1: \documentclass[11pt]{article}
2: \usepackage{epsfig,amssymb}
3: %\usepackage[active]{srcltx}
4: \topmargin=-0.5cm \hoffset=-20mm \textwidth=16cm \textheight=22cm
5: \textheight=22cm
6: \begin{document}
7: %
8: \title{Asymptotic behavior of discrete holomorphic maps $z^c$, $log(z)$ and discrete Painleve transcedents.}
9: %
10: \author{{\Large Agafonov S.I. } \\
11: \\
12: Institut f\"{u}r Algebra und Geometrie\\
13: Martin-Luther-Universit\"{a}t
14: Halle-Wittenberg \\ D-06099 Halle (Saale), Germany\\
15: e-mail: {\tt agafonov@mathematik.uni-halle.de} }
16: \date{}
17:
18: \newtheorem{theorem}{Theorem}
19: \newtheorem{proposition}{Proposition}
20: \newtheorem{lemma}{Lemma}
21: \newtheorem{corollary}{Corollary}
22: \newtheorem{definition}{Definition}
23: \newtheorem{conjecture}{Conjecture}
24:
25: \pagestyle{plain}
26:
27: \maketitle
28:
29: \begin{abstract}
30: It is shown that discrete analogs of $z^{c }$ and ${\rm log}(z)$
31: have the same asymptotic behavior as their smooth counterparts.
32: These discrete maps are described in terms of special solutions of
33: discrete Painleve-II equations, asymptotics of these solutions
34: providing the behaviour of discrete $z^{c }$ and ${\rm log}(z)$
35: at infinity.
36: \end{abstract}
37:
38: \section{Introduction}
39:
40: In this paper we discuss a very interesting and rich object:
41: circle patterns mimicking the holomorphic maps $z^{c }$ and ${\rm
42: log}(z)$. Discrete $z^{2 }$ and ${\rm log}(z)$ were guessed by
43: Schramm and Kenyon (see \cite{www}) as examples from the class of
44: circle patterns with the combinatorics of the square grid
45: introduced by Schramm in \cite{Schramm}. This class opens a new
46: page in the classical theory of circle packings, enjoying new
47: golden age after Thurston's idea \cite{T} about approximating the
48: Riemann mapping by circle packings: it turned out that these
49: circle patterns are governed by discrete integrable equation (the
50: stationary Hirota equation), thus providing one with the whole
51: machinery of the integrable system theory (\cite{BPD}).
52:
53: The striking analogy between circle patterns and holomorphic maps
54: resulted in the development of discrete analytic function theory
55: (for a good survey see\cite{DS}). Classical circle packings
56: comprised of disjoint open disks were later generalized to circle
57: patterns where the disks may overlap. Discrete versions of
58: uniformization theorem, maximum principle, Schwarz's lemma and
59: rigidity properties and Dirichlet principle were established
60: (\cite{MR},\cite{H},\cite{Schramm}).
61:
62: Different underlying combinatorics were considered: Schramm
63: introduced square grid circle patterns, generalized by Bobenko
64: and Hoffmann to hexagonal patterns with constant intersection
65: angles in \cite{BH}, hexagonal circle patterns with constant
66: multi-ratios were studied by Bobenko, Hoffman and Suris in
67: \cite{BHS}.
68:
69: The difficult question of convergence was settled by Rodin and
70: Sullivan \cite{RS} for general circle packings, He and Schramm
71: \cite{HS} showed that the convergence is $C^{\infty}$ for
72: hexagonal packings, the uniform convergence for square grid
73: circle patterns was established by Schramm \cite{Schramm}.
74:
75: On the other hand not very many examples are known: for circle
76: packings with the hexagonal combinatorics the only explicitly
77: described examples are Doyle spirals, which are discrete
78: analogues of exponential maps \cite{Doy}, and conformally
79: symmetric packings, which are analogues of a quotient of Airy
80: functions \cite{BHConf}. For patterns with overlapping circles
81: more examples are constructed: discrete versions of ${\rm exp}
82: (z)$, ${\rm erf}(z)$ (\cite{Schramm}), $z^c$, $ {\rm log} (z)$
83: (\cite{AB}) are constructed for patterns with underlying
84: combinatorics of the square grid; $z^c$, ${\rm log}(z)$ are also
85: described for hexagonal patterns with both multi-ratio
86: (\cite{BHS}) and constant angle (\cite{BH}) properties.
87:
88: Discrete $z^c$ is not only a very interesting example in discrete
89: conformal geometry. It has mysterious relationships to other
90: fields. It is constructed via some discrete isomonodromic problem
91: and is governed by discrete Painlev\'e II equation
92: (\cite{AB},\cite{N}), thus giving geometrical
93: interpretation thereof. Its linearization defines Green's function on critical
94: graphs (see \cite{BMS}) found in \cite{K} in the frames of the
95: theory of Dirac operator. Moreover, it seems to be a rather
96: important tool for investigation of more general circle patterns
97: and discrete minimal surfaces (see \cite{B} for a brief survey and
98: \cite{BHSp} for more details).
99:
100: \subsection{Circle patterns and discrete conformal maps}
101:
102: To visualize the analogy between Schramm's circle patterns and
103: conformal maps, consider regular patterns composed of unit
104: circles and suppose that the radii are being deformed so as to
105: preserve the orthogonality of neighboring circles and the
106: tangency of half-neighboring ones. Discrete maps taking
107: intersection points of
108: the unit circles of the standard regular patterns to the
109: respective points of the deformed patterns mimic classical
110: holomorphic functions, the deformed radii being analogous to
111: $|f'(z)|$ (see Fig.~\ref{CPMap}).
112:
113: \begin{figure}[th]
114: \begin{center}
115: \epsfig{file=PreImage.eps,width=45mm}
116: \begin{picture}(20,50)
117: \put(3,52){\vector(1,0){20}} \put(7,58){ \it \huge f}
118: \end{picture}
119: \epsfig{file=Image.eps,width=50mm} \caption{Schramm's circle
120: patterns as discrete conformal map. Shown is the discrete
121: version of the holomorphic mapping $z^{3/2}$.} \label{CPMap}
122: \end{center}
123: \end{figure}
124: It is easy to show that the lattice comprised of the centers of
125: circles of Schramm's pattern and their intersection points is a
126: special discrete conformal mapping (see Definition \ref{DCM}
127: below). The latter were introduced in \cite{BPdis} in the frames
128: of discrete integrable geometry, originally without any relation
129: to circle patterns.
130:
131:
132: \begin{definition} A map $f\ : \ {\mathbb Z^2\ \rightarrow \ {\mathbb R^2}={\mathbb C}}$
133: is called a discrete conformal map if all its elementary
134: quadrilaterals are conformal squares, i.e., their cross-ratios
135: are equal to -1:
136:
137: $$
138: q(f_{n,m},f_{n+1,m},f_{n+1,m+1},f_{n,m+1}):=
139: $$
140: \begin{equation}
141: \frac{(f_{n,m}-f_{n+1,m})(f_{n+1,m+1}-f_{n,m+1})}
142: {(f_{n+1,m}-f_{n+1,m+1})(f_{n,m+1}-f_{n,m})}=-1. \label{q}
143: \end{equation}
144: \label{DCM}
145: \end{definition}
146: This
147: definition is motivated by the following properties:\\
148: 1) it is
149: M\"obius invariant,\\
150: 2) a smooth map $f:\ D\ \subset {\mathbb C} \to {\mathbb C}$ is
151: conformal (holomorphic or antiholomorphic) if and only if
152: $$
153: \lim _{\epsilon \to 0}q(f(x,y),f(x+\epsilon ,y)f(x+\epsilon
154: ,y+\epsilon )f(x,y+\epsilon ))=-1
155: $$
156: for all $(x,y)\in D$.
157:
158:
159: \begin{definition}
160: A discrete conformal map $f_{n,m}$ is called embedded if
161: interiors of different elementary quadrilaterals
162: $(f_{n,m},f_{n+1,m},f_{n+1,m+1},f_{n,m+1})$ do not intersect.
163: \end{definition}
164: \noindent The condition for discrete conformal map to be embedded
165: can be relaxed as follows.
166: \begin{definition}
167: A discrete conformal map $f_{n,m}$ is called an immersion if
168: interiors of adjacent elementary quadrilaterals
169: $(f_{n,m},f_{n+1,m},f_{n+1,m+1},f_{n,m+1})$ are disjoint.
170: \end{definition}
171: To illustrate the difference between the immersed and embedded
172: discrete conformal maps, let us imagine that the elementary
173: quadrilaterals of the map are made of elastic inextensible
174: material and glued along the corresponding edges to produce
175: a surface with a border. If this surface is immersed it is locally flat. Being dropped down it will not have folds.
176: At first sight it seems to be sufficient to guarantee
177: embeddedness, provided $Z_{n,0}\to \infty$ and $Z_{0,m}\to \infty$
178: as $n \to \infty$ (which follows from $ Z^{c }_{n,0}=k(c)n^{c
179: }\left(1+O\left(\frac{1}{n^2} \right) \right), \ \ n \to \infty $
180: (\cite{AB}). But a surface with such properties still may have
181: some limit curve with self-intersections thus giving overlapping
182: quadrilaterals. Hypothetical example of such a surface is shown
183: in Fig.~\ref{surface}.
184: \begin{figure}[ht]
185: \begin{center}
186: \epsfig{file=surface.eps, width=50mm}
187: \end{center}
188: \caption{Surface glued of quadrilaterals of immersed but
189: non-embedded discrete map. } \label{surface}
190: \end{figure}
191:
192:
193: Equation (\ref{q}) can be supplemented with the following
194: nonautonomous constraint:
195: \begin{equation}
196: c f_{n,m}=2n\frac{(f_{n+1,m}-f_{n,m})(f_{n,m}-f_{n-1,m})}
197: {(f_{n+1,m}-f_{n-1,m})}+2m\frac{(f_{n,m+1}-f_{n,m})(f_{n,m}-f_{n,m-1})}
198: {(f_{n,m+1}-f_{n,m-1})}. \label{c}
199: \end{equation}
200: This constraint, as well as its compatibility with (\ref{q}), is
201: derived from some monodromy problem (see \cite{AB} for detail).
202: Let us assume $0<c < 2$ and denote ${\mathbb Z^2_{+}}=\{ (n,m) \in
203: {\mathbb Z^2}: n,m \ge 0 \}.$ Motivated by the asymptotics of the
204: constraint (\ref{c}) at $n,m \rightarrow \infty$ and the
205: properties
206:
207: $$
208: z^c({\mathbb R_+}) \in {\mathbb R_+}, \ \ z^c(i{\mathbb R_+}) \in
209: e^{c \pi i/2}{\mathbb R_+}
210: $$
211: of the holomorphic mapping $z^c$ we use the following definition
212: (\cite{BPD}) of the "discrete" $z^c$.
213:
214: \begin{definition}
215: The discrete conformal map $Z^c\ : \ {\mathbb Z^2_+\ \rightarrow \
216: {\mathbb C}}, \ 0<c < 2\ $ is the solution of (\ref{q}),(\ref{c})
217: with the initial conditions
218:
219: \begin{equation}
220: Z^c(0,0)=0, \ \ Z^c(1,0)=1, \ \ Z^c(0,1)=e^{c \pi i/2}.
221: \label{initial}
222: \end{equation}
223: \label{def}
224: \end{definition}
225: Obviously, $Z^c(n,0)\in {\mathbb R_{+}}$ and
226: $Z^c(0,m)\in
227: e^{c \pi i/2} ({\mathbb R_{+}})$ for any $ n,m
228: \in {\mathbb N}$ .\\
229: A limit of rescaled $Z^c$ as $c$ approaches 2 gives discrete
230: $Z^2$. Discrete ${\rm Log}$
231: is defined as dual to $Z^2$ (see subsection \ref{s.Z2} for detail).
232: \begin{theorem}(\cite{A}) \label{emb}
233: The discrete maps ${\rm Log}$ and $Z^{c }$ for $0<c \le 2$ are
234: embedded.
235: \end{theorem}
236:
237:
238: Given initial data $f_{0,0}=0$, $f_{1,0}=1$, $f_{0,1}=e^{i \alpha
239: }$ with $\alpha \in \mathbb R$, constraint (\ref{c}) allows one
240: to compute $f_{n,0}$ and $f_{0,m}$ for all $n,m \ge 1.$ Now using
241: equation (\ref{q}) one can successively compute $f_{n,m}$ for any
242: $n,m \in {\mathbb N}$. It turned out that all edges at the vertex
243: $f_{n,m}$ with $n+m=0 \ ({\rm mod} \ 2)$ are of the same length
244: \begin{equation}\label{equi}
245: |f_{n+1,m}-f_{n,m}|=|f_{n,m+1}-f_{n,m}|=|f_{n-1,m}-f_{n,m}|=|f_{n,m-1}-f_{n,m}|
246: \end{equation}
247: and all angles between the neighboring edges at the vertex
248: $f_{n,m}$ with $n+m=1 \ ({\rm mod} \ 2)$ are equal to $\pi /2.$
249: Thus for any $n,m: \ n+m=0 \ ({\rm mod }\ 2)$ the points
250: $f_{n+1,m},$ $f_{n,m+1},$ $f_{n-1,m},$ $f_{n,m-1}$ lie on the
251: circle with the center $f_{n,m}$. All such circles form a circle
252: pattern of Schramm type (see \cite{Schramm}), i.e. the circles of
253: neighboring quadrilaterals intersect orthogonally and the circles
254: of half-neighboring quadrilaterals with common vertex are
255: tangent. Consider the sublattice $\{n,m: \ n+m=0 \ ({\rm mod} \
256: 2)\}$ and denote by $\mathbb V$ its quadrant
257: \begin{equation}\label{V}
258: {\mathbb V}=\{z=N+iM:\ N,M \in {\mathbb Z^2}, M \ge |N| \},
259: \end{equation}
260: where
261: $$
262: N=(n-m)/2, \ \ M=(n+m)/2.
263: $$
264: We will use complex labels $z=N+iM$ for this sublattice. Denote by
265: $C(z)$ the circle of the radius
266: \begin{equation}
267: R(z)=|f_{n,m}-f_{n+1,m}|=|f_{n,m}-f_{n,m+1}|=|f_{n,m}-f_{n-1,m}|=|f_{n,m}-f_{n,m-1}|
268: \label{rmap}
269: \end{equation}
270: with the center at $f_{N+M,M-N}=f_{n,m}.$ Then $Z^c$ is immersed
271: if and only if $R(z)$ satisfy the following equation
272: \begin{equation}
273: R(z)^2=\frac{\left(
274: \frac{1}{R(z+1)}+\frac{1}{R(z+i)}+\frac{1}{R(z-1)}+\frac{1}{R(z-i)}
275: \right) R(z+1)R(z+i)R(z-1)R(z-i) }{ R(z+1)+R(z+i)+R(z-1)+R(z-i) }
276: \label{lnR}
277: \end{equation}
278: The proof of Theorem \ref{emb}
279: is based on the analysis of the solution to the
280: following equations:
281: \begin{equation}
282: \begin{array}{l}
283: R(z)R(z+1)(-2M-c
284: )+R(z+1)R(z+1+i)(2(N+1)-c )+\\
285: \qquad R(z+1+i)R(z+i)(2(M+1)-c )+R(z+i)R(z)(-2N-c )=0,
286: \end{array}
287: \label{square}
288: \end{equation}
289: \begin{equation}
290: \begin{array}{l}
291: (N+M)(R(z)^2-R(z+1)R(z-i))(R(z+i)+R(z+1))+\\
292: \qquad (M-N)(R(z)^2-R(z+i)R(z+1))(R(z+1)+R(z-i))=0,
293: \end{array}
294: \label{Ri}
295: \end{equation}
296: $$
297: (N+M)(R(z)^2-R(z+i)R(z-1))(R(z-1)+R(z-i))+
298: $$
299: \begin{equation}
300: (M-N)(R(z)^2-R(z-1)R(z-i))(R(z+i)+R(z-1))=0, \label{Le}
301: \end{equation}
302: $$
303: (N+M)(R(z)^2-R(z+i)R(z-1))(R(z+1)+R(z+i))+
304: $$
305: \begin{equation}
306: (N-M)(R(z)^2-R(z+1)R(z+i))(R(z+i)+R(z-1))=0, \label{Up}
307: \end{equation}
308: which are compatible with (\ref{lnR}). Namely, $Z^c$ is immersed
309: iff the solutions of these equations in ${\mathbb V}$ with
310: \begin{equation}
311: R(0)=1, \ \ \ R(i)=\tan \frac{c \pi}{4}. \label{Rinitial}
312: \end{equation} is
313: positive (see \cite{AB}). More delicate global property of
314: embeddedness follows from the following proposition.
315: \begin{proposition}(\cite{A})
316: If for a solution R(z) of (\ref{square}, \ref{Ri}) with $c \ne
317: 1$ and initial conditions (\ref{Rinitial}) holds
318: \begin{equation}
319: R(z)>0, \ \ (c -1)(R(z)^2 - R(z-i)R(z+1))\ge 0
320: \label{sign}
321: \end{equation}
322: in interior vertices of ${\mathbb V}$, then $Z^c$ is embedded. \label{convex}
323: \end{proposition}
324:
325:
326:
327: \subsection{The discrete maps $Z^2$ and $\rm Log $. Duality}
328: \label{s.Z2}
329:
330: Definition \ref{def} was given for $0<c <2.$ For $c <0$ or $c
331: >2$, the radius $R(1+i)=c /(2-c)$ of the corresponding
332: circle patterns (found as a solution to equations for $R(z)$)
333: becomes negative and some elementary
334: quadrilaterals around $f_{0,0}$ intersect. But for $c =2$, one
335: can renormalize the initial values of $f$ so that the
336: corresponding map remains an immersion. Let us consider $Z^{c }$,
337: with $0<c <2$,
338: and make the following renormalization
339: for the corresponding radii: $R \to \frac{2-c }{c }R.$ Then as
340: $c \to 2-0$ we have
341: $$
342: R(0)=\frac{2-c }{c }\to +0, \ \ R(1+i)=1, \ \ R(i)=\frac{2-c }{c
343: }\tan \frac{c \pi}{4}\to \frac{2}{\pi}.
344: $$
345: \begin{definition}(\cite{AB})
346: $Z^{2}\ : \ {\mathbb Z^2_+\ \rightarrow \ {\mathbb R^2}={\mathbb
347: C}}$ is the solution of (\ref{q}), (\ref{c}) with $c =2$ and the
348: initial conditions
349: $$
350: Z^{2}(0,0)=Z^{2}(1,0)=Z^{2}(0,1)=0, \ \ Z^{2}(2,0)=1, \ \
351: Z^{2}(0,2)=-1, \ \ Z^{2}(1,1)=i\frac{2}{\pi}.
352: $$
353: \label{def2}
354: \end{definition}
355:
356: \begin{figure}[ht]
357: \begin{center}
358: \epsfig{file=Z2.eps,width=60mm} \caption{Discrete $Z^2$.}
359: \label{f.Z2}
360: \end{center}
361: \end{figure}
362:
363: In this definition, equations (\ref{q}),(\ref{c}) are understood
364: to be regularized through multiplication by their denominators.
365: Note that for the radii on the border one has $R(N+iN)=N.$ \\
366: If $R(z)$ is a solution to (\ref{lnR}) and therefore defines
367: some immersed circle patterns, then $\tilde R(z)=\frac{1}{R(z)}$
368: also solves (\ref{lnR}). This reflects the fact that for any
369: discrete conformal map $f$ there is {\it dual discrete conformal
370: map} $f^*$ defined by (see \cite{BPD})
371: \begin{equation} \label{duality}
372: f^*_{n+1,m}-f^*_{n,m}=-\frac{1}{{f_{n+1,m}}-{f_{n,m}}}, \ \
373: f^*_{n,m+1}-f^*_{n,m}=\frac{1}{{f_{n,m+1}}-{f_{n,m}}}.
374: \end{equation}
375: The smooth limit of the duality (\ref{duality}) is
376: $$
377: ({f^*})'=-\frac{1}{f'}.
378: $$
379: The dual of $f(z)=z^2$ is, up to a constant, ${f^*(z)}=\log z.$
380: Motivated by this observation, we define the discrete logarithm
381: as the discrete map dual to $Z^2$, i.e. the map corresponding to
382: the circle pattern with radii
383: $$
384: R_{\rm{Log}}(z)=\frac{1}{R_{Z^2}(z)},
385: $$
386: where $R_{Z^2}$ are the radii of the circles for $Z^2.$ Here one
387: has $R_{\rm{Log}}(0)=\infty$, i.e. the corresponding circle is a
388: straight line. The corresponding constraint (\ref{c})
389: can be also derived as a limit. Indeed, consider the map
390: $g=\frac{2-c }cZ^{c }-\frac{2-c }c.$ This map satisfies (\ref{q})
391: and the constraint
392: $$
393: c \left( g_{n,m}+ \frac{2-c }c\right)=
394: 2n\frac{(g_{n+1,m}-g_{n,m})(g_{n,m}-g_{n-1,m})}
395: {(g_{n+1,m}-g_{n-1,m})}+
396: 2m\frac{(g_{n,m+1}-g_{n,m})(g_{n,m}-g_{n,m-1})}
397: {(g_{n,m+1}-g_{n,m-1})}.
398: $$
399: Keeping in mind the limit procedure used do determine $Z^2$, it
400: is natural to define the discrete analogue of $\log z$ as the
401: limit of $g$ as $c \to +0$. The corresponding constraint becomes
402: \begin{equation}
403: 1=n\frac{(g_{n+1,m}-g_{n,m})(g_{n,m}-g_{n-1,m})}
404: {(g_{n+1,m}-g_{n-1,m})}+m\frac{(g_{n,m+1}-g_{n,m})(g_{n,m}-g_{n,m-1})}
405: {(g_{n,m+1}-g_{n,m-1})}. \label{cln}
406: \end{equation}
407:
408: \begin{figure}[ht]
409: \begin{center}
410: \epsfig{file=LogBig.eps,width=80mm} \caption{Discrete {\rm Log}.}
411: \label{f.Log}
412: \end{center}
413: \end{figure}
414: \begin{definition} (\cite{AB}) $\rm{Log}$ is the map
415: $\rm{Log} : \ {\mathbb Z^2_+ \rightarrow \ {\mathbb R^2}=\bar
416: {\mathbb C}}$ satisfying (\ref{q}) and (\ref{cln}) with the
417: initial conditions
418: $$
419: \rm{Log}(0,0)=\infty , \ \rm{Log}(1,0)=0, \ \rm{Log}(0,1)=i\pi ,
420: $$
421: $$
422: \ \rm{Log}(2,0)=1, \ \rm{Log}(0,2)=1+i\pi ,
423: \ \rm{Log}(1,1)=i\frac{\pi}{2}.
424: $$
425: \label{defLn}
426: \end{definition}
427: The circle patterns corresponding to the discrete conformal
428: mappings $Z^2$ and $\rm{Log}$ were conjectured
429: by O.~Schramm and R.~Kenyon (see \cite{www}), in \cite{AB} it was proved that
430: they are immersed.
431:
432:
433:
434: \subsection{ $Z^c$ and discrete Painlev\'e equations} The main tool
435: to establish all the above mentioned properties of $R(z)$ was a
436: special case of discrete Painlev\'e-II equation:
437: \begin{equation}
438: \label{dPII}
439: (n+1)(u_n^2-1)\left(\frac{u_{n+1}-iu_n}{i+u_nu_{n+1}}\right)-
440: n(u_n^2+1)\left(\frac{u_{n-1}+iu_n}{i+u_{n-1}u_n}\right) =
441: c u_n.
442: \end{equation}
443: The embedded $Z^c$ corresponds to the unitary solution
444: $u_n=e^{i\alpha_n}$ of this equation with $u_0=e^{ic \pi/4}$
445: $0<\alpha _n < \pi/2.$ ($\alpha_n$ is define by
446: $f_{n,n+1}-f_{n,n}=e^{2i\alpha _n}(f_{n+1,n}-f_{n,n}).$)
447:
448:
449: In \cite{A} was proved:
450: $$
451: \lim _{n\to \infty}Z^c _{n,m}=\infty, \ \ \lim _{m\to \infty}Z^c
452: _{n,m}=\infty.
453: $$
454: In the present paper we use (\ref{Ri}) and (\ref{square}) (which
455: form of course a system of Painlev\'e type) to establish the more
456: accurate result
457: $$R(N_0+iM) \simeq K(c)M^{c-1} \ \ \ {\rm as} \ \ \ M\to
458: \infty
459: $$
460: and corresponding asymptotics for $Z^c$ and $u_n$. Thus asymptotic
461: behavior of discrete $z^c$ and $log(z)$ is exactly that of their
462: smooth counterparts.
463:
464: For smooth Painlev\'e equations similar asymptotic problems have been studied in the frames of the
465: isomonodromic deformation method \cite{IN}. In particular,
466: connection formulas were derived. These formulas describe the
467: asymptotics of solutions for $n \to \infty$ as a function of
468: initial conditions. Some discrete Painlev\'e equations were
469: studied in that framework in \cite{FIK}). The geometric origin of
470: our equations permits us to find our asymptotics by bare-handed
471: approach, studying linearized equations.
472:
473:
474:
475:
476:
477:
478:
479:
480: \section{Asymptotics of discrete $z^c$ and ${\rm log}(z)$}
481:
482: To treat $z^c$ and ${\rm log}(z)$ on equal footing we agree that
483: the case $c=0$ in equations corresponds to discrete ${\rm
484: log}(z)$. For the edges of unit squares with the vertices in $
485: {\mathbb V}$ defined by (\ref{V}) we introduce $X$ and $Y$ via the
486: radius ratios:
487: \begin{equation} \label{t-def}
488: \frac{1+X_{N,M}}{1-X_{N,M}}=\frac{R_{N+1,M}}{R_{N,M}}, \ \ \ \
489: \frac{1+Y_{N,M}}{1-Y_{N,M}}=\frac{R_{N,M}}{R_{N,M-1}}.
490: \end{equation}
491: In these variables equations (\ref{Ri}) and (\ref{square}) read
492: as:
493: \begin{equation} \label{Ri-t}
494: (M-N)\frac{X_{N,M}+Y_{N,M+1}}{1-X_{N,M}Y_{N,M+1}}+(M+N)\frac{X_{N,M}-Y_{N,M}}{1+X_{N,M}Y_{N,M}}=0,
495: \end{equation}
496: \begin{equation} \label{square-t}
497: (M-N)\frac{Y_{N,M+1}-X_{N,M}}{1-X_{N,M}Y_{N,M+1}}+(M+N+1)\frac{X_{N,M+1}+Y_{N,M+1}}{1+X_{N,M+1}Y_{N,M+1}}=c-1.
498: \end{equation}
499: Moreover, $X$ and $Y$ satisfy
500: \begin{equation} \label{inR}
501: \frac{X_{N,M}+Y_{N,M+1}}{1-X_{N,M}Y_{N,M+1}}=\frac{X_{N-1,M}+Y_{N,M}}{1-X_{N-1,M}Y_{N,M}},
502: \end{equation}
503: \begin{equation} \label{outR}
504: \frac{X_{N,M}+Y_{N+1,M+1}}{1+X_{N,M}Y_{N+1,M+1}}=\frac{X_{N,M+1}+Y_{N,M+1}}{1+X_{N,M+1}Y_{N,M+1}},
505: \end{equation}
506: where (\ref{inR}) is equivalent to (\ref{lnR}) and (\ref{outR})
507: is the compatibility condition.
508:
509: Conditions (\ref{sign}), which hold for discrete $z^c$ and ${\rm
510: log}(z)$, turn out to be so restrictive for the corresponding
511: solutions $X_{N,M},Y_{N,M}$ of (\ref{Ri-t}), (\ref{square-t}),
512: that they allow one to compute their asymptotic behavior. For
513: definiteness we consider the case $c>1$.
514: \begin{lemma} \label{estimate}
515: For the solution $X_{N,M},Y_{N,M}$ of
516: (\ref{Ri-t}),(\ref{square-t}) corresponding to $z^c$ in ${\mathbb
517: V}$ with $c>1$ holds true:
518: \begin{equation} \label{estimation}
519: -\frac{c-1}{M-N}\le X_{N,M} \le \frac{c-1}{M+N} \ \ \ {\rm and}\ \
520: 0 \le Y_{N,M+1} \le \frac{c-1}{M+N}+\frac{2(c-1)}{M-N}.
521: \end{equation}
522: \end{lemma}
523: \noindent {\it Proof:} Note that for the studied solutions
524: $R\ge0$ and therefore $-1\le X_{N,M}\le 1$ and $-1\le
525: Y_{N,M}\le1$. Let us denote for brevity $X_{N,M}$ by $X$,
526: $X_{N,M+1}$ by $\bar X$, $Y_{N,M}$ by $Y$, $Y_{N,M+1}$ by $\bar
527: Y$, and $R_{N,M}$ by $R_1$, $R_{N+1,M}$ by $R_2$, $R_{N+1,M+1}$ by
528: $R_3$, $R_{N,M+1}$ by $R_4$, $R_{N,M-1}$ by $R_5$. Then
529: (\ref{sign}) together with (\ref{Ri}),(\ref{Le}),(\ref{Up}) imply
530: \begin{equation} \label{R-inequalities}
531: R_1^2\le R_2R_4, \ \ R_1^2\ge R_2R_5, \ \ R_3^2\ge R_2R_4, \ \
532: R_4^2\ge R_1R_3, \ \ R_2^2\le R_1R_3.
533: \end{equation}
534: Rewriting the first inequality as $\frac{R_1}{R_2}\le
535: \frac{R_4}{R_1}$ and taking into account $1+X\ge 0$ we have
536: \begin{equation}\label{t1+t2}
537: X+\bar Y \ge 0.
538: \end{equation}
539: Similarly the second inequality of (\ref{R-inequalities}) infers
540: $Y-X\ge 0$ or after shifting
541: \begin{equation}\label{t2t6}
542: \bar Y \ge \bar X.
543: \end{equation}
544: Combining the first and the third inequalities of
545: (\ref{R-inequalities}) one gets $R_3\ge R_1$ or
546: \begin{equation}\label{t2+t6}
547: \bar X +\bar Y \ge 0,
548: \end{equation}
549: which is equivalent to
550: \begin{equation}\label{t5+t1}
551: X +\bar Y_{N+1,M+1} \ge 0
552: \end{equation}
553: due to (\ref{outR}). Similarly the forth and the fifth imply
554: $R_4\ge R_2$ or
555: \begin{equation}\label{t2t1}
556: \bar Y \ge X.
557: \end{equation} Comparing (\ref{t2+t6}) with (\ref{t2t6}) one gets $\bar Y\ge 0$.
558: Rewriting (\ref{square}) as
559: $$
560: \frac{R_3}{R_1}=\frac{(2N+c)R_4+(2M+c)R_2}{(2(M+1)-c)R_4+(2(N+1)-c)R_2}
561: $$
562: and taking into account $\frac{R_3}{R_1}\ge1$ and $c \le 2 $ we
563: can estimate $\frac{R_2}{R_4}$ in ${\mathbb V}$:
564: $$
565: \frac{R_2}{R_4}\ge\frac{1-\frac{c-1}{M-N}}{1+\frac{c-1}{M-N}}
566: $$
567: which reads as
568: $$
569: \bar Y-X \le (1-\bar Y X)\epsilon _{\scriptscriptstyle N,M}, \ \ \
570: \epsilon _{\scriptscriptstyle N,M}=\frac{c-1}{M-N}.
571: $$
572: As $(1-\bar Y X)\le 2$ we have
573: \begin{equation}\label{t2-t1e}
574: \bar Y-X \le 2\epsilon _{\scriptscriptstyle N,M}.
575: \end{equation}
576: Similarly we can solve (\ref{square}) with respect to
577: $\frac{R_4}{R_2}$:
578: $$
579: \frac{R_4}{R_2}=\frac{(2M+c)R_1+(-2(N+1)+c)R_3}{(-2N-c)R_1+(2(M+1)-c)R_3}.
580: $$
581: Note that for $M > N$ holds $(-2N-c)R_1+(2(M+1)-c)R_3\ge 0$ as
582: $R_3\ge R_1$, and $c\le 2$. This together with $R_4\ge R_2$ and
583: $(1-\bar Y \bar X)\le 2$ gives after some calculations
584: \begin{equation}\label{t2+t6d}
585: \bar X+\bar Y \le 2\delta _{\scriptscriptstyle N,M},
586: \end{equation}
587: with $\delta _{\scriptscriptstyle N,M}=\frac{c-1}{M+N+1}$. Now
588: inequalities (\ref{t1+t2}),(\ref{t2t1}),(\ref{t2-t1e}) yield
589: $$
590: -\epsilon _{\scriptscriptstyle N,M}\le X
591: $$
592: and (\ref{t2t6}),(\ref{t2+t6}),(\ref{t2+t6d}) imply
593: $$
594: \bar X \le \delta _{\scriptscriptstyle N,M},
595: $$
596: which gives the first inequality of (\ref{estimation}) after
597: shifting backwards from $M+1$ to $M$ . Using
598: (\ref{t2-t1e}) we easily get the second one. $\Box$\\
599:
600: The estimations obtained allow one to find asymptotic behavior
601: for the radius-function
602: in $M-$direction.
603: \begin{theorem}\label{M-ass}
604: For the solution $R_{N,M}$ of (\ref{Ri}),(\ref{square})
605: corresponding to discrete $Z^c$ and $\rm{Log}$ in ${\mathbb V}$
606: holds true:
607: \begin{equation}\label{as2}
608: R_{N_0,M} \simeq K(c)M^{c-1} \ \ \ {\rm as} \ \ \ M\to \infty,
609: \end{equation}
610: with constant $K(c)$ independent of $N_0$.
611: \end{theorem}
612: \noindent {\it Proof:} Because of the duality
613: $$R_{z^{2-c}}=\frac{1}{R_{z^c}}$$ it is enough to consider the case $c > 1$.
614: Let us introduce $n=M-N_0$, $x_n=X_{N_0,M}$, $y_n=Y_{N_0,M}$.
615: Then for large $n$ Lemma \ref{estimate} allows one to rewrite
616: equations (\ref{Ri-t}),(\ref{square-t}) as
617: \begin{equation} \label{liearize}
618: \begin{array}{l}
619: x_{n+1}=5x_n-2y_n+\frac{c-1}{n}+U_{N_0}(n),\\
620: y_{n+1}=-2x_n+y_n+V_{N_0}(n),
621: \end{array}
622: \end{equation}
623: where $U_{N_0}(n)$ and $V_{N_0}(n)$ are defined by discrete $Z^c$
624: and satisfy
625: \begin{equation} \label{restAB}
626: U_{N_0}(n) < C_1\frac{1}{n^2}, \ \ \ V_{N_0}(n) <
627: C_2\frac{1}{n^2},
628: \end{equation}
629: for natural $n$ and some constants $C_1,C_2$ (depending on $N_0$).
630: Thus the solution $(x_n,y_n)$, corresponding to discrete $Z^c$ is
631: a special solution of linear non-homogeneous system
632: (\ref{liearize}) having the order $\frac{1}{n}$ for large $n$.
633: The eigenvalues of the system matrix are positive numbers
634: $\lambda _1=3-2\sqrt 2<1$ and $\lambda _2=3+2\sqrt 2>1.$ In the
635: diagonal form system (\ref{liearize}) takes the form:
636: \begin{equation} \label{diagonal}
637: \varphi _{n+1}= A\varphi _n + s_n+r_n
638: \end{equation}
639: with
640: $$
641: A=\left (\begin{array}{cc} \lambda _1 & 0 \\
642: 0& \lambda _2
643: \end{array} \right ), \ \ \varphi _n=\left (\begin{array}{cc} 1, & 1 \\
644: 1+\sqrt 2 ,& 1-\sqrt 2
645: \end{array} \right )^{-1}\left (\begin{array}{c} x_n \\
646: y_n
647: \end{array} \right ), \ \ \ s_n=\left (\begin{array}{c} 2-\sqrt 2 \\
648: 2+\sqrt 2
649: \end{array} \right )\frac{c-1}{4n},
650: $$
651: and $|r_n|\le \frac{G}{n^2}$ for some $G$. Looking for the
652: solution in the form
653: $$
654: \varphi _n =A^n c_n
655: $$
656: we have the following recurrent formula for $c_n$:
657: $$
658: c_{n+1}=A^{-n}(s_n+r_n)+c_n.
659: $$
660: Integrating one gets
661: $$
662: c_{n+1}=c_1+A^{-2}(s_1+r_1)+A^{-3}(s_2+r_2)+ ... +
663: A^{-1-n}(s_n+r_n).
664: $$
665: For components of $\varphi_n=(a_n,b_n)^T$ this implies
666: $$
667: \begin{array}{l}
668: a_n=\lambda_1^n \left(a_1 + \frac{(2-\sqrt 2)(c-1)}{4}\left(
669: \frac{1}{\lambda_1^2}+\frac{1}{2\lambda_1^3}+...+\frac{1}{(n-1)\lambda_1^n}
670: \right) +\left(
671: \frac{G_1}{\lambda_1^2}+\frac{G_2}{2^2\lambda_1^3}+...+\frac{G_{n-1}}{(n-1)^2\lambda_1^n}
672: \right) \right)=
673: \\
674: = a_1\lambda_1^n + \frac{(2-\sqrt 2)(c-1)}{4}\left(
675: \frac{1}{n-1}+\frac{\lambda_1}{(n-2)}+...+\lambda_1^{n-2}
676: \right) +\left(
677: \frac{G_1}{\lambda_1^2}+\frac{G_2}{2^2\lambda_1^3}+...+\frac{G_{n-1}}{(n-1)^2\lambda_1^n}
678: \right)
679: \end{array}
680: $$
681: with some limited sequence $G_n$: $|G_n|\le G$. The first sum,
682: corresponding to $s_n$, is estimated as follows:
683: $$
684: \begin{array}{l}
685: \frac{1}{n-1}\left(1+\frac{n-1}{n-2}\lambda_1+\frac{n-1}{n-3}\lambda_1^2...+(n-1)\lambda_1^{n-2}\right)=\\
686: =\frac{1}{n-1}\left(1+\lambda_1+\lambda_1^2...+\lambda_1^{n-2}\right)+
687: \frac{\lambda_1}{n-1}\left(\frac{1}{n-2}+\frac{2}{n-3}\lambda_1...+(n-2)\lambda_1^{n-3}\right)=\\
688: =\frac{1}{n-1}\frac{1-\lambda_1^{n-1}}{1-\lambda_1}+
689: \frac{\lambda_1}{(n-1)(n-2)}\left(1+\frac{2(n-2)}{n-3}\lambda_1+\frac{3(n-2)}{n-4}\lambda_1^2...+(n-2)^2\lambda_1^{n-3}\right)=\\
690: =\frac{1}{n-1}\frac{1-\lambda_1^{n-1}}{1-\lambda_1}+
691: \frac{\lambda_1}{(n-1)(n-2)}F_1(n,\lambda_1)
692: \end{array}
693: $$
694: where
695: $$
696: F_1(n,\lambda_1)<(1+2\times 3 \lambda_1+3\times 4
697: \lambda_1^2+...)=\frac{2}{(1-\lambda_1)^3}-1,
698: $$
699: as $|\lambda_1|<1$ and $(n-2)/(n-k)<k$ for $k\le n-1$. The second
700: sum is estimated by
701: $$
702: \begin{array}{l}
703: \frac{G}{(n-1)^2}\left(1+\frac{(n-1)^2}{(n-2)^2}\lambda_1+\frac{(n-1)^2}{(n-3)^2}\lambda_1^2
704: +...+\frac{(n-1)^2}{(n-k)^2}\lambda_1^{k-1}+...+(n-1)^2\lambda_1^{n-2}\right)=\\
705: =\frac{G}{(n-1)^2}(1+\lambda_1+\lambda_1^2+...+\lambda_1^{n-2}+\\
706: +\frac{(n-1)^2-(n-2)^2}{(n-2)^2}\lambda_1+\frac{(n-1)^2-(n-3)^2}{(n-3)^2}\lambda_1^2
707: +...+\frac{(n-1)^2-(n-k)^2}{(n-k)^2}\lambda_1^{k-1}+...+((n-1)^2-1)\lambda_1^{n-2})=\\
708: =\frac{G}{(n-1)^2}\left(\frac{1-\lambda_1^{n-1}}{1-\lambda_1}+
709: \frac{(2(n-2)+1}{(n-2)^2}\lambda_1
710: +...+\frac{(2(n-k)(k-1)+(k-1)^2}{(n-k)^2}\lambda_1^{k-1}+...+(2(n-2)+(n-2)^2)\lambda_1^{n-2}\right)\le
711: \\
712: \le \frac{G}{(n-1)^2}\left(\frac{1}{1-\lambda_1}+
713: 2\lambda_1(1+2\lambda_1+3\lambda_1^2+...)+\lambda_1(1+2^2\lambda_1+3^2\lambda_1^2+...)
714: \right)=\frac{G}{(n-1)^2}F_2(\lambda_1).
715: \end{array}
716: $$
717: Summing up we conclude:
718: \begin{equation}\label{an}
719: a_n=\frac{(2-\sqrt
720: 2)(c-1)}{4(1-\lambda_1)}\frac{1}{n}+O\left(\frac{1}{n^2}\right).
721: \end{equation}
722: For the second component $b_n$ one has
723: $$
724: \begin{array}{l}
725: b_n = b_1\lambda_2^n + \frac{(2+\sqrt 2)(c-1)}{4}\lambda_2^n\left(
726: \frac{1}{\lambda_2^2}+\frac{1}{2\lambda_2^3}+...+\frac{1}{(n-1)\lambda_2^n}
727: \right) +\lambda_2^n\left(
728: \frac{H_1}{\lambda_2^2}+\frac{H_2}{2^2\lambda_2^3}+...+\frac{H_{n-1}}{(n-1)^2\lambda_2^n}
729: \right)
730: \end{array}
731: $$
732: with some limited sequence $H_n$: $|H_n|\le H$. The first sum in
733: the previous formula is estimated as
734: $$
735: \begin{array}{l}
736: \lambda_2^{n-1}\int_0^{\frac{1}{\lambda_2}}
737: (1+x+x^2...+x^{n-2})dx=\lambda_2^{n-1}\int_0^{\frac{1}{\lambda_2}}
738: \frac{1-x^{n-1}}{1-x}dx=\\
739: =\lambda_2^{n-1}\left(\int_0^{\frac{1}{\lambda_2}}
740: \frac{1}{1-x}dx-\int_0^{\frac{1}{\lambda_2}} (x^{n-1}+x^n+...)dx
741: \right)=\\
742: =\lambda_2^{n-1}\left(-\ln
743: (1-x)\mid_0^{\frac{1}{\lambda_2}}-\frac{1}{\lambda_2^n}\left(\frac{1}{n}+\frac{1}{(n+1)\lambda_2}+\frac{1}{((n+2)\lambda_2^2}...
744: \right) \right)=\\
745: =\lambda_2^{n-1}\ln\frac{\lambda_2}{\lambda_2-1}-\frac{1}{n\lambda_2}\left(1+\frac{n}{(n+1)\lambda_2}+\frac{n}{((n+2)\lambda_2^2}...
746: \right)=\\
747: =\lambda_2^{n-1}\ln\frac{\lambda_2}{\lambda_2-1}-\frac{1}{n\lambda_2}\left(
748: \left(1+\frac{1}{\lambda_2}+\frac{1}{\lambda_2^2}+... \right)-
749: \frac{1}{\lambda_2}\left(\frac{1}{(n+1)}+\frac{2}{(n+2)\lambda_2}+...+\frac{k}{(n+k)\lambda_2^{k-1}}+...
750: \right)\right)=\\
751: =\lambda_2^{n}\frac{1}{\lambda_2}\ln\frac{\lambda_2}{\lambda_2-1}-\frac{1}{n(\lambda_2-1)}+\frac{S_n(\lambda_2)}{n(n+1)\lambda_2^2},
752: \end{array}
753: $$
754: Where
755: $$
756: \begin{array}{l}
757: S_n(\lambda_2)=\left(1+\frac{2(n+1)}{(n+2)\lambda_2}+\frac{3(n+1)}{(n+3)\lambda_2^2}+...+\frac{k(n+1)}{(n+k)\lambda_2^{k-1}}+...
758: \right)<\left(1+\frac{2}{\lambda_2}+\frac{3}{\lambda_2^2}+...+\frac{k}{\lambda_2^{k-1}}+...
759: \right)=\frac{\lambda_2^2}{(\lambda_2-1)^2}.
760: \end{array}
761: $$
762: For the second sum in the formula for $b_n$ one has
763: $$
764: \begin{array}{l}
765: \frac{1}{\lambda_2^2}\left(
766: H_1+\frac{H_2}{2^2\lambda_2}+...+\frac{H_{n-1}}{(n-1)^2\lambda_2^{n-2}}
767: \right)=F_3(\lambda_2)-\frac{1}{\lambda_2^2}\left(
768: \frac{H_n}{n^2\lambda_2^{n-1}}+\frac{H_{n+1}}{(n+1)^2\lambda_2^{n}}+...+\frac{H_{n+k}}{(n+k)^2\lambda_2^{n+k-1}}+...
769: \right)
770: \end{array}
771: $$
772: where
773: $$
774: \begin{array}{l}
775: F_3(\lambda_2)=\frac{1}{\lambda_2^2}\left(
776: H_1+\frac{H_2}{2^2\lambda_2}+...+\frac{H_{k-1}}{(k-1)^2\lambda_2^{k-2}}+...
777: \right)
778: \end{array}
779: $$
780: and the second sum is estimated from above as
781: $$
782: \begin{array}{l}
783: \frac{H}{\lambda_2^{n+1}n^2}\left(1+
784: \frac{n^2}{(n+1)^2\lambda_2}+\frac{n^2}{(n+2)^2\lambda_2^2}+...+\frac{n^2}{(n+k)^2\lambda_2^k}+...
785: \right)<\frac{H}{\lambda_2^{n+1}n^2}\left(1+
786: \frac{1}{\lambda_2}+\frac{1}{\lambda_2^2}+\frac{1}{\lambda_2^2}+...
787: \right)=\frac{H}{\lambda_2^{n}n^2(\lambda_2-1)}.
788: \end{array}
789: $$
790: Finally
791: \begin{equation}\label{bn}
792: b_n=\lambda_2^n\left(b_1+\frac{(2+\sqrt
793: 2)(c-1)}{4\lambda_2}\ln\frac{\lambda_2}{\lambda_2-1}+
794: F_3(\lambda_2)\right) -\frac{(2+\sqrt
795: 2)(c-1)}{4(\lambda_2-1)}\frac{1}{n}+O\left(\frac{1}{n^2}\right).
796: \end{equation}
797: As $b_n\to 0$ with $n\to \infty$ and $\lambda_2>1$ one deduces
798: that the coefficient by $\lambda_2^n$ vanishes. For original
799: variables $x_n, y_n$ the found asymptotics (\ref{an}),(\ref{bn})
800: have especially simple form:
801: \begin{equation}\label{xnyn}
802: \left(
803: \begin{array}{l}
804: x_n\\
805: y_n
806: \end{array}\right)
807: =\frac{c-1}{2n} \left(
808: \begin{array}{l}
809: 0\\
810: 1
811: \end{array}\right)+O\left(\frac{1}{n^2}\right).
812: \end{equation}
813: Asymptotic (\ref{xnyn}), the second equation (\ref{t-def}) and
814: $$
815: R_{N_0,N_0+n}=R_{N_0,N_0}\prod
816: \limits_{k=1}^n\frac{2k+(c-1)}{2k-(c-1)}
817: $$
818: imply (\ref{as2}). The independence of $K(c)$ on $N_0$ easily
819: follows from the first equation (\ref{t-def}) and $x_n\to 0$.
820: $\Box$
821:
822: \section{Discussion and concluding remarks}
823: Found asymptotics implies
824: \begin{equation}\label{as4}
825: \tan\alpha _n \simeq \left(1+\frac{1}{n}\right)^{c-1} \ \ \ {\rm
826: as} \ \ \ n\to \infty
827: \end{equation}
828: for the corresponding solution $u_n=e^{i\alpha_n}$ of
829: (\ref{dPII}).\\ Further, equation (\ref{c}) allows one to
830: "integrate" asymptototics (\ref{as2}) to get
831: \begin{equation}\label{as3}
832: Z^c(n_0+n,m_0+n) \simeq e^{c \pi i/4}K(c)n^{c} \ \ \ {\rm as} \ \
833: \ n\to \infty.
834: \end{equation}
835: Thus the circles of $Z^c$ not only cover the whole infinite
836: sector with the angle $c\pi/2$ but the circle centers and
837: intersection points mimic smooth map $z\to z^c$ also
838: asymptotically. Morover, $R(z)$, being analogous to $|f'(z)|$ of
839: the corresponding smooth map, has the "right" asymptotics as well.
840:
841: Discrete map $Z^c$ is defined via constraint (\ref{c}) which is
842: isomonodromy condition for some linear equation. This seems to be
843: rather far-fetched approach. It would be more naturally to define
844: $Z^c$ via circle patterns in a pure geometrical way: square grid
845: $Z^c$ with $0 <c<2$ is an embedded infinite square grid circle
846: pattern, the circles $C(z)$ being labeled by
847: $$
848: {\mathbb V}=\{z=N+iM:\ N,M \in {\bf Z^2}, M \ge |N| \},
849: $$
850: satisfying the following conditions:\\
851: 1) the circles cover the infinite sector with the angle
852: $c\pi/2$,\\
853: 2) the centers of the border circles $C(N+iN)$ and $C(-N+iN)$ lie
854: on the borders of this sector.
855: \begin{conjecture}\label{geodefcon}
856: Up to re-scaling there is unique square grid $Z^c$.
857: \end{conjecture}
858:
859: A naive method to look for so defined discrete $Z^c$ is to start
860: with some equidistant $f_{n,0}\in \mathbb R$, $f_{0,m}\in i^{c
861: }\mathbb R$:
862: $$
863: |f_{2n,0}-f_{2n+1,0}|=|f_{2n,0}-f_{2n-1,0}|=|f_{0,2n}-f_{0,2n+1}|=|f_{0,2n}-f_{0,2n-1}|
864: $$
865: and then compute $f_{n,m}$ for any $n,m > 0$ using equation
866: (\ref{q}). Such $f_{n,m}$ determines some circle patterns as
867: (\ref{equi}) holds true. But so determined map has a not very
868: nice behavior (see an example in Fig.~\ref{correct-wrong}).
869: \begin{figure}[th]
870: \begin{center}
871: \epsfig{file=zalpha_wrong1.eps,width=60mm} \caption{Non-immersed
872: discrete conformal map} \label{correct-wrong}
873: \end{center}
874: \end{figure}
875: For $c=2/k,\ k\in \mathbb N$ the proof of Conjecture
876: \ref{geodefcon} easily follows from the rigidity results obtained
877: in \cite{H}.
878:
879: Infinite embedded circle patterns define some (infinite) convex
880: ideal polyhedron in $\mathbb H^3$ and the group generated by
881: inversions in its faces. The known results (see for example
882: \cite{S}) imply the conjecture claim for rational $c$, but seem
883: to be inapplicable for irrational as the group generated but the
884: circles and the sector borders is not discrete any more.
885: Unfortunately, the rigidity results for finite polyhedra \cite{Ri}
886: does not seem to be carried over to infinite case by induction.
887:
888: One can also consider solutions of (\ref{q}) subjected to
889: (\ref{c}) where $n$ and $m$ are not integer. It turned out that
890: there exist initial data, so that the corresponding solution
891: define immersed circle patterns. (The equations for radii are the
892: same and therefore compatible. Existence of positive solution is
893: provable by the arguments for the corresponding discrete
894: Painlev\'e equation as in \cite{AB}.) It is natural to call such
895: circle patterns discrete map $z\to (z+z_0)^c$. In this case it
896: can not be defined pure geometrically as the centers of border
897: circles are not collinear.
898:
899: One can discard the restriction $0 <c<2$ (as well as some
900: circles $C(z)$ with small $z$) to define discrete $Z^K$ for
901: natural $K>2$ (see \cite{AB}). All the asymptotic results
902: obtained so far can be carried over on this case as well as on
903: hexagonal $Z^c$ with constant intersection angles defined in
904: \cite{BH} since the governing equations are essentially the same
905: (\cite{ABh}).
906:
907:
908:
909:
910: \section{Acknowledgements}
911:
912: The author thanks B.Apanasov, A.Bobenko and Yu.Suris for useful
913: discussions.
914:
915:
916:
917: \begin{thebibliography}{99}
918:
919: \addcontentsline{toc}{section}{References}
920: \bibitem{A}
921: S.I.Agafonov, Imbedded circle patterns with the combinatorics of
922: the square grid and discrete Painlevé equations. {\it Discrete
923: Comput. Geom.,} {\bf 29} (2003), no. 2, 305--319.
924:
925: \bibitem{AB}
926: S.I.Agafonov, A.I.Bobenko, Discrete $Z^{\gamma}$ and Painlev\'e
927: equations. {\it Internat. Math. Res.
928: Notices,} {\bf 4} (2000), 165-193.
929:
930: \bibitem{ABh}
931: S.I.Agafonov, A.I.Bobenko, Hexagonal circle patterns with
932: constant intersection angles and discrete Painlevé and Riccati
933: equations. Integrability, topological solitons and beyond. {\it
934: J. Math. Phys.} {\bf 44} (2003), no. 8, 3455--3469.
935:
936: \bibitem{B} A.I.Bobenko, Surfaces made from circles. Eurographics/ACM SIGGRAPH Symposium
937: "Geometry Processing", Nice 2004. (www-version
938: http://www.math.tu-berlin.de/~bobenko/visualization.html)
939:
940: \bibitem{BHConf} A.I.Bobenko, T.Hoffman,
941: Conformally symmetric circle packings. A generalization of Doyle spirals,
942: {\it Experimental Math.,} {\bf 10:1} (2001), 141-150.
943:
944: \bibitem{BHSp}
945: A.I.Bobenko, T.Hoffmann, B.A.Springborn, Minimal surfaces from
946: circle patterns: Geometry from combinatorics. Preprint (2003)
947: math.DG/0305184
948:
949: \bibitem{BHS} A.I.Bobenko, T.Hoffman, Yu.B.Suris,
950: Hexagonal circle patterns and integrable systems:
951: Patterns with the multi-ratio property and Lax equations on the
952: regular triangular lattice,. {\it Internat. Math. Res.
953: Notices,} {\bf 3} (2002), 111-164.
954:
955: \bibitem{BH} A.I.Bobenko, T.Hoffman,
956: Hexagonal circle patterns and integrable systems: patterns with constant angles. {\it Duke Math. J.} {\bf 116} (2003), no. 3, 525--566.
957:
958:
959: \bibitem{BMS} A.I.Bobenko, Ch.Mercat,Yu.B.Suris, Linear and nonlinear
960: theories of discrete analytic functions. Integrable structure and
961: isomonodromic Green function. Preprint (2004) math.DG/0402097
962:
963: \bibitem{BPdis} A.Bobenko, U.Pinkall,
964: Discrete isothermic surfaces. {\it J. reine angew. Math.} {\bf
965: 475} (1996), 187-208.
966:
967: \bibitem{BPD}
968: A.Bobenko, U.Pinkall, Discretization of surfaces and integrable
969: systems.
970: In: {\it Discrete Integrable Geometry and Physics;} Eds.
971: A.I.Bobenko and
972: R.Seiler, pp. 3-58, Oxford University Press, 1999.
973:
974: \bibitem{Doy}
975: K.Callahan, B.Rodin, Circle packing immersions form regularly
976: exhaustible surfaces. {\it Complex Variables,} {\bf 21} (1993),
977: 171-177.
978:
979: \bibitem{DS} T.Dubejko, K.Stephenson, Circle packings: Experiments in discrete
980: analytic function theory. {\it Experimental Math.} {\bf 4:4}, (1995),
981: 307-348.
982:
983: \bibitem{FIK} A.S.Fokas,
984: A.R.Its, A.V.Kitaev, Discrete Painlev\'e equations and their
985: appearance in quantum gravity. {\it Commun. Math. Phys.} {\bf 142}
986: (1991), 313-344.
987:
988: \bibitem{HS}
989: Z.-X.He, O.Schramm, The $C^{\infty}$ convergence of hexagonal
990: disc packings to the Riemann map. {\it Acta. Math.} {\bf 180}
991: (1998), 219-245.
992:
993: \bibitem{H} Z.-X.He, Rigidity of infinite disk patterns. {\it Annals
994: of Mathematics} {\bf 149} (1999), 1-33.
995:
996: \bibitem{IN} A.R.Its, V.Y.Novokshenov, The
997: isomonodromic deformation method in the theory of Painlev\'e
998: equations. Lecture Notes in Math. Vol. 1191, Springer, Berlin,
999: 1986.
1000:
1001: \bibitem{K} R.Kenyon, The Laplacian and Dirac operators on
1002: critical planar graphs. {\it Invent. Math.} {\bf 150} (2002), no.
1003: 2, 409--439.
1004:
1005: \bibitem{MR} A.Marden, B.Rodin, On Thurston's formulation and proof of
1006: Andreev's theorem. Lecture Notes in Math., Vol. 1435, (1990), 103-115.
1007:
1008: \bibitem{N} F.Nijhoff, On some "Schwarzian" equations and their discrete analogues. In: Algebraic aspects of
1009: integrable systems, In memory of Irene Dorfman, Eds.: A.S.Fokas
1010: and I.M.Gelfand, Birkh\"auser, 1997, 237-260.
1011:
1012: \bibitem{Ri}
1013: I.Rivin, On geometry of convex ideal polyhedra in hyperbolic
1014: $3$-space. {\it Topology}, {\bf 32} (1993), no. 1, 87--92.
1015:
1016: \bibitem{RS}
1017: B.Rodin, D.Sullivan, The convergence of circle packings to
1018: Riemann mapping. {\it J. Diff. Geometry} {\bf 26}
1019: (1987), 349-360.
1020:
1021: \bibitem{S}
1022: D.Sullivan, On the ergodic theory at infinity of an arbitrary
1023: discrete group of hyperbolic motions. Riemann surfaces and
1024: related topics: Proceedings of the 1978 Stony Brook Conference,
1025: (1978), {\it Ann. of Math. Stud.}, 97, Princeton Univ. Press,
1026: Princeton, N.J., (1981), 465--496.
1027:
1028: \bibitem{Schramm} O.Schramm,
1029: Circle patterns with the combinatorics of the square grid. {\it
1030: Duke Math. J.} {\bf 86} (1997), 347-389.
1031:
1032: \bibitem{www} O.Schramm, www-home page, Circle packings and conformal geometry, a survey of selected
1033: topics. http://www.math.weizmann.ac.il/\~\ schramm/talks/
1034:
1035: \bibitem{T} W.P.Thurston, The finite Riemann mapping
1036: theorem. Invited address, International Symposium in Celebration
1037: of the Proof of the Bieberbach Conjecture, Purdue University
1038: (1985).
1039:
1040: \end{thebibliography}
1041:
1042:
1043: \end{document}
1044: