math0503190/im.tex
1: %------------------------------------------------------------------------------
2: % K. Ichihara and S. Mizushima
3: % last update: 2005.9.27 by Mizushima
4: %------------------------------------------------------------------------------
5: 
6: \documentclass{amsart}
7: 
8: \usepackage[dvips]{graphics}
9: 
10: \makeatletter
11: \renewcommand{\p@enumii}{\theenumi-}
12: \makeatother
13: 
14: % Theorem environments
15: \newtheorem{theorem}{Theorem}[section]
16: \newtheorem{corollary}[theorem]{Corollary}
17: \newtheorem{lemma}[theorem]{Lemma}
18: %%% \newtheorem{claim}[theorem]{Claim}
19: %%% \newtheorem*{proposition*}{Proposition}
20: \newtheorem{problem}{Problem}
21: 
22: %%% \theoremstyle{definition}
23: %%% \newtheorem{definition}[theorem]{Definition}
24: 
25: \theoremstyle{remark}
26: \newtheorem{remark}[theorem]{Remark}
27: 
28: \numberwithin{equation}{section}
29: 
30: 
31: % Commands
32: \newcommand{\Diagram}{\mathcal{D}}
33: \newcommand{\Strip}{\mathcal{S}}
34: \newcommand{\Twist}{\tau}
35: \newcommand{\Remainder}{\rho}
36: \newcommand{\Cancel}{\kappa}
37: \newcommand{\Edge}{e}
38: \newcommand{\Edgepath}{\gamma}
39: \newcommand{\ConstantEdgepaths}{\Gamma_{\mathrm{const}}}
40: \newcommand{\NonconstantEdgepaths}{\Gamma_{\mathrm{non-const}}}
41: \newcommand{\EdgepathSystem}{\Gamma}
42: \newcommand{\BasicEdgepath}{\lambda}
43: \newcommand{\BasicEdgepathSystem}{\Lambda}
44: %%% \newcommand{\Diam}{\mathrm{Diam}}
45: 
46: \newcommand{\NumTangles}{N}
47: \newcommand{\Numer}{P}
48: \newcommand{\Denom}{Q}
49: \newcommand{\Slope}{R}
50: 
51: \newcommand{\anglebb}[1]{\langle\langle #1 \rangle\rangle}
52: \newcommand{\angleb}[1]{\langle #1 \rangle} % angle bracket
53: \newcommand{\circlebb}[1]{\langle\langle #1 \rangle\rangle^{\circ}}
54: \newcommand{\circleb}[1]{\langle #1 \rangle^{\circ}} % angle bracket
55: 
56: %%    Blank box placeholder for figures (to avoid requiring any
57: %%    particular graphics capabilities for printing this document).
58: %\newcommand{\blankbox}[2]{%
59: %  \parbox{\columnwidth}{\centering
60: %%    Set fboxsep to 0 so that the actual size of the box will match the
61: %%    given measurements more closely.
62: %    \setlength{\fboxsep}{0pt}%
63: %    \fbox{\raisebox{0pt}[#2]{\hspace{#1}}}%
64: %  }%
65: %}
66: 
67: 
68: %--------------------------%--------------------------%--------------------------
69: 
70: \begin{document}
71: 
72: % #############################################
73: %
74: %           Title, Authors, etc
75: %
76: % #############################################
77: 
78: \title[Bounds on numerical boundary slopes for Montesinos knots]
79: {Bounds on numerical boundary slopes\\ for Montesinos knots}
80: 
81: \author{Kazuhiro Ichihara}
82: \address{
83: College of General Education, 
84: Osaka Sangyo University, 
85: 3--1--1 Nakagaito, Daito, Osaka 574--8530, Japan 
86: }
87: \email{ichihara@las.osaka-sandai.ac.jp}
88: 
89: \author{%
90:     Shigeru Mizushima}
91: \address{%
92:         Department of Mathematical and Computing Sciences \\
93: %       Graduate School of Information Science and Engineering \\
94:         Tokyo Institute of Technology \\
95:         12--1 Ohokayama, Meguro \\
96:         Tokyo 152--8552, Japan}
97: \email{mizusima@is.titech.ac.jp}
98: 
99: \keywords{boundary slopes, Montesinos knots}
100: \subjclass[2000]{Primary 57M50; Secondary 57M25}
101: % 57M50 : Geometric structures on low-dimensional manifolds
102: % 57M25 : Knots and links in S^3
103: % \date{\today}
104: 
105: % #############################################
106: %
107: %                  Abstract
108: %
109: % #############################################
110: 
111: \begin{abstract}
112: We give 
113: an upper bound on the denominators of numerical boundary slopes and
114: an upper bound on the differences between two numerical boundary slopes
115: for Montesinos knots. 
116: \end{abstract}
117: 
118: \maketitle
119: 
120: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
121: %
122: % Section : Introduction
123: %
124: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
125: 
126: \section{Introduction}
127: 
128: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
129: %
130: % MONTESINOS KNOTS and NUMERICAL SLOPES
131: %
132: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
133: 
134: We consider compact connected surfaces properly embedded 
135: in compact orientable irreducible $3$-manifolds with single toral boundary, 
136: which are \textit{essential}, meaning that incompressible and boundary-incompressible. 
137: The boundary of such a surface consists of a parallel family of 
138: non-trivial simple closed curves. 
139: Thus they determine a \textit{slope}, that is, 
140: the isotopy class of non-trivial simple closed curves. 
141: This slope is called the \textit{boundary slope} of the surface. 
142: %
143: Boundary slopes of essential surfaces 
144: have been well-studied, especially, 
145: in a relation to the study of Dehn surgery on knots. 
146: 
147: Recall that, for the knot exteriors in the $3$-sphere $S^3$, 
148: the set of slopes is usually identified with the set 
149: of rational numbers with the infinity $\infty$. 
150: In fact, such an identification can be done 
151: by using the standard meridian-longitude system. 
152: See \cite{Ro} for example. 
153: %In \cite{CS04}, the rational number corresponding to a slope 
154: %is called a \textit{numerical slope}. 
155: %In the rest of this paper, we always regard 
156: %the slopes as the rational number (or infinity) corresponding to the slope, 
157: %and so we omit the adjective ``numerical". 
158: %	From this point of view, we call a non-meridional slope 
159: %	a \textit{finite slope}, and the meridional slope the \textit{infinite slope}. 
160: 
161: 
162: In this paper, we study numerical properties of the boundary slopes, 
163: regarded as rational numbers, for \textit{Montesinos knots}; 
164: the knots composed by a number of rational tangles. 
165: Precisely, the aim of this paper is: To give 
166: (1) an upper bound on the denominator of a boundary slope and
167: (2) an upper bound on differences between two boundary slopes 
168: for a Montesinos knot exterior. 
169: Our bounds are actually described 
170: in terms of the Euler characteristic and some 
171: other topological quantity of the surfaces. 
172: 
173: 
174: For Montesinos knot exteriors, 
175: Hatcher and Oertel studied 
176: the boundary slopes in \cite{HO} intensively. 
177: They gave an algorithm, 
178: based on the arguments developed originally in \cite{HT}, 
179: to list all essential surfaces up for a given Montesinos knot exterior. 
180: Their algorithm has somehow combinatorial workings, 
181: and in fact, was implemented to a computer program 
182: by Dunfield described in \cite{Dun}. 
183: By using this program, we had performed computer-aided experiments, 
184: and got some observations which suggest the existence of 
185: such numerical properties for the boundary slopes. 
186: This is the motivation of our study. 
187: 
188: In the following, 
189: let $K$ be a Montesinos knot $K(K_1, K_2, \cdots, K_\NumTangles)$,
190: where $\NumTangles\ge 3$ is the number of tangles and 
191: each $K_i$ is a non-integral rational number.
192: 
193: 
194: 
195: 
196: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
197: %
198: % BOUND ON DENOMINATORS
199: %
200: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
201: 
202: \subsection{Bound on denominator}
203: 
204: We first give an upper bound on the denominators 
205: of boundary slopes for Montesinos knot exteriors. 
206: 
207: \begin{theorem}
208: \label{Thm:Denom:UpperBound:Main}
209: Let $\chi$ be 
210: the Euler characteristic of the surface corresponding to 
211: a finite boundary slope $\Slope$ 
212: and  $\sharp b$ the number of its boundary components. 
213: %
214: Then, 
215: except for some boundary slopes,
216: the denominator $\Denom$ of $\Slope$ is bounded as 
217: \begin{eqnarray}
218: \Denom \le \frac{-\chi}{\sharp b}.
219: \label{Eq:Denom:UpperBound:Main-A}
220: \end{eqnarray}
221: 
222: The exceptions occur from $(-2,3,t)$-pretzel knots for odd $t\ge 3$ or their mirror images.
223: Some boundary slopes for the knot only satisfies a weaker bound
224: \begin{eqnarray}
225: \frac{-\chi}{\sharp b} < \Denom \le \frac{-\chi}{\sharp b} + 1,
226: \label{Eq:Denom:UpperBound:Main-B}
227: \end{eqnarray}
228: though a stronger condition $\sharp b \ge 2$ 
229: on the number of boundary components holds in these cases. 
230: \end{theorem}
231: %
232: Here, we remark that, as well as $K(-1/2,1/3,1/t)$, 
233: for example, Montesinos knots $K((-1/2)+k,(1/3)+l,(1/t)-k-l)$ for $k,l\in \mathbb{Z}$
234: are also isotopic to the pretzel knot.
235: 
236: From Theorem \ref{Thm:Denom:UpperBound:Main}, 
237: we have the following corollary immediately. 
238: 
239: \begin{corollary}
240: \label{Cor:Denom:UpperBound:Main:ByGenus}
241: Under the same assumption as in Theorem $\ref{Thm:Denom:UpperBound:Main}$, 
242: and if the surface considered is orientable of genus $g$, 
243: then the denominator of the boundary slope is bounded as
244: $\Denom = 1$ if $g = 0 $, 
245: $\Denom \le 2$ if $g = 1 $, and 
246: $\Denom \le 2g-1$ if $g \ge 2 $. 
247: Furthermore, there are no non-torus Montesinos knots 
248: whose exterior contains essential planar surfaces. 
249: Thus non-torus Montesinos knots admit no reducible surgery. 
250: \end{corollary}
251: 
252: The last statement assures that the well-known Cabling Conjecture 
253: is true for Montesinos knots directly. 
254: This fact has already been achieved in \cite{EM92} 
255: as a corollary of the result for strongly invertible knots. 
256: 
257: 
258: %%% \begin{proof}
259: %%% \end{proof}
260: 
261: %%% \begin{corollary}
262: %%% Under the same assumption as in Theorem $\ref{Thm:Denom:UpperBound:Main}$, 
263: %%% and if the surface considered is a once-punctured Klein bottle, 
264: %%% then the denominator of the boundary slope is one, 
265: %%% that is, the slope is an integral slope. 
266: %%% \qed
267: %%% \end{corollary}
268: 
269: The next corollary is the non-orientable version of the above. 
270: Recall that a non-orientable surface is called 
271: of \textit{non-orientable genus} $h$ 
272: if it contains mutually disjoint $h$ Mobius bands. 
273: 
274: \begin{corollary}
275: \label{Cor:Denom:UpperBound:Main:ByNonOrientableGenus}
276: Under the same assumption as in Theorem $\ref{Thm:Denom:UpperBound:Main}$,
277: and if the surface considered is a non-orientable surface
278: of non-orientable genus $h$,
279: then for the denominator $\Denom$ of the boundary slope, we have,
280: \begin{eqnarray}
281: \Denom \le \frac{h}{2} +1.
282: \end{eqnarray}
283: Moreover if $\sharp b = 1$,
284: \begin{eqnarray}
285: \Denom \le h - 1
286: \end{eqnarray}
287: holds.
288: \end{corollary}
289: 
290: 
291: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
292: %
293: % BOUND ON DIFFERENCES
294: %
295: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
296: 
297: \subsection{Bound on difference}
298: 
299: We next give an upper bound on the ``difference" 
300: between two boundary slopes for Montesinos knot exteriors 
301: by a linear function of the ratio $-\chi/\sharp s$ of 
302: the negative of  the Euler characteristic of the surface 
303: and the number of sheets. 
304: 
305: The \textit{number of sheets} is the number of pieces of the surface 
306: in a small neighborhood of a point on a knot. 
307: If small meridian circles of the knot meet the surface in $m$ points, 
308: then the number of sheets is $m$.
309: 
310: 
311: \begin{theorem}
312: \label{Thm:Diff:UpperBound:Main}
313: Let $\chi_i$ be 
314: the Euler characteristic of the surface 
315: corresponding to a finite boundary slope $\Slope_i$ 
316: and $\sharp s_i$ its number of sheets, for $i=1,2$ respectively. 
317: %
318: Then
319: the difference $|\Slope_1-\Slope_2|$ between the boundary slopes $\Slope_1$ and $\Slope_2$
320: is bounded as
321: \begin{eqnarray}
322: |\Slope_1-\Slope_2|\le 2\,(\frac{-\chi_1}{\sharp s_1}+\frac{-\chi_2}{\sharp s_2})+4
323: .
324: \label{Eq:Diff:UpperBound:Main}
325: \end{eqnarray}
326: \end{theorem}
327: 
328: This inequality (\ref{Eq:Diff:UpperBound:Main}) can be rewritten as 
329: \begin{eqnarray}
330: \Delta(\Slope_1,\Slope_2)\le 2\,(\Denom_2 \frac{-\chi_1}{\sharp b_1}+\Denom_1 \frac{-\chi_2}{\sharp b_2})+4 \Denom_1 \Denom_2
331: ,
332: \label{Eq:Dist:UpperBound:Main-C}
333: \end{eqnarray}
334: which may be preferable for understanding the meaning 
335: from the geometric viewpoint. 
336: Here $\Delta(\Slope_1,\Slope_2)$ denotes the \textit{distance} 
337: between the slopes $\Slope_1$ and $\Slope_2$, 
338: which is defined to be the minimal geometric intersection number 
339: of the simple closed curves representing $\Slope_1$ and $\Slope_2$. 
340: Recall that if $\Slope_i$ is expressed by 
341: an irreducible fraction $\Numer_i/\Denom_i$ for $i=1,2$, 
342: then $\Delta(\Slope_1,\Slope_2)$ is equal to $| \Numer_1 \Denom_2 - \Numer_2 \Denom_1 |$. 
343: %
344: However, in the algorithm of Hatcher and Oertel, 
345: $\Slope_i=\Numer_i/\Denom_i$ and $\sharp s_i$ 
346: play significant roles rather than $\Numer_i$, $\Denom_i$ and $\sharp b_i$.
347: %Besides, it seems difficult to find an upper or lower bound of $\Numer_i$ or $\Denom_i$ by $\Slope_i$. 
348: Hence, in the light of the algorithm, it seems natural to consider 
349: the difference $|\Slope_1-\Slope_2|$ and $-\chi_i/\sharp s_i$. 
350: %
351: Note that, in particular case that both $\Slope_1$ and $\Slope_2$ are integers, 
352: $|\Slope_1-\Slope_2|$ coincides with $\Delta(\Slope_1,\Slope_2)$, 
353: and we have an upper bound of the distance simultaneously.
354: 
355: 
356: 
357: From Theorem $\ref{Thm:Diff:UpperBound:Main}$, 
358: we have three corollaries as follows. 
359: 
360: When the surface are both orientable, we immediately have the following. 
361: 
362: \begin{corollary}
363: \label{Cor:Diff:UpperboundByGenus}
364: Under the same assumption as in Theorem $\ref{Thm:Diff:UpperBound:Main}$,
365: and if the surfaces considered are both orientable surfaces
366: of genera $g_1$ and $g_2$ respectively, then,
367: the difference $|\Slope_1-\Slope_2|$ between the boundary slopes $\Slope_1$ and $\Slope_2$
368: is bounded as
369: \begin{eqnarray}
370: |\Slope_1-\Slope_2| \le 4 \,( g_1 + g_2 ) .
371: \label{Eq:Diff:UpperBound:Main-B:ByGenus}
372: \end{eqnarray}
373: \end{corollary}
374: 
375: With respect to a linear bound on the difference, 
376: or a somehow irregular quadratic bound on the distance by Euler characteristics,
377: the following corollary is easily obtained from Theorem \ref{Thm:Diff:UpperBound:Main}.
378: Though the bounds may not be sharp for Montesinos knots with $\NumTangles\ge 3$,
379: the equality holds for boundary slopes of the trefoil knot.
380: 
381: %%% Corollary
382: 
383: \begin{corollary}
384: \label{Cor:DiffDist:UpperBound}
385: For two boundary slopes and their corresponding essential surfaces,
386: we have the inequality
387: \begin{eqnarray}
388: |\Slope_1-\Slope_2|\le 6\,(\frac{-\chi_1}{\sharp s_1}+\frac{-\chi_2}{\sharp s_2}).
389: \label{Eq:Diff:UpperBound:Linear}
390: \end{eqnarray}
391: This is equivalent to the inequality
392: \begin{eqnarray}
393: \Delta(\Slope_1,\Slope_2)\le 6\,(\Denom_2 \frac{-\chi_1}{\sharp b_1}+\Denom_1 \frac{-\chi_2}{\sharp b_2}).
394: \label{Eq:Dist:UpperBound:Semi-Linear}
395: \end{eqnarray}
396: \end{corollary}
397: 
398: 
399: Regarding the upper bound of the distance or difference by the product of Euler characteristics, we have the following.
400: Though the bounds may not be sharp for Montesinos knots with $\NumTangles\ge3$ tangles,
401: the equality holds for boundary slopes of the figure eight knot.
402: 
403: %%% Corollary
404: 
405: \begin{corollary}
406: \label{Cor:Main3b}
407: If both of the Euler characteristics are negative, then we have
408: \begin{eqnarray}
409: \Delta(\Slope_1,\Slope_2)\le 8\cdot \frac{-\chi_1}{\sharp b_1}\cdot \frac{-\chi_2}{\sharp b_2}.
410: \label{Eq:Dist:UpperBound:Quadratic}
411: \end{eqnarray}
412: This is equivalent to the inequality
413: \begin{eqnarray}
414: |\Slope_1-\Slope_2|\le 8\cdot \frac{-\chi_1}{\sharp s_1}\cdot \frac{-\chi_2}{\sharp s_2}.
415: \label{Eq:Diff:UpperBound:Quadratic}
416: \end{eqnarray}
417: \end{corollary}
418: 
419: 
420: 
421: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
422: %
423: % ORGANIZATION
424: %
425: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
426: 
427: This paper is organized as follows. 
428: We review the algorithm in \cite{HO} in Section 2 and 
429: prepare some formulae in Section 3. 
430: Then, Section 4 and 5 are devoted to giving proofs of 
431: Theorem \ref{Thm:Denom:UpperBound:Main} and 
432: \ref{Thm:Diff:UpperBound:Main} respectively.
433: In the last section, 
434: a brief review on related known results is given, 
435: and some open problems are stated. 
436: 
437: 
438: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
439: %
440: % ACKNOWLEDGMENTS
441: %
442: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
443: 
444: \section*{Acknowledgments}
445: The authors would like to thank Professor Sadayoshi Kojima 
446: for his helpful suggestions about earlier drafts. 
447: They also thank to Professor Masakazu Teragaito for 
448: letting them know the related paper \cite{Te}. 
449: 
450: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
451: %
452: % Section : Algorithm of Hatcher and Oertel
453: %
454: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
455: 
456: \section{Algorithm of Hatcher and Oertel}
457: 
458: The proofs of both Theorem \ref{Thm:Denom:UpperBound:Main} and \ref{Thm:Diff:UpperBound:Main} deeply depend on the algorithm in \cite{HO}.
459: Hence, in this section,
460: we review its workings of enumerating all boundary slopes.
461: 
462: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
463: % Paragraph : Montesinos knot
464: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
465: 
466: \subsection*{Montesinos knot}
467: 
468: As mentioned in the introduction,
469: we assume that the knot $K$ is a Montesinos knot $K(K_1,K_2,\ldots,K_\NumTangles)$, where each $K_i$ is a non-integral fraction and $\NumTangles\ge 3$ in this article.
470: By the assumption, 
471: we normalize Montesinos knots and
472: eliminate two-bridge knots from the argument.
473: This is because boundary slopes are enumerated for two-bridge knots in \cite{HT},
474: and two-bridge knot case is excluded in \cite{HO}.
475: Results for two-bridge knots similar to our main results are obtained by \cite{HT}.
476: 
477: Since a knot in this article is basically a Montesinos knot,
478: we use the term ``tangle'' as a rational tangle if not mentioned otherwise particularly.
479: 
480: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
481: % Paragraph : Decomposition
482: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
483: 
484: \subsection*{Decomposition}
485: 
486: First, we regard $S^3$ including a Montesinos knot $K$ as the union of a $\NumTangles$-tuple of 3-balls $B_i~(i=1, 2, \ldots, \NumTangles)$ in $S^3$
487: with following properties.
488: The interiors of all $B_i$'s are disjoint.
489: The intersection of all boundaries $\partial B_i$'s is a circle, which is called the {\em axis} of the knot $K$.
490: Each $\partial B_i$ is divided into two hemispheres by the axis,
491: and the right hemisphere of $\partial B_i$ coincides with the left hemisphere of $\partial B_{i+1}$ (indices are taken modulo $\NumTangles$).
492: Each ball $B_i$ includes the rational tangle $K_i$ of the Montesinos knot $K$.
493: 
494: By this decomposition,
495: a properly embedded essential surface $F$ is also decomposed into a $\NumTangles$-tuple of subsurfaces $S_i$ in $B_i$.
496: The boundary of $S_i$ is the union of the tangle $K_i$ in the interior of $B_i$ and a curve system on the four-punctured sphere $\partial B_i\setminus K_i$,
497: where a {\em curve system} means the union of disjoint circles and arcs connecting distinct punctures.
498: A simple example of a curve system is a {\em $p/q$-tangle} drawn on a four-punctured sphere. 
499: It is denoted by $\anglebb{p/q}$.
500: Note that a tangle usually means two strings in a 3-ball with their four ends fixed on the boundary sphere,
501: in some cases,
502: we use ``tangle'' as a rational tangle projected to, or drawn on the boundary disjointly.
503: Another example of a curve system is a {\em $p/q$-circle},
504: which is a non-trivial circle disjoint from $p/q$-tangle in a sphere,
505: and is denoted by $\circlebb{p/q}$.
506: 
507: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
508: % Paragraph : Subsurfaces
509: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
510: 
511: \subsection*{Subsurfaces}
512: 
513: In the argument,
514: subsurfaces $S_i$'s are arranged to sit in a standard position by isotopy,
515: and are restricted to be saddle subsurfaces or cap subsurfaces
516: as shown in \cite{HO}.
517: 
518: The simplest example of a subsurface $S_i$ is the direct product set of the curve system $\anglebb{p/q}$ in a four-punctured level sphere with an interval.
519: %%% The boundary of the subsurface consists of two copies of the curve system $\anglebb{p/q}$ and four strands obtained as the direct product of four punctures with an interval.
520: %%% One of the copies and four strands form a $p/q$-tangle in $B_i$,
521: %%% which is a part of the boundary of the surface $F$.
522: %%% The other copy is the curve system $\anglebb{p/q}$ on $\partial B_i$.
523: Also this subsurface is denoted by $\anglebb{p/q}$.
524: This subsurface $S_i$ is topologically the union of two disks,
525: and is called {\em base disks}
526: since every subsurface can be regarded to include these kind of disks.
527: %%% The base disk $S_i$ can be regarded as both a saddle subsurface and a cap subsurface.
528: 
529: An example of a saddle subsurface $S_i$ is constructed by connecting two surfaces by a saddle,
530: where both of the two surfaces are ``base disks'' $\anglebb{p/q}$ and $\anglebb{r/s}$ described above
531: for $p/q$ and $r/s$ satisfying $|ps-qr|=1$.
532: See Figure \ref{Fig:Subsurfaces}(a).
533: A saddle is a disk on a level sphere
534: bounded by a simple closed curve 
535: made of $p/q$-tangle, $r/s$-tangle and four punctures.
536: Though there are two choices of disks bounded by the simple closed curve,
537: the choice does not matter in our later argument.
538: Note that $|ps-qr|=1$ ensures that $p/q$-tangle and $r/s$-tangle are disjoint in a level sphere.
539: %%% $r/s$-tangle as a curve system on a level sphere varies into $p/q$-tangle through the subsurface $S_i$.
540: %%%So, 
541: This subsurface is denoted by $\anglebb{p/q}$\,--\,$\anglebb{r/s}$.
542: 
543: We can construct the disjoint union of 
544: $k$ parallel copies of a subsurface $\anglebb{r/s}$
545: and $l$ parallel copies of a saddle subsurface $\anglebb{p/q}$\,--\,$\anglebb{r/s}$.
546: %%% $r/s$-tangle on a level sphere
547: %%% varies into the union of $k$ parallel copies of $p/q$-tangle and $l$ parallel copies of $r/s$-tangle through the subsurface $S_i$.
548: This subsurface $S_i$ is denoted by $(k\anglebb{p/q}+l\anglebb{r/s})$\,--\,$(k+l)\anglebb{r/s}$.
549: 
550: For a sequence
551: $(k\anglebb{p_{j}/q_{j}}+l\anglebb{p_{j-1}/q_{j-1}})$\,--\,$(k+l)\anglebb{p_{j-1}/q_{j-1}}$\,--\,$\ldots$\,--\,$(k+l)\anglebb{p_{1}/q_{1}}$,
552: we can construct a subsurface by preparing components corresponding to each pair of successive two curve systems in the sequence and gluing them together according to the sequence.
553: %%%% The subsurface is denoted by the sequence of curve systems, which is called a non-projectified edgepath.
554: %%% Note that 
555: The rightmost curve system $p_{1}/q_{1}$ is required to coincide with $K_i$ so that the boundary of $S_i$ includes the rational tangle $K_i$.
556: We regard $\anglebb{p_{1}/q_{1}}$ as the starting point of the sequence.
557: Furthermore, we describe curve systems from right to left in a sequence of curve systems as above.
558: %%% since the convention fits with the $uv$-coordinate system mentioned later.
559: 
560: %%%% On the gluing, the number of sheets around four strands must coincide for all components.
561: 
562: A cap subsurface is constructed as follows.
563: We prepare a curve system consisting of $k$ parallel copies of $p/q$-tangle and $l$ parallel copies of $p/q$-circles.
564: Next, we take a direct product of the curve system with the interval,
565: and arrange the product to lie inside the ball $B_i$
566: so that one of the two boundary level spheres is placed at $\partial B_i$.
567: Then, $p/q$-circles of the inner boundary of the product are capped by disks.
568: See Figure \ref{Fig:Subsurfaces}(b).
569: %%%%% we glue disks as caps to the direct product at the $p/q$-circles,
570: %%%%% where the caps lie inside the direct product $S^2 \times I$
571: %%%%% and $S_i$ is still disjoint.
572: Both this curve system and the cap subsurface are denoted by $k\anglebb{p/q}+l\circlebb{p/q}$.
573: 
574: For any type of subsurface, the leftmost curve system of a sequence represents the curve system $S_i \cap \partial B_i$.
575: 
576: Here, we note that some surfaces may correspond to the same representation by $\anglebb{}$ and $\circlebb{}$,
577: because of the two choices of saddles described above.
578: Though, 
579: the ambiguity does not cause trouble,
580: and we regard a representation as if it corresponded to a surface. 
581: 
582: \begin{figure}[hbt]
583:  \begin{picture}(330,130)
584:  \put(40,20){\scalebox{1.0}{\includegraphics{saddle1.mod.eps}}} % 75x101
585:  \put(0,0){(a) A saddle subsurface $\anglebb{\infty}$--$\anglebb{0}$}
586:  \put(220,20){\scalebox{0.8}{\includegraphics{cap.eps}}} % 83x122
587:  \put(180,0){(b) A cap subsurface $\anglebb{0}+\circlebb{0}$}
588:  \end{picture}
589:  \caption{Examples of subsurfaces}
590:  \label{Fig:Subsurfaces}
591: \end{figure}
592: 
593: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
594: % Paragraph : abc-coordinate
595: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
596: 
597: \subsection*{$abc$-coordinates}
598: 
599: As illustrated in Figure \ref{Fig:CurveSystem},
600: we normalize a curve system to a standard form.
601: The curve system in the standard form is represented by integers $a$, $b$ and $c$.
602: $a$, $b$ and $c$ denote the number of subarcs of the curve system lying in a particular region as in the figure.
603: %%% For example, the number $a$ corresponds to subarcs around the puncture and the number $c$ around the axis of the knot.
604: %%% $a$ and $b$ are non-negative integers, but $c$ can be negative.
605: The number of subarcs around the axis, denoted by $c$, can be negative. 
606: If $c$ is negative, $(a,b,c)$ represents a curve system which is the mirror image of the curve system represented by $(a,b,|c|)$ taken with respect to the axis.
607: %
608: Thus, $(a,b,c)$ represents a curve system and these are called {\em $abc$-coordinates} of a curve system.
609: %%% Furthermore, a subsurface is denoted by a sequence of points with $abc$-coordinates.
610: 
611: For example, the coordinates of 
612: a $p/q$-tangle and a $p/q$-circle are $(1,q-1,p)$ and $(0,q,p)$ respectively.
613: The coordinates of the disjoint union of two curve systems
614: are calculated as the vector sum of the coordinates of both curve systems.
615: %%% Of course, the coordinates of $k$ parallel copies of a curve system
616: %%% are the $k$-multiple of the coordinates of the original curve system.
617: 
618:   \begin{figure}[htb]
619:    \begin{center}
620:     \begin{picture}(90,105)
621:      \put(0,0){\scalebox{0.5}{\includegraphics{curvesystem.eps}}} %181x205
622:      %curvesystem.eps 181x206 PseudoClass 256c 35kb PS 0.0u 0:01
623:      \put(18,48){$a$}
624:      \put(50,48){$a$}
625:      \put(18,85){$a$}
626:      \put(50,85){$a$}
627:      \put(66,25){$b$}
628:      \put(91,30){$c$}
629:     \end{picture}
630:    \end{center}
631:    \caption{A curve system with $(a,b,c)=(1,3,2)$}
632:    \label{Fig:CurveSystem}
633:   \end{figure}
634: 
635: 
636: Though there is no explicit description about $\anglebb{\infty}$ in \cite{HO},
637: we only have to introduce the $d$-coordinate for the number of $\infty$-tangles.
638: 
639: With the $abc$-coordinates,
640: we can explicitly describe conditions for subsurfaces $S_i$'s to be glued consistently.
641: For the $abc$-coordinates of the last curve systems for all subsurfaces,
642: one of the conditions is that $a$-coordinates are the same for all $S_i$'s and so are $b$-coordinates.
643: The other condition is that $c$-coordinates for all $S_i$'s sum up to exactly $0$.
644: 
645: %%% Note that for all curve systems in sequences for all subsurfaces,
646: %%% $a$ coordinates coincide each other.
647: %%% The common value of the $a$-coordinate means $\sharp s$, that is, the number of sheets.
648: 
649: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
650: % Paragraph : uv-coordinates
651: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
652: 
653: \subsection*{$uv$-coordinates}
654: 
655: By projectifying the $abc$-coordinates to the {\em $uv$-coordinates}
656: by $u=b/(a+b)$ and $v=c/(a+b)$,
657: we can make the subsequent argument simpler.
658: An important fact is that a curve system consisting of parallel copies of a curve system of arbitrary multiplicity is mapped to the same point in the $uv$-plane.
659: 
660: A curve system $\anglebb{p/q}$ is projected to $(u,v)=((q-1)/q, p/q))$,
661: which is denoted by $\angleb{p/q}$.
662: %
663: %%% For a non-negative $k$,
664: %%% a curve system consisting of $k$ parallel copies of $p/q$-tangle has $abc$-coordinates $(k,k(q-1),kp)$
665: %%% and is projected to $(u,v)=( (q-1)/q, p/q) )$.
666: %%% Thus, curve systems consisting of parallel copies of a common tangle are mapped to the same point on $uv$-plane.
667: %%% This point in the $uv$-plane is denoted by $\angleb{p/q}$.
668: %
669: A curve system $(k\anglebb{p/q}+l\anglebb{r/s})$ has coordinates
670: $(a,b,c)=(k+l,k(q-1)+l(s-1),kp+lr)$,
671: and thus is projected to the point $(u,v)=(kq/(kq+ls))((q-1)/q,p/q)+(ls/(kq+ls))((s-1)/s,r/s)$.
672: Thus, for fixed $p$, $q$, $r$ and $s$, the $uv$-coordinates depend on only the ratio of $k$ to $l$.
673: After projectification,
674: the curve system is represented as a point on the segment connecting $\angleb{p/q}$ and $\angleb{r/s}$.
675: This point is denoted by
676: $((k/(k+l))\angleb{p/q}+(l/(k+l))\angleb{r/s})$.
677: Note that we use the ratio $k/(k+l)$ instead of $kq/(kq+ls)$,
678: since it is suitable in later calculation.
679: 
680: A curve system $\circlebb{p/q}$ is projected to $(u,v)=(1,p/q)$ on a vertical line $u=1$, and is denoted by $\circleb{p/q}$.
681: %%% A curve system $\circlebb{p/q}$
682: %%% has coordinates $(a,b,c)=(0,kq,kp)$,
683: %%% and thus is projected to $(u,v)=(1,p/q)$.
684: %%% Thus, the curve system is represented as a point
685: %%% on a vertical line $u=1$.
686: %%% Note that, the coordinates do not depend on $k$.
687: %%% This point is denoted by $\circleb{p/q}$.
688: %
689: A curve system $(k\anglebb{p/q}+l\circlebb{p/q})$
690: has coordinates $(a,b,c)=(k,k(q-1)+lq,kp+lp)$,
691: and thus is projected to $(u,v)=(k/(k+l))\cdot((q-1)/q,p/q)+(l/(k+l))\cdot(1,p/q)$
692: on the horizontal segment connecting $\angleb{p/q}$ and $\circleb{p/q}$.
693: %%% Thus, the curve system is represented as a point 
694: %%% on a horizontal segment between $\angleb{p/q}$ and $\circleb{p/q}$.
695: Note that, for fixed $p$ and $q$,
696: the $uv$-coordinates depend on only the ratio of $k$ to $l$.
697: This point is denoted by $((k/(k+l))\angleb{p/q}+(l/(k+l))\circleb{p/q})$.
698: 
699: After projectification,
700: from a sequence of points in the $abc$-space representing a subsurface $S_i$,
701: we obtain a sequence of points on the $uv$-plane for the subsurface.
702: 
703: 
704: The conditions for subsurfaces to be glued together consistently
705: with respect to the $abc$-coordinates
706: are translated into the conditions of the $uv$-coordinates.
707: For the $uv$-coordinates of the last one of the sequence of curve systems for all subsurfaces,
708: $u$-coordinates are the same for all subsurfaces
709: and $v$-coordinates for all subsurfaces sum up to $0$.
710: 
711: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
712: % Paragraph : Diagram and edgepath
713: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
714: 
715: \subsection*{The diagram and edgepaths}
716: 
717: A subsurface $S_i$ is formally related to 
718: a piecewise linear path in the $uv$-plane.
719: %%% though information other than vertices is not necessary in fact.
720: Such a path is called an {\em edgepath}, and
721: we will often use $\Edgepath$ as the symbol for it.
722: Edgepaths lie on a ``diagram'' described as follows.
723: 
724: The {\em diagram $\Diagram$} is a graph on the $u$-$v$ plane.
725: A vertex is 
726: a point $\angleb{p/q}$, whose coordinates are $(u,v)=((q-1)/q,p/q)$,
727: a point $\circleb{p/q}$, whose coordinates are $(u,v)=(1,p/q)$, where $p/q$ is an irreducible fraction, 
728: or a point $\angleb{\infty}$, whose coordinates are $(u,v)=(-1,0)$.
729: If two vertices $\angleb{p/q}$ and $\angleb{r/s}$ satisfy the condition $|ps-qr|=1$, the two vertices are connected by a segment.
730: This segment is one of the two types of edges of the diagram and is denoted by $\angleb{p/q}$\,--\,$\angleb{r/s}$.
731: We call the segment a {\em non-horizontal edge}.
732: In particular, for an integer $z$, 
733: there are edges $\angleb{z}$\,--\,$\angleb{z+1}$ and $\angleb{\infty}$\,--\,$\angleb{z}$.
734: The former is called a {\em vertical edge} since it is a segment of the vertical line $u=0$. The latter is called an {\em $\infty$-edge}.
735: %%% which are called a {\em vertical edge} and an {\em $\infty$-edge} respectively.
736: %
737: The other type of edge is called a {\em horizontal edge},
738: which connects vertices $\angleb{p/q}=((q-1)/q,p/q)$ and $\circleb{p/q}=(1,p/q)$.
739: This edge is denoted by $\angleb{p/q}$\,--\,$\circleb{p/q}$.
740: Note that though the edge $\angleb{\infty}$\,--\,$\angleb{0}$ is horizontal in the usual sense,
741: we regard the edge as a non-horizontal edge rather than a horizontal edge
742: for ease in our later argument.
743: %%% since we would like to use the term ``horizontal edge'' 
744: %%% to express an edge which includes points corresponding to cap subsurfaces.
745: The region $-1\le u \le 1$ is triangulated by these kinds of edges,
746: though the triangulation is not locally finite.
747: In particular, the part of the diagram lying in the strip $0\le u\le 1$ 
748: is denoted by $\Strip$.
749: 
750:   \begin{figure}[htb]
751:    \begin{center}
752:     \begin{picture}(70,150)
753:      \put(0,0){\scalebox{0.7}{\includegraphics{diagram.eps}}}
754:      \put(49,135){\rotatebox{90}{\scalebox{1.0}{$\cdots$}}}
755:      \put(49,-10){\rotatebox{90}{\scalebox{1.0}{$\cdots$}}}
756:      \put(42,70){\scalebox{1.5}{\vector(1,0){30}}}
757:      \put(42,70){\scalebox{1.5}{\vector(0,1){50}}}
758:      \put(92,68){$u$}
759:      \put(40,148){$v$}
760:      \put(33,73){$O$}
761:     \end{picture}
762:    \end{center}
763:    \caption{The diagram $\Diagram$}
764:    \label{Fig:Diagram}
765:   \end{figure}
766: 
767: The edgepath $\Edgepath_i$ of a cap subsurface $S_i$ is a point on the horizontal segment $\angleb{K_i}$\,--\,$\circleb{K_i}$.
768: The edgepath $\Edgepath_i$ of a saddle subsurface $S_i$ is a piecewise linear path starting from the vertex $\angleb{K_i}$.
769: The endpoint of an edgepath is either of a vertex of the diagram or 
770: a point on an edge of the diagram.
771: Hence, the last edge of an edgepath may be a part of a non-horizontal edge.
772: We call such an edge a {\em partial edge}.
773: In comparison with this, we use the term {\em complete edge} to express the whole of a non-horizontal edge.
774: An edgepath consisting of only one point is called a {\em constant edgepath}.
775: The other type of edgepath is called a {\em non-constant edgepath}.
776: 
777: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
778: % Paragraph : Edgepath systems
779: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
780: 
781: \subsection*{Edgepath systems}
782: 
783: By collecting edgepaths for subsurfaces $S_i$,
784: we can represent the original surface $F$.
785: We call this kind of $\NumTangles$-tuple ($\Edgepath_1$, $\Edgepath_2$, $\ldots$, $\Edgepath_\NumTangles$) an {\em edgepath system}
786: and will often use $\EdgepathSystem$ as the symbol for it.
787: 
788: Conversely,
789: for an appropriate edgepath system,
790: by unprojecting all vertices in its edgepaths in the $uv$-plane
791: to integral points in the $abc$-space with the common $a$-coordinate,
792: we can construct subsurfaces and a surface $F$, though some ambiguity remains.
793: 
794: 
795: %%% A sequence of curve systems 
796: %%% corresponds to a class of subsurfaces which are isotopic except for saddles.
797: %%% Thus, $\NumTangles$-tuple of such sequences corresponds to a class of surfaces.
798: %%% 
799: %%% An edgepath $\Edgepath_i$ corresponds to a projective equivalent class of 
800: %%% such a class of subsurfaces,
801: %%% and then
802: %%% by collecting such edgepaths,
803: %%% we can describe a projective equivalent class of a class of surfaces by a $\NumTangles$-tuple ($\Edgepath_1$, $\Edgepath_2$, $\ldots$, $\Edgepath_\NumTangles$) of edgepaths.
804: %%% We call this kind of a tuple of edgepaths
805: %%% an {\em edgepath system}, and
806: %%% will often use $\EdgepathSystem$ to denote it.
807: %%% 
808: %%% A point with $uv$-coordinates $(u,v)$ 
809: %%% is projected back to 
810: %%% a point with $abc$-coordinates $(k(1-u),ku,kv)$.
811: %%% For an edgepath system,
812: %%% all points in it must be projected back to points with the same $a$-coordinate.
813: %%% If we choose appropriate rational number $k$ and integer $a$,
814: %%% we can obtain sequences of curve systems
815: %%% whose $abc$-coordinates of all curve systems are not only rational but also integral,
816: %%% and hence, a class of surface is obtained.
817: %%% The class of surface and the corresponding edgepath system are identified each other implicitly.
818: 
819: The set of the edgepath systems are divided into three classes 
820: according to the common $u$-coordinate of the endpoints of edgepaths in their edgepath system.
821: An edgepath system and the corresponding surface are said to be {\em type I}, {\em type II} or {\em type III},
822: if all edgepaths in the edgepath system end at $u>0$, $u=0$ or $u<0$ respectively.
823: 
824: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
825: % Paragraph : Candidate surfaces
826: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
827: 
828: \subsection*{Candidate surfaces}
829: 
830: In the enumeration of boundary slopes, 
831: we first list candidates for essential surfaces,
832: and then omit compressible surfaces from the candidates.
833: Precise conditions for an edgepath system $\EdgepathSystem=(\Edgepath_1, \Edgepath_2,\ldots ,\Edgepath_\NumTangles)$ to be an edgepath system of a {\em candidate surface} are given in \cite{HO} as follows.
834: %
835: \begin{itemize}
836: %
837: \item[(E1)] The starting point of $\Edgepath_i$ lies on the edge $\angleb{K_i}$\,--\,$\circleb{K_i}$, and if this starting point is not the vertex $\angleb{K_i}$, then the edgepath $\Edgepath_i$ is constant.
838: %
839: \item[(E2)] $\Edgepath_i$ is minimal, i.e., it never stops and retraces itself, nor does it ever go along two sides of the same triangle of $\Diagram$ in succession.
840: %
841: \item[(E3)] The ending points of the $\Edgepath_i$'s are rational points of $\Diagram$ which all lie on one vertical line and whose vertical coordinates add up to zero.
842: %
843: \item[(E4)] $\Edgepath_i$ proceeds monotonically from right to left, ``monotonically'' in the weak sense that motion along vertical edges is permitted. 
844: %
845: \end{itemize}
846: 
847: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
848: % Paragraph : Basic edgepath systems
849: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
850: 
851: \subsection*{Basic edgepath systems and consistency on gluing}
852: 
853: The enumeration is performed by use of basic edgepath systems.
854: A {\em basic edgepath} is an edgepath which
855: starts from a vertex $\angleb{p/q}$,
856: goes leftwards monotonically,
857: and ends at the time when the edgepath first reaches $u=0$.
858: Moreover, a {\em basic edgepath system}
859: is an edgepath system which consists of basic edgepaths.
860: %%% For a given Montesinos knot $K(K_1, K_2,\ldots, K_\NumTangles)$,
861: %%% there exist finite basic edgepath systems
862: %%% whose edgepath $\Edgepath_i$ starts from $\angleb{K_i}$.
863: We will often use $\BasicEdgepath$ and $\BasicEdgepathSystem$ as the symbols for
864: a basic edgepath and a basic edgepath system.
865: 
866: In order to seek type I edgepaths,
867: it is convenient to introduce an {\em extended basic edgepath} $\widetilde{\BasicEdgepath}$
868: which is obtained 
869: by connecting a horizontal segment $\angleb{p/q}$\,--\,$\circleb{p/q}$
870: to the starting point $\angleb{p/q}$ of a basic edgepath $\BasicEdgepath$.
871: We define an {\em extended basic edgepath system} $\widetilde{\BasicEdgepathSystem}$ similarly.
872: 
873: Sometimes, we regard 
874: an edgepath $\Edgepath$ as a function from an interval in $\mathbb{R}$ to $\mathbb{R}$, 
875: which maps $u$-coordinate to $v$-coordinate,
876: and then allow ourselves to use expressions like $\Edgepath(u)$,
877: although we cannot define its value for $u=0$ if the edgepath includes vertical edges.
878: The function is piecewise linear.
879: Similar notation is used for
880: a basic edgepath and an extended basic edgepath.
881: %
882: With the notation, a condition of consistency on gluing in (E3) can be described as 
883: \begin{eqnarray}
884: \label{Eq:EquationForEdgepathSystem}
885: \sum_{i=1}^{\NumTangles} \Edgepath_i(u)=0.
886: \end{eqnarray}
887: %%% This is a piecewise linear equation.
888: Especially for type I surfaces,
889: we need to solve (\ref{Eq:EquationForEdgepathSystem}) for some extended basic edgepath system
890: to determine the common $u$-coordinate of the endpoints of its edgepath system.
891: 
892: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
893: % Paragraph : Enumeration
894: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
895: 
896: \subsection*{Enumeration}
897: We have finished introducing notions used in the algorithm in \cite{HO}.
898: Now, we review its workings.
899: All boundary slopes are enumerated as follows.
900: 
901: All basic edgepath systems for the Montesinos knot $K$ are enumerated first.
902: Then, type I, type II and type III candidate edgepath systems $\EdgepathSystem$ are obtained for each basic edgepath system $\BasicEdgepathSystem$ .
903: %
904: A type I edgepath system is obtained by solving the equation (\ref{Eq:EquationForEdgepathSystem}) for the extended basic edgepath system $\widetilde{\BasicEdgepathSystem}$.
905: For a solution $u_0$ of the equation, we construct an edgepath system as follows.
906: Let $\widetilde{\BasicEdgepath_i}$ be the $i$-th extended basic edgepath
907: %%%of the extended basic edgepath system 
908: starting from $\angleb{K_i}$ where $K_i=p_i/q_i$.
909: If $(q_i-1)/q_i < u_0$, then the line $u=u_0$ intersects with the horizontal edge of the extended basic edgepath.
910: Therefore, we prepare a constant edgepath with a single point $(u,v)=(u_0,K_i)$.
911: If otherwise, the line $u=u_0$ intersects with the non-horizontal part of the extended basic edgepath.
912: Hence, we cut out an edgepath $\Edgepath_i$ 
913: starting from $\angleb{K_i}$ and ending at $u=u_0$,
914: from the original basic edgepath $\BasicEdgepath_i$.
915: The edgepath system is obtained by collecting all such edgepaths.
916: %
917: A type II edgepath system $\EdgepathSystem$ is obtained by adding vertical edges to the basic edgepaths of the basic edgepath system $\BasicEdgepathSystem$
918: so that $v$ coordinates of endpoints of the edgepath system $\EdgepathSystem$ sum up to $0$.
919: %%%The choice of the vertical edges is essentially unique.
920: %
921: A type III edgepath system $\EdgepathSystem$ is obtained by adding an $\infty$-edge to each basic edgepath $\BasicEdgepath_i$.
922: %%% After verifying the conditions (E1) to (E4) above,
923: Thus, we can enumerate all candidate surfaces.
924: The process is completed in finite time.
925: 
926: After enumerating all the candidate edgepath systems,
927: we verify their incompressibility.
928: Detailed conditions for the edgepath system of a candidate surface
929: to be incompressible are also described in \cite{HO}.
930: By the conditions, we can eliminate compressible surfaces from the set of candidate surfaces, and complete the enumeration of essential surfaces.
931: Though, the conditions are not so crucial in this paper
932: and we hardly make use of the conditions.
933: Besides, the determination of the orientability is omitted in \cite{HO}.
934: It must be performed by oneself if necessary.
935: 
936: %%% From a remaining edgepath system,
937: %%% via sequences of curve systems,
938: %%% we can reconstruct an essential surface.
939: %%% A boundary slope can be calculated from the edgepath system
940: %%% as in the next section.
941: 
942: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
943: %
944: % Section : Preparation
945: %
946: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
947: 
948: \section{Preparation}
949: \label{Sec:Preparation}
950: 
951: In this section,
952: we prepare some formulae for concrete calculation in the subsequent sections.
953: We also introduce an operation named ``simplification''.
954: 
955: %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
956: %%% % Subsection : Formulae
957: %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
958: %%% 
959: %%% \subsection{Formulae}
960: %%% 
961: %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
962: %%% % Paragraph : Notation
963: %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
964: 
965: \subsection*{Notation}
966: 
967: We first give some notation about edgepaths and edgepath systems.
968: For an edgepath $\Edgepath$,
969: symbols $\Edgepath_{>0}$ and $\Edgepath_{\ge 0}$ denote
970: a part of $\Edgepath$ inside the region $u>0$ and $u\ge 0$ respectively.
971: A part of $\Edgepath$ consisting of vertical edges
972: is denoted by $\Edgepath_{=0}$.
973: 
974: For type II and type III edgepath systems
975: let $\Edgepath(+0)$ denote the value of $v$ at the moment when $u$-coordinate reaches $0$.
976: Furthermore, for an edgepath system $\EdgepathSystem=(\Edgepath_1, \Edgepath_2,\ldots ,\Edgepath_\NumTangles)$, let $\EdgepathSystem(+0)$ denote the sum $\sum_{i=1}^{\NumTangles} \Edgepath_i(+0)$.
977: 
978: 
979: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
980: % Paragraph : Sign of an edge
981: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
982: 
983: \subsection*{Signs of edges}
984: A complete non-horizontal edge $\Edge=\angleb{p/q}$\,--\,$\angleb{r/s}$
985: is said to be {\em increasing} or {\em decreasing}
986: if the $v$-coordinate of a point increases or decreases respectively when it goes from $\angleb{r/s}$ to $\angleb{p/q}$ along $\Edge$.
987: %%% This definition may cause a confusion, since it does not match with the standard one.
988: 
989: We define the sign of the edge $\Edge$ to be $+1$ or $-1$
990: according to whether the edge is increasing or decreasing respectively.
991: The sign of the edge $\Edge$ is denoted by $\sigma(\Edge)$
992: and is calculated by $(ps-qr)$.
993: The sign of a partial edge is defined as the sign of the complete edge including the partial edge.
994: 
995: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
996: % Paragraph : Length of an edgepath
997: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
998: 
999: \subsection*{Lengths of edgepaths}
1000: 
1001: Next, we define the length of an edgepath.
1002: The lengths of a single point and a complete edge are defined to be $0$ and $1$ respectively.
1003: The length of a partial edge $\Edge=(k/(k+l) \angleb{p/q}+l/(k+l) \angleb{r/s})$\,--\,$\angleb{r/s}$ is $k/(k+l)$.
1004: Note that the ratio of the Euclidean length on the $uv$-plane of the partial edge 
1005: to that of the complete edge $\angleb{p/q}$\,--\,$\angleb{r/s}$ is $kq/(kq+ls)$ as calculated in the previous section.
1006: Thus, the length of a partial edge does not coincide with the ratio generally.
1007: The length of an edge $\Edge$ is denoted by $|\Edge|$.
1008: The length of an edgepath is the sum of the lengths of the edges in the edgepath.
1009: A constant edgepath is of length $0$.
1010: The length of an edgepath $\Edgepath$ is denoted by $|\Edgepath|$.
1011: %%% This length is suitable for calculating boundary slopes and Euler characteristics.
1012: 
1013: Here, we prepare another formula of the length of a partial edge.
1014: For a partial edge $e$ of a complete edge $\angleb{p/q}$\,--\,$\angleb{r/s}$,
1015: assume that $u$-coordinate of the endpoint of the partial edge is $u_0$.
1016: If a curve system $k\anglebb{p/q}+l\anglebb{r/s}$ corresponds to the endpoint,
1017: we have $abc$-coordinates $(a_0,b_0,c_0)=k(1,q-1,p)+l(1,s-1,r)$ and $u$-coordinate $u_0=b_0/(a_0+b_0)=\{k(q-1)+l(s-1)\}/(kq+ls)$,
1018: and hence,
1019: \begin{eqnarray}
1020: |e|&=& \frac{k}{k+l}=\frac{1+s(u_0-1)}{(s-q)(u_0-1)}
1021: \label{Eq:Formula:LengthOfPartialEdge}
1022: .
1023: \end{eqnarray}
1024: 
1025: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1026: % Paragraph : Twist
1027: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1028: 
1029: \subsection*{Boundary slopes and twists of surfaces}
1030: 
1031: The boundary slope of a surface is calculated via the total number of twists,
1032: which we call {\em twist} for short.
1033: %%% Especially, a bound of twist plays important role in the proof of the theorem \ref{Thm:Diff:UpperBound:Main}.
1034: Roughly, the twist $\Twist(F)$ of a surface $F$ is a variation of the numerical boundary slope, which fits with the algorithm.
1035: With the twist, the boundary slope $\Slope$ is calculated by $\Slope = \Twist(F) - \Twist(F_S)$
1036: where $F_S$ is a Seifert surface in the list of candidate surfaces of the knot $K$.
1037: 
1038: We define the twist of a subsurface first.
1039: Base disks have twist $0$.
1040: For a non-$\infty$-edge $\anglebb{p/q}$\,--\,$\anglebb{r/s}$,
1041: if we draw both tangles in standard position as in Figure \ref{Fig:CurveSystem},
1042: the saddle for the edge surrounds two of the four punctures
1043: (see Figure \ref{Fig:Twist}).
1044: This means that
1045: two of the four boundary arcs of a saddle component of a subsurface revolve once around the strands of the tangle.
1046: Hence, the edge of the edgepath contributes $\pm 2$ to the twist.
1047: The sign of the value is determined by whether the boundary arc revolves in clockwise or counter-clockwise direction,
1048: and the sign coincides with $-\sigma(\Edge)$ for an edge $\Edge$.
1049: For a partial edge $(k\anglebb{p/q}+l\anglebb{r/s})$\,--\,$(k+l)\anglebb{r/s}$,
1050: at two of the four boundary arcs of a component of a subsurface,
1051: $k$ of $(k+l)$ sheets go around the strands once.
1052: Hence, the partial edge contributes $\pm 2 k/(k+l)$ to the twist.
1053: Note that $k/(k+l)$ coincides with the length of the partial edge.
1054: The twist of a subsurface is the sum of the twists of the edges of the edgepath corresponding to the subsurface.
1055: Naturally, the twist of a cap subsurface is $0$, since the corresponding edgepath is a point.
1056: 
1057:   \begin{figure}[htb]
1058:    \begin{center}
1059:     \begin{picture}(70,102)
1060:      \put(0,0){\scalebox{0.6}{\includegraphics{0-1.eps}}} %116x171
1061:      %0-1.eps 116x171 PseudoClass 256c 24kb PS 0.0u 0:01
1062:     \end{picture}
1063:    \end{center}
1064:    \caption{A saddle corresponding to $\angleb{0}$\,--\,$\angleb{1}$}
1065:    \label{Fig:Twist}
1066:   \end{figure}
1067: 
1068: The twist of the surface $F$ is the sum of the twists of its subsurfaces $S_i$.
1069: The precise definition of the twist is
1070: %
1071: \begin{eqnarray}
1072: \Twist(F)
1073: &=&
1074: \sum_{i=1}^{\NumTangles}
1075: ~
1076: \left\{
1077:  \begin{array}{l}
1078:   0 \\
1079:   ~~~~~(\textrm{  if $\Edgepath_i$ is constant }) \\
1080:   \sum_{\Edge_{i,j}\in \Edgepath_i}
1081:   \left\{
1082:    \begin{array}{l}
1083:     0 \\
1084:     ~~~~~(\textrm{  if $\Edge_{i,j}$ is an $\infty$-edge.} ) \\
1085:     -2\,\sigma(\Edge_{i,j})~|\Edge_{i,j}| \\
1086:     ~~~~~(\textrm{  otherwise. } )\\
1087:    \end{array}
1088:   \right. \\
1089:   ~~~~~(\textrm{  if $\Edgepath_i$ is non-constant }) \\  
1090:  \end{array}
1091: \right.
1092: .
1093: \label{Eq:Formula:Twist}
1094: \end{eqnarray}
1095: %
1096: %
1097: Though the twist is originally defined for a surface,
1098: it is well-defined for an edgepath system.
1099: %%% Though there exist two choices of adding a saddle for each non-horizontal edge,
1100: %%% we can find that the twist does not depend on the choice
1101: %%% and that the twist is well-defined for sequences of curve systems.
1102: %%% Furthermore, since a surface and its multiple have the same twist,
1103: %%% the twist is well-defined also for an edgepath system.
1104: 
1105: 
1106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1107: % Paragraph : surfaces with the same boundary slope
1108: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1109: 
1110: \subsection*{Surfaces with the same boundary slope}
1111: 
1112: We think about surfaces with the same boundary slope.
1113: In the proofs of the theorems,
1114: only a surface of minimal $-\chi/\sharp s$ is important
1115: among such surfaces sharing the common boundary slope.
1116: Therefore, we take the surface of minimal $-\chi/\sharp s$ as their representative,
1117: ignore the others and will simplify the subsequent argument,
1118: especially in Section \ref{Sec:ABoundOnTheDifference}.
1119: We call this operation {\em simplification}.
1120: 
1121: By the simplification, we ignore (1) most of type I surfaces corresponding to non-isolated solutions of (\ref{Eq:EquationForEdgepathSystem}), (2) type II surfaces with redundant vertical edges, (3) type III surfaces with partial $\infty$-edges and (4) augmented type III surfaces mentioned in \cite{HO}.
1122: 
1123: 
1124: %%% \begin{itemize}
1125: %%% \item
1126: %%% Non-isolated solutions for type I surfaces
1127: %%% 
1128: %%% Every type I surface corresponds to an edgepath system whose common $u$-coordinate of the endpoints comes from a solution of an equation comes from (E3).
1129: %%% Though solutions of the equation are isolated generally,
1130: %%% the solutions rarely happen to be non-isolated.
1131: %%% %%% In other words,
1132: %%% %%% a part of the sum $\sum \Edgepath_i(u)$ as a function partially coincides with the $u$-axis ($v=0$).
1133: %%% As described in \cite{HO},
1134: %%% non-isolated solutions correspond to the same boundary slope.
1135: %%% 
1136: %%% If a set of non-isolated solutions forms an interval $u_0\le u\le u_1$, 
1137: %%% among the surfaces corresponding to $u_2 \in [u_0,u_1]$,
1138: %%% either of two surfaces corresponding to $u_0$ or $u_1$
1139: %%% has minimum $-\chi/\sharp s$.
1140: %%% 
1141: %%% \item
1142: %%% Type II surfaces
1143: %%% 
1144: %%% For a type II surface,
1145: %%% there are infinitely many choices on adding vertical edges
1146: %%% to a basic edgepath.
1147: %%% But, in any choice, the twist is found to be same
1148: %%% after the addition of vertical edges.
1149: %%% %
1150: %%% %%% Assume that $x_i=\Edgepath_{i}(+0)$. Let $y_i$ denotes the directed length of the vertical edges,
1151: %%% %%% which is positive when the vertical edges go upward and negative when downward.
1152: %%% %%% Then $\EdgepathSystem(+0)=\sum x_i$ and $\sum (x_i+y_i)=0$.
1153: %%% %%% About contribution on Euler characteristic by saddles corresponding to the vertical edgepaths, we have
1154: %%% %%% $\frac{\chi(\textrm{saddle}_{=0})}{\sharp s}=-\sum |y_i|\le -|\sum y_i|=-|-\sum x_i|=-|\EdgepathSystem(+0)|$.
1155: %%% %%% When the equal in the inequality holds,
1156: %%% %%% Euler characteristic is maximum.
1157: %%% %
1158: %%% $-\chi/\sharp s$ is minimum when an edgepath system has no redundant vertical edges.
1159: %%% 
1160: %%% \item
1161: %%% Type III surfaces with partial edges
1162: %%% 
1163: %%% For a type II surface with $\EdgepathSystem(+0)=0$,
1164: %%% type III surfaces are obtained by adding partial edges of $\infty$-edges.
1165: %%% Since saddles corresponding to $\infty$-edges do not contribute to the twist,
1166: %%% the twist is constant for an arbitrary common $u$-coordinate $-1\le u \le 0$ of endpoints.
1167: %%% The initial type II surface has minimum $-\chi/\sharp s$.
1168: %%% 
1169: %%% \item 
1170: %%% Augmented type III surfaces
1171: %%% 
1172: %%% If $\NumTangles\ge 4$, an operation named ``augmentation'' is possible for a type III surface,
1173: %%% but the operation causes no effect on the twist
1174: %%% as described in \cite{HO}.
1175: %%% %
1176: %%% %%% In augmentation process,
1177: %%% %%% some axis parallel annulus,
1178: %%% %%% some saddle from a band to itself,
1179: %%% %%% and gluing two circles of boundary of surface $S_i$'s occur
1180: %%% %%% in comparison with the non-augmented initial type III surface.
1181: %%% %%% But the annulus and gluing by circles do not change Euler characteristic.
1182: %%% %%% Saddles only decrease Euler characteristic.
1183: %%% %%% Hence, 
1184: %%% %
1185: %%% The non-augmented type III surface has minimal $-\chi/\sharp s$.
1186: %%% 
1187: %%% \end{itemize}
1188: 
1189: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1190: % Paragraph : Euler characteristic
1191: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1192: 
1193: \subsection*{Euler characteristics}
1194: 
1195: Instead of the Euler characteristic itself,
1196: formulae for calculating $-\chi/\sharp s$ are given
1197: since they are more suitable.
1198: Note that 
1199: though the Euler characteristic itself is not well-defined for an edgepath system,
1200: so is the quantity $-\chi/\sharp s$.
1201: 
1202: %%% Type III surface %%%
1203: 
1204: To construct a type III surface,
1205: we have base disks for each tangle first,
1206: add saddles according to non-$\infty$-edges in the edgepath system,
1207: add also saddles according to $\infty$-edges,
1208: and then glue $S_i$'s together at arcs which are halves of $\infty$-tangle on $\partial B_i$.
1209: The Euler characteristic of the surface $F$ so obtained is calculated by
1210: $\chi=
1211: \sum_{i=1}^{\NumTangles} (2\cdot \sharp s)
1212: -\sum_{i=1}^{\NumTangles} (|\Edgepath_{i,>0}|\cdot \sharp s)
1213: -\sum_{i=1}^{\NumTangles} (1\cdot \sharp s)
1214: -\NumTangles \cdot \sharp s
1215: $.
1216: Thus,
1217: \begin{eqnarray}
1218: \frac{-\chi}{\sharp s}
1219: &=&
1220:  \sum_{i=1}^{\NumTangles}|\Edgepath_{i,>0} |
1221: \label{Eq:Formula:EulerCharTypeIII}
1222: .
1223: \end{eqnarray}
1224: 
1225: \bigskip
1226: 
1227: %%% Type II surface %%%
1228: 
1229: To construct a type II surface,
1230: we have base disks first, 
1231: add saddles for the basic edgepath,
1232: add also saddles for vertical edges,
1233: and then glue $S_i$'s together at integral tangles on $S_i \cap \partial B_i$.
1234: %
1235: Euler characteristic is
1236: $\chi=
1237: \sum_{i=1}^{\NumTangles} (2\cdot \sharp s)
1238: -\sum_{i=1}^{\NumTangles} (|\Edgepath_{i,>0}|\cdot \sharp s)
1239: -\sum_{i=1}^{\NumTangles} (|\Edgepath_{i,=0}|\cdot \sharp s)
1240: -2(\NumTangles-1) \cdot \sharp s
1241: $.
1242: %
1243: Thus,
1244: %
1245: \begin{eqnarray}
1246: \frac{-\chi}{\sharp s}
1247: %%% &=&
1248: %%% \left(
1249: %%%  \sum_{\Edgepath_i \in \EdgepathSystem}|\Edgepath_{i,>0} |
1250: %%% \right)
1251: %%% +
1252: %%% \left(
1253: %%%  \sum_{\Edgepath_i \in \EdgepathSystem} |\Edgepath_{i,=0}|
1254: %%% \right) -2
1255: %%% \nonumber
1256: %%% \\
1257: &=&
1258:  \sum_{i=1}^{\NumTangles}(|\Edgepath_{i,>0} |)
1259:  +|\EdgepathSystem(+0) |-2
1260: %%%\nonumber
1261: \label{Eq:Formula:EulerCharTypeII}
1262: .
1263: \end{eqnarray}
1264: 
1265: \bigskip
1266: 
1267: %%% Type I surface %%%
1268: 
1269: To construct a type I surface,
1270: we have base disks first,
1271: add caps for constant edgepaths,
1272: add saddles for non-constant edgepaths,
1273: and then glue $S_i$'s at curve systems on $\partial B_i$'s.
1274: 
1275: Assume that $C$ is a component of a subsurface described by $(k\anglebb{p/q}+l\anglebb{r/s})$\,--\,$(k+l)\anglebb{r/s}$.
1276: $k$ saddles are included in the component $C$,
1277: and contribute $-k$ to the Euler characteristic.
1278: Since $\sharp s=k+l$,
1279: the contribution to $-\chi/\sharp s$ by the partial edge $(k/(k+l)\angleb{p/q}+l/(k+l)\angleb{r/s})$\,--\,$\angleb{r/s}$ is $k/(k+l)$,
1280: which coincides with the length of the partial edge $e$ by definition.
1281: 
1282: Next assume that $C$ is a cap subsurface described by $k\anglebb{p/q}+l\circlebb{p/q}$.
1283: $l$ caps are included in the component $C$,
1284: and contribute $+l$ to Euler characteristic.
1285: By $\sharp s=k$,
1286: the contribution to $-\chi/\sharp s$ by the constant edgepath $k/(k+l)\angleb{p/q}+l/(k+l)\circleb{p/q}$ is $-l/k$.
1287: Since the curve system $k\anglebb{p/q}+l\circlebb{p/q}$ has $abc$-coordinates $k(1,q-1,p)+l(0,q,p)$,
1288: we have $u=b/(a+b)=1-k/\{(k+l)q\}$.
1289: Then the contribution is calculated by $1-1/\{q(1-u)\}$.
1290: 
1291: %%%Assume that 
1292: For every subsurface,
1293: the last curve system of the sequence of curve systems corresponding to the subsurface has the common $a$ and $b$ coordinates, say $a_0$ and $b_0$, respectively.
1294: On every hemisphere of $\partial B_i$ divided by the axis,
1295: $(2a_0+b_0)$ subarcs of $\partial S_i$ exist.
1296: $(2a_0+b_0)$ subarcs on each of $2N$ hemisphere are glued each other first, and then $(2a_0+2b_0)$ disks intersecting the axis are connected next.
1297: Thus, the effect of the gluing on the Euler characteristic is $-(2a_0+b_0)\NumTangles+(2a_0+2b_0)$.
1298: Since $\sharp s=a_0$ and $u_0=b_0/(a_0+b_0)$, the ratio of this effect to the number of sheets
1299: is $\{-(2a_0+b_0)\NumTangles+2a_0+2b_0\}/a_0=-1/(1-u_0)\cdot(\NumTangles-2)-\NumTangles$.
1300: 
1301:   \begin{figure}[htb]
1302:    \begin{center}
1303:     \begin{picture}(242,103)
1304:      \put(0,0){\scalebox{0.5}{\includegraphics{glue.eps}}} %484x206
1305:      %glue.eps 484x206 PseudoClass 256c 22kb PS 0.0u 0:01
1306:      \put(18,32){$a_0$}
1307:      \put(33,32){$a_0$}
1308:      \put(18,53){$a_0$}
1309:      \put(31,53){$a_0$}
1310:      \put(7,28){$b_0$}
1311:      \put(133,32){\scalebox{3.0}{$\cdot\cdot\cdot$}}
1312:     \end{picture}
1313:    \end{center}
1314:    \caption{Gluing $\NumTangles$ subsurfaces at subarcs on $\partial B_i$'s}
1315:    \label{Fig:Gluing}
1316:   \end{figure}
1317: 
1318: Hence,
1319: %
1320: \begin{eqnarray*}
1321: \chi&=&
1322: \sum_{i=1}^{\NumTangles} (2\cdot \sharp s)
1323: +\sum_{\Edgepath_{i}\in\ConstantEdgepaths} 
1324:  \left(
1325:      (\frac{1}{q_i(1-u)}-1) \cdot \sharp s 
1326:  \right)
1327: \\
1328: &&
1329: -\sum_{\Edgepath_{i}\in\NonconstantEdgepaths} 
1330:  \left( |\Edgepath_{i} | \cdot \sharp s \right)
1331: \\
1332: &&
1333: +\{-\frac{1}{1-u}(\NumTangles-2)-\NumTangles\} \cdot\sharp s
1334: ,
1335: \end{eqnarray*}
1336: %
1337: %
1338: \begin{eqnarray}
1339: \frac{-\chi}{\sharp s}
1340: &=&
1341:  \sum_{i=1}^{\NumTangles}
1342:   \left(
1343:    \left\{
1344:     \begin{array}{l}
1345:      0 \\
1346:      ~~~~~(\textrm{  if $\Edgepath_i$ is constant }) \\
1347:      |\Edgepath_{i} |\\
1348:      ~~~~~(\textrm{ otherwise }) \\
1349:     \end{array}
1350:    \right.
1351:   \right)
1352: %%% \nonumber
1353: \label{Eq:Formula:EulerCharTypeI}
1354: \\
1355: && 
1356: +\NumTangles_{\mathrm{const}}-\NumTangles
1357: +\left(
1358: \NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths}\frac{1}{q_i}
1359: \right)
1360: \frac{1}{1-u}
1361: ,
1362: \nonumber
1363: \end{eqnarray}
1364: %
1365: where the edgepath system $\EdgepathSystem$ is divided into
1366: the set $\NonconstantEdgepaths$ of non-constant edgepaths
1367: and the set $\ConstantEdgepaths$ of constant edgepaths,
1368: and $\NumTangles_{\mathrm{const}}$ denotes the number of the constant edgepaths.
1369: 
1370: %%% In the process of proving Theorem \ref{Thm:Denom:UpperBound:Main},
1371: %%% the following form is more suitable than the above.
1372: %%% %
1373: %%% \begin{eqnarray}
1374: %%% \frac{-\chi}{\sharp s}
1375: %%% &=&-\NumTangles+(\NumTangles-2)\frac{1}{1-u} \nonumber \\
1376: %%% &&+
1377: %%%  \sum_{i=1}^{\NumTangles}
1378: %%%   \left(
1379: %%%    \left\{
1380: %%%     \begin{array}{l}
1381: %%% %%%     (1-\frac{1}{q_i})\frac{1}{1-u} \\
1382: %%%      1-\frac{1}{q_i}\frac{1}{1-u} \\
1383: %%%      ~~~~~(\textrm{  if $\Edgepath_i$ is constant }) \\
1384: %%%      |\Edgepath_{i} |\\
1385: %%%      ~~~~~(\textrm{ otherwise }) \\
1386: %%%     \end{array}
1387: %%%    \right.
1388: %%%   \right)
1389: %%% .
1390: %%% \label{Eq:Formula:EulerCharTypeI_AnotherForm}
1391: %%% \end{eqnarray}
1392: 
1393: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1394: % Paragraph : Number of sheets
1395: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1396: 
1397: \subsection*{Number of sheets}
1398: 
1399: When we construct a surface from an edgepath system,
1400: the number of sheets of a surface denoted by $\sharp s$ is determined as follows.
1401: 
1402: Assume first that the last edge of an edgepath of an edgepath system is a partial edge of length $k/m$ where the fraction $k/m$ is irreducible.
1403: Since the number of saddles $k/m \cdot \sharp s$ must be an integer,
1404: $\sharp s$ is a multiple of $m$.
1405: %
1406: Assume next that an edgepath of an edgepath system is a constant edgepath 
1407: $k/m \angleb{p/q}$\,--\,$(1-k/m) \circleb{p/q}$.
1408: Since the number of caps $(m-k)/k \cdot \sharp s$ must be an integer,
1409: $\sharp s$ is a multiple of $k$.
1410: %
1411: Thus, $\sharp s$ is determined as the least common multiple of these integers.
1412: 
1413: %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1414: %%% % Subsection : Remarks for proofs
1415: %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1416: %%% 
1417: %%% \subsection{Remarks for proofs}
1418: 
1419: \subsection*{A remark for proofs}
1420: 
1421: Here, we give an elementary fact for the subsequent sections.
1422: %
1423: \begin{remark}\label{Rem:Condition:ForLinkToBeKnot}
1424: Let $L$ be a Montesinos link $L(K_1,K_2,\ldots,K_\NumTangles)$
1425: where $\NumTangles\ge 3$ is the number of tangles and 
1426: each $K_i$ is a non-integral rational number.
1427: For the link $L$ to be a knot,
1428: fractions $K_1,K_2,\ldots,K_\NumTangles$ must satisfy either of:
1429: \begin{itemize}
1430: \item
1431: Exactly one of the fractions has even denominator.
1432: \item
1433: All denominators are odd and the number of odd numerators is odd.
1434: \end{itemize}
1435: Moreover, $\sum K_i \neq 0$ holds under the condition.
1436: This means that a type I edgepath system with all its edgepaths being constant 
1437: does not exist for a Montesinos knot.
1438: \end{remark}
1439: 
1440: %%% Besides, the check of the compressibility of surfaces seems to be not so necessary.
1441: %%% %
1442: %%% \begin{remark}
1443: %%% %
1444: %%% Though the algorithm in \cite{HO} excludes compressible surfaces in the set of candidate surfaces,
1445: %%% we will omit some check of compressibility
1446: %%% in the proofs.
1447: %%% %%%In short,
1448: %%% %%%the upper bound is obtained without completely excluding compressible surfaces.
1449: %%% %
1450: %%% \end{remark}
1451: 
1452: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1453: %
1454: % Section : A bound on the denominator
1455: %
1456: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1457: 
1458: \section{A bound on the denominator}
1459: \label{Sec:ABoundOnTheDenominator}
1460: 
1461: The purpose of this section is to prove Theorem \ref{Thm:Denom:UpperBound:Main}
1462: about an upper bound of the denominator of a boundary slope.
1463: Proving Lemma \ref{Lem:LowerBound:-XoverS} which claims a lower bound of ${-\chi}/{\sharp s}$,
1464: immediately gives the theorem.
1465: We also show the best possibility and some corollaries.
1466: 
1467: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1468: % Subsection : 
1469: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1470: 
1471: \subsection{A lower bound of $-\chi/\sharp s$}
1472: \label{SubSec:ALowerBoundOf-XoverS}
1473: 
1474: This subsection is devoted to giving a proof of Lemma \ref{Lem:LowerBound:-XoverS}, which claims a lower bound of $-\chi/\sharp s$.
1475: 
1476: \begin{lemma}
1477: \label{Lem:LowerBound:-XoverS}
1478: Let $\Slope=\Numer/\Denom$ be a finite boundary slope.
1479: The corresponding essential surface $F$
1480: has Euler characteristic $\chi$,
1481: the number of sheets $\sharp s$,
1482: and the number of boundary components $\sharp b$.
1483: Then, except for some boundary slopes,
1484: these values satisfy
1485: \[
1486: {-\chi}/{\sharp s}\ge 1.
1487: \]
1488: 
1489: The exceptions occur from $(-2,3,t)$-pretzel knots for odd $t\ge 3$ or their mirror images.
1490: One of the boundary slopes for the knot only satisfies 
1491: \[
1492: {-\chi}/{\sharp s}\ge (\Denom-1)/\Denom
1493: ,
1494: \]
1495: though a slightly stronger condition $\sharp b\ge 2$ holds at the same time.
1496: \end{lemma}
1497: 
1498: \subsubsection{Type II surfaces and Type III surfaces}
1499: 
1500: A lower bound of $-\chi/\sharp s$ is easily obtained
1501: for type II surfaces and type III surfaces.
1502: 
1503: \begin{lemma}
1504: $-\chi/\sharp s \ge 1$
1505: holds for any type II surface and any type III surface 
1506: corresponding to any Montesinos knot with $\NumTangles \ge 3$ tangles.
1507: \end{lemma}
1508: 
1509: \begin{proof}
1510: %
1511: For a type II surface,
1512: since every edgepath has at least $1$ complete edge in $u>0$,
1513: we have
1514: $
1515: -\chi/\sharp s
1516: =
1517:  (\sum_{i=1}^{\NumTangles}|\Edgepath_{i,>0} |)
1518:  +|\EdgepathSystem(+0) |-2
1519: \ge 1
1520: $.
1521: For a type III surface,
1522: $-\chi/\sharp s
1523: =
1524: \left( 
1525:  \sum_{i=1}^{\NumTangles}|\Edgepath_{i,>0} |
1526: \right)
1527: \ge 3
1528: $.
1529: By the simplification, we can ignore the effect by the augmentation and type III surfaces with partial $\infty$-edges.
1530: %%% Though augmentation may be performed upon a surface $F$,
1531: %%% the operation only increases $-\chi/\sharp s$.
1532: %%% Besides, 
1533: %%% there exists a surface whose edgepaths end at the middle of $\infty$-edges.
1534: %%% This is a neutral example between a type II surface and a typical type III surface, and $-\chi/\sharp s\ge 1$ holds in any $-1\le u\le 0$.
1535: %
1536: \end{proof}
1537: 
1538: \subsubsection{Type I surfaces}
1539: 
1540: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1541: % Paragraph : Denominator sequences
1542: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1543: 
1544: For type I surfaces, to verify the bound is not so easy as type II and type III surfaces.
1545: Though, for a major part of the type I surfaces,
1546: the bound is shown by ``denominator sequences'' only.
1547: 
1548: Fix a Montesinos knot $K$ and 
1549: a basic edgepath system $\BasicEdgepathSystem$.
1550: Though type I edgepath systems must correspond to the solution $u$ of the equation (\ref{Eq:EquationForEdgepathSystem}),
1551: we can formally calculate $-\chi/\sharp s$ by the formula (\ref{Eq:Formula:EulerCharTypeI}) for arbitrary $0<u<1$.
1552: Thus, we have a function $X_{\BasicEdgepathSystem}(u):(0,1)\rightarrow \mathbb{R}$.
1553: %
1554: The function depends on the basic edgepath system $\BasicEdgepathSystem$. 
1555: Though, by examining the formula (\ref{Eq:Formula:EulerCharTypeI})
1556: together with the formula (\ref{Eq:Formula:LengthOfPartialEdge}) about lengths of partial edges,
1557: we can confirm that the function $X_{\BasicEdgepathSystem}$ does not depend on the numerators of vertices which edgepaths pass through or reach.
1558: Namely, for a basic edgepath $\BasicEdgepath_i=\angleb{p_{i,j}/q_{i,j}}$\,--\,$\angleb{p_{i,j-1}/q_{i,j-1}}$\,--\,$\cdots$\,--\,$\angleb{p_{i,2}/q_{i,2}}$\,--\,$\angleb{p_{i,1}/q_{i,1}}$,
1559: $X_{\BasicEdgepathSystem}$ depends on only a $N$-tuple of {\em sequences of denominators} of the form $q_{i,j}(=1)$\,--\,$q_{i,j-1}$\,--\,$\cdots$\,--\,$q_{i,2}$\,--\,$q_{i,1}$.
1560: 
1561: We introduce a preorder of basic edgepaths and basic edgepath systems.
1562: For two basic edgepaths $\BasicEdgepath_a$ and $\BasicEdgepath_b$,
1563: we say that $\BasicEdgepath_a \le \BasicEdgepath_b$
1564: if $q_{a,k}\le q_{b,k}$ for all $k=1,2,\ldots,\min(j_a,j_b)$
1565: where $j_a$ and $j_b$ mean the lengths of their denominator sequences. 
1566: For two basic edgepath systems $\BasicEdgepathSystem_a$ and $\BasicEdgepathSystem_b$,
1567: we define a preorder by $\BasicEdgepathSystem_a \le \BasicEdgepathSystem_b$ if $\BasicEdgepath_{a,i} \le \BasicEdgepath_{b,i}$ is satisfied for all indices $i$.
1568: It is easy to confirm that
1569: if $\BasicEdgepathSystem_a\le \BasicEdgepathSystem_b$,
1570: then $X_{\BasicEdgepathSystem_a}(u)\le X_{\BasicEdgepathSystem_b}(u)$ holds for any $u \in (0,1)$.
1571: 
1572: By elementary calculations,
1573: for a basic edgepath system $\BasicEdgepathSystem$
1574: whose set of denominator sequences is one of
1575: %
1576: \begin{eqnarray}
1577: \begin{minipage}{10cm}
1578: \begin{itemize}
1579: \item \{$1$\,--\,$2$, $1$\,--\,$3$, $1$\,--\,$3$, $1$\,--\,$3$, $\ldots$\} ($\NumTangles \ge 4$), 
1580: \item \{$1$\,--\,$2$, $1$\,--\,$7$, $1$\,--\,$3$\,--\,$7$\},
1581: \item \{$1$\,--\,$2$, $1$\,--\,$2$\,--\,$7$, $1$\,--\,$2$\,--\,$7$\},
1582: \item \{$1$\,--\,$3$, $1$\,--\,$4$, $1$\,--\,$7$\},
1583: \item \{$1$\,--\,$3$, $1$\,--\,$4$, $1$\,--\,$2$\,--\,$5$\},
1584: \item \{$1$\,--\,$3$, $1$\,--\,$5$, $1$\,--\,$5$\},
1585: \item \{$1$\,--\,$4$, $1$\,--\,$4$, $1$\,--\,$4$\},
1586: \end{itemize}
1587: \end{minipage}
1588: \label{Items:DenomSequences}
1589: \end{eqnarray}
1590: %
1591: the inequality $X_{\BasicEdgepathSystem}(u)\ge 1$ holds for arbitrary $u$.
1592: %
1593: For instance,
1594: if a basic edgepath system $\BasicEdgepathSystem$ has the set of denominator sequences \{$1$\,--\,$4$, $1$\,--\,$4$, $1$\,--\,$4$\},
1595: $X_{\BasicEdgepathSystem}(u)$ is $1$ for $0<u\le 3/4$ and is $1/\{4(1-u)\}$ for $3/4\le u<1$.
1596: %%% It may be useful to introduce a variable $w=1/(1-u)$.
1597: %
1598: Necessarily, another edgepath system greater than such an edgepath system also satisfies $X_{\BasicEdgepathSystem}(u)\ge 1$.
1599: Note that \{$1$\,--\,$2$, $1$\,--\,$3$, $1$\,--\,$3$, $1$\,--\,$3$, \ldots \} ($\NumTangles\ge4$) is the denominator sequence of one of the smallest basic edgepath systems for fixed $\NumTangles$.
1600: %%% Similarly, the smallest basic edgepath system satisfies $X_{\BasicEdgepathSystem}(u)\ge 1$ for $\NumTangles\ge 5$.
1601: 
1602: Hence, in a sense, a major part of the edgepath systems corresponding to candidate surfaces satisfies $-\chi/\sharp s \ge 1$.
1603: We have:
1604: %
1605: \begin{lemma}
1606: \label{Lem:LowerBound:-XoverS:B}
1607: Assume that a type I edgepath system $\EdgepathSystem$ 
1608: is included in the extended basic edgepath system of a basic edgepath system $\BasicEdgepathSystem$.
1609: If the basic edgepath system $\BasicEdgepathSystem$
1610: is equal to or greater than one of 
1611: the basic edgepath systems listed in (\ref{Items:DenomSequences}).
1612: Then $\EdgepathSystem$ satisfies $-\chi/\sharp s\ge 1$.
1613: Especially,
1614: any type I surface of a Montesinos knot with $\NumTangles\ge4$ tangles 
1615: always satisfies the inequality.
1616: \end{lemma}
1617: 
1618: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1619: % Paragraph : Remaining cases
1620: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1621: 
1622: \subsubsection{Remaining cases}
1623: 
1624: We only have to check for the rest of the edgepath systems.
1625: For the remaining basic edgepath systems $\BasicEdgepathSystem$,
1626: we concretely solve the equation $\sum \widetilde{\BasicEdgepath_i}(u)=0$,
1627: enumerate all the candidate edgepath systems,
1628: and calculate $-\chi/\sharp s$ one by one.
1629: %%% Since detail of the calculation is very complicated and troublesome,
1630: %%% it is given in the next section.
1631: %
1632: %%%In this section,
1633: %%%we perform calculation of $-\chi/\sharp s$ for remaining cases of edgepath systems.
1634: Remaining cases are described by denominator sequences as follows.
1635: \begin{eqnarray}
1636: \begin{minipage}{10cm}
1637: \begin{itemize}
1638:  \item $\BasicEdgepath_1:1$--$2$, $\BasicEdgepath_2:\cdots$\,--$3$, $\BasicEdgepath_3:$ arbitrary,
1639:  \item $\BasicEdgepath_1:1$--$2$, $\BasicEdgepath_2:\cdots$\,--$5$, $\BasicEdgepath_3:$ arbitrary,
1640:  \item $\BasicEdgepath_1:1$--$2$, $\BasicEdgepath_2:1$--$t_1$, $\BasicEdgepath_3:1$--$t_2$ ($t_1,t_2$ odd and $\ge7$),
1641:  \item $\BasicEdgepath_1:1$--$2$, $\BasicEdgepath_2:1$--$t_1$, $\BasicEdgepath_3:1$--$2$--$t_2$ ($t_1,t_2$ odd and $\ge7$),
1642:  \item $\BasicEdgepath_1:1$--$3$ or $1$--$2$--$3$, $\BasicEdgepath_2:1$--$3$ or $1$--$2$--$3$, $\BasicEdgepath_3:$ arbitrary,
1643:  \item $\BasicEdgepath_1:1$--$3$ or $1$--$2$--$3$, $\BasicEdgepath_2:1$--$4$ or $\cdots$\,--$3$--$4$, $\BasicEdgepath_3:1$--$5$.
1644: \end{itemize}
1645: \end{minipage}
1646: \label{Items:RemainingCases}
1647: \end{eqnarray}
1648: Basically, $\BasicEdgepath_1$, $\BasicEdgepath_2$ and $\BasicEdgepath_3$ are arranged
1649: so that the denominators of their starting points $K_i$ are in ascending order.
1650: 
1651: \begin{lemma}
1652: \label{Lem:LowerBound:RemainingCase}
1653: For the rest of type I surfaces,
1654: that is, the type I surfaces to which Lemma \ref{Lem:LowerBound:-XoverS:B} cannot be applied,
1655: nevertheless Lemma \ref{Lem:LowerBound:-XoverS} holds.
1656: Namely,
1657: except for some boundary slopes,
1658: \[
1659: %%%\Denom \le \frac{-\chi}{\sharp b}
1660: -\chi/\sharp s\ge 1
1661: \]
1662: holds.
1663: The exceptions occur from $(-2,3,t)$-pretzel knots for odd $t\ge 3$ or their mirror images.
1664: One of the boundary slopes for the knots only satisfies
1665: \[
1666: %%%\Denom \le \frac{-\chi}{\sharp b} + 1,
1667: -\chi/\sharp s\ge (\Denom-1)/\Denom
1668: \]
1669: though a slightly stronger condition $\sharp b\ge 2$ holds at the same time.
1670: \end{lemma}
1671: 
1672: \begin{proof}
1673: 
1674: First,
1675: we enumerate all choices of a pair ($\BasicEdgepath_1$, $\BasicEdgepath_2$) of basic edgepath systems
1676: whose denominator sequences are included in the list (\ref{Items:RemainingCases}).
1677: Without loss of generality,
1678: we can normalize edgepath systems 
1679: by assuming that the tangles $K_1$ and $K_2$ of the Montesinos knot $K$ satisfy $0<K_1,K_2<1$
1680: and that the last edge of the edgepath $\BasicEdgepath_1$ is decreasing.
1681: There are 27 possible pairs of two edgepaths as listed in the rest of this subsection.
1682: For each choice of $\BasicEdgepath_1$ and $\BasicEdgepath_2$,
1683: we think about the sum $\BasicEdgepath_1+\BasicEdgepath_2$,
1684: which is a function defined by $\BasicEdgepath_1(u)+\BasicEdgepath_2(u)$.
1685: Then, we seek all edges of $\Diagram$ intersecting the sum 
1686: and all vertices on the sum.
1687: For each such an edge or vertex $\Edge$,
1688: we take its mirror image $\Edge_3=-\Edge$ with respect to the $u$-axis,
1689: and then make a basic edgepath $\BasicEdgepath_3$ including $\Edge_3$.
1690: The triple $(\BasicEdgepath_1, \BasicEdgepath_2, \BasicEdgepath_3)$ is ignored if it does not match with the condition in the list (\ref{Items:RemainingCases}).
1691: $\BasicEdgepath_1$, $\BasicEdgepath_2$ and $\BasicEdgepath_3$
1692: have a solution of the equation (\ref{Eq:EquationForEdgepathSystem})
1693: at the $u$-coordinate of the intersection point.
1694: From the solution, we cut the basic edgepaths ($\BasicEdgepath_1$, $\BasicEdgepath_2$, $\BasicEdgepath_3$)
1695: and obtain constant or non-constant edgepaths ($\Edgepath_1$, $\Edgepath_2$, $\Edgepath_3$).
1696: Note that there are many choices of $\BasicEdgepath_3$ which share $\Edge_3$ as the common tail part but have different parts.
1697: Hence, we implicitly discuss many choices of edgepath $\Edgepath_3$ at the same time, though the edgepath with minimum $-\chi/\sharp s$ is important.
1698: 
1699: 
1700: In the detailed argument,
1701: note that the denominator $\Denom$ of the slope is the same as the denominator of the twist.
1702: Besides, 
1703: for a constant edgepath $\Edgepath_i$,
1704: a fraction $k_i/m_i$ denotes a particular ratio
1705: which appears in the description
1706: $((k_i/m_i)\angleb{p_i/q_i}+(1-k_i/m_i)\circleb{p_i/q_i})$
1707: of the unique point of the constant edgepath,
1708: especially when we calculate $\sharp s$.
1709: The following fact is used often in the argument.
1710: 
1711: \begin{remark}\label{Rem:Condition:ForLinkToBeKnot:2}
1712: The starting points of an edgepath system
1713: correspond to the tangles of the Montesinos link.
1714: Therefore, the starting points must satisfy a condition in Remark \ref{Rem:Condition:ForLinkToBeKnot},
1715: for the corresponding Montesinos link to be a knot.
1716: If an edgepath system does not satisfy the condition,
1717: we must add at least one edge to the beginning of some edgepath of an edgepath system.
1718: \end{remark}
1719: 
1720: Here, we briefly show the calculation of $-\chi/\sharp s$ for the 27 cases.
1721: 
1722: \begin{enumerate}
1723: 
1724: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1725: % Item 1 
1726: %   \Edgepath_1:0-1/2
1727: %   \Edgepath_2:0-1/3
1728: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1729: 
1730: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/3}$}.
1731: \label{Item:0-1/2,0-1/3}
1732: %
1733: In Figure \ref{Fig:Edgepaths},
1734: the left and the middle pictures illustrate extended basic edgepath systems $\widetilde{\BasicEdgepath_1}$ and $\widetilde{\BasicEdgepath_2}$,
1735: while the right figure shows $\widetilde{\BasicEdgepath_1}+\widetilde{\BasicEdgepath_2}$ and possible choices of the edge $\Edge$.
1736: 
1737: \begin{figure}[hbt]
1738: \begin{picture}(256,144)%%%(320,190)
1739:  \scalebox{0.8}{
1740:    \put(0,0){
1741:     \put(0,0){\scalebox{1.0}{\includegraphics{e_0_1o2.eps}}}
1742:     \put(90,2){u}
1743:     \put(81,-5){1}
1744:     \put(-5,0){0}
1745:     \put(3,-5){0}
1746:     \put(-3,81){1}
1747:     \put(3,91){v}
1748:    }
1749:    \put(110,0){
1750:     \put(0,0){\scalebox{1.0}{\includegraphics{e_0_1o3.eps}}}
1751:     \put(90,2){u}
1752:     \put(81,-5){1}
1753:     \put(-5,0){0}
1754:     \put(3,-5){0}
1755:     \put(-3,81){1}
1756:     \put(3,91){v}
1757:    }
1758:    \put(220,0){
1759:     \put(0,0){\scalebox{1.0}{\includegraphics{e_0_1o2_e_0_1o3.eps}}}
1760:     \put(90,2){u}
1761:     \put(81,-5){1}
1762:     \put(-5,0){0}
1763:     \put(3,-5){0}
1764:     \put(-3,81){1}
1765:     \put(-3,161){2}
1766:     \put(3,169){v}
1767:    }
1768:  }
1769: \end{picture}
1770: \caption{$\widetilde{\BasicEdgepath_1}$, $\widetilde{\BasicEdgepath_2}$, $\widetilde{\BasicEdgepath_1}$+$\widetilde{\BasicEdgepath_2}$, $\Edge$}
1771: \label{Fig:Edgepaths}
1772: \end{figure}
1773: 
1774: \begin{enumerate}
1775: 
1776:  \item $\Edge_3=\angleb{-1}$--$\angleb{-1/2}$. %%%\\
1777:   \label{Item:0-1/2,0-1/3,A}
1778:    The equation (\ref{Eq:EquationForEdgepathSystem}) gives $3/2\, u=1-u$,
1779:    and we have $u=2/5 ~~(0<2/5<1/2)$.
1780:    From formulae (\ref{Eq:Formula:LengthOfPartialEdge}) and 
1781:    (\ref{Eq:Formula:EulerCharTypeI}),
1782:    $|\Edgepath_1|=\{1+2(2/5 -1)\}/\{(2-1)(2/5 -1)\}
1783:     =1/3$,
1784:    $|\Edgepath_2|=\{1+3(2/5 -1)\}/\{(3-1)(2/5 -1)\}
1785:     =2/3$,
1786:    $|\Edgepath_3|=|\Edgepath_1|
1787:     =1/3$,
1788:    $-\chi/\sharp s =1/3+2/3+1/3+(0-3)+(3-2)\cdot 5/3=4/3 -3+5/3 
1789:     =0$.
1790:   %%\\
1791:   Though, as Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, we must extend the edgepath $\Edgepath_3$ by adding at least one edge. Thus, $-\chi/\sharp s \ge 1$.
1792: 
1793:  \item $\Edge_3=\angleb{-1}$--$\angleb{-2/3}$. %%%\\
1794:   \label{Item:0-1/2,0-1/3,B}
1795: %%%   $1/2 u+1/2 =1-1/2 u$,
1796:    $u=1/2$,
1797:    $|\Edgepath_1|=0$,
1798:    $|\Edgepath_2|%%%=\frac{1+3(1/2 -1)}{(3-1)(1/2 -1)}
1799:     =1/2$,
1800:    $|\Edgepath_3|%%%=|\Edgepath_2|
1801:     =1/2 $,
1802:    $-\chi/\sharp s %%%=0+1/2 +1/2 +(0-3)+(3-2)\cdot 2
1803:     =0$.
1804:   %%%\\
1805:   This example is obtained for the torus knot $K(-1/2,1/3,1/3)$.
1806:   %%% Regarding the twist, we have
1807:   $|\Twist| %%%=2\,||\Edgepath_1|+|\Edgepath_2|+|\Edgepath_3|| =2(0+1/2 +1/2 )
1808:     =2$.
1809:   %%% The denominator of the slope $\Denom$ is the same as the denominator $1$ of the twist. 
1810:   $\Denom=1$.
1811:   Thus, $-\chi/\sharp s =0=(\Denom-1)/\Denom$.
1812: %  Hence, (\ref{Eq:-XoverSbeta-1beta}) holds.
1813:   Moreover, $\sharp s=\mathrm{lcm}(1,2,2)=2$ and $\Denom=1$ give $\sharp b=2$.
1814: 
1815:  \item $\Edge_3=\angleb{-1}$--$\angleb{-3/4}$. %%%\\
1816:   \label{Item:0-1/2,0-1/3,C}
1817: %%%   $1/2 u+1/2 =1-1/3 u$,
1818:    $u=3/5  ~~(1/2 <3/5 <2/3 )$,
1819:    $\Edgepath_1$ is constant,
1820:    $|\Edgepath_2|%%%=\frac{1+3(3/5 -1)}{(3-1)(3/5 -1)}
1821:      =1/4 $,
1822:    $|\Edgepath_3|%%%=\frac{1+4(3/5 -1)}{(4-1)(3/5 -1)}
1823:      =1/2 $,
1824:    $-\chi/\sharp s %%%=1/4+1/2+(1-3)+(3-2-1/2)\cdot 5/2 
1825:      =0$.
1826: %%%  \\
1827:   By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s \ge 1$.
1828: 
1829:  \item $\Edge_3=\angleb{-1}$--$\angleb{-4/5}$. %%%\\
1830:   \label{Item:0-1/2,0-1/3,D}
1831: %%%   $1/2 u+1/2 =1-1/4 u$,
1832:    $u=2/3 $,
1833: %%%   $\frac{1}{1-u}=\frac{1}{1-\frac{2}{3}}=3$,
1834:    $\Edgepath_1$ is constant,
1835:    $|\Edgepath_2|=0$,
1836:    $|\Edgepath_3|%%%=\frac{1+5(2/3 -1)}{(5-1)(2/3 -1)}
1837:      =1/2 $,
1838:    $-\chi/\sharp s %%%=0+1/2 +(1-3)+(3-2-1/2 )\cdot 3
1839:       =0$.
1840: %%%   \\
1841:   This example is obtained for the torus knot $K(-1/2 ,1/3 ,1/5)$.
1842:     $|\Twist| %%%=2||\Edgepath_2|+|\Edgepath_3||
1843:        =1$.
1844: %%%  The denominator of the slope $\Denom$ is the same as the denominator of the twist $1$.
1845:     $\Denom=1$.
1846:     $-\chi/\sharp s =0=\Denom/(\Denom-1)$.
1847: %  Hence, (\ref{Eq:-XoverSbeta-1beta}) holds.
1848: %%%
1849:   Moreover, $k_1/m_1 =2/3 $, $\sharp s=\mathrm{lcm}(2,1,2)=2$ and $\Denom=1$ give $\sharp b=2$.
1850: 
1851:  \item $\Edge_3=\angleb{-5/6}$. %%%\\
1852:   \label{Item:0-1/2,0-1/3,E}
1853:    $u=5/6 $,
1854:    $\Edgepath_1$ and
1855:    $\Edgepath_2$ is constant,
1856:    $|\Edgepath_3|=0$,
1857:    $-\chi/\sharp s %%%=0+(2-3)+(3-2-1/2 -1/3 )\cdot 6
1858:      =0$.
1859: %%%  \\
1860:   By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s \ge 1$.
1861: \end{enumerate}
1862: 
1863: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1864: % Item 2 
1865: %   \Edgepath_1:0-1/2
1866: %   \Edgepath_2:1-1/2-1/3
1867: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1868: 
1869: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$}.
1870: \label{Item:0-1/2,1-1/2-1/3}
1871: \begin{enumerate}
1872:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 5$). %%%\\ 
1873:   \label{Item:0-1/2,1-1/2-1/3,B}
1874: %%%   $3/2 -u=1-1/(t-1) u$,
1875:    $u=(t-1)/\{2(t-2)\} ~~(1/2 <u\le 2/3 )$,
1876:    $\Edgepath_1$ is constant,
1877:    $|\Edgepath_2|%%%=\frac{1+3(\frac{t-1}{2t-4}-1)}{(3-2)(\frac{t-1}{2t-4}-1)}
1878:      =(t-5)/(t-3)$,
1879:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{2t-4}-1)}{(t-1)(\frac{t-1}{2t-4}-1)}
1880:      =(t-4)/(t-3)$,
1881:    $-\chi/\sharp s%%% =(t-5)/(t-3)+(t-4)/(t-3)+(1-3)+(3-2-1/2 )\cdot (2t-4)/(t-3)
1882:      =1-2/(t-3)$.
1883:   \\
1884:     If $t=6$, $-\chi/\sharp s = 1/3 $. 
1885:      By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s = 1/3 +1\ge 1$. 
1886:      If $t$ is even, $-\chi/\sharp s \ge 1$ similarly.
1887:     If $t=5$, $-\chi/\sharp s =0$.
1888:      This example is obtained for the torus knot $K(-1/2 ,1/3 ,1/5 )$.
1889:      For odd $t\ge 7$, 
1890:       $|\Twist|%%%=2||\Edgepath_2|-|\Edgepath_3||=2|(t-5)/(t-3)-(t-4)/(t-3)|
1891:        =2/(t-3)$.
1892:      $\Denom=(t-3)/2$.
1893: %%%      The denominator of the slope $\Denom$ is the same as the denominator of the twist $(t-3)/2$. 
1894:      $(\Denom-1)/\Denom=1-2/(t-3)=-\chi/\sharp s $.
1895:      %
1896:      Moreover, for the constant edgepath $\Edgepath_1$,
1897:      $k_1/m_1 =q_1(1-u)=(t-3)/(t-2)$, $\sharp s=\mathrm{lcm}(t-3,(t-3)/2,t-3)=t-3$, and $\Denom=(t-3)/2$ give $\sharp b=2$.
1898: 
1899:  \item $\Edge_3=\angleb{-5/6}$. %%%\\
1900:    The edgepath system is same as in the item \ref{Item:0-1/2,0-1/3,E}.
1901: \end{enumerate}
1902: 
1903: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1904: % Item 3
1905: %   \Edgepath_1:0-1/2
1906: %   \Edgepath_2:1-2/3
1907: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1908: 
1909: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{2/3}$}.
1910: \label{Item:0-1/2,1-2/3}
1911: \begin{enumerate}
1912: 
1913:  \item $\Edge_3=\angleb{-1}$--$\angleb{-4/3}$. %%%\\
1914:   \label{Item:0-1/2,1-2/3,A}
1915:    $u$ is non-isolated ($0\le u\le 1/2 $),
1916:    $|\Edgepath_1|%%%=\frac{1+2(u-1)}{(2-1)(u-1)}
1917:      =(1-2u)/(1-u)$,
1918:    $|\Edgepath_2|%%%=\frac{1+3(u-1)}{(3-1)(u-1)}
1919:      =(2-3u)/(2-2u)$,
1920:    $|\Edgepath_3|%%%=|\Edgepath_2|
1921:      =(2-3u)/(2-2u)$,
1922:    $-\chi/\sharp s %%%=(1-2u)/(1-u)+(2-3u)/(2-2u)+(2-3u)/(2-2u)+(0-3)+(3-2)/(1-u)
1923:      =2-1/(1-u)$.
1924:   At $u=0$, $-\chi/\sharp s =1$.
1925:   At $u=1/2 $, $-\chi/\sharp s =0$. 
1926:      $0\le -\chi/\sharp s \le 1$.
1927: 
1928:   This edgepath system is obtained for the torus knot $K(-1/2 ,1/3 ,1/3 )$.
1929: 
1930:     $|\Twist|%%%=2|(1-2u)/(1-u)-(2-3u)/(2-2u)-(2-3u)/(2-2u)|
1931:        =2$.
1932:     $\Denom=1$.
1933:   $(\Denom-1)/\Denom=0\le -\chi/\sharp s $.
1934: 
1935:     If $0<u<1/2 $, $|\Edgepath_1|\notin\mathbb{Z}$ gives $\sharp s\ge 2$. Since $\Denom=1$, we have $\sharp b\ge 2$.
1936:     In the case $u=0$, the surface can be regarded as Type II.
1937:     If $u=1/2 $, 
1938:     the argument reduces to the item \ref{Item:0-1/2,0-1/3,B} corresponding to the torus knot $K(-1/2 ,1/3 ,1/3 )$.
1939: 
1940:  \item $\Edge_3=\angleb{-1}$--$\angleb{-5/4}$.
1941:   \label{Item:0-1/2,1-2/3,B}
1942:   The calculation is same as in the item \ref{Item:0-1/2,0-1/3,C}, though the edgepath systems themselves do not coincide with each other.
1943: 
1944:  \item $\Edge_3=\angleb{-1}$--$\angleb{-6/5}$. 
1945:   \label{Item:0-1/2,1-2/3,C}
1946:   The calculation is same as in the item \ref{Item:0-1/2,0-1/3,D}.
1947: 
1948:  \item $\Edge_3=\angleb{-7/6}$. 
1949:   The calculation is same as in the item \ref{Item:0-1/2,0-1/3,E}.
1950: 
1951: \end{enumerate}
1952: 
1953: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1954: % Item 4
1955: %   \Edgepath_1:0-1/2
1956: %   \Edgepath_2:0-1/2-2/3
1957: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1958: 
1959: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$}.
1960: \label{Item:0-1/2,0-1/2-2/3}
1961: \begin{enumerate}
1962: 
1963:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 2$). %%%\\
1964:   \label{Item:0-1/2,0-1/2-2/3,A}
1965: %%%   $2u=1-\frac{1}{t-1}u$,
1966:    $u=(t-1)/(2t-1) ~~(1/3 \le u<1/2 )$,
1967:    $|\Edgepath_1|%%%=\frac{1+2(\frac{t-1}{2t-1}-1)}{(2-1)(\frac{t-1}{2t-1}-1)}
1968:      =1/t $,
1969:    $|\Edgepath_2|%%%=1+|\Edgepath_1|
1970:      =1+1/t $,
1971:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{2t-1}-1)}{(t-1)(\frac{t-1}{2t-1}-1)}
1972:      =(t-1)/t$,
1973:    $-\chi/\sharp s %%%=1/t+1+1/t+(t-1)/t+(0-3)+(3-2)\cdot (2t-1)/t
1974:      =1$.
1975: 
1976:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t+1)/t}$ ($t\ge 5$).
1977:   \label{Item:0-1/2,0-1/2-2/3,B}
1978:   The calculation is same as in the item \ref{Item:0-1/2,1-1/2-1/3,B}.
1979: 
1980:  \item $\Edge_3=\angleb{-7/6}$.
1981:   The calculation is same as in the item \ref{Item:0-1/2,0-1/3,E}.
1982: 
1983: \end{enumerate}
1984: 
1985: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1986: % Item 5
1987: %   \Edgepath_1:0-1/2
1988: %   \Edgepath_2:0-1/5
1989: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1990: 
1991: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/5}$}.
1992: \label{Item:0-1/2,0-1/5}
1993: \begin{enumerate}
1994:  \item $\Edge_3=\angleb{-1}$--$\angleb{-1/2}$. %%%\\
1995:   \label{Item:0-1/2,0-1/5,A}
1996: %%%   $5/4 u=1-u$,
1997:    $u=4/9  ~~(0<4/9 <1/2 )$,
1998:    $|\Edgepath_1|%%%=\frac{1+2(4/9 -1)}{(2-1)(4/9 -1)}
1999:      =1/5$,
2000:    $|\Edgepath_2|%%%=\frac{1+5(4/9 -1)}{(5-1)(4/9 -1)}
2001:      =4/5$,
2002:    $|\Edgepath_3|%%%=|\Edgepath_1|
2003:      =1/5 $,
2004:    $-\chi/\sharp s %%%=1/5+4/5+1/5+(0-3)+(3-2)\cdot 9/5 
2005:      =0$.
2006:    %%\\
2007:    By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s \ge 1$.
2008: 
2009:  \item $\Edge_3=\angleb{-2/3}$. %%%\\
2010:   \label{Item:0-1/2,0-1/5,B}
2011:    $u=2/3$,
2012: %%%   $\frac{1}{1-u}=\frac{1}{1-\frac{2}{3}}=3$,
2013:    $\Edgepath_1$ is constant,
2014:    $|\Edgepath_2|%%%=\frac{1+5(2/3 -1)}{(5-1)(2/3 -1)}
2015:      =1/2 $,
2016:    $|\Edgepath_3|=0$,
2017:    $-\chi/\sharp s %%%=1/2+0+(1-3)+(3-2-1/2)\cdot 3
2018:      =0$.
2019:   For the denominators of tangles to be in ascending order, we must add at least an edge to the edgepath $\Edgepath_3$. Hence, $-\chi/\sharp s \ge 1$.
2020: 
2021:  \item $\Edge_3=\angleb{-2/3}$--$\angleb{-5/7}$. %%%\\
2022:   \label{Item:0-1/2,0-1/5,C}
2023:    $u$ is non-isolated ($2/3 \le u\le 4/5 $),
2024:    $\Edgepath_1$ is constant,
2025:    $|\Edgepath_2|%%%=\frac{1+5(u-1)}{(5-1)(u-1)}
2026:      =(4-5u)/(4-4u)$, 
2027:    $|\Edgepath_3|%%%=\frac{1+7(u-1)}{(7-3)(u-1)}
2028:      =(6-7u)/(4-4u)$,
2029:    $-\chi/\sharp s %%%=(4-5u)/(4-4u)+(6-7u)/(4-4u)+(1-3)+(3-2-1/2)/(1-u)
2030:      =1$.
2031: 
2032:  \item $\Edge_3=\angleb{-7/10}$. %%%\\
2033:   \label{Item:0-1/2,0-1/5,D}
2034:    $u=9/10 $,
2035:    $\Edgepath_1$ and 
2036:    $\Edgepath_2$ is constant,
2037:    $|\Edgepath_3|=0$,
2038:    $-\chi/\sharp s %%%=0+(2-3)+(3-2-1/2 -1/5)\cdot 10
2039:      =2$.
2040: 
2041: \end{enumerate}
2042: 
2043: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2044: % Item 6
2045: %   \Edgepath_1:0-1/2
2046: %   \Edgepath_2:0-1/2-1/3-1/4-1/5
2047: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2048: 
2049: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$--$\angleb{1/4}$--$\angleb{1/5}$}.
2050: \label{Item:0-1-2,1-1/2-1/3-1/4-1/5}
2051: \begin{enumerate}
2052: 
2053:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 5$). %%%\\
2054:   \label{Item:0-1-2,1-1/2-1/3-1/4-1/5,B}
2055:   %%%The calculation is similar to the item \ref{Item:0-1/2,1-1/2-1/3,B}.
2056: %%%  
2057: %%%   $3/2 -u=1-\frac{1}{t-1}u$,
2058:    $u=(t-1)/\{2(t-2)\} ~(1/2 <u\le 2/3 )$,
2059:    $\Edgepath_1$ is constant,
2060:    $|\Edgepath_2|\ge 2$,
2061:    $|\Edgepath_3|\ge 0$,
2062:    $-\chi/\sharp s \ge %%%2+0+(1-3)+(3-2-1/2)\cdot 2=
2063:      1$.
2064: 
2065:  \item $\Edge_3=\angleb{-3/4}$. %%%\\
2066:   \label{Item:0-1-2,1-1/2-1/3-1/4-1/5,A}
2067:    $u=3/4$,
2068:    $\Edgepath_1$ is constant,
2069:    $|\Edgepath_2|=1$,
2070:    $|\Edgepath_3|=0$,
2071:    $-\chi/\sharp s %%%=1+0+(1-3)+(3-2-1/2)\cdot 4
2072:      =1$.
2073: 
2074:  \item $\Edge_3=\angleb{-2/3}$--$\angleb{-5/7}$. %%%\\
2075:   \label{Item:0-1-2,1-1/2-1/3-1/4-1/5,C}
2076: %%%   $3/2 -u=1/4 (u-2/3 )+2/3 $,
2077:    $u=4/5$,
2078:    $\Edgepath_1$ is constant,
2079:    $|\Edgepath_2|=0$,
2080:    $|\Edgepath_3|%%%=\frac{1+7(4/5 -1)}{(7-3)(4/5 -1)}
2081:      =1/2$,
2082:    $-\chi/\sharp s %%%=0+1/2+(1-3)+(3-2-1/2)\cdot 5
2083:      =1$.
2084: 
2085:  \item $\Edge_3=\angleb{-7/10}$.
2086:   The edgepath system is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2087: \end{enumerate}
2088: 
2089: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2090: % Item 7
2091: %   \Edgepath_1:0-1/2
2092: %   \Edgepath_2:1-1/2-2/5
2093: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2094: 
2095: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{2/5}$}.
2096: \label{Item:0-1/2,1-1/2-2/5}
2097: \begin{enumerate}
2098: 
2099:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 9$). %%%\\
2100:   \label{Item:0-1/2,1-1/2-2/5,A}
2101: %%%   $7/6 -1/3 u=1-\frac{1}{t-1}u$,
2102:    $u=(t-1)/\{2(t-4)\} ~(1/2 <u\le 4/5 )$,
2103:    $\Edgepath_1$ is constant,
2104:    $|\Edgepath_2|%%%=\frac{1+5(\frac{t-1}{2t-8}-1)}{(5-2)(\frac{t-1}{2t-8}-1)}
2105:      =(t-9)/(t-7)$,
2106:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{2t-8}-1)}{(t-1)(\frac{t-1}{2t-8}-1)}
2107:      =(t-8)/(t-7)$,
2108:    $-\chi/\sharp s %%%=(t-9)/(t-7)+(t-8)/(t-7)+(1-3)+(3-2-1/2)\cdot(2t-8)/(t-7)
2109:      =1$.
2110: 
2111:  \item $\Edge_3=\angleb{-9/10}$.
2112:   \label{Item:0-1/2,1-1/2-2/5,B}
2113:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2114: 
2115: \end{enumerate}
2116: 
2117: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2118: % Item 8
2119: %   \Edgepath_1:0-1/2
2120: %   \Edgepath_2:0-1/2-2/5
2121: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2122: 
2123: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/5}$}.
2124: \label{Item:0-1/2,0-1/2-2/5}
2125: \begin{enumerate}
2126: 
2127:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 2$) ($0<u<1/2 $). %%%\\
2128:   \label{Item:0-1/2,0-1/2-2/5,A}
2129:   %%%The calculation is similar to the item \ref{Item:0-1/2,0-1/2-2/3,A}.
2130: %%%   $2u=1-\frac{1}{t-1}u$,
2131:    $u=(t-1)/(2t-1) ~(1/3 \le u<1/2 )$,
2132:    $|\Edgepath_1|%%%=\frac{1+2(\frac{t-1}{2t-1}-1)}{(2-1)(\frac{t-1}{2t-1}-1)}
2133:      =1/t$,
2134:    $|\Edgepath_2|%%%=1+|\Edgepath_1|
2135:      =1+1/t $,
2136:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{2t-1}-1)}{(t-1)(\frac{t-1}{2t-1}-1)}
2137:      =(t-1)/t$,
2138:    $-\chi/\sharp s %%%=1/t +1+1/t +(t-1)/t+(0-3)+(3-2)\cdot (2t-1)/t
2139:      =1$.
2140: 
2141:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 9$) ($u>1/2 $).
2142:   \label{Item:0-1/2,0-1/2-2/5,B}
2143:   The edgepath system is same as in the item \ref{Item:0-1/2,1-1/2-2/5,A}.
2144: 
2145:  \item $\Edge_3=\angleb{-9/10}$.
2146: %%  ->\\\ref{Item:0-1/2,1-1/2-2/5} \ref{Item:0-1/2,1-1/2-2/5,B} .
2147:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2148: 
2149: \end{enumerate}
2150: 
2151: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2152: % Item 9
2153: %   \Edgepath_1:0-1/2
2154: %   \Edgepath_2:0-1/3-2/5
2155: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2156: 
2157: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/3}$--$\angleb{2/5}$}.
2158: \label{Item:0-1/2,0-1/3-2/5}
2159: \begin{enumerate}
2160: 
2161:  \item $\Edge_3=\angleb{-1}$--$\angleb{-1/2}$. %%%\\
2162:   %%%The calculation is similar to the item \ref{Item:0-1/2,0-1/3,A}.
2163: %%%   $3/2 u=1-u$,
2164:    $u=2/5  ~~(0<2/5 <1/2 )$,
2165:    $|\Edgepath_1|%%%=\frac{1+2(2/5 -1)}{(2-1)(2/5 -1)}
2166:      =1/3$,
2167:    $|\Edgepath_2|%%%=1+\frac{1+3(2/5 -1)}{(3-1)(2/5 -1)}
2168:      =1+2/3$,
2169:    $|\Edgepath_3|%%%=|\Edgepath_1|
2170:      =1/3 $,
2171:    $-\chi/\sharp s %%%=1/3+1+2/3+1/3+(0-3)+(3-2)\cdot 5/3
2172:      =1$.
2173: 
2174:  \item $\Edge_3=\angleb{-1}$--$\angleb{-2/3}$. %%% \\
2175: %%%   $3/2 u=1-1/2 u$,
2176:    $u=1/2 $,
2177:    $|\Edgepath_1|=0$,
2178:    $|\Edgepath_2|%%%=1+\frac{1+3(1/2 -1)}{(3-1)(1/2 -1)}
2179:     =1+1/2$,
2180:    $|\Edgepath_3|=1/2 $,
2181:    $-\chi/\sharp s %%%=0+1+1/2+1/2+(0-3)+(3-2)\cdot 2
2182:      =1$.
2183: 
2184:  \item $\Edge_3=\angleb{-1}$--$\angleb{-3/4}$. %%% \\
2185:   \label{Item:0-1/2,0-1/3-2/5,C}
2186: %%%   $1/2 +1/2 u=1-1/3 u$,
2187:    $u=3/5  ~~(1/2 <3/5 <2/3 )$,
2188:    $\Edgepath_1$ is constant,
2189:    $|\Edgepath_2|%%%=1+\frac{1+3(3/5 -1)}{(3-1)(3/5 -1)}
2190:      =1+1/4$,
2191:    $|\Edgepath_3|%%%=\frac{1+4(3/5 -1)}{(4-1)(3/5 -1)}
2192:      =1/2$,
2193:    $-\chi/\sharp s %%%=1+1/4+1/2+(1-3)+(3-2-1/2)\cdot 5/2
2194:      =1$.
2195: 
2196:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($5\le t\le 9$). %%%\\
2197:   \label{Item:0-1/2,0-1/3-2/5,D}
2198: %%%   $1/2 +1/2 u=1-\frac{1}{t-1}u$,
2199:    $u=(t-1)/(t+1) ~~(2/3 \le u \le 4/5 )$,
2200:    $\Edgepath_1$ is constant,
2201:    $|\Edgepath_2|%%%=\frac{1+5(\frac{t-1}{t+1}-1)}{(5-3)(\frac{t-1}{t+1}-1)}
2202:      =(9-t)/4$,
2203:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{t+1}-1)}{(t-1)(\frac{t-1}{t+1}-1)}
2204:      =1/2$,
2205:    $-\chi/\sharp s %%%=(9-t)/4+1/2 +(1-3)+(3-2-1/2)\cdot (t+1)/2
2206:      =1$.
2207: 
2208:  \item $\Edge_3=\angleb{-9/10}$.
2209:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2210: %%  \\->\ref{Item:0-1/2,1-1/2-2/5} \ref{Item:0-1/2,1-1/2-2/5,B}.
2211: 
2212: \end{enumerate}
2213: 
2214: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2215: % Item 10
2216: %   \Edgepath_1:0-1/2
2217: %   \Edgepath_2:1-1/2-3/5
2218: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2219: 
2220: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{3/5}$}.
2221: \begin{enumerate}
2222: 
2223:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t+1)/t}$ ($t\ge 9$).
2224:   The calculation is same as in the item \ref{Item:0-1/2,1-1/2-2/5,A}.
2225: 
2226:  \item $\Edge_3=\angleb{-11/10}$.
2227:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2228: %%  \\-> \ref{Item:0-1/2,1-1/2-2/5} \ref{Item:0-1/2,1-1/2-2/5,B}.
2229: 
2230: \end{enumerate}
2231: 
2232: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2233: % Item 11
2234: %   \Edgepath_1:0-1/2
2235: %   \Edgepath_2:0-1/2-3/5
2236: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2237: 
2238: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{3/5}$}.
2239: \begin{enumerate}
2240: 
2241:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$.
2242:   The calculation is same as in the item \ref{Item:0-1/2,0-1/2-2/5,A}.
2243: 
2244:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t+1)/t}$ ($t\ge 9$).
2245:   The calculation is same as in the item \ref{Item:0-1/2,1-1/2-2/5,A}.
2246: 
2247:  \item $\Edge_3=\angleb{-11/10}$.
2248:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2249: %%  \\->\ref{Item:0-1/2,1-1/2-2/5} \ref{Item:0-1/2,1-1/2-2/5,B}.
2250: 
2251: \end{enumerate}
2252: 
2253: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2254: % Item 12
2255: %   \Edgepath_1:0-1/2
2256: %   \Edgepath_2:1-2/3-3/5
2257: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2258: 
2259: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{2/3}$--$\angleb{3/5}$}.
2260: \begin{enumerate}
2261: 
2262:  \item $\Edge_3=\angleb{-1}$--$\angleb{-4/3}$. %%%\\
2263:   %%%The calculation is similar to the item \ref{Item:0-1/2,1-2/3,A}.
2264:    $u$ is non-isolated ($0\le u\le 1/2 $),
2265:    $|\Edgepath_1|%%%=\frac{1+2(u-1)}{(2-1)(u-1)}
2266:      =(1-2u)/(1-u)$,
2267:    $|\Edgepath_3|%%%=\frac{1+3(u-1)}{(3-1)(u-1)}
2268:      =(2-3u)/(2-2u)$,
2269:    $|\Edgepath_2|%%%=1+|\Edgepath_3|
2270:      =1+(2-3u)/(2-2u)$,
2271:    $-\chi/\sharp s %%%=(1-2u)/(1-u)+1+(2-3u)/(2-2u)+(2-3u)/(2-2u)+(0-3)+(3-2)/(1-u)
2272:      =1$.
2273: 
2274:  \item $\Edge_3=\angleb{-1}$--$\angleb{-5/4}$.
2275:   The calculation is same as in the item \ref{Item:0-1/2,0-1/3-2/5,C}.
2276: 
2277:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t+1)/t}$ ($5\le t \le 9$).
2278:   The calculation is same as in the item \ref{Item:0-1/2,0-1/3-2/5,D}.
2279: 
2280:  \item $\Edge_3=\angleb{-11/10}$.
2281:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2282: \end{enumerate}
2283: 
2284: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2285: % Item 13
2286: %   \Edgepath_1:0-1/2
2287: %   \Edgepath_2:1-4/5
2288: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2289: 
2290: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{4/5}$}.
2291: \begin{enumerate}
2292:  \item $\Edge_3=\angleb{-4/3}$.
2293:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,B}.
2294:  \item $\Edge_3=\angleb{-4/3}$--$\angleb{-9/7}$.
2295:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,C}.
2296:  \item $\Edge_3=\angleb{-13/10}$.
2297:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2298: \end{enumerate}
2299: 
2300: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2301: % Item 14
2302: %   \Edgepath_1:0-1/2
2303: %   \Edgepath_2:0-1/2-2/3-3/4-4/5
2304: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2305: 
2306: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$--$\angleb{3/4}$--$\angleb{4/5}$}.
2307: \begin{enumerate}
2308:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge2$). %%% \\
2309: %%%   $2u=1-\frac{1}{t-1}u$,
2310:    $u=(t-1)/(2t-1) ~(1/3 \le u<1/2 )$,
2311:    $|\Edgepath_1|%%%=\frac{1+2(\frac{t-1}{2t-1}-1)}{(2-1)(\frac{t-1}{2t-1}-1)}
2312:      =1/t $,
2313:    $|\Edgepath_2|%%%=3+|\Edgepath_1|
2314:      =3+1/t$,
2315:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{2t-1}-1)}{(t-1)(\frac{t-1}{2t-1}-1)}
2316:      =(t-1)/t$,
2317:    $-\chi/\sharp s %%%=1/t+3+1/t+(t-1)/t+(0-3)+(3-2)\cdot (2t-1)/t
2318:      =3$.
2319:   %%% This is similar to \ref{Item:0-1/2,0-1/2-2/5,B} but slightly different.
2320: 
2321:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t+1)/t}$ ($t\ge5$).
2322:   The calculation is same as in the item \ref{Item:0-1-2,1-1/2-1/3-1/4-1/5,B}.
2323:  \item $\Edge_3=\angleb{-1}$--$\angleb{-5/4}$.
2324:   The calculation is same as in the item \ref{Item:0-1-2,1-1/2-1/3-1/4-1/5,A}.
2325:  \item $\Edge_3=\angleb{-4/3}$--$\angleb{-9/7}$.
2326:   The calculation is same as in the item \ref{Item:0-1-2,1-1/2-1/3-1/4-1/5,C}.
2327:  \item $\Edge_3=\angleb{-13/10}$.
2328:   The calculation is same as in the item \ref{Item:0-1/2,0-1/5,D}.
2329: \end{enumerate}
2330: 
2331: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2332: % Item 15
2333: %   \Edgepath_1:0-1/2
2334: %   \Edgepath_2:0-1/t
2335: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2336: 
2337: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/t}$ (odd $t\ge 7$)}.
2338: \label{Item:0-1/2,0-1/t}
2339: \begin{enumerate}
2340: 
2341:   Properly speaking, $t \le 2p-1$ is required in the items \ref{Item:0-1/2,0-1/t,B}, \ref{Item:0-1/2,0-1/t,D} and \ref{Item:0-1/2,0-1/t,C} below.
2342:   Though, in this part,
2343:   we allow the denominator of the second tangle to be greater than that of the third tangle exceptionally. 
2344:   Thus, we avoid repeating essentially the same calculations.
2345: 
2346:  \item $\Edge_3=\angleb{-1}$--$\angleb{-1/2}$. %%%\\
2347:   \label{Item:0-1/2,0-1/t,A}
2348: %%%   $(1+\frac{1}{t-1})u=1-u$,
2349:    $u=(t-1)/(2t-1)$,
2350:    $|\Edgepath_1|%%%=\frac{1+2(\frac{t-1}{2t-1}-1)}{(2-1)(\frac{t-1}{2t-1}-1)}
2351:      =1/t$,
2352:    $|\Edgepath_2|%%%=\frac{1+t(\frac{t-1}{2t-1}-1)}{(t-1)(\frac{t-1}{2t-1}-1)}
2353:      =(t-1)/t$,
2354:    $|\Edgepath_3|%%%=|\Edgepath_1|
2355:      =1/t$,
2356:    $-\chi/\sharp s %%%=1/t+(t-1)/t+1/t +(0-3)+(3-2)\cdot(2t-1)/t
2357:      =0$.
2358:    %%%\\
2359:   By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s \ge 1$.
2360: 
2361:  \item $\Edge_3=\angleb{-1/2}$--$\angleb{-p/(2p-1)}$ ($p\ge 4$, $u\le (t-1)/t$). %%%\\
2362:   \label{Item:0-1/2,0-1/t,B}
2363:   $p\ge 4$ is derived from $2p-1\ge 7$.
2364:   %
2365:   By the conditions
2366:   $1/2+1/(t-1)\cdot u-\{1/(2p-3)\cdot(u-1/2)+1/2\}=0$ from (\ref{Eq:EquationForEdgepathSystem}),
2367:   $1/2\le u\le (t-1)/t$ and
2368:   $1/2\le u\le (2p-2)/(2p-1)$,
2369:   the solution $u=(t-1)/\{2(t-2p+2)\}$ exists if $t-4p+3\ge 0$ holds.
2370:   %
2371:   $\Edgepath_1$ is constant, 
2372:   $|\Edgepath_2|=(t-4p+4)/(t-4p+5)$,
2373:   $|\Edgepath_3|=(t-4p+3)/(t-4p+5)$,
2374:   $-\chi/\sharp s=1+(2p-6)/(t-4p+5)\ge 1$.
2375: 
2376:  \item $\Edge_3=\angleb{-1/2}$--$\angleb{-p/(2p-1)}$ ($p\ge 2$, $u\ge (t-1)/t$). %%%\\
2377:   \label{Item:0-1/2,0-1/t,D}
2378: %
2379:   By the conditions
2380:   $1/2+1/t-\{1/(2p-3)\cdot(u-1/2)+1/2\}=0$ from (\ref{Eq:EquationForEdgepathSystem}),
2381:   $u\ge (t-1)/t$ and
2382:   $1/2\le u\le (2p-2)/(2p-1)$,
2383:   we have a contradiction $4p-2\le t\le 4p-4$.
2384:   Thus, no solution exists.
2385: 
2386:  \item $\Edge_3=\angleb{-p/(2p-1)}$--$\circleb{-p/(2p-1)}$ ($p\ge 4$). %%%\\
2387:   \label{Item:0-1/2,0-1/t,C}
2388:   %
2389:   By the conditions
2390:   $1/2+1/(t-1)\cdot u-p/(2p-1)=0$,
2391:   $u\ge (2p-2)/(2p-1)$ and
2392:   $1/2\le u\le (t-1)/t$,
2393:   the solution $u=(t-1)/\{2(2p-1)\}$ exists if $4p-3\le t\le 4p-2$ holds.
2394:   The solution for $t=4p-3$ is treated in \ref{Item:0-1/2,0-1/t,B},
2395:   while $t=4p-2$ is even and unsuitable. 
2396: 
2397: \end{enumerate}
2398: 
2399: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2400: % Item 16
2401: %   \Edgepath_1:0-1/2
2402: %   \Edgepath_2:0-(t-1)/t
2403: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2404: 
2405: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{(t-1)/t}$ (odd $t\ge 7$)}.
2406: \begin{enumerate}
2407:  \item $\Edge_3=\angleb{-1/2}$--$\angleb{-p/(2p+1)}$ ($p\ge 2$). %%%\\
2408:   For the denominators of the tangles to be in ascending order, $7\le t\le 2p-1$ and $p\ge 4$ are required.
2409: %
2410:   The calculation is similar to the item \ref{Item:0-1/2,0-1/t,B}.
2411: \end{enumerate}
2412: 
2413: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2414: % Item 17
2415: %   \Edgepath_1:0-1/2
2416: %   \Edgepath_2:0-1/2-p/(2p\pm1)
2417: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2418: 
2419: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{p/(2p\pm 1)}$}.
2420: \begin{enumerate}
2421: 
2422:  \item $\Edge_3=\angleb{-1}$--$\angleb{-1/t}$. %%%\\
2423:   The calculation is same as in the item \ref{Item:0-1/2,0-1/t,B}.
2424: \end{enumerate}
2425: 
2426: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2427: % Item 18
2428: %   \Edgepath_1:0-1/3
2429: %   \Edgepath_2:0-1/3
2430: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2431: 
2432: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/3}$}.
2433: \label{Item:0-1/3,0-1/3}
2434: 
2435: \begin{enumerate}
2436: 
2437:  \item $\Edge_3=\angleb{0}$--$\angleb{-1/2}$.
2438:        %%% (including the case of a constant edgepath $\angleb{-1/2}$.)
2439:        %%%\\
2440:   \label{Item:0-1/3,0-1/3,A}
2441:   %%%The calculation is similar to the item \ref{Item:0-1/2,1-2/3,A}.
2442:    %
2443:    $u$ is non-isolated ($0\le u\le 1/2$), 
2444:    $|\Edgepath_1|%%%=\frac{1+3(u-1)}{(3-1)(u-1)}
2445:      =(3u-2)/(2u-2)$,
2446:    $|\Edgepath_2|%%%=|\Edgepath_1|
2447:      =(3u-2)/(2u-2)$,
2448:    $|\Edgepath_3|%%%=\frac{1+2(u-1)}{(2-1)(u-1)}
2449:      =(2u-1)/(u-1)$,
2450:    $-\chi/\sharp s %%%=(3u-2)/(2u-2)+(3u-2)/(2u-2)+(2u-1)/(u-1)+(0-3)+(3-2)/(1-u)
2451:      =2-1/(1-u)\ge 0$.
2452:   %\\
2453:   Similarly to the item \ref{Item:0-1/2,1-2/3,A},
2454:   $-\chi/\sharp s \ge (\Denom-1)/\Denom$ holds.
2455: 
2456:  \item $\Edge_3=\angleb{-1/2}$--$\angleb{-2/3}$.
2457:    %%% (including a constant edgepath $\angleb{-2/3}$.)
2458:    $u$ is non-isolated ($1/2 \le u \le 2/3 $),
2459:    $|\Edgepath_1|%%% =\frac{1+3(u-1)}{(3-1)(u-1)}
2460:       =(3u-2)/(2u-2)$,
2461:    $|\Edgepath_2|%%%=|\Edgepath_1|
2462:       =(3u-2)/(2u-2)$,
2463:    $|\Edgepath_3|%%%=\frac{1+3(u-1)}{(3-2)(u-1)}
2464:       =(3u-2)/(u-1)$,
2465:    $-\chi/\sharp s %%%=(3u-2)/(2u-2)+(3u-2)/(2u-2)+(3u-2)/(u-1)+(0-3)+(3-2)/(1-u)
2466:       =3-1/(1-u)\ge 0$.
2467:  %%%\\
2468:   By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s \ge 1$.
2469: 
2470: \end{enumerate}
2471: 
2472: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2473: % Item 19
2474: %   \Edgepath_1:0-1/3
2475: %   \Edgepath_2:1-2/3
2476: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2477: 
2478: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{2/3}$}. 
2479: %\\
2480:   No solution exists.
2481: 
2482: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2483: % Item 20
2484: %   \Edgepath_1:0-1/3
2485: %   \Edgepath_2:1-1/2-1/3
2486: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2487: 
2488: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$}.
2489: \begin{enumerate}
2490:  \item $\Edge_3=\angleb{-1}$--$\angleb{-2/3}$. %%%\\
2491:    $u$ is non-isolated ~~($0\le u\le1/2 $),
2492:    $|\Edgepath_1|%%%=\frac{1+3(u-1)}{(3-1)(u-1)}
2493:      =(3u-2)/(2u-2)$,
2494:    $|\Edgepath_2|%%%=1+\frac{1+2(u-1)}{(2-1)(u-1)}
2495:      =1+(2u-1)/(u-1)$,
2496:    $|\Edgepath_3|%%%=|\Edgepath_1|
2497:      =(3u-2)/(2u-2)$,
2498:    $-\chi/\sharp s %%%=(3u-2)/(2u-2)+1+(2u-1)/(u-1)+(3u-2)/(2u-2)+(0-3)+(3-2)/(1-u)
2499:      =3-1/(1-u)\ge 1$.
2500: 
2501:  \item $\Edge_3=\angleb{-1}$--$\angleb{-2/3}$.
2502:       %%%  ~~(including a constant edgepath $\angleb{-2/3}$.). 
2503:       %%%\\
2504:    $u$ is non-isolated ~~($1/2 < u\le2/3 $),
2505:    $|\Edgepath_1|%%%=\frac{1+3(u-1)}{(3-1)(u-1)}
2506:      =(3u-2)/(2u-2)$,
2507:    $|\Edgepath_2|%%%=\frac{1+3(u-1)}{(3-2)(u-1)}
2508:      =(3u-2)/(u-1)$,
2509:    $|\Edgepath_3|%%%=|\Edgepath_1|
2510:      =(3u-2)/(2u-2)$,
2511:    $-\chi/\sharp s %%%=(3u-2)/(2u-2)+(3u-2)/(u-1)+(3u-2)/(2u-2)+(0-3)+(3-2)/(1-u)
2512:      =3-1/(1-u)\ge 0$.
2513:   %%%\\
2514:    By Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, $-\chi/\sharp s \ge 1$.
2515: \end{enumerate}
2516: 
2517: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2518: % Item 21
2519: %   \Edgepath_1:0-1/3
2520: %   \Edgepath_2:0-1/2-2/3
2521: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2522: 
2523: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$}.
2524: \begin{enumerate}
2525: 
2526:  \item $\Edge_3=\angleb{-1}$--$\angleb{-1/2}$. %%%\\
2527: %%%  Similar to the item \ref{Item:0-1/2,0-1/3,A}.
2528: %%%   $3/2 u=1-u$,
2529:    $u=2/5  ~~(0<2/5 <1/2 )$,
2530:    $|\Edgepath_1|%%%=\frac{1+3(2/5 -1)}{(3-1)(2/5 -1)}
2531:      =2/3 $,
2532:    $|\Edgepath_2|%%%=1+\frac{1+2(2/5 -1)}{(2-1)(2/5 -1)}
2533:      =1+1/3 $,
2534:    $|\Edgepath_3|%%%=|\Edgepath_2|
2535:      =1/3 $,
2536:    $-\chi/\sharp s %%%=2/3+1+1/3+1/3+(0-3)+(3-2)\cdot 5/3
2537:      =1$.
2538: 
2539:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ($t\ge 3$). %%%\\
2540: %%%   $3/2 u=1-\frac{1}{t-1}u$,
2541:    $u=(2t-2)/(3t-1) ~(1/2 \le u <2/3 )$,
2542:    $|\Edgepath_1|%%%=\frac{1+3(\frac{2t-2}{3t-1}-1)}{(3-1)(\frac{2t-2}{3t-1}-1)}
2543:      =2/(t+1)$,
2544:    $|\Edgepath_2|%%%=\frac{1+3(\frac{2t-2}{3t-1}-1)}{(3-2)(\frac{2t-2}{3t-1}-1)}
2545:      =4/(t+1)$,
2546:    $|\Edgepath_3|%%%=\frac{1+t(\frac{2t-2}{3t-1}-1)}{(t-1)(\frac{2t-2}{3t-1}-1)}
2547:      =(t-1)/(t+1)$,
2548:    $-\chi/\sharp s %%%=2/(t+1)+4/(t+1)+(t-1)/(t+1)+(0-3)+(3-2)\cdot (3t-1)/(t+1)
2549:      =1$.
2550: \end{enumerate}
2551: 
2552: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2553: % Item 22
2554: %   \Edgepath_1:0-1/2-2/3
2555: %   \Edgepath_2:0-1/2-2/3
2556: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2557: 
2558: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$}.
2559: \begin{enumerate}
2560:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t-1)/t}$ ~($t\ge 2)$. %%%\\
2561: %%%  The calculation is similar to the item \ref{Item:0-1/2,0-1/2-2/5,B}.
2562: %%%   $2u=1-\frac{1}{t-1}u$,
2563:    $u=(t-1)/(2t-1) ~~(1/3 \le u<1/2 )$,
2564:    $|\Edgepath_1|=1+1/t $,
2565:    $|\Edgepath_2|=1+1/t $,
2566:    $|\Edgepath_3|=1-1/t $,
2567:    $-\chi/\sharp s %%%=3+1/t +(0-3)+(3-2)\cdot (2t-1)/t
2568:      =2$.
2569: 
2570:  \item $\Edge_3=\angleb{-1}$--$\angleb{-(t+1)/t}$.
2571:     %%% ~($t\ge 3$) (including a constant edgepath $\angleb{-4/3}$.)
2572:     %%%\\
2573: %%%   $2u=1+\frac{1}{t-1}u$,
2574:    $u=(t-1)/(2t-3) ~~(1/2 < u \le 2/3 )$,
2575:    $|\Edgepath_1|%%%=\frac{1+3(\frac{t-1}{2t-3}-1)}{(3-2)(\frac{t-1}{2t-3}-1)}
2576:      =(t-3)/(t-2)$,
2577:    $|\Edgepath_2|%%%=|\Edgepath_1|
2578:      =(t-3)/(t-2)$,
2579:    $|\Edgepath_3|%%%=\frac{1+t(\frac{t-1}{2t-3}-1)}{(t-1)(\frac{t-1}{2t-3}-1)}
2580:      =(t-3)/(t-2)$,
2581:    $-\chi/\sharp s %%%=(t-3)/(t-2)+(t-3)/(t-2)+(t-3)/(t-2)+(0-3)+(3-2)\cdot(2t-3)/(t-2)
2582:      =2-2/(t-2)$.
2583:   %%%\\
2584:   If $t=3$, though $-\chi/\sharp s =0$, by Remark \ref{Rem:Condition:ForLinkToBeKnot:2}, we have $-\chi/\sharp s \ge 1$.
2585:   If $t\ge 4$, $-\chi/\sharp s \ge 1$ holds.
2586: \end{enumerate}
2587: 
2588: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2589: % Item 23
2590: %   \Edgepath_1:0-1/2-2/3
2591: %   \Edgepath_2:1-1/2-1/3
2592: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2593: 
2594: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$}.
2595: %
2596: No solutions exist.
2597: 
2598: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2599: % Item 24
2600: %   \Edgepath_1:0-1/3 or 1-1/2-1/3
2601: %   \Edgepath_2:0-1/4
2602: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2603: 
2604: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$ or $\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/4}$}.
2605: %
2606: The check is necessary only for the case that the denominators for the edge $\Edge_3$ is $1$--$5$.
2607: Neither $\angleb{0}$--$\angleb{-1/5}$ nor $\angleb{-1}$--$\angleb{-4/5}$ intersects the sum $\BasicEdgepath_1+\BasicEdgepath_2$.
2608: Thus, no solution exists.
2609: 
2610: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2611: % Item 25
2612: %   \Edgepath_1:0-1/3 or 1-1/2-1/3
2613: %   \Edgepath_2:1-3/4
2614: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2615: 
2616: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$ or $\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{3/4}$}.
2617: \begin{enumerate}
2618:  \item $\Edge_3=\angleb{-1}$--$\angleb{-6/5}$. %%%\\
2619: %%%   $2-4/3 u=1+1/4 u$,
2620:    $u=12/19  ~~(1/2 <12/19 <2/3 )$,
2621:    $|\Edgepath_1|%%%=\frac{1+3(12/19 -1)}{(3-2)(12/19 -1)}
2622:      =2/7$,
2623:    $|\Edgepath_2|%%%=\frac{1+4(12/19 -1)}{(4-1)(12/19 -1)}
2624:      =3/7$,
2625:    $|\Edgepath_3|%%%=\frac{1+5(12/19 -1)}{(5-1)(12/19 -1)}
2626:      =4/7$,
2627:    $-\chi/\sharp s %%%=2/7+3/7+4/7+(0-3)+(3-2)\cdot 19/7
2628:      =1$.
2629: \end{enumerate}
2630: 
2631: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2632: % Item 26
2633: %   \Edgepath_1:0-1/3 or 1-1/2-1/3
2634: %   \Edgepath_2:1-1/2-1/3-1/4
2635: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2636: 
2637: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$ or $\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$--$\angleb{1/4}$}.
2638: 
2639: \begin{enumerate}
2640:  \item $\Edge_3=\angleb{-1}$--$\angleb{-6/5}$. %%%\\
2641: %%%   $2-2u=1+1/4 u$,
2642:    $u=4/9  ~~(0< 4/9 <1/2 )$,
2643:    $|\Edgepath_1|%%%=1+\frac{1+2(4/9 -1)}{(2-1)(4/9 -1)}
2644:      =1+1/5$,
2645:    $|\Edgepath_2|=2+1/5 $,
2646:    $|\Edgepath_3|%%%=\frac{1+5(4/9 -1)}{(5-1)(4/9 -1)}
2647:      =4/5$,
2648:    $-\chi/\sharp s %%%=1+1/5+2+1/5+4/5+(0-3)+(3-2)\cdot 9/5
2649:      =3$.
2650: 
2651:  \item $\Edge_3=\angleb{-1}$--$\angleb{-4/5}$. %%%\\
2652: %%%   $2-2u=1-1/4 u$,
2653:    $u=4/7  ~~(1/2 <4/7 <2/3 )$,
2654:    $|\Edgepath_1|%%%=\frac{1+3(4/7 -1)}{(3-2)(4/7 -1)}
2655:      =2/3$,
2656:    $|\Edgepath_2|=1+2/3 $,
2657:    $|\Edgepath_3|%%%=\frac{1+5(4/7 -1)}{(5-1)(4/7 -1)}
2658:      =2/3$,
2659:    $-\chi/\sharp s %%%=2/3+1+2/3+2/3+(0-3)+(3-2)\cdot 7/3 
2660:      =7/3$.
2661: 
2662: \end{enumerate}
2663: 
2664: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2665: % Item 27
2666: %   \Edgepath_1:0-1/3 or 1-1/2-1/3
2667: %   \Edgepath_2:0-1/2-2/3-3/4
2668: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2669: 
2670: \item {$\BasicEdgepath_1=\angleb{0}$--$\angleb{1/3}$ or $\angleb{1}$--$\angleb{1/2}$--$\angleb{1/3}$, $\BasicEdgepath_2=\angleb{0}$--$\angleb{1/2}$--$\angleb{2/3}$--$\angleb{3/4}$}.
2671: 
2672: \begin{enumerate}
2673:  \item $\Edge_3=\angleb{-1}$--$\angleb{-4/5}$. %%%\\
2674: %%%   $3/2 u=1-1/4 u$,
2675:    $u=4/7  ~~(1/2<4/7<2/3 )$,
2676:    $|\Edgepath_1|%%%=\frac{1+3(4/7 -1)}{(3-1)(4/7 -1)}
2677:      =1/3$,
2678:    $|\Edgepath_2|%%%=1+\frac{1+3(4/7 -1)}{(3-2)(4/7 -1)}
2679:      =1+2/3$,
2680:    $|\Edgepath_3|%%%=\frac{1+5(4/7 -1)}{(5-1)(4/7 -1)}
2681:      =2/3$,
2682:    $-\chi/\sharp s %%%=1/3+1+2/3+2/3+(0-3)+(3-2)\cdot 7/3
2683:      =2$.
2684:    \\ %%% just a dummy.
2685:       %%% for avoiding an error ``! Argument of \@elt has an extra }.''.
2686: \end{enumerate}
2687: 
2688: \end{enumerate}
2689: 
2690: Thus, in most of the cases, $-\chi/\sharp s \ge 1$ is satisfied.
2691: The cases in which only $-\chi/\sharp s \ge (\Denom-1)/\Denom$ is satisfied are 
2692: items \ref{Item:0-1/2,0-1/3,B},
2693: \ref{Item:0-1/2,0-1/3,D},
2694: \ref{Item:0-1/2,1-1/2-1/3,B},
2695: \ref{Item:0-1/2,1-2/3,A},
2696: \ref{Item:0-1/2,1-2/3,C},
2697: \ref{Item:0-1/2,0-1/2-2/3,B} and
2698: \ref{Item:0-1/3,0-1/3,A}.
2699: Last three items are reduced to 
2700: \ref{Item:0-1/2,0-1/3,D},
2701: \ref{Item:0-1/2,1-1/2-1/3,B} and
2702: \ref{Item:0-1/2,1-2/3,A}
2703: respectively.
2704: %
2705: In any of these cases,
2706: the knot is essentially a Montesinos knot $K(-1/2 ,1/3 ,1/t )$ ($t\ge 3$, $t$ is odd).
2707: %
2708: Eventually, the candidate surfaces with $0\le -\chi/\sharp s < 1$ are
2709: \begin{itemize}
2710: \item[(a)] annuli for the torus knots $K(-1/2,1/3,1/3)$ and $K(-1/2,1/3,1/5)$,
2711: \item[(b)] a surface with $1/2 \le -\chi/\sharp s = 1-2/(t-3) < 1$ corresponding to the knot $K(-1/2,1/3,1/t)$ for odd $t\ge 7$,
2712: \item[(c)] a family of surfaces with $0 < -\chi/\sharp s <1$ which corresponds to the non-isolated solutions for $K(-1/2,1/3,1/3)$.
2713: \end{itemize}
2714: Note that surfaces in the family (c) in the above list are compressible, in fact.
2715: 
2716: \end{proof}
2717: 
2718: %\input{draft-detail.tex}
2719: 
2720: 
2721: 
2722: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2723: % Subsection : 
2724: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2725: 
2726: \subsection{Corollaries and best possibility of Theorem \ref{Thm:Denom:UpperBound:Main}}
2727: 
2728: \subsubsection{Proofs of corollaries}
2729: 
2730: Once the theorem is shown,
2731: proof of Corollary \ref{Cor:Denom:UpperBound:Main:ByGenus} is straightforward.
2732: Since the argument does not depend on the orientability,
2733: Corollary \ref{Cor:Denom:UpperBound:Main:ByNonOrientableGenus} is also easily obtained.
2734: 
2735: \begin{proof}[Proof of Corollary \ref{Cor:Denom:UpperBound:Main:ByGenus}]
2736: 
2737: First, assume that $\sharp b\ge 2$.
2738: A boundary slope and its corresponding surface satisfy
2739: at least the inequality $\Denom \le -\chi/\sharp b + 1$ 
2740: in Theorem \ref{Thm:Denom:UpperBound:Main}.
2741: With a variable $g=(2-\chi-\sharp b)/2$,
2742: if $g\ge 1$, 
2743: we have 
2744: \begin{eqnarray}
2745: \Denom&\le& \frac{-2+2g+\sharp b}{\sharp b}+1=\frac{-2+2g}{\sharp b}+2\le g+1
2746: .
2747: \end{eqnarray}
2748: If $g=0$, we have $\Denom\le 2-2/{\sharp b}<2$, which means $\Denom=1$.
2749: $\Denom\le g+1$ is satisfied also in this case.
2750: 
2751: Next, assume that $\sharp b=1$.
2752: Then the inequality $\Denom \le -\chi$ in Theorem \ref{Thm:Denom:UpperBound:Main} is satisfied.
2753: Then,
2754: \begin{eqnarray}
2755: \Denom&\le& -2+2g+\sharp b=2g-1.
2756: \end{eqnarray}
2757: 
2758: By taking maximum of $g+1$ and $2g-1$,
2759: we have $\Denom\le g+1$ for $g=0,1$ and $\Denom\le 2g-1$ for $g\ge 2$.
2760: 
2761: 
2762: In the case of $g=0$, equivalently, the surface is planar, 
2763: the inequality $\Denom \le -\chi$ 
2764: in Theorem \ref{Thm:Denom:UpperBound:Main} cannot be satisfied. 
2765: This means that the cases are exceptional, that is, 
2766: the knot is a torus knot or $K(-1/2,1/3,1/t)$ for odd $t\ge 7$, 
2767: as stated in the last of the proof of Lemma \ref{Lem:LowerBound:RemainingCase}. 
2768: However, in the latter case, 
2769: the surface satisfies $-\chi/\sharp s = 1-2/(t-3) $ and $\sharp s = t-3$. 
2770: Therefore non-torus Montesinos knot have no essential planar surfaces. 
2771: \end{proof}
2772: 
2773: \begin{proof}[Proof of Corollary \ref{Cor:Denom:UpperBound:Main:ByNonOrientableGenus}]
2774: We only have to check for an essential surface with non-orientable genus $h=1$ and $\sharp b\ge 2$.
2775: In this case, since $\Denom\le(-2+h+\sharp b)/\sharp b+1=2-1/\sharp b<2$,
2776: we have $\Denom=1\le h/2+1=3/2$.
2777: 
2778: \end{proof}
2779: 
2780: \subsubsection{The best possibility}
2781: 
2782: The upper bounds in Theorem \ref{Thm:Denom:UpperBound:Main}, Corollary \ref{Cor:Denom:UpperBound:Main:ByGenus} and Corollary \ref{Cor:Denom:UpperBound:Main:ByNonOrientableGenus} are best possible in a sense.
2783: Let $F$ denote a surface.
2784: Note that we do not care the orientability of the surface $F$.
2785: In this part, $g$ denotes $(2-\chi-\sharp b)/2$,
2786: which coincides with genus if the surface is orientable
2787: and with $2h$ where $h$ is non-orientable genus if the surface is non-orientable.
2788: 
2789: First, 
2790: we assume that the candidate surface $F$ corresponds to the edgepath system in the item \ref{Item:0-1/2,1-1/2-1/3,B}
2791: %%%in the subsection \ref{SubSec:ALowerBoundOf-XoverS}
2792: in the previous subsection
2793: for odd $t$.
2794: Since the edgepath $\Edgepath_1$ is constant,
2795: $F$ is incompressible by the Proposition 2.1 in \cite{HO},
2796: and thus, is an essential surface.
2797: %
2798: $-\chi/\sharp s=(\Denom-1)/\Denom$ and $\sharp b=2$ hold,
2799: and give $\Denom=-\chi/\sharp b +1$ and $\Denom=g+1$. 
2800: %%%  \begin{eqnarray*}
2801: %%%   $\sharp s=t-3$,
2802: %%%   $\Denom=(t-3)/2$,
2803: %%%   -\chi&=&\frac{\Denom-1}{\Denom}\cdot\sharp s=t-5 \\
2804: %%%   \chi&=&5-t \\
2805: %%%   2-2g-\sharp b&=&5-t \\
2806: %%%   2g&=&t-5 \\
2807: %%%   g&=&\frac{t-5}{2}=\frac{t-3}{2}-1=\Denom-1 \\
2808: %%%  \end{eqnarray*}
2809: %
2810: Since $\Denom=(t-3)/2$,
2811: the value of $g=(t-5)/2$ for $t=5,7,9,\cdots$ is $0,1,2,\cdots$.
2812: This indicates that
2813: if $\sharp b\ge2$ are satisfied,
2814: $\Denom\le -\chi/\sharp b +1$ and $\Denom\le g+1$ are best possible for arbitrary non-negative integer $g$.
2815: 
2816: %%
2817: 
2818: Next,
2819: let $F$ be a candidate surface in the item \ref{Item:0-1/2,0-1/2-2/3,A}
2820: %%%in the subsection \ref{SubSec:ALowerBoundOf-XoverS} 
2821: %%%in the previous subsection
2822: for odd $t$.
2823: %%%Since the type I edgepath system $\EdgepathSystem$ has the cycle of final r-values $(1,1,t-1)$ and their sign are all negative,
2824: It is incompressible by the Proposition 2.6 in \cite{HO}.
2825: %%%  \begin{eqnarray*}
2826: Since $|\Twist|%%%=2(1/t+1+1/t+1-1/t)
2827: =4+2/t$, we have $\Denom=t$.
2828: $\sharp s=\mathrm{lcm}(t,t,t)=t$ gives $\sharp b=\sharp s/\Denom=1$.
2829: Thus $-\chi/\sharp s=1$ gives $\Denom=-\chi/\sharp b$ and $\Denom=2g-1$.
2830: %%%   -\chi&=&\sharp s=t \\
2831: %%%   \chi&=&-t \\
2832: %%%   2-2g-\sharp b&=&-t \\
2833: %%%   t&=&2g-1 \\
2834: %%%  \end{eqnarray*}
2835: %
2836: %%% Since $\Denom=t=2g-1$,
2837: The value of $g=(t+1)/2$ for $t=3,5,7,\cdots$ is $2,3,4,\cdots$.
2838: This indicates that
2839: if $\sharp b=1$ is satisfied,
2840: $\Denom\le -\chi/\sharp b+1$ and $\Denom\le 2g-1$ are best possible for arbitrary integer $g\ge 2$.
2841: %
2842: 
2843: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2844: %
2845: % Section : A bound on the difference
2846: %
2847: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2848: 
2849: \section{A bound on the difference}
2850: \label{Sec:ABoundOnTheDifference}
2851: 
2852: The purpose of this section is to prove Theorem \ref{Thm:Diff:UpperBound:Main},
2853: which claims an upper bound of the difference of two boundary slopes.
2854: A large part of this section is the proof of a technical lemma,
2855: which is used for proving the theorem.
2856: The best possibility and some corollaries are also given.
2857: 
2858: We begin with several remarks on these results. 
2859: 
2860: \begin{enumerate}
2861: 
2862: \item
2863: From the argument, 
2864: we must exclude the meridional boundary slope, 
2865: for it corresponds to ``infinity" numerical boundary slope. 
2866: Note that it can actually appear if $\NumTangles\ge4$. 
2867: See \cite{O} for a detail. 
2868: 
2869: \item
2870: There is an apparent lower bound $|\Slope_1-\Slope_2|\ge 0$.
2871: The lower and upper bounds meet at 
2872: $|\Slope_1-\Slope_2|=0$ and $(-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2)=-2$. 
2873: This corresponds to the boundary slope of the incompressible disk in 
2874: the trivial knot exterior.
2875: 
2876: \item
2877: The upper bound  (\ref{Eq:Diff:UpperBound:Main}) is sharp: 
2878: there is an infinite sequence of Montesinos knots 
2879: each of whose exterior includes two essential surfaces with 
2880: boundary slopes satisfying the equality. 
2881: %%%See Appendix B for more detail. 
2882: See Subsection \ref{Subsec:Diff:ProofEtc} for more detail.
2883: 
2884: \item
2885: No such a ``linear" upper bound can hold for $\Delta$. 
2886: See Subsection \ref{SubSec:LinearBoundOfDist} for example. 
2887: In fact known bounds on $\Delta$ are quadratic with respect to 
2888: $-\chi_i/\sharp s_i$. 
2889: 
2890: \item
2891: In \cite{I}, the first author obtains the same upper bounds for 
2892: 2-bridge knot exterior and Seifert fibered manifolds 
2893: which include torus knot exteriors. 
2894: Therefore, the upper bound (\ref{Eq:Diff:UpperBound:Main}) may be 
2895: applicable for some wider class of knot exteriors or manifolds.
2896: 
2897: \end{enumerate}
2898: 
2899: 
2900: 
2901: 
2902: %%\subsection{Lemma \ref{Lem:Remainder:UpperBound}}
2903: 
2904: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2905: % Subsection : 
2906: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2907: 
2908: \subsection{Linear bound of the distance}
2909: %$\Delta$
2910: \label{SubSec:LinearBoundOfDist}
2911: 
2912: Any linear bounds of the distance of two boundary slopes are impossible.
2913: This is a reason why we consider an upper bound of the difference
2914: rather than of the distance.
2915: 
2916: We give a concrete example of a pair of boundary slopes,
2917: which make any linear bounds impossible.
2918: The example is two boundary slopes $\Slope_1$ and $\Slope_2$ of the Montesinos knot $K(-1/2,1/3,1/t)$ for odd $t\ge 7$.
2919: 
2920: $\Slope_1$ is a boundary slope which appears in
2921: \ref{Item:0-1/2,1-1/2-1/3,B}
2922: in Subsection \ref{SubSec:ALowerBoundOf-XoverS}.
2923: %
2924: It corresponds to a type I edgepath system $\EdgepathSystem_1$
2925: \[
2926: \left\{
2927: \begin{array}{l}
2928: \Edgepath_{1,1}=((t-3)/(t-2))\cdot \angleb{-1/2}+(1/(t-2))\cdot \circleb{-1/2} \\
2929: \Edgepath_{1,2}=(((t-5)/(t-3))\cdot \angleb{1/2}+(2/(t-3))\cdot \angleb{1/3})\textrm{\,--\,}\angleb{1/3} \\
2930: \Edgepath_{1,3}=(((t-4)/(t-3))\cdot \angleb{0}+(1/(t-3))\cdot \angleb{1/t})\textrm{\,--\,}\angleb{1/t}.
2931: \end{array}
2932: \right.
2933: \]
2934: For the edgepath system,
2935: %%% $u_1=(t-1)/\{2(t-2)\}$,
2936: $\Twist_1=2/(t-3)$,
2937: $-\chi_1/\sharp s_1=1-2/(t-3)$,
2938: $\Denom_1=(t-3)/2$,
2939: $-\chi_1/\sharp b_1=(t-3)/2-1$.
2940: 
2941: On the other hand,
2942: the second slope $\Slope_2$ corresponds to a type III edgepath system $\EdgepathSystem_2$
2943: \[
2944: \left\{
2945: \begin{array}{l}
2946: \Edgepath_{2,1}=\angleb{\infty}\textrm{\,--\,}\angleb{0}\textrm{\,--\,}\angleb{-1/2} \\
2947: \Edgepath_{2,2}=\angleb{\infty}\textrm{\,--\,}\angleb{1}\textrm{\,--\,}\angleb{1/2}\textrm{\,--\,}\angleb{1/3} \\
2948: \Edgepath_{2,3}=\angleb{\infty}\textrm{\,--\,}\angleb{1}\textrm{\,--\,}\angleb{1/2}\textrm{\,--\,}\cdots\textrm{\,--\,}\angleb{1/(t-1)}\textrm{\,--\,}\angleb{1/t}.
2949: \end{array}
2950: \right.
2951: \]
2952: For the edgepath system,
2953: $\Twist_2=-2(t+2)$,
2954: $\Denom_2=1$,
2955: $-\chi_2/\sharp s_2=-\chi_2/\sharp b_2=t+2$.
2956: 
2957: Since
2958: we have both
2959: $(-\chi_1/\sharp b_1)+(-\chi_2/\sharp b_2)=(t-3)/2-1+t+2=(3t-1)/2$
2960: and
2961: $\Delta(\Slope_1,\Slope_2)=\Denom_1 \Denom_2 |\Slope_1-\Slope_2|=\Denom_1 \Denom_2 |\Twist_1-\Twist_2| =t^2-t-5$
2962: at the same time,
2963: %
2964: $\Delta$ cannot be bounded by an inequality
2965: \[
2966: \Delta(\Slope_1,\Slope_2)\le X(\frac{-\chi_1}{\sharp b_1}+\frac{-\chi_2}{\sharp b_2})+Y
2967: \]
2968: for any constant $X$ and $Y$.
2969: 
2970: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2971: % Subsection : 
2972: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2973: 
2974: \subsection{An upper bound of the sum of remainder terms}
2975: 
2976: In this subsection,
2977: we state and prove Lemma \ref{Lem:Remainder:UpperBound},
2978: which claims an upper bound of the sum of ``remainder terms'' of two boundary slopes
2979: and is the key to proving Theorem \ref{Thm:Diff:UpperBound:Main}.
2980: 
2981: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2982: % Paragraph : Remainder term
2983: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2984: 
2985: 
2986: By definition, 
2987: the twist $\Twist$ is roughly twice of a sum of signed lengths of edges.
2988: On the other hand,
2989: as we see in Section \ref{Sec:Preparation},
2990: the major part of $-\chi/\sharp s$ is the sum of lengths of the edgepaths in the edgepath system.
2991: Hence, by the triangle inequality,
2992: these facts imply a bound of the twist
2993: by an inequality like $|\Twist|\le 2\,(-\chi/\sharp s)+\alpha$.
2994: Then,
2995: we introduce the {\em remainder term}
2996: $\Remainder(F)
2997: \equiv
2998: |\Twist|-2\,(-\chi/\sharp s)
2999: $. 
3000: With the remainder term, the key lemma is described as follows.
3001: 
3002: \begin{lemma}
3003: \label{Lem:Remainder:UpperBound}
3004: For a Montesinos knot $K$,
3005: after simplification,
3006: distinct two candidate surfaces $F_1$ and $F_2$ satisfy
3007: \[
3008: \Remainder_1+\Remainder_2\le 4,
3009: \]
3010: where $\Remainder_i=\Remainder(F_i)$.
3011: 
3012: In fact, the set of candidate surfaces satisfies following conditions.
3013: \begin{itemize}
3014: \item
3015: For a type I surface $F$, we have $\Remainder\le 4$.
3016: Furthermore, 
3017: there is at most one type I surface with $0<\Remainder\le 4$,
3018: and any other type I surface $F$ satisfies $\Remainder\le 0$.
3019: \item
3020: For a type II surface $F$, we have $\Remainder\le 4$.
3021: Furthermore,
3022: there is at most one type II surface with $0<\Remainder\le 4$,
3023: and any other type II surface $F$ satisfies $\Remainder\le 0$.
3024: \item
3025: For a type III surface $F$, we have $\Remainder\le 0$.
3026: \item
3027: There is at most one surface with $0<\Remainder\le 4$.
3028: Namely, the type I surface $F_1$ with $\Remainder_1>0$ and 
3029:  the type II surface $F_2$ with $\Remainder_2>0$ 
3030:  do not exist for a Montesinos knot $K$ at the same time.
3031: \end{itemize}
3032: \end{lemma}
3033: 
3034: The simplification is mentioned in Section \ref{Sec:Preparation}.
3035: We divide Lemma \ref{Lem:Remainder:UpperBound} into some partial claims,
3036: and prove the lemma in the rest of this subsection.
3037: We first introduce two notions 
3038: which are used in the proof of Lemma \ref{Lem:Remainder:UpperBound}. 
3039: 
3040: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3041: % Paragraph : Cancellation
3042: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3043: 
3044: \subsection*{Cancellation}
3045: 
3046: On summation in (\ref{Eq:Formula:Twist}),
3047: opposite signs of $\sigma(\Edge)$'s for two or more edges cause cancellation.
3048: If such a cancellation occurs,
3049: we call an edgepath system $\EdgepathSystem$ an {\em edgepath system with cancellation},
3050: and the corresponding surface a {\em surface with cancellation}.
3051: %%% It is important in the proof to distinguish between a surface without cancellation and one with cancellation.
3052: %
3053: A constant edgepath does not cause cancellation.
3054: %
3055: Only a surface $F$ without cancellation will be able to have $\Remainder>0$ in Lemma \ref{Lem:Remainder:UpperBound}. 
3056: 
3057: For an edgepath system,
3058: we collect all non-$\infty$-edges of every non-constant edgepath,
3059: divide them into two classes
3060: according to the sign $\sigma(\Edge_{i,j})$ of an edge $\Edge_{i,j}$,
3061: and then sum up the lengths of edges for each class.
3062: With their total lengths $l_{+}$ and $l_{-}$,
3063: let $\Cancel(F)$ denote $\min (l_{+}, l_{-})$.
3064: $\Cancel$ means the amount of the cancellation in calculating the twist for a surface.
3065: %%% where positive sign edges of total length $\Cancel$
3066: %%% and negative sign edges of total length $\Cancel$ cancel each other.
3067: With the variable $\Cancel$, the twist is related to the total length of edgepaths in $\Strip$ 
3068: as
3069: %
3070: \begin{eqnarray}
3071:  |\Twist|
3072: %%% &=&
3073: %%%  \left|
3074: %%%  \sum_{\Edgepath_i \in \NonconstantEdgepaths}
3075: %%%   \left(
3076: %%%    \sum_{\Edge_{i,j} \in \Edgepath_{i}} 
3077: %%%     \left\{
3078: %%%      \begin{array}{l}
3079: %%%       0 \\
3080: %%%       ~~~~~(\textrm{  if $\Edge_{i,j}$ is an $\infty$-edge.} ) \\
3081: %%%       - 2\,\sigma(\Edge_{i,j})|\Edge_{i,j}| \\
3082: %%%       ~~~~~(\textrm{  otherwise. } )\\
3083: %%%      \end{array}
3084: %%%     \right.
3085: %%%   \right)
3086: %%%  \right|
3087: %%% \nonumber
3088: %%% \\
3089: &=& 2\,|l_{+}-l_{-}| = 2\,\{l_{+}+l_{-} -2 \min(l_{+},l_{-}) \}
3090: \nonumber
3091: \\
3092: &=&
3093: 2
3094:  \left(
3095:  \sum_{\Edgepath_i \in \NonconstantEdgepaths}
3096:   |\Edgepath_{i,\ge 0} |
3097:  \right)
3098:  -4 \cdot \Cancel
3099: .
3100: %%% ,
3101: \nonumber
3102: %%% \\
3103: \end{eqnarray}
3104: %%% where the total length of edgepaths is the major part of $-\chi/\sharp s$.
3105: 
3106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3107: % Paragraph : Monotonic edgepath systems
3108: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3109: 
3110: \subsection*{Monotonic edgepath systems}
3111: 
3112: For an edgepath system, 
3113: the edgepath system is said to be {\em monotonically increasing} (resp. {\em decreasing})
3114: if the $v$-coordinates are monotonically increasing (resp. decreasing) for all edgepaths in the edgepath system.
3115: A surface without cancellation corresponds to a monotonic edgepath system.
3116: 
3117: Since each vertex $\angleb{p/q}$ ($q>1$) of the diagram $\Diagram$ has two leftward edges, one is increasing and the other is decreasing, there exist only one monotonically increasing basic edgepath and one monotonically decreasing basic edgepath.
3118: 
3119: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3120: % Subsection : Type II and Type III surfaces
3121: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3122: 
3123: \subsubsection{Type II and type III surfaces}
3124: 
3125: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3126: % Paragraph : Type III surfaces
3127: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3128: 
3129: %%%\paragraph{Type III surfaces}
3130: 
3131: The situation is simplest for type III surfaces among all types of surfaces. 
3132: Even for type II surfaces, the argument is not so complicated.
3133: 
3134: \begin{lemma}
3135: After simplification,
3136: for any type III surface,
3137: the inequality $\Remainder\le 0$ holds.
3138: \end{lemma}
3139: %
3140: \begin{proof}
3141: (\ref{Eq:Formula:Twist}) and (\ref{Eq:Formula:EulerCharTypeIII}) give
3142: $
3143: |\Twist|
3144: \le
3145: 2\cdot\sum_{i=1}^{\NumTangles}
3146: |\Edgepath_{i,>0}|
3147: =2\,(-\chi/\sharp s)
3148: .
3149: $
3150: Hence, $\Remainder=|\Twist|-2\,(-\chi/\sharp s) \le 0$.
3151: \end{proof}
3152: 
3153: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3154: % Paragraph : Type II surfaces
3155: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3156: 
3157: %%%\subsubsection{Type II surfaces}
3158: %%%\paragraph{Type II surfaces}
3159: 
3160: \begin{lemma}
3161: After simplification,
3162: for any type II surface $F$,
3163: the inequality $\Remainder\le 4$ holds.
3164: Moreover, there exists at most one surface without cancellation, for which $0<\Remainder\le 4$,
3165: while any surface with cancellation satisfies $\Remainder\le 0$.
3166: \end{lemma}
3167: %
3168: \begin{proof}
3169: (\ref{Eq:Formula:Twist}) and (\ref{Eq:Formula:EulerCharTypeII}) give
3170: %
3171: \begin{eqnarray}
3172: |\Twist|&\le &
3173: 2(\sum_{i=1}^{\NumTangles}|\Edgepath_{i,>0} | )
3174:  +2\,|\EdgepathSystem(+0) |
3175: =2(-\chi/\sharp s+2)
3176: .
3177:  \label{Eq:Twist:UpperBound:TypeII}
3178: \end{eqnarray}
3179: %
3180: Hence, 
3181: \begin{eqnarray}
3182: \Remainder&=&|\Twist|-2\,(-\chi)/\sharp s \le 4.
3183: \label{Eq:Remainder:UpperBound:TypeII}
3184: \end{eqnarray}
3185: 
3186: When cancellation occurs,
3187: the difference between the both sides of the inequality (\ref{Eq:Twist:UpperBound:TypeII})
3188: increases by $2\,(+1-(-1))=4$ at a pair of complete edges causing cancellation.
3189: Thus,
3190: $
3191: \Remainder=|\Twist|-2\,(-\chi/\sharp s)\le 0
3192: $.
3193: 
3194: If the equality in (\ref{Eq:Remainder:UpperBound:TypeII}) holds for an edgepath system,
3195: the edgepath system satisfies either of:
3196: \begin{itemize}
3197: \item
3198: the corresponding basic edgepath system is monotonically decreasing and $\EdgepathSystem(+0)\ge 0$,
3199: \item
3200: the corresponding basic edgepath system is monotonically increasing and $\EdgepathSystem(+0)\le 0$.
3201: \end{itemize}
3202: Both types of edgepath systems are not obtained simultaneously for a Montesinos knot $K$.
3203: %
3204: By the uniqueness of the monotonically increasing or decreasing basic edgepath system, there is at most one type II surface with $0<\Remainder\le 4$.
3205: \end{proof}
3206: 
3207: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3208: % Subsubsection : Type I surfaces
3209: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3210: 
3211: \subsubsection{Type I surfaces}
3212: 
3213: For type I surfaces, the argument is more complicated than for type II and type III surfaces.
3214: Thus, we here introduce two inequalities for type I surfaces.
3215: 
3216: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3217: % Paragraph : Inequality
3218: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3219: 
3220: \subsection*{An inequality for type I surfaces}
3221: 
3222: \begin{lemma}
3223: For a type I surface $F$,
3224: its remainder term $\Remainder$ is upper-bounded as
3225: \begin{eqnarray}
3226: \Remainder
3227: &\le&
3228: 2(\NumTangles-\NumTangles_{\mathrm{const}})-\left( \NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths}\frac{1}{q_i} \right)\frac{2}{1-u}
3229: .
3230: \label{Eq:Remainder:UpperBound:TypeI:General}
3231: \end{eqnarray}
3232: \end{lemma}
3233: %
3234: \begin{proof}
3235: First, an edgepath system for type I surface does not include any vertical edge or $\infty$-edge. 
3236: (\ref{Eq:Formula:Twist}) gives
3237: \begin{eqnarray*}
3238: \Twist&=& 
3239: 2 \sum_{i=1}^{\NumTangles}
3240:  \left(
3241:   \left\{
3242:    \begin{array}{l}
3243:     0 \\
3244:     ~~~~~(\textrm{ if $\Edgepath_i$ is constant } ) \\
3245:     \sum_{\Edge_{i,j} \in \Edgepath_{i}} 
3246:      -~\sigma(\Edge_{i,j})~|\Edge_{i,j}| \\
3247:     ~~~~~(\textrm{ if $\Edgepath_i$ is non-constant })
3248:    \end{array}
3249:   \right.
3250:  \right)
3251: .
3252: \end{eqnarray*}
3253: 
3254: By (\ref{Eq:Formula:EulerCharTypeI}),
3255: \begin{eqnarray}
3256: \Remainder
3257: &=&
3258: |\Twist|-2\frac{-\chi}{\sharp s}
3259: %%% \nonumber
3260: \label{Eq:Remainder:UpperBound:TypeI:WithCancel}
3261: \\
3262: %%% &=&
3263: %%% \left[~
3264: %%%  \left|
3265: %%%  2 \sum_{\Edgepath_i \in \EdgepathSystem}
3266: %%%   \left(
3267: %%%    \left\{
3268: %%%     \begin{array}{l}
3269: %%%      0 \\
3270: %%%      ~~~~~(\textrm{ if $\Edgepath_i$ is constant } ) \\
3271: %%%      \sum_{\Edge_{i,j} \in \Edgepath_{i}} 
3272: %%%       - \sigma(\Edge_{i,j})|\Edge_{i,j}| \\
3273: %%%      ~~~~~(\textrm{ otherwise })
3274: %%%     \end{array}
3275: %%%    \right.
3276: %%%   \right)
3277: %%%  \right|
3278: %%%  \right.
3279: %%% \nonumber
3280: %%% \\
3281: %%% &&
3282: %%% ~~
3283: %%% \left.
3284: %%% -2
3285: %%%  \sum_{\Edgepath_i \in \EdgepathSystem}
3286: %%%   \left(
3287: %%%    \left\{
3288: %%%     \begin{array}{l}
3289: %%%      0 \\
3290: %%%      ~~~~~(\textrm{  if $\Edgepath_i$ is constant }) \\
3291: %%%      |\Edgepath_{i} |\\
3292: %%%      ~~~~~(\textrm{ otherwise }) \\
3293: %%%     \end{array}
3294: %%%    \right.
3295: %%%   \right)
3296: %%% \right]
3297: %%% \nonumber
3298: %%% \\
3299: %%% &&
3300: %%% -2
3301: %%% \left[
3302: %%%  \NumTangles_{\mathrm{const}}-\NumTangles
3303: %%%  +\left(
3304: %%%  \NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths}\frac{1}{q_i}
3305: %%%  \right)
3306: %%%  \frac{1}{1-u}
3307: %%% \right]
3308: %%% \nonumber
3309: %%% \\
3310: &=&
3311:  -4 \cdot \Cancel
3312:  +2(\NumTangles-\NumTangles_{\mathrm{const}})
3313:  -\left(
3314:  \NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths}\frac{1}{q_i}
3315:  \right)
3316:  \frac{2}{1-u}
3317: .
3318: \nonumber
3319: %%% \\
3320: %%% &&
3321: \end{eqnarray}
3322: %
3323: Even if we ignore the effect of cancellation $\Cancel$, 
3324: we have the upper bound in the statement.
3325: %
3326: \end{proof}
3327: 
3328: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3329: % Paragraph : Inequality for type I surfaces with cancellation
3330: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3331: 
3332: \subsection*{An inequality for type I surfaces with cancellation}
3333: 
3334: Under the assumption that we could prove $\Remainder\le 4$,
3335: we think about the effect of cancellation.
3336: If $\Cancel \ge 1$,
3337: since cancellation works on the twist by $-4\Cancel$,
3338: $\Remainder\le 0$ immediately follows.
3339: The case of $\Cancel < 1$ only remains. 
3340: %
3341: In the formula (\ref{Eq:Remainder:UpperBound:TypeI:WithCancel}),
3342: $\Cancel$ and the term $2/(1-u)$ are in the trade-off relationship.
3343: %%% As $u$ decreases, $\Cancel$ increases and makes $\Remainder$ decrease.
3344: %%% Though, at the same time,
3345: %%% $-\frac{2}{1-u}$ makes $\Remainder$ increase.
3346: %%% Contrary,
3347: %%% as $u$ increases, $\Cancel$ decreases and makes $\Remainder$ increase,
3348: %%% while $-\frac{2}{1-u}$ makes $\Remainder$ decrease.
3349: %%% Note that the coefficient of
3350: %%% $\frac{2}{1-u}$ will be always positive.
3351: We examine the variation of $\Remainder$ in detail
3352: and make an inequality about $\Remainder$ for an edgepath system with cancellation.
3353: 
3354: \begin{lemma}
3355: Let $F$ be a type I surface $F$ with cancellation.
3356: Assume that a partial edge of an edge $\angleb{p/q}$\,--\,$\angleb{r/s}$ ($|ps-qr|=1$, $q<s$) causes cancellation.
3357: %% Then, the common $u$-coordinate of the endpoints of edgepaths satisfies $(q-1)/q<u<(s-1)/s$.
3358: Then, the remainder term $\Remainder$ is upper-bounded as
3359: \begin{eqnarray}
3360: \Remainder
3361: &\le&
3362:  2(\NumTangles-\NumTangles_{\mathrm{const}}) -2x \min \{s,q+2/x\}
3363: \label{Eq:Remainder:UpperBound:TypeI:WithSingleCancel}
3364: ,
3365: \end{eqnarray}
3366: where $x=\NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths} (1/q_i)>0$.
3367: \end{lemma}
3368: %
3369: \begin{proof}
3370: Assume that the length of the partial edge included in $\angleb{p/q}$\,--\,$\angleb{r/s}$ is $k/m$, 
3371: and the edge causes cancellation.
3372: %
3373: We start from (\ref{Eq:Remainder:UpperBound:TypeI:WithCancel}), that is,
3374: $\Remainder
3375: =
3376:  -4 \cdot \Cancel
3377:  +2(\NumTangles-\NumTangles_{\mathrm{const}})
3378:  -x \cdot 2/(1-u)
3379: \le
3380:  -4\, k/m
3381:  +2(\NumTangles-\NumTangles_{\mathrm{const}})
3382:  -x \cdot 2/(1-u)
3383: $%
3384: .
3385: %
3386: Since the length $k/m$ is given by
3387: $k/m=\{1+s(u-1)\}/\{(s-q)(u-1)\}$ as the formula (\ref{Eq:Formula:LengthOfPartialEdge}),  
3388: the above inequality can be deformed into
3389: $
3390: \Remainder
3391: %%% &=&
3392: %%% -4\frac{k}{m}-x\frac{2}{1-u}
3393: %%% \\
3394: %%% &=&
3395: %%% -4\frac{1+s(u-1)}{(s-q)(u-1)}-x\frac{2}{1-u}
3396: %%% \\
3397: \le
3398:  2(\NumTangles-\NumTangles_{\mathrm{const}})
3399:  -4 s/(s-q)
3400:  +\{2/(s-q)-x\}\cdot 2/(1-u)
3401: $.
3402: %
3403: %%% If $(q<)s<q+\frac{2}{x}$, since $f$ is monotonically increasing,
3404: %%% $f(u)\le -2xs$ and $f$ is maximum at $u=\frac{s-1}{s}$.
3405: %%% If $s=q+\frac{2}{x}$, $f(u)$ is constant, and $f(u)\le -4\frac{s}{s-q}=-2xs=-2(xq+2)$.
3406: %%% If $q+\frac{2}{x}<s$, since $f$ is monotonically decreasing,
3407: %%% $f(u)\le -2(xq+2)$ and $f$ is maximum at $u=\frac{q-1}{q}$.
3408: The right-hand side is monotonically increasing, constant, or monotonically decreasing as a function of $u$,
3409: according to the sign of $(2/(s-q)-x)$.
3410: Thus, $\Remainder$ can be upper-bounded as in the statement.
3411: %%% $\Remainder
3412: %%% \le
3413: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})+\max \{-2xs,-2(xq+2)\}
3414: %%% =
3415: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})-2x \min \{s,q+\frac{2}{x}\}
3416: %%% .
3417: %%% $
3418: %%% %
3419: %%% By this inequality and the formula (\ref{Eq:Remainder:UpperBound:TypeI:WithCancel}),
3420: %%% with
3421: %%% $x=\NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths}\frac{1}{q_i}$
3422: %%% ,
3423: %
3424: \end{proof}
3425: 
3426: Now we show the following.
3427: \begin{lemma}
3428: After the simplification,
3429: for all type I surfaces $F$,
3430: the inequality $\Remainder\le 4$ holds.
3431: Moreover, there exists at most one surface without cancellation, for which $0< \Remainder\le 4$,
3432: while any surface with cancellation satisfies $\Remainder\le 0$.
3433: \end{lemma}
3434: 
3435: \begin{proof}
3436: 
3437: First, we here examine type I surfaces without cancellation.
3438: For such a surface, the edgepath system is monotonically increasing or decreasing.
3439: According to the sign of the sum of the tangles $K_i$ as fractions,
3440: only one of the two monotonic basic edgepath systems has a solution of the equation (\ref{Eq:EquationForEdgepathSystem}), that is, $\sum_{i=1}^{\NumTangles} \widetilde{\BasicEdgepath}(u)=0$.
3441: Thus, there exists at most one type I surface without cancellation.
3442: 
3443: For a type I surface,
3444: since the situation is complicated,
3445: we separate the arguments for $\NumTangles=3$ and $\NumTangles\ge4$.
3446: Furthermore, if $\NumTangles=3$,
3447: we check the lemma case by case 
3448: according to the number of the constant edgepaths.
3449: 
3450: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3451: % Paragraph : Type I surface with $\NumTangles\ge4$
3452: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3453: 
3454: \medskip
3455: %\paragraph{Type I surface with $\NumTangles\ge4$}
3456: \noindent{(1)
3457: %% Type I surface with $\NumTangles\ge4$
3458: %% Type I surfaces for a Montesinos knot with $\NumTangles\ge4$ tangles
3459: $\NumTangles\ge4$.
3460: }
3461: 
3462: %%% ### General ### %%%
3463: %
3464: %%% The statement is relatively easy to prove for a type I surface with $\NumTangles\ge 4$ in comparison with a type I surface with $\NumTangles=3$.
3465: 
3466: %
3467: First,
3468: without considering the effect of cancellation,
3469: the inequality (\ref{Eq:Remainder:UpperBound:TypeI:General}) is
3470: %
3471: $\Remainder
3472: %%% &\le&
3473: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})-\{ \NumTangles-2-\sum_{\Edgepath_i \in \ConstantEdgepaths}\frac{1}{q_i} \}\frac{2}{1-u}
3474: %%% \\
3475: \le
3476: 2(\NumTangles-\NumTangles_{\mathrm{const}})-\{ \NumTangles-2-\NumTangles_{\mathrm{const}}(1/2) \}\cdot 2/(1-u)
3477: $.
3478: %
3479: By watching at $u=0$,
3480: we have
3481: $
3482: \Remainder
3483: \le
3484: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})-\{ \NumTangles-2-\NumTangles_{\mathrm{const}}\frac{1}{2} \}\cdot 2
3485: %%% \\
3486: %%% &=&
3487: 4-\NumTangles_{\mathrm{const}}
3488: \le
3489: 4
3490: $.
3491: %
3492: %%% ### With Cancellation ### %%%
3493: %
3494: For a surface $F$ with cancellation,
3495: even if $\Cancel<1$,
3496: by (\ref{Eq:Remainder:UpperBound:TypeI:WithSingleCancel}) for $x=\NumTangles-2-\NumTangles_{\mathrm{const}}\cdot 1/2$, 
3497: $q\ge1$ and $s\ge2$,
3498: we obtain
3499: $
3500: \Remainder
3501: \le
3502: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})
3503: %%% -2x\min\{s,q+\frac{2}{x} \}
3504: %%% \\
3505: %%% &=&
3506: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})
3507: %%% +\max\{-2xs,-2xq-4 \}
3508: %%% \\
3509: %%% &\le&
3510: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})
3511: %%% +\max\{-4x,-2x-4 \}
3512: %%% \\
3513: %%% &=&
3514: %%% \max\{2(\NumTangles-\NumTangles_{\mathrm{const}})-4x,2(\NumTangles-\NumTangles_{\mathrm{const}})-2x-4 \}
3515: %%% \\
3516: %%% &=&
3517: %%% \max\{
3518: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})-4(\NumTangles-2-\NumTangles_{\mathrm{const}}\frac{1}{2}),
3519: %%% 2(\NumTangles-\NumTangles_{\mathrm{const}})-2(\NumTangles-2-\NumTangles_{\mathrm{const}}\frac{1}{2})-4 \}
3520: %%% \\
3521: %%% &=&
3522: \max\{
3523: -2\NumTangles+8,-\NumTangles_{\mathrm{const}}
3524: \}
3525: \le
3526: 0
3527: $.
3528: 
3529: \medskip
3530: 
3531: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3532: % Type I surface with $\NumTangles=3$
3533: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3534: 
3535: \noindent {(2)
3536: %% Type I surface with $\NumTangles=3$
3537: %% Type I surfaces for a Montesinos knot with $\NumTangles=3$ tangles
3538: $\NumTangles=3$.
3539: }
3540: 
3541: First, we introduce a notation.
3542: %
3543: We represent the complete non-horizontal edges including a partial edge $\Edge_i$ of $\Edgepath_i$ by $\angleb{p_i/q_i}$\,--\,$\angleb{r_i/s_i}$,
3544: and the denominators of the $v$-coordinates of constant edgepaths $\Edgepath_a$ and $\Edgepath_b$ by $q_a$ and $q_b$.
3545: %
3546: Though the indices of non-constant edgepaths may not be successive and be something like $\Edgepath_{1}$ and $\Edgepath_{3}$,
3547: we replace the indices so that the non-constant edgepaths have successive indices like $\Edgepath_{1}$ and $\Edgepath_{2}$,
3548: for ease in the argument.
3549: If cancellation occurs,
3550: $\Edgepath_1$ denotes an edgepath whose partial edge causes cancellation.
3551: 
3552: \medskip
3553: 
3554: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3555: % Paragraph : Type I surface with $\NumTangles=3$ and no constant edgepath
3556: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3557: 
3558: %\paragraph{Type I surface with $\NumTangles=3$ and no constant edgepath}
3559: \noindent {(2-1)
3560: %% Type I surface with $\NumTangles=3$ and no constant edgepath
3561: %% The edgepath system of the Type I surface includes no constant edgepath
3562: $\NumTangles=3$, $\NumTangles_{\mathrm{const}}=0$.
3563: }
3564: 
3565: %
3566: %%% ### General ### %%%
3567: %
3568: For an edgepath system with no constant edgepath, the formula (\ref{Eq:Remainder:UpperBound:TypeI:General})
3569: is
3570: %
3571: $\Remainder\le 6-2/(1-u)$.
3572: %
3573: Since $0\le u <1$, we have
3574: %
3575: $\Remainder\le 4$.
3576: %
3577: %%% ($\Remainder$ is maximum at $u=0$). 
3578: This is sufficient for edgepath systems without cancellation.
3579: %
3580: %%% ### With Cancellation ### %%%
3581: %
3582: For an edgepath system with cancellation,
3583: the inequality (\ref{Eq:Remainder:UpperBound:TypeI:WithSingleCancel}) is simplified into
3584: $\Remainder\le 6-2 \min \{s_1,q_1+2\}$.
3585: If $s_1=2$, then $\Remainder\le2$. Otherwise, $\Remainder\le 0$. 
3586: %
3587: In the case of $s_1=2$, there are two possibilities:
3588: (2-1-1) final edges of all edgepaths have common sign,
3589: (2-1-2) final edges of edgepaths have both positive and negative sign.
3590: In the former case,
3591: except the case $s_2=s_3=2$,
3592: by applying the edgepath system to an inequality similar to (\ref{Eq:Remainder:UpperBound:TypeI:WithSingleCancel}),
3593: we have $\Remainder\le 0$.
3594: For the case $s_1=s_2=s_3=2$,
3595: by solving the equation $\sum_{i=1}^{\NumTangles} \widetilde{\BasicEdgepath_i}(u)=0$,
3596: $\Cancel$ is $1$ or greater, and thus $\Remainder\le 0$.
3597: In the latter case,
3598: the solutions of the equation are non-isolated
3599: and their representative is an edgepath system
3600: which causes no cancellation on $\angleb{p/q}$\,--\,$\angleb{r/s}$.
3601: Thus, we have verified the claim for the remaining case of $s_1=2$ such that there is no such a surface $F$ with cancellation and the remainder term $0<\Remainder\le2$.
3602: %
3603: 
3604: \medskip
3605: 
3606: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3607: % Paragraph : Type I surface with $\NumTangles=3$ and one constant edgepath
3608: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3609: 
3610: %\paragraph{Type I surface with $\NumTangles=3$ and one constant edgepath}
3611: \noindent {(2-2)
3612: %% Type I surface with $\NumTangles=3$ and one constant edgepath
3613: %% The edgepath system of the Type I surface includes one constant edgepath
3614: $\NumTangles=3$, $\NumTangles_{\mathrm{const}}=1$.
3615: }
3616: 
3617: %
3618: %%% ### General ### %%%
3619: %
3620: For a type I surface with one constant edgepath,
3621: the formula (\ref{Eq:Remainder:UpperBound:TypeI:General}) is
3622: %
3623: $\Remainder
3624: \le
3625: 4-(1-1/q_a)\cdot 2/(1-u)
3626: $.
3627: %
3628: If $q_a=2$, 
3629: since $u \ge 1/2$,
3630: then $\Remainder\le 2$.
3631: If $q_a\ge3$, then $\Remainder\le 0$.
3632: %
3633: %%% ### With Cancellation ### %%%
3634: %
3635: For an edgepath system with one constant edgepath $q_a=2$ and cancellation,
3636: by the inequality (\ref{Eq:Remainder:UpperBound:TypeI:WithSingleCancel}),
3637: we have $\Remainder\le 4-\min \{s_1,q_1+4\}$,
3638: where the edge
3639: $\angleb{p_1/q_1}$\,--\,$\angleb{r_1/s_1}$ causes cancellation.
3640: The possibility of $\Remainder>0$ remains
3641: when $(q_1,s_1)=(1,2),(1,3),(2,3)$.
3642: In any cases,
3643: we can check that every candidate edgepath system with cancellation
3644: obtained by solving the equation $\sum_{i=1}^{\NumTangles} \widetilde{\BasicEdgepath_i}(u)=0$ actually
3645: satisfies $\Remainder\le 0$ as follows.
3646: 
3647: First, if the last edge $\Edge_1$ is of type $(q_1,s_1)=(1,2)$,
3648: the $u$ coordinate of the endpoints satisfies $u\le \frac{1}{2}$.
3649: At the same time, as $\Edgepath_a$ is a constant edgepath,
3650: $u\ge \frac{1}{2}$ is required.
3651: Hence, we have a contradiction.
3652: 
3653: For the case of $(q_1,s_1)=(1,3)$,
3654: possible edgepath systems are essentially same as the following case:
3655: \[
3656: \left\{
3657: \begin{array}{l}
3658: \textrm{$\Edgepath_a$ is a constant edgepath consisting of a point on the edge $\angleb{1/2}$\,--\,$\circleb{1/2}$,} \\
3659: \textrm{$\Edge_1$ is a partial edge of $\angleb{0}$\,--\,$\angleb{1/3}$,} \\
3660: \textrm{$\Edge_2$ is a partial edge of $\angleb{-1}$\,--\,$\angleb{-(t-1)/t}$ ($t=3,4,5$).}
3661: \end{array}
3662: \right.
3663: \]
3664: %
3665: This appears as the items 
3666: \ref{Item:0-1/2,0-1/3,B},
3667: \ref{Item:0-1/2,0-1/3,C} and
3668: \ref{Item:0-1/2,0-1/3,D}
3669: in the previous section.
3670: Note that $\Edge_1$ and $\Edge_2$ have a common sign.
3671: The value of $|\Edge_1|+|\Edge_2|$ is $1/2+1/2=1$ for $t=3$,
3672: $1/4+1/2=3/4$ for $t=4$.
3673: Since $\Remainder\le 2$ holds even if we ignore the effect of the cancellation, by taking it into account, we have $\Remainder\le 0$ .
3674: When $t=5$, since $|\Edge_1|$ is zero, the edgepath system contradicts the hypothesis that $\Edge_1$ causes cancellation.
3675: 
3676: For the case of $(q_1,s_1)=(2,3)$,
3677: possible edgepath systems are essentially same as the following case:
3678: \[
3679: \left\{
3680: \begin{array}{l}
3681: \textrm{$\Edgepath_a$ is a constant edgepath consisting of a point on the edge $\angleb{1/2}$\,--\,$\circleb{1/2}$,} \\
3682: \textrm{$\Edge_1$ is a partial edge of $\angleb{1/2}$\,--\,$\angleb{1/3}$,} \\
3683: \textrm{$\Edge_2$ is a partial edge of $\angleb{-1}$\,--\,$\angleb{-(t-1)/t}$ ($t\ge 5$).}
3684: \end{array}
3685: \right.
3686: \]
3687: This corresponds to the item \ref{Item:0-1/2,1-1/2-1/3,B}.
3688: Note that $\Edge_1$ and $\Edge_2$ have opposite signs.
3689: In this case, we have 
3690: $u=(t-1)/\{2(t-2)\}$,
3691: $|\Twist| %%%=2||\Edge_1|-|\Edge_2||=2|(t-5)/(t-3)-(t-4)(t-3)|
3692: =2/(t-3)$,
3693: and $-\chi/\sharp s %%%=(t-5)/(t-3)+(t-4)(t-3)+(1-3)+(3-2-1/2)\cdot(2t-4)(t-3)
3694: =1-2/(t-3)$.
3695: Hence, $\Remainder=|\Twist|-2\cdot(-\chi/\sharp s)=6/(t-3)-2$.
3696: If $t=5$, since $|\Edge_1|=0$, $\Edge_1$ does not cause cancellation and the edgepath system contradicts the hypothesis.
3697: If $t\ge 6$, then $\Remainder\le 0$.
3698: %
3699: 
3700: \medskip
3701: 
3702: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3703: % Paragraph : Type I surface for $\NumTangles=3$ with two constant edgepaths
3704: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3705: 
3706: %\paragraph{Type I surface for $\NumTangles=3$ with two constant edgepaths}
3707: \noindent {(2-3)
3708: %% Type I surface for $\NumTangles=3$ with two constant edgepaths
3709: %% The edgepath system of the Type I surface includes two constant edgepaths
3710: $\NumTangles=3$, $\NumTangles_{\mathrm{const}}=2$.
3711: }
3712: 
3713: %
3714: %%% ### General ### %%%
3715: %
3716: First, the inequality (\ref{Eq:Remainder:UpperBound:TypeI:General}) is
3717: %
3718: $\Remainder\le 2-\{ 1-1/q_a-1/q_b \}\cdot 2/(1-u)
3719: $.
3720: %
3721: Let $q_a$ denote the smaller of denominators of the tangles corresponding to constant edgepaths,
3722: and $q_b$ the larger. 
3723: By the condition for the Montesinos link to be a knot,
3724: we have $q_a\ge 2$ and $q_b\ge3$.
3725: By $u \ge (q_b-1)/q_b$,
3726: we also have $1/(1-u)\ge q_b \ge 3$.
3727: %
3728: Thus
3729: $\Remainder\le 2-( 1-1/2-1/3 )~2\cdot 3 =1$.
3730: %
3731: If $q_a=2$ and $q_b=5$, then
3732: $\Remainder\le 2-( 1-1/2-1/5 )~2\cdot 5 = -1$.
3733: If $q_a=3$ and $q_b=3$, then
3734: $\Remainder\le 2-( 1-1/3-1/3 )~2\cdot 3 = 0$.
3735: Similarly, for other edgepath systems, we have $\Remainder\le 0$.
3736: 
3737: %
3738: %%% ### With Cancellation ### %%%
3739: %
3740: For an edgepath with cancellation,
3741: it is sufficient to check for $q_a=2$ and $q_b=3$.
3742: Since $x=1-1/q_a-1/q_b=1/6$, 
3743: by inequality (\ref{Eq:Remainder:UpperBound:TypeI:WithSingleCancel}),
3744: $\Remainder\le 2-1/3 \min\{s_1,q_1+12\}$.
3745: %
3746: If $s_1\ge 6$, then $\Remainder\le 0$.
3747: Hence, an edgepath system with $s_1\le 5$ only remains.
3748: Not so many such concrete examples exist, in fact.
3749: The edgepath system must be
3750: \[
3751: \left\{
3752: \begin{array}{l}
3753: \textrm{$\Edgepath_a$ is a constant edgepath consisting of a point on the edge $\angleb{x/2}$\,--\,$\circleb{x/2}$}, \\
3754: \textrm{$\Edgepath_b$ is a constant edgepath consisting of a point on the edge $\angleb{y/3}$\,--\,$\circleb{y/3}$}, \\ 
3755: \textrm{$\Edge_1$ is a partial edge of $\angleb{p_1/q_1}\textrm{\,--\,}\angleb{r_1/s_1}$} \\
3756: \end{array}
3757: \right.
3758: \]
3759: for some appropriate integer $x$ and $y$.
3760: %
3761: In order for the edgepath system
3762: to satisfy $\sum_{i=1}^{\NumTangles} \Edgepath_i(u)=0$ at the common $u$-coordinate of the endpoints,
3763: the edge $\Edge_1$ must intersect with the horizontal segments
3764: $v=\pm 1/6+z~(z\in\mathbb{Z})$
3765: within a strip region $2/3 \le u < 1$. 
3766: %
3767: %\begin{figure}[hbt]
3768: % \begin{center}
3769: %  \begin{picture}(200,200)
3770: %   \put(0,0){\scalebox{1.0}{\includegraphics{d-2.eps}}}
3771: %  \end{picture}
3772: % \end{center}
3773: % \caption{Two horizontal edgepaths}
3774: % \label{Fig:TwoHorizontalEdgepaths}
3775: %\end{figure}
3776: %
3777: The only example of such an edgepath system 
3778: has the partial edge of $\angleb{0}$\,--\,$\angleb{1/5}$ as $\Edge_1$
3779: (or another example essentially same as this example).
3780: We must add at least one increasing complete edge to $\Edge_1$
3781: so that $\Edge_1$ actually causes cancellation.
3782: Since $|\Edge_1|=1/2$,
3783: we have $\Remainder=1-4\cdot 1/2=-1$ for this edgepath system.
3784: %
3785: 
3786: \medskip
3787: 
3788: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3789: % Paragraph : Type I surface for $\NumTangles=3$ with three constant edgepaths
3790: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3791: 
3792: %\paragraph{Type I surface for $\NumTangles=3$ with three constant edgepaths}
3793: \noindent {(2-4)
3794: %% Type I surface for $\NumTangles=3$ with three constant edgepaths
3795: %% All edgepaths of the edgepath system of the Type I surface are constant edgepaths
3796: $\NumTangles=3$, $\NumTangles_{\mathrm{const}}=3$.
3797: }
3798: 
3799: As mentioned in Remark \ref{Rem:Condition:ForLinkToBeKnot},
3800: no edgepath system with three constant edgepaths exists.
3801: %
3802: \end{proof}
3803: 
3804: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3805: % Subsection : Type I surface and Type II surface without cancellation
3806: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3807: 
3808: \subsubsection{Type I surface and type II surface without cancellation}
3809: 
3810: Now, we have only to show the following.
3811: 
3812: \begin{lemma}
3813: %
3814: The type I surface without cancellation and
3815: the type II surface without cancellation 
3816: do not exist for a Montesinos knot $K$ at the same time.
3817: %
3818: \end{lemma}
3819: %
3820: \begin{proof}
3821: %
3822: For the type I surface without cancellation,
3823: $\sum_{i=1}^{\NumTangles} K_i$ and $\EdgepathSystem(+0)$ have opposite signs.
3824: In contrast with this,
3825: for the type II surface without cancellation,
3826: $\sum_{i=1}^{\NumTangles} K_i$ and $\EdgepathSystem(+0)$ have the same sign or $\EdgepathSystem(+0)=0$.
3827: %
3828: \end{proof}
3829: 
3830: Combining sub-lemmas completes the proof of Lemma \ref{Lem:Remainder:UpperBound}.
3831: %
3832: 
3833: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3834: % Subsection : 
3835: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3836: 
3837: \subsection{Proof, best possibility and corollaries of Theorem \ref{Thm:Diff:UpperBound:Main}}
3838: \label{Subsec:Diff:ProofEtc}
3839: 
3840: \subsubsection{Proof}
3841: 
3842: Theorem \ref{Thm:Diff:UpperBound:Main} follows Lemma \ref{Lem:Remainder:UpperBound}.
3843: %
3844: \begin{proof}[Proof of Theorem \ref{Thm:Diff:UpperBound:Main}]
3845: %
3846: By the triangle inequality and Lemma \ref{Lem:Remainder:UpperBound}
3847: \begin{eqnarray*}
3848: |\Slope_1-\Slope_2| 
3849: &=& |\{\Twist_1-\Twist_{S}\} -\{\Twist_2-\Twist_{S}\} | \\
3850: &=& |\Twist_1-\Twist_2| \\
3851: &\le& |\Twist_1|+|\Twist_2| \\
3852: &=&2\,(\frac{-\chi_1}{\sharp s_1}+\frac{-\chi_2}{\sharp s_2})+(\Remainder_1+\Remainder_2) \\
3853: &\le&2\,(\frac{-\chi_1}{\sharp s_1}+\frac{-\chi_2}{\sharp s_2})+4
3854: ,
3855: \end{eqnarray*}
3856: %
3857: where $\Twist_{S}$ is the twist of a Seifert surface $F_{S}$ of $K$.
3858: 
3859: Note that there is no side effect by simplification.
3860: For example, the inequality (\ref{Eq:Diff:UpperBound:Main})
3861: holds also for two surfaces $F_1$ and $F_2$ with the same boundary slope,
3862: since essential surfaces satisfy $(-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2)\ge -2$ even if they are disks.
3863: 
3864: .
3865: %
3866: \end{proof}
3867: 
3868: \subsubsection{Best possibility}
3869: 
3870: The upper bound in Theorem \ref{Thm:Diff:UpperBound:Main} is best possible in a sense.
3871: %%% That is, 
3872: %%% for arbitrary large $t$,
3873: %%% there is a Montesinos knot $K$ 
3874: %%% and two essential surfaces $F_1$ and $F_2$ in its knot complement,
3875: %%% where the two surfaces are incompressible but not necessarily orientable
3876: %%% and the corresponding boundary slopes satisfy
3877: %%% $|\Slope_1-\Slope_2|=2\{(-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2)\}+4$
3878: %%% and $(-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2) > t$.
3879: 
3880: First, we provide a concrete example of a family of Montesinos knots and pairs of edgepath systems of boundary slopes of the knots. 
3881: The Montesinos knot is $K(1/(2k),1/5,\ldots,1/5)$ with $\NumTangles\ge 3$ tangles and the natural number $k$.
3882: The two edgepath systems $\EdgepathSystem_1=(\Edgepath_{1,1},\Edgepath_{1,2},\ldots,\Edgepath_{1,\NumTangles})$ and $\EdgepathSystem_2=(\Edgepath_{2,1},\Edgepath_{2,2},\ldots,\Edgepath_{2,\NumTangles})$ of $F_1$ and $F_2$ are as follows.
3883: %
3884: \begin{eqnarray*}
3885: \Edgepath_{1,1}&=&\angleb{0}-\angleb{1/(2k)}, \\
3886: \Edgepath_{1,i}&=&\angleb{0}-\angleb{1/5} \textrm{\ \ \ \ for $2\le i \le \NumTangles$}, \\
3887: \Edgepath_{2,1}&=&\angleb{\infty}-\angleb{1}-\angleb{1/2}-\angleb{1/3}-\cdots-\angleb{1/(2k-1)}-\angleb{1/(2k)}, \\ 
3888: \Edgepath_{2,i}&=&\angleb{\infty}-\angleb{1}-\angleb{1/2}-\angleb{1/3}-\angleb{1/4}-\angleb{1/5} \textrm{\ \ \ \ for $2\le i \le \NumTangles$}. 
3889: \end{eqnarray*}
3890: These candidate edgepath systems $\EdgepathSystem_1$ and $\EdgepathSystem_2$ are type II and type III respectively.
3891: Two candidate surfaces $F_1$ and $F_2$ are both incompressible by the Corollary 2.4 and Proposition 2.5 in \cite{HO}.
3892: 
3893: %%% Since the type II system $\EdgepathSystem_1$ has the cycle of final r-values $(2k-1,4,\ldots,4)$,
3894: %%% it is incompressible by the Proposition 2.4 in \cite{HO}.
3895: %%% Besides,
3896: %%% since all edgepaths of the type III system $\EdgepathSystem_2$ go through $\angleb{1}$ and the absolute value of their sum $|1+\ldots+1|=\NumTangles$ is greater than 2,
3897: %%% it is also incompressible by the Proposition 2.5 in \cite{HO}.
3898: %%% Eventually, two candidate surfaces $F_1$ and $F_2$ are both incompressible.
3899: 
3900: Since $\EdgepathSystem_1$ and $\EdgepathSystem_2$ are monotonically decreasing and increasing,
3901: their remainder term $\Remainder$ are easily confirmed to be $4$ and $0$.
3902: $\Twist_1$ and $\Twist_2$ with opposite signs give $|\Twist_1-\Twist_2|=|\Twist_1|+|\Twist_2|$.
3903: %
3904: Thus, $F_1$ and $F_2$ satisfy
3905: \[
3906: |\Slope_1-\Slope_2|= 2\,(\frac{-\chi_1}{\sharp s_1}+\frac{-\chi_2}{\sharp s_2})+4
3907: .
3908: \]
3909: Note that $(-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2)$
3910: is greater than arbitrary $t$ for sufficiently large $k$.
3911: 
3912: %%%\subsubsection{Proof of corollaries}
3913: \subsubsection{Corollaries}
3914: 
3915: As described in Section \ref{Sec:ABoundOnTheDenominator},
3916: if a Montesinos knot $K$ is not the same as or is not isotopic to $(-2,3,t)$-pretzel knots for odd $t\ge 3$,
3917: we have $-\chi/\sharp s\ge 1$ for its boundary slopes.
3918: The $(-2,3,t)$-pretzel knots have boundary slopes and corresponding essential surfaces as follows. Note that they are torus knots if $t=3$ or $5$.
3919: 
3920: \begin{itemize}
3921:  \item $t=3$
3922:  \begin{itemize}
3923:   \item[$\circ$] $\Slope_a=12$, $\Denom_a=1$, $\chi_a=0$, $\sharp s_a=\sharp b_a=2$, $-\chi_a/\sharp s_a=-\chi_a/\sharp b_a=0$,
3924:   \item[$\circ$] $\Slope_b=0$, $\Denom_b=1$, $\chi_b=-5$, $\sharp s_b=\sharp b_b=1$, $-\chi_b/\sharp s_b=-\chi_b/\sharp b_b=5$.
3925:  \end{itemize}
3926:  \item $t=5$
3927:  \begin{itemize}
3928:   \item[$\circ$] $\Slope_a=15$, $\Denom_a=1$, $\chi_a=0$, $\sharp s_a=\sharp b_a=2$, $-\chi_a/\sharp s_a=-\chi_a/\sharp b_a=0$,
3929:   \item[$\circ$] $\Slope_b=0$, $\Denom_b=1$, $\chi_b=-7$, $\sharp s_b=\sharp b_b=1$, $-\chi_b/\sharp s_b=-\chi_b/\sharp b_b=7$.
3930:  \end{itemize}
3931:  \item $t\ge 7$,
3932:  \begin{itemize}
3933:   \item[$\circ$] $\Slope_a=16$, $\Denom_a=1$, $\chi_a=6-t$, $\sharp s_a=\sharp b_a=1$, $-\chi_a/\sharp s_a=-\chi_a/\sharp b_a=t-6$,
3934:   \item[$\circ$] $\Slope_b=(t^2-t-5)/\{(t-3)/2\}$, $\Denom_b=(t-3)/2$, $\chi_b=5-t$, $\sharp s_b=t-3$, $\sharp b_b=2$, $-\chi_b/\sharp s_b=1-2/(t-3)$, $-\chi_b/\sharp b_b=(t-5)/2$,
3935:   \item[$\circ$] $\Slope_c=2t+6$, $\Denom_c=1$, $-\chi_c/\sharp s_c=-\chi_c/\sharp b_c=1$,
3936:   \item[$\circ$] $\Slope_d=0$, $\Denom_d=1$, $-\chi_d/\sharp s_d=-\chi_d/\sharp b_d=t+2$.
3937:  \end{itemize}
3938: \end{itemize}
3939: %
3940: By examining these boundary slopes,
3941: we obtain linear or quadratic upper bounds of the difference and the distance of the two boundary slopes.
3942: 
3943: %%%
3944: 
3945: %With respect to a linear bound of the difference
3946: %or a somehow irregular quadratic bound of the distance 
3947: %by Euler characteristics,
3948: %following corollary is easily obtained from Theorem \ref{Thm:Diff:UpperBound:Main}.
3949: %Though the bounds may not be sharp for Montesinos knots with $\NumTangles\ge 3$,
3950: %the equality holds for boundary slopes of the trefoil knot.
3951: 
3952: %%%% Corollary
3953: 
3954: \begin{proof}[Proof of Corollary \ref{Cor:Diff:UpperboundByGenus}]
3955: 
3956: If the knot $K$ is not $(-2,3,t)$-pretzel,
3957: $-\chi/\sharp s\ge 1$ and $g\ge 1$ hold for all essential surfaces.
3958: Since $g\ge 1$ gives $-\chi/\sharp s\le 2g-1$,
3959: we have the inequality (\ref{Eq:Diff:UpperBound:Main-B:ByGenus}).
3960: %
3961: If the knot $K$ is $(-2,3,t)$-pretzel with $t\ge 7$,
3962: then genus of any essential surface is found to be $1$ or greater.
3963: Similarly to the previous case, we have (\ref{Eq:Diff:UpperBound:Main-B:ByGenus}).
3964: %
3965: For the remaining $(-2,3,3)$ and $(-2,3,5)$-pretzel knots,
3966: the value of $|\Slope_1-\Slope_2| - 4 \,( g_1 + g_2 )$ 
3967: for the two boundary slopes 
3968: is $|12-0|-4\,(0+3)=0$ for $(-2,3,3)$
3969: and $|15-0|-4\,(0+4)=-1<0$ for $(-2,3,5)$.
3970: %
3971: \end{proof}
3972: 
3973: %%%% Corollary
3974: 
3975: %\begin{corollary}
3976: %\label{Cor:DiffDist:UpperBound}
3977: %For two boundary slopes and their corresponding essential surfaces,
3978: %we have the inequality
3979: %\begin{eqnarray}
3980: %|\Slope_1-\Slope_2|\le 6\,(\frac{-\chi_1}{\sharp s_1}+\frac{-\chi_2}{\sharp s_2}).
3981: %\label{Eq:Diff:UpperBound:Linear}
3982: %\end{eqnarray}
3983: %This is equivalent to the inequality
3984: %\begin{eqnarray}
3985: %\Delta(\Slope_1,\Slope_2)\le 6\,(\Denom_2 \frac{-\chi_1}{\sharp b_1}+\Denom_1 \frac{-\chi_2}{\sharp b_2}).
3986: %\label{Eq:Dist:UpperBound:Semi-Linear}
3987: %\end{eqnarray}
3988: %\end{corollary}
3989: 
3990: \begin{proof}[Proof of Corollary \ref{Cor:DiffDist:UpperBound}]
3991: If the knot $K$ is neither $(-2,3,3)$ nor $(-2,3,5)$-pretzel essentially, we have $-\chi/\sharp s\ge 1/2$ for all boundary slopes.
3992: Then, since $(-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2)\ge 1$,
3993: we obtain 
3994: (\ref{Eq:Diff:UpperBound:Linear})
3995: %%% $|\Slope_1-\Slope_2|\le 6\,((-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2))$
3996: from (\ref{Eq:Diff:UpperBound:Main}) in Theorem \ref{Thm:Diff:UpperBound:Main}.
3997: %%% that is, $|\Slope_1-\Slope_2|\le 2\,((-\chi_1/\sharp s_1)+(-\chi_2/\sharp s_2))+4$.
3998: 
3999: For remaining $(-2,3,3)$ and $(-2,3,5)$-pretzel knots,
4000: since the expression $|\Slope_a-\Slope_b|-6\,((-\chi_a/\sharp s_a)+(-\chi_b/\sharp s_b))$
4001: has its value $12-6\cdot 5=-18<0$ and $15-6\cdot 7=-27<0$ respectively,
4002: the inequality (\ref{Eq:Diff:UpperBound:Linear}) holds.
4003: \end{proof}
4004: 
4005: %%%
4006: 
4007: %Regarding the upper bound of the distance or difference by the product of Euler characteristics, we have the following.
4008: %Though the bounds may not be sharp for Montesinos knots with $\NumTangles\ge3$ tangles,
4009: %the equality holds for boundary slopes of the figure eight knot.
4010: 
4011: %%%% Corollary
4012: 
4013: %\begin{corollary}
4014: %\label{Cor:Main3b}
4015: %If both of the Euler characteristics are negative, then we have
4016: %\begin{eqnarray}
4017: %\Delta(\Slope_1,\Slope_2)\le 8\cdot \frac{-\chi_1}{\sharp b_1}\cdot \frac{-\chi_2}{\sharp b_2}.
4018: %\label{Eq:Dist:UpperBound:Quadratic}
4019: %\end{eqnarray}
4020: %This is equivalent to the inequality
4021: %\begin{eqnarray}
4022: %|\Slope_1-\Slope_2|\le 8\cdot \frac{-\chi_1}{\sharp s_1}\cdot \frac{-\chi_2}{\sharp s_2}.
4023: %\label{Eq:Diff:UpperBound:Quadratic}
4024: %\end{eqnarray}
4025: %\end{corollary}
4026: 
4027: \begin{proof}
4028: [Proof of Corollary \ref{Cor:Main3b}]
4029: If both boundary slopes satisfy $-\chi/\sharp s\ge 1$,
4030: or equivalently $\Denom \le -\chi/\sharp b$,
4031: we obtain 
4032: (\ref{Eq:Dist:UpperBound:Quadratic})
4033: %%%$\Delta\le 8\cdot (-\chi_1/\sharp b_1)\cdot (-\chi_2/\sharp b_2)$
4034: easily from (\ref{Eq:Dist:UpperBound:Main-C}).
4035: %%% that is,
4036: %%% $\Delta(\Slope_1,\Slope_2)\le 2\,(\Denom_2 (-\chi_1/\sharp b_1)+\Denom_1 (-\chi_2/\sharp b_2))+4 \Denom_1 \Denom_2$.
4037: Thus, we are done for all Montesinos knots but $(-2,3,t)$-pretzel knots.
4038: 
4039: For $(-2,3,3)$ and $(-2,3,5)$-pretzel knots,
4040: since one of the two essential surfaces has Euler characteristic $0$,
4041: there are no pairs of boundary slopes to be applied to the inequality.
4042: For $(-2,3,t)$-pretzel knots with odd $t\ge 7$,
4043: the value of the expression $\Delta(\Slope_i,\Slope_j)-8\cdot (-\chi_i/\sharp b_i)\cdot (-\chi_j/\sharp b_j)$ for $(i,j)=(a,b),(b,c),(b,d)$ are $-3t^2+35t-101$, $-3t+16$ and $-3t^2+11t+35$ respectively, which are all negative for any $t\ge 7$.
4044: \end{proof}
4045: 
4046: 
4047: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4048: %
4049: % Known results and open problems
4050: %
4051: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4052: 
4053: 
4054: \section{Known results and open problems}
4055: 
4056: Here we give a brief review about 
4057: the study of boundary slopes of essential surfaces 
4058: related to our results, and collects some open problems. 
4059: 
4060: 
4061: \subsection*{}
4062: 
4063: For the existence and the number of boundary slopes, 
4064: the following are fundamental. 
4065: It was shown by Hatcher in \cite{H} that 
4066: there are just finitely many boundary slopes 
4067: of essential surfaces for a compact, orientable, irreducible $3$-manifold 
4068: with boundary a single torus. 
4069: Also it was shown by Culler and Shalen in \cite{CS84} that 
4070: there are at least two boundary slopes 
4071: for a non-trivial knot exterior in the $3$-sphere $S^3$. 
4072: See also \cite{CS04}. 
4073: 
4074: Boundary slopes for some class of knots 
4075: have been intensively studied. 
4076: As a pioneering work, 
4077: for two-bridge knots, Hatcher and Thurston gave a complete enumeration 
4078: of boundary slopes in \cite{HT}. 
4079: Following this work, 
4080: Hatcher and Oertel \cite{HO} developed a procedure to compute 
4081: the boundary slopes for Montesinos knots, 
4082: on which our arguments heavily depend. 
4083: Recently boundary slopes of genus one essential surface 
4084: for Montesinos knots of length three are completely determined 
4085: by Wu \cite{W}. 
4086: 
4087: 
4088: 
4089: \subsection*{}
4090: %%% The denominators of boundary slopes ....
4091: 
4092: The denominators of boundary slopes have also been studied
4093: in relation to the study of Dehn surgery.
4094: 
4095: In the following,
4096: let $F$ be an essential surface properly embedded
4097: in the exterior of a non-trivial knot $K$ in $S^3$.
4098: The surface $F$ is of Euler characteristic $\chi$ and
4099: has the boundary slope $\Slope$ of $F$,
4100: which is represented by an irreducible fraction $\Slope = \Numer / \Denom$.
4101: Let $\sharp s$ denote the number of sheets of $F$ and
4102: $\sharp b$ the number of boundary components of $F$,
4103: where they are related to each other by $\sharp s = \Denom \,\sharp b$.
4104: 
4105: Please note that
4106: the results cited below will often be modified from the original  
4107: statements.
4108: It is for making easy to see their relationship and to compare with our  
4109: results.
4110: 
4111: First, for genus 0 case, by Gordon and Luecke in \cite{GLu87},
4112: it was shown that $\Denom \le 1$, that is, $\Slope$ is integral.
4113: On the other hand, for genus one case,
4114: no corresponding results are found in literature at least by the author.
4115: The related result was obtained by Gordon and Luecke in \cite{GLu95,  
4116: GLu00}.
4117: That is,
4118: if a Dehn surgery on a hyperbolic knot in $S^3$ along a slope $\Slope$
4119: yields a closed $3$-manifold containing an incompressible torus, then  
4120: $\Denom \leq 2$.
4121: Note that if such a surgery can occur, then $\Slope$ is a boundary slope of
4122: an essential surface of genus one. 
4123: However the converse does not hold in general. 
4124: 
4125: Following these results, it is natural to ask:
4126: 
4127: \begin{problem}
4128: Find a generalization to these results 
4129: for the higher genus case or the non-orientable surface case. 
4130: \end{problem}
4131: 
4132: Concerning this problem, some results are already known. 
4133: By using the argument used in \cite[Proposition 6.1]{GLi},
4134: together with Gabai's thin position argument \cite{Ga87},
4135: we have
4136: 
4137: $$ \Denom \le 6 \frac{ - \chi }{ \sharp b } $$
4138: %
4139: if the knot $K$ is non-cabled.
4140: Originally in their argument, $F$ is assumed to be orientable,
4141: but the assumption might be not necessary.
4142: This result had not been included in \cite{GLi}, but was suggested in  
4143: \cite{R}.
4144: 
4145: In \cite{T}, other generalization was developed
4146: if $K$ admits some tangle decomposition.
4147: In particular if $K$ has non-trivial $t$ connected summands, he obtained
4148: 
4149: $$ \Denom \le \frac{ g }{ t-1 } ,$$
4150: %
4151: where $F$ is assumed to be orientable and $g$ denotes the genus of $F$.
4152: 
4153: Under restriction to the class of knots, two excellent results are  
4154: known.
4155: One is for two-bridge knots, by Hatcher and Thurston \cite{HT}.
4156: They gave a classification of essential surfaces in two-bridge knot  
4157: exteriors, and
4158: as a corollary, it was shown that all such surfaces have integral  
4159: slopes.
4160: Another one is for alternating knots, by Menasco and Thistlethwaite  
4161: \cite{MT}.
4162: They presented that
4163: 
4164: $$ \Denom \le \frac{ - \chi }{ \sharp b } $$
4165: %
4166: for non-torus alternating knots.
4167: As a corollary they achieved the affirmative answer to
4168: the well-known Cabling Conjecture for alternating knots.
4169: We also remarked that
4170: it is known that a torus knot exterior contains only two essential  
4171: surfaces
4172: and their boundary slopes are both integers.
4173: 
4174: 
4175: Recently, Matignon and Sayari \cite{MS04}obtained
4176: similar bounds for non-orientable surfaces.
4177: Their result was obtained by using Dehn surgery method, 
4178: and in fact, they do not assume 
4179: that the surfaces they are considering are essential. 
4180: However, as pointing out in \cite{Te}, 
4181: the condition that the surfaces are essential is necessary. 
4182: Under the essentiality condition, 
4183: their results could be interpreted in terms of boundary  slopes as follows. 
4184: They actually showed that
4185: 
4186: $$ \Denom \le \frac{ - \chi }{ \sharp b } + 4 $$
4187: %
4188: if $F$ is non-orientable and $\sharp b > 1$.
4189: If $\sharp b = 1$, they had
4190: 
4191: $$ \Denom \le  - 3 \chi  + 1 $$
4192: %
4193: if $K$ is not a cable knot, and
4194: 
4195: $$ \Denom \le  - 5 \chi  + 3 $$
4196: %
4197: if $K$ is cable knot.
4198: They also showed that
4199: 
4200: $$ \Denom \le \frac{-\chi}{\sharp b} $$
4201: %
4202: if $K$ is a composite knot,
4203: and
4204: 
4205: $$ \Denom \le \frac{-\chi}{\sharp b} +1$$
4206: %
4207: if $K$ admits a Conway sphere.
4208: 
4209: Remark that, for a non-orientable genus two case, namely,
4210: punctured Klein bottle case,
4211: it was obtained in \cite{GLu95} that $\Denom = 1$, that is, $\Slope$ is integral.\\
4212: 
4213: 
4214: 
4215: 
4216: 
4217: There are many results which give upper bounds on 
4218: the distances between boundary slopes. 
4219: The main problem would be:
4220: 
4221: \begin{problem}
4222: Establish a sharp estimate on the distances between boundary slopes 
4223: in terms of the genera of the corresponding essential surfaces. 
4224: \end{problem}
4225: 
4226: In the following let $M$ be a compact orientable irreducible $3$-manifold
4227: whose boundary $\partial M$ is homeomorphic to the torus $T^2$.
4228: For $i =1,2$, let $F_i$ be an essential surface properly embedded in $M$ 
4229: of Euler characteristic $\chi_i$. 
4230: The boundary slope $\Slope_i$ of $F_i$ is 
4231: represented by an irreducible fraction $\Slope_i=\Numer_i/\Denom_i$.
4232: Let $\sharp s_i$ denote the number of sheets.
4233: Note that 
4234: if the number of boundary of $F_i$ is denoted by $\sharp b_i$,
4235: $\sharp b_i$ and $\sharp s_i$ are related to each other by $\sharp s_i = \Denom_i \sharp b_i$.
4236: 
4237: 
4238: For small genus surface case, intensively fine results have been achieved 
4239: in relation to the study of the exceptional Dehn surgery. 
4240: If both $F_i$'s are planar, 
4241: Gordon and Luecke proved in \cite[Theorem~1.1]{GLu96} that $\Delta \leq 4$ holds. 
4242: If both $F_i$'s are punctured torus, Gordon proved in \cite[Theorem~1.1]{G} 
4243: that $\Delta \leq 8$ holds. 
4244: Moreover, he gave $\Delta \leq 5$ with just five exceptional manifolds, 
4245: which are completely characterized. 
4246: 
4247: As a generalization to the higher genera case, 
4248: Gordon and Litherland obtained 
4249: in \cite[Proposition~6.1]{GLi} the following: 
4250: Suppose that $M$ contains no cable spaces. 
4251: If both $F_i$'s are orientable and $F_1$ is planar, 
4252: then $ \Delta < 6 \,(-\chi_2/\sharp b_2) $ holds. 
4253: 
4254: As a natural extension of \cite[Proposition~6.1]{GLi}, 
4255: Torisu obtained in \cite[Theorem~1]{T} the following: 
4256: Suppose that $M$ contains no essential annuli. 
4257: If both $F_i$'s are orientable and of genus $g_i \geq 1$, 
4258: then $ \Delta < 36 (2 g_1 -1 ) (2 g_2 -1)$ holds. 
4259: 
4260: On the other hand, Rieck obtained in \cite{R} a slightly sharper bound. 
4261: Suppose that $M$ contains no essential annuli. 
4262: If both $F_i$'s are orientable and of genus $g_i \geq 1$, 
4263: $ \Delta < 18 (2 g_1 +1 ) (g_2 +1)$ holds. 
4264: Moreover if $\sharp b_i \geq 2$ for $i=1,2$, 
4265: then $ \Delta < 18 (g_1 +1 ) (g_2 +1)$ holds. 
4266: This is slightly different from the original form. 
4267: Please refer \cite[Theorem~5.2]{R} as the original form. 
4268: In fact, from his proof, we can find 
4269: $$ \Delta < 
4270: 18 \left(2 \frac{g_1}{\sharp b_1} +1 \right) 
4271: \left(2 \frac{g_2}{\sharp b_2} +1 \right).$$ 
4272: 
4273: These above are all proved by 
4274: the combinatorial analysis of the graph 
4275: constructed from the intersection of the two surfaces. 
4276: 
4277: On the other hand, by a differential geometric approach, 
4278: the following bound was shown by Hass, Rubinstein and Wang in \cite{HRW}: 
4279: Suppose that the interior of $M$ admits 
4280: a complete hyperbolic metric of finite volume. 
4281: Then 
4282: $$ \Delta \leq \frac{(2\pi)^2}{3.35} 
4283: \frac{ - \chi_1 }{\sharp b_1} \frac{ - \chi_2 }{\sharp b_2}
4284: \doteq 11.8 
4285: \frac{ - \chi_1 }{\sharp b_1} \frac{ - \chi_2 }{\sharp b_2}
4286: $$
4287: holds. 
4288: This bound still holds in the case that 
4289: the surface $F_i$ is immersed essential surface. 
4290: We remark that this also differs from the original form. 
4291: In the original form, the surfaces are assumed to be orientable, 
4292: but this orientability condition is not necessary in their argument. 
4293: Please refer \cite[Theorem~4.5]{HRW} as the original form. 
4294: %
4295: Moreover, by using the result of Agol \cite[Theorem~5.1]{A}, 
4296: this bound is improved as 
4297: $$ \Delta \leq \frac{36}{3.35} 
4298: \frac{ - \chi_1 }{\sharp b_1} \frac{ - \chi_2 }{\sharp b_2}
4299: \doteq 
4300: \frac{43}{4} 
4301: \frac{ - \chi_1 }{\sharp b_1} \frac{ - \chi_2 }{\sharp b_2} .
4302: $$
4303: 
4304: 
4305: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4306: %
4307: % Reference
4308: %
4309: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4310: 
4311: \begin{thebibliography}{9}
4312: 
4313: \bibitem[A00]{A} 
4314: I. Agol,
4315: \textit{Bounds on exceptional Dehn filling}, 
4316: Geom. Topol. \textbf{4} (2000), 431--449. 
4317: 
4318: \bibitem[CS84]{CS84}
4319: M. Culler and P.B. Shalen, 
4320: \textit{Bounded, separating, incompressible surfaces in knot manifolds}, 
4321: Invent. Math. \textbf{75} (1984), 537--545. 
4322: 
4323: %%% \bibitem[CS99]{CS99}
4324: %%% M. Culler and P.B. Shalen, 
4325: %%% \textit{Boundary slopes of knots}, 
4326: %%% Comment. Math. Helv. \textbf{74} (1999), 530--547.
4327: 
4328: \bibitem[CS04]{CS04}
4329: M. Culler and P.B. Shalen, 
4330: \textit{Knots with only two strict essential surfaces}, 
4331: Geometry and Topology Monographs, Vol. 7 (2004), 
4332: Proceedings of the Casson Fest, Paper no. 14, pages 335--430. 
4333: 
4334: \bibitem[Dun01]{Dun}
4335: N. Dunfield,
4336: \textit{A table of boundary slopes of Montesinos knots}, 
4337: Topology \textbf{40} (2001), no. 2, 309--315.
4338: 
4339: \bibitem[EM92]{EM92}
4340: M. Eudave-Mu\~{n}oz, 
4341: \textit{Band sums of links which yield composite links. The cabling conjecture for strongly invertible knots}, 
4342: Trans, Amer. Math. Soc. \textbf{330} (1992), 463--501. 
4343: 
4344: \bibitem[Ga87]{Ga87}
4345: D. Gabai,
4346: \textit{Foliations and the topology of $3$-manifolds. $III$},
4347: J. Diff. Geom. \textbf{26} (1987), 479--536.
4348: 
4349: \bibitem[G98]{G}
4350: C.McA. Gordon, 
4351: \textit{Boundary slopes of punctured tori in $3$-manifolds}, 
4352: Trans. Amer. Math. Soc. \textbf{350} (1998), 1713--1790. 
4353: 
4354: \bibitem[GLi84]{GLi}
4355: C.McA. Gordon and R.A. Litherland, 
4356: \textit{Incompressible planar surfaces in $3$-manifolds}, 
4357: Topology Appl. \textbf{18} (1984), 121--144.
4358: 
4359: \bibitem[GLu87]{GLu87}
4360: C.McA. Gordon and J. Luecke,
4361: \textit{Only integral Dehn surgeries can yield reducible manifolds},
4362: Math. Proc. Camb. Phil. Soc. \textbf{102} (1987), 97--101.
4363: 
4364: \bibitem[GLu95]{GLu95}
4365: C.McA. Gordon and J. Luecke, 
4366: \textit{Dehn surgeries on knots creating essential tori. $I$},
4367: Commun. Anal. Geom. \textbf{3} (1995), no.4, 597--644.
4368: 
4369: \bibitem[GLu96]{GLu96}
4370: C.McA. Gordon and J. Luecke,
4371: \textit{Reducible manifolds and Dehn surgery}, 
4372: Topology \textbf{35} (1996), no.2, 385--409.
4373: 
4374: \bibitem[GLu00]{GLu00}
4375: C.McA. Gordon and J. Luecke,
4376: \textit{Dehn surgeries on knots creating essential tori. $II$},
4377: Commun. Anal. Geom. \textbf{8} (2000), no.4, 671--725.
4378: 
4379: \bibitem[HRW99]{HRW} 
4380: J. Hass, J.H. Rubinstein and S.-C. Wang, 
4381: \textit{Boundary slopes of immersed surfaces in $3$-manifolds}, 
4382: J. Differential Geom. \textbf{52} (1999), no. 2, 303--325. 
4383: 
4384: \bibitem[H82]{H}
4385: A. Hatcher, 
4386: \textit{On the boundary curves of incompressible surfaces}, 
4387: Pacific J. Math. \textbf{99} (1982), 373--377. 
4388: 
4389: \bibitem[HO89]{HO}
4390: A. Hatcher and U. Oertel, 
4391: \textit{Boundary slopes for Montesinos knots},
4392: Topology \textbf{28} (1989), no. 4, 453--480.
4393: 
4394: \bibitem[HT85]{HT}
4395: A.E. Hatcher and W.P. Thurston,
4396: \textit{Incompressible surfaces in 2-bridge knot complements},
4397: Invent. Math. \textbf{79} (1985), 225--246.
4398: 
4399: \bibitem[I]{I}
4400: K. Ichihara, 
4401: \textit{Distances between immersed boundary slopes}, 
4402: in preparation. 
4403: 
4404: %%% \bibitem[IMS04]{IMS04}
4405: %%% M. Ishikawa, T.W. Mattman and K. Shimokawa, 
4406: %%% \textit{Exceptional surgery and boundary slopes},
4407: %%% preprint available at arXiv:math.GT/0211147 
4408: 
4409: %%% \bibitem[L00]{L} 
4410: %%% M. Lackenby,
4411: %%% \textit{Word hyperbolic Dehn surgery}, 
4412: %%% Invent. Math. \textbf{140} (2000), 243--282. 
4413: 
4414: \bibitem[MS04]{MS04}
4415: D. Matignon and N. Sayari,
4416: \textit{Non-Orientable Surfaces and Dehn Surgeries},
4417: Canadian J. Math. \textbf{56} (2004), no.5, 1022--1033. 
4418: 
4419: \bibitem[MT92]{MT}
4420: W.W. Menasco and M.B. Thistlethwaite,
4421: \textit{Surfaces with boundary in alternating knot exteriors},
4422: J. Reine Angew. Math. \textbf{426} (1992), 47--65.
4423: 
4424: \bibitem[O84]{O}
4425: U. Oertel, 
4426: \textit{Closed incompressible surfaces in complements of star links}, 
4427: Pacific J. Math. \textbf{111} (1984), 209--230. 
4428: 
4429: \bibitem[R00]{R}
4430: Y. Rieck, 
4431: \textit{Heegaard structure of manifolds in the Dehn filling space}, 
4432: Topology \textbf{39} (2000), 619--641. 
4433: 
4434: \bibitem[R76]{Ro}
4435: D. Rolfsen, 
4436: \textit{Knots and Links}, 
4437: Publish or Perish, Berkeley, California, 1976.
4438: 
4439: \bibitem[T]{Te}
4440: M. Teragaito, 
4441: \textit{On non-integral Dehn surgeries creating non-orientable surfaces},
4442: to appear in Bull. Canadian Math. 
4443: 
4444: 
4445: \bibitem[T96]{T}
4446: I. Torisu, 
4447: \textit{Boundary slopes for knots},
4448: Osaka J. Math. \textbf{33} (1996), no. 1, 47--55.
4449: 
4450: \bibitem[W]{W}
4451: Y.Q. Wu, 
4452: \textit{The classification of toroidal Dehn surgeries on Montesinos knots}, 
4453: preprint. 
4454: 
4455: \end{thebibliography}
4456: 
4457: \end{document}
4458: 
4459: 
4460: 
4461:  
4462: