44 |
|
%\preprint{AIP/123-QED} |
45 |
|
|
46 |
|
\title{Real space alternatives to the Ewald |
47 |
< |
Sum. I. Taylor-shifted and Gradient-shifted electrostatics for multipoles} |
47 |
> |
Sum. I. Shifted electrostatics for multipoles} |
48 |
|
|
49 |
|
\author{Madan Lamichhane} |
50 |
|
\affiliation{Department of Physics, University |
1446 |
|
self-neutralization schemes, the primary quantity of interest is the |
1447 |
|
analytic energy constant for the perfect arrays. Convergence to these |
1448 |
|
constants are shown as a function of both the cutoff radius, $r_c$, |
1449 |
< |
and the damping parameter, $\alpha$ in Figs. \ref{fig:Dipoles_rCut} |
1449 |
> |
and the damping parameter, $\alpha$ in Figs.\ref{fig:Dipoles_rCut} |
1450 |
|
and \ref{fig:Dipoles_alpha}. We have simultaneously tested a hard |
1451 |
|
cutoff (where the kernel is simply truncated at the cutoff radius), as |
1452 |
|
well as a shifted potential (SP) form which includes a |
1454 |
|
forces and torques smoothly at the cutoff radius. The SP method is |
1455 |
|
essentially an extension of the original Wolf method for multipoles. |
1456 |
|
|
1457 |
< |
The Hard cutoff exhibits oscillations around the analytic energy |
1457 |
> |
The hard cutoff exhibits oscillations around the analytic energy |
1458 |
|
constants, and converges to incorrect energies when the complementary |
1459 |
|
error function damping kernel is used. The shifted potential (SP) and |
1460 |
|
gradient-shifted force (GSF) approximations converge to the correct |
1461 |
< |
energy smoothly by $r_c / 6 a$ even for the undamped case. This |
1461 |
> |
energy smoothly by $r_c = 6 a$ even for the undamped case. This |
1462 |
|
indicates that the correction provided by the self term is required |
1463 |
|
for obtaining accurate energies. The Taylor-shifted force (TSF) |
1464 |
|
approximation appears to perturb the potential too much inside the |
1465 |
|
cutoff region to provide accurate measures of the energy constants. |
1466 |
+ |
|
1467 |
|
{\it Quadrupolar} analogues to the Madelung constants were first |
1468 |
|
worked out by Nagai and Nakamura who computed the energies of selected |
1469 |
|
quadrupole arrays based on extensions to the Luttinger and Tisza |
1470 |
< |
approach.\cite{Nagai01081960,Nagai01091963} We have compared the |
1470 |
< |
energy constants for the lowest energy configurations for linear |
1471 |
< |
quadrupoles. |
1470 |
> |
approach.\cite{Nagai01081960,Nagai01091963} |
1471 |
|
|
1472 |
|
In analogy to the dipolar arrays, the total electrostatic energy for |
1473 |
|
the quadrupolar arrays is: |
1483 |
|
(For the traceless quadrupole tensor, $\Theta = 3 Q - \text{Tr} Q$, |
1484 |
|
the effective moment, $\bar{Q}^2 = \frac{2}{3} \Theta : \Theta$.) |
1485 |
|
|
1486 |
+ |
To test the new electrostatic methods for quadrupoles, we have |
1487 |
+ |
constructed very large, $N=$ 8,000~(sc), 16,000~(bcc), and |
1488 |
+ |
32,000~(fcc) arrays of linear quadrupoles in the orientations |
1489 |
+ |
described in Ref. \onlinecite{Nagai01081960}. We have compared the |
1490 |
+ |
energy constants for the lowest energy configurations for these linear |
1491 |
+ |
quadrupoles. Convergence to these constants are shown as a function |
1492 |
+ |
of both the cutoff radius, $r_c$, and the damping parameter, $\alpha$ |
1493 |
+ |
in Figs.~\ref{fig:Quadrupoles_rCut} and \ref{fig:Quadrupoles_alpha}. |
1494 |
+ |
|
1495 |
|
\begin{figure} |
1496 |
|
\includegraphics[width=\linewidth]{Quadrupoles_rcutConvergence.pdf} |
1497 |
|
\caption{Convergence of the lattice energy constants as a function of |
1501 |
|
are indicated with grey dashed lines. The left panel shows results |
1502 |
|
for the undamped kernel ($1/r$), while the damped error function |
1503 |
|
kernel, $B_0(r)$ was used in the right panel.} |
1504 |
< |
\label{fig:QuadrupolesrcutCovergence} |
1504 |
> |
\label{fig:Quadrupoles_rCut} |
1505 |
|
\end{figure} |
1506 |
|
|
1507 |
|
|
1508 |
|
\begin{figure}[!htbp] |
1509 |
< |
\includegraphics[width=3.5in]{Quadrupoles_alphaConvergence-crop.pdf} |
1510 |
< |
\caption{Convergence to the analytic energy constants as a function of |
1511 |
< |
cutoff damping alpha for the different |
1512 |
< |
real-space methods for (a) dipolar and (b) quadrupolar crystals.The energy constants for hard, SP, GSF, TSF, and analytic methods are represented by the black sold-circle, red solid-square, green solid-diamond, and grey dashed line respectively. |
1513 |
< |
The left panel shows results for the undamped kernel ($1/r$), while the damped |
1514 |
< |
error function kernel, $B_0(r)$ was used in the right panel. } |
1515 |
< |
\label{fig:Quadrupoles_alphaCovergence-crop.pdf} |
1509 |
> |
\includegraphics[width=\linewidth]{Quadrupoles_newAlpha.pdf} |
1510 |
> |
\caption{Convergence to the lattice energy constants as a function of |
1511 |
> |
the reduced damping parameter ($\alpha^* = \alpha a$) for the |
1512 |
> |
different real-space methods in the same three quadrupolar crystals |
1513 |
> |
in Figure \ref{fig:Quadrupoles_rCut}. The left panel shows |
1514 |
> |
results for a relatively small cutoff radius ($r_c = 4.5 a$) while a |
1515 |
> |
larger cutoff radius ($r_c = 6 a$) was used in the right panel. } |
1516 |
> |
\label{fig:Quadrupoles_alpha} |
1517 |
|
\end{figure} |
1518 |
|
|
1519 |
+ |
Again, we find that the hard cutoff exhibits oscillations around the |
1520 |
+ |
analytic energy constants. The shifted potential (SP) and |
1521 |
+ |
gradient-shifted force (GSF) approximations converge to the correct |
1522 |
+ |
energy smoothly by $r_c = 4 a$ even for the undamped case. The |
1523 |
+ |
Taylor-shifted force (TSF) approximation again appears to perturb the |
1524 |
+ |
potential too much inside the cutoff region to provide accurate |
1525 |
+ |
measures of the energy constants. |
1526 |
|
|
1527 |
+ |
|
1528 |
|
\section{Conclusion} |
1529 |
< |
We have presented two efficient real-space methods for computing the |
1529 |
> |
We have presented three efficient real-space methods for computing the |
1530 |
|
interactions between point multipoles. These methods have the benefit |
1531 |
|
of smoothly truncating the energies, forces, and torques at the cutoff |
1532 |
|
radius, making them attractive for both molecular dynamics (MD) and |
1534 |
|
(GSF) and the Shifted-Potential (SP) methods converge rapidly to the |
1535 |
|
correct lattice energies for ordered dipolar and quadrupolar arrays, |
1536 |
|
while the Taylor-Shifted Force (TSF) is too severe an approximation to |
1537 |
< |
provide accurate convergence to lattice energies. |
1537 |
> |
provide accurate convergence to lattice energies. |
1538 |
|
|
1539 |
|
In most cases, GSF can obtain nearly quantitative agreement with the |
1540 |
|
lattice energy constants with reasonably small cutoff radii. The only |