
983: Z. Zhang, {\it et al.}, 
984: J. Solid State Comm. {\bf 108}, 402 (1994); {\bf 117} (1995) 236.
985: 
986: \bibitem{Kobayashi}
987: Y. Kobayashi, S. Taniguchi, M. Kasai, M. Sato, T. Nishioka, M. Kontani,
988: J. Phys. Soc. Jpn. {\bf 65}, 3978 (1996).
989: 
990: \bibitem{Kusunose}
991: H. Kusunose, S. Yotsuhashi, and K. Miyake, Phys. Rev. B {\bf 62}, 4403 (2000).
992: 
993: \bibitem{tilting}
994: %M. Braden, {\it et al.}, Phys. Rev. B {\bf 58}, 847 (1998);
995: O. Friedt, {\it et al.}, Phys. Rev. B {\bf 63}, 174432 (2001). 
996: 
997: \bibitem{Metzner}
998: W. Metzner and D. Vollhardt, Phys. Rev. Lett. {\bf 62}, 324 (1989).
999: 
1000: \bibitem{Muller}
1001: E. M\"uller-Hartmann, Z. Phys. B: Condens. Matter {\bf 74}, 507 (1989).
1002: 
1003: \bibitem{Georges}
1004: A. Georges, G. Kotliar, W. Krauth and M. J. Rozenberg,
1005: Rev. Mod. Phys. {\bf 68}, 13 (1996).
1006: 
1007: \bibitem{Pruschke}
1008: T. Pruschke, M. Jarrell, and J.K. Freericks, Adv. Phys. {\bf 42}, 187 (1995).
1009: 
1010: \bibitem{Hirsch}
1011: J. E. Hirsch and R. M. Fye, Phys. Rev. Lett. {\bf 56}, 2521 (1986).
1012: 
1013: \bibitem{Sakai}
1014: S. Sakai, R. Arita, and H. Aoki, cond-mat/0405503.
1015: 
1016: 
1017: %\bibitem{ARPES}
1018: %S.-C. Wang {\it et al.},
1019: %cond-mat/0407040.
1020: 
1021: 
1022: 
1023: %%%%%%% PAM
1024: \bibitem{Coleman}
1025: P. Coleman, Phys. Rev. B {\bf 28}, 5255 (1983); {\bf 29}, 3035 (1984).
1026: 
1027: \bibitem{Rice}
1028: T.M. Rice and K. Ueda, Phys. Rev. Lett. {\bf 55}, 995 (1985);
1029: Phys. Rev. B {\bf 34}, 6420 (1986).
1030: 
1031: \bibitem{Yamada}
1032: K. Yamada, K. Yosida, and K. Hanzawa, Prog. Thero, Phys. Suppl.
1033: {\bf 108}, 141 (1992).
1034: 
1035: \bibitem{Kuramoto}
1036: Y. Kuramoto, Z. Phys. B {\bf 53}, 37 (1983).
1037: 
1038: \bibitem{Kim}
1039: C. Kim, Y. Kuramoto, and T. Kasuya, J. Phys. Soc. Jpn. {\bf 59}, 2414 (1990).
1040: 
1041: %%%%%%%%%% Double exchange
1042: \bibitem{Zener}
1043: C. Zener, Phys. Rev. {\bf 82}, 4031 (1951).
1044: 
1045: \bibitem{Anderson}
1046: P. W. Anderson and H. Hasegawa, Phys. Rev. {\bf 100}, 675 (1955).
1047: 
1048: \bibitem{Kubo}
1049: K. Kubo and N. Ohata, J. Phys. Soc. Jpn. {\bf 33}, 21 (1975).
1050: 
1051: \bibitem{Furukawa}
1052: N. Furukawa, J. Phys. Soc. Jpn. {\bf 64}, 2734 (1995).
1053: 
1054: %%%%%%%%%%%%
1055: 
1056: 
1057: %%%%%%%%%%%%%%%%%% single %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1058: \bibitem{Caffarel}
1059: M. Caffarel and W. Krauth, Phys. Rev. Lett. {\bf 72}, 1545 (1994).      
1060: 
1061: \bibitem{2site}
1062: R. Bulla and M. Potthof, Eur. Phys. J. B {\bf 13}, 257 (2000);
1063: M. Potthoff, Phys. Rev. B {\bf 64}, 165114 (2001).
1064: 
1065: \bibitem{single1}
1066: Th. Pruschke, D. L. Cox and M. Jarrell:  Phys. Rev. B 47 (1993) 3553.
1067: 
1068: \bibitem{OSakai}
1069: O. Sakai and Y. Kuramoto, Solid State Comm. {\bf 89}, 307 (1994).
1070: 
1071: \bibitem{single2}
1072: R. Chitra and G. Kotliar, Phys. Rev. Lett. {\bf 83}, 2386 (1999).
1073: 
1074: \bibitem{single3}
1075: J. Joo and V. Oudovenko, Phys. Rev. B {\bf 64}, 193102 (2001).
1076: 
1077: \bibitem{single4}
1078: M. S. Laad, L. Craco and E. M\"uller-Hartmann,
1079: Phys. Rev. B {\bf 64}, 195114 (2001). 
1080: 
1081: \bibitem{BullaNRG}
1082: R. Bulla, Phys. Rev. Lett. {\bf 83}, 136 (1999).
1083: 
1084: %%%%%%%%%%%%%%%%% multi-band %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1085: \bibitem{2band1}
1086: A. Georges, G. Kotliar and W. Krauth, Z. Phys. B {\bf 92}, 313 (1993).
1087: 
1088: \bibitem{2band2}
1089: Th. Maier, M. B. Z\"olfl, Th. Pruschke and J. Keller, 
1090: Eur. Phys. J. B {\bf 7}, 377 (1999).
1091: 
1092: \bibitem{OnoED}
1093: Y. Ono, R. Bulla and A. C. Hewson, Eur. Phys. J. B {\bf 19}, 375 (2001);
1094: Y. Ohashi and Y. Ono, J. Phys. Soc. Jpn. {\bf 70}, 2989 (2001).
1095: 
1096: \bibitem{Koga}
1097: A. Koga, Y. Imai and N. Kawakami, Phys. Rev. B {\bf 66}, 165107 (2002);
1098: %
1099: %\bibitem{KogaSN}
1100: A. Koga, T. Ohashi, Y. Imai, S. Suga and N. Kawakami,
1101: J. Phys. Soc. Jpn. {\bf 72}, 1306 (2003).
1102: 
1103: 
1104: \bibitem{Momoi}
1105: T. Momoi and K. Kubo, Phys. Rev. B {\bf 58}, R567 (1998).
1106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1107: 
1108: \bibitem{multi}
1109: G. Kotliar and H. Kajueter, Phys. Rev. B {\bf 54}, R14221 (1996);
1110: M. J. Rozenberg, Phys. Rev. B {\bf 55}, R4855 (1997);
1111: K. Held and D. Vollhardt, Eur. Phys. J. B {\bf 5}, 473 (1998);
1112: J. E. Han, M. Jarrell and D. L. Cox, Phys. Rev. B {\bf 58}, R4199 (1998);
1113: Y. Imai and N. Kawakami, J. Phys. Soc. Jpn. {\bf 70}, 2365 (2001);
1114: V. S. Oudovenko and G. Kotliar, Phys. Rev. B {\bf 65}, 075102 (2002);
1115: Y. Ono, M. Potthoff and R. Bulla, Phys. Rev. B {\bf 67}, 035119 (2003);
1116: Y. Tomio and T. Ogawa, cond-mat/0407314;
1117: Th. Pruschke and R. Bulla, cond-mat/0411186.
1118: 
1119: 
1120: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1121: 
1122: \bibitem{PAM}
1123: M. Jarrell, H. Akhlaghpour and T. Pruschke,
1124: Phys. Rev. Lett. {\bf 70}, 1670 (1993). 
1125: 
1126: \bibitem{Mutou}
1127: T. Mutou and D. Hirashima, J. Phys. Soc. Jpn. {\bf 63}, 4475 (1994);
1128: 
1129: \bibitem{Saso}
1130: T. Saso and M. Itoh, Phys. Rev. B {\bf 53}, 6877 (1996);  
1131: %T. Muto, Phys. Rev. B {\bf 64}, 165103 (2001); {\bf 64}, 245102 (2001).
1132: 
1133: \bibitem{Sato}
1134: R. Sato, T. Ohashi, A. Koga, N. Kawakami, 
1135: J. Phys. Soc. Jpn. {\bf 73}, 1864 (2004).
1136: 
1137: \bibitem{Ohashi}
1138: T. Ohashi, A. Koga, S. Suga, and N. Kawakami, 
1139: Phys. Rev. B {\bf 70} 245104 (2004).
1140: 
1141: \bibitem{Medici}
1142: L. de' Medici, A. Georges, G. Kotliar, and S. Biermann, cond-mat/0502563.
1143: 
1144: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1145: 
1146: 
1147: 
1148: \bibitem{MEM1}
1149: S. F. Gull, in {\it Maximum Entropy and Bayesian Methods in Science 
1150: and Engineering}, ed. G. J. Erickson and C. R. Smith 
1151: (Kluwer Academic, Dordrecht, 1988) p. 53;
1152: J. Skilling(Kluwer Academic, Dordrecht, 1989) p. 45;
1153: S. F. Gull, {\it ibid}. p. 53.
1154: 
1155: \bibitem{MEM2}
1156: R. N. Silver, D. S. Sivia and J. E. Gubernatis, 
1157: Phys. Rev. B {\bf 41}, 2380 (1990); 
1158: J. E. Gubernatis, M. Jarrell, R. N. Silver and D. S. Sivia, 
1159: Phys. Rev. B {\bf 44}, 6011 (1991).                                                
1160: 
1161: \bibitem{MEM3}
1162: W. F. Press, S. A. Teukolsky, W. T. Vetterling and B. R. Flannery, 
1163: {\it Numerical Recipes} (Cambridge University Press, Cambridge, 1992) p. 809.
1164: 
1165: \end{thebibliography}
1166: 
1167: %%%
1168: 
1169: 
1170: 
1171: \end{document}
1172: