1:
2:
3: \documentclass[11pt]{article}
4: \special{header=xxxprolog1.ps}
5:
6:
7: \renewcommand{\baselinestretch}{1.1}
8: \setlength{\textwidth}{6.0in}
9: %\setlength{\textwidth}{6.5in}
10: \setlength{\topmargin}{-0.5in}
11: \setlength{\textheight}{9in}
12: \setlength{\oddsidemargin}{0in}
13:
14: %%---%%---%%---%%---%%---%%
15:
16:
17: %%---%%---%%---%%---%%---%%
18:
19: % EPSF.TEX macro file:
20: % Written by Tomas Rokicki of Radical Eye Software, 29 Mar 1989.
21: % Revised by Don Knuth, 3 Jan 1990.
22: % Revised by Tomas Rokicki to accept bounding boxes with no
23: % space after the colon, 18 Jul 1990.
24: %
25: % TeX macros to include an Encapsulated PostScript graphic.
26: % Works by finding the bounding box comment,
27: % calculating the correct scale values, and inserting a vbox
28: % of the appropriate size at the current position in the TeX document.
29: %
30: % To use with the center environment of LaTeX, preface the \epsffile
31: % call with a \leavevmode. (LaTeX should probably supply this itself
32: % for the center environment.)
33: %
34: % To use, simply say
35: % \input epsf % somewhere early on in your TeX file
36: % \epsfbox{filename.ps} % where you want to insert a vbox for a figure
37: %
38: % Alternatively, you can type
39: %
40: % \epsfbox[0 0 30 50]{filename.ps} % to supply your own BB
41: %
42: % which will not read in the file, and will instead use the bounding
43: % box you specify.
44: %
45: % The effect will be to typeset the figure as a TeX box, at the
46: % point of your \epsfbox command. By default, the graphic will have its
47: % `natural' width (namely the width of its bounding box, as described
48: % in filename.ps). The TeX box will have depth zero.
49: %
50: % You can enlarge or reduce the figure by saying
51: % \epsfxsize=<dimen> \epsfbox{filename.ps}
52: % (or
53: % \epsfysize=<dimen> \epsfbox{filename.ps})
54: % instead. Then the width of the TeX box will be \epsfxsize and its
55: % height will be scaled proportionately (or the height will be
56: % \epsfysize and its width will be scaled proportiontally). The
57: % width (and height) is restored to zero after each use.
58: %
59: % A more general facility for sizing is available by defining the
60: % \epsfsize macro. Normally you can redefine this macro
61: % to do almost anything. The first parameter is the natural x size of
62: % the PostScript graphic, the second parameter is the natural y size
63: % of the PostScript graphic. It must return the xsize to use, or 0 if
64: % natural scaling is to be used. Common uses include:
65: %
66: % \epsfxsize % just leave the old value alone
67: % 0pt % use the natural sizes
68: % #1 % use the natural sizes
69: % \hsize % scale to full width
70: % 0.5#1 % scale to 50% of natural size
71: % \ifnum#1>\hsize\hsize\else#1\fi % smaller of natural, hsize
72: %
73: % If you want TeX to report the size of the figure (as a message
74: % on your terminal when it processes each figure), say `\epsfverbosetrue'.
75: %
76: \newread\epsffilein % file to \read
77: \newif\ifepsffileok % continue looking for the bounding box?
78: \newif\ifepsfbbfound % success?
79: \newif\ifepsfverbose % report what you're making?
80: \newif\ifepsfdraft % use draft mode?
81: \newdimen\epsfxsize % horizontal size after scaling
82: \newdimen\epsfysize % vertical size after scaling
83: \newdimen\epsftsize % horizontal size before scaling
84: \newdimen\epsfrsize % vertical size before scaling
85: \newdimen\epsftmp % register for arithmetic manipulation
86: \newdimen\pspoints % conversion factor
87: %
88: \pspoints=1bp % Adobe points are `big'
89: \epsfxsize=0pt % Default value, means `use natural size'
90: \epsfysize=0pt % ditto
91: %
92: \def\epsfbox#1{\global\def\epsfllx{72}\global\def\epsflly{72}%
93: \global\def\epsfurx{540}\global\def\epsfury{720}%
94: \def\lbracket{[}\def\testit{#1}\ifx\testit\lbracket
95: \let\next=\epsfgetlitbb\else\let\next=\epsfnormal\fi\next{#1}}%
96: %
97: \def\epsfgetlitbb#1#2 #3 #4 #5]#6{\epsfgrab #2 #3 #4 #5 .\\%
98: \epsfsetgraph{#6}}%
99: %
100: \def\epsfnormal#1{\epsfgetbb{#1}\epsfsetgraph{#1}}%
101: %
102: \def\epsfgetbb#1{%
103: %
104: % The first thing we need to do is to open the
105: % PostScript file, if possible.
106: %
107: \openin\epsffilein=#1
108: \ifeof\epsffilein\errmessage{I couldn't open #1, will ignore it}\else
109: %
110: % Okay, we got it. Now we'll scan lines until we find one that doesn't
111: % start with %. We're looking for the bounding box comment.
112: %
113: {\epsffileoktrue \chardef\other=12
114: \def\do##1{\catcode`##1=\other}\dospecials \catcode`\ =10
115: \loop
116: \read\epsffilein to \epsffileline
117: \ifeof\epsffilein\epsffileokfalse\else
118: %
119: % We check to see if the first character is a % sign;
120: % if not, we stop reading (unless the line was entirely blank);
121: % if so, we look further and stop only if the line begins with
122: % `%%BoundingBox:'.
123: %
124: \expandafter\epsfaux\epsffileline:. \\%
125: \fi
126: \ifepsffileok\repeat
127: \ifepsfbbfound\else
128: \ifepsfverbose\message{No bounding box comment in #1; using defaults}\fi\fi
129: }\closein\epsffilein\fi}%
130: %
131: % Now we have to calculate the scale and offset values to use.
132: % First we compute the natural sizes.
133: %
134: \def\epsfclipon{\def\epsfclipstring{ clip}}%
135: \def\epsfclipoff{\def\epsfclipstring{\ifepsfdraft\space clip\fi}}%
136: \epsfclipoff
137: %
138: \def\epsfsetgraph#1{%
139: \epsfrsize=\epsfury\pspoints
140: \advance\epsfrsize by-\epsflly\pspoints
141: \epsftsize=\epsfurx\pspoints
142: \advance\epsftsize by-\epsfllx\pspoints
143: %
144: % If `epsfxsize' is 0, we default to the natural size of the picture.
145: % Otherwise we scale the graph to be \epsfxsize wide.
146: %
147: \epsfxsize\epsfsize\epsftsize\epsfrsize
148: \ifnum\epsfxsize=0 \ifnum\epsfysize=0
149: \epsfxsize=\epsftsize \epsfysize=\epsfrsize
150: \epsfrsize=0pt
151: %
152: % We have a sticky problem here: TeX doesn't do floating point arithmetic!
153: % Our goal is to compute y = rx/t. The following loop does this reasonably
154: % fast, with an error of at most about 16 sp (about 1/4000 pt).
155: %
156: \else\epsftmp=\epsftsize \divide\epsftmp\epsfrsize
157: \epsfxsize=\epsfysize \multiply\epsfxsize\epsftmp
158: \multiply\epsftmp\epsfrsize \advance\epsftsize-\epsftmp
159: \epsftmp=\epsfysize
160: \loop \advance\epsftsize\epsftsize \divide\epsftmp 2
161: \ifnum\epsftmp>0
162: \ifnum\epsftsize<\epsfrsize\else
163: \advance\epsftsize-\epsfrsize \advance\epsfxsize\epsftmp \fi
164: \repeat
165: \epsfrsize=0pt
166: \fi
167: \else \ifnum\epsfysize=0
168: \epsftmp=\epsfrsize \divide\epsftmp\epsftsize
169: \epsfysize=\epsfxsize \multiply\epsfysize\epsftmp
170: \multiply\epsftmp\epsftsize \advance\epsfrsize-\epsftmp
171: \epsftmp=\epsfxsize
172: \loop \advance\epsfrsize\epsfrsize \divide\epsftmp 2
173: \ifnum\epsftmp>0
174: \ifnum\epsfrsize<\epsftsize\else
175: \advance\epsfrsize-\epsftsize \advance\epsfysize\epsftmp \fi
176: \repeat
177: \epsfrsize=0pt
178: \else
179: \epsfrsize=\epsfysize
180: \fi
181: \fi
182: %
183: % Finally, we make the vbox and stick in a \special that dvips can parse.
184: %
185: \ifepsfverbose\message{#1: width=\the\epsfxsize, height=\the\epsfysize}\fi
186: \epsftmp=10\epsfxsize \divide\epsftmp\pspoints
187: \vbox to\epsfysize{\vfil\hbox to\epsfxsize{%
188: \ifnum\epsfrsize=0\relax
189: \special{PSfile=\ifepsfdraft psdraft.ps\else#1\fi\space
190: llx=\epsfllx\space lly=\epsflly\space
191: urx=\epsfurx\space ury=\epsfury\space rwi=\number\epsftmp
192: \epsfclipstring}%
193: \else
194: \epsfrsize=10\epsfysize \divide\epsfrsize\pspoints
195: \special{PSfile=\ifepsfdraft psdraft.ps\else#1\fi\space
196: llx=\epsfllx\space lly=\epsflly\space
197: urx=\epsfurx\space ury=\epsfury\space rwi=\number\epsftmp\space
198: rhi=\number\epsfrsize \epsfclipstring}%
199: \fi
200: \hfil}}%
201: \global\epsfxsize=0pt\global\epsfysize=0pt}%
202: %
203: % We still need to define the tricky \epsfaux macro. This requires
204: % a couple of magic constants for comparison purposes.
205: %
206: {\catcode`\%=12 \global\let\epsfpercent=%\global\def\epsfbblit{%BoundingBox}}%
207: %
208: % So we're ready to check for `%BoundingBox:' and to grab the
209: % values if they are found.
210: %
211: \long\def\epsfaux#1#2:#3\\{\ifx#1\epsfpercent
212: \def\testit{#2}\ifx\testit\epsfbblit
213: \epsfgrab #3 . . . \\%
214: \epsffileokfalse
215: \global\epsfbbfoundtrue
216: \fi\else\ifx#1\par\else\epsffileokfalse\fi\fi}%
217: %
218: % Here we grab the values and stuff them in the appropriate definitions.
219: %
220: \def\epsfempty{}%
221: \def\epsfgrab #1 #2 #3 #4 #5\\{%
222: \global\def\epsfllx{#1}\ifx\epsfllx\epsfempty
223: \epsfgrab #2 #3 #4 #5 .\\\else
224: \global\def\epsflly{#2}%
225: \global\def\epsfurx{#3}\global\def\epsfury{#4}\fi}%
226: %
227: % We default the epsfsize macro.
228: %
229: \def\epsfsize#1#2{\epsfxsize}
230: %
231: % Finally, another definition for compatibility with older macros.
232: %
233:
234:
235: \let\epsffile=\epsfbox
236:
237:
238:
239: \newtheorem{theorem}{Theorem}[section]
240: \newtheorem{lemma}[theorem]{Lemma}
241: \newtheorem{corollary}[theorem]{Corollary}
242: \newtheorem{proposition}[theorem]{Proposition}
243: \newtheorem{example}[theorem]{Example}
244: \newtheorem{definition}[theorem]{Definition}
245: \newtheorem{sublemma}[theorem]{Sublemma}
246: %%---%%---%%---%%---%%---%%
247:
248: \newcommand{\sqt}{\mbox{ $\qed \hspace{-9.1pt} \raisebox{1.1pt}{$\times$} $} }
249: \newcommand{\Hom}{\mbox{\rm Hom}}
250: \newcommand{\lk}{\mbox{\rm Link}}
251: \newcommand{\tlk}{\mbox{\rm Tlk}}
252: \newcommand{\dlk}{\mbox{\rm Dlk}}
253: \newcommand{\glk}{\mbox{\rm Glk}}
254: \newcommand{\h}{\mbox{\rm H}}
255: \newcommand{\Int}{\mbox{\rm Int}}
256: \newcommand{\qed}{\hbox{\rule{6pt}{6pt}}}
257:
258: %%---%%---%%---%%---%%---%%
259:
260:
261: \newcommand{\R}{\mathbb{R}}
262: \newcommand{\Q}{\mathbb{Q}}
263: \newcommand{\Z}{\mathbb{Z}}
264: \newcommand{\N}{\mathbb{N}}
265: \newcommand{\C}{\mathbb{C}}
266: \newcommand{\sphere} {\mathbb{S}}
267: \newcommand{\T}{\mathbb{T}}
268:
269:
270: %%---%%---%%---%%---%%---%%
271:
272: \newcommand{\cmapright}[2]{
273: \smash{\mathop{
274: \hbox to 1cm{\rightarrowfill}}\limits^{#1}_{#2}}}
275:
276: \newcommand{\cmapleft}[2]{
277: \smash{\mathop{
278: \hbox to 1cm{\leftarrowfill}}\limits^{#1}_{#2}}}
279:
280: \newcommand{\cmapdown}[2]
281: {\Big\downarrow
282: \llap{$\vcenter{\hbox{$\scriptstyle#1\ \ $}}$}
283: \rlap{$\vcenter{\hbox{$\scriptstyle#2$}}$ }}
284:
285: \newcommand{\cmapup}[2]
286: {\Big\uparrow
287: \llap{$\vcenter{\hbox{$\scriptstyle#1\ \ $}}$}
288: \rlap{$\vcenter{\hbox{$\scriptstyle#2$}}$ }}
289:
290:
291: \begin{document}
292:
293:
294: \title{Bordism of Unoriented Surfaces in $4$-Space}
295:
296:
297: \author{
298: J. Scott Carter \\
299: University of South Alabama \\
300: Mobile, AL 36688 \\ carter@mathstat.usouthal.edu
301: \and
302: Seiichi Kamada \\
303: Osaka City University \\
304: Osaka 558-8585, JAPAN\\
305: kamada@sci.osaka-cu.ac.jp
306: \and
307: Masahico Saito \\
308: University of South Florida \\
309: Tampa, FL 33620 \\ saito@math.usf.edu
310: \and
311: Shin Satoh \\
312: Kyoto University \\
313: Kyoto 606-8502, JAPAN\\
314: satoh@kurims.kyoto-u.ac.jp
315: }
316: \maketitle
317:
318: %%---%%---%%---%%---%%---%%
319:
320:
321:
322: \begin{abstract}
323: The group of bordism classes of
324: unoriented surfaces
325: in $4$-space
326: is determined.
327: The bordism classes are characterized
328: by normal Euler numbers, double linking numbers,
329: and triple linking numbers.
330: \end{abstract}
331:
332:
333: %%---%%---%%---%%---%%---%%
334: %%---%%---%%---%%---%%---%%
335:
336: \section{Introduction}
337:
338:
339: B. J. Sanderson \cite{Sanderson1, Sanderson2}
340: studied
341: the group $L_{m,n}$
342: of bordism classes of
343: `oriented' closed $(m-2)$-manifolds
344: of $n$ components in ${\bf R}^m$.
345: He showed that
346: $L_{m,n}$ is isomorphic to the homotopy group
347: $\pi_m\left(\bigvee_{i=1}^{n-1} S^2\right)$;
348: in particular,
349: the bordism group $L_{m,n}$ for $m=4$ is given
350: as follows.
351:
352: %%---%%---%%---%%---%%---%%
353: \begin{theorem}[Sanderson]
354: \begin{eqnarray*}
355: L_{4,n}\cong
356: (\underbrace{{\bf Z}_2\oplus\dots
357: \oplus{\bf Z}_2}_{\frac{n(n-1)}{2}})
358: \oplus
359: (\underbrace{{\bf Z} \oplus\dots
360: \oplus{\bf Z}}_{\frac{n(n-1)(n-2)}{3}}).
361: \end{eqnarray*}
362: In particular, we have
363: $L_{4,1}\cong\{0\}$,
364: $L_{4,2}\cong{\bf Z}_2$, and
365: $L_{4,3}\cong{\bf Z}_2^3\oplus{\bf Z}^2$.
366: \end{theorem}
367: %%---%%---%%---%%---%%---%%
368:
369:
370: Similarly,
371: there is a group of
372: bordism classes of `unoriented' closed $(m-2)$-manifolds
373: of $n$ components in ${\bf R}^m$.
374: We denote the group by $UL_{m,n}$.
375: The aim of this paper is
376: to determine
377: the bordism group $UL_{m,n}$ for $m=4$
378: via purely geometric techniques.
379:
380: An {\it $n$-component surface-link} $F=F_1\cup\dots\cup F_n$
381: is a disjoint union of closed surfaces embedded in ${\bf R}^4$
382: (smoothly, or PL and locally flatly) such that each component
383: $F_i$, $i=1, \ldots, n$, is an
384: orientable or non-orientable embedded surface,
385: which may or may not be connected.
386: Two $n$-component surface-links $F$ and $F'$ are
387: {\it unorientedly bordant}
388: if there is a compact $3$-manifold
389: $W=\cup_{i=1}^n W_i$ properly embedded in
390: ${\bf R}^4\times[0,1]$
391: such that
392: $\partial W_i=F_i\times\{0\}\cup F_i'\times\{1\}$
393: for $i=1,\dots,n$.
394: In this paper,
395: $F\simeq_B F'$ means that $F$ and $F'$ are
396: unorientedly bordant, and
397: $F\cong_A F'$ means that they are ambient isotopic
398: in ${\bf R}^4$.
399: The unoriented bordism classes of
400: $n$-component surface-links form
401: an abelian group $UL_{4,n}$ such that
402: the sum $[F]+[F']$ is defined
403: to be the class $[F\amalg F']$ of
404: the split union $F\amalg F'$.
405: The identity is represented by the empty $F=\emptyset$
406: and the inverse $-[F]$ is
407: represented by the mirror image of $F$.
408: The following is our main theorem.
409:
410:
411:
412: %%---%%---%%---%%---%%---%%
413:
414: \begin{theorem} \label{thm:mainA}
415: \begin{eqnarray*}
416: UL_{4,n}\cong
417: (\underbrace{{\bf Z}\oplus\dots
418: \oplus{\bf Z}}_{n})
419: \oplus
420: (\underbrace{{\bf Z}_4\oplus\dots
421: \oplus{\bf Z}_4}_{\frac{n(n-1)}{2}})
422: \oplus
423: (\underbrace{{\bf Z}_2 \oplus\dots
424: \oplus{\bf Z}_2}_{\frac{n(n-1)(n-2)}{3}}).
425: \end{eqnarray*}
426: In particular,
427: we have
428: $UL_{4,1}\cong{\bf Z}$,
429: $UL_{4,2}\cong{\bf Z}^2\oplus{\bf Z}_4$,
430: and
431: $UL_{4,3}\cong{\bf Z}^3
432: \oplus{\bf Z}_4^3
433: \oplus{\bf Z}_2^2$.
434:
435: \end{theorem}
436:
437: %%---%%---%%---%%---%%---%%
438:
439:
440:
441:
442:
443:
444: This paper is organized as follows.
445: In \S$2$,
446: we give definitions of
447: three kinds of unoriented bordism invariants ---
448: normal Euler numbers, double linkings,
449: and triple linkings ---
450: by the projection method.
451: In \S$3$,
452: we study $1$-component surface-links.
453: In \S$4$,
454: we introduce a family of surface-links
455: which are called {\it necklaces}.
456: \S$5$ is devoted to study
457: crossing changes
458: that produce necklaces.
459: In \S$6$, we prove Theorem $1.2$.
460:
461:
462: %%---%%---%%---%%---%%---%%
463: %%---%%---%%---%%---%%---%%
464:
465: \section{Unoriented Bordism Invariants}
466:
467:
468: All of our bordism invariants will be defined using the diagram of a knotted or linked surface. We begin by recalling this notion.
469: Consider
470: a surface-link $F$.
471: We may assume that
472: the restriction
473: $\pi|_F:F\rightarrow{\bf R}^3$
474: of a projection
475: $\pi:{\bf R}^4\rightarrow{\bf R}^3$
476: is a generic map,
477: that is, the singularity set of
478: the image $\pi(F)\subset{\bf R}^3$ consists of
479: double points and isolated branch/triple points.
480: See Figure 1.
481: The closure of the self-intersection set on $\pi(F)$
482: is regarded as
483: a union of immersed arcs and loops,
484: which we call double curves.
485: Branch points (or Whitney umbrella points) occur at
486: the end of the double curves,
487: and triple points occur when double curves intersect.
488:
489:
490:
491: %%---%%---%%---%%---%%---%%
492: \begin{figure}[htb]
493: \begin{center}
494: \mbox{
495: \epsfxsize=3in
496: \epsfbox{NLkH01.eps}
497: }
498: \end{center}
499: \caption{Generic intersections of surfaces in 3-space}
500: %\vspace{3.5cm}
501: %\hspace{3.3cm}
502: %\special{illustration NLkH01.eps scaled 1000}
503: %\begin{center}
504: %Figure 1
505: %\end{center}
506: \end{figure}
507: %%---%%---%%---%%---%%---%%
508:
509:
510:
511: A {\it surface diagram} of $F$
512: is the image $\pi(F)$ equipped with
513: over/under-information along each double curve
514: with respect to the projection direction.
515: To indicate such over/under-information,
516: we remove a neighborhood of
517: a double curve on the sheet
518: (the {\it lower sheet})
519: which lies lower than the other sheet
520: (the {\it upper sheet}).
521: See Figure 2.
522: Notice that the removal of this neighborhood is merely a convention in depicting illustrations. In particular, we still speak of ``double curves'' and triple points, and we locally parametrize the surfaces using immersions.
523:
524:
525:
526: There are seven kinds of local moves on surface diagrams,
527: called {\it Roseman moves},
528: analogues of Reidemeister moves for classical knots,
529: which are sufficient to
530: relate diagrams of
531: ambient isotopic surface-links
532: (cf. \cite{CS:Reidemeister, CS:book, Roseman}).
533: Specifically, two diagrams represent ambiently isotopic surface-links if and only if one can be obtained from the other by means a finite sequence of moves taken from the list of Roseman moves. We remark that the Roseman moves are used here only in relation their affect on the cobordism invariants.
534:
535:
536: %%---%%---%%---%%---%%---%%
537: \begin{figure}[htb]
538: \begin{center}
539: \mbox{
540: \epsfxsize=4in
541: \epsfbox{NLkH02.eps}
542: }
543: \end{center}
544: \caption{The broken surface diagrams at intersection points}
545: %\vspace{3.5cm}
546: %\hspace{2cm}
547: %\special{illustration NLkH02.eps scaled 1000}
548: %\begin{center}
549: %Figure 2
550: %\end{center}
551: \end{figure}
552: %%---%%---%%---%%---%%---%%
553:
554:
555:
556: For a surface-knot $K$
557: (that is a connected closed surface
558: embedded in ${\bf R}^4$),
559: H. Whitney defined the {\it normal Euler number} $e(K)$
560: of $K$
561: to be the
562: Euler number
563: of a tubular neighborhood of $K$ in ${\bf R}^4$
564: considered as a $2$-plane bundle
565: (cf. \cite{Massey, Whitney}).
566: It is known that
567:
568: %%---%%---%%---%%---%%---%%
569: \begin{itemize}
570: \setlength{\itemsep}{-3pt}
571: \item[{\rm (i)}]
572: $e(K)=0$ if $K$ is orientable,
573:
574: \item[{\rm (ii)}] (Whitney's congruence)
575: $e(K)\equiv 2\chi(K)$ (mod $4$), and
576:
577: \item[{\rm (iii)}] (Whitney-Massey
578: theorem)
579: $|e(K)|\leq 4-2\chi(K)$,
580: \end{itemize}
581: %%---%%---%%---%%---%%---%%
582:
583: \noindent
584: where $\chi(K)$ denotes
585: the Euler characteristic of $K$
586: (cf. \cite{Kamada, Massey}).
587: For a $1$-component surface-link $F$,
588: we define the normal Euler number $e(F)$ of $F$
589: to be
590: the sum of $e(K)$
591: for the connected components $K$ of $F$.
592: The normal Euler number
593: $e(F)$ can be calculated
594: by use of a projection of $F$ in ${\bf R}^3$;
595: it is equal to the number of positive type branch points
596: (see Figure~$2$) minus that of negative type ones,
597: \cite{CS:branch}.
598:
599:
600:
601:
602: Let $F=F_1\cup F_2$
603: be a $2$-component surface-link
604: and $D$ a surface diagram of $F$.
605: A double curve of $D$ is said to be
606: {\it of type $(i,j)$}
607: if the upper sheet
608: belongs to
609: $F_i$
610: and the lower
611: belongs to
612: $F_j$,
613: where $i,j\in\{1,2\}$.
614: If a double curve is an immersed arc,
615: then its endpoints are branch points, and hence,
616: the type is $(1,1)$ or $(2,2)$.
617: Let $C=c_1\cup\dots\cup c_m$
618: be the set of the double curves of type $(1,2)$
619: on the surface diagram $D$.
620: Each double curve $c_i$ is
621: an immersed loop in ${\bf R}^3$.
622:
623:
624:
625: We take
626: a $2$-disk $B^2$ and a union of intervals $X$
627: in ${\bf R}^2$ as follows:
628:
629: %%---%%---%%---%%---%%---%%
630: \begin{itemize}
631: \setlength{\itemsep}{-3pt}
632: \item[{\rm (i)}]
633: $B^2=\{(x,y)|x^2+y^2\leq 1\}$, and
634:
635: \item[{\rm (ii)}]
636: $X=\{(x,y)|-1\leq x\leq 1,y=0\}\cup
637: \{(x,y)|x=0,-1\leq y\leq 1\}$.
638: \end{itemize}
639: %%---%%---%%---%%---%%---%%
640:
641: \noindent
642: For a regular neighborhood $N(c_i)$ of $c_i$
643: in ${\bf R}^3$,
644: the pair $(N(c_i),D\cap N(c_i))$
645: is regarded as the image of
646: an immersion, say $\varphi$, of one of
647: the following manifold pairs:
648:
649: %%---%%---%%---%%---%%---%%
650: \begin{itemize}
651: \setlength{\itemsep}{-3pt}
652: \item[{\rm (i)}]
653: $(B^2,X)\times [0,1]/
654: (x,0)\sim (x,1)$ for $x\in B^2$, and
655:
656: \item[{\rm (ii)}]
657: $(B^2,X)\times [0,1]/
658: (x,0)\sim (-x,1)$ for $x\in B^2$.
659: \end{itemize}
660: %%---%%---%%---%%---%%---%%
661:
662:
663: \noindent
664: Let $c_i'$ be a loop or a pair of loops
665: immersed in $N(c_i)$
666: such that
667: $c_i'=\varphi\Bigl(\{z,-z\}\times [0,1]/\sim\Bigr)$
668: for some $z\in B^2\setminus X$.
669: We give orientations to $c_i$
670: and $c_i'$ such that $[c_i'] = 2[c_i] \in
671: H_1(N(c_i); {\bf Z})$.
672: See Figure $3$.
673: We put $C'=c_1'\cup\dots\cup c_m'$.
674: Since $C$ and $C'$ are mutually disjoint
675: 1-cycles in ${\bf R}^3$,
676: the linking number Lk$(C,C')$
677: between $C$ and $C'$ is defined;
678: let $d$ be a 2-cycle in ${\bf R}^3$
679: with $\partial d =C$ such that $d$ and $C'$
680: intersect transversely. Then Lk$(C, C')$
681: is the algebraic intersection number of
682: $C'$ with $d$.
683: This number is well-defined modulo $4$;
684: it does not depend on a choice of
685: $z \in B^2 \setminus X$ and an orientation
686: of $c_i$ for each $i$.
687: Furthermore,
688: its congruence class modulo 4 remains unchanged under the
689: Roseman moves.
690: Hence, the mod $4$ reduction of Lk$(C,C')$
691: is an ambient isotopy invariant of
692: $F=F_1\cup F_2$.
693:
694:
695: %%---%%---%%---%%---%%---%%
696: \begin{figure}[htb] \begin{center}
697: \mbox{
698: \epsfxsize=5in
699: \epsfbox{NLkH03.eps}
700: }
701: \end{center}
702: \caption{The double curve push-offs}
703: %\vspace{3.5cm}
704: %\hspace{3cm}
705: %\special{illustration NLkH03.eps scaled 1000}
706: %\begin{center}
707: %Figure $3$
708: %\end{center}
709: \end{figure}
710: %%---%%---%%---%%---%%---%%
711:
712:
713:
714:
715: %%---%%---%%---%%---%%---%%
716: \begin{definition}
717: {\rm The {\it double linking number}
718: between $F_1$ and $F_2$,
719: denoted by $d(F_1,F_2)$,
720: is a value in ${\bf Z}_4={\bf Z}/4{\bf Z}=\{0,1,2,3\}$
721: that is the linking number
722: ${\rm Lk}(C,C')$ modulo $4$. }
723: \end{definition}
724: %%---%%---%%---%%---%%---%%
725:
726: It is proved later that the double linking number
727: is asymmetric; $d(F_1,F_2)=-d(F_2,F_1)$.
728:
729:
730:
731:
732: Let $F=F_1\cup F_2\cup F_3$
733: be a $3$-component surface-link.
734: At a triple point on a surface diagram of $F$,
735: there are three sheets
736: called {\it top}, {\it middle}, and {\it bottom}
737: with respect to the projection direction.
738: A triple point is
739: {\it of type $(i,j,k)$}
740: if the top sheet comes from $F_i$,
741: the middle comes from $F_j$,
742: and the bottom comes from $F_k$,
743: where $i,j,k\in\{1,2,3\}$.
744: Let $N(i,j,k)$ denote
745: the number of the triple points of type $(i,j,k)$.
746: The mod $2$ reduction of $N(i,j,k)$
747: is preserved under Roseman moves and hence
748: an ambient isotopy invariant of $F$
749: provided $i\ne j$ and $j\ne k$
750: (possibly $i=k$), \cite{Satoh}.
751:
752:
753:
754: %%---%%---%%---%%---%%---%%
755: \begin{definition}
756: {\rm The {\it triple linking number}
757: among $F_i,F_j$, and $F_k$,
758: denoted by $t(F_i,F_j,F_k)$,
759: is a value in ${\bf Z}_2={\bf Z}/2{\bf Z}=\{0,1\}$
760: that is the number $N(i,j,k)$ modulo $2$
761: provided $i\ne j$ and $j\ne k$.}
762: \end{definition}
763: %%---%%---%%---%%---%%---%%
764:
765:
766:
767: %%---%%---%%---%%---%%---%%
768: \begin{lemma}
769: The ambient isotopy invariants
770: $e,d$ and $t$ are
771: unoriented bordism invariants.
772: \end{lemma}
773: %%---%%---%%---%%---%%---%%
774:
775:
776:
777: {\it Proof.}
778: We take surface diagrams $D$ and $D'$
779: of surface-links $F$ and $F'$ respectively.
780: If $F$ and $F'$ are unorientedly bordant,
781: then $D'$ is obtained from $D$
782: by a finite sequence of
783: moves from the following list:
784: \begin{itemize}
785: \setlength{\itemsep}{-3pt}
786: \item an ambient isotopy of ${\bf R}^3$,
787: \item a Roseman move,
788: \item adding or deleting embedded $2$-spheres
789: in ${\bf R}^3$ which are disjoint from $D$,
790: \item a $1$-handle surgery on $D$ in ${\bf R}^3$
791: whose core is a simple arc $\gamma$ with
792: $\gamma\cap D=\partial\gamma$
793: (see Figure $4$), and
794: \item a $2$-handle surgery on $D$ in ${\bf R}^3$
795: whose core is a simple $2$-disk $\delta$
796: with $\delta\cap D=\partial\delta$.
797: \end{itemize}
798: %%---%%---%%---%%---%%---%%
799:
800: \noindent
801: Recall that Roseman moves do not change
802: the invariants $e,d$ and $t$.
803: The other deformations listed above
804: do not change the singularity set
805: of the diagram.
806: Hence $e,d$ and $t$ are unoriented bordism invariants.
807: \ \qed
808:
809:
810: %%---%%---%%---%%---%%---%%
811: \begin{figure}[htb] \begin{center}
812: \mbox{
813: \epsfxsize=4.5in
814: \epsfbox{NLkH04.eps}
815: }
816: \end{center}
817: \caption{Attaching $1$ and $2$-handles}
818: %\vspace{3.5cm}
819: %\hspace{2.7cm}
820: %\special{illustration NLkH04.eps scaled 1000}
821: %\begin{center}
822: %Figure $4$
823: %\end{center}
824: \end{figure}
825: %%---%%---%%---%%---%%---%%
826:
827:
828: %%---%%---%%---%%---%%---%%
829: %%---%%---%%---%%---%%---%%
830:
831: \section{1-Component Surface-Links}
832:
833:
834: A projective plane embedded
835: in ${\bf R}^4$ is {\it standard}
836: if it has a surface diagram
837: as shown in Figure $5$.
838: A non-orientable surface-knot is
839: said to be {\it trivial}
840: if it is a connected sum of some
841: standard projective planes in ${\bf R}^4$.
842: Two
843: trivial non-orientable surface-knots $F$ and $F'$
844: are ambient isotopic if and only if
845: $e(F)=e(F')$ and $\chi(F)=\chi(F')$.
846: The following lemma is folklore.
847:
848:
849: %%---%%---%%---%%---%%---%%
850: \begin{figure}[htb] \begin{center}
851: \mbox{
852: \epsfxsize=3in
853: \epsfbox{NLkH05.eps}
854: }
855: \end{center}
856: \caption{The positive and negative projective planes}
857: %\vspace{3.5cm}
858: %\hspace{4.3cm}
859: %\special{illustration NLkH05.eps scaled 1000}
860: %\begin{center}
861: %Figure $5$
862: %\end{center}
863: \end{figure}
864: %%---%%---%%---%%---%%---%%
865:
866:
867: %%---%%---%%---%%---%%---%%
868: \begin{lemma}
869: Two $1$-component surface-links
870: $F$ and $F'$ are unorientedly bordant
871: if and only if $e(F)=e(F')$.
872: \end{lemma}
873: %%---%%---%%---%%---%%---%%
874:
875:
876:
877: {\it Proof.}
878: The only if part is obvious.
879: We prove the if part.
880: It is known that
881: any non-orientable surface-link
882: is transformed into
883: a trivial non-orientable surface-knot
884: by some $1$-handle surgeries,
885: \cite{Kamada}.
886: Thus we may assume that
887: $F$ and $F'$ are trivial non-orientable surface-knots
888: with $e(F)=e(F')$.
889: By Whitney's congruence,
890: we have $\chi(F)\equiv \chi(F')$ (mod~$2$).
891: Doing $1$-handle surgeries if necessary,
892: we may assume that $\chi(F)=\chi(F')$.
893: Then $F$ and $F'$ are ambient isotopic.
894: Thus original $1$-component surface-links
895: $F$ and $F'$ are unorientedly bordant.
896: \ \qed
897:
898:
899:
900: %%---%%---%%---%%---%%---%%
901: %%---%%---%%---%%---%%---%%
902:
903: \section{Necklaces}
904:
905:
906: We introduce a family of surface-links,
907: called {\it necklaces},
908: which is used to prove Theorem $1.2$.
909: In the upper $3$-space
910: ${\bf R}^3_+=\{(x,y,z)|z\geq 0\}$,
911: we take a $3$-ball
912: $$B^3=\{(x,y,z)|x^2+y^2+(z-2)^2\leq 1\}.$$
913: Let $f=\{f_t\}_{0\leq t\leq 1}$
914: and $g=\{g_t\}_{0\leq t\leq 1}$
915: be ambient isotopies of $B^3$
916: which present
917: a $180^\circ$-rotation around the $z$-axis and
918: a $360^\circ$-rotation around the axis
919: $(y=0,z=2)$ respectively.
920: We put a Hopf link $k_1\cup k_2$ in $B^3$
921: as in Figure $6$
922: so that $f_1(k_i)=k_i$
923: and $g_1(k_i)=k_i$ for $i=1,2$.
924: By a {\it motion} of $k_1\cup k_2$,
925: we mean an ambient isotopy $h=\{h_t\}_{0\leq t\leq 1}$
926: of $B^3$ with $h_1(k_1\cup k_2)=k_1\cup k_2$.
927: Two motions $h$ and $h'$ are {\it equivalent}
928: if there is a $1$-parameter family
929: of motions between $h$ and $h'$.
930:
931:
932:
933:
934: %%---%%---%%---%%---%%---%%
935: \begin{figure}[htb] \begin{center}
936: \mbox{
937: \epsfxsize=2in
938: \epsfbox{NLkH06.eps}
939: }
940: \end{center}
941: \caption{Spinning the Hopf link in two directions}
942: %\vspace{5cm}
943: %\hspace{5.8cm}
944: %\special{illustration NLkH06.eps scaled 1000}
945: %\begin{center}
946: %Figure $6$
947: %\end{center}
948: %\vspace{-0.5cm}
949: %
950: \end{figure}
951: %%---%%---%%---%%---%%---%%
952:
953:
954: We consider that ${\bf R}^4$ is obtained by
955: spinning ${\bf R}^3_+$ around $\partial{\bf R}^3_+$
956: by use of a map
957: $\mu:{\bf R}^3_+\times [0,1]\rightarrow {\bf R}^4$
958: defined by
959: $\Bigl((x,y,z) \times \{ t \}\Bigr)\mapsto (x,y,z\cos 2\pi t,z\sin 2\pi t)$.
960: For integers $p,q$,
961: we construct a $2$-component surface-link
962: $S^{p,q}=T_1\cup T_2$,
963: called a {\it strand}, as follows:
964: $$T_i=\mu\left(\bigcup_{t\in[0,1]}
965: \Bigl(f^p\cdot g^q\Bigr)_t(k_i)\times\{t\}\right)
966: \subset{\bf R}^4 \ (i=1,2).$$
967:
968:
969: \noindent
970: Each $T_i$ $(i=1,2)$ is homeomorphic to
971: a torus (or a Klein bottle, respectively)
972: if $p$ is even (or odd).
973:
974:
975: %%---%%---%%---%%---%%---%%
976: \begin{lemma}
977: {\rm (i)} Strands $S^{p,q}$ and $S^{p+2q,0}$ are
978: ambient isotopic.
979:
980: {\rm (ii)} If $p\equiv p'$ {\rm (mod $4$)},
981: then two strands $S^{p,q}$ and $S^{p',q}$ are
982: ambient isotopic.
983: \end{lemma}
984: %%---%%---%%---%%---%%---%%
985:
986:
987: {\it Proof.}
988: By the belt trick (cf. \cite{Kauffman}),
989: the motion $g$ is equivalent to $f^2$,
990: and
991: $f^4$ is equivalent to the identity.
992: Hence we have the result.
993: \qed
994:
995:
996: \bigskip
997:
998: The above lemma implies that ambient isotopy classes of strands are
999: represented by $S^{p,q}$ with $q=0$ and $p\in{\bf Z}_4$.
1000: We shall abbreviate $S^{p,0}$ to $S^p$.
1001:
1002:
1003: %%---%%---%%---%%---%%---%%
1004: \begin{lemma}
1005: For a strand $S^p=T_1\cup T_2$,
1006: we have
1007:
1008: {\rm (i)} $e(T_1)=e(T_2)=0\in{\bf Z}$,
1009:
1010: {\rm (ii)} $d(T_1,T_2)=-d(T_2,T_1)=p
1011: \in{\bf Z}_4$, and
1012:
1013: {\rm (iii)} $t(T_1,T_2,T_1)=t(T_2,T_1,T_2)=p
1014: \in{\bf Z}_2$.
1015: \end{lemma}
1016: %%---%%---%%---%%---%%---%%
1017:
1018:
1019: {\it Proof.}
1020: (i)
1021: For each $i=1,2$,
1022: we take a $2$-disk $D_i$ embedded in $B^3$ with
1023: $\partial D_i=k_i$ and $f_1(D_i)=D_i$.
1024: The image $\mu(D_i\times[0,1])$ is a $3$-manifold
1025: whose boundary is $T_i$.
1026: Thus $e(T_i)=0$.
1027:
1028: (ii)
1029: In Figure $7$,
1030: we illustrate the motion $f$ of the Hopf link $k_1\cup k_2$.
1031: Since we can obtain a diagram of $S^p$
1032: by taking $p$ copies of the motion and connecting them,
1033: we have $d(T_1,T_2)=p$.
1034: Similarly, we have $d(T_2,T_1)=-p$.
1035:
1036: (iii)
1037: The motion in Figure $7$ contains
1038: two Reidemeister moves of type III.
1039: One of them corresponds to a triple point
1040: of type (top, middle, bottom)$=(T_1,T_2,T_1)$ and
1041: the other corresponds to that of $(T_2,T_1,T_2)$.
1042: Thus we have $t(T_1,T_2,T_1)=t(T_2,T_1,T_2)=p$.
1043: \qed
1044:
1045: \bigskip
1046:
1047: %%---%%---%%---%%---%%---%%
1048: \begin{figure}[htb] \begin{center}
1049: \mbox{
1050: \epsfxsize=5in
1051: \epsfbox{NLkH07.eps}
1052: }
1053: \end{center}
1054: \caption{The movie of the strand $S^1$}
1055: %\vspace{4.5cm}
1056: %\hspace{1.5cm}
1057: %\special{illustration NLkH07.eps scaled 1000}
1058: %\begin{center}
1059: %Figure $7$
1060: %\end{center}
1061: \end{figure}
1062: %%---%%---%%---%%---%%---%%
1063:
1064:
1065:
1066: For $m$ numbers $\{t_i\}_{i=1,\dots,m}$
1067: with $0<t_1<\dots<t_m<1$,
1068: we consider a surface-link
1069: defined
1070: as follows:
1071: $$S^p\cup
1072: \mu\Bigl(\partial B^3\times\{t_1\}\Bigr)
1073: \cup\dots\cup
1074: \mu\Bigl(\partial B^3\times\{t_m\}\Bigr).$$
1075: We call such a surface-link a {\it necklace}
1076: and a spherical component
1077: $B_i=\mu\Bigl(\partial B^3\times\{t_i\}\Bigr)$
1078: a {\it bead} of the necklace.
1079:
1080:
1081: %%---%%---%%---%%---%%---%%
1082: \begin{lemma}
1083: Let $S^p\cup B_1\cup\dots\cup B_m$
1084: be a necklace with the strand $S^p=T_1\cup T_2$.
1085: For each $i=1,\dots,m$, we have
1086: $$\left\{
1087: \begin{array}{l}
1088: t(T_2,T_1,B_i)=t(B_i,T_1,T_2)=1, \\
1089: t(T_1,T_2,B_i)=t(B_i,T_2,T_1)=1, \\
1090: t(T_1,B_i,T_2)=t(T_2,B_i,T_1)=0.
1091: \end{array}\right.$$
1092:
1093: \end{lemma}
1094: %%---%%---%%---%%---%%---%%
1095:
1096:
1097: {\it Proof.}
1098: In a surface diagram,
1099: a bead introduces four triple points
1100: $\tau_1,\dots,\tau_4$
1101: as shown in Figure $8$.
1102: The top, middle and bottom sheets around $\tau_1$
1103: come from $T_2$, $T_1$ and $B_i$ respectively.
1104: For $\tau_2$, they are
1105: $T_1$, $T_2$ and $B_i$.
1106: For $\tau_3$, they are
1107: $B_i$, $T_1$ and $T_2$.
1108: For $\tau_4$, they are
1109: $B_i$, $T_2$ and $T_1$.
1110: \qed
1111:
1112:
1113: %%---%%---%%---%%---%%---%%
1114: \begin{figure}[htb] \begin{center}
1115: \mbox{
1116: \epsfxsize=3in
1117: \epsfbox{NLkH08.eps}
1118: }
1119: \end{center}
1120: \caption{The local picture of a bead on a strand}
1121: %\vspace{6cm}
1122: %\hspace{4.2cm}
1123: %\special{illustration NLkH08.eps scaled 1000}
1124: %\begin{center}
1125: %Figure $8$
1126: %\end{center}
1127: \end{figure}
1128: %%---%%---%%---%%---%%---%%
1129:
1130: \bigskip
1131:
1132: We denote by
1133: $N^p(i,j;k_1,\dots,k_m)$
1134: an $n$-component surface-link which is a necklace
1135: $S^p\cup B_1\cup\dots\cup B_m$
1136: with the strand $S^p=T_1\cup T_2$ and
1137: with a label
1138: $\alpha:\{T_1,T_2,B_1,\dots,B_m\}\rightarrow
1139: \{1,\dots,n\}$
1140: such that
1141: $\alpha(T_1)=i$, $\alpha(T_2)=j$, and
1142: $\alpha(B_1)=k_1,\dots,\alpha(B_m)=k_m$.
1143:
1144:
1145:
1146: %%---%%---%%---%%---%%---%%
1147: \begin{lemma}
1148: {\rm (i)}
1149: $N^p(i,i;k_1,\dots,k_m)$
1150: is unorientedly null-bordant.
1151:
1152: {\rm (ii)}
1153: $N^p(i,j;k_1,\dots,k_m)
1154: \cong_A N^{-p}(j,i;k_1,\dots,k_m)$.
1155:
1156: {\rm (iii)}
1157: $N^p(i,j;k_1,k_2,k_3,\dots,k_m)
1158: \simeq_B N^p(i,j;k_3,\dots,k_m)$
1159: provided $k_1=k_2$.
1160:
1161: {\rm (iv)}
1162: $N^p(i,j;k_1,k_2,\dots,k_m)
1163: \simeq_B N^{p+2}(i,j;k_2,\dots,k_m)$
1164: provided $k_1=i$.
1165:
1166:
1167: {\rm (v)}
1168: $N^p(i,j;k_1,\dots,k_m)\amalg
1169: N^{p'}(i,j;k_1',\dots,k_l')\simeq_B
1170: N^{p+p'}(i,j;k_1,\dots,k_m,k_1',\dots,k_l')$.
1171:
1172: {\rm (vi)}
1173: $N^0(i,j;\emptyset)$ is
1174: unorientedly null-bordant.
1175:
1176: \end{lemma}
1177: %%---%%---%%---%%---%%---%%
1178:
1179:
1180: {\it Proof.}
1181: (i)
1182: Consider an annulus $A$ in int$B^3$
1183: with $\partial A=k_1\cup k_2$
1184: and $f_1(A)=A$.
1185: The image $\mu(A\times[0,1])$ is
1186: a $3$-manifold
1187: whose boundary is the strand $S^p=T_1\cup T_2$.
1188: Thus, we can remove $S^p$
1189: and then all the beads $B_i$ $(i=1,\dots,m)$
1190: up to unoriented bordism.
1191:
1192:
1193: (ii)
1194: By a $180^{\circ}$-rotation of $B^3$
1195: around the axis $(y=0,z=2)$,
1196: the components $k_1$ and $k_2$
1197: are switched.
1198: Then the direction of the rotation $f$
1199: is reversed.
1200:
1201:
1202: (iii)
1203: We can remove the beads
1204: $B_i=\mu\Bigl(\partial B^3\times\{t_i\}\Bigr)$
1205: $(i=1,2)$ from $F$ by a $3$-manifold
1206: $\mu\Bigl(\partial B^3\times[t_1,t_2]\Bigr)$.
1207:
1208:
1209:
1210: (iv)
1211: We perform a $1$-handle surgery
1212: between $K_1$ and $B_1$
1213: as shown in Figure $9(1)\rightarrow(2)$.
1214: The diagram illustrated in $(2)$ is
1215: ambient isotopic to that in (3).
1216: This surface is realized by a motion as in $(4)$,
1217: which is the $360^\circ$-rotation around the axis
1218: $(x=0,z=2)$.
1219: This motion is equivalent to $f^2$.
1220:
1221: (v)
1222: Connect the strands of
1223: $N^p(i,j;k_1,\dots,k_m)$ and
1224: $N^{p'}(i,j;k_1',\dots,k_l')$
1225: up to unoriented bordism
1226: to obtain
1227: $N^{p+p'}(i,j;k_1,\dots,k_m,k_1',\dots,k_l')$
1228: (see Lemma $2.3$ of \cite{CKSS2}).
1229:
1230: (vi)
1231: Consider the Hopf link $k_1\cup k_2$
1232: to use the definition of a strand.
1233: Let $\gamma_i$ $(i=1,2)$
1234: be disjoint simple arcs
1235: in ${\bf R}_3^+$ connecting between
1236: $k_i$ and $\partial{\bf R}^3_+$
1237: in an obvious way.
1238: Then the $2$-handle surgeries along the $2$-disks
1239: $\mu\Bigl(\gamma_1\times[0,1]\Bigr)$ and
1240: $\mu\Bigl(\gamma_2\times[0,1]\Bigr)$ make $S^0$
1241: a split union of $2$-spheres.
1242: \qed
1243:
1244:
1245: %%---%%---%%---%%---%%---%%
1246: \begin{figure}[htb] \begin{center}
1247: \mbox{
1248: \epsfxsize=5in
1249: \epsfbox{NLkH09.eps}
1250: }
1251: \end{center}
1252: \caption{The movies of surgeries and isotopies}
1253: %\vspace{11.5cm}
1254: %\hspace{0.5cm}
1255: %\special{illustration NLkH09.eps scaled 1000}
1256: %\begin{center}
1257: %Figure $9$
1258: %\end{center}
1259: \end{figure}
1260: %%---%%---%%---%%---%%---%%
1261:
1262: %%---%%---%%---%%---%%---%%
1263: %%---%%---%%---%%---%%---%%
1264:
1265: \section{Crossing Changes}
1266:
1267:
1268: In this section,
1269: we study a crossing change
1270: along a double curve of a surface diagram.
1271: The idea is similar to a crossing change for a classical link
1272: in ${\bf R}^3$.
1273: For a classical link,
1274: a crossing change can be
1275: realized by $1$-handle(=band) surgeries
1276: as shown in Figure $10$(i).
1277: Hence, any classical link $k_1\cup\dots\cup k_n$
1278: is bordant to a split union
1279: $k_1\amalg\dots\amalg k_n\amalg($Hopf links$)$.
1280:
1281:
1282:
1283: %%---%%---%%---%%---%%---%%
1284: \begin{figure}[htb] \begin{center}
1285: \mbox{
1286: \epsfxsize=5in
1287: \epsfbox{NLkH10.eps}
1288: }
1289: \end{center}
1290: \caption{A movie of a crossing change and the resulting surface}
1291: %\vspace{3.5cm}
1292: %\hspace{1cm}
1293: %\special{illustration NLkH10.eps scaled 1000}
1294: %\begin{center}
1295: %Figure $10$
1296: %\end{center}
1297: \end{figure}
1298: %%---%%---%%---%%---%%---%%
1299:
1300: %\clearpage
1301: %%---%%---%%---%%---%%---%%
1302: \begin{lemma}
1303: Any $n$-component surface-link
1304: $F_1\cup\dots\cup F_n$ is
1305: unorientedly bordant to a split union
1306: $F_1\amalg\dots\amalg F_n\amalg({\rm necklaces})$.
1307: \end{lemma}
1308: %%---%%---%%---%%---%%---%%
1309:
1310: In a surface diagram, we use a symbol $E$
1311: (as in Figure $10$(ii)) for a
1312: local diagram which is
1313: obtained from Figure $10$(i) by
1314: regarding it a motion picture.
1315:
1316:
1317: Let $c$ be a double curve of a surface diagram
1318: such that it is an immersed loop in ${\bf R}^3$.
1319: As mentioned in Section $2$,
1320: a regular neighborhood $N(c)$ of $c$
1321: is regarded as the image of an immersion $\varphi$ of
1322: $B^2 \times [0,1]/ \sim$.
1323: See Figure $11$.
1324: For each $c$,
1325: we choose a pair of diagonal regions
1326: $Y$ of $B^2\setminus X$
1327: and put
1328: $R(c)=\varphi(Y\times[0,1]/\sim)$.
1329:
1330:
1331: %%---%%---%%---%%---%%---%%
1332: \begin{figure}[htb] \begin{center}
1333: \mbox{
1334: \epsfxsize=3in
1335: \epsfbox{NLkH11.eps}
1336: }
1337: \end{center}
1338: \caption{A neighborhood of a double curve}
1339: %\vspace{3.5cm}
1340: %\hspace{4cm}
1341: %\special{illustration NLkH11.eps scaled 1000}
1342: %\begin{center}
1343: %Figure $11$
1344: %\end{center}
1345: \end{figure}
1346: %%---%%---%%---%%---%%---%%
1347:
1348: Let $A_1$ and $A_2$ be the sheets
1349: which intersects along the curve $c$
1350: such that $A_1$ is higher than $A_2$
1351: with respect to the projection direction;
1352: that is, $c$ is of type (upper, lower)=$(A_1,A_2)$.
1353: To prove Lemma $5.1$,
1354: we introduce four kinds of local deformations
1355: under which unoriented bordism classes are preserved.
1356:
1357:
1358: {\bf (1) Local crossing change.}
1359: Consider two $1$-handle surgeries
1360: as shown in Figure $12$
1361: in the motion picture method.
1362: We call this a {\it local crossing change} along $c$.
1363: We always assume that
1364: the `local strand' (Hopf link)$\times[0,1]$
1365: obtained by a local crossing change
1366: is in the specified region $R(c)$.
1367: Let $t$ be a triple point on $c$ and
1368: $H$ a sheet which is transverse to $c$ at $t$.
1369: If $H$ is a top or bottom sheet,
1370: then we can perform a local crossing change
1371: along the curve $c$ as in Figure $13$.
1372: For example,
1373: if $H$ is a top sheet,
1374: then the type of $t$ is changed from
1375: (top, middle, bottom)=$(H,A_1,A_2)$ to
1376: $(H,A_2,A_1)$, and the local strand
1377: goes under the sheet $H$.
1378:
1379: %%---%%---%%---%%---%%---%%
1380: \begin{figure}[htb] \begin{center}
1381: \mbox{
1382: \epsfxsize=5in
1383: \epsfbox{NLkH12.eps}
1384: }
1385: \end{center}
1386: \caption{The local crossing change}
1387: %\vspace{8.5cm}
1388: %\hspace{0cm}
1389: %\special{illustration NLkH12.eps scaled 1000}
1390: %\begin{center}
1391: %Figure $12$
1392: %\end{center}
1393: \end{figure}
1394: %%---%%---%%---%%---%%---%%
1395:
1396: %%---%%---%%---%%---%%---%%
1397: \begin{figure}[htb] \begin{center}
1398: \mbox{
1399: \epsfxsize=4in
1400: \epsfbox{NLkH13.eps}
1401: }
1402: \end{center}
1403: \caption{A local crossing change near a triple point}
1404: %\vspace{3.5cm}
1405: %\hspace{1.5cm}
1406: %\special{illustration NLkH13.eps scaled 1000}
1407: %\begin{center}
1408: %Figure $13$
1409: %\end{center}
1410: \end{figure}
1411: %%---%%---%%---%%---%%---%%
1412:
1413: \clearpage
1414:
1415: {\bf (2) End-change.}
1416: Consider a composite of three ambient isotopies
1417: as shown in Figure $14$.
1418: We call this an {\it end-change} of a local strand.
1419: This deformation moves
1420: an end-part of the local strand
1421: into the diagonal region of $R(c)$.
1422: The new local strand has
1423: additional intersections with $A_1$ and $A_2$.
1424: In the bottom of Figure $14$,
1425: the boxed ``$f$'' means a local diagram
1426: corresponding to a $180^\circ$-rotation
1427: of the Hopf link.
1428:
1429:
1430:
1431:
1432: %%---%%---%%---%%---%%---%%
1433: \begin{figure}[htb] \begin{center}
1434: \mbox{
1435: \epsfxsize=5in
1436: \epsfbox{NLkH14.eps}
1437: }
1438: \end{center}
1439: \caption{The movies and diagrams determining an end change}
1440: %\vspace{12cm}
1441: %\hspace{0cm}
1442: %\special{illustration NLkH14.eps scaled 1000}
1443: %\begin{center}
1444: %\Figure $14$
1445: %\end{center}
1446: %\vspace{-10pt}
1447: \end{figure}
1448: %%---%%---%%---%%---%%---%%
1449:
1450:
1451: {\bf (3) Canceling adjacent ends.}
1452: Assume that two adjacent ends
1453: of local strands are in the same region of $R(c)$.
1454: Consider a deformation illustrated in Figure $15$,
1455: which is realized by two $2$-handle surgeries
1456: on $A_1$ and $A_2$.
1457: We call this a {\it canceling of adjacent ends}
1458: of local strands along $c$.
1459:
1460:
1461: %%---%%---%%---%%---%%---%%
1462: \begin{figure}[htb] \begin{center}
1463: \mbox{
1464: \epsfxsize=4in
1465: \epsfbox{NLkH15.eps}
1466: }
1467: \end{center}
1468: \caption{Canceling adjacent ends for a crossing change}
1469: %\vspace{3cm}
1470: %\hspace{3cm}
1471: %\special{illustration NLkH15.eps scaled 1000}
1472: %\begin{center}
1473: %Figure $15$
1474: %\end{center}
1475: %\vspace{-10pt}
1476: \end{figure}
1477: %%---%%---%%---%%---%%---%%
1478:
1479:
1480:
1481: {\bf (4) Making a bead.}
1482: Assume that a strand intersects a sheet $H$
1483: transversely.
1484: In case that the strand is under $H$,
1485: we consider a $2$-handle surgery
1486: on $H$ as shown in Figure $16$.
1487: This deformation makes the strand
1488: over $H$ with producing a bead.
1489:
1490:
1491: %%---%%---%%---%%---%%---%%
1492: \begin{figure}[htb] \begin{center}
1493: \mbox{
1494: \epsfxsize=4in
1495: \epsfbox{NLkH16.eps}
1496: }
1497: \end{center}
1498: \caption{Making a bead}
1499: %\vspace{3cm}
1500: %\hspace{4cm}
1501: %\special{illustration NLkH16.eps scaled 1000}
1502: %\begin{center}
1503: %Figure $16$
1504: %\end{center}
1505: \end{figure}
1506: %%---%%---%%---%%---%%---%%
1507:
1508:
1509: In a surface diagram of
1510: an $n$-component surface-link
1511: $F=F_1\cup\dots\cup F_n$,
1512: a double curve of type $(i,j)$
1513: is {\it preferred} if $i\leq j$,
1514: and a triple point of type $(i,j,k)$ is {\it preferred}
1515: if $i\leq j$ and $j\leq k$.
1516:
1517:
1518:
1519: \bigskip
1520:
1521: {\it Proof of Lemma $5.1$.}
1522: Consider a surface diagram of $F=F_1\cup\dots\cup F_n$.
1523: If there is a non-preferred double curve
1524: without triple points
1525: (that is an embedded loop),
1526: we apply a local crossing change along the curve
1527: followed by canceling adjacent ends of the local strand.
1528: This makes the double curve preferred and
1529: yields a strand (necklace without beads),
1530: which is separated from $F$.
1531: Thus we may assume that
1532: there is at least one triple point on
1533: any non-preferred double curve of the surface diagram.
1534:
1535:
1536:
1537: For each non-preferred triple point $t$
1538: of type $(i,j,k)$,
1539: we perform local crossing changes so that
1540: $t$ changes into a preferred triple point
1541: as follows:
1542:
1543:
1544: \begin{itemize}
1545: \setlength{\itemsep}{-3pt}
1546: \item
1547: If $j<i\leq k$, then
1548: we perform a local crossing change
1549: along the double curve of type $(i,j)$
1550: over the bottom sheet labeled $k$.
1551: \item
1552: If $i\leq k<j$, then
1553: we perform a local crossing change
1554: along the double curve of type $(j,k)$
1555: under the top sheet labeled $i$.
1556: \item
1557: If $k<i\leq j$, then
1558: we perform a local crossing change
1559: along the double curve of type $(j,k)$
1560: and then perform another
1561: along the curve of type $(i,k)$.
1562: \item
1563: If $j\leq k<i$, then
1564: we perform a local crossing change
1565: along the double curve of type $(i,j)$
1566: and then perform anther
1567: along the curve of type $(i,k)$.
1568: \item
1569: If $k<j<i$, then
1570: we perform a local crossing change
1571: along the double curve of type $(i,j)$,
1572: next along the curve of type $(i,k)$,
1573: and then along the curve of type $(j,k)$.
1574: \end{itemize}
1575:
1576:
1577:
1578: \noindent
1579: Figure $17$ shows the case of $k<j<i$.
1580:
1581:
1582: %%---%%---%%---%%---%%---%%
1583: \begin{figure}[htb] \begin{center}
1584: \mbox{
1585: \epsfxsize=5in
1586: \epsfbox{NLkH17.eps}
1587: }
1588: \end{center}
1589: \caption{A three-fold crossing change at a triple point of type $k<j< i$}
1590: %\vspace{3.3cm}
1591: %\hspace{1.3cm}
1592: %\special{illustration NLkH17.eps scaled 1000}
1593: %\begin{center}
1594: %Figure $17$
1595: %\end{center}
1596: \end{figure}
1597: %%---%%---%%---%%---%%---%%
1598:
1599: All the local crossing changes
1600: that are
1601: described above
1602: are
1603: performed
1604: along non-preferred double curves.
1605: After applying suitable end-changes,
1606: we can cancel all the adjacent ends of local strands
1607: along non-preferred double curves.
1608: Then we obtain a surface diagram of
1609: $(F_1\amalg\dots\amalg F_n)\cup($strands$)$;
1610: for any double curves between $F_i$ and $F_j$ are preferred.
1611: By making beads if necessary,
1612: we can split necklaces from $F_1\amalg\dots\amalg F_n$.
1613: Thus $F$ is unorientedly bordant to
1614: $F_1\amalg\dots\amalg F_n\amalg($necklaces$)$.
1615: \qed
1616:
1617: \bigskip
1618:
1619:
1620: %%---%%---%%---%%---%%---%%
1621: \begin{lemma}
1622: $N^0(i,j;k)\simeq_B N^0(k,i;j)\amalg N^0(k,j;i)$.
1623: \end{lemma}
1624: %%---%%---%%---%%---%%---%%
1625:
1626:
1627: {\it Proof.}
1628: Consider a surface diagram of $N^0(i,j;k)$
1629: which contains the local diagram
1630: illustrated in Figure $8$.
1631: In the local diagram,
1632: Along the double curves of type $(i,k)$ and $(j,k)$,
1633: we perform (global) crossing changes
1634: as in the proof of Lemma $5.1$.
1635: Then $N^0(i,j;k)$ is unorientedly bordant to
1636: a split union of $N^0(i,k;j)\cong_A N^0(k,i;j)$
1637: along the curve of type $(i,k)$,
1638: $N^0(j,k;i)\cong_A N^0(k,j;i)$ along the curve of
1639: type $(j,k)$, and a surface-link $F$
1640: obtained by the crossing change.
1641: Then $F$ is a split union of $S^0(i,j)$
1642: and a trivial $2$-sphere labeled $k$,
1643: which is unorientedly null-bordant
1644: by Lemma $4.4$(vi).
1645: \qed
1646:
1647: %%---%%---%%---%%---%%---%%
1648: %%---%%---%%---%%---%%---%%
1649:
1650: \section{Proof of Theorem 1.2}
1651:
1652:
1653: Let $P_+$ and $P_-$
1654: be the standard projective plane in ${\bf R}^4$
1655: with $e(P_+)=2$ and $e(P_-)=-2$ respectively.
1656: See Figure $5$.
1657: We denote by $P^m$
1658: the connected sum of $m$ copies of
1659: $P_+$ if $m>0$,
1660: $-m$ copies of $P_-$ if $m<0$,
1661: and the empty if $m=0$.
1662: Regarding $P^m$ as an $n$-component surface-link
1663: with a label $\alpha(P^m)=i$,
1664: we denote it by $P^m(i)$.
1665: Regarding a strand $S^p=T_1\cup T_2$
1666: as an $n$-component surface-link
1667: such that $\alpha(T_1)=i$ and $\alpha(T_2)=j$,
1668: we denote it by $S^p(i,j)$.
1669: Then $S^p(i,j)=N^p(i,j;\emptyset)$
1670: in the notation used in Lemma $4.4$.
1671:
1672:
1673:
1674: %%---%%---%%---%%---%%---%%
1675: \begin{lemma}
1676: Any $n$-component surface-link $F$ is
1677: unorientedly bordant to a split union
1678: $$\Biggl(\coprod_{i\in\Gamma_1} P^{m_i}(i)\Biggr)\coprod
1679: \Biggl(\coprod_{(i,j)\in\Gamma_2} S^{p_{ij}}(i,j)\Biggr)\coprod
1680: \Biggl(\coprod_{(i,j,k)\in\Gamma_3} N^0(i,j;k)\Biggr).$$
1681: Here,
1682: $\Gamma_i$ $(i=1,2,3)$
1683: is a subset of the $i$-fold Cartesian product
1684: of $\{1,\dots,n\}$,
1685: $m_i\in{\bf Z}$ $(i\in\Gamma_1)$ and
1686: $p_{ij}\in{\bf Z}_4$ $\Bigl((i,j)\in\Gamma_2\Bigr)$
1687: satisfying the following.
1688:
1689: \begin{itemize}
1690: \setlength{\itemsep}{-3pt}
1691: \item[{\rm (i)}]
1692: $m_i\ne 0$ for any $i\in\Gamma_1$.
1693:
1694:
1695: \item[{\rm (ii)}]
1696: $i<j$ and $p_{ij}\ne 0$
1697: for any $(i,j)\in\Gamma_2$.
1698:
1699: \item[{\rm (iii)}]
1700: $i<j<k$ or $i<k<j$ for any $(i,j,k)\in\Gamma_3$.
1701: \end{itemize}
1702: \end{lemma}
1703: %%---%%---%%---%%---%%---%%
1704:
1705:
1706:
1707: {\it Proof.}
1708: By Lemma $5.1$,
1709: any $n$-component surface-link $F=F_1\cup\dots\cup F_n$
1710: is unorientedly bordant to
1711: $(F_1\amalg\dots\amalg F_n)\amalg F'$,
1712: where $F'$ is a split union of necklaces.
1713: Put
1714: $\Gamma_1=\{i\ |e(F_i)\ne 0\}$ and $m_i=e(F_i)/2\in{\bf Z}$
1715: $(i\in\Gamma_1)$.
1716: By Lemma $3.1$,
1717: we see that
1718: $F_1\amalg\dots\amalg F_n$ is unorientedly bordant to
1719: $\coprod_{i\in\Gamma_1} P^{m_i}(i)$
1720: satisfying the condition (i).
1721:
1722: Since a necklace
1723: $N^p(i,j;k_1,\dots,k_m)$ is unorientedly bordant to
1724: $S^p(i,j)\amalg N^0(i,j;k_1)\amalg\dots
1725: \amalg N^0(i,j;k_m)$
1726: by Lemma $4.4$(v),
1727: $F'$ is unorientedly bordant to $F''\amalg F'''$
1728: such that $F''$ is a split union of
1729: some $S^p(i,j)$'s and
1730: $F'''$ is a split union of some $N^0(i,j;k)$'s.
1731:
1732:
1733:
1734: We may assume that $i<j$
1735: for any $S^p(i,j)$ appearing in $F''$
1736: by Lemma $4.4$(i) and (ii).
1737: Moreover,
1738: by Lemma $4.4$(v) and (vi),
1739: we see that there exist a subset
1740: $\Gamma_2\subset\{1,\dots,n\}^2$ and
1741: $p_{ij}\in{\bf Z}_4$
1742: such that $F''$ is unorientedly bordant to
1743: $\coprod_{(i,j)\in\Gamma_2} S^{p_{ij}}(i,j)$
1744: and that the condition (ii) is satisfied.
1745:
1746:
1747: By Lemma $4.4$(i), (ii), (iv) and (vi),
1748: we may assume that $i<j$,
1749: $i\ne k$ and $j\ne k$ for any $N^0(i,j;k)$
1750: appearing in $F'''$.
1751: Applying Lemma $5.2$ for $N^0(i,j;k)$
1752: with $k<i<j$,
1753: we may assume that $(i,j,k)$ satisfies
1754: the condition (iii) for any $N^0(i,j;k)$
1755: appearing in $F'''$.
1756: By Lemma $4.4$(iii), (v) and (vi),
1757: we see that there exists a subset
1758: $\Gamma_3\subset\{1,\dots,n\}^3$
1759: such that $F'''$ is unorientedly bordant to
1760: $\coprod_{(i,j,k)\in\Gamma_3} N^0(i,j;k)$
1761: and that the condition (iii) is satisfied.
1762: \qed
1763:
1764:
1765: \bigskip
1766:
1767:
1768:
1769: For the unoriented bordism group $UL_{4,n}$,
1770: we consider three types of
1771: homomorphisms $e_i$ $(i=1,\dots,n)$,
1772: $d_{ij}$ $(i\ne j)$, and
1773: $t_{ijk}$ ($i\ne j$ and $j\ne k$) as follows:
1774: $$\left\{\begin{array}{rll}
1775: e_i:&UL_{4,n}\longrightarrow{\bf Z}
1776: &{\rm for}\ [F]\mapsto e(F_i)/2,\\
1777: d_{ij}:&UL_{4,n}\longrightarrow{\bf Z}_4
1778: &{\rm for}\ [F]\mapsto d(F_i,F_j), \\
1779: t_{ijk}:&UL_{4,n}\longrightarrow{\bf Z}_2
1780: &{\rm for}\ [F]\mapsto t(F_i,F_j,F_k),
1781: \end{array}\right.
1782: $$
1783: where $F=F_1\cup\dots\cup F_n$.
1784:
1785:
1786: %%---%%---%%---%%---%%---%%
1787: \begin{lemma}
1788: For an $n$-component surface-link $F$,
1789: let $\Gamma_i$ $(i=1,2,3)$,
1790: $m_i\in{\bf Z}$ $(i\in\Gamma_1)$ and
1791: $p_{ij}\in{\bf Z}_4$ $\Bigl((i,j)\in\Gamma_2\Bigr)$
1792: be as in Lemma $6.1$.
1793: Then we have the following.
1794: %%---%%---%%---%%---%%---%%
1795: \begin{itemize}
1796: \setlength{\itemsep}{-2pt}
1797: \item[{\rm (i)}]
1798: $e_i([F])=m_i$ if $i\in\Gamma_1$ and
1799: $e_i([F])=0$ if $i \not\in\Gamma_1$.
1800: \item[{\rm (ii)}]
1801: For $i<j$,
1802: $d_{ij}([F])=p_{ij}$ if $(i,j)\in\Gamma_2$
1803: and $d_{ij}([F])=0$ if $(i,j)\not\in\Gamma_2$.
1804:
1805: \item[{\rm (iii)}]
1806: For $i<j<k$,
1807: $t_{ijk}([F])=1$ if $(i,j,k)\in\Gamma_3$
1808: and $t_{ijk}([F])=0$ if $(i,j,k)\not\in\Gamma_3$.
1809:
1810: \item[{\rm (iv)}]
1811: For $i<k<j$,
1812: $t_{ijk}([F])=1$ if $(i,j,k)\in\Gamma_3$
1813: and $t_{ijk}([F])=0$ if $(i,j,k)\not\in\Gamma_3$.
1814:
1815: \end{itemize}
1816: %%---%%---%%---%%---%%---%%
1817: \end{lemma}
1818: %%---%%---%%---%%---%%---%%
1819:
1820:
1821: {\it Proof.}
1822: (i)
1823: Since $e(P^m)=2m$,
1824: we have $e_i([F])=m_i$ if $i\in\Gamma_1$
1825: and otherwise $e_i([F])=0$.
1826:
1827: (ii)
1828: This follows from Lemma $4.2$(ii).
1829:
1830: (iii) and (iv)
1831: Note that $(j,i,k)$, $(j,k,i)$,
1832: $(k,i,j)$, $(k,j,i)\notin\Gamma_3$.
1833: Since $t_{ijk}\Bigl([N^0(i,j;k)]\Bigr)=1$
1834: and $t_{ijk}\Bigl([N^0(i,k;j)]\Bigr)=0$
1835: by Lemma $4.3$,
1836: $t_{ijk}([F])=1$ if and only if
1837: $(i,j,k)\in\Gamma_3$.
1838: \qed
1839:
1840:
1841:
1842: \bigskip
1843:
1844: {\it Proof of Theorem $1.2$.}
1845: Consider a homomorphism
1846: $$UH:UL_{4,n}\longrightarrow
1847: (\underbrace{{\bf Z}\oplus\dots
1848: \oplus{\bf Z}}_{n})
1849: \oplus
1850: (\underbrace{{\bf Z}_4\oplus\dots
1851: \oplus{\bf Z}_4}_{\frac{n(n-1)}{2}})
1852: \oplus
1853: (\underbrace{{\bf Z}_2 \oplus\dots
1854: \oplus{\bf Z}_2}_{\frac{n(n-1)(n-2)}{3}})$$
1855: defined by
1856: $UH=\Bigl(\bigoplus_{i=1}^n e_i\Bigr)\oplus
1857: \Bigl(\bigoplus_{i<j} d_{ij}\Bigr)\oplus
1858: \Bigl(\bigoplus_{i<j<k\mbox{ or }i<k<j}
1859: t_{ijk}\Bigr)$.
1860: This homomorphism is injective by Lemma $6.2$.
1861: Also,
1862: $UH$ is surjective;
1863: indeed,
1864: $UH\Bigl([P^1(i)]\Bigr)$ $(i=1,\dots,n)$,
1865: $UH\Bigl([S^1(i,j)]\Bigr)$ $(i<j)$, and
1866: $UH\Bigl([N^0(i,j;k)]\Bigr)$
1867: $(i<j<k$ or $i<k<j)$ are
1868: generators of
1869: ${\bf Z}$, ${\bf Z}_4$, and ${\bf Z}_2$
1870: respectively.
1871: \qed
1872:
1873:
1874: \bigskip
1875:
1876: In the definition of the homomorphism $UH$,
1877: we do not use all double linking numbers and
1878: triple linking numbers.
1879: The unused ones are determined as follows.
1880:
1881:
1882:
1883:
1884: %%---%%---%%---%%---%%---%%
1885: \begin{proposition}
1886: For distinct $i,j,k$ and
1887: an $n$-component surface-link $F$,
1888: we have
1889:
1890: {\rm (i)}
1891: $d_{ji}([F])=-d_{ij}([F])$,
1892:
1893: {\rm (ii)}
1894: $t_{iji}([F])=t_{jij}([F])=
1895: \lambda\Bigl(d_{ij}([F])\Bigr)$,
1896: where $\lambda:{\bf Z}_4\rightarrow{\bf Z}_2$
1897: is the natural projection,
1898:
1899: {\rm (iii)}
1900: $t_{ijk}([F])=t_{kji}([F])$, and
1901:
1902: {\rm (iv)}
1903: $t_{jik}([F])+t_{ijk}([F])+t_{ikj}([F])=0$.
1904:
1905: \end{proposition}
1906: %%---%%---%%---%%---%%---%%
1907:
1908:
1909: {\it Proof.}
1910: It is sufficient to prove (i) and (ii)
1911: in case that $F$ is as in Lemma $6.1$.
1912: We use Lemma $4.2$.
1913: We have $t_{iji}([F])=p_{ij}\in{\bf Z}_2$ if $i<j$, and
1914: $t_{iji}([F])=p_{ji}\in{\bf Z}_2$ if $i>j$.
1915: On the other hand,
1916: we have $d_{ij}([F])=p_{ij}\in{\bf Z}_4$ if $i<j$,
1917: and
1918: $d_{ij}([F])=-p_{ji}\in{\bf Z}_4$ if $i>j$.
1919: Hence, we have
1920: $\lambda\Bigl(d_{ij}([F])\Bigr)=t_{iji}([F])$.
1921: Similarly,
1922: since $d_{ij}([F])=p_{ij}\in{\bf Z}_4$ and
1923: $d_{ji}([F])=-p_{ij}\in{\bf Z}_4$ for $i<j$,
1924: we have $d_{ij}([F])=-d_{ji}([F])$.
1925: (iii) and (iv) are proved
1926: in Theorem $3.2$ of \cite{Satoh}.
1927: \qed
1928:
1929:
1930: \bigskip
1931:
1932:
1933:
1934: We consider the homomorphism
1935: $f:L_{4,n}\rightarrow UL_{4,n}$
1936: induced by the map
1937: forgetting the orientations of surface-links.
1938: For an oriented $n$-component surface-link $F$,
1939: we can define two kinds of bordism invariants;
1940: double linking invariants
1941: $D_{ij}:L_{4,n}\rightarrow{\bf Z}_2={\bf Z}/2{\bf Z}$
1942: and triple linking invariants
1943: $T_{ijk}:L_{4,n}\rightarrow{\bf Z}$
1944: (cf. \cite{CKSS2}).
1945: Then Sanderson's isomorphism
1946: $$H:L_{4,n}\longrightarrow
1947: (\underbrace{{\bf Z}_2\oplus\dots
1948: \oplus{\bf Z}_2}_{\frac{n(n-1)}{2}})
1949: \oplus
1950: (\underbrace{{\bf Z} \oplus\dots
1951: \oplus{\bf Z}}_{\frac{n(n-1)(n-2)}{3}})$$
1952: is given by
1953: $H=\Bigl(\bigoplus_{i<j} D_{ij}\Bigr)\oplus
1954: \Bigl(\bigoplus_{i<j<k\mbox{ or }i<k<j}
1955: T_{ijk}\Bigr)$.
1956: From the definitions of these invariants,
1957: the forgetful map $f$ is regarded as
1958: $$(\oplus 0)\oplus(\oplus\kappa)\oplus
1959: (\oplus\nu):(\oplus\{0\})\oplus(\oplus{\bf Z}_2)
1960: \oplus(\oplus{\bf Z})\rightarrow
1961: (\oplus{\bf Z})\oplus(\oplus{\bf Z}_4)
1962: \oplus(\oplus{\bf Z}_2),$$
1963: under the isomorphisms $U$ and $UH$,
1964: where $\kappa:{\bf Z}_2\rightarrow{\bf Z}_4$
1965: is the natural inclusion and
1966: $\nu:{\bf Z}\rightarrow{\bf Z}_2$ is
1967: the natural projection.
1968:
1969:
1970: \bigskip
1971:
1972: \noindent
1973: {\large\bf Acknowledgments} \
1974: JSC is being supported by NSF grant DMS-9988107.
1975: SK is being supported by Fellowships
1976: from the Japan Society for the Promotion of Science.
1977: MS is being supported by NSF grant DMS-9988101.
1978: The authors would like to thank Uwe Kaiser for providing helpful information.
1979:
1980: %%---%%---%%---%%---%%---%%
1981:
1982:
1983: \newcommand{\bysame}{%
1984: \leavevmode\hbox to 3em{\hrulefill}\,}
1985: \begin{thebibliography}{99}
1986:
1987:
1988: %\bibitem{BT} A. Bartels and P. Teichner,
1989: %{\it All two dimensional links are null homotopic,}
1990: %Geometry and Topology, Vol. 3 (1999), 235--252.
1991:
1992: %\bibitem{CJKLS}
1993: %J.S. Carter, D. Jelsovsky, S. Kamada, L.
1994: %Langford and M. Saito,
1995: %{\it Quandle cohomology and state-sum invariants
1996: %of knotted curves and surfaces,}
1997: %preprint at
1998: %\begin{verbatim}
1999: %http://xxx.lanl.gov/abs/math.GT/9903135
2000: %\end{verbatim}
2001:
2002: %\bibitem{CKSS}
2003: %J.S. Carter, S. Kamada, M. Saito, and S. Satoh,
2004: %{\it Triple linking of surfaces in $4$-space,}
2005: %preprint.
2006:
2007:
2008: \bibitem{CKSS2}
2009: J.S. Carter, S. Kamada, M. Saito, and S. Satoh,
2010: {\it A theorem of Sanderson on link bordisms in dimension $4$},
2011: Algebraic and Geometric Topology, {\bf 1} (2001) 299--310.
2012:
2013:
2014: \bibitem{CS:branch}
2015: J. S. Carter and M. Saito,
2016: {\it Canceling branch points on projections
2017: of surfaces in $4$-space},
2018: Proc. Amer. Math. Soc. {\bf 116} (1992), 229--237.
2019:
2020:
2021: \bibitem{CS:Reidemeister}
2022: \bysame,
2023: {\it Reidemeister moves for surface isotopies and
2024: their interpretations as moves to movies,}
2025: J. Knot Theory Ramifications {\bf 2} (1993), 251--284.
2026:
2027:
2028: \bibitem{CS:book}
2029: \bysame,
2030: {\it Knotted surfaces and their diagrams,}
2031: the American Mathematical Society, Surveys and Monograph Series
2032: (Providence 1998).
2033:
2034:
2035:
2036: % \bibitem{CS:seif} Carter, J.S.; Saito, M.,
2037: % {\it A Seifert algorithm for knotted surfaces,}
2038: % Topology 36 (1997), no. 1, 179--201.
2039:
2040:
2041: %\bibitem{Tim1}
2042: %T. D. Cochran,
2043: %{\it On an invariant of link cobordism in dimension four,}
2044: %Topology Appl. 18 (1984), no. 2-3, 97--108.
2045:
2046: %\bibitem{TO}
2047: %T. D. Cochran and K. E. Orr,
2048: %{\it Not all links are concordant to boundary
2049: %links, } Ann. of Math. (2) 138 (1993), no. 3, 519--554.
2050:
2051: %\bibitem{FennRolf}
2052: %R. Fenn and D. Rolfsen,
2053: %{\it Spheres may link homotopically in
2054: %$4$-space,}
2055: %J. London Math. Soc. (2) 34 (1986), no. 1, 177--184.
2056:
2057: \bibitem{Kauffman} L. H. Kauffman,
2058: {\it Knots and physics},
2059: Series on Knots and Everything, 1,
2060: World Scientific Publ.
2061:
2062:
2063: \bibitem{Kamada} S. Kamada,
2064: {\it Non-orientable surfaces in $4$-space},
2065: Osaka J. Math., {\bf 26} (1989), 367--385.
2066:
2067:
2068:
2069:
2070: %\bibitem{Kawa:book} A. Kawauchi,
2071: %{\it A survey of knot theory,}
2072: %Birkhauser, 1996.
2073:
2074: %\bibitem{Kirk} P. A. Kirk,
2075: %{\it Link maps in the four sphere,}
2076: %Differential topology (Siegen, 1987), 31--43,
2077: %Lecture Notes in Math., 1350,
2078: %Springer, Berlin-New York, 1988.
2079:
2080: %\bibitem{KK}
2081: %P. A. Kirk and U. Koschorke,
2082: %{\it Generalized Seifert surfaces and linking numbers,}
2083: %Topology Appl. 42 (1991), no. 3, 247--262.
2084:
2085: %\bibitem{Koschorke}
2086: %U. Koschorke,
2087: %{\it Homotopy, concordance and bordism of link maps,}
2088: %Global analysis in modern mathematics
2089: %(Orono, ME, 1991; Waltham, MA, 1992),
2090: %283--299,
2091: %Publish or Perish, Houston, TX, 1993.
2092:
2093:
2094: \bibitem{Massey} W. S. Massey,
2095: {\it Proof of a conjecture of Whitney},
2096: Pacific J. Math., {\bf 31} (1969),
2097: 143--156.
2098:
2099:
2100:
2101: %\bibitem{Massey} W. S. Massey,
2102: %{\it A basic course in algebraic topology,}
2103: %Springer-Verlag, 1991.
2104:
2105: %\bibitem{MassayRolf} W. S. Massey and D. Rolfsen,
2106: %{\it Homotopy classification of higher-dimensional links,}
2107: %Indiana Univ. Math. J. 34 (1985), no. 2, 375--391.
2108:
2109: %\bibitem{Rolfsen} D. Rolfsen,
2110: %{\it Knots and links,}
2111: %Publish or Perish, Inc., 1976.
2112:
2113: \bibitem{Roseman} D. Roseman,
2114: {\it Reidemeister-type moves for surfaces in four dimensional space, }
2115: in Banach Center Publications {\bf 42} (1998) Knot theory, 347--380.
2116:
2117: %\bibitem{Ruberman}
2118: %D. Ruberman,
2119: %{\it Concordance of links in $S\sp 4$,}
2120: % Four-manifold
2121: %theory (Durham, N.H., 1982), 481--483, Contemp. Math., 35, Amer. Math.
2122: %Soc., Providence, R.I.,
2123: %1984.
2124:
2125:
2126: \bibitem{Sanderson1} B. J. Sanderson,
2127: {\it Bordism of links in codimension $2$},
2128: J. London Math. Soc. {\bf 35} (1987),
2129: 367--376.
2130:
2131:
2132:
2133: \bibitem{Sanderson2}
2134: \bysame,
2135: {\it Triple links in codimension $2$},
2136: Colloquia Mathematica Societatis
2137: J\'anos Bolyai,
2138: Topology, P\'ecs (Hungary),
2139: {\bf 55} (1989),
2140: 457--471.
2141:
2142:
2143:
2144: %\bibitem{NSato} N. Sato,
2145: %{\it Cobordisms of semiboundary links.}
2146: %Topology Appl. 18
2147: % (1984), no. 2-3, 225--234.
2148:
2149:
2150: \bibitem{Satoh} S. Satoh,
2151: {\it Triple point invariants of non-orientable surface-links},
2152: Proc. of the first topology meeting of Japan-Mexico
2153: (Molelia, July 1999),
2154: Topology and its Applications (to appear).
2155:
2156: \bibitem{Whitney} H. Whitney,
2157: {\it On the topology of differentiable manifolds},
2158: Lectures in topology, Michigan Univ. Press, 1940.
2159:
2160: \end{thebibliography}
2161:
2162: \end{document}
2163: