1: \documentclass[12pt,a4paper]{article}

2: %\documentstyle[12pt]{article}

3: \usepackage[english]{babel}

4: \usepackage{amssymb}

5: \usepackage[dvips]{graphicx}

6: 

7: \setlength{\textwidth}{6.5in}

8: \setlength{\oddsidemargin}{0.0in}

9: \setlength{\evensidemargin}{0.0in} 

10: \setlength{\textheight}{9.6in}

11: \addtolength{\topmargin}{-1.1in}

12: 

13: 

14: \def\C{{\mathbb C}}

15: \def\Tn{T^*{\bf C}^n}

16: \def\f{\varphi}

17: \def\L{\Lambda}

18: \def\l{\lambda}

19: \def\D{\Delta}

20: \def\S{\Sigma}

21: \def\s{\sigma}

22: \def\Z{{\bf Z}}

23: \def\sub{\underline}

24: \def\w{{\omega}}

25: \def\ww{{\overline \w}}

26: %{\wedge}

27: \def\LL{{\mathcal L}}

28: \def\M{{\cal M}}

29: \def\SS{{\cal S}}

30: \def\p{\partial}

31: \def\su{{\rm Su}}

32: \def\O{{\cal O}}

33: \def\J{J^1({\bf C}^n,{\bf C})}

34: \def\Bif{{\rm Bif}}

35: \def\D{\Delta}

36: \def\A{{\mathcal A}}

37: \def\F{{\mathcal F}}

38: \def\K{{\mathcal K}}

39: \def\R{{\mathcal R}}

40: \def\U{{\mathcal U}}

41: \def\Ud{U^\circ}

42: \def\beq{\begin{equation}}

43: \def\eeq{\end{equation}}

44: \def\un{\underline}

45: \def\st{\stackrel}

46: \def\a{\alpha}

47: \def\b{\beta}

48: \def\g{\gamma}

49: \def\G{\Gamma}

50: \def\ds{\displaystyle}

51: \def\hf{{\widehat f}}

52: \def\hfi{{\widehat\f}}

53: \def\k{\kappa}

54: \def\we{\wedge}

55: \def\wt{\widetilde}

56: \def\T{{\cal T}}

57: \def\e{\varepsilon}

58: \def\ov{\overline}

59: \def\o{\omega}

60: \def\oo{\ov\o}

61: \def\ZZ{{\mathbb Z}}

62: \def\aa{{\mathbf a}}

63: 

64: 

65: %%%%%%%%%%%%%%%%

66: 

67: 

68: %\newcommand{\p}{\partial}

69: \renewcommand{\l}{\lambda}

70: \renewcommand{\L}{\Lambda}

71: \renewcommand{\d}{\delta}

72: \newtheorem{thm}{Theorem}[section]

73: %\newtheorem{pr}{Proposition}[section]

74: \newtheorem{pr}[thm]{Proposition}

75: %%%%\newtheorem{cor}{Corollary}[section]

76: \newtheorem{cor}[thm]{Corollary}

77: %%%%\newtheorem{lm}{Lemma}[section]

78: \newtheorem{lm}[thm]{Lemma}

79: %%%%\newtheorem{conj}{Conjecture}[section]

80: \newtheorem{conj}[thm]{Conjecture}

81: \newtheorem{dfn}[thm]{Definition}

82: 

83: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

84: 

85: \begin{document}

86: 

87: \title{{\bf Automorphisms of $P_8$ singularities \\ and the complex crystallographic groups}}

88: 

89: \author{Victor Goryunov and Dmitry Kerner}

90: \date{}

91: \maketitle

92: 

93: \begin{abstract}

94: The paper completes the study of symmetries of parabolic function singularities

95: with relation to complex crystallographic groups that was started in \cite{GM,X9}. 

96: We classify smoothable automorphisms of $P_8$ singularities 

97: which split the kernel of the intersection form on the second homology. For such automorphisms,

98: the monodromy groups acting on the duals to the eigenspaces with degenerate intersection form are then identified as some of 

99: complex affine reflection groups tabled in \cite{P}.

100: \end{abstract} 

101: 

102: 

103: Singularity theory has been maintaining close relations with reflection groups since its early days, starting with the famous

104: works of Arnold and Brieskorn \cite{A,Br}. The first to emerge, as simple function singularities, were the 

105: Weyl groups $A_k, D_k$ and $E_k$. They were followed by the $B_k$,  $C_k$ and $F_4$ as simple boundary 

106: singularities or, equivalently, functions invariant under the involution \cite{Ab,Slo}. Since then appearance of Weyl groups 

107: in a singularity problem became a kind of criterion of naturalness of the problem \cite{AGV2,AGLV1,AGLV2}. 

108: 

109: The next to receive a singularity realisation was the list of all Coxeter groups, in the classification of 

110: stable Lagrangian maps \cite{Gi}. Some of the

111: Shephard-Todd groups appeared in \cite{G1,G2,GB} in the context of simple functions equivariant with respect to a cyclic

112: group action. And finally, the first examples of complex crystallographic groups came out in \cite{GM,X9} 

113: in connection with the symmetries of parabolic functions $J_{10}$ and $X_9$. 

114: The affine reflection groups appeared there as monodromy groups, which 

115:  is similar to the first realisations of other classes of reflection groups. This time it was the equivariant monodromy

116: corresponding to the symmetry eigenspaces $H_\chi$ in the vanishing second homology on which the intersection form 

117: $\s$ has corank 1.

118: 

119: This paper completes the study of cyclically equivariant parabolic functions started in \cite{GM,X9} and considers the $P_8$ singularities.

120: The approach we are developing here is considerably shorter, with less calculations and more self-contained. This is allowed by 

121: a preliminary observation that the modulus parameter may take on only exceptional values if corank$(\s|_{H_{\chi}}) = 1$: the

122: $j$-invariant of the underlying elliptic curves must be either 0 or 1728 (see Section \ref{Sker}).

123: 

124: \medskip

125: The structure of the paper is as follows.

126: 

127: Section 1 describes the crystallographic groups which will be involved.

128: 

129: Section 2 gives a classification of smoothable cyclic symmetries of $P_8$ singularities possessing eigenspaces $H_{\chi}$

130: with the property as already mentioned.

131: 

132: In Section 3, we obtain distinguished sets of generators in such eigenspaces and calculate the intersection numbers of the

133: elements in the sets. This allows us to describe the Picard-Lefschetz operators generating, as complex reflections,  the equivariant

134: monodromy action on the $H_{\chi}$.

135: 

136: Finally, in Section 4, we pass to the vanishing cohomology. In its subspaces dual to the $H_{\chi}$, we consider hyperplanes

137: formed by all the cocycles taking the same non-zero value on a fixed generator of the kernel of the hermitian intersection form

138: $\s|_{H_{\chi}}$. We show that the equivariant monodromy group acting on each of these hyperplanes is a complex 

139: crystallographic group.

140:  

141: \medskip

142: Both authors gratefully acknowledge the support of the EC Marie Curie Training Site (LIMITS) on Algebraic Geometry, 

143: Singularities and Knot Theory which funded the visits of the second author to Liverpool.

144: The research of the second author was also supported by the Skirball postdoctoral fellowship of the Center 

145: of Advanced Studies in Mathematics, Be'er Sheva, Israel.

146: 

147: 

148: %8. Reflection groups on the non-kernel character subspaces

149: %\\

150: %9. Odd versions of the affine groups, i.e. of the Dynkin diagrams

151: %\\

152: %10. Non-smoothable automorphisms splitting the kernel 

153: 

154: 

155: \section{The crystallographic groups}

156: We remind the description of the complex affine reflection groups 

157: which will be relevant to our singularity constructions. In the notations of \cite{P}, the groups

158: are

159: $$

160: [K_3(3)], [K_3(4)], [K_3(6)], [G(3,1,2)]_2,  [K_5], [K_8], [K_{25}]\,.

161: $$ 

162: The linear parts of these groups are Shephard-Todd groups $L$  shown in Figure \ref{Fgr}, in the related notation of \cite{ST}. 

163: There and in what follows $\w=exp(2\pi i/3)$.

164: The rank of $L$ is the number of the vertices in its diagram. 

165: Each of the groups involved is a semi-direct product of its linear part and translation lattice. 

166: The lattice of $[G(3,1,2)]_2$ is spanned over $\ZZ$ by the

167: $G(3,1,2)$-orbit of any order 2 root, while for any other group it is spanned the $L$-orbit of any root. For example, our list contains

168: all three one-dimensional groups, the $[K_3(m)]$, each generated by an order $m$ rotation in $\C$ and the lattice $\ZZ[1,i]$ if $m=4$ or 

169: $\ZZ[1,\w]$ if $m=3,6$.

170: 

171: \begin{figure}[hbtp]

172: \begin{center}

173: %%\epsfxsize=12.0cm

174: %%\leavevmode

175: %%\epsffile{p8di.eps}

176: \includegraphics[width=0.7\textwidth]{p8gr.eps}

177: %%\end{center}

178: \caption{Linear parts of the crystallographic groups. Each vertex represents a unit root. The order of the corresponding reflection 

179: is written inside the vertex (order 2 omitted). An egde $a \to b$ is equipped with the hermitian product $\left< a, b \right>$ 

180: (no orientation is needed when the product is real). There are no edges between orthogonal roots.}

181: \label{Fgr}

182: \end{center}

183: \end{figure}

184: 

185: 

186: 

187: 

188: \section{The symmetries}

189: 

190: \subsection{Cubic curves}

191: Our first step in the study of linear automorphisms of the surfaces

192: $$x^3+y^3+z^3+3\a xyz =0\,, \qquad \a^3 \ne -1,$$

193: in $\C^3$, is consideration of the projective version of the question. This is a

194: description of automorphisms of projective curves $C_\a$ given by the same

195: equations in $\C P^2$. The question is classical, but, for the sake of the exposition,

196: we indicate a possible elementary approach.

197: 

198: A curve $C_\a$ has nine inflection points. There are four triplets of lines

199: in $\C P^2$ passing through these points, each line containing exactly three inflections

200: and each triplet containing all nine inflections. The lines are zero sets of the linear forms

201: $$

202: \begin{array}{lll}

203: x, & y, & z; \\

204: x+y+z, \quad & x+\w y + \ww z, \quad & x + \ww y + \w z; \\

205: x+y+\w z, \quad & x + \w y + z, \quad & \w x + y + z; \\

206: x+y+\ww z, \quad & x + \ww y + z, \quad & \ww x + y + z\,.

207: \end{array}

208: $$

209: A projective automorphism of $C_\a$ permutes the triplets and reorders the lines

210: within the triplets. It is sufficient to consider the images of the first triplet

211: only. A routine study of all the possibilities, followed by diagonalisation of each 

212: projective transformation and rewriting the equation of an appropriate curve in the 

213: projective eigencoordinates, give us 

214: 

215: \begin{pr}\label{Pproj}

216: Any projective automorphism of a cubic curve can be reduced, by a choice of projective

217: coordinates, either to the diagonal projective transformation 

218: $(x:y:z) \mapsto (\k_xx:\k_yy:\k_zz)$ 

219: %or its inverse

220: of one of the following

221: cubics $f=0$ or to the inverse of such a transformation:

222: $$

223: \begin{array}{|l|l|l|l|} 

224: \hline 

225: &&& \vspace{-8pt} \\

226:  & \qquad\qquad\qquad  f & \k_x:\k_y:\k_z & 

227: {\rm monomial \  basis \  of \  the \  local \  ring \  of \  } f

228: %\qquad\quad{\rm versal \  monomials} 

229: \\ 

230: &&& \vspace{-8pt} \\

231:  \hline &&& \vspace{-8pt} \\

232: 1 & x^3 + y^3 + yz^2   & \,\, \w : 1 : -1 & 

233: 1, x, y, z, xy, xz, y^2\sim z^2, xy^2 \sim xz^2 \\ &&& \vspace{-8pt} \\

234: 2 & x^3 + y^3 + yz^2   \simeq x^3 + y^3 + z^3 & \,\, \w : 1 : 1 &

235: 1, x, y, z, xy, xz, y^2\sim z^2, xy^2 \simeq xyz \\ &&& \vspace{-8pt} \\

236: 3 & x^2y + y^2z + z^2x & \,\, 1 : \w : \ww &

237: 1,x,y,z,x^2 \sim yz, y^2 \sim xz, z^2 \sim xy, xyz  

238: \\ &&&  \vspace{-8pt} \\

239: 4 & x^2z + xy^2 + z^3 & \,\, 1 : i : -1 &

240: 1,x,y,z,yz,x^2\sim z^2,xz\sim y^2,x^3 \\  &&& \vspace{-8pt} \\

241: 5 & x^3 + y^3 + yz^2 + 3 \a xy^2\,, \,\, \a^3 \ne -\frac14 & \,\, 1:1:-1 & 

242: 1, x, y, z, xy, xz, y^2\sim z^2, xy^2 \sim xz^2 \\ &&& \vspace{-8pt} \\

243: 6 & x^3+y^3+z^3+ 3 \a xyz\,, \,\, \a^3 \ne -1  & \,\, 1:\w:\ww  &

244: 1,x,y,z,xy,xz,yz,xyz \\ &&& \vspace{-8pt} \\

245: 7 & x^3+y^3+z^3+3\a xyz\,, \,\, \a^3 \ne -1 & \,\, 1:1:1 &

246: 1,x,y,z,xy,xz,yz,xyz \vspace{-8pt} \\ &&& \\ \hline

247: \end{array}

248: $$

249: \end{pr}

250: 

251: \medskip

252: The contents of the last column of the table will be used in Section \ref{Ssmosym}.

253:  

254: \subsection{Kernel character constraints}\label{Sker} 

255: We now lift a projective automorphism of a cubic curve $f=0$ to a linear

256: transformation $g$ of $\C^3$. It multiplies $f$ by a non-zero constant 

257: %$\kappa$ 

258: and hence acts on the target $\C$ of the function $f$. 

259: We consider deformations of the function-germ $f: (\C^3,0) \to (\C,0)$ 

260: as usual: choosing a small ball $B$ in $\C^3$ centred at the origin and deforming $f$ within the ball. However, 

261: we restrict our attention to deformations which are $g$-equivariant with respect to the actions of $g$ on

262: $\C^3$ and $\C$. 

263:  

264: \begin{dfn} {\em

265: If the function-germ $f$ admits $g$-equivariant deformations 

266: with non-singular zero levels, we say that the symmetry $g$ of $f$

267: is} smoothable.

268: \end{dfn}

269: 

270: In this case, the smooth zero level $V \subset B$ of a small $g$-equivariant 

271: deformation of $f$ is a Milnor fibre of $f$, and $g$ acts on $V$ and on its 

272: second homology. Assuming from now on the order $r$ of $g$ finite, we obtain the 

273: splitting into a direct sum of character subspaces 

274: \begin{equation}

275: H_2(V,\C) = \oplus_{\chi^r=1} H_{2,\chi}

276: \label{Esum}

277: \end{equation}

278: so that $g$ acts on each summand as multiplication by the relevant 

279: $r$th root $\chi$ of unity.

280: 

281: We extend the intersection form $\sigma$ from the lattice $H_2(V,\ZZ)$ to

282: the hermitian form $\sigma_h$ on $H_2(V,\C)$. The symmetric intersection form 

283: of a parabolic function singularity in 3 variables is negative semi-definite, with a two-dimensional kernel. 

284: 

285: \begin{dfn} {\em

286: We say that a smoothable symmetry $g$} splits {\em the kernel of

287: $\sigma_h$ if this kernel is shared by two different character subspaces

288: $H_{2,\chi}$. If the restriction of $\sigma_h$ to a $H_{2,\chi}$ is 

289: degenerate, we call $\chi$ a} kernel character.

290: \end{dfn}     

291: 

292: The aim of our construction, a complex crystallographic group, will be acting 

293: on a hyperplane $\G$ in the cohomology subspace $H_{\chi}^2 = \{\gamma \in H^2(V,\C): g^*(\gamma) = \chi\gamma\}$ where

294: $\chi$ is a kernel character (cf. \cite{GM,X9}).

295: For this, our symmetry $g$ should split the kernel of $\sigma_h$, and the hyperplane $\G$ is defined as 

296: the set of all elements of $H_{\chi}^2$ taking the same non-zero value

297: on a fixed generator of the line $Ker(\sigma_h) \cap H_{2,\chi}$.

298: 

299: \medskip

300: Our next step is to show that there is no kernel splitting in any of the modular

301: cases of Proposition \ref{Pproj}. 

302: 

303: Assume $g$ is a lifting of one of the projective automorphisms from the table there.

304: %The cyclic group generated by $g$ will be denoted $G$.

305: Take a monomial basis $\f_0=1, \f_1, \dots, \f_7$ of the local ring

306: of $f$, with $\f_7$ of degree 3.

307: Consider the sub-unfolding of the $\R$-miniversal unfolding of $f$ in

308: which the modulus parameter does not participate:

309: $$

310: \F: \C^3 \times \C^{6} \to \C \times \C^{6}\,, 

311: \quad 

312: (x,y,z,\l_1,\dots,\l_{6}) \mapsto

313: (f+\l_1\f_1+\dots\l_{6}\f_{6},\l_1,\dots,\l_{6})\,.

314: $$ 

315: This map is defined globally. Its smooth fibres are Milnor fibres of $f$.

316: The action of $g$ naturally extends to the source and target of the map,

317: making $\F$ $g$-equivariant. We will denote the target of $\F$ by $\U$,

318: points there will be $u=(u_0,\dots,u_{6})$.

319: %, and the $g$-invariant part of $\U$ will be $U$.

320: 

321: We simultaneously compactify all fibres $V_u$, 

322: $u \in \U$, of $\F$ adding to each of them a copy 

323: $C_u = \ov V_u \setminus V_u$ of the same projective curve $f=0$. 

324: Each $\ov V_u \subset \C P^3$ is smooth near $C_u$. 

325: %The $\ov V_u$ are Del Pezzo surfaces.

326: 

327: For non-critical values of $u$, take the intersection homomorphism

328: $\s:  H_2(V_u;\ZZ) \to H^2(V_u;\ZZ)$.

329: %[later extend this to coefficients $\ZZ[\chi]$ "antiholomorphically")

330: The Leray map sends $H_1(C_u; \ZZ)$ isomorphically onto the 2-dimensional 

331: kernel of $\s$ \cite{L}.

332: 

333: Consider the 2-form $w = dx \we dy \we dz \we d\l_1 \we \dots \we d\l_{6}/ 

334: du_0 \we \dots \we du_{6}$ on the fibres

335: $V_u$. According to \cite{L}, its residue $\b$ along $C_u$ is a non-zero

336: holomorphic 1-form which does not depend on $u$.

337: 

338: The action of the automorphism $g$ on a $g$-symmetric fibre $V_u$ 

339: extends to $\ov V_u$.  The residue map

340: $Res: H^2(V_u;\C) \to H^1(C_u;\C)$ is $g$-equivariant: if $g^*$ multiplies a 2-form

341: by $\chi$, then it multiplies the residue of the form by $\chi$ too. Therefore, if $g$

342: splits $Ker(\s_h)$ on $V_u$, then its action on $H^1(C_u;\C)$ must also

343: have two distinct conjugate eigenvalues, with $\b$ and $\ov \b$ as eigenvectors. 

344: Hence the curve $C_u=\{f=0\}$ must be either $\C/\ZZ[1,i]$ or $\C/\ZZ[1,\w]$. 

345: Respectively, the kernel characters are either $\pm i$ or 

346: $(\w,\ov \w)$ or $(-\w,-\ov\w)$.  

347: %The form $w$ is multiplied by $g$ by one of these numbers.

348: 

349: Thus we have:

350: 

351: \begin{pr}

352: A smoothable automorphism $g$ splits the kernel of the intersection form

353: if and only if the section $w=dx \we dy \we dz /df$ of the vanishing cohomology

354: fibration of $f$ is a $g^*$-eigenvector with the eigenvalue from the set

355: $\{\pm i, \pm \w, \pm\ov\w\}$. The eigenvalue and its conjugate are the 

356: kernel characters. 

357: \end{pr}

358: 

359: If $g^*(w) = \chi w$ then for the hyperplane $\G$

360: mentioned above  one can take $Res^{-1}(Res(w)) \cap H_{\chi}^2$.

361: 

362: Since the degree of $f$ is 3, the eigenvalue of $g^*$ on $w=dx \we dy \we dz /df$

363: depends only on the projectivisation of $g$. 

364: 

365: \begin{cor}\label{Clift}

366: All smoothable liftings of the first four automorphisms from Proposition

367: \ref{Pproj} split $Ker(\s_h)$. On the other hand, none of the liftings of the

368: last three cases does the same. 

369: \end{cor}

370: 

371: \noindent

372: {\bf Remark.} The argument can be easily modified to the 

373: symmetries of $X_9$ and $J_{10}$ function singularities, thus explaining

374: the absence of the moduli in the classification tables in \cite{GM,X9}.

375: 

376: \subsection{Smoothable symmetries}\label{Ssmosym}

377: Assume we have two actions of a finite cyclic group, on $(\C^k,0)$ and on $(\C,0)$.

378: Consider two function-germs $f_1,f_2:  (\C^k,0) \to (\C,0)$ equivariant with

379: respect to these actions: $f_i \circ \rho = \rho \circ f_i$ where $\rho$ is a generator of

380: the group. We say that the two functions are {\em $\R_\rho$-equivalent\/} if one

381: can be transformed into the other by a $\rho$-equivariant diffeomorphism-germ 

382: of $(\C^k,0)$. 

383: 

384: We apply this notion in the context of our cubic function $f$, its {\em diagonal\/}

385: symmetry $g$ and the induced action of $g$ on $\C$.  With a minor abuse of the

386: language, we will still be calling the corresponding equivalence the $\R_g$-equivalence.

387: %If, for example, $g$ splits $Ker(\s)$, then 

388: For example, an $\R_g$-miniversal unfolding of $f$ 

389: is the restriction $F$ of the unfolding $\F$ of the previous subsection to 

390: $\F^{-1}(U)$, where $U \subset \U$ is the set of fixed points of the natural

391: action of $g$ on $\U$. 

392: Equivalently, in this case, for an $\R_g$-miniversal deformation of 

393: $f$ one can take $f +$ arbitrary linear combinations of the elements $\psi_1, \dots, \psi_\tau$ of 

394: a monomial basis of the local ring of $f$ that are multiplied by $g$ by the

395: same factor as $f$. The number $\tau$ appearing here will be called the 

396: {\em $\R_g$-codimension\/} of $f$.

397: 

398: \begin{pr}\label{Proff0}

399: Assume a symmetry $g$ is smoothable. Then $g$ multiplies $f$ by the same

400: factor as it multiplies one of the monomials $1, x, y, z$.

401: \end{pr} 

402: 

403: Indeed the conclusion is equivalent to generic fibres of an $\R_g$-miniversal unfolding

404: not being singular at $0 \in \C^3$. 

405:  

406: With the help of Corollary \ref{Clift} and the last Proposition, we obtain after

407: straightforward calculations which we prefer to omit:

408: 

409: \begin{thm}\label{Tclass} 

410: The complete list of  smoothable symmetries of $P_8$ function singularities 

411: on $\C^3$ which split the kernel of the intersection form is given in Table \ref{Ta}. 

412: Our classification is up to a choice of generators of the same cyclic group.

413: \end{thm}

414: 

415: 

416: \begin{table}[p]

417: \begin{center}

418: \caption{Smoothable symmetries of $P_8$ singularities splitting $Ker(\s)$}

419: \vspace{-15pt}

420: $$\begin{array}{|c|c|c|c|c|c|c|}

421: \hline 

422: &&&&&&\\

423: f & g: x,y,z \mapsto & |g| & 

424: \begin{array}{c}

425: {\rm versal} \\

426: {\rm monomials}

427: \end{array}

428: &

429: \begin{array}{c}

430: \!\!\!{\rm kernel}\!\!\! \\

431: \chi

432: \end{array}

433: &

434: \begin{array}{c}

435: {\rm affine}\\

436: {\rm group}

437: \end{array}

438:  & 

439: {\rm notation} \\

440: &&&&&&\\

441: \hline 

442: \hline

443: &&&&&&\\

444: x^3    &   \w x,y,-z   &   6   &   1, y, y^2    &   -\w, -\ww  &  \!\! [K_5] \!\!  &   C_3^{(3,3)} \\

445: +y^3 &&&&&& \\

446: +yz^2 &   x, \ww y, -\ww z    &   6   &      1,x        &   -\w, -\ww   &  [K_3(6)]    &   (P_8|\ZZ_6)'   \\

447: &&&&&& \\

448: & \ww x, \w y, -\w z  &  6  & 1,xy  &  -\w, -\ww & [K_3(6)] & (P_8|\ZZ_{6})'' \\

449: &&&&&& \\

450: & -x, -\ww y, \ww z &  6  &  x  &  -\w, -\ww  &  -  &  (P_8/\ZZ_6)' \\

451: &&&&&& \\

452: & i\w x, iy, -iz  &  12  &  z  &  -\w, -\ww  &  -  &  P_8/\ZZ_{12} \\

453: &&&&&& \\

454: &  \w x, y, z  &  3  &  1,y,y^2,z  & \w, \ww & [K_{25}] & D_4^{(3)} \\

455: &&&&&& \\

456: &  x, \ww y, \ww z  &  3  &  1,x  & \w,\ww  &  [K_3(6)] & (P_8|\ZZ_{3})' \\

457: &&&&&& \\

458: &  \ww x, \w y, \w z  &  3   &  1,xy,xz & \w, \ww & [G(3,1,2)]_2 & P_8|\ZZ_3 \\

459: &&&&&& \\

460: \hline

461: &&&&&& \\

462: %x^3  &  \ww x, \w y, \w z  &  3   &  1,xy,xz & \w, \ww & [G(3,1,2)]_2 & P_8|\ZZ_3 \\

463: %+y^3 &&&&&& \\

464: x^3 &  -x,-\ww y, -\ww z  &  6  &  x  & \w, \ww & - & (P_8/\ZZ_6)'' \\

465: +y^3 &&&&&& \\

466: +z^3 & -\w x, -y, -z  &  6  &  y,z  &  \w, \ww  &  [K_3(3)]  & P_8/\ZZ_6 \\

467: &&&&&& \\

468: \hline

469: &&&&&& \\

470: \begin{array}{c}

471: x^2y \\ + y^2z \\ + z^2x

472: \end{array}  

473: &  \ov\e_9 x, \w\ov\e_9 y, \ov{\w\e}_9 z  &  9   &  1 & \w, \ww & - & P_8|\ZZ_9 \\

474: &&&&&& \\

475: \hline

476: &&&&&& \\

477: x^2z   &  -x, -iy, z  &  4   &  1,z,z^2 & \pm i & [K_8] & C_3^{(2,4)} \\

478: +xy^2 &&&&&& \\

479: +z^3 & -\w x, -i\w y, \w z & 12 &  1  & \pm i  &  -  &  P_8|\ZZ_{12} \\

480: &&&&&& \\

481: & ix, -y, -iz  &  4  & x, yz  & \pm i & [K_3(4)]  &  P_8/\ZZ_4 \\

482: &&&&&& \\

483: & \ov\e_8 x , \e_8 y, -\ov\e_8 z  &  8  &  y  &  \pm i  &  -  &  P_8/\ZZ_8 \\

484: &&&&&& \\

485: \hline

486: \end{array}

487: $$

488: \label{Ta}

489: \end{center}

490: \end{table}

491: 

492: In the Table: the versal monomials are the monomials participating in $\R_g$-miniversal deformations 

493: (these are selected from the last column of the

494: table in Proposition \ref{Pproj} as those multiplied by $g^*$ by the same factor as $f$), 

495: the affine groups are those which will come out later as monodromy

496: groups acting on the hyperplanes in the character subspaces in the cohomology,

497: and $\e_k=exp(2\pi i/k)$.

498: 

499: \medskip\noindent

500: {\bf Remark.} During the proof of the Theorem, one obtains a few non-smoothable symmetries allowing linear terms in deformations:

501: $$

502: \begin{array}{|c|c|c|c|}

503: \hline 

504: &&&   \vspace{-8pt} \\

505: f & g: x,y,z \mapsto & |g| & 

506: {\rm versal \   monomials}    \vspace{-8pt}  \\

507: &&&\\

508: \hline 

509: \hline

510: &&&   \vspace{-8pt} \\

511: x^3 + y^3 + yz^2 & -\w x, -y, z  &  6  &  y   \vspace{-8pt} \\

512: &&&\\

513: x^2y + y^2z + z^2x  &  \w x, \ww y, z  &  3  &  x, xz    \vspace{-8pt}  \\

514: &&&\\

515: %x^2y + yz^2 + z^2x   

516: &  -\w x, -\ww y, -z  &  6  &  x    \vspace{-8pt}  \\

517: &&&\\

518: x^2z + xy^2 + z^3  &  x, iy, -z & 4  &  z, xz    \vspace{-8pt}  \\

519: &&& \\

520: %x^2z + xy^2 + z^3  

521: & -ix, y, iz  &  4  &  x   \vspace{-8pt} \\

522: &&& \\

523: \hline

524: \end{array}

525: $$

526: It is possible to relate complex crystallographic groups to the two singularities

527: with 2-parameter miniversal deformations, but this will be done in another paper.   

528: 

529: \bigskip

530: Of course, one may consider a bit more general problem of finding all possible 

531: smoothable symmetry {\em groups\/} $G$ of parabolic singularities, which are not necessarily cyclic.

532: In this situation the second homology splits into irreducible represenations of $G$, and we may be still looking

533: for cases when the kernel of the intersection form is shared by two of them.

534: %, let us denote these $\mathcal H'$ and $\mathcal H''$. 

535: For $P_8$, for example, this means that $G$ should contain one of the symmetries $g$ of the Table, 

536: and hence the affine reflection group related to $G$ will be a lower rank subgroup of the one we are relating to $g$.

537: This does not leave too much dimensional room for further interesting crystallographic groups.

538:  

539: 

540: %\section{The monodromy groups}

541: %The aim of this Section is to obtain the monodromy actions on the subspaces $H^2_{\chi}$

542: %in the cohomology of symmetric Milnor fibres, where the $\chi$ are the kernel

543: %characters. These will be the representations of the fundamental groups of the 

544: %complements to the discriminants in the $\R_g$-miniversal deformations 

545: %of the singularities. 

546: 

547: \section{Dynkin diagrams}

548: The equivariant monodromy of a $g$-equivariant function singularity $f$, that is, the one

549: within an $\R_g$-miniversal deformation of $f$ preserves the direct sum decomposition (\ref{Esum}).

550: Its action on an individual summand $H_{2,\chi}$ is generated by the Picard-Lefschetz operators $h_e$

551: corresponding to vanishing $\chi$-cycles $e$:

552: $$

553: h_e(c) = c - (1-\l_e) \left< c,e \right> e /\left< e, e \right>\,.

554: $$

555: Here $\l_e \ne 1$ is the eigenvalue of $h_e$: $h_e = \l_e e$. 

556: 

557: In this Section we obtain all the necessary information describing the monodromy on

558: the kernel character subspaces of the singularities of Table \ref{Ta}. We choose 

559: distinguished sets of generators of the subspaces $H_{2,\chi}$ (cf. \cite{Brmon,GZ, AGV2,AGLV1}), calculate 

560: the intersection numbers of the elements

561: of the sets, and find the eigenvalues $\l_e$. The data will be collected into Dynkin diagrams of the singularities.

562: 

563: %We also calculate the eigenvalues $\l_e \ne 1$ of the 

564: %Picard-Lefschetz operators $h_e$ corresponding to vanishing $\chi$-cycles $e$. All this information defines

565: %the equivariant monodromy action on the kernel character subspaces in the homology:

566: %$$

567: %h_e(c) = c - (1-\l_e) \left< c,e \right> e /\left< e, e \right>\,.

568: %$$

569: 

570: 

571: \begin{pr}\label{Pdim}

572: If $\chi$ is a kernel character of a symmetric singularity from Table \ref{Ta},

573: then the rank of $H^2_{\chi}$, equivalently the rank of $H_{2,\chi}$, coincides

574: with the $\R_g$-codimension $\tau$ of the singularity.

575: \end{pr}

576: 

577: {\em Proof.} If $f$ is $g$-invariant, then a basis of one of the two $H^2_{\chi}$

578: is formed, in the notations of Section \ref{Sker}, by the sections

579: $\f_i w$, where the $\f_i$ are all $g$-invariant monomials within 

580: $\{\f_0, \dots, \f_7\}$. Their number is exactly $\tau$. (Cf. \cite{OS,W}.)

581: 

582: If $f$ is $g$-equivariant rather than invariant, the claim can be verified by

583: a direct case-by-case computation of the eigenvalues of $g^*$ on the 

584: sections $\f_0 w, \dots, \f_7 w$. The observation needs a case-free understanding.

585: \hfill{$\Box$}      

586: 

587: \medskip

588: Thus, if $\tau=1$ then the subspaces $H_{2,\chi}$ in the case of kernel $\chi$

589: are one-dimensional, hence contained in $Ker(\s_h)$, and therefore the monodromy

590: we are interested in is trivial. So, from now on we are considering only the cases

591: of at least two-parameter $\R_g$-miniversal deformations.

592: 

593: We will also forget about the $(P_8|\ZZ_3)'$ singularity. Indeed its symmetry $g$

594: is the inverse of the square of the $(P_8|\ZZ_6)'$ symmetry and the two 

595: equivariant miniversal deformations coincide. 

596: Hence the character subspaces with degenerate intersection form are

597: the same (only the actual character assignments differ: $\chi_3 = \chi_6^{-2}$)

598: and the monodromy is the same.  

599: 

600: \bigskip

601: For each remaining symmetry, we will choose a distinguished set of generators for $H_{2,\chi}$ 

602: in the way it has been done in \cite{G1,G2,GB}.

603: Namely, let $u_*$ be a point  in the complement $\C^{\tau} \setminus \Delta$ to the discriminant in the

604: base of the $\R_g$-miniversal deformation of our function. Set $V_{u_*}/g$ to be the quotient of the 

605: fibre $V_{u_*}$ under the cyclic

606: $g$-action and $(V_{u_*}/g)'$ its part corresponding to irregular orbits.  Denote

607: by $\pi$ the factorisation map $V_{u_*} \to V_{u_*}/g$. 

608: The homology group $H_2(V_{u_*}/g,(V_{u_*}/g)';\ZZ)$ is spanned by a 

609: distinguished set of relative vanishing cycles (such a set is defined in the traditional singularity theory manner). Denote them $c_1, \dots, c_k$.

610: The inverse image $\pi^{-1}(c_j)$ is the cyclic orbit of one of its components, 

611: let it be $\wt c_j$. For a character $\chi$, the chain

612: $$e_j=\sum_{i=0}^{order(g)-1} \chi^{-i}g^i(\wt c_j)$$

613: is an element of $H_{2,\chi}$.  The {\em vanishing $\chi$-cycles\/} 

614: $e_1, \dots, e_k$ span $H_{2,\chi}=H_{2,\chi}(V_{u_*},\C)$, not necessarily as a basis.

615: They are defined up to multiplication by powers of $\chi$ and the sign change.

616: %Therefore, for a kernel character $\chi = \pm i, \pm \w, \pm \ov\w$, we have a well-defined lattice

617: %$$\LL_\chi = \ZZ[1,\chi]\langle e_1,\dots,e_k \rangle \subset H_{2,\chi}\,.$$

618: 

619: \medskip

620: \noindent

621: {\bf Examples.}  (a) The function $x^n+y^2+z^2$ invariant under the transformation $g: (x,y,z) \mapsto (\e_n x,y,z)$ was denoted $A_1^{(n)}$

622: in \cite{G1}. 

623: The factorisation by the group action gives the boundary singularity $A_1: w + y^2 + z^2$, $w=x^n$, whose relative

624: vanishing homology is spanned by 

625: one semi-cycle $\{ w + y^2 + z^2 = 1: w,y,z \in {\mathbf R}, w \ge 0\}$. The corresponding vanishing $\chi$-cycle, $\chi^n=1$, $\chi \ne 1$, 

626: is formed by $n$ hemi-spheres with appropriate coefficients and has self-intersection $-n$ \cite{G1}, which is consistent

627: with the standard Morse vanishing cycle having self-intersection $-2$. 

628: The Picard-Lefschetz operator $h$ is the classical monodromy of the ordinary $A_{n-1}$ singularity. Its eigenvalue

629: on $H_{2,\chi}$ is $\l=\chi$, since the quasi-homogeneous isotopy in the family $x^n + y^2 + z^2 = e^{it}$, $t \in [0,2\pi]$, finishes

630: with the transformation $(x,y,z) \mapsto (\e_n x,-y,-z)$ whose action on the homology coincides with that of $g$. 

631: 

632: (b) The $A_2$-version of the previous singularity is $A_2^{(n)}: x^n+y^3+z^2$, with the same symmetry.

633: For it, each of the $H_{2,\chi}$ is spanned by two similarly defined 

634: $\chi$-cycles which may be chosen so that their intersection number is $n/(1-\chi)$. The two Picard-Lefschetz operators $h_j$ 

635: %are both of the order equal to the order of $\chi$, each having eigenvalues 1 and $\chi$. They 

636: satisfy the standard braiding relation $aba=bab$. 

637: %Each of the operators is actually the classical monodromy operator of the ordinary $A_{n-1}$ singularity.

638: Diagrammatically, the $A_2^{(n)}$ singularity is represented by fusing the

639: rectangular $2 \times (n-1)$ Dynkin diagram to the $A_2$ diagram which may be equipped with the markings indicating the orders of the

640: vertices and the intersection numbers. Cf. the $P_8 \to D_4^{(3)}$ part of Figure \ref{Ffus}. 

641: 

642: (c) The singularity $A_m/\ZZ_m$ is the function $x^{m+1}+yz$ with the equivariant symmetry $g: (x,y,z) \mapsto (\e_m x,\e_m y,z)$. As it has

643: been shown in \cite{G2}, its $\chi$-cycle has self-intersection $-m$, $\chi$ being any $m$th root of unity. The quasi-homogeneous

644: argument applied to the $\R_g$-miniversal deformation $x^{m+1}+yz + \a x$ demonstrates that the Picard-Lefschetz operator

645: on $H_{2,\chi}$ is multiplication by $\chi$.  

646: 

647: \begin{thm}\label{Tdd}

648: For the symmetric $P_8$ singularities and their kernel characters $\chi$, there exist distinguished sets of vanishing 

649: $\chi$-cycles described by Dynkin diagrams of Figure \ref{Fdi}.

650: \end{thm}

651: 

652: \begin{figure}[hbtp]

653: \begin{center}

654: %%\epsfxsize=12.0cm

655: %%\leavevmode

656: %%\epsffile{p8di.eps}

657: \includegraphics[width=0.8\textwidth]{p8di.eps}

658: %%\end{center}

659: \caption{Dynkin diagrams of the symmetric $P_8$ singularities. The numbers inside vertices are the orders of the Picard-Lefschetz operators

660: (order 2 omitted). The number next to a vertex is the self-intersection of the $\chi$-cycle. As earlier, edges are marked with the intersection numbers

661: of the cycles. In the diagrams of all {\em invariant\/} singularities, the multiplicity of an edge indicates the length of the pair-wise braiding relation

662: inherited from the fundamental group of the complement to the discriminant:

663: $aba=bab$ for a simple edge, $(ab)^2=(ba)^2$ for a double and $(ab)^3=(ba)^3$ for a triple.} 

664: \label{Fdi}

665: \end{center}

666: \end{figure}

667: 

668: {\em Proof\/}. We proceed on the case-by-case basis. We are considering only kernel characters of the $P_8$ singularities. 

669: 

670: \medskip

671: $\mathbf{D_4^{(3)}}$. Factorisation by the action of $\ZZ_3$ provides the boundary $D_4$ singularity. Hence, according to the above

672: Examples, a Dynkin diagram of $D_4^{(3)}$ can be obtained by the modification of the standard $D_4$ diagram: the 

673: roots should have squares $-3$ instead of $-2$, the Picard-Lefschetz operators $h_j$ become of order 3, and the intersection

674: numbers 1 of pairs of cycles change to $3/(1-\chi)$. Since $\chi=\w,\ov\w$, the latter may be reduced to $1-\w$ 

675: (for both values of $\chi$) using the ambiguities in the choice of the $\chi$-cycles. 

676: The diagram may be obtained by the fusion of the cylindrical $P_8$ diagram as shown in Figure \ref{Ffus}.   

677: 

678: \begin{figure}[hbtp]

679: \begin{center}

680: %%\epsfxsize=7.0cm

681: %%\leavevmode

682: %%\epsffile{p8fus.eps}

683: \includegraphics[width=0.48\textwidth]{p8fus.eps}

684: %%\end{center}

685: \caption{Obtaining some of the diagrams by fusion and folding.}

686: \label{Ffus}

687: \end{center}

688: \end{figure}

689: 

690: \medskip

691: $\mathbf{C_3^{(3,3)}}$. We extend the previous $\ZZ_3$-symmetry by $\ZZ_2$ acting by the sign change on $z$. This embeds the character

692: subspaces $H_{2,-\w}$ and $H_{2,-\ov\w}$ of the symmetric singularity $C_3^{(3,3)}$ into  

693: $H_{2,\w}$ and $H_{2,\ov\w}$ of $D_4^{(3)}$ as subspaces anti-invariant under the involution. 

694: This is absolutely similar to the relation between the $\ZZ_2$-symmetric singularity $C_3$ and the absolute $D_4$ singularity,cf.

695: \cite{Ab,AGV2,AGLV1}.  

696: Hence the $C_3^{(3,3)}$ diagram of Figure \ref{Fdi} is provided by the folding of the $D_4^{(3)}$ shown in Figure \ref{Ffus}.

697: 

698: 

699: \medskip

700: $\mathbf{(P_8|\Z_6)'}$. The discriminant in this case is a semi-cubic cusp, hence the relation $aba=bab$ between  

701: the two Picard-Lefschetz operators. A generic point of the discriminant corresponds to the $D_4|\ZZ_6$ singularity considered in 

702: \cite{G2}, where the self-intersection number of its vanishing $\chi$-cycle was shown to be $-3$ for $\chi=-\w,-\ov\w$. 

703: Since we are considering the kernel characters, the intersection form on the $H_{2,\chi}$ must be degenerate which implies that

704: the absolute value of the intersection number of the two $\chi$-cycles is 3. Since this number is in $\ZZ[1,\w]$, we can make it 3 

705: using the $\chi$-cycle choice ambiguities once again.

706:  

707: Finally, the Picard-Lefschetz operator $h_{D_4|\ZZ_6}$ has order 6 as the classical monodromy operator of the absolute $D_4$ singularity. 

708: Its non-trivial eigenvalue on $H_{2,\chi}$ is $\l=\ov\chi$ since the quasi-homogeneous isotopy $x^2+y^3+yz^2=e^{it}$, $t\in[0,2\pi]$,

709: finishes with the transformation $(x,y,z) \mapsto (-x, \w y, \w z)$ which coincides with $g^{-1}$ on the local homology.  

710: 

711: Thus we have arrived at the $(P_8|\ZZ_6)'$ diagram of Figure \ref{Fdi}. Due to the generating reflections coming from the classical 

712: monodromy of the $D_4$ singularities, it is natural to consider it as obtained by the fusion of Figure \ref{Ffus}.

713: 

714: \medskip

715: $\mathbf{(P_8|\Z_6)''}$. The discriminant of the versal family $x^3+y^3+yz^2 + \b xy + \a$ consist of two strata:

716: $$ 

717: A_5 = \{\a = 0\} \qquad {\rm and} \qquad 3A_1 = \{27\a+\b^3=0\}\,.

718: $$

719: The cubic tangency of the strata gives the braiding relation between the generators. 

720: Of course, the self-intersection of the $3A_1$ vanishing $\chi$-cycle is $-6$ and the order of the reflection $h_{3A_1}$ is 2.

721: As for the $A_5$ $\chi$-cycle, it is possible to homotope it to the standard $A_1^{(6)}$ vanishing $\chi$-cycle, with the self-intersection

722: $-6$ according to Example (a). Now the argument similar to that for the previous singularity gives the intersection number $6$ of the

723: two cycles.  Quasi-homogeneous considerations of the $A_5$ local vanishing form $xy + z^6 = \e$ shows that the non-trivial 

724: eigenvalue of its monodromy operator on $H_{2,\chi}$ is $\ov\chi$ since its local action coincides with $g^{-1}$. Hence the 

725: operator $h_{A_5}$ is of order 6. 

726: 

727: \medskip

728: $\mathbf{P_8|\Z_3}$. The discriminants of two real versions $x^3+y^3 \pm yz^2 + \a + \b xy + \g xz$ of a versal family are

729: the unions of the $A_2$ stratum $\a=0$ and $3A_1$ stratum $(54\a+\b^3 \pm 9\b\g^2)^2 = (\b^2 \mp 3\g^2)^3$. They are shown

730: in Figure \ref{Fnew}. The two strata are simply tangent to each other along their meeting lines.

731: 

732: Like for the previous singularity, it is possible to homotope the $A_2$ vanishing $\chi$-cycle to the standard

733: $A_1^{(3 )}$ vanishing $\chi$-cycle, with the self-intersection $-3$. An operator $h_{A_2}$ has the eigenvalue $\chi$. 

734: The self-intersection of a $3A_1$ $\chi$-cycle is $-6$ and its monodromy reflection is an involution. 

735: 

736: Assume the base point $u_*$ is chosen inside the front lips region of the left diagram of Figure \ref{Fnew}. A generic line through it is vertical.

737: Take on it a path system from $u_*$: two paths going straight to the $3A_1$ points and the third nearly straight to the $A_2$ point, bypassing

738: the upper $3A_1$ point on its way (the side is not important). Then the vanishing $\chi$-cycles may be chosen so that

739: $$ 

740: \left< e_{3A_1,lower},   e_{3A_1,upper} \right> = 3 \qquad {\rm and} \qquad  

741: \left< e_{A_2},   e_{3A_1,upper} \right> = 3\,.

742: $$

743: Indeed the absolute values of these intersection numbers must be 3 following the braiding relations between the

744: two pairs of the Picard-Lefschetz operators

745: coming from the singularities of the discriminants. Moreover, both numbers are in $\ZZ[1,\w]$ and the $\chi$-cycles may be multiplied

746: by $-1$ and powers of $\w$. 

747: 

748: %the latter due to the fact that the strata have a simple tangency and hence the intersection form on  

749: %$\{e_{A_2},e_{3A_1,upper}\}$ is proportional to the $B_2$ form.

750: 

751: Now we see that the last intersection number, $\left< e_{A_2},   e_{3A_1,lower} \right>$, is either $3\w$ or $3\ov\w$. This is guaranteed

752: by the degeneracy of the intersection form on $H_{2,\chi}$ and the number being in $\ZZ[1,\w]$. However, both options turn out to be

753: equivalent up to a braid transformation. 

754: 

755: \begin{figure}[hbtp]

756: \begin{center}

757: %%\epsfxsize=5.0cm

758: %%\leavevmode

759: %%\epsffile{new.eps}

760: \includegraphics[width=0.7\textwidth]{new.eps}

761: %%\end{center}

762: \caption{Two real versions of the $P_8|\ZZ_3$ discriminant, for $x^3+y^3+yz^2$ and $x^3+y^3-yz^2$.}

763: \label{Fnew}

764: \end{center}

765: \end{figure}

766: 

767: \medskip

768: $\mathbf{C_3^{(2,4)}}$. The discriminant coincides with the standard $C_3$ discriminant. Its smooth stratum is $2A_1$

769: and the cuspidal is $A_3$. The latter is nearly $A_1^{(4)}$ of Example (a), but with the order 4 symmetry changing the sign of one of the squared

770: variables, hence the self-intersection of a vanishing $\chi$-cycle $e_{A_3}$ is $-4$, while its operator $h_{A_3}$ has eigenvalue $-\ov\chi=\chi$.

771: Additionally, $\left<e_{2A_1},e_{2A_1}\right>$ is also $-4$ and $h_{2A_1}$ is an involution. 

772: 

773: A distinguished set of vanishing $\chi$-cycles consists of  $e_{2A_1,1},e_{2A_1,2}, e_{2A_1}$. 

774: The singularities of the discriminant guarantee that the set may be chosen so that the braiding relations between the Picard-Lefschetz

775: operators are exactly those encoded in the diagram of Figure \ref{Fdi}. These imply 

776: $$

777: \left<e_{2A_1,1}, e_{2A_1}\right> = 0 \qquad {\rm and} \qquad 

778: |\left<e_{2A_1,1},e_{2A_1,2}\right>| = |\left<e_{2A_1,2}, e_{2A_1}\right>| = 2\sqrt2\,. 

779: $$ 

780: Since the intersection numbers are in $\ZZ[1,i]$, we can make  

781: $\left<e_{2A_1,1},e_{2A_1,2}\right> = \left<e_{2A_1,2}, e_{2A_1}\right> = 2(1-i)$ by possible multiplication of the cycles by powers of $i$.

782: 

783: {\em Remark.} We prefer calling this singularity $C$ rather than $B$ since it has only one vanishing cycle coming from more

784: than one critical point on one level (cf. \cite{Ab}). 

785: 

786: 

787: \medskip

788: $\mathbf{P_8/\Z_6}$. The discriminant of the versal family $x^3+y^3+z^3 + \a y + \b z$ consists of 3 lines corresponding to the

789: three divisors $\a y + \b z$ of the cubic form $y^3 + z^3$. Each of the lines is a $2A_1^{(3)}$ stratum: each $A_1^{(3)}$ singularity

790: is off the origin and has symmetry $g^2: (x,y,z) \mapsto (\ov\w x,y,z)$. Hence all vanishing $\chi$-cycles have self-intersection $-6$.

791: Any Picard-Lefschetz operator has eigenvalue $\l = \ov\chi^2 = \chi$.

792: 

793: A distinguished set of $\chi$-cycles consists of 3 elements. 

794: Since the rank of the intersection form on $H_{2,\chi}$ is 1, the absolute value of the intersection number of any pair is 6.

795: Since the number itself is in $\ZZ[1,\w]$, we can use the ambiguities in the choice of the cycles and make 

796: $\left<e_1,e_2\right> = \left<e_2,e_3\right> = -6\w$. The rank condition implies that 

797: $\left<e_3,e_1\right>$ is also $-6\w$.

798: 

799: {\em Remark.} It is possible to show that the relation between the cycles may be assumed to be $e_1+e_2+e_3=0$. 

800: 

801: \medskip

802: $\mathbf{P_8/\Z_4}$. The discriminant of the versal family $x^2z + xy^2 + z^3 + \a x + \b yz$ has 3 strata:

803: $$

804: A_4: \a = 0\,, \qquad 2A_2: \b = 0 \,, \qquad 4A_1: 4\a + \b^2 =0\,.

805: $$

806: The $A_4$ degeneration reduces to the $A_4/\ZZ_4$ singularity of Example (c). Hence the self-intersection of $e_{A_4}$ is $-4$.

807: By the quasi-homogeneity, we see that the operator $h_{A_4}$ coincides locally with $g$ and thus has eigenvalue $\chi$ on $H_{2,\chi}$.

808: Similarly, each of the two $A_2$ degenerations is the $g^2$-symmetric singularity $A_2/\ZZ_2$. 

809: Therefore, the self-intersection of $e_{2A_2}$ is also $-4$, while $h_{2A_2}$ is an involution. It is clear

810: that  a $\chi$-cycle vanishing at $4A_1$ has the self-intersection $-8$, and $h_{4A_1}$ is also an involution.

811: 

812: A distinguished set of vanishing $\chi$-cycles contains one cycle of each of the three kinds. Since the rank of the $H_{2,\chi}$ is

813: 2, the $3 \times 3$ intersection matrix must have rank 1. Due to this, after possible multiplication of the cycles by powers of $i$,

814: we can assume that

815: $$

816: \left<e_{4A_1},e_{A_4}\right> = \left<e_{4A_1},e_{2A_2}\right> = 4(1+i)\,.

817: $$

818: This forces $\left<e_{A_4},e_{2A_2}\right> = -4$. 

819: 

820: {\em Remark.} The relation between the cycles may be assumed to be either 

821: $e_{A_4}+e_{4A_1}+ ie_{2A_2}=0$ or $ie_{A_4}+e_{4A_1}+ e_{2A_2}=0$.  \hfill{$\Box$}

822: 

823: 

824: \section{Affine monodromy}

825: 

826: We start with a reminder how a corank 1 semi-definite hermitian form $\wt q$ on $\wt V = \C^{n+1}$ defines an affine

827: reflection group on a hyperplane in the dual space \cite{Bou,GM,X9}.

828: 

829: First of all, choose a basis $e_0',e_1,\dots,e_n$ in $\wt V$ so that $e'_0$ is in the kernel $K$ of the form. We denote

830: the span of the $e_{j>0}$ by $V$ and write $v$ for the $V$-component of $\wt v \in \wt V$: $\wt v = v_0 e_0' + v$. 

831: Set $Q$ to be the matrix of the restriction $q = \wt q|_V$: $Q=(\wt q (e_i,e_j))_{i,j>0}$.

832: 

833: In the dual space $\wt V^* \simeq K^* \oplus V$, we use coordinates $\a_0,\a_1,\dots,\a_n$ so that a linear functional

834: $\wt \a$ on $\wt V$ is written as 

835: $$\wt \a (\wt v) = v_0\a_0 + v^T Q \a = v_0 \a_0 + q(v,\ov \a)\,.$$

836: 

837: Consider a reflection on $\wt V$ with a root $\wt u \notin K$ and the eigenvalue $\l$:

838: $$

839: A: \wt v \mapsto \wt v - (1-\l) \wt q (\wt v, \wt u) \wt u / \wt q (\wt u, \wt u)\,.

840: $$

841: Then its dual $A^*$ sends each hyperplane $\a_0=const$ in $\wt V^*$ into itself and on such a hyperplane it acts as

842: $$

843: \a \mapsto \a - (1 - \ov \l){\a_0u_0 + \ov q(\a,\ov u) \over \ov q(\ov u, \ov u)} \ov u\,,

844: $$

845: where $\ov q$ if the hermitian form on $V$ conjugate to $q$, with the matrix $\ov Q = Q^T$ in the chosen coordinates. 

846: For $\a_0 \ne 0$, this is an affine reflection on the hyperplane $\G = \{\a_0 = const\} \simeq V$, with the root $\ov u$,

847: mirror $\wt a (\wt u) = \a_0 u_0 + \ov q (\a, \ov u) =0$ and eigenvalue $\ov\l \,.$ For $u_0=0$, the transformation is linear.

848: 

849: \bigskip

850: We are now ready to prove our main result

851: 

852: \begin{thm} Assume a symmetry of a function singularity $P_8$ on $\C^3$ splits the kernel of the intersection form $\s_h$ 

853: and $\chi$ is a kernel character. Assume the cohomology character subspace $H^2_{\chi}$ is at least of rank 2.

854: Let $\G \subset H^2_\chi$ be the hyperplane formed by all 2-cocycles taking a fixed non-zero value on a fixed generator

855: of the line $H_{2,\chi} \cap Ker(\s_h)$.

856: Then the equivariant monodromy acts on $\G$ as a complex crystallographic group. The correspondence between the symmetric

857: singularities and affine groups is given by Table \ref{Ta}.

858: \end{thm}

859: 

860: {\em Proof\/}.

861: Number the vanishing $\chi$-cycles of each diagram in Figure \ref{Fdi} from the left to the right

862: (the numbering of the last two vertices in $D_4^{(3)}$ is not important). 

863: Omit the leftmost cycle $e_0$. In equivariant cases omit also $e_2$. It is easy to notice that

864: the remaining cycles may be multiplied by appropriate factors so that the encoded hermitian form becomes the negative 

865: of the relevant form of Figure \ref{Fgr} while the orders of the related vertices coincide. Therefore,

866: taking the remaining vanishing cycles  

867: as the basic vectors $e_{j>0}$ in $\wt V = H_{2,\chi}$ in the

868: introduction to this Section, we see that their Picard-Lefschetz operators $h_j$

869: generate the required Shephard-Todd group $L$ on $\G \subset H^2_{\chi} = \wt V^*$. 

870: 

871: The next task is to obtain the translation lattices of the crystallographic groups. 

872: The kernel of any intersection form in Figure \ref{Fdi} is spanned by the $e_0'=e_0 + {\mathbf a}$, where ${\mathbf a}$ is

873: a linear combination of the $e_{j>0}$. For example, in all $\tau =2$ cases, $\mathbf a$ is a non-zero multiple of $e_1$

874: (for equivariant singularities this is due to the two Remarks by the end of the previous Section).

875: %We will call $\mathbf a$ the {\em truncated kernel vector\/}.

876: 

877: %Assume the order of the operator $h_0$ coincides with the order of a root of $L$

878: %which generates the lattice of the required affine group, that is, with the order of some operator $h_k$, $k>0$. 

879: 

880: Assume the order of a root of $L$

881: which generates the lattice of the required affine group coincides with the order of an operator $h_k$, $k>0$. 

882: According to what has been said before the Theorem, its validity for such singularity will follow from  

883: $\mathbf a$ being a multiple of an element of the $L$-orbit of $e_k$. Hence only $\tau > 2$ singularities are still to be checked.

884: And we have for them:

885: $$

886: \begin{array}{rcclcl}

887: C_3^{(3,3)}: & \mathbf a & = & (1-\ov\w, 1) & = & A_1A_2^2 e_1 \\

888: %(P_8|\ZZ_6)' ,  (P_8|\ZZ_6)'':  & \mathbf a & = & e_1 \\

889: D_4^{(3)}:    & \mathbf a & = & (1-\ov\w,1,1)             & =  & A_1A_3^2A_1 e_2  \\

890: P_8|\ZZ_3:    & \mathbf a & = & (\ov\w - \w, -\w)    & =   & \ov\w A_2A_1^2 e_2 \\

891: C_3^{(2,4)}: & \mathbf a & = & (1+i,i) & = & iA_1^2e_2 

892: \end{array}

893: $$

894: To make the calculations suitable for any of the two kernel characters, the $A_j$ here are either the Picard-Lefschetz

895: operators or their inverses, but always with the eigenvalues either $\w$ or $-1$ or $i$. The expressions obtained show that

896: the vectors $\mathbf a$ are maximal roots of the groups $L$ in the sense \cite{H}.

897: \hfill{$\Box$}

898: 

899: \bigskip

900: \noindent

901: {\bf Remark.} The multiplicities of vertices and edges in Figure \ref{Fdi} provide nearly complete presentations of all our rank $>1$ 

902: crystallographic groups as abstract groups. To obtain all defining relations, one should add:

903: \begin{itemize}

904: \item[(i)] $(h_1h_0h_2h_0)^2 = (h_0h_2h_0h_1)^2$ to the $P_8|\ZZ_3$ diagram of $[G(3,1,2)]_2$ (cf. \cite{M}), which corresponds

905: to one of the tangency lines in Figure \ref{Fnew}, right;

906: \item[(ii)] the condition that the classical monodromy (that is, the product of all the generators) of each

907: singularity has order 3. This is exactly the extra relation from \cite{M}.

908: \end{itemize} 

909: 

910: 

911: 

912: \begin{thebibliography}{99}

913: 

914: \bibitem{A}

915: V.~I.~Arnold, {\em Normal forms of functions near degenerate critical points, 

916: the Weyl groups $A_{k}$, $D_{k}$, $E_{k}$ and Lagrangian singularities\/},  

917: Functional Analysis and its Applications {\bf 6} (1972), 254--272.

918: 

919: \bibitem{Ab}Arnold, V. I., \emph{Critical points of functions on a manifold with 

920: boundary, the simple Lie groups $B_k, C_k$ and $F_4$, and singularities of evolutes}, 

921: Russian Math. Surveys {\bf 33}:5 (1978), 99--116.

922: 

923: \bibitem{AGV2} 

924: V.~I.~Arnold, S.~M.~Gusein-Zade and A.~N.~Varchenko, 

925: {\em Singularities of Differentiable

926: maps\/}, vol. 2, Monographs in Mathematics {\bf 83}, Birkh\"auser, Boston, 1988.

927: 

928: \bibitem{AGLV1} V.~I.~Arnold, V.~V.~Goryunov, O.~V.~Lyashko and V.~A.~Vassiliev, {\em

929: Singularities I. Local and global theory\/}, Encyclopaedia of Mathematical 

930: Sciences, vol.6,

931: Dynamical Systems VI, Springer Verlag, Berlin a.o., 1993.

932: 

933: \bibitem{AGLV2} V.~I.~Arnold, V.~V.~Goryunov, O.~V.~Lyashko and V.~A.~Vassiliev, {\em

934: Singularities I. Classification and applications\/}, Encyclopaedia of Mathematical 

935: Sciences, vol.39,

936: Dynamical Systems VIII, Springer Verlag, Berlin a.o., 1993.

937: 

938: \bibitem{Bou} 

939: N.~Bourbaki, {\em Groupes et alg\`ebres de Lie\/}, Chapitres IV--VI,

940: Hermann, Paris, 1968.

941: 

942: \bibitem{Brmon}

943: E.~Brieskorn, {\em Die Monodromie der isolierten Singularit\"aten von Hyperfl\"achen\/}, Manuscr. Math. {\bf 2} (1970), 103--161.

944: 

945: \bibitem{Br}

946: E.~Brieskorn, {\em Singular elements of semi-simple algebraic groups\/},

947: Actes du Congr\`es International des Math\'ematiciens (Nice, 1970), Tome 2, 

948: Gauthier-Villars, Paris, 1971, 279--284.

949: 

950: %\bibitem{BM}  M.~Brou\'e and G.~Malle, {\em Zyklotomische Heckealgebren\/}, Ast\'erisque {\bf 212}

951: %(1993), 119--189.

952: 

953: %\bibitem{BMR} M.~Brou\'e, G.~Malle and R.~Rouquier,

954: %{\em Complex reflection groups, braid groups, Hecke algebras\/}, 

955: %J. reine angew. Math. {\bf 500} (1998), 127--190.

956: 

957: %\bibitem{C} A.~M.~Cohen, 

958: %\emph{Finite complex reflection groups}, 

959: %Annales Scientifiques de l' \'{E}cole Normale Sup\'{e}rieure {\bf 4}:9 (1976), 379--436.

960: 

961: %\bibitem{D} J.~N.~Damon, \emph{The unfolding and determinacy theorems for subgroups of

962: %$\A$ and $\K$\/}, Memoirs Amer. Math. Soc. {\bf 50}, no.306, 1984.

963: 

964: \bibitem{Gi} A.~B.~Givental,

965: \emph{Singular Lagrangian manifolds and their Lagrangian mappings\/}, 

966:  J. Soviet Math. {\bf 52} (1990), no. 4, 3246--3278. 

967: %Itogi Nauki i Tekhniki,  Current problems in mathematics. 

968: %Newest results, Vol. 33 (Russian), 55--112, 236, Akad. Nauk SSSR, Vsesoyuz. Inst. Nauchn. i Tekhn. Inform., Moscow, 1988. 

969: 

970: \bibitem{G1} V.~V.~Goryunov, \emph{Unitary reflection groups associated with singularities of 

971: functions with cyclic symmetry\/}, Russian Math. Surveys {\bf 54}:5 (1999), 873--893.

972: 

973: \bibitem{G2} V.~V.~Goryunov, \emph{Unitary reflection groups and automorphisms of simple hypersurface

974: singularities\/}, New Developments in Singularity Theory, Kluwer Academic Publishers (2001), 305--328.

975: 

976: \bibitem{X9} V.~V.~Goryunov, \emph{Symmetric $X_9$ singularities and the complex affine reflection groups\/}, 

977: Proceedings of the Steklov Institute of Mathematics {\bf 258} (2007), 44--52.

978: 

979: \bibitem{GB} V.~V.~Goryunov and C.~E.~Baines, \emph{Cyclically equivariant function singularities 

980: and unitary reflection groups $G(2m,2,n), G_9, G_{31}$}, St. Petersburg Math. J. {\bf 11}:5 (2000),

981: 761--774.

982: 

983: \bibitem{GM} V.~V.~Goryunov and S.~H.~Man, \emph{The complex crystallographic

984: groups and symmetries of $J_{10}$}, Singularity Theory and its Applications,

985: Advanced Studies in Pure Mathematics {\bf 43} (2006), Mathematical Society of Japan, Kinokuniya, Tokyo,

986: 55--72.

987: 

988: \bibitem{GZ} S.~M.~Husein-Zade, \emph{

989: The monodromy groups of isolated singularities of hypersurfaces},

990: Russian Mathematical Surveys {\bf 32} : 2 (1977), 23--69.

991: 

992: \bibitem{H}  M.~C.~Hughes, \emph{

993: Complex reflection groups\/}, Communications in Algebra {\bf 18} (1990), 3999-4029.

994: 

995: \bibitem{L} E.~Looijenga, {\em On the semi-universal deformation

996: of a simple-elliptic hypersurface singularity, part II\/},

997: Topology {\bf 17} (1978), 23--40.

998: 

999: \bibitem{M} G.~Malle, \emph{Presentations for crystallographic complex reflection groups\/},

1000: Transformation Groups {\bf 1} (1996), no.3, 1--19.

1001: 

1002: \bibitem{OS} P.~Orlik and L.~Solomon, \emph{Singularities II: automorphisms of forms\/},  Math. Ann. {\bf 231} (1978), 229--240.

1003: 

1004: \bibitem{P} V.~L.~Popov, \emph{Discrete complex reflection groups\/}, 

1005: Communications of the Mathematical Institute Rijksuniversiteit Utrecht 15--1982 (1982), 

1006: 89 pp. 

1007: 

1008: \bibitem{ST} G.~C.~Shephard and J.~A.~Todd, \emph{Finite unitary reflection groups}, 

1009: Canad. J. Math. {\bf 6} (1954), 274--304.

1010: 

1011: \bibitem{Slo}

1012: P.~Slodowy, \emph{Simple Singularities and Simple Algebraic Groups\/}, Lecture Notes in Mathematics {\bf 815}, Springer Verlag, 1980.

1013: 

1014: \bibitem{W} C. T. C. Wall, \emph{A Note on Symmetry of Singularities\/}, Bull. London Math. Soc. {\bf 12} (1980), 169--175.

1015: 

1016: 

1017: \end{thebibliography}

1018: 

1019: \ 

1020: 

1021: \begin{tabbing}

1022: Department of Mathematical Sciences,\= \qquad \qquad \qquad\quad\= Department of Mathematics,\\

1023: University of Liverpool, \> \> Ben Gurion University of the Negev,\\

1024: Liverpool L69 7ZL, \> \> P.O.B. 653, Be'er Sheva 84105,\\

1025: United Kingdom\> \> Israel\\

1026: {\em E-mail}: {\tt goryunov@liv.ac.uk} \> \> {\em E-mail}: {\tt kernerdm@math.bgu.ac.il}

1027: \end{tabbing} 

1028: 

1029: \end{document}

1030: 

1031: 

1032: \ 

1033: 

1034: \ 

1035: 

1036: \ 

1037: 

1038: \ 

1039: 

1040: \ 

1041: 

1042: \ 

1043: 

1044: \ 

1045: 

1046: \newpage 

1047: 

1048: {\bf INVARIANT diagonal actions on the standard forms}

1049: 

1050: {\bf --- those which may split hence without modulus}

1051: 

1052: \medskip

1053: \noindent

1054: (1) $x^3 + y^2z + z^3$ with $a(\ww, -1, 1)$ 

1055: \begin{itemize}

1056: \item

1057: $a=1$: \quad $(\ww, -1, 1)$ \quad $\ZZ_6$ \quad $1, z, z^2$ \quad $\chi = -\w,-\ww$

1058: \hfill{{\bf DONE}}

1059: \\

1060: discriminant: $C_3$ along $z$, with smooth stratum $2A_2$ 

1061: ($\l=\chi^2$) and cuspidal $A_2$ ($\l=-\bar\chi$);

1062: \\

1063: diagram: $C_3^{(3,3)} =$  star $D_4^{(3)}$ from ($2'$) folded like $D_4 \to C_3$, 

1064: all vertices of order 3: the $B_2^{(3,3)}$ subdiagram is that from the last 

1065: table of my first paper and the $A_2^{(3)}$ subdiagram is also from there 

1066: (mind the change of the characters which were $\w,\ww$ there)\\

1067: The intersection matrix is \qquad

1068: $-\left(\begin{array}{ccc}

1069: 3 & -{3 \over 1+\chi} & 0 \\ -{3\over 1 + \bar\chi} & 3 & -{6 \over 1+\chi} \\

1070: 0& -{6\over 1 + \bar\chi} & 6

1071: \end{array}\right)$\\

1072: Dropping $e_1$ gives $[K_5]$ \hfill{{\bf DONE}} \\

1073: $\chi=-\w$: ${\bf a}= (1-\ww,-\ww)$ (i.e. $(1,1-\ww,-\ww)\in Ker$; have not conjugated

1074: the matrix yet), in the basis $\{e_2,e_3\}$

1075: $h_2=\left(\begin{array}{cc} \ww & 2 \\ 0 & 1 \end{array}\right)$,

1076: $h_3=\left(\begin{array}{cc} 1 & 0 \\ -\ww & \ww \end{array}\right)$,

1077: ${\bf a} = h_2^2 h_3 e_2$

1078: \item

1079: $a=\w$: \quad $(1, -\w, \w)$ \quad $\ZZ_6$ \quad $1, x$ \quad $\chi = -\w,-\ww$

1080: \hfill{{\bf DONE}}

1081: \\

1082: discriminant: $A_2$ along $x$ with generic $D_4$ ($\l = \chi$)\\

1083: %the eigenvalues of both PLOs are $\chi$ \\

1084: intersection matrix for $\chi=-\w,-\ww$: \qquad 

1085: $\left(\begin{array}{ccc}

1086: -3 & 3 \\ 3 & -3

1087: \end{array}\right)$ \qquad --- Here the squares are from the $D_4|\ZZ_6$ singularity. 

1088: The absolute values of the off-diagonal elements should be 3 from the

1089: $det=0$ condition. Since the off-diagonal elements are in $\ZZ[\w]$, by the

1090: freedom in the choice of the vanishing cycles (up to multiplication

1091: by powers of $\chi$) we can reduce them to $-3$. \\

1092: Gives $[K_3(6)]$: clear as $h_2(e_2)=\chi e_2$ and ${\bf a}=e_2$ \hfill{{\bf DONE}}

1093: \item

1094: $a = \ww$: \quad $(\w, -\ww, \ww)$ \quad $\ZZ_6$ \quad $1,xz$\quad $\chi = -\w,-\ww$

1095: \hfill{{\bf DONE}}

1096: \\

1097: discriminant: $\a=0$ ($A_5$ at the origin $\sim xz+y^6$, $\l=\chi$) and $27\a + \beta^3=0$ ($3A_1$ on $y=0$, $\l=-1$) \\

1098: the $\chi$-cycle vanishing on $\a=0$ can be homotoped to the 

1099: standard vanishing $\chi$-cycle for $A_1^{(6)}$ (by the homotopies $(x e^{-2\pi i kt/3},

1100: y, z e^{2\pi i kt/3})$, $t \in [0,1]$, which slide the common base 

1101: $\{(e^{i\theta},0,e^{-i\theta}), \theta\in[0,2\pi]\}$

1102: of the thimbles $g^k \sigma$ in $xz+y^6=1$  along itself) \\

1103: now by the same argument as in the previous case, we bring the intersection form

1104: for $\chi= -\w,-\ww$ \quad to \quad

1105: $\left(\begin{array}{cc} -6 & 6 \\ 6 & -6 \end{array} \right)$   \\

1106: Gives $[K_3(6)]$: clear as $h_2(e_2)=\chi e_2$ and ${\bf a}=e_2$ (here $e_2$ 

1107: corresponds to $A_5$ and $e_1$ to $3A_1$) \hfill{{\bf DONE}}

1108: \end{itemize}

1109: 

1110: 

1111: \medskip

1112: \noindent

1113: ($2'$) $x^3 + y^2z + z^3$ with $a(\w, 1, 1)$ 

1114: \begin{itemize}

1115: \item

1116: $a=1$: \quad $(\w, 1, 1)$ \quad $\ZZ_3$ \quad $1, y, z, z^2$ \quad $\chi = \w,\ww$

1117: \quad $\l=\chi$

1118: \qquad $D_4^{(3)}$ 

1119: \hfill{{\bf DONE}}\\

1120: Dropping one of the leafs must give $[K_{25}]$ \hfill{{\bf DONE}} \\

1121: Intersection matrix:

1122: $-\left(\begin{array}{cccc} 3 & 0 & 0 & -{3 \over 1-\chi} \\

1123: 0 & 3 & 0 &  -{3 \over 1-\chi} \\ 0 & 0 & 3 &  -{3 \over 1-\chi} \\

1124:  -{3 \over 1-\bar\chi} &  -{3 \over 1-\bar\chi} &  -{3 \over 1-\bar\chi} & 3

1125: \end{array}\right)$\\ dropping $e_1$ gives

1126: ${\bf a}= (1,1,1-\bar\chi)$, 

1127: $h_2=\left(\begin{array}{ccc} \chi & 0 & -\chi \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{array}\right)$,

1128: $h_3=\left(\begin{array}{ccc} 1 & 0 & 0 \\ 0 & \chi & -\chi \\ 0 & 0 & 1 \end{array}\right)$,

1129: $h_4=\left(\begin{array}{ccc} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 1 & 1 & \chi \end{array}\right)$,

1130: $h_4h_3^2h_4e_2={\bf a}$

1131: \item

1132: $a=\w$: \quad $(\ww, \w, \w)$ \quad $\ZZ_3$ \quad $1, xy,xz$ \quad $\chi = \w,\ww$

1133: \quad square of the 3rd case in (1) 

1134: \hfill{{\bf DONE}} \\

1135: discriminant in $\dots + a + bxy + cxz$: 

1136: $18(3a+9b^2c+c^3)^2+(3b^2-c^2)^3=0$

1137: ($3A_1$, $\l=-1$) and $a=0$ ($A_2$ at the origin, $\l=\chi$). 

1138: {\bf Done} in ($2''$).

1139: \\

1140: {\bf Better and corrected version:} take $x^3+y^3-yz^2+a+bxy+cxz$ with $(\ww,\w,\w)$.

1141: This is the "-" version of the previous and later. Discriminant:

1142: $(54a+b^3-9bc^2)^2-(b^2+3c^2)^3=0$ of $3A_1$ and $a=0$ of $A_2$. The curved

1143: surface has its only singularity at the origin, its two sheets are simply tangent 

1144: to $a=0$ alternatingly along $c(c^2-b^2)=0$.   

1145: \item

1146: $a = \ww$: \quad $(1, \ww, \ww)$ \quad $\ZZ_3$ \quad $1,x$ \quad $\chi = \w,\ww$

1147: \quad square of the 2nd symmetry in (1)

1148: \hfill{{\bf DONE}}  \\

1149: so the same $\chi$-basis, same intersection form and same PLOs, however

1150: mind the difference in the senses of the $\chi$s: $\chi(here) = \chi^2(there)$\\

1151: Gives $[K_3(6)]$: clear as $h_2(e_2)=-\bar\chi e_2$ 

1152: and ${\bf a}=e_2$ \hfill{{\bf DONE}}

1153: \end{itemize}

1154: 

1155: \medskip

1156: \noindent

1157: ($2''$) $x^3 + y^3 + z^3$ with $a(\w, 1, 1)$ 

1158: \begin{itemize}

1159: \item

1160: $a=1$: \quad $(\w, 1, 1)$ \quad $\ZZ_3$ \quad $1, y, z, yz$ \qquad $D_4^{(3)}$ 

1161: \item

1162: $a=\w$: \quad $(\ww, \w, \w)$ \quad $\ZZ_3$ \quad $1, xy,xz$ \hfill{{\bf DONE}}

1163: \\

1164: discriminant in $\dots + a + bxy + cxz$: $a=0$ ($A_2$, $\l=\chi$) and

1165: $(27a+b^3+c^3)^2-4(bc)^3=0$ ($3A_1$, $\l=-1$) \\

1166: its real section $b+c=1$ is the line $a=0$, with $(b \leftrightarrow c)$-symmetric lips

1167: in $a \le 0$ which are simply tangent to $a=0$ at $b=c$. The tangency point is 

1168: $B_2^{(3)}$, the cusps are $3A_2$. Hence we can choose a vanishing

1169: basis in which the intersections are

1170: $\left(\begin{array}{ccc} -3 & 3 & 3A \\ 3 & -6 & 3 \\ 3 \bar A & 3 & -6 \end{array}\right)$   

1171: where the first two cycles give $B_2^{(3)}$ and the last two give $\ZZ_3$-symmetric

1172: $3A_2$. From $det=0$ and $3A \in \ZZ[\w]$, $A=\w,\ww$. The two choices are 

1173: actually equivalent since we can apply the braid transformation $t_2$ here

1174: ($e_2' : = h_2(e_3) = e_3+e_2$, $e_3':=e_2$) which sends the above matrix to

1175: $\left(\begin{array}{ccc} -3 & -3\bar A & 3 \\ -3A & -6 & -3 \\ 3 & -3 & -6 \end{array}\right)$.

1176: Now the twist $e_2':=-Ae_2''$, $e_3':=Ae_3''$ brings the last matrix to the

1177: first in which $A$ is replaced by $\bar A$. \\

1178: draw the discriminant in the two real versions and the section described above

1179: \hfill{{\bf DO!}}\\

1180: Dropping a $-6$-cycle gives $[G(3,1,2)]_2$: \hfill{{\bf DONE}}\\

1181: Drop $e_3$ for the first matrix with $A=\w$, up to proportionality can take ${\bf a}= 

1182: (\ww-\w,-\w)$; for $\chi=\w$, $h_1=\left(\begin{array}{cc} \w & 1-\w \\ 0 & 1 \end{array}\right)$,

1183: $h_2=\left(\begin{array}{cc} 1 & 0 \\ 1 & -1 \end{array}\right)$, $h_2h_1^2e_2=\w {\bf a}$ 

1184: 

1185: %Here we can change the orientation of $e_2'$ and notice that we could have reordered

1186: %cycles number 2 and 3. 

1187: %\qquad square of the 3rd symmetry in (1) 

1188: \item

1189: $a = \ww$: \quad $(1, \ww, \ww)$ \quad $\ZZ_3$ \quad $1,x$ 

1190: %\qquad square of the 2nd symmetry in (1)

1191: \end{itemize}

1192: 

1193: \medskip

1194: \noindent

1195: (3) $x^2y + y^2z + z^2x$ with $a(1,\w,\ww)$ 

1196: \begin{itemize}

1197: \item

1198: $a= {\overline \e}_9, \w{\overline \e}_9,\ww{\overline \e}_9$ give the same $\ZZ_9$ \quad $1$ \quad hence not interesting \quad $\chi = \w, \ww$

1199: \end{itemize}

1200: 

1201: \medskip

1202: \noindent

1203: (4) $x^2z + xy^2 + z^3$ with $a(-1,i,1)$ 

1204: \begin{itemize}

1205: \item

1206: $a=1$: \quad $(-1, i, 1)$ \quad $\ZZ_4$ \quad $1, z, z^2$ \quad $\chi = \pm i$

1207: \hfill{{\bf DONE}}

1208: \\

1209: discriminant: $B_3$ along $z$, with smooth stratum $2A_1$ ($\l=-1$) 

1210: and cuspidal $A_3$ ($\l=-\chi$); the cuspidal edge is $A_2^{(4)}=G_8$ and

1211: the tangency of the two 2-dimensional strata is $D_5|\ZZ_4$. Hence the 

1212: intersection form for $\chi=\pm i$ can be reduced to \\

1213: $\left(\begin{array}{ccc} -4 & 2(1-i) & 0 \\ 2(1+i) & -4 & 2(1-i) \\ 0 & 2(1+i) & -4

1214: \end{array} \right)$. This is up to multiplication of the first two cycles by 

1215: powers of $i$. The first two PLOs are of order 4 (making up $A_2^{(4)}$)

1216: and the third of order 2. \\

1217: Must give $[K_8]$ after leaving $2A_1$ out: ${\bf a}=(1-i,-i)$; for $\chi=-i$,

1218: $h_2= \left(\begin{array}{cc}i & 1 \\ 0 & 1\end{array}\right)$ and

1219: $h_3= \left(\begin{array}{cc}1 & 0 \\ -i & i\end{array}\right)$, $h_2^2e_3=i{\bf a}$

1220:   \hfill{{\bf DONE}}

1221: \end{itemize}

1222: 

1223: \ 

1224: 

1225: {\bf SMOOTHABLE EQUIVARIANT actions}

1226: 

1227: \medskip

1228: \noindent

1229: (1) $x^3 + y^2z + z^3$ with $a(\ww, -1, 1)$ 

1230: \begin{itemize}

1231: \item

1232: $a=-1$: \quad $(-\ww,1,-1;-1)$ ($f$ included) \quad $\ZZ_6$ \quad $z$ \quad

1233: hence not interesting

1234: \item

1235: $a=-\w$: \quad $(-1,\w,-\w;-1)$ \quad $\ZZ_6$ \quad $x$ \quad hence not interesting

1236: \item

1237: $a=\pm i$: \quad $(\pm i\ww,\mp i, \pm i;\mp i)$ \quad $\ZZ_{12}$ \quad $y$ \quad

1238: hence not interesting

1239: \end{itemize}

1240: 

1241: \medskip

1242: \noindent

1243: ($2''$) $x^3 + y^3 + z^3$ with $a(\w, 1, 1)$ 

1244: \begin{itemize}

1245: \item

1246: $a=-\ww$ \quad $(-1,-\ww,-\ww;-1)$ \quad $\ZZ_6$ \quad $x$ \quad hence not interesting

1247: \item

1248: $a=-1$ \quad $(-\w,-1,-1;-1)$ \quad $\ZZ_6$ \quad $y,z$  \hfill{{\bf DONE}} \\

1249: discriminant: 3 straight lines of $2A_2$ similar to $D_4/\ZZ_2$ \\

1250: diagram: version of the $D_4/\ZZ_2$ with order 3 vertices\\

1251: the monodromy $h$ in the family $\dots+ay$, $a=e^{2\pi i t}$, is $(-1,-1,-1)=g^3$. 

1252: Since the kernel of the intersection form must $h$-fixed, we see that $\chi^3=1$

1253: for kernel characters. In each of the $2A_2$ degenerations, near each of

1254: the critical points, the local monodromy 

1255: transformation $g^2=(\ww,1,1)$ acts as the $\ZZ_3$ ordinary symmetry of

1256: $A_2$, hence $g^2$ has eigenvalues $\w,\ww$. Therefore,  $H_{\chi=1}=0$

1257: and the kernel splits between $\chi(g)=\w,\ww$ for each of which $H_\chi$

1258: has rank 2 (same rank for $\chi = -\w,-\ww$ as well). Now the square of each 

1259: vanishing $\chi$-cycle is $-6$. Since the rank of the intersection form

1260: for the kernel characters is 1, we see that the intersection number of two 

1261: such cycles has absolute value 6. Since this number must be in $\ZZ[\w]$,

1262: and the vanishing $\w,\ww$-cycles are defined up to multiplications by powers

1263: of $-\w$, we can assume that the intersection of the two cycles is 6. Similarly,

1264: we can assume that the intersection of the 1st and 3rd vaniching $\chi$-cycles is 6

1265: as well. Now the rank$=1$ condition gives the intersection matrix 

1266: $\left(\begin{array}{ccc} -6 & 6 & 6 \\ 6 & -6 & -6 \\ 6 & -6 & -6 \end{array}\right)$ and

1267: for all the PLOs $\l=\chi^2$ for $\chi=\w,\ww$.\\

1268: The relation between the 3 vanishing $\chi$-cycles:  \hfill{{\bf DONE}} \\

1269: take for a generic line in the base of versal deformation $\dots+\a y + z$ ($*$). 

1270: Take a distinguished set of $\chi$-vanishing cycles those vanishing along the 

1271: straight paths on $\C_\a$ from the origin to the discriminantal points 

1272: $\a=1,\w,\ww$. Due to the symmetry $k=(\a\cdot\w, x\cdot 1, y\cdot\ww, z\cdot 1)$

1273: of ($*$), we can assume that the three $\chi$-cycles are chosen so that

1274: $k: e_1 \mapsto e_2 \mapsto e_3 \mapsto e_1$. Then the intersection form 

1275: is 

1276: $6\left(\begin{array}{ccc} -1 & B & \bar B \\ \bar B & -1 & B \\ B & \bar B & -1 \end{array}\right)$. The rank=1 condition implies $B^3=-1$.  \\

1277: The relation between the cycles must be $k$-invariant, hence it is

1278: either $e_1+e_2+e_3=0$ or $e_1+\w e_2+\ww e_3=0$ or $e_1+\w e_2+ \ww e_3=0$.\\

1279: Assume it is $e_1+e_2+e_3=0$.  Since this is in the kernel of the intersection

1280: matrix, this implies $-1+B+ \bar B=0$ which means $B=-\w,-\ww$.

1281: Up to swapping $e_2$ and $e_3$ (which does not affect the actual monodromy group),

1282: we can assume $B=-\w$. \\

1283: Now assume the relation is $e_1+\w e_2+\ww e_3=0$. In the set 

1284: $e_1,e_2'=\w e_2,e'_3=\ww e_3$, the intersection matrix is the one

1285: above with $\ww B$ instead of $B$. Since $e_1+e'_2+e'_3$ must be in its kernel,

1286: we again obtain $\ww B= -\w,-\ww$. \\

1287: Conclusion: we may assume that the intersection matrix is

1288: $-6\left(\begin{array}{ccc} 1 & \w & \ww \\ \ww & 1 & \w \\ \w & \ww & 1 \end{array}\right)$

1289: and that the relation is $e_1+e_2+e_3=0$. This is up to possible 

1290: swapping of $e_2$ and $e_3$.\\

1291: Gives $[K_3(3)]$, following formula (1) of the $X_9$ paper. The group can be

1292: considered as spanned by three order 3 rotations about the origin, $-1$ and $-\w$ or

1293: $-\ww$. It preserves the lattice $(\w-1)\ZZ[1,\w]$  \hfill{{\bf DONE}}

1294: \end{itemize}

1295: 

1296: 

1297: \medskip

1298: \noindent

1299: (3) $x^2y + y^2z + z^2x$ with $a(1,\w,\ww)$ 

1300: \begin{itemize}

1301: \item

1302: $a=\w$: \quad $(\w,\ww,1;\w)$ \quad $\ZZ_{3}$ \quad $x,xz$ \quad is not smoothable 

1303: \item

1304: $a=-\w$: \quad $(-\w,-\ww,-1;-\w)$ \quad $\ZZ_{6}$ \quad $x$ \quad

1305: hence not interesting

1306: \item

1307: $a=\pm 1, \pm \ww$ (when either $y$ or $z$ are in) just repeat (with cyclic permutation 

1308: of the axes) the above two cases

1309: \end{itemize}

1310: 

1311: 

1312: \medskip

1313: \noindent

1314: (4) $x^2z + xy^2 + z^3$ with $a(-1,i,1)$ 

1315: \begin{itemize}

1316: \item

1317: $a=i$: \quad $(-i,-1,i;-i)$ \quad $\ZZ_{4}$ \quad $x,yz$  \hfill{{\bf DONE}} \\

1318: discriminant in $\dots +\a x + \beta yz$: \\ $\a=0$ ($A_4$), 

1319: $\beta =0$ ($2A_2$, $A_2 \sim xy+z^3$),  $4\a+\beta^2=0$ ($4A_1$) \\

1320: the square of the $\chi$-cycle vanishing at $A_4$ is $-20$ if $\chi=1$ and $-4$

1321: otherwise; the square of the $\chi$-cycle vanishing at $4A_1$ is $-8$

1322: for any $\chi$. The intersection number $A$ of these two cycles is in $\ZZ[\chi]$.

1323: If $\chi$ is a kernel character, then the rank of the matrix

1324: $\left(\begin{array}{cc} -20 & A \\ A & -8 \end{array}\right)$ if $\chi=1$

1325: or $\left(\begin{array}{cc} -4 & A \\  \bar A & -8 \end{array}\right)$ must be 1.

1326: For $\chi=\pm 1$ this is impossible since the corresponding values of $A$

1327: are not integer. Therefore, the kernel splits between $\chi=\pm i$. \\

1328: Now $\chi=\pm i$. The self-itersection of each of the two $\ZZ_2$-anti-invariant 

1329: $A_2$ cycles in $2A_2$ is $-2$. Hence the intersection matrix of the vanishing

1330: $\chi$-cycles is

1331: $\left(\begin{array}{ccc} -4 & A & B \\ \bar A & -8 & C \\

1332: \bar B & \bar C & -4 \end{array}\right)$. It has rank 1 and $A,B,C \in \ZZ[i]$. 

1333: Hence $|A|=4\sqrt2$ and $|B|=4$. Up to multiplying the vanishing $\chi$-cycles by

1334: powers of $i$, we can assume $A=4(1-i)$ and $B=-4$:

1335: $\left(\begin{array}{ccc} -4 & 4(1-i) & -4 \\ 4(1+i) & -8 & C \\

1336: -4 & \bar C & -4 \end{array}\right)$. Then $C=4(1+i)$:

1337: $\left(\begin{array}{ccc} -4 & 4(1-i) & -4 \\ 4(1+i) & -8 & 4(1+i) \\

1338: -4 & 4(1-i) & -4 \end{array}\right)$. \\

1339: For the first PLO ($A_4$), $\l=\bar \chi$; for the second ($4A_1$), $\l=-1$;

1340: for the third ($2A_2$), $\l=\chi^2=-1$ \\

1341: The relation between the three vanishing $\chi$-cycles:

1342: either $e_1+e_2+ie_3=0$ or $ie_1+e_2+e_3=0$

1343: (sense: $h_2^{1/2}(e_1)=e_1+ce_2 = ke_3$, where $c=1,-i$ and $k=\pm1,\pm i$) \hfill{{\bf DONE}} \\

1344: Leaving the $A_4$-cycle only gives $[K_3(4)]$ as the group preserving

1345: the lattice $\ZZ[2,2i]$ with the three generators being the rotations

1346: about the origin (of order 4), about 1 (of order 2) and either about $1-i$ or

1347: about $1+i$ (order 2) \hfill{{\bf DONE}}

1348: \item

1349: $a=-i$: \quad $(i,1,-i;i)$ \quad $\ZZ_{4}$ \quad $x$ \quad hence not interesting

1350: \item

1351: $a=\pm \e_8$ \quad give only $y$ \quad hence not interesting

1352: \item

1353: $a=-1$: \quad $(1,-i,-1;-1)$ \quad $\ZZ_4$ \quad $z,y^2\sim xz$ \quad is not smoothable

1354: \end{itemize}

1355: 

1356: \medskip

1357: \noindent

1358: (5) $x^3+y^3+z^3+ 3 \beta xyz$ with $a(1,\w,\ww)$ \quad 

1359: \begin{itemize}

1360: \item

1361: $a=-1$: \quad $(-1,-\w,-\ww;-1)$ \quad $\ZZ_6$ \quad $x,xyz$ \quad hence just 1 loop

1362: \item

1363: $a=-\ww,-\w$ when either $y$ or $z$ are in reduce to the above by permuting coordinates 

1364: \end{itemize}

1365: 

1366: \medskip

1367: \noindent

1368: (6) $x^3 + y^2z + z^3  + 3 \gamma xz^2$ with $a(1,-1,1)$

1369: \begin{itemize}

1370: \item

1371: $a=-1$: \quad $(-1,1,-1;-1)$ \quad $\ZZ_2$ \quad $x,z,xy,xz^2$ \quad [should not split]

1372: \hfill{{\bf DONE}}\\

1373: this is the square of the 1st symmetry in (4). That symmetry splits the kernel between

1374: $\chi=\pm i$. Therefore now all the kernel is in $\chi=-1$. Hence no splitting now.

1375: \item

1376: $a=i$: \quad $(i,-i,i;-i)$ \quad $\ZZ_4$ \quad $y,xz^2$ \quad hence just 1 loop 

1377: \end{itemize}

1378: 

1379: 

1380: \medskip

1381: \noindent

1382: (7) $x^3+y^3+z^3+3\a xyz$ with $a(1,1,1)$

1383: \begin{itemize}

1384: \item

1385: $a=-1$: \quad $(-1,-1,-1;-1)$ \quad $\ZZ_2$ \quad $x,y,z,xyz$ \quad [should not split]

1386: \\

1387: Consider monodromy $h$ in the family $\dots+\b z$, $b=e^{2\pi i t}$. $h$ coincides

1388: with $g$, hence all the kernel (preserved by $h$) is also preserved by $g$ and

1389: thus lies in $\chi=1$. Therefore the kernel does not split. 

1390: \hfill{{\bf DONE}}

1391: \end{itemize}

1392: 

1393: \ 

1394: 

1395: {\bf NON-SMOOTHABLE EQUIVARIANT actions}

1396: 

1397: Only one such (already mentioned above, but we take it now with the inverse

1398: symmetry): 

1399: 

1400: (4) with $(1,i,-1;-1)$ \quad $\ZZ_4$ \quad $z,y^2\sim xz$. A general member of the miniversal deformation

1401: $$

1402: xy^2 + z(x^2 + z^2+\a+\b x) \qquad\qquad (*)

1403: $$ 

1404: has $2A_1$ at $y=z=0$, $x^2+\b x + \a =0$. Discriminant is that of $B_2$:

1405: $\a=0$ corresponds to $A_5$ at the origin 

1406: (further degeneration of one of the existing $A_1$'s)

1407: and $\b^2+4\a=0$ to $D_4$ (involves collapsing of both existing $A_1$'s).

1408: The vanishing cycles can be easily understod from the curve

1409: $ z(x^2 + z^2+\a+\b x)=0$

1410: 

1411: Set $x=x$, $y=vw$ and $z=w^2$ with the $\ZZ_4$-action $x\cdot1, v \cdot1, w\cdot i$.

1412: This lifts the zero levels in the above family to

1413: $$

1414: w^2(xv^2 + x^2 + w^4 + \a + \b x)=0 \qquad\qquad (**)

1415: $$

1416: The expression in the brackets is a symmetric $X_9$ singularity $X_9|\ZZ_4 = A_3^{(4)}$

1417: with an additional involution changing the sign of $v$. Cycles anti-invariant under this

1418: involution are the ordinary $A_1^{(4)}$ vanishing $\chi$-cycle ($\chi^4=1$) corresponding

1419: to $\b^2+4\a=0$ (short) and the difference of two similar cycles corresponding

1420: to $\a=0$ (long) 

1421: \marginpar{got to understand the need of the long cycle and of the extra involution}

1422: (basically we fold the $A_3^{(4)}$ diagram in two like $A_3 \to B_3$).

1423: Notice that $\chi=1$ is allowed here as the plane $w=0$ is in the zero level.

1424: 

1425: Described short and long $\chi$-cycles of $A_3^{(4)}|\ZZ_2$ are mapped 2-to-1 onto

1426: the $\chi$-cycles of the initial singularity. In the $\chi=\pm 1$ cases the degree

1427: of the mapping is 0 (i.e. such $\chi$-cycles are going to 0). In the $\chi=\pm i$

1428: cases we get twice the $\chi$-cycles of (*). Both for (*) and (**), the PLOs 

1429: multiply the corresponding vanishing cycles by $\chi$. 

1430: 

1431: We see that the monodromy in (*) can be desribed in terms of the monodromy in

1432: (**). Since we are not interested in $\chi=1$, we consider (**) without the

1433: first factor. The intersection matrix for (**) is

1434: $$

1435: \left( \begin{array}{cc}

1436: -4 & {8 \over \chi -1} \\ {8 \over \bar\chi -1} & -8

1437: \end{array} \right) \qquad\qquad \chi=\pm i, -1

1438: $$

1439: We can take either this matrix or (more sensible) its quater for the intersection

1440: matrix of (*). This choice does not affect the PLOs for (*) which come out as

1441: $$

1442: \left( \begin{array}{cc}

1443: \chi & 2\chi \\ 0 & 1 \end{array} \right) \qquad {\rm and } \qquad

1444: \left( \begin{array}{cc}

1445: 1 & 0 \\ -1 & \chi \end{array} \right) \qquad i=\pm i

1446: $$

1447: Both operators have $\l=1$ eigenvector $(\chi-1,1)$. Now we go the linear forms

1448: on $\C^2$ which take on this vector a fixed non-zero value, say, 1:

1449: $$

1450: a(\chi-1) + b =1

1451: $$

1452: The dual action of the two PLOs corresponds to transposing their matrices. 

1453: Writing these actions in the $a$-chart gives

1454: $$

1455: a \mapsto \chi a \qquad {\rm and} \qquad a \mapsto \chi a -1\,.

1456: $$

1457: These generate the group $[K_3(4)]$.

1458: 

1459: \ 

1460: 

1461: {\bf Remark.} Absolutely the same approach gives the Shephard-Todd group $A_2^{(4)}$

1462: for the non-smoothable singularity $E_7/\ZZ_4$ 

1463: $$

1464: y^2 + z(z^2 + x^3 + \a x + \b)\,, \qquad x\cdot 1, y\cdot i, z \cdot (-1), f \cdot(-1)\,.

1465: $$

1466: The 2-covering singularity this time is $E_6|\ZZ_4 = A_2^{(4)}$ (no need of the extra

1467: involution, both cycles are $\chi$-short in the above sense).

1468: 

1469: 

1470: %Possible approach: multiply the miniversal deformation by $x$, set $y':=xy$

1471: %and then perhaps $z':=z^2$ in the planar singularity [may think either about 

1472: %planar $X_9$ or about a parabola $z'+x^2+\b x+\a=0$

1473: %on the plane with a corner $xz'=0$]

1474: 

1475: 

1476: \end{document}

1477: 

1478: 

1479: 

1480: \ 

1481: 

1482: \ 

1483: 

1484: \noindent

1485: {\bf Standard projective diagonal forms:}

1486: 

1487: (1) $x^3 + y^2z + z^3$ with $(\ww, -1, 1)$ \\

1488: Basis $1, x, y, z, xy, xz, y^2\sim z^2, xy^2 \sim xz^2$

1489: 

1490: (2) $x^3 + y^2z + z^3 \simeq x^3 + y^3 + z^3$ with $(\w, 1, 1)$ \\

1491: Basis $1, x, y, z, xy, xz, y^2\sim z^2, xy^2 \sim xz^2 \simeq xyz$

1492: 

1493: (3) $x^2y + y^2z + z^2x$ with $(1,\w,\ww)$ \\

1494: Basis $1,x,y,z,x^2 \sim yz, y^2 \sim xz, z^2 \sim xy, xyz \sim x^3 \sim y^3 \sim z^3$

1495: 

1496: (4) $x^2z + xy^2 + z^3$ with $(-1,i,1)\equiv (1,-i,-1)$ \\

1497: Basis $1,x,y,z,yz,x^2\sim z^2,xz\sim y^2,x^3 \sim xz^2 \sim y^2z$

1498: 

1499: (5) $x^3+y^3+z^3+ 3 \beta xyz$ with $(1,\w,\ww)$ \\

1500: Basis $1,x,y,z,xy,xz,yz,xyz$

1501: 

1502: (6) $x^3 + y^2z + z^3  + 3 \gamma xz^2$ with $(1,-1,1)$ \\

1503: Basis $1, x, y, z, xy, xz, y^2\sim z^2, xy^2 \sim xz^2$

1504: 

1505: (7) $x^3+y^3+z^3+3\a xyz$ with $(1,1,1)$ \\

1506: Basis $1,x,y,z,xy,xz,yz,xyz$

1507: 

1508: 

1509: \ 

1510: 

1511: 

1512: 

1513: 

1514: \medskip

1515: Let $V_u/g$ be the quotient of a generic fibre $V_u$ of $F$ under the 

1516: $g$-action and $(V_u/g)'$ its part corresponding to irregular orbits. Denote

1517: by $\pi$ the factorisation map $V_u \to V_u/g$. 

1518: The homology group $H_2(V_u/g,(V_u/g)';\ZZ)$ is spanned by a 

1519: distinguished set of relative vanishing cycles. Denote them $c_1, \dots, c_k$.

1520: The inverse image $\pi^{-1}(c_j)$ is the $G$-orbit of one of its components, 

1521: let it be $\wt c_j$. For a character $\chi$, the chain

1522: $$e_j=\sum_{i=0}^{order(g)-1} \chi^{-i}g^i(\wt c_j)$$

1523: is an element of $H_\chi$.  The $\chi$-cycles $e_1, \dots, e_k$ span $H_\chi$.

1524: They are defined up to multiplication by powers of $\chi$ and the sign change.

1525: Therefore, for a kernel character, we have a well-defined lattice

1526: $$\LL_\chi = \ZZ[\chi]\langle e_1,\dots,e_k \rangle \subset H_\chi\,.$$

1527: 

1528: \ 

1529: 

1530: {\bf Note:} The hermitian intersection map $\s_h: H_{2,\chi} \to H^2_{\ov\chi}$ 

1531: (i.e. the intersection

1532: with a fixed cycle gives a $\C$-linear from on $H_2$) is anti-holomorphic:

1533: $q(\cdot,\b e) = \ov \b q(\cdot,e)$. 

1534: 

1535: [Is the translation lattice of the monodromy group on the cohomology 

1536: $\s_h(\LL_\chi)$?]

1537: 

1538: 

1539: 

1540: 

1541: 

1542: 

1543: \begin{table}[p]

1544: \begin{center}

1545: \caption{Modular symmetries of  $P_8$}

1546: \vspace{-15pt}

1547: $$\begin{array}{|c|c|c|c|c|c|c|}

1548: \hline 

1549: &&&&&&\\

1550: f & g: x,y,z \mapsto & |g| & 

1551: \begin{array}{c}

1552: {\rm versal} \\

1553: {\rm monomials}

1554: \end{array}

1555: &

1556: \begin{array}{c}

1557: \!\!\!{\rm kernel}\!\!\! \\

1558: \chi

1559: \end{array}

1560: &

1561: \begin{array}{c}

1562: {\rm affine}\\

1563: {\rm group}

1564: \end{array}

1565:  & 

1566: {\rm notation} \\

1567: &&&&&&\\

1568: \hline 

1569: \hline

1570: &&&&&&\\

1571: x^3 + & -x,-\w y - \ww z  &  6  &  x, xyz  &  1  &  -  &  (P_8/\ZZ_6)  \\

1572: y^3 + z^ 3 &&&&&& \\

1573: + \alpha xyz  & -x,-y,-z  &   2   &  x,y,z,xyz & 1 & - & (P_8/\ZZ_2)' \\

1574: &&&&&&\\

1575: \hline

1576: &&&&&&\\

1577: x^3 + & -x,y,-z  &  2  &  x,z,xy,xz^2  & -1 & - & (P_8/\ZZ_2)'' \\

1578: yz^2+z^3 &&&&&& \\

1579: +\alpha xz^2  &  ix, -iy, iz  &  4  & y, xz^2  & -1 &  - & (P_8/Z_4)'' \\

1580: &&&&&& \\

1581: \hline

1582: \end{array}

1583: $$

1584: \label{Tam}

1585: \end{center}

1586: \end{table}

1587: 

1588: 

1589: 

1590: 

1591: \end{document}