1: \section{Multifractal (or Topological) Zeta Functions}\label{mzf}
2:
3: The multifractal zeta function, which made its first appearance in
4: \cite{LLVR}, was initially developed to investigate the properties
5: of multifractal measures. Its definition also relies on the notion
6: of regularity, but the lengths come from a much larger and more
7: complicated family than the family given by $\mathfrak{P}$ for the
8: partition zeta functions. This larger collection creates many
9: computational and theoretical difficulties, yet two regularity
10: values ($\pm \infty$) yield new and existing results for fractal
11: strings. The results presented in this section are originally from
12: \cite{LLVR} (viewed from a slightly different perspective), and can
13: also be found in \cite{Rock} (viewed from the current perspective of
14: this work).
15:
16: \ndnt In order to bring fractal strings into a framework that uses
17: regularity, an appropriate measure must be defined. For a fractal
18: string $\Omega$ in the unit interval, the measure $\mu_{\Omega}$ is
19: the measure which has a unit point-mass at every endpoint of the
20: fractal string. Thus, any interval which does not contain an
21: endpoint of $\Omega$ does not have mass, hence its regularity is
22: $\infty$. On the other hand, any interval which contains a
23: neighborhood of a limit point of the endpoints of $\Omega$ has
24: infinite mass, hence its regularity is $-\infty$. These measures
25: combine with the multifractal zeta functions to recover and extend
26: the results obtained for fractal strings through the geometric zeta
27: functions.
28:
29: \ndnt Using intervals whose lengths appear in a sequence
30: $\mathcal{N}$ (which decreases to zero) and collecting them
31: according to their regularity $\alpha$ allow for the definition of
32: multifractal zeta function given below, where $k_{n}(\alpha)$ is the
33: number of new disjoint intervals $K^{n}_{p}(\alpha)$ which arise at
34: stage $n$. The intervals $K^{n}_{p}(\alpha)$ do not necessarily have
35: length in $\mathcal{N}$, rather they are the disjoint intervals of
36: the set which is the union of all closed intervals of length $\eta_n
37: \in \mathcal{N}$ and the same regularity $\alpha$.
38:
39: \begin{dfn}\label{def:mzf} The \underline{multifractal zeta function} of a measure
40: $\mu$, sequence $\mathcal{N}$, associated regularity value $\alpha
41: \in [-\infty,\infty]$ and is given by
42: \[
43: \zeta^{\mu}_{\mathcal{N}}(\alpha,s) =
44: \sum_{n=1}^{\infty}\sum_{p=1}^{k_{n}(\alpha)}|K^{n}_{p}(\alpha)|^s
45: \]
46: for \textnormal{Re}$(s)$ large enough.
47: \end{dfn}
48:
49:
50:
51: \begin{figure}
52: \epsfysize=3cm\epsfbox{threecantorstrings.eps}
53: \caption{\textit{Three strings with the same lengths
54: $\mathcal{L}$, the lengths of the Cantor String $\Omega_1$,
55: but different topological configuration. }}
56: \end{figure}
57:
58:
59: \ndnt In this setting, we have the following theorem. The full proof
60: can be found in \cite{LLVR,Rock}. The basic idea of the proof is
61: that an interval with regularity $\alpha = \infty$ has no mass, thus
62: this interval must be a subset of the complement of the support of
63: the measure, the fractal string $\Omega_{\mu} =(supp(\mu))^c$. The
64: decreasing sequence $\mathcal{N}$ ensures that every disjoint open
65: interval in this fractal string is recovered, in turn enabling us to
66: recover the geometric zeta function. Unlike some of the other
67: results on fractal strings mentioned in this work, the following
68: theorem does not require the fractal string to have total length 1,
69: to be a subset of $[0,1]$, nor to have boundary equal to the
70: complement in the smallest compact interval which contains the
71: fractal string.
72:
73: \begin{thm}\label{thm:recovergzf}
74: The multifractal zeta function of a positive Borel measure $\mu$,
75: any sequence $\mathcal{N}$ such that \(\eta_{n} \searrow 0\) and
76: regularity $\alpha = \infty$ is the geometric zeta function of
77: $\Omega_{\mu} =(supp(\mu))^c$ (where the complement is taken in the
78: smallest compact interval containing $supp(\mu)$), with lengths
79: $\mathcal{L}_{\mu}$. That is, \( \zeta^{\mu}_{\mathcal{N}}(\infty,s)
80: = \zeta_{\mathcal{L}_{\mu}}(s). \)
81: \end{thm}
82:
83: \ndnt When a fractal string which has a boundary (complement in the
84: unit interval) that is a perfect set, such as the Cantor String, we
85: get the following theorem with omitted proof. The full development
86: and proof can be found in \cite{LLVR,Rock}. The regularity value
87: $-\infty$ allows us to distinguish between fractal strings with
88: identical lengths $\mathcal{L}$ and, hence, the same Minkowski
89: dimension, but with different topological arrangements. See Figure 6
90: for approximations of three fractal strings which have the same
91: $\mathcal{L}$ (the lengths of the Cantor String $\Omega_1$), but
92: have obviously distinct topological properties. In light of the
93: following theorem, one may refer to the multifractal zeta function
94: of a measure $\mu_{\Omega}$ with regularity $\alpha = -\infty$ as
95: the {\it topological} zeta function of the fractal string $\Omega$.
96:
97:
98:
99: \begin{thm}\label{thm:perfect}
100: Let $\Omega$ be a fractal string with sequence of lengths
101: $\mathcal{L}$ and perfect boundary. Suppose that $\mathcal{N}$ is a
102: sequence which decreases to zero such that \(l_n>\eta_n \geq
103: l_{n+1}\) and \(l_n>2\eta_n,\) for all \(n \in \mathbb{N}.\) Then
104:
105: \begin{eqnarray}
106: \zeta^{\mu_{\Omega}}_{\mathcal{N}}(\infty,s) &=&
107: \zeta_{\mathcal{L}}(s) = \sum_{n=1}^{\infty}m_nl_{n}^{s}, \textnormal{ and}\\
108: %\zeta^{\mu_{\Omega}}_{\mathcal{N}}(0,s) &=&
109: %\sum_{n=1}^{\infty}2m_n\eta_n^s, \textit{ and }\\
110: \zeta^{\mu_{\Omega}}_{\mathcal{N}}(-\infty,s) &=&
111: h(s)+\sum_{n=2}^{\infty}m_n(l_n-2\eta_{n})^s,
112: \end{eqnarray}
113:
114: where $h(s)$ is the entire function given by \(h(s)=
115: \sum_{p=1}^{k_1(-\infty)}|K^{1}_{p}(-\infty)|^s. \)
116: \end{thm}
117:
118:
119: \ndnt For $q=1,2,3$ and fractal strings $\Omega_q$, Theorem
120: \ref{thm:perfect} yields
121:
122: \[
123: \zeta^{\mu_q}_{\mathcal{N}}(\infty,s) = \zeta_{CS}(s) =
124: \frac{3^{-s}}{1-2\cdot3^{-s}}.
125: \]
126:
127: However, the multifractal zeta functions corresponding to the
128: regularity value $-\infty$ have the following forms, where the
129: function $h_3$ is entire:
130:
131: \begin{eqnarray*}
132: \zeta^{\mu_1}_{\mathcal{N}}(-\infty,s) &=&
133: 2\left( \frac{1}{3}+\frac{1}{9} \right)^s
134: +\sum_{n=2}^{\infty} 2^{n-1} \left(
135: \frac{1}{3^n}-\frac{2}{3^{n+1}} \right)^s\\
136: &=& 2\left( \frac{4}{9} \right)^s
137: +\frac{2}{27^s} \left(\frac{1}{1-2\cdot3^{-s}} \right),\\
138: \zeta^{\mu_2}_{\mathcal{N}}(-\infty,s) &=& \eta_{1}^s =
139: \frac{1}{9^s},\\
140: \zeta^{\mu_3}_{\mathcal{N}}(-\infty,s) &=& h_3(s)
141: +\sum_{n=2}^{\infty}m_n \left(l_{2n-1}+l_{2n}
142: -2\eta_{2n-1}\right)^s\\
143: &=& h_3(s) +\sum_{n=2}^{\infty}2^{n-1} \left( \frac{1}{3^{2n-1}}
144: + \frac{1}{3^{2n}} - \frac{2}{3^{2n}}\right)^s\\
145: &=& h_3(s)+ \left(\frac{2^{s+1}}{81^s}\right) \left(
146: \frac{1}{1-2\cdot9^{-s}} \right).\\
147: \end{eqnarray*}
148:
149: \ndnt More definitively, it follows from the above discussion that
150: the poles (complex dimensions) of these multifractal zeta functions
151: differ completely:
152:
153: \begin{eqnarray*}
154: \mathcal{D}^{\mu_1}_{\mathcal{N}}(-\infty) &=& \mathcal{D}_{CS} =
155: \left\{ \log_3{2} +
156: \frac{2i\pi m}{\log3} \hs | \hs m \in \mathbb{Z} \right\},\\
157: \mathcal{D}^{\mu_2}_{\mathcal{N}}(-\infty) &=& \emptyset,
158: \textnormal{\ndnt and}\\
159: \mathcal{D}^{\mu_3}_{\mathcal{N}}(-\infty) &=& \left\{ \log_9{2}+
160: \frac{2i\pi m}{\log9} \hs | \hs m \in \mathbb{Z} \right\}.
161: \end{eqnarray*}
162:
163: \ndnt Thus, multifractal zeta functions for at least two regularity
164: values provide useful information about the properties of fractal
165: strings when certain measures are considered, specifically the
166: measures which have unit mass at every endpoint of the disjoint open
167: intervals which define the fractal string. These multifractal zeta
168: functions were the starting point for the development of the more
169: refined and relevant (with respect to multifractal analysis)
170: partition zeta functions.
171: