1: \documentclass[showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2:
3: \usepackage{amsfonts} % For ams fonts
4: \usepackage[dvips]{graphicx} % For graphics input
5: \usepackage{type1cm} % For arbitrary size font selection
6: \usepackage{color}
7:
8: %%% DEFINITIONS %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
9:
10: \newcommand{\eqnref}[1]{Eqn.~\eqref{#1}} % For referencing equations (label and list)
11: \newcommand{\figref}[1]{Fig.~\ref{#1}} % For figure references
12: \newcommand{\secref}[1]{Sec.~\ref{#1}} % For referencing sections
13: \newcommand{\textprog}[1]{#1} % For highlighting computer programs
14: \newcommand{\ket}[1]{|#1\rangle} % For ket notation
15: \newcommand{\expecval}[1]{\langle #1\rangle} % For expectation values
16: \newcommand{\Bexpecval}[1]{\left\langle #1\right\rangle} % For BIG expectation values
17: \newcommand{\diffd}{\text{d}} % For calculus "d"
18: \newcommand{\e}[1]{\text{e}^{#1}} % For exponentials
19: \newcommand{\cmplxi}{\text{i}} % For complex numbers
20: \newcommand{\bracetextsize}{\displaystyle} % For correct size of text in over(under)brace
21: \newcommand{\tr}{\operatorname{tr}} % For text traces
22: \renewcommand{\vec}[1]{\mathbf{#1}} % For vectors
23: \newcommand{\obrace}[2]{\overbrace{#1}^{\bracetextsize{#2}}} % For overbrace
24: \newcommand{\ubrace}[2]{\underbrace{#1}_{\bracetextsize{#2}}} % For% underbrace
25: \newcommand{\inlineeqn}[1]{$#1$}
26: \newcommand{\perc}{\,\%}
27: \newcommand{\unit}[1]{\,\text{#1}}
28: \newcommand{\punc}[1]{\,#1}
29:
30: %%% DOCUMENT %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
31:
32: \begin{document}
33: \title{Diffusion Monte Carlo study of a valley
34: degenerate electron gas and application to quantum dots}
35:
36: \author{G.~J. Conduit}
37: \email{gjc29@cam.ac.uk}
38: \affiliation{Theory of Condensed Matter, Department of Physics, University of
39: Cambridge, Cavendish Laboratory, 19, J.~J. Thomson Avenue, Cambridge, CB3~0HE,
40: United Kingdom}
41: \author{P.~D. Haynes}
42: \affiliation{Departments of Physics and Materials, Imperial College London,
43: Exhibition Road, London, SW7~2AZ, United Kingdom}
44: \date{\today}
45:
46: \begin{abstract} A many-flavor electron gas (MFEG) in a semiconductor with a
47: valley degeneracy ranging between 6 and 24 was analyzed using diffusion Monte
48: Carlo (DMC) calculations. The DMC results compare well with an analytic
49: expression derived by one of us [Phys. Rev. B \textbf{78}, 035111 (2008)] for
50: the total energy to within $\pm1\perc$ over an order of magnitude range of
51: density, which increases with valley degeneracy. For
52: $\text{Bi}_{2}\text{Te}_{3}$ (six-fold valley degeneracy) the applicable charge
53: carrier densities are between $7\times10^{19}\unit{cm}^{-3}$ and
54: $2\times10^{20}\unit{cm}^{-3}$. DMC calculations distinguished between an exact
55: and a useful approximate expression for the 24-fold degenerate MFEG
56: polarizability for wave numbers $2p_{\text{F}}<q<7p_{\text{F}}$. The analytical
57: result for the MFEG is generalized to inhomogeneous systems by means of a
58: gradient correction, the validity range of this approach is obtained. Employed
59: within a density-functional theory calculation this approximation compares well
60: with DMC results for a quantum dot. \end{abstract}
61:
62: \pacs{71.15.Mb,71.10.Ca,02.70.Ss}
63:
64: \maketitle
65:
66: \section{Introduction}\label{sec:Introduction}
67:
68: Good quantum numbers, that describe conserved quantities as a quantum system
69: evolves, derive their significance from their connection to the powerful
70: conservation laws of physics. In addition to the familiar examples of spin and
71: crystal momentum, under some circumstances electrons in solids can have an
72: additional quantum number that distinguishes them, which we call the
73: \emph{flavor}; we denote the total number of flavors by $\nu$. One example of
74: such a system are semiconductors and semimetals that have degenerate
75: conduction-band valleys, the flavor denotes the electron's
76: valley. Examples of multi-valley semiconductors include Ge, which as shown in
77: \figref{fig:GeBandStructure} has four degenerate valleys (N.B. not eight, as
78: valleys at the Brillouin zone vertices overlap), Si has six degenerate valleys,
79: a Ge-Si alloy has ten degenerate valleys, and
80: $\text{Pb}_{1-x-y}\text{Sn}_{x}\text{Mn}_{y}\text{Te}$ has twelve valleys in the
81: $\Sigma$ band \cite{90sksg12}. The system has been experimentally realized as an
82: electron-hole liquid that forms in drops \cite{76abkos08,08c07}. In these systems
83: the number of flavors (the number of valleys) is well defined and there are
84: strong Coulomb interactions between particles which motivates the analysis. This
85: is in contrast to several other systems in which the number of flavors is poorly
86: defined such as heavy fermions \cite{02zcja03,05k12,06bi02}, charged domain
87: walls \cite{98eogsz09}, a super-strong magnetic field \cite{76ko07}, and spin
88: instabilities \cite{86gq04,00mqg03}; or where the number of flavors is well
89: defined but interactions between particles are weak such as ultracold atoms in
90: optical lattices \cite{04hh04,04hh09,07crd09}.
91:
92: The properties of a many-flavor electron gas (MFEG) in a semiconductor were
93: first studied analytically for the normal phase by \citet{76abkos08}, and for
94: the superconducting phase by \citet{64c04}. Recently one of us \cite{08c07}
95: extended the MFEG analysis by finding an energy functional and gradient
96: expansion, which allowed the study of inhomogeneous systems. However, the
97: analytical treatment was limited to consider the same contributions to the
98: energy as in the random phase approximation (these contributions dominate in the
99: many-flavor limit). To go further requires numerical calculations, the only
100: example of which for a MFEG to date \cite{94g08} used a self-consistent approach
101: for the local field correction formulated by \citet{68stls12} (STLS), see also
102: Ref.~\cite{81st06}. The method was later applied to charge impurities by
103: \citet{97bat12}. The calculations of Ref.~\cite{94g08} were performed for
104: $\nu\leq6$, too few flavors to gauge the applicability of the analytic
105: many-flavor approximation, which is estimated to apply at around six or more
106: flavors \cite{08c07}.
107:
108: \begin{figure}
109: \includegraphics{GeBandStructure}
110: \caption{The Ge band-structure in the $[111]$ direction calculated using a
111: plane-wave pseudopotential method \cite{05csphprp05}. The Fermi energy is at
112: $E=0\unit{eV}$; below are valence bands with the holes centered around H, above
113: are conduction bands. The first conduction band valley is highlighted in bold,
114: low-lying conduction-band electrons are centered around C.}
115: \label{fig:GeBandStructure}
116: \end{figure}
117:
118: In this paper we follow the suggestion of \citet{94g08}, and present the results
119: of what are expected to be more accurate diffusion Monte Carlo (DMC)
120: \cite{94hlr10,97l07,99nu01,01fmnr01} calculations on the MFEG for $\nu\leq24$,
121: which should allow us to verify the analytical MFEG approach. We then examine
122: aspects of the many-flavor approximation that have not yet been studied
123: computationally: in \secref{sec:CASINOElectronGasDensityResponse} we compare the
124: analytical density-density response function derived in
125: \secref{sec:AnalyticalPolarisability} with that predicted using DMC. Once
126: verified this allows us in \secref{sec:QuantumDotsQMC} to employ a gradient
127: expansion within density-functional theory (DFT) to find the ground state of a
128: quantum dot, we compare results with DMC calculations and examine the validity
129: of the gradient expansion.
130:
131: We adopt the atomic system of units: that is
132: $e^{2}=\hbar=m=1/(4\pi\epsilon_{0})=1$. The mass $m=m_{\text{e}}m^{*}$ is
133: defined to be the electron mass, $m_{\text{e}}$, multiplied by a dimensionless
134: effective mass $m^{*}$ appropriate for the conduction-band valleys, which when
135: $m^{*}=1$ will recover standard atomic units. We assume the valleys all have the
136: same dispersion profile and so the same effective mass, \citet{76abkos08}
137: outlined a method of calculating a scalar effective mass for anisotropic
138: valleys. With the above definitions, energy is given in terms of an exciton
139: $E_{\text{h}}^{*}=E_{\text{h}}m^{*}$, where $E_{\text{h}}$ is the Hartree
140: energy, and length $a_{0}^{*}=a_{0}/m^{*}$ in terms of the Bohr radius
141: $a_{0}$. To denote density we use both the number density of conduction-band
142: electrons $n$ and the Wigner-Seitz radius $r_{\text{s}}$.
143:
144: Before presenting the numerical results, to orient the discussion, we describe
145: the basic physics of the MFEG and review the analytical results of
146: Ref.~\cite{08c07} that will be computationally verified in this paper.
147:
148: \subsection{Introduction to a MFEG}\label{sec:AnalyticalResults}
149:
150: In a low temperature MFEG, the number of flavors $\nu$, number density of
151: conduction-band electrons $n$, and Fermi momentum $p_{\text{F}}$ are related
152: through
153: \begin{equation}
154: \label{flavoursdensityofstates}
155: n=\frac{\nu p_{\text{F}}^{3}}{3\pi^{2}}\punc{.}
156: \end{equation}
157: At fixed electron density, the Fermi momentum reduces with increasing number of
158: flavors as $p_{\text{F}}\propto\nu^{-1/3}$, so each Fermi surface encloses fewer
159: states. The semiconductor hole band-structure often has a single valence-band
160: minimum at the $\Gamma$ point, such as in Ge, see \figref{fig:GeBandStructure},
161: hence we assume the holes are heavy and are uniformly distributed, providing a
162: jellium background.
163:
164: For a constant number density of particles, the density of states at the Fermi
165: surface, $g$, rises with increasing number of flavors as
166: $g\propto\nu\sqrt{E_{\text{F}}}\propto\nu^{2/3}$. Therefore, the screening
167: length estimated with the Thomas-Fermi approximation \cite{04k11} is
168: $\kappa^{-1}=(4\pi e^{2}g)^{-1/2}\propto\nu^{-1/3}$, and the ratio of the
169: screening to Fermi momentum length-scale varies with number of flavors as
170: $p_{\text{F}}/\kappa\propto\nu^{-2/3}$. In the many-flavor limit $\nu\gg1$, the
171: screening length is much smaller than the inverse Fermi momentum,
172: $\kappa^{-1}\ll p_{\text{F}}^{-1}$, and so the dominant electron-electron
173: interactions have characteristic wave vectors which obey $q\gg p_{\text{F}}$.
174: This is in direct contrast to the random phase approximation (RPA) where
175: $p_{\text{F}}\gg\kappa$, although in both the many-flavor and the RPA, the same
176: Green function contributions with empty electron loops dominate diagrammatically
177: \cite{08c07,76abkos08}. These diagrams contain the greatest number of different
178: flavors of electrons, and as $\nu\gg1$ therefore have the largest matrix
179: element. Since $q\gg p_{\text{F}}$, the typical length-scales of the MFEG are
180: short, this indicates that a local density approximation (LDA) could be
181: applied. This motivation is in addition to the usual reasons for the success of
182: the LDA in DFT \cite{92ptaaj10}, namely that the LDA exchange-correlation hole
183: need only provide a good approximation for the spherical average of the
184: exchange-correlation hole and obey the sum rule \cite{89jg06}.
185:
186: \subsection{Polarizability}\label{sec:AnalyticalPolarisability}
187:
188: In the many-flavor limit the exact result for the polarizability of a MFEG at
189: wave vector $\vec{q}$, and Matsubara frequency $\omega$ is
190: \cite{76abkos08,78br07,08c07}
191: \begin{widetext}
192: \begin{eqnarray}
193: \label{refeqnpolarisability}
194: \Pi_{0}(\omega,q)&=&\frac{\nu}{2\pi^{2}}\Biggl[\frac{\omega}{q}\left({\text{tan}}^{-1}\left(\frac{
195: q/2+p_{\text{F}}}{\omega/q}\right)-{\text{tan}}^{-1}\left(\frac{
196: q/2-p_{\text{F}}}{\omega/q}\right)\right)\nonumber\\
197: &-&\frac{\left(\omega/q\right)^{2}+p_{\text{F}}^2-q^{2}/4}{2q}{\ln}\left(\frac{\left(\omega/q\right)^{2}+\left(q/2+p_{\text{F}}\right)^{2}}{\left(\omega/q\right)^{2}+\left(q/2-p_{\text{F}}\right)^{2}}\right)-p_{\text{F}}\Biggr]\punc{,}
198: \end{eqnarray}
199: \end{widetext}
200: which in the many-flavor limit is approximately
201: \begin{equation}
202: \Pi_{0}(\omega,q)=-\frac{n}{(\omega/q)^2+q^{2}/4}+\mathcal{O}(\nu^{-2/3})\punc{.}
203: \label{refeqnpolarisabilityexpansion}
204: \end{equation}
205: This quantity governs the density-density response of the MFEG so is important
206: to verify. Since \eqnref{refeqnpolarisabilityexpansion} has a simple form it
207: can be used to calculate further properties of the MFEG \cite{08c07}, such as
208: homogeneous energy in \secref{sec:AnalyticalUniformEnergy} and the gradient
209: expansion in \secref{sec:AnalyticalGradient}, which further motivates its
210: numerical verification.
211:
212: \subsection{Homogeneous energy}\label{sec:AnalyticalUniformEnergy}
213:
214: Starting from the approximate expression for polarizability,
215: \eqnref{refeqnpolarisabilityexpansion}, it can be shown that the total energy of
216: a MFEG, including all the exchange and correlation contributions is \cite{08c07}
217: \begin{equation}
218: E=\frac{3}{10}\left(\frac{3\pi^{2}}{\nu}\right)^{2/3}n^{5/3}-\ubrace{A_{\text{3D}}n^{5/4}}{E_{\text{int}}}\punc{,}
219: \label{eqn:CASINOHomogeneousFreeElectronGasTotalEnergy}
220: \end{equation}
221: where $A_{3\text{D}}=\Gamma(-5/4)\Gamma(3/4)/(2\pi^{5/4})(E_{\text{h}}^{*}
222: a_{0}^{*3/4})\approx0.574447(E_{\text{h}}^{*} a_{0}^{*3/4})$ and
223: $E_{\text{int}}$ denotes the interacting energy (which would be zero if
224: electron-electron interactions were ignored).
225:
226: In Ref.~\cite{08c07} it was suggested that this relation for the total energy
227: applies over a density range, at $99\perc$ accuracy, $0.03\nu\ll
228: na_{0}^{*3}\ll(0.074\nu)^{4}$, which widens with number of flavors as $\nu^{4}$
229: (see also Ref.~\cite{76abkos08}). Considering the number of flavors where the
230: range of validity vanishes indicates that the many-flavor limit will apply if
231: there are ten or more flavors. An alternative estimate for the density range is
232: found in \secref{sec:GroundStateEnergyVariation} by comparing the analytical
233: result with DMC calculations.
234:
235: \subsection{Gradient correction}\label{sec:AnalyticalGradient}
236:
237: The applicability of the LDA in a MFEG motivates the search for a gradient
238: expansion to the energy \eqnref{eqn:CASINOHomogeneousFreeElectronGasTotalEnergy}
239: as a way to analyze inhomogeneous systems such as electron-hole drops and
240: quantum dots. The typical momentum transfer in the MFEG is $q\sim4(\hbar
241: a_{0}^{*-1/4})n^{1/4}$, which defines the shortest length-scale over which a LDA
242: can be made, therefore, the maximum permissible gradient in electron density is
243: $|\nabla n|_{\text{max}}\sim qn\sim4(\hbar a_{0}^{*-1/4})n^{5/4}$. A gradient
244: expansion will break down for phenomena with short length-scales, for example
245: mass enhancement \cite{93m08}. If electron density is smoothly varying then
246: starting from \eqnref{refeqnpolarisabilityexpansion}, the gradient correction to
247: the energy for a MFEG is \cite{08c07}
248: \begin{equation}
249: E=E_{0}+\frac{1}{8}\frac{(\nabla n)^{2}}{n}\punc{,}
250: \label{eqn:GradientExpansion}
251: \end{equation}
252: where $E_{0}$ is the energy of a homogeneous MFEG with density $n$,
253: see \eqnref{eqn:CASINOHomogeneousFreeElectronGasTotalEnergy}. As discussed in
254: \secref{sec:AnalyticalResults}, this gradient expansion would be useful for DFT
255: calculations and so its computational verification is important.
256:
257: \section{Computational method}
258:
259: In this section we briefly describe the two computational methods that we used,
260: variational Monte Carlo (VMC) and Diffusion Monte Carlo (DMC)
261: \cite{01fmnr01}. These are quantum Monte Carlo (QMC) methods, chosen since DMC
262: gives the exact ground state energy subject to the fixed node approximation, and
263: both are expected to give more accurate results than the STLS approach used by
264: \citet{94g08}.
265:
266: The VMC method uses a normalizable and differentiable trial wave function
267: $\Psi_{\text{T}}$, of the form discussed below. The Metropolis algorithm
268: \cite{53mrtt06} is used to sample the wave function probability density
269: $|\Psi_{\text{T}}|^{2}$ using a random walk, and make an estimate of the local
270: energy
271: $E_{\text{L}}(\vec{r})=\Psi_{\text{T}}(\vec{r})^{-1}\hat{H}\Psi_{\text{T}}(\vec{r})$.
272: In order to obtain the ground state one could minimize the spatial average of
273: the local energy with respect to the free parameters in the trial wave
274: function. However, it is computationally more stable to minimize the variance in
275: the estimates of the local energy. As VMC obeys the variational principle by
276: construction, it yields an upper bound to the true ground state energy.
277:
278: The more accurate DMC algorithm is a stochastic method that begins with a trial
279: or guiding wave function, in this case the optimized VMC trial wave
280: function. The DMC method is based on imaginary time evolution, which when using the
281: operator $\text{e}^{-t(\hat{H}-E_{\text{T}})}$ projects out the ground state
282: wave function from the trial wave function, and yields an estimate of the ground state
283: energy, $E_{\text{T}}$. The nodal surface on which the wave function is zero (and
284: across which it changes sign) is fixed \cite{82rcal12,99nu01} to be that of the
285: trial wave function, this ensures that the fermionic exchange symmetry is
286: maintained. The DMC algorithm produces the exact ground state energy subject to the
287: fixed node approximation, and is also variational so gives an accurate upper
288: bound to the true ground state energy once the population control bias and finite
289: time-step bias are eliminated. The algorithm used closely follows that
290: described in Ref.~\cite{93unr08}.
291:
292: In our QMC calculations we use a Slater-Jastrow \cite{55j06,01fmnr01,04dtn12}
293: trial wave function. The Slater part of the wave function is a product of
294: determinants, each one corresponding to a different electron spin or
295: flavor. Each determinant is over the spatial orbitals of electrons occupying the
296: lowest energy levels. The determinant changes sign when rows or columns are
297: swapped, this ensures that the wave function is antisymmetric under exchange of
298: electrons with the same flavor and spin. The Slater wave function itself is not
299: the ground state of an interacting electron gas, so to improve the wave
300: function, variational degrees of freedom that account for two-body correlations
301: are included within a Jastrow factor. The Jastrow factor is symmetric under
302: particle exchange so does not alter the particle exchange symmetry of the wave
303: function. Furthermore, the Jastrow factor is always positive so does not alter
304: the wave function nodal surface. The Jastrow factor contains a two-body
305: polynomial term $u(r_{ij})=F(r_{ij})\sum_{l=2}^{6}\alpha_{l}r_{ij}^{l}$, a power
306: series form \cite{04dtn12} in electron separation $r_{ij}$ with optimizable
307: parameters, $\alpha_{l}$. The term $F(r_{ij})$ ensures that the Kato cusp
308: conditions are satisfied \cite{66pb02}. To ensure that electron-electron
309: correlations do not extend beyond the simulation cell, the term is cutoff at the
310: Wigner-Seitz radius. To treat longer-ranged correlations, the Jastrow factor
311: includes a two-body plane-wave expansion,
312: $p(\vec{r}_{ij})=\sum_{A,\vec{G}_{A}}a_{A}\cos(\vec{G}_{A}\cdot\vec{r}_{ij})$.
313: Those reciprocal lattice vectors, $\{\vec{G}_{A}\}$, that are related by the
314: point group symmetry (denoted by $A$) of the Bravais lattice share the same
315: optimizable parameters, $a_{A}$. To ensure accuracy we checked the stability of
316: the VMC results when the expansion order of the $u$ and $p$ terms was increased.
317: At all densities the Jastrow factor optimized cutoff lengths took the maximum
318: allowed value (the Wigner-Seitz radius).
319:
320: The DMC calculations were performed with $57$ different reciprocal lattice
321: vectors and, following \citet{94ob07}, \citet{78c10}, and \citet{80ca08},
322: further VMC calculations were performed at other system sizes (27, 33, 57, and
323: 81 reciprocal lattice vectors) to derive the parameters to extrapolate the DMC
324: energy to infinite system size. Additionally, all the DMC results were
325: extrapolated to have zero time-step between successive steps in the electron
326: random walk. In DMC simulations the acceptance probability of a proposed step in
327: the random walk exceeded $99\perc$. We used $300$ DMC configurations, comparable
328: to the 200-300 used by \citet{94ob07}, and checked for population control bias
329: by ensuring that ground state energy estimates did not vary with a changing
330: number of configurations. All the QMC calculations were performed using the
331: \textprog{CASINO} computer program \cite{06ntdl07}.
332:
333: \section{Homogeneous MFEG}\label{sec:CASINOHomogeneousFreeElectronGas}
334:
335: We start with the simplest possible system to analyze numerically, the
336: homogeneous MFEG, this provides not only a suitable system to validate both
337: theory (\secref{sec:AnalyticalUniformEnergy}) and the QMC many-flavor
338: calculations, but should also confirm the range of densities over which the
339: many-flavor approximation applies. The 3D homogeneous electron gas ($\nu=1$) has
340: been studied before using QMC \cite{78c10,80ca08,94ob07} and these studies
341: provide a useful guide to the method we should follow.
342:
343: To calculate the interaction energy $E_{\text{int}}$ we subtracted the
344: theoretical Thomas-Fermi kinetic energy from the DMC ground state energy (see
345: \eqnref{eqn:CASINOHomogeneousFreeElectronGasTotalEnergy}). At each of 6, 12, 18,
346: and 24 flavors we performed five DMC calculations and interpolated to find where
347: theory and DMC results agree to within $\pm1\%$. Results in
348: \figref{fig:CASINOhomogeneousgas} show that for $\nu>6$ the theory applies over
349: at least an order of magnitude in density to an accuracy of $\pm1\perc$ -- the
350: theory can be applied at fewer flavors than expected. For fewer than $\sim12$
351: flavors the valid logarithmic range of the theory increases with $\nu$, the $18$
352: and $24$ flavor results show a dramatic increase in the range of validity,
353: especially on the high density side. In the limit of many flavors $(\nu>12)$ the
354: expected $99\perc$ range of validity $0.03\nu\ll na_{0}^{*3}\ll(0.074\nu)^{4}$
355: is approximately consistent with the computationally predicted $\pm1\perc$
356: region, therefore the minimum number of flavors required for all aspects of
357: the many-flavor theory to be valid is approximately ten.
358:
359: For Si with $m^{*}=1.08$ the many-flavor limit applies to an accuracy of
360: $\pm1\perc$ for a charge carrier concentration between
361: $4\times10^{23}\unit{cm}^{-3}$ and $1\times10^{24}\unit{cm}^{-3}$, this is
362: greater than the typical maximum carrier density
363: $\sim1\times10^{21}\unit{cm}^{-3}$ and so in Si the formalism is not applicable.
364: In systems with a low effective mass, for example the $\nu=6$ material
365: $\text{Bi}_{2}\text{Te}_{3}$ used in thermoelectric cooling, which has
366: $m^{*}=0.06$ \cite{98hhosk06,06la01,08nkopkrvkjhl03}, the required charge
367: carrier concentration is between $7\times10^{19}\unit{cm}^{-3}$ and
368: $2\times10^{20}\unit{cm}^{-3}$, which compares favorably with the typical
369: maximum carrier density $\sim1\times10^{21}\unit{cm}^{-3}$ and so the
370: many-flavor limit formalism could be applied to low effective mass
371: materials.
372:
373: \begin{figure}
374: \includegraphics{CASINOhomogeneousgas}
375: \caption{The lower panel shows the fractional difference of DMC interaction
376: energy $E_{\text{DMC}}$ from the model $E_{\text{int}}$ with MFEG density $n$
377: (and Wigner-Seitz radius $r_{\text{s}}$) for different numbers of flavors, the
378: dotted lines show $\pm1\%$ disagreement. The central panel bars
379: highlight the numerical region of applicability, the gray shaded area
380: represents the analytically predicted region of $\pm1\%$ applicability. The
381: upper panel shows the total energy for 6, 12, 18 and 24 flavor electron gases.
382: }
383: \label{fig:CASINOhomogeneousgas}
384: \end{figure}
385:
386: The STLS results of \citet{94g08} at $\nu=1$ were $\sim3.4\%$ less negative than
387: the DMC results of \citet{94ob07}, and at $\nu=6$ were $\sim3.1\%$ less negative
388: than our DMC results. This represents a significant difference between our and
389: the STLS results when looking for the $1\%$ range of validity, highlighting the
390: need for the more accurate DMC calculations. The range of validity at
391: $\pm1\perc$ up to at least 24 flavors is to the high density side of the minimum
392: in the total energy seen in \figref{fig:CASINOhomogeneousgas}, but the minimum
393: $n_{\text{min}}\propto\nu^{8/5}$ lies within the region of validity for higher
394: $\nu$. $n_{\text{min}}$ is the density expected to be seen in physical systems
395: such as electron-hole drops, the good agreement of the theory with DMC results
396: at this density indicates that the theory could be usefully applied to
397: investigate the properties of physical systems, see for example
398: Ref.\cite{08c07}.
399:
400: \section{Static density-density response}\label{sec:CASINOElectronGasDensityResponse}
401:
402: Having verified the homogeneous system behavior we may now proceed and
403: computationally examine inhomogeneous behavior through the static
404: density-density (linear) response function
405: \eqnref{refeqnpolarisabilityexpansion}. The polarizability is an important
406: quantity used \cite{08c07} to develop both homogeneous theory and the gradient
407: correction, the density-density response function itself also governs the
408: electrical response properties, for example polarization, screening, and
409: behavior in an external potential; it is therefore useful to verify this
410: response before applying the theory to model systems. We examine
411: $1/\epsilon(\vec{q})$, the quantity probed experimentally \cite{99np10}.
412:
413: DMC has previously been used to find the static density-density response of
414: single-flavor systems: \citet{92sba11} applied the method to charged bosons, the
415: density-density response of the electron gas was calculated by \citet{92mcs09}
416: (in two dimensions), and \citet{94bsa11} and \citet{95mcs07} (three dimensions).
417: However, density-density response has not been studied numerically in a
418: many-flavor system. Here we employ two methods to find the density-density
419: response function. The more accurate and computationally efficient method of
420: calculating the response is to examine the ground state energy, calculated using
421: DMC. A VMC energy based estimate and an estimate using the induced electron
422: density are used to check the accuracy of the trial wave function.
423:
424: Before the results are described in
425: \secref{sec:StaticDielectricResponseResults}, we outline the theory behind the
426: two methods used to estimate the response, firstly in
427: \secref{sec:GroundStateEnergyVariation} by using the ground state energy
428: variation, and secondly in \secref{sec:InducedChargeDensityMeasurement} through
429: the magnitude of the periodic density modulation.
430:
431: \subsection{Ground state energy variation}\label{sec:GroundStateEnergyVariation}
432:
433: To calculate the density-density response we use a weak probe so that the
434: density response is solely due to the properties of the homogeneous system. We
435: apply a static ($\omega=0$) monochromatic perturbative external potential
436: $U(\vec{r})=U_{\vec{q}}\cos(\vec{q}\cdot\vec{r})$ to the homogeneous MFEG,
437: corresponding to the background charge having an additional sinusoidal variation
438: $n_{\text{ext}}(\vec{r})=n_{\vec{q}}\cos(\vec{q}\cdot\vec{r})$. The external
439: potential and external charge are linked \cite{92sba11} through Poisson's
440: equation by
441: \begin{equation}
442: \label{eqn:ExternalChargeDensity}
443: n_{\text{ext}}(\vec{k})=\frac{U_{\vec{q}}q^{2}}{8\pi}(\delta_{\vec{k},\vec{q}}+\delta_{\vec{k},\vec{-q}})\punc{.}
444: \end{equation}
445: We assume that different Fourier components are independent, the density
446: response to an external potential with wave vector $\vec{q}$ and frequency
447: $\omega$ is only at that wave vector and frequency so the induced charge is
448: $n_{\text{ind}}(\vec{k})=\left(\Bexpecval{\hat{n}_{\vec{k}}}_{U_{\vec{q}}}-\Bexpecval{\hat{n}_{\vec{k}}}_{0}\right)(\delta_{\vec{k},\vec{q}}+\delta_{\vec{k},\vec{-q}})$.
449: Here $\expecval{\hat{n}_{\vec{k}}}_{U_{\vec{q}}}$ is the expectation value of
450: the charge density Fourier component at wave vector $\vec{k}$ with an applied
451: external potential $U_{\vec{q}}$, and $\expecval{\hat{n}_{\vec{k}}}_{0}$ is the
452: same but in the homogeneous case with no external potential. Linear response
453: theory gives the static density-density response function as the ratio of the
454: induced charge density and the perturbing external charge density so
455: \begin{equation}
456: \label{eqn:StaticDielectricFunctionDefinition}
457: \frac{1}{\epsilon(\vec{q})}=1+\frac{8\pi}{U_{\vec{q}}q^{2}}\left(\Bexpecval{\hat{n}_{\vec{q}}}_{U_{\vec{q}}}-\Bexpecval{\hat{n}_{\vec{q}}}_{0}\right)\punc{.}
458: \end{equation}
459: If the external potential is small relative to other typical energies the
460: density response is determined solely by the properties of the homogeneous
461: MFEG. We can expand in small $U_{\vec{q}}$ so that
462: \begin{eqnarray}
463: \Bexpecval{\hat{n}_{\vec{k}}}_{U_{\vec{q}}}-\Bexpecval{\hat{n}_{\vec{k}}}_{0}&\approx&\left.U_{\vec{q}}\frac{\diffd\Bexpecval{\hat{n}_{\vec{k}}}}{\diffd U_{\vec{q}}}\right|_{U_{\vec{q}}=0}=\left.U_{\vec{q}}\frac{\diffd ^{2}E}{\diffd U_{\vec{q}}^{2}}\right|_{U_{\vec{q}}=0}\punc{,}
464: \end{eqnarray}
465: where the induced charge density is calculated by considering the dependence of
466: the ground state energy $E$ on the magnitude of the external field. Substituting
467: this into \eqnref{eqn:StaticDielectricFunctionDefinition} gives an expression
468: for the density-density response
469: \begin{equation}
470: \frac{1}{\epsilon(\vec{q})}=1+\frac{8\pi}{q^{2}}\left.\frac{\diffd ^{2}E}{\diffd U_{\vec{q}}^{2}}\right|_{U_{\vec{q}}=0}\punc{.}
471: \end{equation}
472:
473: To recover the density-density response function at a particular wave vector,
474: several QMC calculations were performed at that wave vector for different
475: amplitudes of the external field. A polynomial fit was made to the ground state
476: energy so as to extract the second derivative. To investigate the lowest order
477: polarizability the applied external field should be as small as possible yet
478: still give statistically significant results, to ensure this we checked that the
479: ground state energy showed only quadratic behavior with applied field amplitude.
480: A further convenient way to check the perturbing field is sufficiently small is
481: to ensure the electric field of the external potential is less than the typical
482: electric field strength between two neighboring electrons, $e/r_{\text{s}}^{2}$.
483:
484: \subsection{Induced charge density measurement}\label{sec:InducedChargeDensityMeasurement}
485:
486: As the external potential is perturbative we use the same plane-wave basis set
487: as employed for the calculations on the homogeneous MFEG described in
488: \secref{sec:CASINOHomogeneousFreeElectronGas}. To account for the modulating
489: density, following \citet{92mcs09}, \citet{94bsa11}, and \citet{95mcs07} we
490: introduce a new $q$ term into the Jastrow factor of the form
491: \begin{equation}
492: \label{eqn:PolarisabilityJastrowqTerm}
493: q(\vec{r}_{i})=b\cos(\vec{q}\cdot\vec{r}_{i})\punc{,}
494: \end{equation}
495: where $b$ is an optimizable parameter, $\vec{r}_{i}$ the position of the $i$th
496: electron, and the wave vector $\vec{q}$ corresponds to that of the perturbative
497: external potential. As $b$ is small, the charge density induced by the
498: perturbative external potential is
499: $n_{\text{ind}}\approx2b\cos(\vec{q}\cdot\vec{r}_{i})$. From
500: \eqnref{eqn:ExternalChargeDensity} and
501: \eqnref{eqn:StaticDielectricFunctionDefinition} it follows that
502: \begin{equation}
503: \label{eqn:InversePermittivityFromJastrowCoefficient}
504: \frac{1}{\epsilon(\vec{q})}=1+\frac{8\pi b}{q^{2}U_{\mathbf{q}}}\punc{.}
505: \end{equation}
506: The optimized value of $b$ was found by variance minimization during a VMC
507: calculation. The relationship then allows us to derive an estimate for the
508: density-density response function for each separate $U_{\vec{q}}$, typically
509: four values were averaged to give a final estimate for the density-density
510: response.
511:
512: \subsection{Results}\label{sec:StaticDielectricResponseResults}
513:
514: We chose to find the polarizability for a MFEG with $\nu=24$ and
515: $r_{\text{s}}=0.6a_{0}^{*}$. This lies at the lower bound of the range of
516: validity near to the minimum in the energy (see
517: \figref{fig:CASINOhomogeneousgas}) at a density expected to be seen in physical
518: systems. This density was also chosen since it had most of the polarizability
519: curve $0.25<1/\epsilon\le1$ in the region of applicability
520: $q\ge2p_{\text{F}}$. Boundary conditions mean that the external potential must
521: be periodic over the simulation cell, therefore the external potential wave
522: vector $\vec{q}$ must be a reciprocal lattice vector. We checked that if the
523: Jastrow factor $q$ term wave vector was changed so that it was incommensurate
524: with the external potential then following optimization $b=0$ within statistical
525: errors; this verified the linear response assumption that Fourier components are
526: independent.
527:
528: \begin{figure}
529: \includegraphics{CASINOpolarisability}
530: \caption{The density-density response $1/\epsilon$ versus the wave vector
531: $|\vec{q}|$ of a MFEG with $\nu=24$ and $r_{\text{s}}=0.6a_{0}^{*}$. The
532: solid curve shows the exact result $1/\epsilon_{\text{exact}}$
533: (\eqnref{refeqnpolarisability}), the dotted curve the
534: \eqnref{refeqnpolarisabilityexpansion} approximation. The shaded grey region
535: $|\vec{q}|/2<p_{\text{F}}$ is where the many-flavor limit breaks down. The
536: points show the values for the permittivity calculated from QMC results, the
537: circle is from modulated charge measurements (CM), the triangle from VMC energy
538: and the cross from DMC energy. The lower panel plots the actual response, the
539: upper panel shows the deviation of response from the exact theoretical result
540: \eqnref{refeqnpolarisability} with standard error bars.}
541: \label{fig:CASINOpolarisability}
542: \end{figure}
543:
544: The results of the calculation are shown in
545: \figref{fig:CASINOpolarisability}. The DMC results obtained by considering the
546: variation in ground state energy (see \secref{sec:GroundStateEnergyVariation})
547: better fits the exact than approximate expression for the polarizability, and
548: though error bars are large can distinguish between the two within one standard
549: deviation. This shows that QMC results can exceed the accuracy of the
550: approximation made in \eqnref{refeqnpolarisabilityexpansion}, though that
551: estimate remains useful. The positive agreement verifies the theory and confirms
552: the accuracy of the \textprog{CASINO} simulations.
553:
554: The ground state energies calculated by VMC were used in the same way as the DMC
555: results to find the density response and provide a reasonable fit, though here
556: error bars are large so comparison is difficult. Following the prescription in
557: \secref{sec:InducedChargeDensityMeasurement} we also derived values for the
558: density-density response function using the charge density modulation at the
559: wavelength of the perturbing potential, $U_{\vec{q}}$. These values agreed
560: within statistics though carried a larger uncertainty than those derived using
561: the ground state energy. Both of these alternative methods appear to
562: underestimate the density-density response. These results are consistent, a
563: smaller charge density response gives a smaller coefficient in the Jastrow
564: factor $q$ term and a smaller reduction in ground state energy. Nevertheless,
565: the reasonable agreement of both VMC estimates and to the DMC results indicates
566: that the trial wave function had an adequate nodal surface.
567:
568: \section{Gradient correction}\label{sec:QuantumDotsQMC}
569:
570: It was important to verify the density-density response as it is a key component
571: to the many-flavor formalism and could be applied to other many-flavor systems
572: where density is expected to be inhomogeneous, for example junctions and the
573: response to defects and impurities. Now that it has been verified, we may
574: proceed to consider a quantity derived from it: the gradient expansion,
575: \eqnref{eqn:GradientExpansion}, which is also useful for analyzing systems with
576: inhomogeneous density. Once we have investigated the validity of such an
577: expansion we can apply the formalism to quantum dots, chosen since they have a
578: large controllable variation in electron density so should provide a good test
579: of the gradient expansion. Quantum dots are commonly made in many-flavor
580: semiconductor materials so can be modeled using a many-flavor formalism, and are
581: a system in which there is current research interest.
582:
583: Quantum dots \cite{93bk03,02rm10} have not previously been studied in the
584: many-flavor limit though there have been several previous computational studies
585: of a single-flavor electron gas confined in a quantum dot. Previous QMC
586: simulations of quantum dots include \citet{91pk05}, \citet{99hsnh02} performed
587: VMC calculations for parabolically confined electrons in circular
588: dots. \citet{96b08} performed fixed-phase DMC simulations. Path-integral QMC
589: calculations have also been performed \cite{92lm06,98mew11,99ehmg04}, these
590: showed poor agreement with results from exact
591: diagonalization~\cite{97m03}. \citet{03bpwggkrp09, 02wghpg11, 03pwrg10} all
592: compared the optical band-gap between DMC calculations and results from other
593: methods. For circular quantum dots \citet{00pul09} found the ground-state using
594: both DMC, a local spin density approximation method, and Hartree-Fock, they then
595: directly compared the ground-state energy, correlation energy, and spin density
596: profiles. \citet{06gguub05} also used DMC to investigate circular quantum
597: dots. Quantum dots have successfully been investigated using DFT
598: \cite{94fv11,97kmr08,99hw02,00pul09}, \citet{00pul09} found the local spin
599: density approximation method predicted ground-state energies that were typically
600: $2\perc$ greater than DMC energies, \citet{94fv11} obtained a $3\perc$ agreement
601: between current-density-functional theory and exact diagonalization results.
602:
603: \subsection{Method}
604:
605: Before describing the study of quantum dots using a many-flavor functional in
606: detail we first outline the general strategy of the numerical calculations.
607: Firstly, a DFT calculation using the many-flavor functional (including the
608: gradient approximation) was performed using a plane-wave basis set. This
609: produced an estimate of the ground-state energy and density according to the
610: many-flavor theory. It also provided a trial wave function that was converted to
611: a B-spline basis set and, with Jastrow factor, was optimized in a VMC calculation,
612: in preparation for a DMC calculation. Finally, the DMC calculation gave a second
613: estimate of the ground state energy and density, exact only for the fixed node
614: approximation. This estimate was compared with the DFT calculation, and
615: also gave an insight into the accuracy of the many-flavor theory.
616:
617: Here we carried out simulations on a quantum dot with a harmonic external
618: potential of the form $V=kr^{2}/2$, where $r$ is the distance to the center of
619: the quantum dot containing a MFEG with 12 flavors. This potential was chosen as
620: it is simple, continuous, realistic \cite{96tahhk10,98sat12}, and has been used
621: in previous computational studies
622: \cite{94fv11,96a02,97kmr08,97kodeaht12,99hsnh02,99hw02,00pul09,03r03}. Filled
623: shells in this potential correspond to $1, 4, 10, 20, 35,\dots$ orbitals (whose
624: degeneracy may be reduced by electron-electron interactions). In DFT we used a
625: supercell containing a single dot to model the aperiodic system with periodic
626: boundary conditions, in DMC non-periodic calculations with just a single quantum
627: dot were performed. The cubic cell was large enough that the trial wave
628: functions had reduced by at least a factor of $10^{-4}$ at its boundary.
629:
630: Trial wave functions were generated using the DFT program \textprog{3Ddotdft},
631: an extended version of \textprog{DOTDFT} \cite{05h01}. This used the many-flavor
632: functional with gradient approximation so had energy density
633: \begin{equation}
634: \label{eqn:QntmDotDFTFunctional}
635: \varepsilon(n(\vec{r}))=-A_{3\text{D}}n^{5/4}+\xi\frac{|\nabla n|^{2}}{8n}\punc{.}
636: \end{equation}
637: A new parameter $\xi$ was introduced that multiplies the gradient term, which
638: allowed us to adjust its size; $\xi=1$ gives the correct analytical expression,
639: and $\xi=0$ the functional without a gradient expansion.
640:
641: The VMC simulations, run in \textprog{CASINO}, used a B-spline basis set
642: \cite{97hgg05,04ag10} because a localized basis set offers
643: significant performance advantages over plane-waves. The wave function was
644: optimized in VMC with a Jastrow factor containing the two-body polynomial $u$
645: term and two-body plane-wave term $p$ with the same form as used in
646: \secref{sec:CASINOHomogeneousFreeElectronGas} and
647: \secref{sec:CASINOElectronGasDensityResponse}, and a one-body electron-potential
648: term $\chi(r_{i})=F(r_{i})\sum_{m=2}^{6}\beta_{m}r_{i}^{m}$ with $F$ determining
649: behavior at the cutoff length, $r_{i}$ the distance of the $i$th electron from
650: the center of the potential, and the $\beta_{m}$ being optimizable parameters;
651: we also note that the $\chi$ term has no central cusp.
652:
653: The many-flavor functional incorrectly adds in the \emph{self-interaction}
654: energy of each electron to its own Coulomb potential. One way to correct for
655: this is to add an additional term to the density functional
656: \cite{81pz05,01dg01}. However, as the number of flavors is increased the ratio
657: of the correct interaction ($\propto\nu^{2}$) to incorrect self-interaction
658: ($\propto\nu$) increases as $\sim2\nu-1$ so in the many-flavor limit the
659: self-interacting energy error may be neglected. To ensure the B-spline grid was
660: sufficiently fine, we compared the trial wave function kinetic and external
661: potential energy before and after conversion the B-spline basis set. We also
662: checked the choice of DMC time-step was sufficiently small, the number of
663: configurations was suitably large, and the simulation cell size was adequately
664: large. On changing these variables the variation in the ground state energy was
665: $\Delta E\approx0.02E_{\text{h}}^{*}$, sufficiently small to allow us to compare
666: the ground state energy as the potential strength and gradient expansion
667: coefficient were varied.
668:
669: \subsection{Results}
670:
671: We analyzed a quantum dot containing a MFEG of 12 flavors and 4 bands (shells),
672: containing a total of 96 electrons. This was chosen since it had a full shell so
673: is expected to have a zero spin ground-state \cite{02rm10} that can be analyzed
674: with the many-flavor functional, was computationally feasible, and contained
675: enough electrons to be in the LDA regime, where the many-flavor functional is
676: expected to apply.
677:
678: Two different investigations were carried out to probe effects of changing the
679: density gradient, firstly strength of the dot confining potential $k$ was
680: changed, and secondly the gradient expansion coefficient $\xi$ was varied.
681:
682: \subsubsection{Varying the external potential strength $k$}
683:
684: \begin{figure}
685: \includegraphics{QntmDotVaryK}
686: \caption{In the upper plot the crosses and solid line show the
687: difference between DFT $(E_{\text{DFT}})$ and DMC $(E)$ energies with varying
688: external potential strength $k$, if agreement were exact, points would lie
689: along the horizontal dotted line. In the lower plot the circles and solid line
690: show the maximum density gradient of the dots with varying $k$, and the squares
691: and dashed line the gradient at which the theory breaks down.}
692: \label{fig:QntmDotVaryK}
693: \end{figure}
694:
695: At the strong external potential $k=8$, corresponding to steep gradients,
696: \figref{fig:QntmDotVaryK} shows the DFT energy is overestimated compared with
697: the DMC result, indicating that the gradient approximation is not applicable and
698: that the next order term in a gradient expansion is
699: negative. \figref{fig:PlotDotRadDensVaryk} shows that the DFT density profile
700: underestimates the true density towards the center of the dot and overestimates
701: density in the outer regions, indicating that the DFT functional does not favor
702: steep enough gradients. This is consistent with the next term in the gradient expansion
703: being negative. The breakdown corresponds to a coefficient of $\alpha\approx1.8$
704: in $q_{\text{max}}=\alpha(\hbar a_{0}^{*-1/4})n^{1/4}$, close to the
705: $\alpha\approx4$ which corresponds to the maximum contribution to the
706: interacting energy.
707:
708: At the intermediate potential $k=1$ the DFT and DMC estimates of energy and
709: the density profile agree, in this region the gradient approximation
710: applies. The DFT density profile shows a slight over-density at the center,
711: consistent with self-interaction energy being included in the DFT
712: calculation. At the weak potential $k=0.1$ electron densities are low meaning
713: the homogeneous interacting energy is outside of its region of applicability
714: (see \figref{fig:CASINOhomogeneousgas}), therefore the DFT energy is an
715: overestimate.
716:
717: \begin{figure}
718: \includegraphics{PlotDotRadDensVaryk}
719: \caption{The lower panel shows the density profile of quantum dots estimated
720: using both DFT and DMC at external potential strengths of $k=1$ and $k=8$. The
721: difference between the DFT and DMC results at $k=1$ and $k=8$ is shown in the
722: upper panel. The DMC statistical error is less than the size of the points.}
723: \label{fig:PlotDotRadDensVaryk}
724: \end{figure}
725:
726: \subsubsection{Varying the gradient term coefficient $\xi$}
727:
728: \begin{figure}
729: \includegraphics{QntmDotVaryChi}
730: \caption{The upper panel shows, for dots with external potential strength
731: $k=1$, the density profile maximum gradient as a function of
732: $\xi$. The central panel shows the variation of external
733: potential DFT energy based on the primary $y$-axis using square points and the solid line,
734: the secondary $y$-axis shows electron-electron DFT energy using crosses and the
735: dashed line. The lower panel solid line shows the variation of DFT ground state
736: energy with $\xi$, and the horizontal dotted line the ground-state energy
737: predicted using DMC from the $\xi=1$ trial wave function. The DMC statistical
738: error is less than the size of the points.}
739: \label{fig:QntmDotVaryChi}
740: \end{figure}
741:
742: \figref{fig:QntmDotVaryChi} shows results of simulations on dots, chosen to have
743: a potential strength $k=1$, which is at the center of agreement of the previous
744: results. The best agreement between the DFT and DMC ground state energy is at
745: $\xi\sim0.9$. This is in good agreement with the expected $\xi=1$, the
746: difference may be due to systematic errors such as the self-interacting energy
747: or higher order gradient terms. As expected, the energy is overestimated for
748: dots with too large a gradient expansion term, and underestimated for dots with
749: too small a gradient correction term.
750:
751: The maximum gradient seen in the dot density profile decreases as $\xi$
752: increases (see \figref{fig:QntmDotVaryChi}). The dot becomes more spread out so
753: the external energy $E_{\text{ext}}$ increases whilst the total
754: electron-electron Coulomb energy $E_{\text{e-e}}$ decreases. Overall the total
755: DFT energy increases. Three quantum dot electron density profiles for gradient
756: term coefficients $\xi=1$, 2, and 3 are shown in
757: \figref{fig:PlotDotRadDensVaryChi}. Compared with the dot calculated with
758: $\xi=1$, the dot generated with no energy penalty for gradients, $\xi=0$, has a
759: high central and low outer density showing that it has a higher gradient in the
760: density. Conversely dot with increased energy cost for gradients, $\xi=2$, has a
761: more shallow profile.
762:
763: \begin{figure}
764: \includegraphics{PlotDotRadDensVaryChi}
765: \caption{The upper panel shows the variation of the $\xi=0$ (dotted line,
766: squares) and the $\xi=2$ (dashed line, crosses) DFT density profiles from the
767: $\xi=1$ DFT density profile. The lower panel shows DFT density profiles for
768: $\xi=0$, $\xi=1$ (solid line, triangles), and $\xi=2$ in a $k=1$ dot.}
769: \label{fig:PlotDotRadDensVaryChi}
770: \end{figure}
771:
772: The density profiles seen in \figref{fig:PlotDotRadDensVaryk} and
773: \figref{fig:PlotDotRadDensVaryChi} can be further analyzed in light of other
774: theoretical studies of quantum dots reviewed in Ref.~\cite{02rm10}. The density
775: profile calculated using the many-flavor functional is not flat at the center,
776: but instead has correlation-induced density inhomogeneity evidenced by a
777: characteristic minimum in the density at $r\approx2a_{0}^{*}$. The
778: intermediate density regime in which this occurs is consistent with the strong
779: correlations causing a minimum in the total many-flavor energy density
780: \cite{08c07}. It is also akin to the intermediate density regime seen in other
781: quantum dot systems \cite{99ehmg04,02rm10,06gguub05,08krpgpgmpw06}, in the high
782: density limit the quantum dot has properties like a Fermi liquid with
783: de-localized electrons \cite{99ehmg04,02rm10,05gv04}, whereas in the low density
784: limit the electrons become crystalline
785: \cite{90lm04,92lm06,93br03,98mvn07,99ehmg04,02rm10} inside the dot. As the
786: many-flavor functional was successful in predicting correlation-induced
787: inhomogeneities, it could be used to investigate other many-flavor quantum dot
788: effects including the Kondo effect in multi-valley semiconductors
789: \cite{07scj05,07sj11}, the reduction of valley degeneracy of coupled quantum
790: dots \cite{03he10,05a08,07bb07}, and harmonically trapped cold atoms with an
791: additional quantum number denoting energy level
792: \cite{04hh04,04hh09,07crd09,07yl11}.
793:
794: \section{Conclusions}
795:
796: We have computationally verified the theory of the MFEG presented in
797: \cite{08c07} using QMC simulations. In a homogeneous system, DMC estimates for
798: the ground state energy are consistent with theory and the theoretically
799: estimated density range over which the theory applies is consistent with
800: numerical results. The applicable density for $\text{Bi}_{2}\text{Te}_{3}$
801: $(\nu=6)$ corresponds to a charge carrier density between
802: $7\times10^{19}\unit{cm}^{-3}$ and $2\times10^{20}\unit{cm}^{-3}$.
803:
804: The density response function for a MFEG with 24 flavors was found using three
805: methods: density modulation predicted by VMC, and the variation in ground state
806: energy predicted by VMC and also by DMC. The two VMC results underestimated the
807: response $1/\epsilon$, but the DMC results agreed with theory and could
808: distinguish between the exact and a useful approximate expression for
809: polarizability.
810:
811: We used a many-flavor functional including a local gradient approximation in DFT
812: calculations of large quantum dots. The DFT calculation estimated the
813: ground-state energy and wave function, which were verified by a DMC
814: calculation. We found the high gradient breakdown of the expansion was at
815: $q_{\text{max}}\approx1.8(\hbar a_{0}^{*-1/4})n^{1/4}$, the low gradient
816: breakdown was consistent with the homogeneous MFEG lowest applicable density,
817: and that the gradient expansion was applicable in the intermediate regime. The
818: many-flavor functional, used as part of DFT calculations, could be a useful tool
819: for analyzing other multi-valley semiconductor systems.
820:
821: \begin{acknowledgements}
822: G.J.C. acknowledges the financial support of an EPSRC studentship, P.D.H. was
823: supported by a Royal Society University Research Fellowship. We thank
824: N.D.M.~Hine for providing the DOTDFT code, P.~L\'opez R\'ios and N.D.~Drummond
825: for help modifying and running CASINO, R.~Needs for providing computing time, and
826: A.J.~Morris for careful reading of the manuscript.
827: \end{acknowledgements}
828:
829: %\bibliography{./references}
830:
831: \begin{thebibliography}{84}
832: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
833: \expandafter\ifx\csname bibnamefont\endcsname\relax
834: \def\bibnamefont#1{#1}\fi
835: \expandafter\ifx\csname bibfnamefont\endcsname\relax
836: \def\bibfnamefont#1{#1}\fi
837: \expandafter\ifx\csname citenamefont\endcsname\relax
838: \def\citenamefont#1{#1}\fi
839: \expandafter\ifx\csname url\endcsname\relax
840: \def\url#1{\texttt{#1}}\fi
841: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
842: \providecommand{\bibinfo}[2]{#2}
843: \providecommand{\eprint}[2][]{\url{#2}}
844:
845: \bibitem[{\citenamefont{Story et~al.}(1990)\citenamefont{Story, Karczewski,
846: \'Swierkowski, and Galazka}}]{90sksg12}
847: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Story}},
848: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Karczewski}},
849: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{\'Swierkowski}},
850: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.~R.}
851: \bibnamefont{Galazka}}, \bibinfo{journal}{Phys. Rev. B}
852: \textbf{\bibinfo{volume}{42}}, \bibinfo{pages}{10477} (\bibinfo{year}{1990}).
853:
854: \bibitem[{\citenamefont{Andryushin et~al.}(1976)\citenamefont{Andryushin,
855: Babichenko, Keldysh, Onishchenko, and Silin}}]{76abkos08}
856: \bibinfo{author}{\bibfnamefont{E.~A.} \bibnamefont{Andryushin}},
857: \bibinfo{author}{\bibfnamefont{V.~S.} \bibnamefont{Babichenko}},
858: \bibinfo{author}{\bibfnamefont{L.~V.} \bibnamefont{Keldysh}},
859: \bibinfo{author}{\bibfnamefont{Y.~A.} \bibnamefont{Onishchenko}},
860: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.~P.} \bibnamefont{Silin}},
861: \bibinfo{journal}{Pis'ma Zh. Eksp. Teor. Fiz.} \textbf{\bibinfo{volume}{24}},
862: \bibinfo{pages}{210} (\bibinfo{year}{1976}).
863:
864: \bibitem[{\citenamefont{Conduit}(2008)}]{08c07}
865: \bibinfo{author}{\bibfnamefont{G.~J.} \bibnamefont{Conduit}},
866: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{78}},
867: \bibinfo{pages}{035111} (\bibinfo{year}{2008}).
868:
869: \bibitem[{\citenamefont{Zar\'and et~al.}(2002)\citenamefont{Zar\'and, Costi,
870: Jerez, and Andrei}}]{02zcja03}
871: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Zar\'and}},
872: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Costi}},
873: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Jerez}}, \bibnamefont{and}
874: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Andrei}},
875: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{65}},
876: \bibinfo{pages}{134416} (\bibinfo{year}{2002}).
877:
878: \bibitem[{\citenamefont{Kim}(2005)}]{05k12}
879: \bibinfo{author}{\bibfnamefont{K.~S.} \bibnamefont{Kim}},
880: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{72}},
881: \bibinfo{pages}{245106} (\bibinfo{year}{2005}).
882:
883: \bibitem[{\citenamefont{Bolech and Iucci}(2006)}]{06bi02}
884: \bibinfo{author}{\bibfnamefont{C.~J.} \bibnamefont{Bolech}} \bibnamefont{and}
885: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Iucci}},
886: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{96}},
887: \bibinfo{pages}{056402} (\bibinfo{year}{2006}).
888:
889: \bibitem[{\citenamefont{Eskes et~al.}(1998)\citenamefont{Eskes, Osman,
890: Grimberg, van Saarloos, and Zaanen}}]{98eogsz09}
891: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Eskes}},
892: \bibinfo{author}{\bibfnamefont{O.~Y.} \bibnamefont{Osman}},
893: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Grimberg}},
894: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{van Saarloos}},
895: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Zaanen}},
896: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{58}},
897: \bibinfo{pages}{6963} (\bibinfo{year}{1998}).
898:
899: \bibitem[{\citenamefont{Keldysh and Onishchenko}(1976)}]{76ko07}
900: \bibinfo{author}{\bibfnamefont{L.~V.} \bibnamefont{Keldysh}} \bibnamefont{and}
901: \bibinfo{author}{\bibfnamefont{T.~A.} \bibnamefont{Onishchenko}},
902: \bibinfo{journal}{Pis'ma Zh. Eksp. Teor. Fiz.} \textbf{\bibinfo{volume}{24}},
903: \bibinfo{pages}{70} (\bibinfo{year}{1976}).
904:
905: \bibitem[{\citenamefont{Giuliani and Quinn}(1986)}]{86gq04}
906: \bibinfo{author}{\bibfnamefont{G.~F.} \bibnamefont{Giuliani}} \bibnamefont{and}
907: \bibinfo{author}{\bibfnamefont{J.~J.} \bibnamefont{Quinn}},
908: \bibinfo{journal}{Surface Science} \textbf{\bibinfo{volume}{170}},
909: \bibinfo{pages}{316} (\bibinfo{year}{1986}).
910:
911: \bibitem[{\citenamefont{Marinescu et~al.}(2000)\citenamefont{Marinescu, Quinn,
912: and Giuliani}}]{00mqg03}
913: \bibinfo{author}{\bibfnamefont{D.~C.} \bibnamefont{Marinescu}},
914: \bibinfo{author}{\bibfnamefont{J.~J.} \bibnamefont{Quinn}}, \bibnamefont{and}
915: \bibinfo{author}{\bibfnamefont{G.~F.} \bibnamefont{Giuliani}},
916: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{61}},
917: \bibinfo{pages}{7245} (\bibinfo{year}{2000}).
918:
919: \bibitem[{\citenamefont{Honerkamp and Hofstetter}(2004{\natexlab{a}})}]{04hh04}
920: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Honerkamp}} \bibnamefont{and}
921: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Hofstetter}},
922: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{92}},
923: \bibinfo{pages}{170403} (\bibinfo{year}{2004}{\natexlab{a}}).
924:
925: \bibitem[{\citenamefont{Honerkamp and Hofstetter}(2004{\natexlab{b}})}]{04hh09}
926: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Honerkamp}} \bibnamefont{and}
927: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Hofstetter}},
928: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{70}},
929: \bibinfo{pages}{094521} (\bibinfo{year}{2004}{\natexlab{b}}).
930:
931: \bibitem[{\citenamefont{Cherng et~al.}(2007)\citenamefont{Cherng, Refael, and
932: Demler}}]{07crd09}
933: \bibinfo{author}{\bibfnamefont{R.~W.} \bibnamefont{Cherng}},
934: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Refael}}, \bibnamefont{and}
935: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Demler}},
936: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{99}},
937: \bibinfo{pages}{130406} (\bibinfo{year}{2007}).
938:
939: \bibitem[{\citenamefont{Cohen}(1964)}]{64c04}
940: \bibinfo{author}{\bibfnamefont{M.~L.} \bibnamefont{Cohen}},
941: \bibinfo{journal}{Phys. Rev.} \textbf{\bibinfo{volume}{134}},
942: \bibinfo{pages}{A511} (\bibinfo{year}{1964}).
943:
944: \bibitem[{\citenamefont{Gold}(1994)}]{94g08}
945: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Gold}}, \bibinfo{journal}{Phys.
946: Rev. B} \textbf{\bibinfo{volume}{50}}, \bibinfo{pages}{4297}
947: (\bibinfo{year}{1994}).
948:
949: \bibitem[{\citenamefont{Singwi et~al.}(1968)\citenamefont{Singwi, Tosi, Land,
950: and Sj\"olander}}]{68stls12}
951: \bibinfo{author}{\bibfnamefont{K.~S.} \bibnamefont{Singwi}},
952: \bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Tosi}},
953: \bibinfo{author}{\bibfnamefont{R.~H.} \bibnamefont{Land}}, \bibnamefont{and}
954: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Sj\"olander}},
955: \bibinfo{journal}{Phys. Rev.} \textbf{\bibinfo{volume}{176}},
956: \bibinfo{pages}{589} (\bibinfo{year}{1968}).
957:
958: \bibitem[{\citenamefont{Singwi and Tosi}(1981)}]{81st06}
959: \bibinfo{author}{\bibfnamefont{K.~S.} \bibnamefont{Singwi}} \bibnamefont{and}
960: \bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Tosi}},
961: \bibinfo{journal}{Solid State Physics} \textbf{\bibinfo{volume}{36}},
962: \bibinfo{pages}{177} (\bibinfo{year}{1981}).
963:
964: \bibitem[{\citenamefont{Bulutay et~al.}(1997)\citenamefont{Bulutay, Al-Hayek,
965: and Tomak}}]{97bat12}
966: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Bulutay}},
967: \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Al-Hayek}}, \bibnamefont{and}
968: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Tomak}},
969: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{56}},
970: \bibinfo{pages}{15115} (\bibinfo{year}{1997}).
971:
972: \bibitem[{\citenamefont{Clark et~al.}(2005)\citenamefont{Clark, Segall,
973: Pickard, Hasnip, Probert, Refson, and Payne}}]{05csphprp05}
974: \bibinfo{author}{\bibfnamefont{S.~J.} \bibnamefont{Clark}},
975: \bibinfo{author}{\bibfnamefont{M.~D.} \bibnamefont{Segall}},
976: \bibinfo{author}{\bibfnamefont{C.~J.} \bibnamefont{Pickard}},
977: \bibinfo{author}{\bibfnamefont{P.~J.} \bibnamefont{Hasnip}},
978: \bibinfo{author}{\bibfnamefont{M.~J.} \bibnamefont{Probert}},
979: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Refson}}, \bibnamefont{and}
980: \bibinfo{author}{\bibfnamefont{M.~C.} \bibnamefont{Payne}},
981: \bibinfo{journal}{Zeitschrift f\"ur Kristallographie}
982: \textbf{\bibinfo{volume}{220}}, \bibinfo{pages}{567} (\bibinfo{year}{2005}).
983:
984: \bibitem[{\citenamefont{Lester}(1997)}]{97l07}
985: \bibinfo{author}{\bibfnamefont{W.~A.} \bibnamefont{Lester}},
986: \emph{\bibinfo{title}{Recent Advances in Quantum Monte Carlo Methods Part
987: II}} (\bibinfo{publisher}{World Scientific}, \bibinfo{year}{1997}).
988:
989: \bibitem[{\citenamefont{Nightingale and Umrigar}(1999)}]{99nu01}
990: \bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Nightingale}}
991: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{C.~J.}
992: \bibnamefont{Umrigar}}, \emph{\bibinfo{title}{Quantum Monte Carlo Methods in
993: Physics and Chemistry}} (\bibinfo{publisher}{Kluwer Academic Publishers},
994: \bibinfo{year}{1999}).
995:
996: \bibitem[{\citenamefont{Foulkes et~al.}(2001)\citenamefont{Foulkes, Mitas,
997: Needs, and Rajagopal}}]{01fmnr01}
998: \bibinfo{author}{\bibfnamefont{W.~M.~C.} \bibnamefont{Foulkes}},
999: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Mitas}},
1000: \bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Needs}}, \bibnamefont{and}
1001: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Rajagopal}},
1002: \bibinfo{journal}{Rev. of Modern Phys.} \textbf{\bibinfo{volume}{73}},
1003: \bibinfo{pages}{33} (\bibinfo{year}{2001}).
1004:
1005: \bibitem[{\citenamefont{Hammond et~al.}(1994)\citenamefont{Hammond, Lester~Jr.,
1006: and Reynolds}}]{94hlr10}
1007: \bibinfo{author}{\bibfnamefont{B.~L.} \bibnamefont{Hammond}},
1008: \bibinfo{author}{\bibfnamefont{W.~A.} \bibnamefont{Lester~Jr.}},
1009: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{P.~J.}
1010: \bibnamefont{Reynolds}}, \emph{\bibinfo{title}{Monte Carlo Methods in Ab
1011: Initio Quantum Chemistry}} (\bibinfo{publisher}{World Scientific},
1012: \bibinfo{year}{1994}).
1013:
1014: \bibitem[{\citenamefont{Kittel}(2004)}]{04k11}
1015: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Kittel}},
1016: \emph{\bibinfo{title}{Introduction to Solid State Physics}}
1017: (\bibinfo{publisher}{Wiley}, \bibinfo{year}{2004}).
1018:
1019: \bibitem[{\citenamefont{Payne et~al.}(1992)\citenamefont{Payne, Teter, Allan,
1020: Arias, and Joannopoulos}}]{92ptaaj10}
1021: \bibinfo{author}{\bibfnamefont{M.~C.} \bibnamefont{Payne}},
1022: \bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Teter}},
1023: \bibinfo{author}{\bibfnamefont{D.~C.} \bibnamefont{Allan}},
1024: \bibinfo{author}{\bibfnamefont{T.~A.} \bibnamefont{Arias}}, \bibnamefont{and}
1025: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Joannopoulos}},
1026: \bibinfo{journal}{Rev. of Modern Phys.} \textbf{\bibinfo{volume}{64}},
1027: \bibinfo{pages}{1045} (\bibinfo{year}{1992}).
1028:
1029: \bibitem[{\citenamefont{Jones and Gunnarsson}(1989)}]{89jg06}
1030: \bibinfo{author}{\bibfnamefont{R.~O.} \bibnamefont{Jones}} \bibnamefont{and}
1031: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Gunnarsson}},
1032: \bibinfo{journal}{Rev. of Modern Phys.} \textbf{\bibinfo{volume}{61}},
1033: \bibinfo{pages}{689} (\bibinfo{year}{1989}).
1034:
1035: \bibitem[{\citenamefont{Beni and Rice}(1978)}]{78br07}
1036: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Beni}} \bibnamefont{and}
1037: \bibinfo{author}{\bibfnamefont{T.~M.} \bibnamefont{Rice}},
1038: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{18}},
1039: \bibinfo{pages}{768} (\bibinfo{year}{1978}).
1040:
1041: \bibitem[{\citenamefont{Mahan}(1993)}]{93m08}
1042: \bibinfo{author}{\bibfnamefont{G.~D.} \bibnamefont{Mahan}},
1043: \emph{\bibinfo{title}{Many-Particle Physics}} (\bibinfo{publisher}{Plenum
1044: Press}, \bibinfo{year}{1993}).
1045:
1046: \bibitem[{\citenamefont{Metropolis et~al.}(1953)\citenamefont{Metropolis,
1047: Rosenbluth, Teller, and Teller}}]{53mrtt06}
1048: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Metropolis}},
1049: \bibinfo{author}{\bibfnamefont{M.~N.} \bibnamefont{Rosenbluth}},
1050: \bibinfo{author}{\bibfnamefont{A.~H.} \bibnamefont{Teller}},
1051: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Teller}},
1052: \bibinfo{journal}{J. of Chem. Phys.} \textbf{\bibinfo{volume}{21}},
1053: \bibinfo{pages}{1087} (\bibinfo{year}{1953}).
1054:
1055: \bibitem[{\citenamefont{Reynolds et~al.}(1982)\citenamefont{Reynolds, Ceperley,
1056: Alder, and Lester}}]{82rcal12}
1057: \bibinfo{author}{\bibfnamefont{P.~J.} \bibnamefont{Reynolds}},
1058: \bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Ceperley}},
1059: \bibinfo{author}{\bibfnamefont{B.~J.} \bibnamefont{Alder}}, \bibnamefont{and}
1060: \bibinfo{author}{\bibfnamefont{W.~A.} \bibnamefont{Lester}},
1061: \bibinfo{journal}{J. of Chem. Phys.} \textbf{\bibinfo{volume}{77}},
1062: \bibinfo{pages}{5593} (\bibinfo{year}{1982}).
1063:
1064: \bibitem[{\citenamefont{Umrigar et~al.}(1993)\citenamefont{Umrigar,
1065: Nightingale, and Runge}}]{93unr08}
1066: \bibinfo{author}{\bibfnamefont{C.~J.} \bibnamefont{Umrigar}},
1067: \bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Nightingale}},
1068: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{K.~J.} \bibnamefont{Runge}},
1069: \bibinfo{journal}{J. of Chem. Phys.} \textbf{\bibinfo{volume}{99}},
1070: \bibinfo{pages}{2865} (\bibinfo{year}{1993}).
1071:
1072: \bibitem[{\citenamefont{Jastrow}(1955)}]{55j06}
1073: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Jastrow}},
1074: \bibinfo{journal}{Phys. Rev.} \textbf{\bibinfo{volume}{98}},
1075: \bibinfo{pages}{1479} (\bibinfo{year}{1955}).
1076:
1077: \bibitem[{\citenamefont{Drummond et~al.}(2004)\citenamefont{Drummond, Towler,
1078: and Needs}}]{04dtn12}
1079: \bibinfo{author}{\bibfnamefont{N.~D.} \bibnamefont{Drummond}},
1080: \bibinfo{author}{\bibfnamefont{M.~D.} \bibnamefont{Towler}},
1081: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Needs}},
1082: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{70}},
1083: \bibinfo{pages}{235119} (\bibinfo{year}{2004}).
1084:
1085: \bibitem[{\citenamefont{Pack and Brown}(1966)}]{66pb02}
1086: \bibinfo{author}{\bibfnamefont{R.~T.} \bibnamefont{Pack}} \bibnamefont{and}
1087: \bibinfo{author}{\bibfnamefont{W.~B.} \bibnamefont{Brown}},
1088: \bibinfo{journal}{J. of Chem. Phys.} \textbf{\bibinfo{volume}{45}},
1089: \bibinfo{pages}{556} (\bibinfo{year}{1966}).
1090:
1091: \bibitem[{\citenamefont{Ortiz and Ballone}(1994)}]{94ob07}
1092: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Ortiz}} \bibnamefont{and}
1093: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Ballone}},
1094: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{50}},
1095: \bibinfo{pages}{1391} (\bibinfo{year}{1994}).
1096:
1097: \bibitem[{\citenamefont{Ceperley}(1978)}]{78c10}
1098: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Ceperley}},
1099: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{18}},
1100: \bibinfo{pages}{3126} (\bibinfo{year}{1978}).
1101:
1102: \bibitem[{\citenamefont{Ceperley and Alder}(1980)}]{80ca08}
1103: \bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Ceperley}} \bibnamefont{and}
1104: \bibinfo{author}{\bibfnamefont{B.~J.} \bibnamefont{Alder}},
1105: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{45}},
1106: \bibinfo{pages}{566} (\bibinfo{year}{1980}).
1107:
1108: \bibitem[{\citenamefont{Needs et~al.}(2006)\citenamefont{Needs, Towler,
1109: Drummond, and L\'opez~R\'ios}}]{06ntdl07}
1110: \bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Needs}},
1111: \bibinfo{author}{\bibfnamefont{M.~D.} \bibnamefont{Towler}},
1112: \bibinfo{author}{\bibfnamefont{N.~D.} \bibnamefont{Drummond}},
1113: \bibnamefont{and}
1114: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{L\'opez~R\'ios}},
1115: \emph{\bibinfo{title}{CASINO version 2.0 User Manual}},
1116: \bibinfo{organization}{Cambridge University} (\bibinfo{year}{2006}).
1117:
1118: \bibitem[{\citenamefont{Hyun et~al.}(1998)\citenamefont{Hyun, Hwang, Oh, Shim,
1119: and Kolomoets}}]{98hhosk06}
1120: \bibinfo{author}{\bibfnamefont{D.~B.} \bibnamefont{Hyun}},
1121: \bibinfo{author}{\bibfnamefont{J.~S.} \bibnamefont{Hwang}},
1122: \bibinfo{author}{\bibfnamefont{T.~S.} \bibnamefont{Oh}},
1123: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Shim}}, \bibnamefont{and}
1124: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Kolomoets}},
1125: \bibinfo{journal}{J. Phys. and Chem. of Solids}
1126: \textbf{\bibinfo{volume}{59}}, \bibinfo{pages}{1039} (\bibinfo{year}{1998}).
1127:
1128: \bibitem[{\citenamefont{Lee and von Allmen}(2006)}]{06la01}
1129: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Lee}} \bibnamefont{and}
1130: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{von Allmen}},
1131: \bibinfo{journal}{Applied Physics Letters} \textbf{\bibinfo{volume}{88}},
1132: \bibinfo{pages}{022107} (\bibinfo{year}{2006}).
1133:
1134: \bibitem[{\citenamefont{Noh et~al.}(2008)\citenamefont{Noh, Koh, Oh, Park, Kim,
1135: Rameau, Valla, Kidd, Johnson, Hu et~al.}}]{08nkopkrvkjhl03}
1136: \bibinfo{author}{\bibfnamefont{H.-J.} \bibnamefont{Noh}},
1137: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Koh}},
1138: \bibinfo{author}{\bibfnamefont{S.-J.} \bibnamefont{Oh}},
1139: \bibinfo{author}{\bibfnamefont{J.-H.} \bibnamefont{Park}},
1140: \bibinfo{author}{\bibfnamefont{H.-D.} \bibnamefont{Kim}},
1141: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Rameau}},
1142: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Valla}},
1143: \bibinfo{author}{\bibfnamefont{T.~E.} \bibnamefont{Kidd}},
1144: \bibinfo{author}{\bibfnamefont{P.~D.} \bibnamefont{Johnson}},
1145: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Hu}}, \bibnamefont{et~al.},
1146: \bibinfo{journal}{Europhys. Lett.} \textbf{\bibinfo{volume}{81}},
1147: \bibinfo{pages}{57006} (\bibinfo{year}{2008}).
1148:
1149: \bibitem[{\citenamefont{Nozi\'eres and Pines}(1999)}]{99np10}
1150: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Nozi\'eres}} \bibnamefont{and}
1151: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Pines}},
1152: \emph{\bibinfo{title}{The Theory of Quantum Liquids}}
1153: (\bibinfo{publisher}{Perseus Books, Cambridge, Mass.}, \bibinfo{year}{1999}).
1154:
1155: \bibitem[{\citenamefont{Sugiyama et~al.}(1992)\citenamefont{Sugiyama, Bowen,
1156: and Alder}}]{92sba11}
1157: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Sugiyama}},
1158: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Bowen}}, \bibnamefont{and}
1159: \bibinfo{author}{\bibfnamefont{B.~J.} \bibnamefont{Alder}},
1160: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{46}},
1161: \bibinfo{pages}{13042} (\bibinfo{year}{1992}).
1162:
1163: \bibitem[{\citenamefont{Moroni et~al.}(1992)\citenamefont{Moroni, Ceperley, and
1164: Senatore}}]{92mcs09}
1165: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Moroni}},
1166: \bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Ceperley}},
1167: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Senatore}},
1168: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{69}},
1169: \bibinfo{pages}{1837} (\bibinfo{year}{1992}).
1170:
1171: \bibitem[{\citenamefont{Bowen et~al.}(1994)\citenamefont{Bowen, Sugiyama, and
1172: Alder}}]{94bsa11}
1173: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Bowen}},
1174: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Sugiyama}}, \bibnamefont{and}
1175: \bibinfo{author}{\bibfnamefont{B.~J.} \bibnamefont{Alder}},
1176: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{50}},
1177: \bibinfo{pages}{14838} (\bibinfo{year}{1994}).
1178:
1179: \bibitem[{\citenamefont{Moroni et~al.}(1995)\citenamefont{Moroni, Ceperley, and
1180: Senatore}}]{95mcs07}
1181: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Moroni}},
1182: \bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Ceperley}},
1183: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Senatore}},
1184: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{75}},
1185: \bibinfo{pages}{689} (\bibinfo{year}{1995}).
1186:
1187: \bibitem[{\citenamefont{Reimann and Manninen}(2002)}]{02rm10}
1188: \bibinfo{author}{\bibfnamefont{S.~M.} \bibnamefont{Reimann}} \bibnamefont{and}
1189: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Manninen}},
1190: \bibinfo{journal}{Rev. of Modern Phys.} \textbf{\bibinfo{volume}{74}},
1191: \bibinfo{pages}{1283} (\bibinfo{year}{2002}).
1192:
1193: \bibitem[{\citenamefont{B\'anyai and Koch}(1993)}]{93bk03}
1194: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{B\'anyai}} \bibnamefont{and}
1195: \bibinfo{author}{\bibfnamefont{S.~W.} \bibnamefont{Koch}},
1196: \emph{\bibinfo{title}{Semiconductor Quantum Dots}} (\bibinfo{publisher}{World
1197: Scientific}, \bibinfo{year}{1993}).
1198:
1199: \bibitem[{\citenamefont{Pollock and Koch}(1991)}]{91pk05}
1200: \bibinfo{author}{\bibfnamefont{E.~L.} \bibnamefont{Pollock}} \bibnamefont{and}
1201: \bibinfo{author}{\bibfnamefont{S.~W.} \bibnamefont{Koch}},
1202: \bibinfo{journal}{J. of Chem. Phys.} \textbf{\bibinfo{volume}{94}},
1203: \bibinfo{pages}{6776} (\bibinfo{year}{1991}).
1204:
1205: \bibitem[{\citenamefont{Harju et~al.}(1999)\citenamefont{Harju, Sverdlov,
1206: Nieminen, and Halonen}}]{99hsnh02}
1207: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Harju}},
1208: \bibinfo{author}{\bibfnamefont{V.~A.} \bibnamefont{Sverdlov}},
1209: \bibinfo{author}{\bibfnamefont{R.~M.} \bibnamefont{Nieminen}},
1210: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Halonen}},
1211: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{59}},
1212: \bibinfo{pages}{5622} (\bibinfo{year}{1999}).
1213:
1214: \bibitem[{\citenamefont{Bolton}(1996)}]{96b08}
1215: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Bolton}},
1216: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{54}},
1217: \bibinfo{pages}{4780} (\bibinfo{year}{1996}).
1218:
1219: \bibitem[{\citenamefont{Mak et~al.}(1998)\citenamefont{Mak, Egger, and
1220: Weber-Gottschick}}]{98mew11}
1221: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Mak}},
1222: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Egger}}, \bibnamefont{and}
1223: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Weber-Gottschick}},
1224: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{81}},
1225: \bibinfo{pages}{4533} (\bibinfo{year}{1998}).
1226:
1227: \bibitem[{\citenamefont{Egger et~al.}(1999)\citenamefont{Egger, H\"ausler, Mak,
1228: and Grabert}}]{99ehmg04}
1229: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Egger}},
1230: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{H\"ausler}},
1231: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Mak}}, \bibnamefont{and}
1232: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Grabert}},
1233: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{82}},
1234: \bibinfo{pages}{3320} (\bibinfo{year}{1999}).
1235:
1236: \bibitem[{\citenamefont{Lozovik and Mandelshtam}(1992)}]{92lm06}
1237: \bibinfo{author}{\bibfnamefont{Y.~E.} \bibnamefont{Lozovik}} \bibnamefont{and}
1238: \bibinfo{author}{\bibfnamefont{V.~A.} \bibnamefont{Mandelshtam}},
1239: \bibinfo{journal}{Phys. Lett. A} \textbf{\bibinfo{volume}{165}},
1240: \bibinfo{pages}{469} (\bibinfo{year}{1992}).
1241:
1242: \bibitem[{\citenamefont{Eto}(1997)}]{97m03}
1243: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Eto}}, \bibinfo{journal}{Jpn.
1244: J. Appl. Phys.} \textbf{\bibinfo{volume}{36}}, \bibinfo{pages}{3924}
1245: (\bibinfo{year}{1997}).
1246:
1247: \bibitem[{\citenamefont{Benedict et~al.}(2003)\citenamefont{Benedict, Puzder,
1248: Williamson, Grossman, Galli, Klepeis, Raty, and Pankratov}}]{03bpwggkrp09}
1249: \bibinfo{author}{\bibfnamefont{L.~X.} \bibnamefont{Benedict}},
1250: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Puzder}},
1251: \bibinfo{author}{\bibfnamefont{A.~J.} \bibnamefont{Williamson}},
1252: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{Grossman}},
1253: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Galli}},
1254: \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Klepeis}},
1255: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Raty}}, \bibnamefont{and}
1256: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Pankratov}},
1257: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{68}},
1258: \bibinfo{pages}{085310} (\bibinfo{year}{2003}).
1259:
1260: \bibitem[{\citenamefont{Williamson et~al.}(2002)\citenamefont{Williamson,
1261: Grossman, Hood, Puzder, and Galli}}]{02wghpg11}
1262: \bibinfo{author}{\bibfnamefont{A.~J.} \bibnamefont{Williamson}},
1263: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{Grossman}},
1264: \bibinfo{author}{\bibfnamefont{R.~Q.} \bibnamefont{Hood}},
1265: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Puzder}}, \bibnamefont{and}
1266: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Galli}},
1267: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{89}},
1268: \bibinfo{pages}{196803} (\bibinfo{year}{2002}).
1269:
1270: \bibitem[{\citenamefont{Puzder et~al.}(2003)\citenamefont{Puzder, Williamson,
1271: Reboredo, and Galli}}]{03pwrg10}
1272: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Puzder}},
1273: \bibinfo{author}{\bibfnamefont{A.~J.} \bibnamefont{Williamson}},
1274: \bibinfo{author}{\bibfnamefont{F.~A.} \bibnamefont{Reboredo}},
1275: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Galli}},
1276: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{91}},
1277: \bibinfo{pages}{157405} (\bibinfo{year}{2003}).
1278:
1279: \bibitem[{\citenamefont{Pederiva et~al.}(2000)\citenamefont{Pederiva, Umrigar,
1280: and Lipparini}}]{00pul09}
1281: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Pederiva}},
1282: \bibinfo{author}{\bibfnamefont{C.~J.} \bibnamefont{Umrigar}},
1283: \bibnamefont{and}
1284: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Lipparini}},
1285: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{62}},
1286: \bibinfo{pages}{8120} (\bibinfo{year}{2000}).
1287:
1288: \bibitem[{\citenamefont{Ghosal et~al.}(2006)\citenamefont{Ghosal,
1289: G\"u\c{c}l\"u, Umrigar, Ullmo, and Baranger}}]{06gguub05}
1290: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Ghosal}},
1291: \bibinfo{author}{\bibfnamefont{A.~D.} \bibnamefont{G\"u\c{c}l\"u}},
1292: \bibinfo{author}{\bibfnamefont{C.~J.} \bibnamefont{Umrigar}},
1293: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Ullmo}}, \bibnamefont{and}
1294: \bibinfo{author}{\bibfnamefont{H.~U.} \bibnamefont{Baranger}},
1295: \bibinfo{journal}{Nature Physics} \textbf{\bibinfo{volume}{2}},
1296: \bibinfo{pages}{336} (\bibinfo{year}{2006}).
1297:
1298: \bibitem[{\citenamefont{Ferconi and Vignale}(1994)}]{94fv11}
1299: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ferconi}} \bibnamefont{and}
1300: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Vignale}},
1301: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{50}},
1302: \bibinfo{pages}{14722} (\bibinfo{year}{1994}).
1303:
1304: \bibitem[{\citenamefont{Koskinen et~al.}(1997)\citenamefont{Koskinen, Manninen,
1305: and Reimann}}]{97kmr08}
1306: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Koskinen}},
1307: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Manninen}}, \bibnamefont{and}
1308: \bibinfo{author}{\bibfnamefont{S.~M.} \bibnamefont{Reimann}},
1309: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{79}},
1310: \bibinfo{pages}{1389} (\bibinfo{year}{1997}).
1311:
1312: \bibitem[{\citenamefont{Hirose and Wingreen}(1999)}]{99hw02}
1313: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Hirose}} \bibnamefont{and}
1314: \bibinfo{author}{\bibfnamefont{N.~S.} \bibnamefont{Wingreen}},
1315: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{59}},
1316: \bibinfo{pages}{4604} (\bibinfo{year}{1999}).
1317:
1318: \bibitem[{\citenamefont{Tarucha et~al.}(1996)\citenamefont{Tarucha, Austing,
1319: Honda, van~der Hage, and Kouwenhoven}}]{96tahhk10}
1320: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Tarucha}},
1321: \bibinfo{author}{\bibfnamefont{D.~G.} \bibnamefont{Austing}},
1322: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Honda}},
1323: \bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{van~der Hage}},
1324: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{L.~P.}
1325: \bibnamefont{Kouwenhoven}}, \bibinfo{journal}{Phys. Rev. Lett.}
1326: \textbf{\bibinfo{volume}{77}}, \bibinfo{pages}{3613} (\bibinfo{year}{1996}).
1327:
1328: \bibitem[{\citenamefont{Sasaki et~al.}(1998)\citenamefont{Sasaki, Austinga, and
1329: Taruchaa}}]{98sat12}
1330: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Sasaki}},
1331: \bibinfo{author}{\bibfnamefont{D.~G.} \bibnamefont{Austinga}},
1332: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Taruchaa}},
1333: \bibinfo{journal}{Physica B} \textbf{\bibinfo{volume}{256-258}},
1334: \bibinfo{pages}{157} (\bibinfo{year}{1998}).
1335:
1336: \bibitem[{\citenamefont{Ashoori}(1996)}]{96a02}
1337: \bibinfo{author}{\bibfnamefont{R.~C.} \bibnamefont{Ashoori}},
1338: \bibinfo{journal}{Nature (London)} \textbf{\bibinfo{volume}{379}},
1339: \bibinfo{pages}{413} (\bibinfo{year}{1996}).
1340:
1341: \bibitem[{\citenamefont{Kouwenhoven et~al.}(1997)\citenamefont{Kouwenhoven,
1342: Oosterkamp, Danoesastro, Eto, Austing, Honda, and Tarucha}}]{97kodeaht12}
1343: \bibinfo{author}{\bibfnamefont{L.~P.} \bibnamefont{Kouwenhoven}},
1344: \bibinfo{author}{\bibfnamefont{T.~H.} \bibnamefont{Oosterkamp}},
1345: \bibinfo{author}{\bibfnamefont{M.~W.~S.} \bibnamefont{Danoesastro}},
1346: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Eto}},
1347: \bibinfo{author}{\bibfnamefont{D.~G.} \bibnamefont{Austing}},
1348: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Honda}}, \bibnamefont{and}
1349: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Tarucha}},
1350: \bibinfo{journal}{Science} \textbf{\bibinfo{volume}{278}},
1351: \bibinfo{pages}{1788} (\bibinfo{year}{1997}).
1352:
1353: \bibitem[{\citenamefont{Reusch}(2003)}]{03r03}
1354: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Reusch}}, Ph.D. thesis,
1355: \bibinfo{school}{Heinrich-Heine-Universit\"at D\"usselforf}
1356: (\bibinfo{year}{2003}).
1357:
1358: \bibitem[{\citenamefont{Hine}()}]{05h01}
1359: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Hine}}, \bibinfo{note}{computer
1360: code DOTDFT, 2005.}
1361:
1362: \bibitem[{\citenamefont{Hern\'andez et~al.}(1997)\citenamefont{Hern\'andez,
1363: Gillan, and Goringe}}]{97hgg05}
1364: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Hern\'andez}},
1365: \bibinfo{author}{\bibfnamefont{M.~J.} \bibnamefont{Gillan}},
1366: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{C.~M.}
1367: \bibnamefont{Goringe}}, \bibinfo{journal}{Phys. Rev. B}
1368: \textbf{\bibinfo{volume}{55}}, \bibinfo{pages}{13485} (\bibinfo{year}{1997}).
1369:
1370: \bibitem[{\citenamefont{Alf\`e and Gillan}(2004)}]{04ag10}
1371: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Alf\`e}} \bibnamefont{and}
1372: \bibinfo{author}{\bibfnamefont{M.~J.} \bibnamefont{Gillan}},
1373: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{70}},
1374: \bibinfo{pages}{161101(R)} (\bibinfo{year}{2004}).
1375:
1376: \bibitem[{\citenamefont{Dreizler and Gross}(2001)}]{01dg01}
1377: \bibinfo{author}{\bibfnamefont{R.~M.} \bibnamefont{Dreizler}} \bibnamefont{and}
1378: \bibinfo{author}{\bibfnamefont{E.~K.~U.} \bibnamefont{Gross}},
1379: \emph{\bibinfo{title}{Density Functional Theory: An Approach to the Quantum
1380: Many-Body Problem}} (\bibinfo{publisher}{Springer-Verlag},
1381: \bibinfo{year}{2001}).
1382:
1383: \bibitem[{\citenamefont{Perdew and Zunger}(1981)}]{81pz05}
1384: \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Perdew}} \bibnamefont{and}
1385: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Zunger}},
1386: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{23}},
1387: \bibinfo{pages}{5048} (\bibinfo{year}{1981}).
1388:
1389: \bibitem[{\citenamefont{Kalliakos et~al.}(2008)\citenamefont{Kalliakos,
1390: Rontani, Pellegrini, Garc\u{\i}a, Pinczuk, Goldoni, Molinari, Pfeiffer, and
1391: West}}]{08krpgpgmpw06}
1392: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Kalliakos}},
1393: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Rontani}},
1394: \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Pellegrini}},
1395: \bibinfo{author}{\bibfnamefont{C.~P.} \bibnamefont{Garc\u{\i}a}},
1396: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Pinczuk}},
1397: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Goldoni}},
1398: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Molinari}},
1399: \bibinfo{author}{\bibfnamefont{L.~N.} \bibnamefont{Pfeiffer}},
1400: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{K.~W.} \bibnamefont{West}},
1401: \bibinfo{journal}{Nature Physics} \textbf{\bibinfo{volume}{4}},
1402: \bibinfo{pages}{467} (\bibinfo{year}{2008}).
1403:
1404: \bibitem[{\citenamefont{Giuliani and Vignale}(2005)}]{05gv04}
1405: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Giuliani}} \bibnamefont{and}
1406: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Vignale}},
1407: \emph{\bibinfo{title}{Quantum Theory of the Electron Liquid}}
1408: (\bibinfo{publisher}{Cambridge University Press}, \bibinfo{year}{2005}).
1409:
1410: \bibitem[{\citenamefont{Lozovik and Mandelshtam}(1990)}]{90lm04}
1411: \bibinfo{author}{\bibfnamefont{Y.~E.} \bibnamefont{Lozovik}} \bibnamefont{and}
1412: \bibinfo{author}{\bibfnamefont{V.~A.} \bibnamefont{Mandelshtam}},
1413: \bibinfo{journal}{Phys. Lett. A} \textbf{\bibinfo{volume}{145}},
1414: \bibinfo{pages}{269} (\bibinfo{year}{1990}).
1415:
1416: \bibitem[{\citenamefont{Bolton and R\"ossler}(1993)}]{93br03}
1417: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Bolton}} \bibnamefont{and}
1418: \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{R\"ossler}},
1419: \bibinfo{journal}{Superlattices and Microstructures}
1420: \textbf{\bibinfo{volume}{13}}, \bibinfo{pages}{139} (\bibinfo{year}{1993}).
1421:
1422: \bibitem[{\citenamefont{Date et~al.}(1998)\citenamefont{Date, Murthy, and
1423: Vathsan}}]{98mvn07}
1424: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Date}},
1425: \bibinfo{author}{\bibfnamefont{M.~V.~N.} \bibnamefont{Murthy}},
1426: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Vathsan}},
1427: \bibinfo{journal}{J. Phys.: Condens. Matter} \textbf{\bibinfo{volume}{10}},
1428: \bibinfo{pages}{5875} (\bibinfo{year}{1998}).
1429:
1430: \bibitem[{\citenamefont{Shiau et~al.}(2007)\citenamefont{Shiau, Chutia, and
1431: Joynt}}]{07scj05}
1432: \bibinfo{author}{\bibfnamefont{S.~Y.} \bibnamefont{Shiau}},
1433: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Chutia}}, \bibnamefont{and}
1434: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Joynt}},
1435: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{75}},
1436: \bibinfo{pages}{195345} (\bibinfo{year}{2007}).
1437:
1438: \bibitem[{\citenamefont{Shiau and Joynt}(2007)}]{07sj11}
1439: \bibinfo{author}{\bibfnamefont{S.~Y.} \bibnamefont{Shiau}} \bibnamefont{and}
1440: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Joynt}},
1441: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{76}},
1442: \bibinfo{pages}{205314} (\bibinfo{year}{2007}).
1443:
1444: \bibitem[{\citenamefont{Hada and Eto}(2003)}]{03he10}
1445: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Hada}} \bibnamefont{and}
1446: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Eto}},
1447: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{68}},
1448: \bibinfo{pages}{155322} (\bibinfo{year}{2003}).
1449:
1450: \bibitem[{\citenamefont{Ahn}(2005)}]{05a08}
1451: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Ahn}}, \bibinfo{journal}{J.
1452: Appl. Phys.} \textbf{\bibinfo{volume}{98}}, \bibinfo{pages}{033709}
1453: (\bibinfo{year}{2005}).
1454:
1455: \bibitem[{\citenamefont{Belyakov and Burdov}(2007)}]{07bb07}
1456: \bibinfo{author}{\bibfnamefont{V.~A.} \bibnamefont{Belyakov}} \bibnamefont{and}
1457: \bibinfo{author}{\bibfnamefont{V.~A.} \bibnamefont{Burdov}},
1458: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{76}},
1459: \bibinfo{pages}{045335} (\bibinfo{year}{2007}).
1460:
1461: \bibitem[{\citenamefont{Yannouleas and Landman}(2007)}]{07yl11}
1462: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Yannouleas}} \bibnamefont{and}
1463: \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{Landman}},
1464: \bibinfo{journal}{Rep. Prog. Phys} \textbf{\bibinfo{volume}{70}},
1465: \bibinfo{pages}{2067} (\bibinfo{year}{2007}).
1466:
1467: \end{thebibliography}
1468:
1469:
1470: \end{document}
1471:
1472: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1473: