1: \documentclass[twocolumn,aps,prl,amsmath,floatfix]{revtex4}
2: \usepackage{graphicx}
3: \begin{document}
4: \title{Comment on ``Orbital-selective Mott transitions in
5: two-band Hubbard models''}
6: \author{A.~Liebsch}
7: \affiliation{Institut f\"ur Festk\"orperforschung,
8: Forschungszentrum J\"ulich,
9: 52425 J\"ulich, Germany}
10: \begin{abstract}
11: A recent paper by Bl\"umer {\it et al.} [cond-mat/0609758] again
12: criticizes earlier QMC/DMFT results by Liebsch
13: [Phys. Rev. B {\bf 70}, 165103 (2004)].
14: This criticism is shown to be unfounded.
15: \end{abstract}
16: \maketitle
17: In Ref.~1 Bl\"umer {\it et al.}~continue to criticize
18: earlier QMC/DMFT calculations by Liebsch \cite{prb70} for the
19: non-isotropic two-band Hubbard model.
20: Now it is claimed: ``We quantify numerical errors in earlier
21: QMC data which had obscured the second transition'' and:
22: ``The second transition is lost in the noise of earlier data [2]
23: with errors exceeding 100 \% at both transitions''.
24:
25: We point out that Ref.~1 once again does not provide any
26: comparisons of self-energies or spectral distributions [3], nor
27: does it refer to recent work [4,5] which confirms the results
28: of Ref.~2. A direct comparison demonstrates, as we
29: show here, that both QMC calculations are in good agreement
30: and that the above claims are unfounded.
31:
32: \begin{figure}[b!] %1
33: \vskip-2mm
34: \begin{center}
35: \includegraphics[width=5.0cm,height=8cm,angle=-90]{fig13b.ps}
36: \includegraphics[width=5.0cm,height=8cm,angle=-90]{Gnubl.ps}
37: \end{center}
38: \vskip-3mm
39: \caption{
40: (a) QMC subband self-energies $\Sigma_i(i\omega_n)$
41: for different $U$, from Fig.~10 of Ref.~2.
42: Solid red curves: narrow band; dashed blue curves: wide band.
43: (b) QMC self-energy of wide band, adapted from Fig.~2 of Ref.~1 (KBD).
44: The narrow band is insulating in this range.
45: }\end{figure}
46:
47: Fig.~1 shows the comparison of QMC self-energies calculated
48: in Ref.~2 with those of Ref.~1. Fig.~2 shows analogous results
49: obtained via exact diagonalization (ED) and
50: numerical renormalization group (NRG) \cite{al+costi}.
51:
52: \begin{figure}[b!] %2
53: \vskip-2mm
54: \begin{center}
55: \includegraphics[width=5.0cm,height=8cm,angle=-90]{fig13a.ps}
56: % \includegraphics[width=5.5cm,height=8cm,angle=-90]{fig13b.ps}
57: \includegraphics[width=5.0cm,height=8cm,angle=-90]{fig13c.ps}
58: \end{center}
59: \vskip-3mm
60: \caption{
61: (a) ED subband self-energies for the same parameters as in Fig.~1(a).
62: Solid red curves: narrow band; dashed blue curves: wide band.
63: (b) Comparison of ED and NRG self-energies of wide band.
64: The narrow band is insulating in this range.
65: (Both results from Ref.~5).
66: }\end{figure}
67:
68: Evidently, all calculations give the same trend: When the
69: narrow band becomes insulating, the self-energy of the wide band no
70: longer exhibits $\sim i\omega_n$ behavior at low frequencies, as
71: would be characteristic of a Fermi-liquid. Instead, it shows
72: progressive bad-metallic behavior, approaching a finite value in
73: the $i\omega_n\rightarrow0$ limit. This value grows with increasing
74: $U$, until it diverges near 2.7~eV. Ref.~2 states:
75: ``$\Sigma_2(i\omega_n)$ becomes inversely proportional to $\omega_n$
76: at 2.7~eV, i.e., a gap opens up.''
77:
78: Precisely this behavior is seen in the quasi-particle spectra derived
79: in Refs.~1 and 2 (see Fig.~3).
80: Despite the differences caused by different maximum entropy
81: fitting parameters, the low-frequency region is in perfect agreement.
82: Both spectra show that, when the narrow band becomes insulating, the
83: wide band reveals a pseudogap which gets progressively deeper with
84: increasing $U$, until this band becomes fully insulating near
85: 2.7~eV.
86:
87: \begin{figure}[t!] %3
88: \begin{center}
89: \includegraphics[width=4.1cm,height=3.9cm,angle=-90]{Fig2a.ps}
90: \includegraphics[width=4.1cm,height=3.9cm,angle=-90]{Fig2d.ps}
91: \includegraphics[width=4.1cm,height=3.9cm,angle=-90]{Fig2b.ps}
92: \includegraphics[width=4.1cm,height=3.9cm,angle=-90]{Fig2e.ps}
93: \includegraphics[width=4.1cm,height=3.9cm,angle=-90]{Fig2c.ps}
94: \includegraphics[width=4.1cm,height=3.9cm,angle=-90]{Fig2f.ps}
95: \end{center}
96: \vskip-3mm
97: \caption{
98: Quasiparticle spectra from Ref.~2 at $T=31$~meV (left panels)
99: and from Ref.~1 (KBD) at $T=25$~meV (right panels).
100: Solid red curves: narrow band; dashed blue curves: wide band;
101: dotted curves: bare densities of states. (See Ref.~3.)
102: } \vskip-2mm
103: \end{figure}
104:
105: The self-energies and spectral distributions indicate that there
106: is good agreement between the QMC results of Refs.~1 and 2.
107: Both describe identical physics and are consistent with the ED
108: and NRG results. There is no evidence of any sort of disparity.
109: Thus, the claims in Ref.~1 have no basis.
110:
111: Fig.~4 compares $Z_i(U)= 1/[1-{\rm Im}\Sigma_i(i\omega_0)/i\omega_0]$.
112: Again, there is good agreement, consistent with the results in Figs.~1--3.
113: The main difference is that the Coulomb energies at which subbands
114: become insulating are slightly lower in Ref.~1 than in Ref.~2 (see
115: also Fig.~3).
116:
117: Although the $Z_i(U)$ are derived from the self-energies in Fig.~1,
118: Ref.~1 claims that $\Delta Z_i(U)$ (obtained by subtracting results
119: from Refs.~1 and 2) reveals a qualitative difference:
120: ``second transition lost in noise'', etc.
121:
122: A proper analysis of $\Delta Z_i(U)$ should, of course, include
123: (i) the different error margins resulting from QMC statistical
124: uncertainties, number of sweeps and time slices, and
125: (ii) the different Coulomb energies at which subbands become insulating,
126: for instance, as a result of a different $U$ mesh, different number of
127: iterations and critical slowing down. These issues are particularly important
128: when $\Sigma_i(i\omega_n)$ becomes singular and $Z_i(U)$ becomes small.
129: Since all of this is ignored in Ref.~1, it is no surprise that the
130: agreement seen in Fig.~4 can be turned, at specific points, into
131: ficticious disagreement of $\Delta Z_i(U)$ of arbitrary magnitude.
132:
133: Evidently the criticism in Ref.~1 is based on a fundamentally inadequate
134: analysis of $\Delta Z_i(U)$. Moreover, Ref.~1 does
135: not provide the reader with direct comparisons of self-energies or
136: spectral functions, such as given here in Figs.~1,3 or in Ref.~3,
137: which demonstrate good agreement. Finally, Ref.~1 ignores that the
138: QMC results of Ref.~2 were fully confirmed by ED and NRG calculations [4,5].
139:
140: \begin{figure}[t!] %4
141: \begin{center}
142: \includegraphics[width=5.0cm,height=8cm,angle=-90]{Fig1.ps}
143: \end{center}
144: \vskip-3mm
145: \caption{
146: Comparison of $Z_i(U)$ derived within QMC for
147: $T=31$~meV, Ising exchange with $J=U/4$ \cite{comment}.
148: $Z_1$: narrow band; $Z_2$: wide band.
149: Solid and open dots (blue): results of Ref.~2 (L);
150: crosses (red): results of Ref.~1 (KBD).
151: }\vskip-3mm
152: \end{figure}
153:
154: We conclude that the QMC/DMFT results of Ref.~2 are correct:
155: The non-isotropic two-band Hubbard
156: model with Ising exchange exhibits a first-order Mott transition near
157: $U=2.1$~eV when the narrow band becomes insulating, with characteristic
158: hysteresis behavior, and there is no sign of first-order behavior
159: when the wide band becomes insulating near $2.7$~eV.
160:
161: As also shown by the ED/DMFT calculations in Ref.~4, to obtain sequential
162: first-order Mott transitions, it is essential to go beyond Ising exchange
163: and include full Hund's coupling.
164: \begin{thebibliography}{99}
165: \bibitem{bl}
166: N. Bl\"umer, C. Knecht, K. Po$\rm\check{z}$gaj$\rm\check{c}$i\'c,
167: and P. G. J. van Dongen,
168: cond-mat/0609758 (accepted for JMMM).
169: See also:
170: C. Knecht, N. Bl\"umer, and P. G. J. van Dongen (KBD),
171: Phys. Rev. B {\bf 72}, 081103(R) (2005);
172: P. G. J. van Dongen, C. Knecht, and N. Bl\"umer,
173: Phys. Stat. Sol. (b) {\bf 243}, 116 (2006).
174:
175: \bibitem{prb70}
176: A. Liebsch,
177: Phys. Rev. B {\bf 70}, 165103 (2004).
178:
179: \bibitem{comment}
180: A. Liebsch, cond-mat/0506138.
181:
182: \bibitem{prl05}
183: A. Liebsch,
184: Phys. Rev. Lett. {\bf 95}, 116402 (2005).
185:
186: \bibitem{al+costi}
187: A. Liebsch and T. A. Costi,
188: Eur. Phys. J. B {\bf 51}, 523 (2006).
189: \end{thebibliography}
190: \vfill
191: \end{document}
192:
193:
194: