ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/multipole/multipole1.tex
Revision: 4213
Committed: Mon Aug 18 15:15:15 2014 UTC (10 years ago) by gezelter
Content type: application/x-tex
File size: 73113 byte(s)
Log Message:
Spelling errors

File Contents

# User Rev Content
1 gezelter 3906 % ****** Start of file aipsamp.tex ******
2     %
3     % This file is part of the AIP files in the AIP distribution for REVTeX 4.
4     % Version 4.1 of REVTeX, October 2009
5     %
6     % Copyright (c) 2009 American Institute of Physics.
7     %
8     % See the AIP README file for restrictions and more information.
9     %
10     % TeX'ing this file requires that you have AMS-LaTeX 2.0 installed
11     % as well as the rest of the prerequisites for REVTeX 4.1
12     %
13     % It also requires running BibTeX. The commands are as follows:
14     %
15     % 1) latex aipsamp
16     % 2) bibtex aipsamp
17     % 3) latex aipsamp
18     % 4) latex aipsamp
19     %
20     % Use this file as a source of example code for your aip document.
21     % Use the file aiptemplate.tex as a template for your document.
22     \documentclass[%
23 gezelter 3980 aip,jcp,
24 gezelter 3906 amsmath,amssymb,
25     preprint,%
26 gezelter 4172 % reprint,%
27 gezelter 3906 %author-year,%
28     %author-numerical,%
29 gezelter 3984 jcp]{revtex4-1}
30 gezelter 3906
31     \usepackage{graphicx}% Include figure files
32     \usepackage{dcolumn}% Align table columns on decimal point
33 gezelter 3980 %\usepackage{bm}% bold math
34 gezelter 3982 \usepackage{times}
35 gezelter 3980 \usepackage[version=3]{mhchem} % this is a great package for formatting chemical reactions
36     \usepackage{url}
37 gezelter 3985 \usepackage{rotating}
38 gezelter 4172
39 gezelter 3906 %\usepackage[mathlines]{lineno}% Enable numbering of text and display math
40     %\linenumbers\relax % Commence numbering lines
41    
42     \begin{document}
43    
44 gezelter 3988 %\preprint{AIP/123-QED}
45 gezelter 3906
46 gezelter 4198 \title{Real space electrostatics for multipoles. I. Development of methods}
47 gezelter 3906
48     \author{Madan Lamichhane}
49     \affiliation{Department of Physics, University
50     of Notre Dame, Notre Dame, IN 46556}
51    
52     \author{J. Daniel Gezelter}
53     \email{gezelter@nd.edu.}
54     \affiliation{Department of Chemistry and Biochemistry, University
55     of Notre Dame, Notre Dame, IN 46556}
56    
57     \author{Kathie E. Newman}
58     \affiliation{Department of Physics, University
59     of Notre Dame, Notre Dame, IN 46556}
60    
61    
62     \date{\today}% It is always \today, today,
63     % but any date may be explicitly specified
64    
65     \begin{abstract}
66 gezelter 3980 We have extended the original damped-shifted force (DSF)
67 gezelter 4179 electrostatic kernel and have been able to derive three new
68 gezelter 3980 electrostatic potentials for higher-order multipoles that are based
69 gezelter 4179 on truncated Taylor expansions around the cutoff radius. These
70     include a shifted potential (SP) that generalizes the Wolf method
71     for point multipoles, and Taylor-shifted force (TSF) and
72     gradient-shifted force (GSF) potentials that are both
73     generalizations of DSF electrostatics for multipoles. We find that
74     each of the distinct orientational contributions requires a separate
75     radial function to ensure that pairwise energies, forces and torques
76     all vanish at the cutoff radius. In this paper, we present energy,
77     force, and torque expressions for the new models, and compare these
78     real-space interaction models to exact results for ordered arrays of
79     multipoles. We find that the GSF and SP methods converge rapidly to
80     the correct lattice energies for ordered dipolar and quadrupolar
81     arrays, while the Taylor-Shifted Force (TSF) is too severe an
82     approximation to provide accurate convergence to lattice energies.
83     Because real-space methods can be made to scale linearly with system
84 gezelter 4183 size, SP and GSF are attractive options for large Monte
85     Carlo and molecular dynamics simulations, respectively.
86 gezelter 3906 \end{abstract}
87    
88 gezelter 3988 %\pacs{Valid PACS appear here}% PACS, the Physics and Astronomy
89 gezelter 3906 % Classification Scheme.
90 gezelter 3988 %\keywords{Suggested keywords}%Use showkeys class option if keyword
91 gezelter 3906 %display desired
92     \maketitle
93    
94     \section{Introduction}
95 gezelter 3982 There has been increasing interest in real-space methods for
96     calculating electrostatic interactions in computer simulations of
97     condensed molecular
98 gezelter 3980 systems.\cite{Wolf99,Zahn02,Kast03,BeckD.A.C._bi0486381,Ma05,Fennell:2006zl,Chen:2004du,Chen:2006ii,Rodgers:2006nw,Denesyuk:2008ez,Izvekov:2008wo}
99     The simplest of these techniques was developed by Wolf {\it et al.}
100     in their work towards an $\mathcal{O}(N)$ Coulombic sum.\cite{Wolf99}
101 gezelter 4200 For systems of point charges, Zahn {\it et al.} extended the Wolf
102     approach by adding a force-shifting term to the enhanced damped
103     Coulomb potential.\cite{Zahn02,Kast03} Fennell and Gezelter showed
104     that the simple damped shifted force (DSF) potential can give nearly
105     quantitative agreement with smooth particle mesh Ewald
106 gezelter 3982 (SPME)\cite{Essmann95} configurational energy differences as well as
107     atomic force and molecular torque vectors.\cite{Fennell:2006zl}
108 gezelter 3906
109 gezelter 3980 The computational efficiency and the accuracy of the DSF method are
110     surprisingly good, particularly for systems with uniform charge
111     density. Additionally, dielectric constants obtained using DSF and
112 gezelter 3986 similar methods where the force vanishes at $r_{c}$ are
113 gezelter 3980 essentially quantitative.\cite{Izvekov:2008wo} The DSF and other
114     related methods have now been widely investigated,\cite{Hansen:2012uq}
115 gezelter 3985 and DSF is now used routinely in a diverse set of chemical
116     environments.\cite{doi:10.1021/la400226g,McCann:2013fk,kannam:094701,Forrest:2012ly,English:2008kx,Louden:2013ve,Tokumasu:2013zr}
117     DSF electrostatics provides a compromise between the computational
118     speed of real-space cutoffs and the accuracy of fully-periodic Ewald
119     treatments.
120 gezelter 3906
121 gezelter 3980 One common feature of many coarse-graining approaches, which treat
122     entire molecular subsystems as a single rigid body, is simplification
123     of the electrostatic interactions between these bodies so that fewer
124     site-site interactions are required to compute configurational
125 gezelter 3986 energies. To do this, the interactions between coarse-grained sites
126     are typically taken to be point
127     multipoles.\cite{Golubkov06,ISI:000276097500009,ISI:000298664400012}
128 gezelter 3906
129 gezelter 3986 Water, in particular, has been modeled recently with point multipoles
130     up to octupolar
131     order.\cite{Chowdhuri:2006lr,Te:2010rt,Te:2010ys,Te:2010vn} For
132     maximum efficiency, these models require the use of an approximate
133     multipole expansion as the exact multipole expansion can become quite
134     expensive (particularly when handled via the Ewald
135     sum).\cite{Ichiye:2006qy} Point multipoles and multipole
136     polarizability have also been utilized in the AMOEBA water model and
137 gezelter 3980 related force fields.\cite{Ponder:2010fk,schnieders:124114,Ren:2011uq}
138 gezelter 3906
139 gezelter 3980 Higher-order multipoles present a peculiar issue for molecular
140     dynamics. Multipolar interactions are inherently short-ranged, and
141     should not need the relatively expensive Ewald treatment. However,
142     real-space cutoff methods are normally applied in an orientation-blind
143     fashion so multipoles which leave and then re-enter a cutoff sphere in
144     a different orientation can cause energy discontinuities.
145 gezelter 3906
146 gezelter 3980 This paper outlines an extension of the original DSF electrostatic
147 gezelter 4183 kernel to point multipoles. We describe three distinct real-space
148     interaction models for higher-order multipoles based on truncated
149 gezelter 3982 Taylor expansions that are carried out at the cutoff radius. We are
150 gezelter 4183 calling these models {\bf Taylor-shifted} (TSF), {\bf
151     gradient-shifted} (GSF) and {\bf shifted potential} (SP)
152 gezelter 3982 electrostatics. Because of differences in the initial assumptions,
153 gezelter 4198 the two methods yield related, but distinct expressions for energies,
154 gezelter 4183 forces, and torques.
155 gezelter 3906
156 gezelter 3982 In this paper we outline the new methodology and give functional forms
157     for the energies, forces, and torques up to quadrupole-quadrupole
158     order. We also compare the new methods to analytic energy constants
159 gezelter 3986 for periodic arrays of point multipoles. In the following paper, we
160 gezelter 3985 provide numerical comparisons to Ewald-based electrostatics in common
161 gezelter 4213 simulation environments.
162 gezelter 3982
163 gezelter 3980 \section{Methodology}
164 gezelter 3986 An efficient real-space electrostatic method involves the use of a
165     pair-wise functional form,
166     \begin{equation}
167 gezelter 4198 U = \sum_i \sum_{j>i} U_\mathrm{pair}(\mathbf{r}_{ij}, \mathsf{A}_i, \mathsf{B}_j) +
168 gezelter 4175 \sum_i U_i^\mathrm{self}
169 gezelter 3986 \end{equation}
170     that is short-ranged and easily truncated at a cutoff radius,
171     \begin{equation}
172 gezelter 4198 U_\mathrm{pair}(\mathbf{r}_{ij},\mathsf{A}_i, \mathsf{B}_j) = \left\{
173 gezelter 3986 \begin{array}{ll}
174 gezelter 4198 U_\mathrm{approx} (\mathbf{r}_{ij}, \mathsf{A}_i, \mathsf{B}_j) & \quad \left| \mathbf{r}_{ij} \right| \le r_c \\
175 gezelter 4098 0 & \quad \left| \mathbf{r}_{ij} \right| > r_c ,
176 gezelter 3986 \end{array}
177     \right.
178     \end{equation}
179 gezelter 3989 along with an easily computed self-interaction term ($\sum_i
180 gezelter 4175 U_i^\mathrm{self}$) which scales linearly with the number of
181 gezelter 4198 particles. Here $\mathsf{A}_i$ and $\mathsf{B}_j$ represent orientational
182     coordinates of the two sites (expressed as rotation matrices), and $\mathbf{r}_{ij}$ is the vector
183 gezelter 4098 between the two sites. The computational efficiency, energy
184 gezelter 3986 conservation, and even some physical properties of a simulation can
185 gezelter 4175 depend dramatically on how the $U_\mathrm{approx}$ function behaves at
186 gezelter 3986 the cutoff radius. The goal of any approximation method should be to
187     mimic the real behavior of the electrostatic interactions as closely
188     as possible without sacrificing the near-linear scaling of a cutoff
189     method.
190 gezelter 3906
191 gezelter 3980 \subsection{Self-neutralization, damping, and force-shifting}
192     The DSF and Wolf methods operate by neutralizing the total charge
193     contained within the cutoff sphere surrounding each particle. This is
194     accomplished by shifting the potential functions to generate image
195     charges on the surface of the cutoff sphere for each pair interaction
196 gezelter 4098 computed within $r_c$. Damping using a complementary error function is
197     applied to the potential to accelerate convergence. The interaction
198     for a pair of charges ($C_i$ and $C_j$) in the DSF method,
199 gezelter 3980 \begin{equation*}
200 gezelter 4175 U_\mathrm{DSF}(r_{ij}) = C_i C_j \Biggr{[}
201 gezelter 3980 \frac{\mathrm{erfc}\left(\alpha r_{ij}\right)}{r_{ij}}
202 gezelter 3986 - \frac{\mathrm{erfc}\left(\alpha r_c\right)}{r_c} + \left(\frac{\mathrm{erfc}\left(\alpha r_c\right)}{r_c^2}
203 gezelter 3980 + \frac{2\alpha}{\pi^{1/2}}
204 gezelter 3986 \frac{\exp\left(-\alpha^2r_c^2\right)}{r_c}
205     \right)\left(r_{ij}-r_c\right)\ \Biggr{]}
206 gezelter 3980 \label{eq:DSFPot}
207     \end{equation*}
208 gezelter 4098 where $\alpha$ is the adjustable damping parameter. Note that in this
209     potential and in all electrostatic quantities that follow, the
210     standard $1/4 \pi \epsilon_{0}$ has been omitted for clarity.
211 gezelter 3980
212     To insure net charge neutrality within each cutoff sphere, an
213     additional ``self'' term is added to the potential. This term is
214     constant (as long as the charges and cutoff radius do not change), and
215     exists outside the normal pair-loop for molecular simulations. It can
216     be thought of as a contribution from a charge opposite in sign, but
217     equal in magnitude, to the central charge, which has been spread out
218     over the surface of the cutoff sphere. A portion of the self term is
219     identical to the self term in the Ewald summation, and comes from the
220     utilization of the complimentary error function for electrostatic
221 gezelter 3986 damping.\cite{deLeeuw80,Wolf99} There have also been recent efforts to
222     extend the Wolf self-neutralization method to zero out the dipole and
223     higher order multipoles contained within the cutoff
224 gezelter 3985 sphere.\cite{Fukuda:2011jk,Fukuda:2012yu,Fukuda:2013qv}
225 gezelter 3982
226 gezelter 3985 In this work, we extend the idea of self-neutralization for the point
227     multipoles by insuring net charge-neutrality and net-zero moments
228 gezelter 4198 within each cutoff sphere. In Figure \ref{fig:shiftedMultipoles}, point dipole
229     $\mathbf{D}_i$ at the center site $i$ is interacting with point dipole
230     $\mathbf{D}_j$ and point quadrupole $\mathsf{Q}_k$. The
231 gezelter 3985 self-neutralization scheme for point multipoles involves projecting
232     opposing multipoles for sites $j$ and $k$ on the surface of the cutoff
233     sphere. There are also significant modifications made to make the
234     forces and torques go smoothly to zero at the cutoff distance.
235 gezelter 3982
236 gezelter 4172 \begin{figure}
237 gezelter 4189 \includegraphics[width=3in]{SM.eps}
238 gezelter 4172 \caption{Reversed multipoles are projected onto the surface of the
239     cutoff sphere. The forces, torques, and potential are then smoothly
240     shifted to zero as the sites leave the cutoff region.}
241     \label{fig:shiftedMultipoles}
242     \end{figure}
243 gezelter 3980
244 gezelter 3986 As in the point-charge approach, there is an additional contribution
245     from self-neutralization of site $i$. The self term for multipoles is
246 gezelter 3982 described in section \ref{sec:selfTerm}.
247 gezelter 3906
248 gezelter 3982 \subsection{The multipole expansion}
249    
250 gezelter 4198 Consider two discrete rigid collections of point charges, denoted as objects
251     $a$ and $b$. In the following, we assume that the two objects
252 gezelter 3982 interact via electrostatics only and describe those interactions in
253     terms of a standard multipole expansion. Putting the origin of the
254 gezelter 4198 coordinate system at the center of mass of $a$, we use vectors
255     $\mathbf{r}_k$ to denote the positions of all charges $q_k$ in
256     $a$. Then the electrostatic potential of object $a$ at
257 gezelter 3980 $\mathbf{r}$ is given by
258 gezelter 3906 \begin{equation}
259 gezelter 4175 \phi_a(\mathbf r) =
260 gezelter 4198 \sum_{k \, \text{in } a} \frac{q_k}{\lvert \mathbf{r} - \mathbf{r}_k \rvert}.
261 gezelter 3906 \end{equation}
262 gezelter 3982 The Taylor expansion in $r$ can be written using an implied summation
263     notation. Here Greek indices are used to indicate space coordinates
264     ($x$, $y$, $z$) and the subscripts $k$ and $j$ are reserved for
265 gezelter 4198 labeling specific sites for charges in $a$ and $b$ respectively. The
266 gezelter 3982 Taylor expansion,
267 gezelter 3906 \begin{equation}
268     \frac{1}{\lvert \mathbf{r} - \mathbf{r}_k \rvert} =
269     \left( 1
270     - r_{k\alpha} \frac{\partial}{\partial r_{\alpha}}
271     + \frac{1}{2} r_{k\alpha} r_{k\beta} \frac{\partial^2}{\partial r_{\alpha} \partial r_{\beta}} +\dots
272     \right)
273 gezelter 3982 \frac{1}{r} ,
274 gezelter 3906 \end{equation}
275 gezelter 4198 can then be used to express the electrostatic potential on $a$ in
276 gezelter 3982 terms of multipole operators,
277 gezelter 3906 \begin{equation}
278 gezelter 4198 \phi_a(\mathbf{r}) =M_a \frac{1}{r}
279 gezelter 3906 \end{equation}
280     where
281     \begin{equation}
282 gezelter 4198 M_a = C_a - D_{a\alpha} \frac{\partial}{\partial r_{\alpha}}
283     + Q_{a\alpha\beta}
284 gezelter 3906 \frac{\partial^2}{\partial r_{\alpha} \partial r_{\beta}} + \dots
285     \end{equation}
286 gezelter 4198 Here, the point charge, dipole, and quadrupole for object $a$ are
287     given by $C_a$, $\mathbf{D}_a$, and $\mathsf{Q}_a$, respectively. These are the primitive multipoles
288 gezelter 3982 which can be expressed as a distribution of charges,
289     \begin{align}
290 gezelter 4198 C_a =&\sum_{k \, \text{in }a} q_k , \label{eq:charge} \\
291     D_{a\alpha} =&\sum_{k \, \text{in }a} q_k r_{k\alpha}, \label{eq:dipole}\\
292     Q_{a\alpha\beta} =& \frac{1}{2} \sum_{k \, \text{in } a} q_k
293 gezelter 3996 r_{k\alpha} r_{k\beta} . \label{eq:quadrupole}
294 gezelter 3982 \end{align}
295     Note that the definition of the primitive quadrupole here differs from
296     the standard traceless form, and contains an additional Taylor-series
297 gezelter 3996 based factor of $1/2$. We are essentially treating the mass
298     distribution with higher priority; the moment of inertia tensor,
299 gezelter 4198 $\mathsf I$, is diagonalized to obtain body-fixed
300 gezelter 3996 axes, and the charge distribution may result in a quadrupole tensor
301     that is not necessarily diagonal in the body frame. Additional
302     reasons for utilizing the primitive quadrupole are discussed in
303     section \ref{sec:damped}.
304 gezelter 3906
305 gezelter 4198 It is convenient to locate charges $q_j$ relative to the center of mass of $b$. Then with $\bf{r}$ pointing from
306     the center of mass of $a$ to the center of mass of $b$ ($\mathbf{r}=\mathbf{r}_b - \mathbf{r}_a $), the interaction energy is given by
307 gezelter 3906 \begin{equation}
308 gezelter 4198 U_{ab}(r)
309     = M_a \sum_{j \, \text{in } b} \frac {q_j}{\vert \mathbf{r}+\mathbf{r}_j \vert} .
310 gezelter 3982 \end{equation}
311     This can also be expanded as a Taylor series in $r$. Using a notation
312 gezelter 4198 similar to before to define the multipoles in object $b$,
313 gezelter 3982 \begin{equation}
314 gezelter 4198 M_b = C_b + D_{b\alpha} \frac{\partial}{\partial r_{\alpha}}
315     + Q_{b\alpha\beta}
316 gezelter 3906 \frac{\partial^2}{\partial r_{\alpha} \partial r_{\beta}} + \dots
317     \end{equation}
318 gezelter 3982 we arrive at the multipole expression for the total interaction energy.
319 gezelter 3906 \begin{equation}
320 gezelter 4198 U_{ab}(r)=M_a M_b \frac{1}{r} \label{kernel}.
321 gezelter 3906 \end{equation}
322 gezelter 3982 This form has the benefit of separating out the energies of
323     interaction into contributions from the charge, dipole, and quadrupole
324 gezelter 4198 of $a$ interacting with the same types of multipoles in $b$.
325 gezelter 3906
326 gezelter 3982 \subsection{Damped Coulomb interactions}
327 gezelter 3996 \label{sec:damped}
328 gezelter 3982 In the standard multipole expansion, one typically uses the bare
329     Coulomb potential, with radial dependence $1/r$, as shown in
330     Eq.~(\ref{kernel}). It is also quite common to use a damped Coulomb
331     interaction, which results from replacing point charges with Gaussian
332     distributions of charge with width $\alpha$. In damped multipole
333     electrostatics, the kernel ($1/r$) of the expansion is replaced with
334     the function:
335 gezelter 3906 \begin{equation}
336     B_0(r)=\frac{\text{erfc}(\alpha r)}{r} = \frac{2}{\sqrt{\pi}r}
337     \int_{\alpha r}^{\infty} \text{e}^{-s^2} ds .
338     \end{equation}
339 gezelter 3982 We develop equations below using the function $f(r)$ to represent
340 gezelter 3986 either $1/r$ or $B_0(r)$, and all of the techniques can be applied to
341     bare or damped Coulomb kernels (or any other function) as long as
342     derivatives of these functions are known. Smith's convenient
343 gezelter 4179 functions $B_l(r)$, which are used for derivatives of the damped
344     kernel, are summarized in Appendix A. (N.B. there is one important
345     distinction between the two kernels, which is the behavior of
346     $\nabla^2 \frac{1}{r}$ compared with $\nabla^2 B_0(r)$. The former is
347     zero everywhere except for a delta function evaluated at the origin.
348     The latter also has delta function behavior, but is non-zero for $r
349     \neq 0$. Thus the standard justification for using a traceless
350 gezelter 3996 quadrupole tensor fails for the damped case.)
351 gezelter 3906
352 gezelter 3982 The main goal of this work is to smoothly cut off the interaction
353     energy as well as forces and torques as $r\rightarrow r_c$. To
354     describe how this goal may be met, we use two examples, charge-charge
355 gezelter 3986 and charge-dipole, using the bare Coulomb kernel, $f(r)=1/r$, to
356     explain the idea.
357 gezelter 3906
358 gezelter 3984 \subsection{Shifted-force methods}
359 gezelter 3982 In the shifted-force approximation, the interaction energy for two
360 gezelter 4198 charges $C_a$ and $C_b$ separated by a distance $r$ is
361 gezelter 3982 written:
362 gezelter 3906 \begin{equation}
363 gezelter 4198 U_{C_aC_b}(r)= C_a C_b
364 gezelter 3906 \left({ \frac{1}{r} - \frac{1}{r_c} + (r - r_c) \frac{1}{r_c^2} }
365     \right) .
366     \end{equation}
367 gezelter 3982 Two shifting terms appear in this equations, one from the
368 gezelter 3984 neutralization procedure ($-1/r_c$), and one that causes the first
369     derivative to vanish at the cutoff radius.
370 gezelter 3982
371     Since one derivative of the interaction energy is needed for the
372     force, the minimal perturbation is a term linear in $(r-r_c)$ in the
373     interaction energy, that is:
374 gezelter 3906 \begin{equation}
375     \frac{d\,}{dr}
376     \left( {\frac{1}{r} - \frac{1}{r_c} + (r - r_c) \frac{1}{r_c^2} }
377     \right) = \left(- \frac{1}{r^2} + \frac{1}{r_c^2}
378     \right) .
379     \end{equation}
380 gezelter 3985 which clearly vanishes as the $r$ approaches the cutoff radius. There
381     are a number of ways to generalize this derivative shift for
382 gezelter 3984 higher-order multipoles. Below, we present two methods, one based on
383     higher-order Taylor series for $r$ near $r_c$, and the other based on
384     linear shift of the kernel gradients at the cutoff itself.
385 gezelter 3906
386 gezelter 3984 \subsection{Taylor-shifted force (TSF) electrostatics}
387 gezelter 3982 In the Taylor-shifted force (TSF) method, the procedure that we follow
388     is based on a Taylor expansion containing the same number of
389     derivatives required for each force term to vanish at the cutoff. For
390     example, the quadrupole-quadrupole interaction energy requires four
391     derivatives of the kernel, and the force requires one additional
392 gezelter 3986 derivative. For quadrupole-quadrupole interactions, we therefore
393     require shifted energy expressions that include up to $(r-r_c)^5$ so
394     that all energies, forces, and torques are zero as $r \rightarrow
395     r_c$. In each case, we subtract off a function $f_n^{\text{shift}}(r)$
396     from the kernel $f(r)=1/r$. The subscript $n$ indicates the number of
397     derivatives to be taken when deriving a given multipole energy. We
398     choose a function with guaranteed smooth derivatives -- a truncated
399     Taylor series of the function $f(r)$, e.g.,
400 gezelter 3906 %
401     \begin{equation}
402 gezelter 3984 f_n^{\text{shift}}(r)=\sum_{m=0}^{n+1} \frac {(r-r_c)^m}{m!} f^{(m)}(r_c) .
403 gezelter 3906 \end{equation}
404     %
405     The combination of $f(r)$ with the shifted function is denoted $f_n(r)=f(r)-f_n^{\text{shift}}(r)$.
406     Thus, for $f(r)=1/r$, we find
407     %
408     \begin{equation}
409     f_1(r)=\frac{1}{r}- \frac{1}{r_c} + (r - r_c) \frac{1}{r_c^2} - \frac{(r-r_c)^2}{r_c^3} .
410     \end{equation}
411     %
412 gezelter 4198 Continuing with the example of a charge $a$ interacting with a
413     dipole $b$, we write
414 gezelter 3906 %
415     \begin{equation}
416 gezelter 4198 U_{C_a\mathbf{D}_b}(r)=
417     C_a D_{b\alpha} \frac {\partial f_1(r) }{\partial r_\alpha}
418     = C_a D_{b\alpha}
419 gezelter 3906 \frac {r_\alpha}{r} \frac {\partial f_1(r)}{\partial r} .
420     \end{equation}
421     %
422 gezelter 4198 The force that dipole $ b$ exerts on charge $a$ is
423 gezelter 3906 %
424     \begin{equation}
425 gezelter 4198 F_{C_a \mathbf{D}_b \beta} = C_a D_{b \alpha}
426 gezelter 3906 \left[ \frac{\delta_{\alpha\beta}}{r} \frac {\partial}{\partial r} +
427     \frac{r_\alpha r_\beta}{r^2}
428     \left( -\frac{1}{r} \frac {\partial} {\partial r}
429     + \frac {\partial ^2} {\partial r^2} \right) \right] f_1(r) .
430     \end{equation}
431     %
432 gezelter 3984 For undamped coulombic interactions, $f(r)=1/r$, we find
433 gezelter 3906 %
434     \begin{equation}
435 gezelter 4198 F_{C_a \mathbf{D}_b \beta} =
436     \frac{C_a D_{b\beta}}{r}
437 gezelter 3906 \left[ -\frac{1}{r^2}+\frac{1}{r_c^2}-\frac{2(r-r_c)}{r_c^3} \right]
438 gezelter 4198 +C_a D_{b \alpha}r_\alpha r_\beta
439 gezelter 3906 \left[ \frac{3}{r^5}-\frac{3}{r^3r_c^2} \right] .
440     \end{equation}
441     %
442     This expansion shows the expected $1/r^3$ dependence of the force.
443    
444 gezelter 3984 In general, we can write
445 gezelter 3906 %
446     \begin{equation}
447 gezelter 4098 U^{\text{TSF}}= (\text{prefactor}) (\text{derivatives}) f_n(r)
448 gezelter 3906 \label{generic}
449     \end{equation}
450     %
451 gezelter 3985 with $n=0$ for charge-charge, $n=1$ for charge-dipole, $n=2$ for
452     charge-quadrupole and dipole-dipole, $n=3$ for dipole-quadrupole, and
453     $n=4$ for quadrupole-quadrupole. For example, in
454     quadrupole-quadrupole interactions for which the $\text{prefactor}$ is
455 gezelter 4198 $Q_{a \alpha\beta}Q_{b \gamma\delta}$, the derivatives are
456 gezelter 3985 $\partial^4/\partial r_\alpha \partial r_\beta \partial
457     r_\gamma \partial r_\delta$, with implied summation combining the
458 gezelter 4183 space indices. Appendix \ref{radialTSF} contains details on the
459     radial functions.
460 gezelter 3906
461 gezelter 3984 In the formulas presented in the tables below, the placeholder
462     function $f(r)$ is used to represent the electrostatic kernel (either
463     damped or undamped). The main functions that go into the force and
464 gezelter 3985 torque terms, $g_n(r), h_n(r), s_n(r), \mathrm{~and~} t_n(r)$ are
465     successive derivatives of the shifted electrostatic kernel, $f_n(r)$
466     of the same index $n$. The algebra required to evaluate energies,
467     forces and torques is somewhat tedious, so only the final forms are
468 gezelter 3986 presented in tables \ref{tab:tableenergy} and \ref{tab:tableFORCE}.
469 gezelter 3996 One of the principal findings of our work is that the individual
470     orientational contributions to the various multipole-multipole
471     interactions must be treated with distinct radial functions, but each
472     of these contributions is independently force shifted at the cutoff
473     radius.
474 gezelter 3906
475 gezelter 3982 \subsection{Gradient-shifted force (GSF) electrostatics}
476 gezelter 3985 The second, and conceptually simpler approach to force-shifting
477     maintains only the linear $(r-r_c)$ term in the truncated Taylor
478     expansion, and has a similar interaction energy for all multipole
479     orders:
480 gezelter 3906 \begin{equation}
481 gezelter 4198 U^{\text{GSF}} = \sum \left[ U(\mathbf{r}, \mathsf{A}, \mathsf{B}) -
482     U(r_c \hat{\mathbf{r}},\mathsf{A}, \mathsf{B}) - (r-r_c)
483     \hat{\mathbf{r}} \cdot \nabla U(r_c \hat{\mathbf{r}},\mathsf{A}, \mathsf{B}) \right]
484 gezelter 3985 \label{generic2}
485 gezelter 3906 \end{equation}
486 gezelter 4183 where $\hat{\mathbf{r}}$ is the unit vector pointing between the two
487     multipoles, and the sum describes a separate force-shifting that is
488     applied to each orientational contribution to the energy. Both the
489     potential and the gradient for force shifting are evaluated for an
490     image multipole projected onto the surface of the cutoff sphere (see
491     fig \ref{fig:shiftedMultipoles}). The image multipole retains the
492 gezelter 4198 orientation (rotation matrix $\mathsf{B}$) of the interacting multipole. No
493 gezelter 3990 higher order terms $(r-r_c)^n$ appear. The primary difference between
494     the TSF and GSF methods is the stage at which the Taylor Series is
495     applied; in the Taylor-shifted approach, it is applied to the kernel
496     itself. In the Gradient-shifted approach, it is applied to individual
497 gezelter 4183 radial interaction terms in the multipole expansion. Energies from
498 gezelter 3990 this method thus have the general form:
499 gezelter 3906 \begin{equation}
500 gezelter 3985 U= \sum (\text{angular factor}) (\text{radial factor}).
501     \label{generic3}
502 gezelter 3906 \end{equation}
503    
504 gezelter 3986 Functional forms for both methods (TSF and GSF) can both be summarized
505 gezelter 4198 using the form of Eq.~\ref{generic3}). The basic forms for the
506 gezelter 3985 energy, force, and torque expressions are tabulated for both shifting
507 gezelter 3986 approaches below -- for each separate orientational contribution, only
508 gezelter 3985 the radial factors differ between the two methods.
509 gezelter 3906
510 gezelter 4179 \subsection{Generalization of the Wolf shifted potential (SP)}
511     It is also possible to formulate an extension of the Wolf approach for
512     multipoles by simply projecting the image multipole onto the surface
513     of the cutoff sphere, and including the interactions with the central
514     multipole and the image. This effectively shifts the pair potential
515     to zero at the cutoff radius,
516     \begin{equation}
517 gezelter 4198 U^{\text{SP}} = \sum \left[ U(\mathbf{r}, \mathsf{A}, \mathsf{B}) -
518     U(r_c \hat{\mathbf{r}},\mathsf{A}, \mathsf{B}) \right]
519 gezelter 4179 \label{eq:SP}
520     \end{equation}
521     independent of the orientations of the two multipoles. The sum again
522     describes separate potential shifting that is applied to each
523     orientational contribution to the energy.
524    
525     The shifted potential (SP) method is a simple truncation of the GSF
526     method for each orientational contribution, leaving out the $(r-r_c)$
527     terms that multiply the gradient. Functional forms for the
528     shifted-potential (SP) method can also be summarized using the form of
529 gezelter 4181 Eq.~\ref{generic3}. The energy, force, and torque expressions are
530 gezelter 4179 tabulated below for all three methods. As in the GSF and TSF methods,
531     for each separate orientational contribution, only the radial factors
532     differ between the SP, GSF, and TSF methods.
533    
534    
535 gezelter 3906 \subsection{\label{sec:level2}Body and space axes}
536 gezelter 4198 Although objects $a$ and $b$ rotate during a molecular
537 gezelter 3989 dynamics (MD) simulation, their multipole tensors remain fixed in
538     body-frame coordinates. While deriving force and torque expressions,
539     it is therefore convenient to write the energies, forces, and torques
540     in intermediate forms involving the vectors of the rotation matrices.
541 gezelter 4198 We denote body axes for objects $a$ and $b$ using unit vectors
542     $\hat{\mathbf{A}}_m$ and $\hat{\mathbf{B}}_m$, respectively, with the index $m=(123)$.
543 gezelter 4179 In a typical simulation, the initial axes are obtained by
544 gezelter 3989 diagonalizing the moment of inertia tensors for the objects. (N.B.,
545     the body axes are generally {\it not} the same as those for which the
546     quadrupole moment is diagonal.) The rotation matrices are then
547     propagated during the simulation.
548 gezelter 3906
549 gezelter 4198 The rotation matrices $\mathsf {A}$ and $\mathsf {B}$ can be
550 gezelter 3985 expressed using these unit vectors:
551 gezelter 3906 \begin{eqnarray}
552 gezelter 4198 \mathsf {A} =
553 gezelter 3906 \begin{pmatrix}
554 gezelter 4198 \hat{\mathbf{A}}_1 \\
555     \hat{\mathbf{A}}_2 \\
556     \hat{\mathbf{A}}_3
557 gezelter 3989 \end{pmatrix}, \qquad
558 gezelter 4198 \mathsf {B} =
559 gezelter 3906 \begin{pmatrix}
560 gezelter 4198 \hat{\mathbf{B}}_1 \\
561     \hat{\mathbf{B}}_2 \\
562     \hat{\mathbf{B}}_3
563 gezelter 3906 \end{pmatrix}
564     \end{eqnarray}
565     %
566 gezelter 3985 These matrices convert from space-fixed $(xyz)$ to body-fixed $(123)$
567 gezelter 3989 coordinates.
568    
569     Allen and Germano,\cite{Allen:2006fk} following earlier work by Price
570     {\em et al.},\cite{Price:1984fk} showed that if the interaction
571 gezelter 4198 energies are written explicitly in terms of $\hat{\mathbf{r}}$ and the body
572     axes ($\hat{\mathbf{A}}_m$, $\hat{\mathbf{B}}_n$) :
573 gezelter 3906 %
574 gezelter 3985 \begin{equation}
575 gezelter 4198 U(r, \{\hat{\mathbf{A}}_m \cdot \hat{\mathbf{r}} \},
576     \{\hat{\mathbf{B}}_n\cdot \hat{\mathbf{r}} \},
577     \{\hat{\mathbf{A}}_m \cdot \hat{\mathbf{B}}_n \}) .
578 gezelter 3989 \label{ugeneral}
579     \end{equation}
580     %
581     the forces come out relatively cleanly,
582     %
583     \begin{equation}
584 gezelter 4198 \mathbf{F}_a=-\mathbf{F}_b = \nabla U =
585     \frac{\partial U}{\partial \mathbf{r}}
586     = \frac{\partial U}{\partial r} \hat{\mathbf{r}}
587 gezelter 3989 + \sum_m \left[
588 gezelter 4198 \frac{\partial U}{\partial (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{r}})}
589     \frac { \partial (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{r}})}{\partial \mathbf{r}}
590     + \frac{\partial U}{\partial (\hat{\mathbf{B}}_m \cdot \hat{\mathbf{r}})}
591     \frac { \partial (\hat{\mathbf{B}}_m \cdot \hat{\mathbf{r}})}{\partial \mathbf{r}}
592 gezelter 3989 \right] \label{forceequation}.
593     \end{equation}
594    
595     The torques can also be found in a relatively similar
596     manner,
597     %
598     \begin{eqnarray}
599 gezelter 4198 \mathbf{\tau}_a =
600 gezelter 3989 \sum_m
601 gezelter 4198 \frac{\partial U}{\partial (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{r}})}
602     ( \hat{\mathbf{r}} \times \hat{\mathbf{A}}_m )
603 gezelter 3989 -\sum_{mn}
604 gezelter 4198 \frac{\partial U}{\partial (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{B}}_n)}
605     (\hat{\mathbf{A}}_m \times \hat{\mathbf{B}}_n) \\
606 gezelter 3989 %
607 gezelter 4198 \mathbf{\tau}_b =
608 gezelter 3989 \sum_m
609 gezelter 4198 \frac{\partial U}{\partial (\hat{\mathbf{B}}_m \cdot \hat{\mathbf{r}})}
610     ( \hat{\mathbf{r}} \times \hat{\mathbf{B}}_m)
611 gezelter 3989 +\sum_{mn}
612 gezelter 4198 \frac{\partial U}{\partial (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{B}}_n)}
613     (\hat{\mathbf{A}}_m \times \hat{\mathbf{B}}_n) .
614 gezelter 3989 \end{eqnarray}
615    
616 gezelter 4179 Note that our definition of $\mathbf{r}=\mathbf{r}_b - \mathbf{r}_a $
617 gezelter 3989 is opposite in sign to that of Allen and Germano.\cite{Allen:2006fk}
618     We also made use of the identities,
619     %
620     \begin{align}
621 gezelter 4198 \frac { \partial (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{r}})}{\partial \mathbf{r}}
622     =& \frac{1}{r} \left( \hat{\mathbf{A}}_m - (\hat{\mathbf{A}}_m \cdot \hat{\mathbf{r}})\hat{\mathbf{r}}
623 gezelter 3989 \right) \\
624 gezelter 4198 \frac { \partial (\hat{\mathbf{B}}_m \cdot \hat{\mathbf{r}})}{\partial \mathbf{r}}
625     =& \frac{1}{r} \left( \hat{\mathbf{B}}_m - (\hat{\mathbf{B}}_m \cdot \hat{\mathbf{r}})\hat{\mathbf{r}}
626 gezelter 4179 \right).
627 gezelter 3989 \end{align}
628    
629     Many of the multipole contractions required can be written in one of
630 gezelter 4198 three equivalent forms using the unit vectors $\hat{\mathbf{r}}$, $\hat{\mathbf{A}}_m$,
631     and $\hat{\mathbf{B}}_n$. In the torque expressions, it is useful to have the
632 gezelter 3989 angular-dependent terms available in all three fashions, e.g. for the
633     dipole-dipole contraction:
634     %
635     \begin{equation}
636 gezelter 4198 \mathbf{D}_a \cdot \mathbf{D}_b
637     = D_{a \alpha} D_{b \alpha} =
638     \sum_{mn} D_{am} \hat{\mathbf{A}}_m \cdot \hat{\mathbf{B}}_n D_{bn}.
639 gezelter 3985 \end{equation}
640 gezelter 3906 %
641 gezelter 3985 The first two forms are written using space coordinates. The first
642     form is standard in the chemistry literature, while the second is
643     expressed using implied summation notation. The third form shows
644     explicit sums over body indices and the dot products now indicate
645     contractions using space indices.
646 gezelter 3906
647 gezelter 3989 In computing our force and torque expressions, we carried out most of
648     the work in body coordinates, and have transformed the expressions
649     back to space-frame coordinates, which are reported below. Interested
650     readers may consult the supplemental information for this paper for
651 gezelter 4199 the intermediate body-frame expressions.\cite{supplemental}
652 gezelter 3906
653 gezelter 3982 \subsection{The Self-Interaction \label{sec:selfTerm}}
654    
655 gezelter 3985 In addition to cutoff-sphere neutralization, the Wolf
656     summation~\cite{Wolf99} and the damped shifted force (DSF)
657 gezelter 4098 extension~\cite{Fennell:2006zl} also include self-interactions that
658 gezelter 3985 are handled separately from the pairwise interactions between
659     sites. The self-term is normally calculated via a single loop over all
660     sites in the system, and is relatively cheap to evaluate. The
661     self-interaction has contributions from two sources.
662    
663     First, the neutralization procedure within the cutoff radius requires
664     a contribution from a charge opposite in sign, but equal in magnitude,
665     to the central charge, which has been spread out over the surface of
666     the cutoff sphere. For a system of undamped charges, the total
667     self-term is
668 gezelter 3980 \begin{equation}
669 gezelter 4198 U_\textrm{self} = - \frac{1}{r_c} \sum_{a=1}^N C_a^{2}.
670 gezelter 3980 \label{eq:selfTerm}
671     \end{equation}
672 gezelter 3985
673     Second, charge damping with the complementary error function is a
674     partial analogy to the Ewald procedure which splits the interaction
675     into real and reciprocal space sums. The real space sum is retained
676     in the Wolf and DSF methods. The reciprocal space sum is first
677     minimized by folding the largest contribution (the self-interaction)
678     into the self-interaction from charge neutralization of the damped
679     potential. The remainder of the reciprocal space portion is then
680     discarded (as this contributes the largest computational cost and
681     complexity to the Ewald sum). For a system containing only damped
682     charges, the complete self-interaction can be written as
683 gezelter 3980 \begin{equation}
684 gezelter 4175 U_\textrm{self} = - \left(\frac{\textrm{erfc}(\alpha r_c)}{r_c} +
685 gezelter 4198 \frac{\alpha}{\sqrt{\pi}} \right)
686     \sum_{a=1}^N
687     C_a^{2}.
688 gezelter 3980 \label{eq:dampSelfTerm}
689     \end{equation}
690    
691     The extension of DSF electrostatics to point multipoles requires
692 gezelter 4183 treatment of the self-neutralization \textit{and} reciprocal
693 gezelter 3980 contributions to the self-interaction for higher order multipoles. In
694     this section we give formulae for these interactions up to quadrupolar
695     order.
696    
697     The self-neutralization term is computed by taking the {\it
698     non-shifted} kernel for each interaction, placing a multipole of
699     equal magnitude (but opposite in polarization) on the surface of the
700     cutoff sphere, and averaging over the surface of the cutoff sphere.
701     Because the self term is carried out as a single sum over sites, the
702     reciprocal-space portion is identical to half of the self-term
703 gezelter 4179 obtained by Smith, and also by Aguado and Madden for the application
704     of the Ewald sum to multipoles.\cite{Smith82,Smith98,Aguado03} For a
705 gezelter 4213 given site which possesses a charge, dipole, and quadrupole, both types
706 gezelter 4179 of contribution are given in table \ref{tab:tableSelf}.
707 gezelter 3980
708     \begin{table*}
709     \caption{\label{tab:tableSelf} Self-interaction contributions for
710 gezelter 4198 site ($a$) that has a charge $(C_a)$, dipole
711     $(\mathbf{D}_a)$, and quadrupole $(\mathsf{Q}_a)$}.
712 gezelter 3980 \begin{ruledtabular}
713     \begin{tabular}{lccc}
714     Multipole order & Summed Quantity & Self-neutralization & Reciprocal \\ \hline
715 gezelter 4198 Charge & $C_a^2$ & $-f(r_c)$ & $-\frac{\alpha}{\sqrt{\pi}}$ \\
716     Dipole & $|\mathbf{D}_a|^2$ & $\frac{1}{3} \left( h(r_c) +
717 gezelter 3980 \frac{2 g(r_c)}{r_c} \right)$ & $-\frac{2 \alpha^3}{3 \sqrt{\pi}}$\\
718 gezelter 4198 Quadrupole & $2 \mathsf{Q}_a:\mathsf{Q}_a + \text{Tr}(\mathsf{Q}_a)^2$ &
719 gezelter 3980 $- \frac{1}{15} \left( t(r_c)+ \frac{4 s(r_c)}{r_c} \right)$ &
720     $-\frac{4 \alpha^5}{5 \sqrt{\pi}}$ \\
721 gezelter 4198 Charge-Quadrupole & $-2 C_a \text{Tr}(\mathsf{Q}_a)$ & $\frac{1}{3} \left(
722 gezelter 3980 h(r_c) + \frac{2 g(r_c)}{r_c} \right)$& $-\frac{2 \alpha^3}{3 \sqrt{\pi}}$ \\
723     \end{tabular}
724     \end{ruledtabular}
725     \end{table*}
726    
727     For sites which simultaneously contain charges and quadrupoles, the
728     self-interaction includes a cross-interaction between these two
729     multipole orders. Symmetry prevents the charge-dipole and
730     dipole-quadrupole interactions from contributing to the
731     self-interaction. The functions that go into the self-neutralization
732 gezelter 3985 terms, $g(r), h(r), s(r), \mathrm{~and~} t(r)$ are successive
733     derivatives of the electrostatic kernel, $f(r)$ (either the undamped
734     $1/r$ or the damped $B_0(r)=\mathrm{erfc}(\alpha r)/r$ function) that
735     have been evaluated at the cutoff distance. For undamped
736     interactions, $f(r_c) = 1/r_c$, $g(r_c) = -1/r_c^{2}$, and so on. For
737     damped interactions, $f(r_c) = B_0(r_c)$, $g(r_c) = B_0'(r_c)$, and so
738     on. Appendix \ref{SmithFunc} contains recursion relations that allow
739     rapid evaluation of these derivatives.
740 gezelter 3980
741 gezelter 3985 \section{Interaction energies, forces, and torques}
742     The main result of this paper is a set of expressions for the
743     energies, forces and torques (up to quadrupole-quadrupole order) that
744 gezelter 4183 work for the Taylor-shifted, gradient-shifted, and shifted potential
745     approximations. These expressions were derived using a set of generic
746     radial functions. Without using the shifting approximations mentioned
747     above, some of these radial functions would be identical, and the
748     expressions coalesce into the familiar forms for unmodified
749     multipole-multipole interactions. Table \ref{tab:tableenergy} maps
750     between the generic functions and the radial functions derived for the
751     three methods. The energy equations are written in terms of lab-frame
752     representations of the dipoles, quadrupoles, and the unit vector
753     connecting the two objects,
754 gezelter 3906
755     % Energy in space coordinate form ----------------------------------------------------------------------------------------------
756     %
757     %
758     % u ca cb
759     %
760 gezelter 3983 \begin{align}
761 gezelter 4198 U_{C_a C_b}(r)=&
762     C_a C_b v_{01}(r) \label{uchch}
763 gezelter 3983 \\
764 gezelter 3906 %
765     % u ca db
766     %
767 gezelter 4198 U_{C_a \mathbf{D}_b}(r)=&
768     C_a \left( \mathbf{D}_b \cdot \hat{\mathbf{r}} \right) v_{11}(r)
769 gezelter 3906 \label{uchdip}
770 gezelter 3983 \\
771 gezelter 3906 %
772     % u ca qb
773     %
774 gezelter 4198 U_{C_a \mathsf{Q}_b}(r)=& C_a \Bigl[ \text{Tr}\mathsf{Q}_b
775     v_{21}(r) + \left( \hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot
776     \hat{\mathbf{r}} \right) v_{22}(r) \Bigr]
777 gezelter 3906 \label{uchquad}
778 gezelter 3983 \\
779 gezelter 3906 %
780     % u da cb
781     %
782 gezelter 3983 %U_{D_{\bf a}C_{\bf b}}(r)=&
783     %-\frac{C_{\bf b}}{4\pi \epsilon_0}
784     %\left( \mathbf{D}_{\mathbf{a}} \cdot \hat{r} \right) v_{11}(r) \label{udipch}
785     %\\
786 gezelter 3906 %
787     % u da db
788     %
789 gezelter 4198 U_{\mathbf{D}_a \mathbf{D}_b}(r)=&
790     -\Bigr[ \left( \mathbf{D}_a \cdot
791     \mathbf{D}_b \right) v_{21}(r)
792     +\left( \mathbf{D}_a \cdot \hat{\mathbf{r}} \right)
793     \left( \mathbf{D}_b \cdot \hat{\mathbf{r}} \right)
794 gezelter 3906 v_{22}(r) \Bigr]
795     \label{udipdip}
796 gezelter 3983 \\
797 gezelter 3906 %
798     % u da qb
799     %
800     \begin{split}
801     % 1
802 gezelter 4198 U_{\mathbf{D}_a \mathsf{Q}_b}(r) =&
803 gezelter 3985 -\Bigl[
804 gezelter 4198 \text{Tr}\mathsf{Q}_b
805     \left( \mathbf{D}_a \cdot \hat{\mathbf{r}} \right)
806     +2 ( \mathbf{D}_a \cdot
807     \mathsf{Q}_b \cdot \hat{\mathbf{r}} ) \Bigr] v_{31}(r) \\
808 gezelter 3906 % 2
809 gezelter 4198 &- \left( \mathbf{D}_a \cdot \hat{\mathbf{r}} \right)
810     \left( \hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}} \right) v_{32}(r)
811 gezelter 3906 \label{udipquad}
812     \end{split}
813 gezelter 3983 \\
814 gezelter 3906 %
815     % u qa cb
816     %
817 gezelter 3983 %U_{Q_{\bf a}C_{\bf b}}(r)=&
818     %\frac{C_{\bf b }}{4\pi \epsilon_0} \Bigl[ \text{Tr}\mathbf{Q}_{\bf a} v_{21}(r)
819     %\left( \hat{r} \cdot \mathbf{Q}_{{\mathbf a}} \cdot \hat{r} \right) v_{22}(r) \Bigr]
820     %\label{uquadch}
821     %\\
822 gezelter 3906 %
823     % u qa db
824     %
825 gezelter 3983 %\begin{split}
826 gezelter 3906 %1
827 gezelter 3983 %U_{Q_{\bf a}D_{\bf b}}(r)=&
828     %\frac{1}{4\pi \epsilon_0} \Bigl[
829     %\text{Tr}\mathbf{Q}_{\mathbf{a}}
830     %\left( \mathbf{D}_{\mathbf{b}} \cdot \hat{r} \right)
831     %+2 ( \mathbf{D}_{\mathbf{b}} \cdot
832     %\mathbf{Q}_{\mathbf{a}} \cdot \hat{r}) \Bigr] v_{31}(r)\\
833 gezelter 3906 % 2
834 gezelter 3983 %&+\frac{1}{4\pi \epsilon_0}
835     %\left( \mathbf{D}_{\mathbf{b}} \cdot \hat{r} \right)
836     %\left( \hat{r} \cdot \mathbf{Q}_{{\mathbf a}} \cdot \hat{r} \right) v_{32}(r)
837     %\label{uquaddip}
838     %\end{split}
839     %\\
840 gezelter 3906 %
841     % u qa qb
842     %
843     \begin{split}
844     %1
845 gezelter 4198 U_{\mathsf{Q}_a \mathsf{Q}_b}(r)=&
846 gezelter 3985 \Bigl[
847 gezelter 4198 \text{Tr} \mathsf{Q}_a \text{Tr} \mathsf{Q}_b
848 gezelter 3989 +2
849 gezelter 4198 \mathsf{Q}_a : \mathsf{Q}_b \Bigr] v_{41}(r)
850 gezelter 3906 \\
851     % 2
852 gezelter 4198 &+\Bigl[ \text{Tr}\mathsf{Q}_a
853     \left( \hat{\mathbf{r}} \cdot
854     \mathsf{Q}_b \cdot \hat{\mathbf{r}} \right)
855     +\text{Tr}\mathsf{Q}_b
856     \left( \hat{\mathbf{r}} \cdot \mathsf{Q}_a
857     \cdot \hat{\mathbf{r}} \right) +4 (\hat{\mathbf{r}} \cdot
858     \mathsf{Q}_a \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}})
859 gezelter 3906 \Bigr] v_{42}(r)
860     \\
861     % 4
862 gezelter 3985 &+
863 gezelter 4198 \left( \hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot \hat{\mathbf{r}} \right)
864     \left( \hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}} \right) v_{43}(r).
865 gezelter 3906 \label{uquadquad}
866     \end{split}
867 gezelter 3983 \end{align}
868 gezelter 3985 %
869 gezelter 4198 Note that the energies of multipoles on site $b$ interacting
870     with those on site $a$ can be obtained by swapping indices
871     along with the sign of the intersite vector, $\hat{\mathbf{r}}$.
872 gezelter 3906
873     %
874     %
875     % TABLE of radial functions ----------------------------------------------------------------------------------------------------------------
876     %
877    
878 gezelter 3985 \begin{sidewaystable}
879     \caption{\label{tab:tableenergy}Radial functions used in the energy
880 gezelter 4183 and torque equations. The $f, g, h, s, t, \mathrm{and~} u$
881     functions used in this table are defined in Appendices
882     \ref{radialTSF} and \ref{radialGSF}. The gradient shifted (GSF)
883     functions include the shifted potential (SP)
884     contributions (\textit{cf.} Eqs. \ref{generic2} and
885     \ref{eq:SP}).}
886 gezelter 4177 \begin{tabular}{|c|c|l|l|l|} \hline
887     Generic&Bare Coulomb&Taylor-Shifted (TSF)&Shifted Potential (SP)&Gradient-Shifted (GSF)
888 gezelter 3906 \\ \hline
889     %
890     %
891     %
892     %Ch-Ch&
893     $v_{01}(r)$ &
894     $\frac{1}{r}$ &
895     $f_0(r)$ &
896 gezelter 4177 $f(r)-f(r_c)$ &
897     SP $-(r-r_c)g(r_c)$
898 gezelter 3906 \\
899     %
900     %
901     %
902     %Ch-Di&
903     $v_{11}(r)$ &
904     $-\frac{1}{r^2}$ &
905     $g_1(r)$ &
906 gezelter 4177 $g(r)-g(r_c)$ &
907     SP $-(r-r_c)h(r_c)$ \\
908 gezelter 3906 %
909     %
910     %
911     %Ch-Qu/Di-Di&
912     $v_{21}(r)$ &
913     $-\frac{1}{r^3} $ &
914     $\frac{g_2(r)}{r} $ &
915 gezelter 4177 $\frac{g(r)}{r}-\frac{g(r_c)}{r_c}$ &
916 gezelter 4179 SP $-(r-r_c) \left( -\frac{g(r_c)}{r_c^2} + \frac{h(r_c)}{r_c} \right)$ \\
917     %
918     %
919     %
920 gezelter 3906 $v_{22}(r)$ &
921     $\frac{3}{r^3} $ &
922     $\left(-\frac{g_2(r)}{r} + h_2(r) \right)$ &
923 gezelter 4179 $\left(-\frac{g(r)}{r}+h(r) \right) -\left(-\frac{g(r_c)}{r_c}+h(r_c) \right)$
924     & SP $-(r-r_c) \left( \frac{g(r_c)}{r_c^2}-\frac{h(r_c)}{r_c}+s(r_c) \right)$\\
925 gezelter 3906 %
926     %
927     %
928     %Di-Qu &
929     $v_{31}(r)$ &
930     $\frac{3}{r^4} $ &
931     $\left(-\frac{g_3(r)}{r^2} + \frac{h_3(r)}{r} \right)$ &
932 gezelter 4179 $\left( -\frac{g(r)}{r^2}+\frac{h(r)}{r}\right)-\left(-\frac{g(r_c)}{r_c^2}+\frac{h(r_c)}{r_c} \right)$
933     & SP $-(r-r_c) \left(\frac{2g(r_c)}{r_c^3}-\frac{2h(r_c)}{r_c^2}+\frac{s(r_c)}{r_c} \right)$ \\
934 gezelter 3906 %
935 gezelter 4179 %
936     %
937 gezelter 3906 $v_{32}(r)$ &
938     $-\frac{15}{r^4} $ &
939     $\left( \frac{3g_3(r)}{r^2} - \frac{3h_3(r)}{r} + s_3(r) \right)$ &
940 gezelter 4179 $\left( \frac{3g(r)}{r^2} - \frac{3h(r)}{r} + s(r) \right)$&
941     SP $-(r-r_c) \left( \frac{-6g(r_c)}{r_c^3}+\frac{6h(r_c)}{r_c^2}\right.$ \\
942     &&& $~~~-\left(\frac{3g(r_c)}{r_c^2} - \frac{3h(r_c)}{r_c} + s(r_c)\right)$ &
943     $\phantom{SP-(r-r_c)}\left.-\frac{3s(r_c)}{r_c}+t(r_c) \right)$\\
944 gezelter 3906 %
945     %
946     %
947     %Qu-Qu&
948     $v_{41}(r)$ &
949     $\frac{3}{r^5} $ &
950     $\left(-\frac{g_4(r)}{r^3} +\frac{h_4(r)}{r^2} \right) $ &
951 gezelter 4179 $\left( -\frac{g(r)}{r^3} + \frac{h(r)}{r^2} \right)- \left(-\frac{g(r_c)}{r_c^3} + \frac{h(r_c)}{r_c^2} \right)$ &
952     SP $-(r-r_c) \left( \frac{3g(r_c)}{r_c^4}-\frac{3h(r_c)}{r_c^3}+\frac{s(r_c)}{r_c^2} \right)$
953 gezelter 3906 \\
954     % 2
955     $v_{42}(r)$ &
956     $- \frac{15}{r^5} $ &
957     $\left( \frac{3g_4(r)}{r^3} - \frac{3h_4(r)}{r^2}+\frac{s_4(r)}{r} \right)$ &
958 gezelter 4179 $\left( \frac{3g(r)}{r^3} - \frac{3h(r)}{r^2}+\frac{s(r)}{r} \right)$ &
959     SP$-(r-r_c) \left(- \frac{9g(r_c)}{r_c^4}+\frac{9h(r_c)}{r_c^3}\right.$ \\
960     &&& $~~~-\left( \frac{3g(r_c)}{r_c^3} - \frac{3h(r_c)}{r_c^2}+\frac{s(r_c)}{r_c} \right)$ &
961     $\phantom{SP-(r-r_c)}\left. -\frac{4s(r_c)}{r_c^2} + \frac{t(r_c)}{r_c}\right)$\\
962 gezelter 3906 % 3
963 gezelter 4179 %
964     %
965 gezelter 3906 $v_{43}(r)$ &
966     $ \frac{105}{r^5} $ &
967     $\left(-\frac{15g_4(r)}{r^3}+\frac{15h_4(r)}{r^2}-\frac{6s_4(r)}{r} + t_4(r)\right) $ &
968 gezelter 4179 $ \left(-\frac{15g(r)}{r^3} +\frac{15h(r)}{r^2}-\frac{6s(r)}{r}+t(r)\right) $ &
969     SP $-(r-r_c)\left(\frac{45g(r_c)}{r_c^4}-\frac{45h(r_c)}{r_c^3}\right.$\\
970     &&& $~~~-\left(-\frac{15g(r_c)}{r_c^3}+\frac{15h(r_c)}{r_c^2}-\frac{6s(r_c)}{r_c}+ t(r_c)\right)$ &
971     $\phantom{SP-(r-r_c)}\left.+\frac{21s(r_c)}{r_c^2}-\frac{6t(r_c)}{r_c}+u(r_c) \right)$\\
972 gezelter 4177 \hline
973 gezelter 3906 \end{tabular}
974 gezelter 3985 \end{sidewaystable}
975 gezelter 3906 %
976     %
977     % FORCE TABLE of radial functions ----------------------------------------------------------------------------------------------------------------
978     %
979    
980 gezelter 3985 \begin{sidewaystable}
981 gezelter 4179 \caption{\label{tab:tableFORCE}Radial functions used in the force
982     equations. Gradient shifted (GSF) functions are constructed using the shifted
983     potential (SP) functions. Some of these functions are simple
984     modifications of the functions found in table \ref{tab:tableenergy}}
985     \begin{tabular}{|c|c|l|l|l|} \hline
986     Function&Definition&Taylor-Shifted (TSF)& Shifted Potential (SP)
987     &Gradient-Shifted (GSF)
988 gezelter 3906 \\ \hline
989     %
990     %
991     %
992     $w_a(r)$&
993 gezelter 3985 $\frac{d v_{01}}{dr}$&
994     $g_0(r)$&
995 gezelter 4179 $g(r)$&
996     SP $-g(r_c)$ \\
997 gezelter 3906 %
998     %
999     $w_b(r)$ &
1000 gezelter 3985 $\frac{d v_{11}}{dr} - \frac{v_{11}(r)}{r} $&
1001     $\left( -\frac{g_1(r)}{r}+h_1(r) \right)$ &
1002 gezelter 4179 $h(r) - \frac{v_{11}(r)}{r} $ &
1003     SP $- h(r_c)$ \\
1004 gezelter 3906 %
1005     $w_c(r)$ &
1006 gezelter 3985 $\frac{v_{11}(r)}{r}$ &
1007     $\frac{g_1(r)}{r} $ &
1008 gezelter 4179 $\frac{v_{11}(r)}{r}$&
1009 gezelter 3985 $\frac{v_{11}(r)}{r}$\\
1010 gezelter 3906 %
1011     %
1012     $w_d(r)$&
1013 gezelter 3985 $\frac{d v_{21}}{dr}$&
1014     $\left( -\frac{g_2(r)}{r^2} + \frac{h_2(r)}{r} \right) $ &
1015 gezelter 4179 $\left( -\frac{g(r)}{r^2} + \frac{h(r)}{r} \right)$ &
1016     SP $-\left( -\frac{g(r_c)}{r_c^2} + \frac{h(r_c)}{r_c} \right) $ \\
1017 gezelter 3906 %
1018     $w_e(r)$ &
1019 gezelter 3985 $\left(-\frac{g_2(r)}{r^2} + \frac{h_2(r)}{r} \right)$ &
1020     $\frac{v_{22}(r)}{r}$ &
1021 gezelter 4179 $\frac{v_{22}(r)}{r}$ &
1022 gezelter 3906 $\frac{v_{22}(r)}{r}$ \\
1023     %
1024     %
1025     $w_f(r)$&
1026 gezelter 3985 $\frac{d v_{22}}{dr} - \frac{2v_{22}(r)}{r}$&
1027     $\left( \frac{3g_2(r)}{r^2}-\frac{3h_2(r)}{r}+s_2(r) \right)$ &
1028 gezelter 4179 $ \left( \frac{g(r)}{r^2}-\frac{h(r)}{r}+s(r) \right) -\frac{2v_{22}(r)}{r}$&
1029     SP $- \left( \frac{g(r_c)}{r_c^2}-\frac{h(r_c)}{r_c}+s(r_c) \right)$\\
1030 gezelter 3906 %
1031     $w_g(r)$&
1032 gezelter 3985 $\frac{v_{31}(r)}{r}$&
1033     $ \left( -\frac{g_3(r)}{r^3}+\frac{h_3(r)}{r^2} \right)$&
1034 gezelter 4179 $\frac{v_{31}(r)}{r}$&
1035 gezelter 3906 $\frac{v_{31}(r)}{r}$\\
1036     %
1037     $w_h(r)$ &
1038 gezelter 3985 $\frac{d v_{31}}{dr} -\frac{v_{31}(r)}{r}$&
1039     $\left(\frac{3g_3(r)}{r^3} -\frac{3h_3(r)}{r^2} +\frac{s_3(r)}{r} \right) $ &
1040 gezelter 4179 $ \left(\frac{2g(r)}{r^3} -\frac{2h(r)}{r^2} +\frac{s(r)}{r} \right) -\frac{v_{31}(r)}{r}$ &
1041     SP $ - \left(\frac{2g(r_c)}{r_c^3} -\frac{2h(r_c)}{r_c^2} +\frac{s(r_c)}{r_c} \right) $ \\
1042 gezelter 3906 % 2
1043     $w_i(r)$ &
1044 gezelter 3985 $\frac{v_{32}(r)}{r}$ &
1045     $\left(\frac{3g_3(r)}{r^3} -\frac{3h_3(r)}{r^2} +\frac{s_3(r)}{r} \right) $ &
1046 gezelter 4179 $\frac{v_{32}(r)}{r}$&
1047 gezelter 3985 $\frac{v_{32}(r)}{r}$\\
1048 gezelter 3906 %
1049     $w_j(r)$ &
1050 gezelter 3985 $\frac{d v_{32}}{dr} - \frac{3v_{32}}{r}$&
1051     $\left(\frac{-15g_3(r)}{r^3} + \frac{15h_3(r)}{r^2} - \frac{6s_3(r)}{r} + t_3(r) \right) $ &
1052 gezelter 4179 $\left(\frac{-6g(r)}{r^3} +\frac{6h(r)}{r^2} -\frac{3s(r)}{r} +t(r) \right) -\frac{3v_{32}}{r}$ &
1053     SP $-\left(\frac{-6g(_cr)}{r_c^3} +\frac{6h(r_c)}{r_c^2}
1054     -\frac{3s(r_c)}{r_c} +t(r_c) \right)$ \\
1055 gezelter 3906 %
1056     $w_k(r)$ &
1057 gezelter 3985 $\frac{d v_{41}}{dr} $ &
1058     $\left(\frac{3g_4(r)}{r^4} -\frac{3h_4(r)}{r^3} +\frac{s_4(r)}{r^2} \right)$ &
1059 gezelter 4179 $\left(\frac{3g(r)}{r^4} -\frac{3h(r)}{r^3} +\frac{s(r)}{r^2}
1060     \right)$ &
1061     SP $-\left(\frac{3g(r_c)}{r_c^4} -\frac{3h(r_c)}{r_c^3} +\frac{s(r_c)}{r_c^2} \right)$ \\
1062 gezelter 3906 %
1063     $w_l(r)$ &
1064 gezelter 3985 $\frac{d v_{42}}{dr} -\frac{2v_{42}(r)}{r}$ &
1065     $\left(-\frac{15g_4(r)}{r^4} +\frac{15h_4(r)}{r^3} -\frac{6s_4(r)}{r^2} +\frac{t_4(r)}{r} \right)$ &
1066 gezelter 4179 $\left(-\frac{9g(r)}{r^4} +\frac{9h(r)}{r^3} -\frac{4s(r)}{r^2}
1067     +\frac{t(r)}{r} \right) -\frac{2v_{42}(r)}{r}$&
1068     SP$-\left(-\frac{9g(r_c)}{r_c^4} +\frac{9h(r_c)}{r_c^3} -\frac{4s(r_c)}{r_c^2} +\frac{t(r_c)}{r_c} \right)$\\
1069 gezelter 3906 %
1070     $w_m(r)$ &
1071 gezelter 3985 $\frac{d v_{43}}{dr} -\frac{4v_{43}(r)}{r}$&
1072 gezelter 4179 $\left(\frac{105g_4(r)}{r^4} - \frac{105h_4(r)}{r^3} \right.$ &
1073     $\left(\frac{45g(r)}{r^4} -\frac{45h(r)}{r^3} +\frac{21s(r)}{r^2}\right.$ &
1074     SP $- \left(\frac{45g(r_c)}{r_c^4} -\frac{45h(r_c)}{r_c^3}\right.$ \\
1075     && $~~~\left.+ \frac{45s_4(r)}{r^2} - \frac{10t_4(r)}{r} +u_4(r) \right)$
1076     & $~~~\left. -\frac{6t(r)}{r} +u(r) \right) -\frac{4v_{43}(r)}{r}$ &
1077     $\phantom{SP-} \left.+\frac{21s(r_c)}{r_c^2} -\frac{6t(r_c)}{r_c} +u(r_c) \right) $\\
1078 gezelter 3906 %
1079     $w_n(r)$ &
1080 gezelter 3985 $\frac{v_{42}(r)}{r}$ &
1081     $\left(\frac{3g_4(r)}{r^4} -\frac{3h_4(r)}{r^3} +\frac{s_4(r)}{r^2} \right)$ &
1082 gezelter 4179 $\frac{v_{42}(r)}{r}$&
1083 gezelter 3985 $\frac{v_{42}(r)}{r}$\\
1084 gezelter 3906 %
1085     $w_o(r)$ &
1086 gezelter 3985 $\frac{v_{43}(r)}{r}$&
1087     $\left(-\frac{15g_4(r)}{r^4} +\frac{15h_4(r)}{r^3} -\frac{6s_4(r)}{r^2} +\frac{t_4(r)}{r} \right)$ &
1088 gezelter 4179 $\frac{v_{43}(r)}{r}$&
1089 gezelter 3985 $\frac{v_{43}(r)}{r}$ \\ \hline
1090 gezelter 3906 %
1091    
1092     \end{tabular}
1093 gezelter 3985 \end{sidewaystable}
1094 gezelter 3906 %
1095     %
1096     %
1097    
1098     \subsection{Forces}
1099 gezelter 4198 The force on object $a$, $\mathbf{F}_a$, due to object
1100     $b$ is the negative of the force on $b$ due to $a$. For
1101 gezelter 3985 a simple charge-charge interaction, these forces will point along the
1102 gezelter 4198 $\pm \hat{\mathbf{r}}$ directions, where $\mathbf{r}=\mathbf{r}_b -
1103 gezelter 3985 \mathbf{r}_a $. Thus
1104 gezelter 3906 %
1105     \begin{equation}
1106 gezelter 4198 F_{a \alpha} = \hat{r}_\alpha \frac{\partial U_{C_a C_b}}{\partial r}
1107     \quad \text{and} \quad F_{b \alpha}
1108     = - \hat{r}_\alpha \frac{\partial U_{C_a C_b}} {\partial r} .
1109 gezelter 3906 \end{equation}
1110     %
1111 gezelter 3985 We list below the force equations written in terms of lab-frame
1112 gezelter 4183 coordinates. The radial functions used in the three methods are listed
1113 gezelter 3985 in Table \ref{tab:tableFORCE}
1114 gezelter 3906 %
1115 gezelter 3985 %SPACE COORDINATES FORCE EQUATIONS
1116 gezelter 3906 %
1117     % **************************************************************************
1118     % f ca cb
1119     %
1120 gezelter 3985 \begin{align}
1121 gezelter 4198 \mathbf{F}_{a {C_a} {C_b}} =&
1122     C_a C_b w_a(r) \hat{\mathbf{r}} \\
1123 gezelter 3906 %
1124     %
1125     %
1126 gezelter 4198 \mathbf{F}_{a {C_a} {\mathbf{D}_b} } =&
1127     C_a \Bigl[
1128     \left( \hat{\mathbf{r}} \cdot \mathbf{D}_b \right)
1129     w_b(r) \hat{\mathbf{r}}
1130     + \mathbf{D}_b w_c(r) \Bigr] \\
1131 gezelter 3906 %
1132     %
1133     %
1134 gezelter 4198 \mathbf{F}_{a {C_a} {\mathsf{Q}_b}} =&
1135     C_a \Bigr[
1136     \text{Tr}\mathsf{Q}_b w_d(r) \hat{\mathbf{r}}
1137     + 2 \mathsf{Q}_b \cdot \hat{\mathbf{r}} w_e(r)
1138     + \left( \hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}}
1139     \right) w_f(r) \hat{\mathbf{r}} \Bigr] \\
1140 gezelter 3906 %
1141     %
1142     %
1143 gezelter 3985 % \begin{equation}
1144     % \mathbf{F}_{{\bf a}D_{\bf a}C_{\bf b}} =
1145     % -C_{\bf{b}} \Bigl[
1146     % \left( \hat{r} \cdot \mathbf{D}_{\mathbf{a}} \right) w_b(r) \hat{r}
1147     % + \mathbf{D}_{\mathbf{a}} w_c(r) \Bigr]
1148     % \end{equation}
1149 gezelter 3906 %
1150     %
1151     %
1152 gezelter 3985 \begin{split}
1153 gezelter 4198 \mathbf{F}_{a \mathbf{D}_a \mathbf{D}_b} =&
1154     - \mathbf{D}_a \cdot \mathbf{D}_b w_d(r) \hat{\mathbf{r}}
1155     + \left( \mathbf{D}_a
1156     \left( \mathbf{D}_b \cdot \hat{\mathbf{r}} \right)
1157     + \mathbf{D}_b \left( \mathbf{D}_a \cdot \hat{\mathbf{r}} \right) \right) w_e(r)\\
1158 gezelter 3906 % 2
1159 gezelter 4198 & - \left( \hat{\mathbf{r}} \cdot \mathbf{D}_a \right)
1160     \left( \hat{\mathbf{r}} \cdot \mathbf{D}_b \right) w_f(r) \hat{\mathbf{r}}
1161 gezelter 3985 \end{split}\\
1162 gezelter 3906 %
1163     %
1164     %
1165     \begin{split}
1166 gezelter 4198 \mathbf{F}_{a \mathbf{D}_a \mathsf{Q}_b} =& - \Bigl[
1167     \text{Tr}\mathsf{Q}_b \mathbf{ D}_a
1168     +2 \mathbf{D}_a \cdot
1169     \mathsf{Q}_b \Bigr] w_g(r)
1170 gezelter 3985 - \Bigl[
1171 gezelter 4198 \text{Tr}\mathsf{Q}_b
1172     \left( \hat{\mathbf{r}} \cdot \mathbf{D}_a \right)
1173     +2 ( \mathbf{D}_a \cdot
1174     \mathsf{Q}_b \cdot \hat{\mathbf{r}}) \Bigr] w_h(r) \hat{\mathbf{r}} \\
1175 gezelter 3906 % 3
1176 gezelter 4198 & - \Bigl[\mathbf{ D}_a (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}})
1177     +2 (\hat{\mathbf{r}} \cdot \mathbf{D}_a ) (\hat{\mathbf{r}} \cdot \mathsf{Q}_b ) \Bigr]
1178 gezelter 3906 w_i(r)
1179     % 4
1180 gezelter 3985 -
1181 gezelter 4198 (\hat{\mathbf{r}} \cdot \mathbf{D}_a )
1182     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}}) w_j(r) \hat{\mathbf{r}} \end{split} \\
1183 gezelter 3906 %
1184     %
1185 gezelter 3985 % \begin{equation}
1186     % \mathbf{F}_{{\bf a}Q_{\bf a}C_{\bf b}} =
1187     % \frac{C_{\bf b }}{4\pi \epsilon_0} \Bigr[
1188     % \text{Tr}\mathbf{Q}_{\bf a} w_d(r) \hat{r}
1189     % + 2 \mathbf{Q}_{{\mathbf a}} \cdot \hat{r} w_e(r)
1190     % + \left( \hat{r} \cdot \mathbf{Q}_{{\mathbf a}} \cdot \hat{r} \right) w_f(r) \hat{r} \Bigr]
1191     % \end{equation}
1192     % %
1193     % \begin{equation}
1194     % \begin{split}
1195     % \mathbf{F}_{{\bf a}Q_{\bf a}D_{\bf b}} =
1196     % &\frac{1}{4\pi \epsilon_0} \Bigl[
1197     % \text{Tr}\mathbf{Q}_{\mathbf{a}} \mathbf{D}_{\mathbf{b}}
1198     % +2 \mathbf{D}_{\mathbf{b}} \cdot \mathbf{Q}_{\mathbf{a}} \Bigr] w_g(r)
1199     % % 2
1200     % + \frac{1}{4\pi \epsilon_0} \Bigl[ \text{Tr}\mathbf{Q}_{\mathbf{a}}
1201     % (\hat{r} \cdot \mathbf{D}_{\mathbf{b}})
1202     % +2 (\mathbf{D}_{\mathbf{b}} \cdot
1203     % \mathbf{Q}_{\mathbf{a}} \cdot \hat{r}) \Bigr] w_h(r) \hat{r} \\
1204     % % 3
1205     % &+ \frac{1}{4\pi \epsilon_0} \Bigl[ \mathbf{D}_{\mathbf{b}}
1206     % (\hat{r} \cdot \mathbf{Q}_{{\mathbf a}} \cdot \hat{r})
1207     % +2 (\hat{r} \cdot \mathbf{D}_{\mathbf{b}})
1208     % (\hat{r} \cdot \mathbf{Q}_{{\mathbf a}} ) \Bigr] w_i(r)
1209     % % 4
1210     % +\frac{1}{4\pi \epsilon_0}
1211     % (\hat{r} \cdot \mathbf{D}_{\mathbf{b}})
1212     % (\hat{r} \cdot \mathbf{Q}_{{\mathbf a}} \cdot \hat{r}) w_j(r) \hat{r}
1213     % \end{split}
1214     % \end{equation}
1215 gezelter 3906 %
1216     %
1217     %
1218     \begin{split}
1219 gezelter 4198 \mathbf{F}_{a \mathsf{Q}_a \mathsf{Q}_b} =&
1220 gezelter 3985 \Bigl[
1221 gezelter 4198 \text{Tr}\mathsf{Q}_a \text{Tr}\mathsf{Q}_b
1222     + 2 \mathsf{Q}_a : \mathsf{Q}_b \Bigr] w_k(r) \hat{\mathbf{r}} \\
1223 gezelter 3906 % 2
1224 gezelter 3985 &+ \Bigl[
1225 gezelter 4198 2\text{Tr}\mathsf{Q}_b (\hat{\mathbf{r}} \cdot \mathsf{Q}_a )
1226     + 2\text{Tr}\mathsf{Q}_a (\hat{\mathbf{r}} \cdot \mathsf{Q}_b )
1227 gezelter 3906 % 3
1228 gezelter 4198 +4 (\mathsf{Q}_a \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}})
1229     + 4(\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot \mathsf{Q}_b) \Bigr] w_n(r) \\
1230 gezelter 3906 % 4
1231 gezelter 3985 &+ \Bigl[
1232 gezelter 4198 \text{Tr}\mathsf{Q}_a (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}})
1233     + \text{Tr}\mathsf{Q}_b
1234     (\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot \hat{\mathbf{r}})
1235 gezelter 3906 % 5
1236 gezelter 4198 +4 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot
1237     \mathsf{Q}_b \cdot \hat{\mathbf{r}}) \Bigr] w_l(r) \hat{\mathbf{r}} \\
1238 gezelter 3906 %
1239 gezelter 3985 &+ \Bigl[
1240 gezelter 4198 + 2 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a )
1241     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}})
1242 gezelter 3906 %6
1243 gezelter 4198 +2 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot \hat{\mathbf{r}})
1244     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b ) \Bigr] w_o(r) \\
1245 gezelter 3906 % 7
1246 gezelter 3985 &+
1247 gezelter 4198 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot \hat{\mathbf{r}})
1248     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}}) w_m(r) \hat{\mathbf{r}} \end{split}
1249 gezelter 3985 \end{align}
1250 gezelter 4198 Note that the forces for higher multipoles on site $a$
1251     interacting with those of lower order on site $b$ can be
1252 gezelter 3985 obtained by swapping indices in the expressions above.
1253    
1254 gezelter 3906 %
1255 gezelter 3985 % Torques SECTION -----------------------------------------------------------------------------------------
1256 gezelter 3906 %
1257     \subsection{Torques}
1258 gezelter 3989
1259 gezelter 3906 %
1260 gezelter 4183 The torques for the three methods are given in space-frame
1261     coordinates:
1262 gezelter 3906 %
1263     %
1264 gezelter 3985 \begin{align}
1265 gezelter 4198 \mathbf{\tau}_{b C_a \mathbf{D}_b} =&
1266     C_a (\hat{\mathbf{r}} \times \mathbf{D}_b) v_{11}(r) \\
1267 gezelter 3906 %
1268     %
1269     %
1270 gezelter 4198 \mathbf{\tau}_{b C_a \mathsf{Q}_b} =&
1271     2C_a
1272     \hat{\mathbf{r}} \times ( \mathsf{Q}_b \cdot \hat{\mathbf{r}}) v_{22}(r) \\
1273 gezelter 3906 %
1274     %
1275     %
1276 gezelter 3985 % \begin{equation}
1277     % \mathbf{\tau}_{{\bf a}D_{\bf a}C_{\bf b}} =
1278     % -\frac{C_{\bf b}}{4\pi \epsilon_0}
1279     % (\hat{r} \times \mathbf{D}_{\mathbf{a}}) v_{11}(r)
1280     % \end{equation}
1281 gezelter 3906 %
1282     %
1283     %
1284 gezelter 4198 \mathbf{\tau}_{a \mathbf{D}_a \mathbf{D}_b} =&
1285     \mathbf{D}_a \times \mathbf{D}_b v_{21}(r)
1286 gezelter 3906 % 2
1287 gezelter 3985 -
1288 gezelter 4198 (\hat{\mathbf{r}} \times \mathbf{D}_a )
1289     (\hat{\mathbf{r}} \cdot \mathbf{D}_b ) v_{22}(r)\\
1290 gezelter 3906 %
1291     %
1292     %
1293 gezelter 3985 % \begin{equation}
1294     % \mathbf{\tau}_{{\bf b}D_{\bf a}D_{\bf b}} =
1295     % -\frac{1}{4\pi \epsilon_0} \mathbf{D}_{\mathbf {a}} \times \mathbf{D}_{\mathbf{b}} v_{21}(r)
1296     % % 2
1297     % +\frac{1}{4\pi \epsilon_0}
1298     % (\hat{r} \cdot \mathbf{D}_{\mathbf {a}} )
1299     % (\hat{r} \times \mathbf{D}_{\mathbf {b}} ) v_{22}(r)
1300     % \end{equation}
1301 gezelter 3906 %
1302     %
1303     %
1304 gezelter 4198 \mathbf{\tau}_{a \mathbf{D}_a \mathsf{Q}_b} =&
1305 gezelter 3985 \Bigl[
1306 gezelter 4198 -\text{Tr}\mathsf{Q}_b
1307     (\hat{\mathbf{r}} \times \mathbf{D}_a )
1308     +2 \mathbf{D}_a \times
1309     (\mathsf{Q}_b \cdot \hat{\mathbf{r}})
1310 gezelter 3906 \Bigr] v_{31}(r)
1311     % 3
1312 gezelter 4198 - (\hat{\mathbf{r}} \times \mathbf{D}_a )
1313     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}}) v_{32}(r)\\
1314 gezelter 3906 %
1315     %
1316     %
1317 gezelter 4198 \mathbf{\tau}_{b \mathbf{D}_a \mathsf{Q}_b} =&
1318 gezelter 3985 \Bigl[
1319 gezelter 4198 +2 ( \mathbf{D}_a \cdot \mathsf{Q}_b ) \times
1320     \hat{\mathbf{r}}
1321     -2 \mathbf{D}_a \times
1322     (\mathsf{Q}_b \cdot \hat{\mathbf{r}})
1323 gezelter 3906 \Bigr] v_{31}(r)
1324     % 2
1325 gezelter 3985 +
1326 gezelter 4198 (\hat{\mathbf{r}} \cdot \mathbf{D}_a)
1327     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b) \times \hat{\mathbf{r}} v_{32}(r)\\
1328 gezelter 3906 %
1329     %
1330     %
1331 gezelter 3985 % \begin{equation}
1332     % \mathbf{\tau}_{{\bf a}Q_{\bf a}D_{\bf b}} =
1333     % \frac{1}{4\pi \epsilon_0} \Bigl[
1334     % -2 (\mathbf{D}_{\mathbf{b}} \cdot \mathbf{Q}_{\mathbf{a}} ) \times \hat{r}
1335     % +2 \mathbf{D}_{\mathbf{b}} \times
1336     % (\mathbf{Q}_{\mathbf{a}} \cdot \hat{r})
1337     % \Bigr] v_{31}(r)
1338     % % 3
1339     % - \frac{2}{4\pi \epsilon_0}
1340     % (\hat{r} \cdot \mathbf{D}_{\mathbf{b}} )
1341     % (\hat{r} \cdot \mathbf
1342     % {Q}_{{\mathbf a}}) \times \hat{r} v_{32}(r)
1343     % \end{equation}
1344 gezelter 3906 %
1345     %
1346     %
1347 gezelter 3985 % \begin{equation}
1348     % \mathbf{\tau}_{{\bf b}Q_{\bf a}D_{\bf b}} =
1349     % \frac{1}{4\pi \epsilon_0} \Bigl[
1350     % \text{Tr}\mathbf{Q}_{\mathbf{a}}
1351     % (\hat{r} \times \mathbf{D}_{\mathbf{b}} )
1352     % +2 \mathbf{D}_{\mathbf{b}} \times
1353     % ( \mathbf{Q}_{\mathbf{a}} \cdot \hat{r}) \Bigr] v_{31}(r)
1354     % % 2
1355     % +\frac{1}{4\pi \epsilon_0}
1356     % (\hat{r} \times \mathbf{D}_{\mathbf{b}} )
1357     % (\hat{r} \cdot \mathbf{Q}_{{\mathbf a}} \cdot \hat{r}) v_{32}(r)
1358     % \end{equation}
1359 gezelter 3906 %
1360     %
1361     %
1362     \begin{split}
1363 gezelter 4198 \mathbf{\tau}_{a \mathsf{Q}_a \mathsf{Q}_b} =&
1364 gezelter 3985 -4
1365 gezelter 4198 \mathsf{Q}_a \times \mathsf{Q}_b
1366 gezelter 3906 v_{41}(r) \\
1367     % 2
1368 gezelter 3985 &+
1369 gezelter 4198 \Bigl[-2\text{Tr}\mathsf{Q}_b
1370     (\hat{\mathbf{r}} \cdot \mathsf{Q}_a ) \times \hat{\mathbf{r}}
1371     +4 \hat{\mathbf{r}} \times
1372     ( \mathsf{Q}_a \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}})
1373 gezelter 3906 % 3
1374 gezelter 4198 -4 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a )\times
1375     ( \mathsf{Q}_b \cdot \hat{\mathbf{r}} ) \Bigr] v_{42}(r) \\
1376 gezelter 3906 % 4
1377 gezelter 3985 &+ 2
1378 gezelter 4198 \hat{\mathbf{r}} \times ( \mathsf{Q}_a \cdot \hat{\mathbf{r}})
1379     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b \cdot \hat{\mathbf{r}}) v_{43}(r) \end{split}\\
1380 gezelter 3906 %
1381     %
1382     %
1383     \begin{split}
1384 gezelter 4198 \mathbf{\tau}_{b \mathsf{Q}_a \mathsf{Q}_b} =
1385 gezelter 3985 &4
1386 gezelter 4198 \mathsf{Q}_a \times \mathsf{Q}_b v_{41}(r) \\
1387 gezelter 3906 % 2
1388 gezelter 4198 &+ \Bigl[- 2\text{Tr}\mathsf{Q}_a
1389     (\hat{\mathbf{r}} \cdot \mathsf{Q}_b ) \times \hat{\mathbf{r}}
1390     -4 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot
1391     \mathsf{Q}_b ) \times
1392     \hat{\mathbf{r}}
1393     +4 ( \hat{\mathbf{r}} \cdot \mathsf{Q}_a ) \times
1394     ( \mathsf{Q}_b \cdot \hat{\mathbf{r}})
1395 gezelter 3906 \Bigr] v_{42}(r) \\
1396     % 4
1397 gezelter 3985 &+2
1398 gezelter 4198 (\hat{\mathbf{r}} \cdot \mathsf{Q}_a \cdot \hat{\mathbf{r}})
1399     \hat{\mathbf{r}} \times ( \mathsf{Q}_b \cdot \hat{\mathbf{r}}) v_{43}(r)\end{split}
1400 gezelter 3985 \end{align}
1401     %
1402     Here, we have defined the matrix cross product in an identical form
1403     as in Ref. \onlinecite{Smith98}:
1404     \begin{equation}
1405 gezelter 4198 \left[\mathsf{A} \times \mathsf{B}\right]_\alpha = \sum_\beta
1406     \left[\mathsf{A}_{\alpha+1,\beta} \mathsf{B}_{\alpha+2,\beta}
1407     -\mathsf{A}_{\alpha+2,\beta} \mathsf{B}_{\alpha+1,\beta}
1408 gezelter 3985 \right]
1409 gezelter 4179 \label{eq:matrixCross}
1410 gezelter 3906 \end{equation}
1411 gezelter 3985 where $\alpha+1$ and $\alpha+2$ are regarded as cyclic
1412     permuations of the matrix indices.
1413 gezelter 3980
1414 gezelter 3985 All of the radial functions required for torques are identical with
1415     the radial functions previously computed for the interaction energies.
1416 gezelter 4183 These are tabulated for all three methods in table
1417 gezelter 3985 \ref{tab:tableenergy}. The torques for higher multipoles on site
1418 gezelter 4198 $a$ interacting with those of lower order on site
1419     $b$ can be obtained by swapping indices in the expressions
1420 gezelter 3985 above.
1421    
1422 gezelter 3980 \section{Comparison to known multipolar energies}
1423    
1424     To understand how these new real-space multipole methods behave in
1425     computer simulations, it is vital to test against established methods
1426     for computing electrostatic interactions in periodic systems, and to
1427     evaluate the size and sources of any errors that arise from the
1428 gezelter 4183 real-space cutoffs. In this paper we test SP, TSF, and GSF
1429 gezelter 3990 electrostatics against analytical methods for computing the energies
1430     of ordered multipolar arrays. In the following paper, we test the new
1431     methods against the multipolar Ewald sum for computing the energies,
1432     forces and torques for a wide range of typical condensed-phase
1433     (disordered) systems.
1434 gezelter 3980
1435     Because long-range electrostatic effects can be significant in
1436     crystalline materials, ordered multipolar arrays present one of the
1437     biggest challenges for real-space cutoff methods. The dipolar
1438     analogues to the Madelung constants were first worked out by Sauer,
1439     who computed the energies of ordered dipole arrays of zero
1440     magnetization and obtained a number of these constants.\cite{Sauer}
1441     This theory was developed more completely by Luttinger and
1442 gezelter 3986 Tisza\cite{LT,LT2} who tabulated energy constants for the Sauer arrays
1443 gezelter 3990 and other periodic structures.
1444 gezelter 3986
1445 gezelter 3990 To test the new electrostatic methods, we have constructed very large,
1446     $N=$ 16,000~(bcc) arrays of dipoles in the orientations described in
1447     Ref. \onlinecite{LT}. These structures include ``A'' lattices with
1448     nearest neighbor chains of antiparallel dipoles, as well as ``B''
1449     lattices with nearest neighbor strings of antiparallel dipoles if the
1450     dipoles are contained in a plane perpendicular to the dipole direction
1451     that passes through the dipole. We have also studied the minimum
1452 gezelter 3980 energy structure for the BCC lattice that was found by Luttinger \&
1453 gezelter 4179 Tisza. The total electrostatic energy density for any of the arrays
1454     is given by:
1455 gezelter 3980 \begin{equation}
1456     E = C N^2 \mu^2
1457     \end{equation}
1458 gezelter 3990 where $C$ is the energy constant (equivalent to the Madelung
1459     constant), $N$ is the number of dipoles per unit volume, and $\mu$ is
1460     the strength of the dipole. Energy constants (converged to 1 part in
1461 gezelter 4199 $10^9$) are given in the supplemental information.\cite{supplemental}
1462 gezelter 3980
1463 gezelter 4172 \begin{figure}
1464 gezelter 4198 \includegraphics[width=\linewidth]{Dipoles_rcut_threeAlpha.eps}
1465 gezelter 4172 \caption{Convergence of the lattice energy constants as a function of
1466     cutoff radius (normalized by the lattice constant, $a$) for the new
1467     real-space methods. Three dipolar crystal structures were sampled,
1468     and the analytic energy constants for the three lattices are
1469     indicated with grey dashed lines. The left panel shows results for
1470 gezelter 4198 the undamped kernel ($1/r$), while the damped kernel, $B_0(r)$ was
1471     used in the center and right panels.}
1472 gezelter 4172 \label{fig:Dipoles_rCut}
1473     \end{figure}
1474 mlamichh 4163
1475 gezelter 3990 For the purposes of testing the energy expressions and the
1476     self-neutralization schemes, the primary quantity of interest is the
1477     analytic energy constant for the perfect arrays. Convergence to these
1478 gezelter 4198 constants are shown as a function of the cutoff radius, $r_c$, for
1479     three different values of the damping coefficient, $\alpha$ in
1480     Fig.\ref{fig:Dipoles_rCut}. We have simultaneously tested a hard
1481 gezelter 4183 cutoff (where the kernel is simply truncated at the cutoff radius) in
1482     addition to the three new methods.
1483 gezelter 3986
1484 gezelter 4174 The hard cutoff exhibits oscillations around the analytic energy
1485 gezelter 3988 constants, and converges to incorrect energies when the complementary
1486 gezelter 4179 error function damping kernel is used. The shifted potential (SP)
1487     converges to the correct energy smoothly by $r_c = 4.5 a$ even for the
1488 gezelter 4198 undamped case. This indicates that the shifting and the correction
1489     provided by the self term are required for obtaining accurate
1490     energies. The Taylor-shifted force (TSF) approximation appears to
1491     perturb the potential too much inside the cutoff region to provide
1492     accurate measures of the energy constants. GSF is a compromise,
1493     converging to the correct energies within $r_c = 6 a$.
1494 gezelter 4174
1495 gezelter 3986 {\it Quadrupolar} analogues to the Madelung constants were first
1496     worked out by Nagai and Nakamura who computed the energies of selected
1497     quadrupole arrays based on extensions to the Luttinger and Tisza
1498 gezelter 4174 approach.\cite{Nagai01081960,Nagai01091963}
1499 gezelter 3980
1500     In analogy to the dipolar arrays, the total electrostatic energy for
1501     the quadrupolar arrays is:
1502     \begin{equation}
1503 gezelter 3996 E = C N \frac{3\bar{Q}^2}{4a^5}
1504 gezelter 3980 \end{equation}
1505 gezelter 3996 where $a$ is the lattice parameter, and $\bar{Q}$ is the effective
1506     quadrupole moment,
1507     \begin{equation}
1508 gezelter 4198 \bar{Q}^2 = 2 \left(3 \mathsf{Q} : \mathsf{Q} - (\text{Tr} \mathsf{Q})^2 \right)
1509 gezelter 3996 \end{equation}
1510     for the primitive quadrupole as defined in Eq. \ref{eq:quadrupole}.
1511 gezelter 4198 (For the traceless quadrupole tensor, $\mathsf{\Theta} = 3 \mathsf{Q} - \text{Tr} \mathsf{Q}$,
1512     the effective moment, $\bar{Q}^2 = \frac{2}{3} \mathsf{\Theta} : \mathsf{\Theta}$.)
1513 gezelter 3980
1514 gezelter 4174 To test the new electrostatic methods for quadrupoles, we have
1515     constructed very large, $N=$ 8,000~(sc), 16,000~(bcc), and
1516     32,000~(fcc) arrays of linear quadrupoles in the orientations
1517     described in Ref. \onlinecite{Nagai01081960}. We have compared the
1518 gezelter 4179 energy constants for these low-energy configurations for linear
1519     quadrupoles. Convergence to these constants are shown as a function of
1520 gezelter 4198 the cutoff radius, $r_c$, for three different values of the damping
1521     parameter, $\alpha$ in Fig.~\ref{fig:Quadrupoles_rCut}.
1522 gezelter 4174
1523 gezelter 4172 \begin{figure}
1524 gezelter 4198 \includegraphics[width=\linewidth]{Quadrupoles_rcut_threeAlpha.eps}
1525 gezelter 4173 \caption{Convergence of the lattice energy constants as a function of
1526     cutoff radius (normalized by the lattice constant, $a$) for the new
1527 gezelter 4198 real-space methods. Three quadrupolar crystal structures were
1528 gezelter 4173 sampled, and the analytic energy constants for the three lattices
1529 gezelter 4198 are indicated with grey dashed lines. The left panel shows results
1530     for the undamped kernel ($1/r$), while the damped kernel, $B_0(r)$
1531     was used in the center and right panels. Note that for quadrupoles,
1532     $\alpha^* = 0.4$ overdamps contributions from repulsive orientations
1533     in the perfect crystal.}
1534 gezelter 4174 \label{fig:Quadrupoles_rCut}
1535 gezelter 4172 \end{figure}
1536    
1537 gezelter 4174 Again, we find that the hard cutoff exhibits oscillations around the
1538 gezelter 4179 analytic energy constants. The shifted potential (SP) approximation
1539     converges to the correct energy smoothly by $r_c = 3 a$ even for the
1540     undamped case. The Taylor-shifted force (TSF) approximation again
1541     appears to perturb the potential too much inside the cutoff region to
1542     provide accurate measures of the energy constants. GSF again provides
1543     a compromise between the two methods -- energies are converged by $r_c
1544     = 4.5 a$, and the approximation is not as perturbative at short range
1545     as TSF.
1546 gezelter 4172
1547 gezelter 4198 It is also useful to understand the behavior of the lattice energy
1548     constants for different values of the reduced damping parameter
1549     ($\alpha^* = \alpha a$) for the real-space methods. All of the methods
1550     (except for TSF) have excellent behavior for the undamped or
1551 gezelter 4199 weakly-damped cases. Overdamping can cause problems in perfect
1552     crystals for the quadrupoles in particular ({\it cf.} the right panel
1553     in Fig. \ref{fig:Quadrupoles_rCut}). In the perfect crystals, only a
1554     few orientations are being sampled. E.g. in the simple cubic (SC)
1555     lattice of linear quadrupoles aligned in the 111 direction, the
1556     nearest-neighbor quadrupoles only sample 3 distinct orientations
1557     relative to the vector between the sites. The damping alters the
1558     radial function for the direct quadrupolar contraction, $v_{41}(r)$,
1559     differently than the radial functions for the terms involving the
1560     product of the separation vector with the quadrupoles, $v_{42}(r)$ and
1561     $v_{43}(r)$. Because these terms are altered by different amounts by
1562     the complementary error function damping, the effect of damping is
1563     non-spherical for multipoles, and the balance between attractive and
1564     repulsive interactions in the crystal is therefore altered
1565     significantly in overdamped situations.
1566 gezelter 4198
1567 gezelter 4183 In the second paper in the series, we discuss how large values of
1568 gezelter 4198 $\alpha$ can perturb the force and torque vectors, but weakly-damped
1569     electrostatics appear to generate reasonable values for the total
1570     electrostatic energies under both the SP and GSF approximations. We
1571     also discuss the effects that $\alpha$ can have on convergence to the
1572     average electrostatic energies in liquids (which sample a much wider
1573     range of local orientations).
1574 gezelter 4174
1575 gezelter 3985 \section{Conclusion}
1576 gezelter 4174 We have presented three efficient real-space methods for computing the
1577 gezelter 4181 interactions between point multipoles. One of these (SP) is a
1578     multipolar generalization of Wolf's method that smoothly shifts
1579 gezelter 4198 electrostatic energies to zero at the cutoff radius. Two of these
1580 gezelter 4181 methods (GSF and TSF) also smoothly truncate the forces and torques
1581     (in addition to the energies) at the cutoff radius, making them
1582     attractive for both molecular dynamics and Monte Carlo simulations. We
1583     find that the Gradient-Shifted Force (GSF) and the Shifted-Potential
1584     (SP) methods converge rapidly to the correct lattice energies for
1585     ordered dipolar and quadrupolar arrays, while the Taylor-Shifted Force
1586 gezelter 4199 (TSF) is too severe an approximation to provide convergence to lattice
1587     energies within reasonable cutoff radii.
1588 gezelter 3980
1589 gezelter 4199 Although the TSF method appears to perform poorly for the analytical
1590     energy constants, the structure of the radial functions used in the
1591     force and torque expressions in the other two methods would not have
1592     been revealed without first developing the TSF approach. TSF also
1593     generates a set of electrostatic kernels that have multiple
1594     derivatives that vanish at the cutoff radius, a property that is
1595     valuable in estimating dielectric constants using the conducting
1596     boundary fluctuation formula.\cite{Izvekov:2008wo}
1597 gezelter 4198
1598 gezelter 3985 In most cases, GSF can obtain nearly quantitative agreement with the
1599     lattice energy constants with reasonably small cutoff radii. The only
1600     exception we have observed is for crystals which exhibit a bulk
1601     macroscopic dipole moment (e.g. Luttinger \& Tisza's $Z_1$ lattice).
1602     In this particular case, the multipole neutralization scheme can
1603     interfere with the correct computation of the energies. We note that
1604     the energies for these arrangements are typically much larger than for
1605     crystals with net-zero moments, so this is not expected to be an issue
1606     in most simulations.
1607 gezelter 3980
1608 gezelter 4199 Relatively weak damping is sufficient to converge to the analytical
1609     energy constants within moderately short cutoff distances. Because
1610     overdamping can present additional issues with higher order
1611     multipoles, our results indicate that the damping coefficient should
1612     be taken as small as possible, and that the {\it undamped} GSF and SP
1613     methods may be the best choice in crystalline systems.
1614    
1615 gezelter 4179 The techniques used here to derive the force, torque and energy
1616     expressions can be extended to higher order multipoles, although some
1617     of the objects (e.g. the matrix cross product in
1618     Eq. \ref{eq:matrixCross}) will need to be generalized for higher-rank
1619     tensors. We also note that the definitions of the multipoles used
1620     here are in a primitive form, and these need some care when comparing
1621     with experiment or other computational techniques.
1622    
1623 gezelter 3985 In large systems, these new methods can be made to scale approximately
1624     linearly with system size, and detailed comparisons with the Ewald sum
1625     for a wide range of chemical environments follows in the second paper.
1626 gezelter 3980
1627 gezelter 3906 \begin{acknowledgments}
1628 gezelter 3985 JDG acknowledges helpful discussions with Christopher
1629     Fennell. Support for this project was provided by the National
1630 gezelter 4172 Science Foundation under grant CHE-1362211. Computational time was
1631 gezelter 3985 provided by the Center for Research Computing (CRC) at the
1632     University of Notre Dame.
1633 gezelter 3906 \end{acknowledgments}
1634    
1635 gezelter 3984 \newpage
1636 gezelter 3906 \appendix
1637    
1638 gezelter 3984 \section{Smith's $B_l(r)$ functions for damped-charge distributions}
1639 gezelter 3985 \label{SmithFunc}
1640 gezelter 3984 The following summarizes Smith's $B_l(r)$ functions and includes
1641     formulas given in his appendix.\cite{Smith98} The first function
1642     $B_0(r)$ is defined by
1643 gezelter 3906 %
1644     \begin{equation}
1645 gezelter 4198 B_0(r)=\frac{\text{erfc}(\alpha r)}{r} = \frac{2}{\sqrt{\pi}r}
1646 gezelter 3906 \int_{\alpha r}^{\infty} \text{e}^{-s^2} ds .
1647     \end{equation}
1648     %
1649     The first derivative of this function is
1650     %
1651     \begin{equation}
1652     \frac{dB_0(r)}{dr}=-\frac{1}{r^2}\text{erfc}(\alpha r)
1653     -\frac{2\alpha}{r\sqrt{\pi}}\text{e}^{-{\alpha}^2r^2}
1654     \end{equation}
1655     %
1656 gezelter 3984 which can be used to define a function $B_1(r)$:
1657 gezelter 3906 %
1658     \begin{equation}
1659     B_1(r)=-\frac{1}{r}\frac{dB_0(r)}{dr}
1660     \end{equation}
1661     %
1662 gezelter 3984 In general, the recurrence relation,
1663 gezelter 3906 \begin{equation}
1664     B_l(r)=-\frac{1}{r}\frac{dB_{l-1}(r)}{dr}
1665     = \frac{1}{r^2} \left[ (2l-1)B_{l-1}(r) + \frac {(2\alpha^2)^l}{\alpha \sqrt{\pi}}
1666     \text{e}^{-{\alpha}^2r^2}
1667 gezelter 3984 \right] ,
1668 gezelter 3906 \end{equation}
1669 gezelter 4183 is very useful for building up higher derivatives. As noted by Smith,
1670     it is possible to approximate the $B_l(r)$ functions,
1671 gezelter 3906 %
1672     \begin{equation}
1673     B_l(r)=\frac{(2l)!}{l!2^lr^{2l+1}} - \frac {(2\alpha^2)^{l+1}}{(2l+1)\alpha \sqrt{\pi}}
1674     +\text{O}(r) .
1675     \end{equation}
1676 gezelter 3984 \newpage
1677     \section{The $r$-dependent factors for TSF electrostatics}
1678 gezelter 4183 \label{radialTSF}
1679 gezelter 3906
1680     Using the shifted damped functions $f_n(r)$ defined by:
1681     %
1682     \begin{equation}
1683 gezelter 3984 f_n(r)= B_0(r) -\sum_{m=0}^{n+1} \frac {(r-r_c)^m}{m!} B_0^{(m)}(r_c) ,
1684 gezelter 3906 \end{equation}
1685     %
1686 gezelter 3984 where the superscript $(m)$ denotes the $m^\mathrm{th}$ derivative. In
1687     this Appendix, we provide formulas for successive derivatives of this
1688     function. (If there is no damping, then $B_0(r)$ is replaced by
1689     $1/r$.) First, we find:
1690 gezelter 3906 %
1691     \begin{equation}
1692     \frac{\partial f_n}{\partial r_\alpha}=\hat{r}_\alpha \frac{d f_n}{d r} .
1693     \end{equation}
1694     %
1695 gezelter 3984 This formula clearly brings in derivatives of Smith's $B_0(r)$
1696     function, and we define higher-order derivatives as follows:
1697 gezelter 3906 %
1698 gezelter 3984 \begin{align}
1699     g_n(r)=& \frac{d f_n}{d r} =
1700     B_0^{(1)}(r) -\sum_{m=0}^{n} \frac {(r-r_c)^m}{m!} B_0^{(m+1)}(r_c) \\
1701     h_n(r)=& \frac{d^2f_n}{d r^2} =
1702     B_0^{(2)}(r) -\sum_{m=0}^{n-1} \frac {(r-r_c)^m}{m!} B_0^{(m+2)}(r_c) \\
1703     s_n(r)=& \frac{d^3f_n}{d r^3} =
1704     B_0^{(3)}(r) -\sum_{m=0}^{n-2} \frac {(r-r_c)^m}{m!} B_0^{(m+3)}(r_c) \\
1705     t_n(r)=& \frac{d^4f_n}{d r^4} =
1706     B_0^{(4)}(r) -\sum_{m=0}^{n-3} \frac {(r-r_c)^m}{m!} B_0^{(m+4)}(r_c) \\
1707     u_n(r)=& \frac{d^5f_n}{d r^5} =
1708     B_0^{(5)}(r) -\sum_{m=0}^{n-4} \frac {(r-r_c)^m}{m!} B_0^{(m+5)}(r_c) .
1709     \end{align}
1710 gezelter 3906 %
1711 gezelter 3984 We note that the last function needed (for quadrupole-quadrupole interactions) is
1712 gezelter 3906 %
1713     \begin{equation}
1714 gezelter 3984 u_4(r)=B_0^{(5)}(r) - B_0^{(5)}(r_c) .
1715 gezelter 3906 \end{equation}
1716 gezelter 3989 % The functions
1717     % needed are listed schematically below:
1718     % %
1719     % \begin{eqnarray}
1720     % f_0 \quad f_1 \qquad \qquad \quad & \nonumber \\
1721     % g_0 \quad g_1 \quad g_2 \quad g_3 \quad &g_4 \nonumber \\
1722     % h_1 \quad h_2 \quad h_3 \quad &h_4 \nonumber \\
1723     % s_2 \quad s_3 \quad &s_4 \nonumber \\
1724     % t_3 \quad &t_4 \nonumber \\
1725     % &u_4 \nonumber .
1726     % \end{eqnarray}
1727 gezelter 3984 The functions $f_n(r)$ to $u_n(r)$ can be computed recursively and
1728 gezelter 3989 stored on a grid for values of $r$ from $0$ to $r_c$. Using these
1729     functions, we find
1730 gezelter 3906 %
1731 gezelter 3984 \begin{align}
1732     \frac{\partial f_n}{\partial r_\alpha} =&r_\alpha \frac {g_n}{r} \label{eq:b9}\\
1733     \frac{\partial^2 f_n}{\partial r_\alpha \partial r_\beta} =&\delta_{\alpha \beta}\frac {g_n}{r}
1734     +r_\alpha r_\beta \left( -\frac{g_n}{r^3} +\frac{h_n}{r^2}\right) \\
1735 gezelter 3989 \frac{\partial^3 f_n}{\partial r_\alpha \partial r_\beta \partial r_\gamma} =&
1736 gezelter 3906 \left( \delta_{\alpha \beta} r_\gamma + \delta_{\alpha \gamma} r_\beta +
1737     \delta_{ \beta \gamma} r_\alpha \right)
1738 gezelter 3989 \left( -\frac{g_n}{r^3} +\frac{h_n}{r^2} \right) \nonumber \\
1739     & + r_\alpha r_\beta r_\gamma
1740 gezelter 3984 \left( \frac{3g_n}{r^5}-\frac{3h_n}{r^4} +\frac{s_n}{r^3} \right) \\
1741 gezelter 3989 \frac{\partial^4 f_n}{\partial r_\alpha \partial r_\beta \partial
1742     r_\gamma \partial r_\delta} =&
1743 gezelter 3906 \left( \delta_{\alpha \beta} \delta_{\gamma \delta}
1744     + \delta_{\alpha \gamma} \delta_{\beta \delta}
1745     +\delta_{ \beta \gamma} \delta_{\alpha \delta} \right)
1746     \left( - \frac{g_n}{r^3} + \frac{h_n}{r^2} \right) \nonumber \\
1747 gezelter 3984 &+ \left( \delta_{\alpha \beta} r_\gamma r_\delta
1748     + \text{5 permutations}
1749 gezelter 3906 \right) \left( \frac{3 g_n}{r^5} - \frac{3h_n}{r^4} + \frac{s_n}{r^3}
1750     \right) \nonumber \\
1751 gezelter 3984 &+ r_\alpha r_\beta r_\gamma r_\delta
1752 gezelter 3906 \left( -\frac{15g_n}{r^7} + \frac{15h_n}{r^6} - \frac{6s_n}{r^5}
1753 gezelter 3984 + \frac{t_n}{r^4} \right)\\
1754 gezelter 3906 \frac{\partial^5 f_n}
1755 gezelter 3989 {\partial r_\alpha \partial r_\beta \partial r_\gamma \partial
1756     r_\delta \partial r_\epsilon} =&
1757 gezelter 3906 \left( \delta_{\alpha \beta} \delta_{\gamma \delta} r_\epsilon
1758 gezelter 3984 + \text{14 permutations} \right)
1759 gezelter 3906 \left( \frac{3g_n}{r^5}-\frac{3h_n}{r^4} +\frac{s_n}{r^3} \right) \nonumber \\
1760 gezelter 3984 &+ \left( \delta_{\alpha \beta} r_\gamma r_\delta r_\epsilon
1761     + \text{9 permutations}
1762 gezelter 3906 \right) \left(- \frac{15g_n}{r^7}+\frac{15h_n}{r^7} -\frac{6s_n}{r^5} +\frac{t_n}{r^4}
1763     \right) \nonumber \\
1764 gezelter 3984 &+ r_\alpha r_\beta r_\gamma r_\delta r_\epsilon
1765 gezelter 3906 \left( \frac{105g_n}{r^9} - \frac{105h_n}{r^8} + \frac{45s_n}{r^7}
1766 gezelter 3984 - \frac{10t_n}{r^6} +\frac{u_n}{r^5} \right) \label{eq:b13}
1767     \end{align}
1768 gezelter 3906 %
1769     %
1770     %
1771 gezelter 3984 \newpage
1772     \section{The $r$-dependent factors for GSF electrostatics}
1773 gezelter 4183 \label{radialGSF}
1774 gezelter 3906
1775 gezelter 3984 In Gradient-shifted force electrostatics, the kernel is not expanded,
1776 gezelter 4183 and the expansion is carried out on the individual terms in the
1777     multipole interaction energies. For damped charges, this still brings
1778     multiple derivatives of the Smith's $B_0(r)$ function into the
1779     algebra. To denote these terms, we generalize the notation of the
1780     previous appendix. For either $f(r)=1/r$ (undamped) or $f(r)=B_0(r)$
1781     (damped),
1782 gezelter 3906 %
1783 gezelter 3984 \begin{align}
1784 gezelter 4179 g(r) &= \frac{df}{d r} && &&=-\frac{1}{r^2}
1785     &&\mathrm{or~~~} -rB_1(r) \\
1786     h(r) &= \frac{dg}{d r} &&= \frac{d^2f}{d r^2} &&= \frac{2}{r^3} &&\mathrm{or~~~}-B_1(r) + r^2 B_2(r) \\
1787     s(r) &= \frac{dh}{d r} &&= \frac{d^3f}{d r^3} &&=-\frac{6}{r^4}&&\mathrm{or~~~}3rB_2(r) - r^3 B_3(r)\\
1788     t(r) &= \frac{ds}{d r} &&= \frac{d^4f}{d r^4} &&= \frac{24}{r^5} &&\mathrm{or~~~} 3
1789     B_2(r) - 6r^2 B_3(r) + r^4 B_4(r) \\
1790     u(r) &= \frac{dt}{d r} &&= \frac{d^5f}{d r^5} &&=-\frac{120}{r^6} &&\mathrm{or~~~} -15
1791     r B_3(r) + 10 r^3B_4(r) -r^5B_5(r).
1792 gezelter 3984 \end{align}
1793 gezelter 3906 %
1794 gezelter 4179 For undamped charges, Table I lists these derivatives under the Bare
1795     Coulomb column. Equations \ref{eq:b9} to \ref{eq:b13} are still
1796 gezelter 3989 correct for GSF electrostatics if the subscript $n$ is eliminated.
1797 gezelter 3906
1798 gezelter 4172 \newpage
1799    
1800 gezelter 3980 \bibliography{multipole}
1801    
1802 gezelter 3906 \end{document}
1803     %
1804     % ****** End of file multipole.tex ******