1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % LaTeX 2E document, using iopart.cls with graphicx and iopams packages. %
3: % 10 pages, 3 figures (4 eps files). Typeset twice! %
4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
5: % Overlapping Unit Cells in 3-d Quasicrystal Structure %
6: % by Helen Au-Yang and Jacques H.H. Perk, %
7: % Department of Physics, Oklahoma State University, %
8: % Stillwater, OK 74078-3072, USA. Email: perk@okstate.edu %
9: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
10: \documentclass[12pt]{iopart}
11: %Uncomment next line if AMS fonts required
12: \usepackage{iopams,graphicx}
13: \begin{document}
14:
15: \title[Overlapping Unit Cells in 3-d Quasicrystal Structure]{Overlapping
16: Unit Cells in 3-d Quasicrystal Structure}
17:
18: \author{Helen Au-Yang and Jacques H H Perk}
19:
20: \address{Department of Physics, Oklahoma State University, %\\
21: 145 Physical Sciences, Stillwater, OK 74078-3072, USA}
22: \ead{perk@okstate.edu}
23: \begin{abstract}
24: A 3-dimensional quasiperiodic lattice, with overlapping unit cells and
25: periodic in one direction, is constructed using grid and projection
26: methods pioneered by de Bruijn. Each unit cell consists of 26 points, of
27: which 22 are the vertices of a convex polytope $\cal P$, and 4 are
28: interior points also shared with other neighboring unit cells. Using
29: Kronecker's theorem the frequencies of all possible types of overlapping
30: are found.
31: \end{abstract}
32:
33: %Uncomment for PACS numbers title message
34: \pacs{05.50.+q, 61.44.Br}
35: % Keywords required only for MST, PB, PMB, PM, JOA, JOB?
36: \vspace{2pc}
37: %\noindent{\it Keywords}: Quasiperiodicity, pentagrid, Penrose tiles,
38: %overlapping unit cells
39: % Uncomment for Submitted to journal title message
40: %\submitto{\JPA}
41: % Comment out if separate title page not required
42: %\maketitle
43:
44: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45: \font\mymsbm=msbm10 at 12pt
46: \def\myb{\boldsymbol}
47: \def\br{{\myb R}}
48: \def\bd{{\myb d}}
49: \def\bk{{\myb k}}
50: \def\bD{{\myb D}}
51: \def\bw{{\myb w}}
52: \def\bW{{\myb W}}
53: \def\bP{{\myb P}}
54: \def\bQ{{\myb Q}}
55: \def\bn{{\myb n}}
56: \def\be{{\myb e}}
57: \def\T{{\rm T}}
58: %\def\cal{\mathcal}
59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60: \section{Introduction}
61: Since the startling discovery of five-fold symmetry in quasiperiodic
62: materials in 1984 \cite{SBGC}, a great deal of research has been done on
63: this subject by both physicists and mathematicians. Originally,
64: quasicrystals were constructed by filling the space aperiodically with
65: nonoverlapping tiles, such as in Penrose tilings
66: \cite{Penrose,GrSh,Bruijn1}\footnote{For a recent review of the
67: theory of quasicrystals, see, e.g., \cite{KrPa}.}. However, recently,
68: Gummelt
69: \cite{Gummelt} motivated by physical considerations, proposed a
70: description of quasicrystals in terms of overlappings of decagons. Further
71: research \cite{SJ,SJSTAT,LR,LRK,Kramer1,DuGr,GGA,ALS} has shown that this
72: may be a more sensible way to understand quasicrystalline materials---made
73: of overlapping unit cells sharing atoms of nearby neighbors \cite{SJSTAT}.
74: Here we use a multigrid method to produce a new example of 3-dimensional
75: overlapping unit cells.
76:
77: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
78: \section{Multigrid construction}
79: It is well-known that a Penrose tiling can be obtained by the projection
80: of a slab of the 5-d euclidean lattice onto a particular 2-d plane
81: ${\cal D}$ \cite{Bruijn1,Bruijn2,GRh}, and hence its diffraction pattern
82: \cite{Elser,DK,Mackay2} has ten-fold symmetry. It is also known
83: that not all lattice points $\bk$ in ${\mbox{\mymsbm Z}}^5$ are allowed
84: (in the sense that they can be mapped onto vertices of a Penrose tiling);
85: only those points whose projections into the 3-dimensional orthogonal
86: space ${\cal W}$ are inside the window of acceptance \cite{Bruijn2,BJKS}
87: contribute. The window has been shown \cite {Bruijn2} to be the projection
88: of the 5-d unit cell Cu(5) with $2^5$ vertices into this 3-d space
89: $\cal W$. Each facet shared by two neighboring 5-d unit cell cubes is
90: 4-dimensional and when projected into 3-d space it produces a polyhedron
91: $\cal K$ with 12 faces. Therefore, the projections of two adjacent 5-d
92: unit cells into 3d must share a common projected facet $\cal K$. Thus the
93: idea of overlapping decagons must have its extension to three dimension,
94: by projecting the 5-d lattice into the space $\cal W$.
95:
96: If $\bd_j$ are the generators of the plane ${\cal D}$ and $\bw_j$ are the
97: generators of its orthogonal space ${\cal W}$, then the projection
98: operators are the matrices $\bD^\T=(\bd_0,\ldots,\bd_4)$ and
99: $\bW^\T=(\bw_0,\ldots,\bw_4)$, such that $\bD^\T\bW=\bW^\T\bD=0$, where
100: the superscript T denotes matrix transpose. More specifically, we choose
101: \begin{equation}
102: \bd^\T_j=(\cos j\theta,\sin j\theta),\quad
103: \bw^\T_j=(\cos 2j\theta,\sin 2j\theta,1)=(\bd^\T_{2j},1),
104: \label{generators}
105: \end{equation}
106: where $j=0,\ldots,4$ and $\theta=2\pi/5$. Using notations and ideas
107: introduced by de Bruijn \cite{Bruijn1}, we consider the five grids
108: consisting of bundles of equidistant planes defined by
109: \begin{equation}
110: x\cos 2j\theta+y\sin 2j\theta+z+\gamma_j=\bw^\T_j\br+\gamma_j=k_j,
111: \label{grid}
112: \end{equation}
113: for $\quad j=0,\ldots,4$, $k_j\in{\mbox{\mymsbm Z}}$. In (\ref{grid}),
114: $\br^\T=(x,y,z)$, and the $\gamma_j$ are real numbers which shift the
115: grids from the origin. We denote their sum by
116: \begin{equation}
117: \gamma_{0}+\gamma_{1}+\gamma_{2}+\gamma_{3}+\gamma_{4}=c.
118: \label{shiftc}
119: \end{equation}
120: Without loss of generality, we may restrict $c$ to $0\le c<1$.
121:
122: It has been shown by de Bruijn \cite{Bruijn1} that the Penrose tiling
123: associated with a 2-d pentagrid has simple matching rules only for
124: $c=0$. For $0<c<1$ the corresponding generalized Penrose tilings do not
125: satisfy simple matching rules, and have seven different sets of vertices
126: corresponding to the different intervals of $c$ \cite{APwindow,APquasi}.
127: Nevertheless, the diffraction patterns are believed to be the same
128: for all values of $c$ \cite {LSt,SoSt,IsYa}.
129:
130: Let the integer $k_j$ be assigned to all points sandwiched between the
131: grid planes defined by $k_j-1$ and $k_j$. Then, five integers
132: \begin{equation}
133: K_j(\br)=\lceil\bw^\T_j\br+\gamma_j\rceil,\quad j=0,\ldots,4,
134: \label{mesh}
135: \end{equation}
136: with $\lceil x\rceil$ the smallest integer greater than or equal to $x$,
137: are uniquely assigned to every point $\br$ in ${\mbox{\mymsbm R}}^3$.
138: A mesh in ${\mbox{\mymsbm R}}^3$ is now an interior volume, enclosed by
139: grid planes, containing points with the same five integers. One next
140: maps each mesh to a vertex in $\cal W$ by
141: \begin{equation}
142: \fl
143: {\myb g}(\br)=\sum_{j=0}^4 K_j(\br)\bw_j=\bW^\T{\myb K}(\br),\quad
144: {\myb K}^\T(\br)=(K_0(\br),\ldots,K_4(\br)).
145: \label{map}\end{equation}
146: The resulting collection of vertices
147: ${\cal L}=\{{\myb g}(\br)|\br\in{\mbox{\mymsbm R}}^3\}$ is a
148: 3-dimensional aperiodic lattice. It has been proven by de Bruijn
149: \cite{Bruijn2}, that a point $\bk$ in ${\mbox{\mymsbm Z}}^5$ satisfies
150: the so-called {\em mesh condition} and therefore can be mapped into
151: ${\cal L}$ if and only if $\bD^\T(\bk-{\myb\gamma})=\bD^\T{\myb\lambda}$,
152: where ${\myb\gamma}^\T=(\gamma_0,\ldots,\gamma_4)$ and
153: ${\myb\lambda}^\T=(\lambda_0,\ldots,\lambda_4)$ with $0<\lambda_j<1$ so
154: that ${\myb\lambda}$ is a point inside the 5-d unit cube Cu(5). Thus, the
155: {\em window of acceptance} is the interior of the convex hull of the
156: points $\bD^\T{\bn_i}$, where $\bn_i$ are the $2^5$ vertices of the 5-d
157: unit cube Cu(5), see figure~1. We choose the 32 $\bn_i$'s as follows
158: \begin{eqnarray}
159: &&\bn_0=0,\quad\bn_{31}=\be_0+\be_1+\be_2+\be_3+\be_4,\nonumber\\
160: &&\bn_{j+1}=\be_{1-2j},\quad
161: \bn_{j+6}=\be_{1-2j}+\be_{4-2j},\quad
162: \bn_{j+11}=\be_{1-2j}+\be_{2-2j},\nonumber\\
163: &&\bn_{j+16}=\be_{1-2j}+\be_{2-2j}+\be_{5-2j},\quad
164: \bn_{j+21}=\be_{1-2j}+\be_{2-2j}+\be_{4-2j},\nonumber\\
165: &&\bn_{j+26}=\be_{1-2j}+\be_{2-2j}+\be_{4-2j}+\be_{5-2j},\quad
166: (j=0,\ldots,4),
167: \label{howto}
168: \end{eqnarray}
169: where $\be_0,\ldots,\be_4$ are the standard unit vectors in
170: ${\mbox{\mymsbm R}}^5$, with subscripts counted mod 5
171: ($\be_j\equiv\be_{j\pm5}$).
172: %--------------------------------------------------------------------------
173: \begin{figure}[tbh]
174: \begin{center}
175: \includegraphics[width=0.7\hsize]{window4.eps}
176: \end{center}
177:
178: \caption{The projection of the 5-d unit cube Cu(5) into the
179: orthogonal 2-d space $\cal D$. The window of acceptance is the interior of
180: outer decagon $\cal Q$ whose vertices are given by (\ref{decagonout}).
181: The innermost decagon is denoted by $\hat{\cal Q}$ and the middle decagon
182: by $\bar{\cal Q}$ with vertices given by (\ref{decagonin}). There are
183: ten triangles of type (a), which are all further subdivided in eight
184: regions of type (a1$),\ldots,($a8), as is indicated for one case in the
185: magnification on the left. This is determined by possible overlaps of
186: this triangle with the ten regions each of types (b) and (c), see text.}
187: \end{figure}
188: %--------------------------------------------------------------------------
189:
190: The projection of Cu(5) with these 32 points into $\cal W$ is a polytope
191: $\cal P$ with 40 edges connecting the 22 vertices, and with 20 faces. We
192: let $\bP_i=\bW^\T\bn_i$ for $i=0,\ldots,31$. The bottom is $\bP_0=(0,0,0)$
193: and top is $\bP_{31}=(0,0,5)$; they are called the tips of the polytope.
194: The remaining twenty vertices of $\cal P$ are
195: \begin{eqnarray}
196: &\bP_{j+1}=(\bd_j,1),\quad \bP_{j+6}=(\bd_j+\bd_{j+1},2),\nonumber\\
197: &\bP_{j+21}=(-\bd_{j-2}-\bd_{j-1},3),\quad \bP_{j+26}=(-\bd_{j-1},4),
198: \label{hpoly}
199: \end{eqnarray}
200: for $j=0,\ldots,4$. The other 10 points $\bP_{11},\ldots,\bP_{20}$
201: are in the interior of the polytope and are given by
202: \begin{equation}
203: \bP_{11+j}=(\bd_j+\bd_{j+2},2),\quad \bP_{16+j}=(-\bd_{j+1}-\bd_{j-1},3),
204: \label{decagon}\end{equation}
205: again for $j=0,\ldots,4$.
206:
207: The orthogonal projection of Cu(5) into $\cal D$ is a decagon $\cal Q$
208: with 10 edges connecting the 10 vertices, see figure~1. Let
209: $\bQ_i=\bD^\T\bn_i$ for $i=0,\ldots,31$. Then the vertices of the decagon
210: are
211: \begin{equation}
212: \bQ_{11+j}=-p\,\bd_{3-2j},\quad \bQ_{16+j}=p\,\bd_{5-2j},
213: \label{decagonout}
214: \end{equation}
215: with $j=0,\ldots,4$, and $p=(\sqrt5+1)/2$. The remaining 22 points
216: $\bQ_{0},\ldots,\bQ_{10}$ and $\bQ_{21},\ldots,\bQ_{31}$ are in the
217: interior; they are given by
218: \begin{eqnarray}
219: &\bQ_{0}=\bQ_{31}=0,\quad \bQ_{j+1}=\bd_{5-2j},\quad
220: \bQ_{26+j}=-\bd_{2-2j},\nonumber\\
221: &\bQ_{j+6}=p^{-1}\bd_{4-2j},\quad \bQ_{21+j}=-p^{-1}\bd_{3-2j}.
222: \label{decagonin}
223: \end{eqnarray}
224: Thus if the orthogonal projection $D^\T(\bk-{\myb\gamma})$ is in $\cal Q$,
225: then its projection $\bW^\T\bk$ is in ${\cal L}$.
226:
227: Consider a polytope $\cal P$ in ${\cal L}$ whose bottom is the
228: projection of a point $\hat\bk$ which is orthogonally projected onto the
229: center of the decagon, then the ten vertices of $\cal P$ at height
230: $z=1$ or $z=4$ correspond to the vertices of the middle decagon
231: $\bar{\cal Q}$, see figure~1; the ten vertices of the polytope at height
232: $z=2$ or $z=3$, correspond to the vertices of the innermost decagon
233: $\hat{\cal Q}$; while the ten interior points of the polytope correspond
234: to the vertices of the (outer) decagon ${\cal Q}$ shown in figure~1.
235:
236: The points in the quasiperiodic lattice $\cal L$ need careful analysis. We
237: shall show that $\cal L$ consists of polytopes $\cal P$ with 22 vertices
238: and 4 interior points which are shared with neighbouring polytopes. We
239: shall also show that it is periodic in the $z$-direction---the direction
240: of the line joining the two tips of $\cal P$---with period 5 corresponding
241: to the height of $\cal P$.
242:
243: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
244: \section{Overlapping unit cells}
245: The window $\cal Q$ is known \cite{Bruijn1,Bruijn3} to be everywhere dense
246: and uniformly distributed. Thus each point in $\cal Q$ corresponds to
247: a point in the quasiperiodic lattice $\cal L$. Using an idea of de Bruijn
248: \cite{Bruijn1}, we may find out the condition for both $\bk$ and
249: $\bk+\bn_i$ to satisfy the mesh condition---to lie both in the window of
250: acceptance, which give insights to how points in $\cal L$ are related to
251: one another.
252:
253: It is straightforward to show that every point inside the innermost
254: decagon $\hat{\cal Q}$ corresponds to a point in $\cal L$ that is
255: connected with 10 neighbors, and is in fact a tip of a polytope, as the
256: middle decagon $\bar{\cal Q}$ whose center is shifted to a point inside
257: $\hat{\cal Q}$ still lies inside $\cal Q$. This innermost decagon
258: $\hat{\cal Q}$ is further divided into 10 triangles \footnote{A rotation
259: of points in $\hat{\cal Q}$ by an angle of $2\pi\ell/5$, corresponds to a
260: rotation of the polytopes in $\cal L$ by an angle of $4\pi\ell/5$ about
261: the $z$-axis, while a rotation by $\pi$---inversion of points in
262: $\hat{\cal Q}$ through the origin of decagon $\hat{\cal Q}$---corresponds
263: to the 3-d inversion of polytopes in $\cal L$ through the center of each
264: polytope.}. Whenever the center of a decagon $\cal Q$ is shifted to a
265: point inside one of the triangles, four of the vertices of the shifted
266: outer decagon now lie inside $\cal Q$. More precisely, if
267: $\bD^\T(\bk-{\myb \gamma})$ is inside triangle (a) in figure~1, then
268: $\bW^\T\bk$ is a tip of a polytope $\cal P$ whose interior points
269: $\bW^\T(\bk+\bn_{20})$, $\bW^\T(\bk+\bn_{13})$, $\bW^\T(\bk+\bn_{18})$ and
270: $\bW^\T(\bk+\bn_{11})$---corresponding to the four points on $\cal Q$ on
271: the opposite side of triangle (a)---are now also in $\cal L$. This means
272: each polytope in ${\cal L}$ can have only four interior points which are
273: also in ${\cal L}$. Thus each such unit cell contains 26 atoms, 22
274: exterior, and 4 interior sites.
275:
276: It is also easy to find out how the polytopes share these interior points.
277: This is equivalent to finding the condition that both $\bk$ and
278: $\bk+\bn_{i}$ for $i=1,\ldots,10$ or $i=21,\ldots,30$ are
279: orthogonally projected into $\hat{\cal Q}$.
280:
281: Consider the ten vertices of the polytope at height $z=1$ or $z=4$
282: corresponding to the vertices of middle decagon $\bar{\cal Q}$. When the
283: center of the decagon is shifted to a point in one of ten rhombs of type
284: (b) shown in figure~1, then one of the ten vertices of the shifted
285: $\bar{\cal Q}$ is inside $\hat{\cal Q}$. For the example in figure~1,
286: $\bW^\T(\bk)$ and $\bW^\T(\bk+\bn_{28})$ are both tips of polytopes, with
287: $\bn_{28}$ the point on $\bar{\cal Q}$ on the opposite side of rhomb (b).
288: As a consequence, the polytope whose top is $\bW^\T(\bk+\bn_{28})$ shares
289: with the polytope $\cal P$ whose bottom is $\bW^\T(\bk)$ a polyhedron
290: $\cal K$ with 12 faces.
291:
292: Moreover, the vertices of the polytope at $z=2$ or $z=3$ correspond to
293: the vertices of $\hat{\cal Q}$. When the center of this decagon is shifted
294: to a point in one of ten octagons of type (c) shown also in figure~1, then
295: one of the ten vertices of the shifted $\hat{\cal Q}$ moves inside
296: $\hat{\cal Q}$. For the example in figure~1, $\bW^\T(\bk)$ and
297: $\bW^\T(\bk+\bn_{25})$ are then both tips of polytopes, while $\bn_{25}$
298: is the point on $\hat{\cal Q}$ on the side opposite to the octagon. Such
299: two polytopes share a polyhedron $\cal J$ with 6 faces. When vertices of
300: $\cal P$ at both $z=3$ and $z=4$ are inside $\hat{\cal Q}$, they are tops
301: of some polytopes\footnote{We shall later show that lattice ${\cal L}$
302: is periodic in the $z$-direction with period 5.} that share with
303: ${\cal P}$ either a polyhedron ${\cal K}$ or ${\cal J}$.
304:
305: By considering how these twenty regions intersect, we find that each of
306: the triangles in $\hat{\cal Q}$ is further divided into eight regions,
307: as shown for one case in figure~1, with this triangle magnified on the
308: left. When the orthogonal projection of $\bk-{\myb \gamma}$ is in (a1),
309: its projection into $\cal W$ is a polytope $\cal P$ intersecting with four
310: others whose tips are at $\bW^\T(\bk+\bn_{21})$, $\bW^\T(\bk+\bn_{8})$,
311: $\bW^\T(\bk+\bn_{24})$ and $\bW^\T(\bk+\bn_{6})$, and sharing with each
312: of them a polyhedron of type $\cal J$. When the projection is in (a2) or
313: (a3), either $\bW^\T(\bk+\bn_{26})$ or $\bW^\T(\bk+\bn_{1})$ becomes also
314: a tip of a polytope, so that $\cal P$ intersects with five polytopes,
315: sharing with one of them a polyhedron of type $\cal K$. In (a8),
316: $\cal P$ intersects with all of the above six polytopes. In regions
317: (a5) or (a7), $\bW^\T(\bk+\bn_{21})$ or $\bW^\T(\bk+\bn_{6})$ is no
318: longer a tip and $\cal P$ intersects with five polytopes sharing with two
319: of them a polyhedron of type $\cal K$. Finally, while in (a4) or (a6),
320: either $\bW^\T(\bk+\bn_{1})$ or $\bW^\T(\bk+\bn_{26})$ is no longer a tip,
321: and $\cal P$ intersects with four polytopes sharing with one of them
322: a polyhedron $\cal K$ and with the other three a $\cal J$.
323:
324: Therefore, in summary, there are only five different possibilities:
325: \begin{enumerate}
326: \item{The points inside a quadrilateral of type (a1) correspond to a
327: polytope intersecting with four other polytopes sharing with each a
328: polyhedron of type $\cal J$. An example of this case is shown in
329: figure~2a.}
330: \item{Points inside triangles of type (a2) or (a3) correspond to a
331: polytope intersecting with five other polytopes, sharing with one of them
332: a polyhedron $\cal K$ and with the other four polyhedra of type $\cal J$.
333: Such a case is shown in figure~2b.}
334: \item{If the point is inside triangles of type (a4) or (a6), the polytope
335: intersects with four other polytopes sharing with one of them a polyhedron
336: $\cal K$ and with the other three polyhedra of type $\cal J$.}
337: \item{If the point
338: is in a triangle (a5) or (a7), the polytope intersects with five other
339: polytopes sharing with two of them polyhedra of type $\cal K$ and with the
340: other three polyhedra $\cal J$.}
341: \item{Finally, if the point is inside a pentagon
342: (a8), the polytope intersects with six other polytopes sharing with two
343: of them a $\cal K$ and with the other four a $\cal J$, as is shown in
344: figure 3.}
345: \end{enumerate}
346: The relative frequencies are related to the ratios of their
347: areas and therefore the normalized probabilities are given by
348: \begin{eqnarray}
349: &P_{\rm a1}=2p^{-3},\quad P_{\rm a2}=P_{\rm a3}=p^{-6},\quad
350: P_{\rm a4}=P_{\rm a6}=p^{-5},\nonumber\\
351: &P_{\rm a5}=P_{\rm a7}=p^{-6},\quad P_{\rm a8}=p^{-5}+p^{-7}.
352: \label{prob}
353: \end{eqnarray}
354: %--------------------------------------------------------------------------
355: \begin{figure}[tbh]
356: \begin{center}
357: \includegraphics[width=0.4\hsize]{4j.eps}\hfil
358: \includegraphics[width=0.4\hsize]{4j1k.eps}
359: \end{center}
360: \vskip6pt
361: \hbox to\hsize{\hspace*{12pt}\footnotesize
362: \hfil (a) \hfil\hfil (b)\hfil\hspace*{-2pt}}
363:
364: \caption{(a) When the orthogonal projection of $\bk-{\myb \gamma}$ is in
365: region (a1), its projection in $\cal L$ is a polytope sharing an interior
366: point with each of four neighboring polytopes whose tips are projections
367: of $\bk+\bn_{21}$, $\bk+\bn_{8}$, $\bk+\bn_{24}$ and $\bk+\bn_{6}$; the
368: shared interior points are projections of $\bk+\bn_{11}$, $\bk+\bn_{16}$,
369: $\bk+\bn_{14}$ and $\bk+\bn_{19}$. (b) When it is in region (a2), the
370: polytope has five neighboring polytopes sharing the same interior points.
371: The additional polytope has its top at $\bW^\T(\bk+\bn_{26})$.}
372: \end{figure}
373: %--------------------------------------------------------------------------
374: %--------------------------------------------------------------------------
375: \begin{figure}[tbh]
376: \begin{center}
377: \includegraphics[width=0.4\hsize]{2k4jv.eps}\hfil
378: \includegraphics[width=0.4\hsize]{2k4jh.eps}
379: \end{center}
380: \vskip6pt
381: \hbox to\hsize{\hspace*{12pt}\footnotesize
382: \hfil (a) \hfil\hfil (b)\hfil\hspace*{-2pt}}
383:
384: \caption{The polytope $\cal P$ and its six neighbors in $\cal L$ for case
385: (a8). Their intersections with $\cal P$ are indicated with different
386: colors. In (a) the orientation is $(\theta,\phi)=(0,80^{\circ})$; in (b)
387: $(\theta,\phi)=(90^{\circ},0)$, so that each polytope appears as a
388: decagon.}
389: \end{figure}
390: %--------------------------------------------------------------------------
391:
392: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
393: \section{Parallelepipeds and windows}
394: Now the condition that $\bk$ satisfies the mesh condition can be further
395: simplified by considering parallelepiped $P(k_4,k_0,k_1)$ sandwiched
396: between the six grid planes $k_4-1$, $k_4$, $k_0-1$, $k_0$, $k_1-1$,
397: and $k_1$. Using (\ref{grid}) for the grid planes, we can solve for the
398: points of intersections, and find that for every point $\br$ in
399: $P(k_4,k_0,k_1)$, we may write
400: \begin{eqnarray}
401: &K_0(\br)=k_0,\quad K_1(\br)=k_1,\quad
402: K_2(\br)=\lfloor\alpha\rfloor+k_4+m,\nonumber\\
403: &\, K_3(\br)=\lfloor\beta\rfloor+k_1+n,\quad K_4(\br)=k_4,
404: \label{piped}
405: \end{eqnarray}
406: where $\lfloor x\rfloor$ is the greatest integer less or equal to $x$,
407: while $m$ and $n$ are integers satisfying $-1\le m,n\le 2$, and
408: \begin{eqnarray}
409: &&\alpha=p^{-1}(k_0-k_1-\gamma_0+\gamma_1)+\gamma_2-\gamma_4,\nonumber\\
410: &&\beta=p^{-1}(k_0-k_4-\gamma_0+\gamma_4)+\gamma_3-\gamma_1.
411: \label{alpha}
412: \end{eqnarray}
413: The range of $m$ and $n$ in equation (\ref{piped}) is limited by their
414: possible values at the eight corners of the parallelepiped, but not all
415: 16 choices are allowed by the mesh condition. Therefore, we now project
416: ${\myb K}(\br)$ given by (\ref{piped}) to $\cal D$ and find
417: \begin{equation}
418: \bD^\T({\myb K}(\br)-{\myb \gamma})=\sum_{j=0}^4
419: (K_j(\br)-\gamma_j)\bd_j=(m-a)\bd_2+(n-b)\bd_3,
420: \label{window}
421: \end{equation}
422: in which $a\equiv\{\alpha\}\equiv\alpha-\lfloor\alpha\rfloor$ and
423: $b\equiv\{\beta\}\equiv\beta-\lfloor\beta\rfloor$. The vector
424: (\ref{window}) must lie within the decagon $\cal Q$. Thus the allowed
425: values of $m$ and $n$ are determined by $a$ and $b$ only. As the
426: differences $k_0-k_4$ and $k_0-k_1$ run through all integer values, we
427: find from Kronecker's theorem \cite{HW} that $a$ and $b$ are everywhere
428: dense and uniformly distributed in the interval $(0,1)$. Proofs of such
429: ``ergodicity" in more general situations can be found, e.g., in the works
430: of Hof and Schlottmann \cite{Hof,Schlott}.
431:
432: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
433: \section{Periodicity in third direction}
434: Furthermore, because of the difference property shown in (\ref{alpha}),
435: $\alpha$ and $\beta$, which determine the configuration of the
436: parallelepiped, remain the same, if
437: $(k_0,k_1,k_4)\to(k_0+\ell,k_1+\ell,k_4+\ell)$. As a consequence we find
438: ${\myb K}(\br)\to{\myb K}(\br)+\ell\bn_{31}$, and its projection is
439: periodic in the $z$ direction with period equal to 5. It is also
440: interesting to note that if
441: ${\myb \gamma}\to{\myb \gamma}-{\frac15}c\,\bn_{31}$, $\alpha$ and $\beta$
442: in (\ref{alpha}) are also unchanged, so that the projection into the
443: 3d space does not show drastic changes when $c\ne0$, which is
444: behavior very different from the 2-d case \cite{APwindow}.
445:
446: It is not difficult to find values of $(a-m,b-n)$ in
447: (\ref{window}) corresponding to the ten vertices of the decagon $\cal Q$
448: given by (\ref{decagonout}). We find
449: \begin{eqnarray}
450: \begin{array}{ll}
451: \bQ_{16}\leftrightarrow (a-0,b-0)=(1,1),
452: &\bQ_{11}\leftrightarrow (a-0,b+1)=(0,p),\cr
453: \bQ_{18}\leftrightarrow (a-1,b+1)=(-1,p),
454: &\bQ_{13}\leftrightarrow (a-2,b+1)=(-p,1),\cr
455: \bQ_{20}\leftrightarrow (a-2,b-0)=(-p,0),
456: &\bQ_{15}\leftrightarrow (a-1,b-1)=(0,-1),\cr
457: \bQ_{17}\leftrightarrow (a-1,b-2)=(0,-p),
458: &\bQ_{12}\leftrightarrow (a-0,b-2)=(1,-p),\cr
459: \bQ_{19}\leftrightarrow (a+1,b-2)=(p,-1),
460: &\bQ_{14}\leftrightarrow (a+1,b-1)=(p,0),
461: \end{array}
462: \label{decagonval}
463: \end{eqnarray}
464: up to ambiguities when $a$ or $b$ is integer, as $m$ or $n$ changes by 1
465: when choosing $a$ or $b$ to be 0 or 1. The edges of decagon $\cal Q$ lead
466: to linear equations in $(a-m,b-n)$ and the mesh condition for
467: ${\myb K}(\br)$ becomes a set of inequalities in $a-m$ and $b-n$ as shown
468: in figure 4. Hence, it is very easy to create a routine to generate
469: $\cal L$.
470: %--------------------------------------------------------------------------
471: \begin{figure}[tbh]
472: \begin{center}
473: \includegraphics[width=0.5\hsize]{abwindow.eps}
474: \end{center}
475:
476: \caption{The image of $\cal Q$ under the inverse mapping of (\ref{window})
477: into the $(a-m,b-n)$-plane. The boundaries of regions with different
478: $(m,n)$ are given by linear equations in terms of $x\equiv a-m$ and
479: $y\equiv b-n$.}
480: \end{figure}
481: %--------------------------------------------------------------------------
482:
483: We note that $\cal L$ is periodic in the longitudinal or $z$-direction
484: with period 5, and aperiodic in the horizontal directions. This is a model
485: with five layers which repeat in the $z$-direction periodically. In each
486: of the layers, the points behave similarly, and the allowed sites for
487: $z=3$ are shown in figure 5$\,$(a). This shows decagonal symmetry, but the
488: layer may not be covered by Penrose tiles. In figure 5$\,$(b), we have
489: sites for atoms in all five layers plotted, with sites in different layers
490: represented by different symbols and colors.
491: %--------------------------------------------------------------------------
492: \begin{figure}[tbh]
493: \begin{center}
494: \includegraphics[width=0.4\hsize]{v3pt.eps}\hfil
495: \includegraphics[width=0.4\hsize]{v12345.eps}
496: \end{center}
497: \vskip6pt
498: \hbox to\hsize{\hspace*{12pt}\footnotesize
499: \hfil (a) \hfil\hfil (b)\hfil\hspace*{-2pt}}
500:
501: \caption{The allowed positions in each layer. (a) Allowed sites at
502: height $z=3$ are plotted. (b) All layers are plotted: points for $z=1$ are
503: represented by black boxes; for $z=2$ by red circles; for $z=3$ by
504: blue diamonds; for $z=4$ by green crosses; for $z=5$ by brown points.}
505: \end{figure}
506: %--------------------------------------------------------------------------
507:
508: Therefore, $\cal L$ is another example of a decagonal quasicrystal.
509: Decagonal quasicrystals exist in nature, and have been extensively studied
510: experimentally and theoretically \cite{YaIs,Steurer,SteKuo,
511: CoWidom,SteH,CHSt}. From Steurer's review \cite{Steurer}, one finds
512: translational periods along the tenfold $z$-axis varying from 4$\,$\AA\ to
513: 16$\,$\AA\ for different alloys, which may allow 2 layers \cite{CoWidom},
514: or even 5 or more layers, within a period.
515:
516: A situation somewhat similar to ours is found by Ben-Abraham, Lerer and
517: Snapir \cite{ALS}, who find that the projection of a 6-d lattice to 3d for
518: a certain choice of bases produces a quasicrystal lattice which is also
519: periodic in the $z$-direction. Their model has six-fold rather than
520: five-fold symmetry.
521:
522: The projection of ${\mbox{\mymsbm Z}}^5\to{\mbox{\mymsbm R}}^3$ is
523: well-known \cite{YaIs,SteH,Kramer2}. However, interpreting the 3-d
524: quasicrystal lattice $\cal L$ as overlapping polytopes $\cal P$ provides a
525: more systematic way to understand this lattice. We find that each unit
526: cell has 26 sites, sharing the four interior sites with its neighbors. The
527: lattice is periodic in the $z$-direction with period 5, and quasiperiodic
528: in the
529: $xy$-directions. Unlike the
530: projection of ${\mbox{\mymsbm Z}}^5\to{\mbox{\mymsbm R}}^2$, which
531: undergoes drastic change in behavior (no inflation and deflation rules) if
532: $c\ne0$, the projections into 3d are in the same class for $c=0$ and
533: $c\ne0$.
534:
535: Since real quasicrystals have icosahedra or triacontahedra as unit cells,
536: which are projections of a six-dimensional hypercube to a 3-d
537: space \cite{LRK}, the above method perhaps can also be used to determine
538: all possible overlappings and their frequencies.
539: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
540: \ack
541: We are most grateful to Dr.\ M.\ Widom, Dr.\ C.\ Richard, and Dr.\ M.\
542: Baake for providing us with many useful references. Comments on our
543: manuscript by Dr.\ M.\ Widom and Dr.\ M.\ Baake are also much
544: appreciated.
545:
546: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
547: \section*{References}
548: \begin{thebibliography}{99}
549: %
550: \bibitem{SBGC}{
551: Shechtman D, Blech I, Gratias D R and Cahn J W 1984
552: Metallic phase with long-range orientational
553: order and no translational symmetry
554: \PRL {\bf 53} 1951--3.}
555: %
556: \bibitem{Penrose}{
557: Penrose R 1989
558: Tilings and quasi-crystals; a non-local growth problem?
559: {\it Introduction to The Mathematics of Quasicrystals}
560: ({\it Aperiodicity and Order} vol~2)
561: ed M~V Jari\'c
562: (Boston: Academic Press) pp~53--79.}
563: %
564: \bibitem{GrSh}{
565: Gr\"unbaum B and Shephard G C 1987
566: {\it Tilings and Patterns}
567: (New York: W~H Freeman and Co) ch~10.}
568: %
569: \bibitem{Bruijn1}{
570: de Bruijn N G 1981
571: Algebraic theory of Penrose's non-periodic
572: tilings of the plane.~I
573: {\it Indagationes Mathematicae} {\bf 84} 38--52.}
574: \nonum{
575: de Bruijn N G 1981
576: Algebraic theory of Penrose's non-periodic
577: tilings of the plane.~II
578: {\it Indagationes Mathematicae} {\bf 84} 53--66.}
579: %
580: \bibitem{KrPa}{
581: Kramer P and Papadopolos Z (eds) 2002
582: {\it Coverings of Discrete Quasiperiodic Sets}
583: (Berlin: Springer).}
584: %
585: \bibitem{Gummelt}{
586: Gummelt P 1996
587: Penrose tilings as coverings of congruent decagons
588: {\it Geometriae Dedicata} {\bf 62} 1--17.}
589: %
590: \bibitem{SJ}{
591: Steinhardt P J and Jeong H C 1996
592: A simpler approach to Penrose tilings with implications for
593: quasicrystal formation
594: {\it Nature} {\bf 382} 433--5.}
595: %
596: \bibitem{SJSTAT}{
597: Steinhardt P J, Jeong H C, Saitoheong K, Tanaka M, Abe E
598: and Tsai A P 1996
599: Experimental verification of the quasi-unit-cell model of
600: quasicrystal structure
601: {\it Nature} {\bf 396} 55--7.}
602: %
603: \bibitem{LR}{
604: Lord E A and Ranganathan S 2001
605: The Gummelt decagon as a `quasi unit cell'
606: \AC A {\bf 57} 531--9.}
607: %
608: \bibitem{LRK}{
609: Lord E A, Ranganathan S, and Kulkarni U D 2001
610: Quasicrystals: tiling versus clustering
611: {\it Philosophical Magazine} A {\bf 81} 2645--51.}
612: %
613: \bibitem{Kramer1}{
614: Kramer P 2002
615: Covering of discrete quasiperiodic sets: concepts and theory
616: {\it Coverings of Discrete Quasiperiodic Sets}
617: ed P Kramer and Z Papadopolos
618: (Berlin: Springer) pp~1--21.}
619: %
620: \bibitem{DuGr}{
621: Duneau M and Gratias D 2002
622: Covering clusters in icosahedral quasicrystals
623: {\it Coverings of Discrete Quasiperiodic Sets}
624: ed P Kramer and Z Papadopolos
625: (Berlin: Springer) pp~23--61.}
626: %
627: \bibitem{GGA}{
628: G\"ahler F, Gummelt P and Ben-Abraham S I 2002
629: Generation of quasiperiodic order by maximal cluster covering
630: {\it Coverings of Discrete Quasiperiodic Sets}
631: ed P Kramer and Z Papadopolos
632: (Berlin: Springer) pp~63--95.}
633: %
634: \bibitem{ALS}{
635: Ben-Abraham S I, Lerer Y and Snapir Y 2004
636: Constructing a dodecagonal structure by projection in two stages
637: {\it J. Non-Cryst. Solids} {\bf 334\&335} 71--6.}
638: %
639: \bibitem{Bruijn2}{
640: de Bruijn N G 1986
641: Quasicrystal and their Fourier transform
642: {\it Indagationes Mathematicae} {\bf 89} 123--52.}
643: %
644: \bibitem{GRh}{
645: G\"ahler F and Rhyner J 1986
646: Equivalence of the generalised grid and projection methods
647: for the construction of quasiperiodic tilings
648: \JPA {\bf 19} 267--77.}
649: %
650: \bibitem{Elser}{
651: Elser V 1986
652: The diffraction pattern of projected structures
653: \AC A {\bf 42} 36--43.}
654: %
655: \bibitem{DK}{
656: Duneau M and Katz A 1985
657: Quasiperiodic patterns
658: \PRL {\bf 54} 2688--91.}
659: %%
660: \bibitem{Mackay2}{
661: Mackay A L 1982
662: Crystallography on the Penrose pattern
663: {\it Physica} A {\bf 114} 609--13.}
664: %
665: \bibitem{BJKS}{
666: Baake M, Joseph D, Kramer P and Schlottmann M 1990
667: Root lattices and quasicrystals
668: \JPA {\bf 23} L1037--41.}
669: %
670: \bibitem{APwindow}{
671: Au-Yang H and Perk J H H 2005
672: Generalized Penrose Tilings
673: {\it Preprint}.}
674: %
675: \bibitem{APquasi}{
676: Au-Yang H and Perk J H H 2005
677: Quasicrystals: projection of 5d lattice into 2 or 3 dimensions
678: {\it Preprint}.}
679: %
680: \bibitem{LSt}{
681: Levine D and Steinhardt P J 1986
682: Quasicrystals. I. Definition and structure
683: \PR B {\bf 34} 596--616.}
684: %
685: \bibitem{SoSt}{
686: Socolar J E S and Steinhardt P J 1986
687: Quasicrystals. II. Unit-cell configurations
688: \PR B {\bf 34} 617--47.}
689: %
690: \bibitem{IsYa}{
691: Ishihara K N and Yamamoto A 1988
692: Penrose patterns and related structures. I.
693: Superstructure and generalized Penrose patterns.
694: \AC A {\bf 44} 508-16.}
695: %
696: \bibitem{Bruijn3}{
697: de Bruijn N G 1992
698: Penrose patterns are almost entirely determined by two points
699: {\it Discrete Mathematics} {\bf 106/107} 97--104.}
700: %
701: \bibitem{HW}{
702: Hardy G H and Wright E M 1960
703: {\it An Introduction to the Theory of Numbers}
704: 4th edition (London: Oxford University Press)
705: ch~XXIII {\it Kronecker's Theorem}.}
706: %
707: \bibitem{Hof}{
708: Hof A 1998
709: Uniform distribution and the projection method
710: {\it Quasicrystals and Discrete Geometry}
711: ed J~Patera
712: (Providence RI: Am. Math. Soc.) pp~201--6.}
713: %
714: \bibitem{Schlott}{
715: Schlottmann M 1998
716: Cut-and-project sets in locally compact abelian groups
717: {\it Quasicrystals and Discrete Geometry}
718: ed J~Patera
719: (Providence RI: Am. Math. Soc.) pp~247--64.}
720: %
721: \bibitem{YaIs}{
722: Yamamoto A and Ishihara K N 1988
723: Penrose patterns and related structures. II.
724: Decagonal quasicrystals.
725: \AC A {\bf 44} 707-14.}
726: %
727: \bibitem{Steurer}{
728: Steurer W 1990
729: The structure of quasicrystals
730: {\it Zeitschrift f\"ur Kristallographie}
731: {\bf 190} 179--234.}
732: %
733: \bibitem{SteKuo}{
734: Steurer W and Kuo K H 1990
735: Five-dimensional structure analysis of decagonal
736: Al$_{65}$Cu$_{20}$Co$_{15}$
737: \AC B {\bf 46} 703--12.}
738: %
739: \bibitem{CoWidom}{
740: Cockayne E and Widom M 1998
741: Ternary model of an Al-Cu-Co decagonal quasicrystal
742: \PRL {\bf 81} 598--601.}
743: %
744: \bibitem{SteH}{
745: Steurer W and Haibach T 1999
746: The periodic average structure of particular quasicrystals
747: \AC A {\bf 55} 48--57.}
748: %
749: \bibitem{CHSt}{
750: Cervellino A, Haibach T and Steurer W 2002
751: Structure solution of the basic decagonal Al--Co--Ni
752: phase by the atomic surfaces modelling method
753: \AC B {\bf 58} 8--33.}
754: %
755: \bibitem{Kramer2}{
756: Kramer P 2002
757: Voronoi and Delone clusters in dual quasiperiodic tilings
758: {\it Coverings of Discrete Quasiperiodic Sets}
759: ed P Kramer and Z Papadopolos
760: (Berlin: Springer) pp~185--225.}
761: \end{thebibliography}
762: \end{document}
763: