1: 

2: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3: \documentclass[10pt,oneside]{jm}

4: \usepackage{amsfonts,amssymb,amsmath,url} %,showkeys}

5: 

6: \newcommand{\abs}{\vskip 0.5em\noindent}

7: \newcommand{\Abs}{\paragraph{}\hspace{-1em}}

8: \newcommand{\AbsT}[1]{\paragraph{\hspace{-1em} #1}}

9: \newcommand{\Bew}{\paragraph*{Beweis.}}

10: \newcommand{\BewT}[1]{\paragraph*{Beweis: #1.}}

11: \newcommand{\Pf}{\paragraph*{Proof.}}

12: 

13: \newcommand{\Not}{\AbsT{Notation.}}

14: \newcommand{\Def}{\AbsT{Definition.}}

15: \newcommand{\DefT}[1]{\AbsT{Definition: #1.}}

16: \newcommand{\DefBem}{\AbsT{Definition und Bemerkung.}}

17: \newcommand{\DefRem}{\AbsT{Definition and Remark.}}

18: \newcommand{\Bem}{\AbsT{Bemerkung.}}

19: \newcommand{\BemT}[1]{\AbsT{Bemerkung: #1.}}

20: \newcommand{\Rem}{\AbsT{Remark.}}

21: \newcommand{\RemT}[1]{\AbsT{Remark: #1.}}

22: \newcommand{\Bsp}{\AbsT{Beispiel.}}

23: \newcommand{\BspT}[1]{\AbsT{Beispiel: #1.}}

24: \newcommand{\Ex}{\AbsT{Example.}}

25: \newcommand{\ExT}[1]{\AbsT{Example: #1.}}

26: \newcommand{\La}{\AbsT{Lemma.}}

27: \newcommand{\Prop}{\AbsT{Proposition.}}

28: \newcommand{\PropT}[1]{\AbsT{Proposition: #1.}}

29: \newcommand{\Satz}{\AbsT{Satz.}}

30: \newcommand{\SatzT}[1]{\AbsT{Satz: #1.}}

31: \newcommand{\Thm}{\AbsT{Theorem.}}

32: \newcommand{\ThmT}[1]{\AbsT{Theorem: #1.}}

33: \newcommand{\Kor}{\AbsT{Korollar.}}

34: \newcommand{\KorT}[1]{\AbsT{Korollar: #1.}}

35: \newcommand{\Cor}{\AbsT{Corollary.}}

36: \newcommand{\CorT}[1]{\AbsT{Corollary: #1.}}

37: \newcommand{\Conj}{\AbsT{Conjecture.}}

38: \newcommand{\Alg}{\AbsT{Algorithm.}}

39: \newcommand{\AlgT}[1]{\AbsT{Algorithm: #1.}}

40: \newcommand{\Aufg}{\AbsT{Aufgabe.}}

41: \newcommand{\Exc}{\AbsT{Exercise.}}

42: \newcommand{\AufgT}[1]{\AbsT{Aufgabe: #1.}}

43: \newcommand{\ExcT}[1]{\AbsT{Exercise: #1.}}

44: \newcommand{\Hinw}{\abs{\bf Hinweis.}}

45: \newcommand{\Hint}{\abs{\bf Hint.}}

46: \newcommand{\HinwT}[1]{\abs{\bf Hinweis #1.}}

47: \newcommand{\HintT}[1]{\abs{\bf Hint #1.}}

48: \newcommand{\bfi}{\noindent {\bf i)} }

49: \newcommand{\bfii}{\noindent {\bf ii)} }

50: \newcommand{\bfiii}{\noindent {\bf iii)} }

51: \newcommand{\bfiv}{\noindent {\bf iv)} }

52: \newcommand{\bfa}{\noindent {\bf a)} }

53: \newcommand{\bfb}{\noindent {\bf b)} }

54: \newcommand{\bfc}{\noindent {\bf c)} }

55: \newcommand{\bfd}{\noindent {\bf d)} }

56: \newcommand{\bfe}{\noindent {\bf e)} }

57: \newcommand{\bff}{\noindent {\bf f)} }

58: \newcommand{\bfg}{\noindent {\bf g)} }

59: 

60: \newcommand{\eg}{e.~g.\hskip 0.5em}

61: \newcommand{\ie}{i.~e.\hskip 0.5em}

62: \newcommand{\oE}{o.~E.\hskip 0.5em}

63: \newcommand{\QED}{\hfill $\sharp$}

64: \newcommand{\QQQ}[1]{{\it ??? #1 ...}}

65: 

66: \newcommand{\N}{{\mathbb N}}

67: \newcommand{\Z}{{\mathbb Z}}

68: \newcommand{\Q}{{\mathbb Q}}

69: \newcommand{\R}{{\mathbb R}}

70: \newcommand{\C}{{\mathbb C}}

71: \newcommand{\F}{{\mathbb F}}

72: 

73: \newcommand{\cA}{{\mathcal A}}

74: \newcommand{\hcA}{{\hat{\mathcal A}}}

75: \newcommand{\tcA}{\tilde{{\mathcal A}}}

76: \newcommand{\cB}{{\mathcal B}}

77: \newcommand{\cBS}{{\mathcal BS}}

78: \newcommand{\cC}{{\mathcal C}}

79: \newcommand{\cD}{{\mathcal D}}

80: \newcommand{\cCl}{{\mathcal Cl}}

81: \newcommand{\cE}{{\mathcal E}}

82: \newcommand{\hcE}{{\hat{\mathcal E}}}

83: \newcommand{\cF}{{\mathcal F}}

84: \newcommand{\hcF}{{\hat{\mathcal F}}}

85: \newcommand{\cG}{{\mathcal G}}

86: \newcommand{\cH}{{\mathcal H}}

87: \newcommand{\tcH}{\tilde{{\mathcal H}}}

88: \newcommand{\cI}{{\mathcal I}}

89: \newcommand{\cJ}{{\mathcal J}}

90: \newcommand{\cL}{{\mathcal L}}

91: \newcommand{\cM}{{\mathcal M}}

92: \newcommand{\cK}{{\mathcal K}}

93: \newcommand{\cO}{{\mathcal O}}

94: \newcommand{\cP}{{\mathcal P}}

95: \newcommand{\cPS}{{\mathcal PS}}

96: \newcommand{\cR}{{\mathcal R}}

97: \newcommand{\cS}{{\mathcal S}}

98: \newcommand{\tcS}{\tilde{{\mathcal S}}}

99: \newcommand{\cT}{{\mathcal T}}

100: \newcommand{\cU}{{\mathcal U}}

101: \newcommand{\cV}{{\mathcal V}}

102: \newcommand{\cX}{{\mathcal X}}

103: \newcommand{\cY}{{\mathcal Y}}

104: 

105: \newcommand{\fE}{{\mathfrak E}}

106: \newcommand{\fG}{{\mathfrak G}}

107: \newcommand{\fO}{{\mathfrak O}}

108: \newcommand{\fp}{{\mathfrak p}}

109: \newcommand{\fV}{{\mathfrak V}}

110: 

111: \newcommand{\bC}{\mbox{\bf C}}

112: \newcommand{\smbC}{\mbox{\footnotesize\bf C}}

113: \newcommand{\tc}{\tilde{c}}

114: \newcommand{\tf}{\tilde{f}}

115: \newcommand{\tg}{\tilde{g}}

116: \newcommand{\ttg}{\tilde{\tilde{g}}}

117: \newcommand{\tih}{\tilde{h}}

118: \newcommand{\tH}{\tilde{H}}

119: \newcommand{\tii}{\tilde{i}}

120: \newcommand{\tj}{\tilde{j}}

121: \newcommand{\tL}{\tilde{L}}

122: \newcommand{\tir}{\tilde{r}}

123: \newcommand{\tv}{\tilde{v}}

124: \newcommand{\tV}{\tilde{V}}

125: \newcommand{\hi}{\hat{i}}

126: \newcommand{\hj}{\hat{j}}

127: \newcommand{\hk}{\hat{k}}

128: \newcommand{\hl}{\hat{l}}

129: \newcommand{\hq}{\hat{q}}

130: \newcommand{\hM}{\hat{M}}

131: \newcommand{\hV}{\hat{V}}

132: 

133: \newcommand{\al}{\alpha}

134: \newcommand{\hal}{\hat{\alpha}}

135: \newcommand{\tal}{\tilde{\alpha}}

136: \newcommand{\bt}{\beta}

137: \newcommand{\gm}{\gamma}

138: \newcommand{\tgamma}{\tilde{\gamma}}

139: \newcommand{\Gm}{\Gamma}

140: \newcommand{\tGamma}{\tilde{\Gamma}}

141: \newcommand{\dt}{\delta}

142: \newcommand{\Dt}{\Delta}

143: \newcommand{\eps}{\epsilon}

144: \newcommand{\ia}{\iota}

145: \newcommand{\ka}{\kappa}

146: \newcommand{\lb}{\lambda}

147: \newcommand{\hlb}{\hat{\lambda}}

148: \newcommand{\tlb}{{\tilde{\lambda}}}

149: \newcommand{\Lb}{\Lambda}

150: \newcommand{\om}{\omega}

151: \newcommand{\tom}{\tilde{\omega}}

152: \newcommand{\Om}{\Omega}

153: \newcommand{\tOm}{\tilde{\Omega}}

154: \newcommand{\ph}{\varphi}

155: \newcommand{\tph}{{\tilde{\varphi}}}

156: \newcommand{\htau}{\hat{\tau}}

157: \newcommand{\sg}{\sigma}

158: \newcommand{\Sg}{\Sigma}

159: \newcommand{\te}{\vartheta}

160: \newcommand{\Th}{\Theta}

161: \newcommand{\tchi}{{\tilde{\chi}}}

162: \newcommand{\yps}{\upsilon}

163: 

164: \newcommand{\alg}{\mbox{\footnotesize algebra}}

165: \newcommand{\Aut}{\mbox{Aut}}

166: \newcommand{\Bild}{\mbox{Bild}}

167: \newcommand{\chr}{\mbox{char}}

168: \newcommand{\codim}{\mbox{codim}}

169: \newcommand{\coim}{\mbox{coim}\,}

170: \newcommand{\cok}{\mbox{cok}\,}

171: \newcommand{\whdet}{\wh{\det}}

172: \newcommand{\depth}{\mbox{depth}}

173: \newcommand{\diag}{\mbox{diag}}

174: \newcommand{\diam}{\mbox{diam}}

175: \newcommand{\End}{\mbox{End}}

176: \newcommand{\smEnd}{\mbox{\footnotesize End}}

177: \newcommand{\Ext}{\mbox{Ext}}

178: \newcommand{\Fix}{\mbox{Fix}}

179: \newcommand{\Gal}{\mbox{Gal}}

180: \newcommand{\ggT}{\mbox{ggT}}

181: \newcommand{\Hom}{\mbox{Hom}}

182: \newcommand{\smHom}{\mbox{\footnotesize Hom}}

183: \newcommand{\id}{\mbox{id}}

184: \newcommand{\smid}{\mbox{\footnotesize id}}

185: \newcommand{\im}{\mbox{im}\,}

186: \newcommand{\smim}{\mbox{\footnotesize im}\,}

187: \newcommand{\IBr}{\mbox{IBr}}

188: \newcommand{\Ind}{\mbox{ind}}

189: \newcommand{\ind}{\mbox{ind}}

190: \newcommand{\Irr}{\mbox{Irr}}

191: \newcommand{\smIrr}{\mbox{\footnotesize Irr}}

192: \newcommand{\Kern}{\mbox{Kern}}

193: \newcommand{\lc}{\mbox{lc}}

194: \newcommand{\modcat}[2]{\mbox{mod}_{#1}\mbox{-}#2}

195: \newcommand{\smmodcat}[2]{\mbox{\footnotesize mod}_{#1}\mbox{-}#2}

196: \newcommand{\lmodcat}[2]{#2\mbox{-}\mbox{mod}_{#1}}

197: \newcommand{\no}{\mbox{no}}

198: \newcommand{\ord}{\mbox{ord}}

199: \newcommand{\Pot}{\mbox{Pot}}

200: \newcommand{\Quot}{\mbox{Quot}}

201: \newcommand{\Qt}{\mbox{Q}}

202: \newcommand{\rad}{\mbox{rad}}

203: \newcommand{\Res}{\mbox{Res}}

204: \newcommand{\res}{\mbox{res}}

205: \newcommand{\rk}{\mbox{rk}}

206: \newcommand{\Stab}{\mbox{Stab}}

207: \newcommand{\soc}{\mbox{soc}}

208: \newcommand{\sgn}{\mbox{sgn}}

209: \newcommand{\smsgn}{\mbox{\footnotesize sgn}}

210: \newcommand{\st}{\mbox{stop}}

211: \newcommand{\Syl}{\mbox{Syl}}

212: \newcommand{\tr}{\mbox{tr}}

213: \newcommand{\smtr}{\mbox{\footnotesize tr}}

214: \newcommand{\Tr}{\mbox{Tr}}

215: \newcommand{\trdeg}{\mbox{tr.deg}}

216: \newcommand{\wt}{\mbox{wt}}

217: \newcommand{\yes}{\mbox{yes}}

218: 

219: \newcommand{\baraut}{\ov{\mbox{\hspace{0.5em}\rule{0em}{0.55em}}}}

220: \newcommand{\hataut}{\wh{\mbox{\hspace{0.5em}\rule{0em}{0.55em}}}}

221: %\newcommand{\bl}{\un{\mbox{\hspace{0.5em}\rule{0em}{0.55em}}}}

222: \newcommand{\bl}{\mbox{?}}

223: \newcommand{\ld}{,\ldots\hskip0em ,}

224: \newcommand{\lr}[1]{\langle #1\rangle}

225: \newcommand{\ov}[1]{\overline{#1}}

226: \newcommand{\un}[1]{\underline{#1}}

227: \newcommand{\fl}[1]{\lfloor #1\rfloor}

228: \newcommand{\ti}[1]{\tilde{#1}}

229: \newcommand{\wti}[1]{\widetilde{#1}}

230: \newcommand{\wh}[1]{\widehat{#1}}

231: \newcommand{\mt}{\mapsto}

232: \newcommand{\da}{\downarrow}

233: \newcommand{\la}{\leftarrow}

234: \newcommand{\ra}{\rightarrow}

235: \newcommand{\lra}{\longrightarrow}

236: \newcommand{\llra}{\longleftrightarrow}

237: \newcommand{\Lra}{\Longrightarrow}

238: \newcommand{\cn}{\colon}

239: \newcommand{\dcup}{\stackrel{.}{\cup}}

240: \newcommand{\ptl}{\partial}

241: \newcommand{\spmid}{\,\mid\,}

242: \newcommand{\spnmid}{\,\nmid\,}

243: \newcommand{\sh}{\sharp}

244: \newcommand{\bu}{\bullet}

245: \newcommand{\wg}{\wedge}

246: \newcommand{\lemax}{<\!\!\!\cdot\,\,}

247: \newcommand{\sseq}{\subseteq}

248: 

249: \newcommand{\MAPLE}{{\sf MAPLE}}

250: \newcommand{\MAGMA}{{\sf MAGMA}}

251: \newcommand{\GAP}{{\sf GAP}}

252: \newcommand{\MA}{{\sf MeatAxe}}

253: \newcommand{\TIME}{\mbox{\sf TIME}}

254: \newcommand{\NTIME}{\mbox{\sf NTIME}}

255: \newcommand{\sfP}{\mbox{\sf P}}

256: \newcommand{\sfcoP}{\mbox{\sf coP}}

257: \newcommand{\sfNP}{\mbox{\sf NP}}

258: \newcommand{\sfcoNP}{\mbox{\sf coNP}}

259: \newcommand{\sfRP}{\mbox{\sf RP}}

260: \newcommand{\sfcoRP}{\mbox{\sf coRP}}

261: \newcommand{\sfZPP}{\mbox{\sf ZPP}}

262: \newcommand{\sfBPP}{\mbox{\sf BPP}}

263: \newcommand{\sfcoBPP}{\mbox{\sf coBPP}}

264: \newcommand{\sfFP}{\mbox{\sf FP}}

265: \newcommand{\sfFNP}{\mbox{\sf FNP}}

266: 

267: 

268: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

269: \begin{document}

270: \raggedbottom

271: \setcounter{page}{1}

272: \pagestyle{myheadings}

273: \markboth{}{}

274: \thispagestyle{empty}

275: \begin{center} \Large\bf

276: On the ring of invariants of ordinary \\ 

277: quartic curves in characteristic $2$

278: \vspace*{1em} \\

279: \normalsize\rm

280: J\"urgen M\"uller and Christophe Ritzenthaler 

281: \footnote{C. Ritzenthaler acknowledges financial support provided

282: through the European Community's Human Potential Programme under

283: contract HPRN-CT-2000-00114, GTEM.} 

284: \vspace*{1em} \\

285: \end{center}

286: 

287: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

288: 

289: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

290: \section{Introduction}\label{intro}

291: 

292: \abs

293: Traditionally, non-singular curves of a fixed genus $g$ are classified 

294: in two categories, those which are hyperelliptic and those which are not. 

295: The first ones are usually considered to be the simplest, since their 

296: geometry relies on their Weierstraß points and is condensed on a line. 

297: Thus, the hyperelliptic locus $\cM_g^{h}$ of the corresponding moduli 

298: space $\cM_g$ is easily described and well understood. Seen from an

299: invariant theoretical viewpoint, this is reflected by the fact

300: that one only needs to deal with binary forms. To the contrary, 

301: non-hyperelliptic curves are more involved. Even in the simplest case,

302: non-singular quartic curves over $\C$, no complete description of the 

303: relevant invariant ring is known. More precisely, for the invariant ring 

304: of the natural action of $SL_3(\C)$ on the vector space of homogeneous 

305: polynomials of degree $4$ in $3$ variables a set of primary invariants,

306: %homogeneous system of parameters of degrees $\{3,6,9,12,15,18,27\}$

307: see \cite{dixmier}, but no complete algebra generating set is known;

308: a conjecture of Shioda says that the invariant ring is generated

309: as an algebra by $13$ elements.

310: 

311: \abs

312: Hence the question arises whether we can describe the situation more  

313: precisely over other fields? If $F$ is a finite field of characteristic $2$, 

314: in \cite{nartritz} a complete classification of the $F$-isomorphism types 

315: of non-singular quartic curves defined over $F$ has been obtained.

316: Moreover, the stratification of the non-hyperelliptic locus $\cM_3^{nh}$

317: of the moduli space $\cM_3$ with respect to the $2$-rank of the Jacobian, 

318: and the $F$-rational points on the various strata, have been described there.

319: Here, the generic case is the one of ordinary non-singular quartic curves,

320: where the $2$-rank of the Jacobian is maximal, hence equal to $3$.

321: In \cite{nartritz}, a precise description of the ordinary non-singular 

322: quartic curves is given, and it is shown that the invariant ring associated 

323: to their moduli space $\cM_3^{ord}$ is given by a linear action of the 

324: finite group $G=GL_3(\F_2)$ on a $6$-dimensional vector space $W'^\ast$,

325: being defined over the field $\F_2$ with $2$ elements.

326: The aim of the present note is to give a complete and precise description

327: of this invariant ring, being denoted $S[W'^\ast]^G$ in the sequel,

328: thereby answering the corresponding question posed in \cite{nartritz}.

329: 

330: \abs

331: Despite the precise description of $G$ and $W'^\ast$, being suitable to be

332: handled by computer algebra systems,

333: brute force computer calculations to find primary and secondary 

334: invariants of $S[W'^\ast]^G$, using the standard press-button algorithms 

335: implemented in \MAGMA{} \cite{MAGMA}, as well as 

336: at most $2$ Gigabytes of memory and several hours of computing time, 

337: had to be abandoned unsuccessfully.

338: Hence the strategy employed here is to intertwine theoretical and 

339: computational analysis of $S[W'^\ast]^G$, which in effect leads both to some 

340: structural understanding of $S[W'^\ast]^G$ and finally to explicitly given 

341: invariants.

342: Actually, the theoretical analysis indicates how to combine ideas from 

343: computational invariant theory and tools already available in

344: computer algebra systems to obtain specially tailored techniques

345: applicable to the examples at hand.

346: Finally, the necessary computations have been carried out

347: the computer algebra systems \MAGMA{} \cite{MAGMA} and \GAP{} \cite{GAP}.

348: After all, to check the correctness of the results only needs a few 

349: seconds of computing time and approximately $10$ Megabytes of memory.

350: More details of the computations, a \MAGMA{} input file, as well as 

351: the primary and secondary invariants calculated, can be found on:

352: \url{http://www.math.jussieu.fr/~ritzenth}.

353: 

354: \abs

355: More precisely, the ring $S[W'^\ast]^G$ turns out to be Cohen-Macaulay, an 

356: optimal set of primary invariants has degrees $\{2,3,3,4,6,7\}$, and 

357: a corresponding minimal set of secondary invariants has cardinality $18$. 

358: Moreover, the algebra $S[W'^\ast]^G$ is generated by at most $12=6+6$ 

359: invariants, namely the $6$ primary invariants and $6$ of the secondary 

360: invariants, the latter having degrees $\{0,4,5,5,6,7\}$. Hence in 

361: particular $S[W'^\ast]^G$ is generated by invariants of degree at most $7$.

362: By the way, the number $12$ rings a bell: The authors wonder whether 

363: there is a connection to Shioda's conjecture mentioned above.

364: 

365: \abs

366: The paper is organized as follows:

367: In Section \ref{curves} we prepare the setting on ordinary quartic curves,

368: recall the necessary facts from \cite{nartritz}, and exhibit the $G$-module

369: $W'^\ast$ whose invariant ring $S[W'^\ast]^G$ we are going to examine.

370: In Section \ref{inv} we recall a few notions from commutative algebra and

371: derive some general facts about invariant rings needed in the sequel.

372: Finally, in Section \ref{ex} we carry out the analysis of the 

373: invariant ring $S[W'^\ast]^G$.

374: Note that we consider right group actions throughout, as this is common

375: is the computer algebra community and assumed in the computer algebra 

376: systems used.

377: 

378: 

379: 

380: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

381: \section{Ordinary quartic curves}\label{curves}

382: 

383: \Abs

384: Let $\F_2$ be the finite field with $2$ elements, and let

385: $\ov{\F}_2$ be its algebraic closure.

386: Let $M:=\F_2^3$ be the $3$-dimensional (row) 

387: vector space over $\F_2$, and let $\ov{M}:=M\otimes_{\F_2}\ov{\F}_2$.

388: Moreover, let $W:=\F_2^7$ and $\ov{W}:=W\otimes_{\F_2}\ov{\F}_2$, and let

389: $\cC:=\{C_{a,b,c,d,e,f,g};0\neq [a,b,c,d,e,f,g]\in\ov{W}\}$ 

390: be the family of quartic curves given by

391: $$ C_{a,b,c,d,e,f,g}\cn Q_{a,b,c,d,e,f}^2=g^2\cdot x y z(x+y+z),$$ 

392: where $Q_{a,b,c,d,e,f}:=ax^2+by^2+cz^2+dxy+eyz+fzx$.

393: This family is important because of the following

394: 

395: \Prop\label{sing} See \cite[Prop.1.1]{nartritz}.\\

396: Let $C$ be a non-singular quartic curve defined over $\ov{\F}_2$.

397: Then the following conditions are equivalent:

398: 

399: \bfi

400: The Jacobian variety $J_C$ of $C$ is ordinary, \ie we have

401: $|J_C[2](\ov{\F}_2)|=2^3$.

402: 

403: \bfii

404: The curve $C$ has seven bitangents.

405: 

406: \bfiii

407: The curve $C$ is isomorphic to some curve $C_{a,b,c,d,e,f,g}\in\cC$ 

408: such that

409: 

410: \abs\abs

411: $(\ast)$\hfill $abcg (a+b+d) (b+c+e) (a+c+f) (a+b+c+d+e+f+g)\neq 0$. \QED

412: 

413: \abs\abs\abs

414: Moreover, non-singular quartic curves $C$ and $C'$ are isomorphic,

415: if and only if there is an element $\gm\in GL_3(\ov{\F}_2)$ such that 

416: $C^\gm=C'$, where $\mathbb{P}(\ov{M})$ is considered as the natural

417: right module for the projective general linear group $PGL_3(\ov{\F}_2)$

418: of rank $3$ over $\ov{\F}_2$, and $C^\gm$ is the curve defined by

419: $$ V(C^\gm):=\{[x,y,z]\in\mathbb{P}(\ov{M});

420:                [x,y,z]\cdot\gm^{-1}\in V(C)\} ,$$

421: where $V(\cdot)$ denotes the locus of points of the curve.

422: Using the action on the $7$ bitangents, the isomorphism issue is 

423: reduced to the finite group $G:=GL_3(\F_2)=PGL_3(\F_2)\leq PGL_3(\ov{\F}_2)$. 

424: Indeed, non-singular curves 

425: $C_{a,b,c,d,e,f,g}\in\cC$ and $C_{a',b',c',d',e',f',g'}\in\cC$ 

426: are isomorphic, if and only if there is an element $\gm\in G$ such that 

427: $(C_{a,b,c,d,e,f,g})^\gm=C_{a',b',c',d',e',f',g'}$, see \cite{nartritz}.

428: 

429: \Abs\label{gdef}

430: In the sequel let $G:=GL_3(\F_2)$ be the general linear group 

431: of degree $3$ over $\F_2$, which up to isomorphism is the unique 

432: simple group of order $168$, see \cite[p.3]{Atlas}.

433: Let $A,B,C\in G$ be the elements of order $2$, $3$ and $7$, 

434: respectively, defined as 

435: $$ A:=\left[\begin{array}{ccc}

436: 1 & . & 1 \\

437: . & 1 & . \\

438: . & . & 1 \\

439: \end{array}\right], \quad

440: B:=\left[\begin{array}{ccc}

441: . & . & 1 \\

442: 1 & . & . \\

443: . & 1 & . \\

444: \end{array}\right], \quad

445: C:=\left[\begin{array}{ccc}

446: . & 1 & . \\

447: . & . & 1 \\

448: 1 & 1 & . \\ \end{array}\right].$$

449: 

450: \abs

451: It is easily checked using \GAP{} or \MAGMA{}

452: that $G=\langle A,B\rangle$.

453: The $\F_2$-vector space $M$ can be 

454: considered as the natural right module for $G$. As the above

455: action of $PGL_3(\ov{\F}_2)$ restricts to an action of $G$ on $\cC$,

456: it is easily checked that we get an $\F_2$-linear action of $G$ on $W$ as

457: $$D_W\cn\quad A\mt\left[\begin{array}{ccc|ccc|c}

458: 1 & . & . & . & . & . & . \\

459: . & 1 & . & . & . & . & . \\

460: 1 & . & 1 & . & . & . & . \\

461: \hline

462: . & . & . & 1 & . & . & . \\

463: . & . & . & 1 & 1 & . & . \\

464: 1 & . & . & . & . & 1 & . \\

465: \hline

466: . & . & . & 1 & . & . & 1 \\

467: \end{array}\right], \quad

468: B\mt\left[\begin{array}{ccc|ccc|c}

469: . & . & 1 & . & . & . & . \\

470: 1 & . & . & . & . & . & . \\

471: . & 1 & . & . & . & . & . \\

472: \hline

473: . & . & . & . & . & 1 & . \\

474: . & . & . & 1 & . & . & . \\

475: . & . & . & . & 1 & . & . \\

476: \hline

477: . & . & . & . & . & . & 1 \\

478: \end{array}\right] .$$

479: 

480: \abs

481: The $\F_2$-subspace 

482: $W':=\{[a,b,c,d,e,f,g]\in W;g=0\}\leq W$ is a

483: $G$-submodule of $W$, and the above matrices show that

484: $W/W'\cong\F_2$ is the trivial $G$-module. Hence

485: we have an extension of $G$-modules

486: 

487: \abs\abs

488: $(\ast\ast)$ \hfill $\{0\}\ra W'\ra W\ra\F_2\ra\{0\}.$ \hfill \mbox{}

489: 

490: \abs\abs

491: Moreover, there is a $G$-submodule $W''\leq W'$, 

492: such that $\dim_{\F_2}(W'')=3$, as is also indicated above. 

493: The characteristic polynomials of the action of $C\in G$ on $W''$ and 

494: $W'/W''$ are $t^3+t+1\in\F_2[t]$ and $t^3+t^2+1\in\F_2[t]$,

495: respectively, which both are irreducible. Hence by \cite[p.3]{ModAtlas} 

496: we conclude that $W''$ and $W'/W''$ are non-isomorphic

497: absolutely irreducible $G$-modules, where $W''\cong M$ is isomorphic to 

498: the natural representation of $G$, and $W'/W''$

499: is obtained from $W''$ by applying the automorphism of $G$ given

500: by inverting and transposing matrices.

501: 

502: \Abs\label{ext}

503: Let $\eta\in Z^1(G,W')$ be the cocycle describing the extension $(\ast\ast)$,

504: and let $\cS_4\cong H\leq G$ be the subgroup permuting the set

505: $\{x^\ast,y^\ast,z^\ast,(x+y+z)^\ast\}\sseq M^\ast$, where 

506: $M^\ast:=\Hom_{\F_2}(M,\F_2)$ is the $G$-module contragredient to $M$

507: and $\{x^\ast,y^\ast,z^\ast\}\sseq M^\ast$ is

508: the $\F_2$-basis dual to the standard basis of $M$.

509: 

510: \abs

511: By construction, for the restriction of $\eta$ to the subgroup

512: $H$ we have $\eta|_H=0\in Z^1(H,W')$.

513: As $[G\cn H]=7$ is invertible in $\F_2$,

514: we by \cite[Cor.3.6.18]{benson} conclude that

515: $\eta=0\in H^1(G,W')\cong\mbox{Ext}_G^1(\F_2,W')$. Thus the

516: extension $(\ast\ast)$ splits, and we have $W=W'\oplus\F_2$ as $G$-modules.

517: More concretely, going over from the standard basis of $W$ 

518: to the $\F_2$-basis where the last standard basis vector is replaced by

519: $[1,1,1,1,1,1,1]\in W$, we indeed obtain

520: $$D'_W\cn\quad A\mt\left[\begin{array}{ccc|ccc|c}

521: 1 & . & . & . & . & . & . \\

522: . & 1 & . & . & . & . & . \\

523: 1 & . & 1 & . & . & . & . \\

524: \hline

525: . & . & . & 1 & . & . & . \\

526: . & . & . & 1 & 1 & . & . \\

527: 1 & . & . & . & . & 1 & . \\

528: \hline

529: . & . & . & . & . & . & 1 \\

530: \end{array}\right], \quad

531: B\mt\left[\begin{array}{ccc|ccc|c}

532: . & . & 1 & . & . & . & . \\

533: 1 & . & . & . & . & . & . \\

534: . & 1 & . & . & . & . & . \\

535: \hline

536: . & . & . & . & . & 1 & . \\

537: . & . & . & 1 & . & . & . \\

538: . & . & . & . & 1 & . & . \\

539: \hline

540: . & . & . & . & . & . & 1 \\

541: \end{array}\right] .$$

542: 

543: \abs

544: Actually, this basis change amounts to 

545: substituting the curves $C_{a,b,c,d,e,f,g}\in\cC$ by curves

546: defined by

547: $$ (a x^2 +b y^2 +c z^2+d xy+eyz+f zx)^2=g^2\cdot C_K(x,y,z) ,$$

548: where $C_K(x,y,z)=x^4+y^4+z^4+(xy)^2+(yz)^2+(zx)^2+xyz(x+y+z)$. 

549: We have 

550: $x^{\ast 4}+y^{\ast 4}+z^{\ast 4}

551:  +(x^\ast y^\ast)^2+(y^\ast z^\ast)^2+(z^\ast x^\ast)^2

552:  +x^\ast y^\ast z^\ast (x^\ast+y^\ast+z^\ast)\in S[M^\ast]^G$

553: by construction, where $S[M^\ast]^G\sseq S[M^\ast]$ 

554: denotes the $\F_2$-subalgebra of $G$-invariants of the symmetric 

555: algebra $S[M^\ast]$ over $M^\ast$. 

556: As for the homogeneous component $S[M^\ast]^G_4$ of $S[M^\ast]^G$ 

557: of degree $4$ we have $\dim_{\F_2}(S[M^\ast]^G_4)=1$, see \cite{elkies}, 

558: the curve $C_K\sseq\mathbb{P}(\ov{M})$ is a twist of the Klein quartic curve.

559: 

560: \abs

561: As the extension $(\ast\ast)$ splits, for the corresponding contragredient

562: $G$-modules we have $W^\ast\cong W'^\ast\oplus\F_2^\ast$. 

563: With respect to the $\F_2$-basis

564: $\{a^\ast,b^\ast,c^\ast,d^\ast,e^\ast,f^\ast\}\sseq W'^\ast$

565: dual to the standard basis of $W'$, the $G$-action on $W'^\ast$ is given as

566: $$D_{W'^\ast}\cn\quad A\mt\left[\begin{array}{ccc|ccc}

567: 1 & . & 1 & . & . & 1 \\

568: . & 1 & . & . & . & . \\ 

569: . & . & 1 & . & . & . \\

570: \hline

571: . & . & . & 1 & 1 & . \\ 

572: . & . & . & . & 1 & . \\

573: . & . & . & . & . & 1 \\

574: \end{array}\right], \quad

575: B\mt\left[\begin{array}{ccc|ccc}

576: . & . & 1 & . & . & . \\

577: 1 & . & . & . & . & . \\

578: . & 1 & . & . & . & . \\

579: \hline

580: . & . & . & . & . & 1 \\

581: . & . & . & 1 & . & . \\

582: . & . & . & . & 1 & . \\

583: \end{array}\right] .$$

584: 

585: \abs

586: Let $\ov{W'^\ast}:=W'^\ast\otimes_{\F_2}\ov{\F}_2$ and 

587: $\ov{W'}:=W'\otimes_{\F_2}\ov{\F}_2$. Moreover, let

588: $S[W'^\ast]$ denote the symmetric $\F_2$-algebra over $W'^\ast$.

589: Hence we have $S[\ov{W'^\ast}]\cong S[W'^\ast]\otimes_{\F_2}\ov{\F}_2$.

590: Moreover, by Remark \ref{fieldextrem} below we have

591: $S[\ov{W'^\ast}]^G\cong S[W'^\ast]^G\otimes_{\F_2}\ov{\F}_2$.

592: The embedding of affine rings $S[\ov{W'^\ast}]^G\sseq S[\ov{W'^\ast}]$, 

593: where $S[\ov{W'^\ast}]^G$ is the ring of $G$-invariants in $S[\ov{W'^\ast}]$,

594: defines a morphism $\ov{W'}\ra\ov{W'}/G$ of affine varieties 

595: over $\ov{\F}_2$, which as $G$ is finite is a geometric quotient,

596: see \cite[Ch.2.3]{derksen}. Hence we have proved the following

597: 

598: \Prop See \cite[Prop.1.3]{nartritz}. \\

599: The moduli space $\cM_3^{ord}$ of the ordinary quartic curves

600: is isomorphic to the open subset of the affine variety 

601: $\mbox{Spec}(S[W'^\ast]^G\otimes_{\F_2}\ov{\F}_2)$

602: given by the non-singularity conditions $(\ast)$, 

603: see Proposition \ref{sing}.

604: \QED

605: 

606: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

607: \section{Invariant rings}\label{inv}

608: 

609: \abs

610: We recall a few notions from commutative algebra and

611: derive some general facts about invariant rings;

612: as general references see \eg \cite{benson2,derksen}.

613: Note that we use the following piece of notation frequently:

614: If $\cE$ is a subset of a commutative ring $R$, then 

615: $\cE R$ denotes the ideal of $R$ 

616: generated by $\cE$.

617: 

618: \Def

619: Let $R=\bigoplus_{d\geq 0}R_d$ be a finitely generated commutative 

620: $\Z^{\geq 0}$-graded algebra over a field $F$, such that 

621: $\dim_F(R_d)<\infty$ for $d\in\N_0$, and $R_0\cong F$. 

622: Let $R_+:=\bigoplus_{d>0}R_d\lhd R$ be the {\bf irrelevant ideal},

623: and let  $H_R(t)\in\Q(t)\sseq \Q((t))$ denote the 

624: {\bf Hilbert series} of $R$.

625: 

626: \abs

627: A set $\cF:=\{f_1\ld f_n\}\sseq R_+$ of homogeneous elements,

628: where $n=\dim(R)$ is the {\bf Krull dimension} of $R$,

629: is called a {\bf homogeneous system of parameters}, if $\cF$ is 

630: algebraically independent and if $R$ is a finitely generated 

631: $F[\cF]$-module, where $F[\cF]\sseq R$ is the polynomial 

632: $F$-subalgebra of $R$ generated by $\cF$.

633: 

634: \Def

635: The $F$-algebra $R$ is called {\bf Cohen-Macaulay} if there is a 

636: homogeneous system of parameters $\cF$ such that $R$ is a free 

637: $F[\cF]$-module.

638: If $R$ is Cohen-Macaulay, then $R$ is a free $F[\cF]$-module 

639: for each homogeneous system of parameters $\cF$. 

640: Moreover, $R$ is Cohen-Macaulay if and only if $\dim(R)=\mbox{depth}(R)$.

641: Here the {\bf depth} of $R$ is the common length of 

642: all maximal {\bf regular} homogeneous sequences in $R$, 

643: see \cite[Ch.4.3]{benson2}.

644: Note that each regular sequence can be extended to a maximal one; and

645: if $R$ is Cohen-Macaulay, then the homogeneous systems of parameters of 

646: $R$ and the maximal regular homogeneous sequences in $R$ coincide.

647: 

648: \Rem\label{hilserrem}

649: Let $R$ be Cohen-Macaulay, and let $\cF=\{f_1\ld f_n\}\sseq R$ 

650: be a homogeneous system of parameters.

651: Hence $R$ is a free $F[\cF]$-module, and 

652: for a minimal homogeneous $\cF$-module generating set 

653: $\cG=\{g_1\ld g_s\}$ of $R$ we have

654: $$ H_R(t)=\frac{\sum_{j=1}^s t^{e_j}}{\prod_{i=1}^n (1-t^{d_i})}\in\Q(t) ,$$

655: where $d_i=\deg(f_i)$ and $e_j= \deg(g_j)$.

656: Hence in the Cohen-Macaulay case the degrees $d_i$ and $e_j$ 

657: of the elements of $\cF$ and $\cG$ can be read off from $H_R$.

658: 

659: \Def

660: Let $G$ be a finite group, let $V$ be an $FG$-module

661: and let $S[V]^G$ denote the ring of {\bf $G$-invariants} in

662: the symmetric algebra $S[V]$ over $V$.

663: %:=\bigoplus_{d\geq 0}\left(V^{\otimes d}/V'_d\right)$ 

664: %$V^{\otimes d}$ denotes the $d$-fold tensor power over $F$ and

665: %$$ V'_d:=\lr{v_1\otimes\cdots\otimes v_d -

666: %             v_{1\pi^{-1}}\otimes\cdots\otimes v_{d\pi^{-1}};

667: %             v_i\in V,\pi\in\cS_d}_F\leq V^{\otimes d} .$$

668: By \cite[Thm.1.3.1]{benson2} 

669: the $F$-algebra $S[V]^G$ is finitely generated and we

670: have $\dim(S[V]^G)=\dim(S[V])=\dim_F(V)$.

671: A homogeneous system of parameters $\cF$ of $S[V]^G$ is called a set of 

672: {\bf primary invariants}. A minimal set of homogeneous $F[\cF]$-module 

673: generators of $R$ is called a set of {\bf secondary invariants}.

674: 

675: \Rem\label{fieldextrem}

676: Let $F\sseq L$ be a field extension, and let $V_L:=V\otimes_F L$.

677: As for $d\in\N_0$ we have

678: $S[V]^G_d=\bigcap_{\sg\in G}\ker_{S[V]_d}(\sg -1)$, 

679: we conclude that $\dim_F(S[V]^G_d)=\dim_L(S[V_L]^G_d)$, for $d\in\N_0$. 

680: Thus we have $S[V]^G\otimes_F L=S[V_L]^G$.

681: 

682: \Def

683: Let $H\leq G$ be a subgroup such that the characteristic $\chr(F)$ 

684: of the field $F$ does not divide the index $[G\cn H]$ of $H$ in $G$.

685: Then the {\bf relative Reynolds operator} with respect to $H$ and $G$

686: is defined as

687: $$ \cR_H^G\cn S[V]^H \ra S[V]^G\cn 

688:               f \mt\frac{1}{[G\cn H]}\cdot\sum_{\sg\in H|G}f\cdot\sg ,$$

689: where the sum runs over a set of representatives of the right cosets

690: $H|G$ of $H$ in $G$. 

691: Note that we have $S[V]^G\sseq S[V]^H$, and hence

692: $\cR_H^G$ is an $S[V]^G$-module projection onto $S[V]^G$.

693: 

694: \abs\abs

695: The following Proposition is a slight generalization of the 

696: Hochster-Eagon Theorem, see \cite[Thm.4.3.6]{benson2}, saying that

697: %in the non-modular case}, \ie 

698: if $\chr(F)$ does not divide $|G|$, 

699: then $S[V]^G$ is Cohen-Macaulay; see also \cite[Prop.8.3.1]{smith}.

700: 

701: \Prop\label{cmsubgprop}

702: Let $H\leq G$ be a subgroup such that $\chr(F)$ does not divide $[G\cn H]$.

703: Provided $S[V]^H$ is Cohen-Macaulay, then $S[V]^G$ is Cohen-Macaulay as well.

704: 

705: \Pf

706: Let $\cF\sseq S[V]^G$ be a set of primary invariants of $S[V]^G$.

707: As both $F[\cF]\sseq S[V]^G$ and $S[V]^G\sseq S[V]^H$ are

708: finite ring extensions, this also holds for $F[\cF]\sseq S[V]^H$,

709: and hence $\cF\sseq S[V]^H$ is a set of primary invariants of $S[V]^H$.

710: As $S[V]^H$ is Cohen-Macaulay, it hence is a free $F[\cF]$-module.

711: Using the relative Reynolds operator $\cR_H^G$, we conclude that

712: $S[V]^G$ is a direct summand of the graded $F[\cF]$-module $S[V]^H$.

713: Hence $S[V]^G$ is a finitely generated graded projective $F[\cF]$-module, 

714: and thus by \cite[La.4.1.1]{benson2} is a free $F[\cF]$-module.

715: \QED

716: 

717: \Rem\label{primrem}

718: To compute primary invariants in our particular situation in

719: Section \ref{ex}, we will exploit the following setting.

720: 

721: \abs

722: Let $\{0\}\ra U'\stackrel{\al}{\ra} U\stackrel{\bt}{\ra}V\ra\{0\}$ 

723: be an extension of $G$-modules. Hence $\al$ induces an embedding 

724: $S[U']\sseq S[U]$, and $\bt$ induces an isomorphism 

725: $S[U]/U'S[U]\ra S[V]$. 

726: Hence we have $S[U']^G\sseq S[U]^G$ and 

727: $(S[U]/U'S[U])^G\cong S[V]^G$, 

728: where $U'S[U]\lhd S[U]$ is a $G$-submodule. Unfortunately,

729: in general we only have an embedding 

730: $S[U]^G/(U'S[U])^G\sseq (S[U]/U'S[U])^G$, but not an 

731: isomorphism, and in general $(U'S[U])^G\lhd S[U]^G$ 

732: is not generated by $S[U']_+^G$.

733: 

734: \abs

735: Let us assume that the above extension splits, and let 

736: $\gm\cn V\ra U$ such that $\gm\bt=\id_{V}$.

737: Hence $\gm$ induces an embedding $S[V]\sseq S[U]$, and we have 

738: $S[U]=S[V]\oplus U'S[U]$ as $G$-modules. Thus from

739: $S[U]^G=S[V]^G\oplus (U'S[U])^G$ we conclude that $\bt$ 

740: induces an isomorphism $S[U]^G/(U'S[U])^G\cong S[V]^G$.

741: 

742: \abs

743: Let us moreover assume that $U'\cong F$ is the trivial $G$-module,

744: and let $0\neq \hat{f}\in U'\sseq S[U]^G_1$.

745: Then we have $(U'S[U])^G=\hat{f}\cdot S[U]^G\lhd S[U]^G$, 

746: and thus $\bt$ induces an isomorphism $S[U]^G/\hat{f}S[U]^G\cong S[V]^G$.

747: As $\hat{f}\in S[U]$ is not a zero-divisor, 

748: for the corresponding Hilbert series 

749: we obtain $H_{S[V]^G}(t)=(1-t)\cdot H_{S[U]^G}(t)\in\Q(t)$.

750: 

751: \abs

752: Let $\hcF:=\{\hat{f}_0\ld\hat{f}_{n-1}\}\sseq S[U]^G$ 

753: be a set of primary invariants, such that $\hat{f}_0=\hat{f}$,

754: and let $f_i:=\hat{f}_i\bt\in S[V]^G$, for $i\in\{1\ld n-1\}$,

755: as well as $\cF:=\{f_1\ld f_{n-1}\}$.

756: Note that we have $n=\dim(S[U]^G)=\dim_F(U)$ and

757: $n-1=\dim(S[V]^G)=\dim_F(V)$.

758: As $\hcF\sseq S[U]^G$ is a set of primary invariants,

759: by the Graded Nakayama Lemma, see \cite[La.3.5.1]{derksen}, 

760: we have $\dim(S[U]^G/\hcF S[U]^G)=0$.

761: Hence using $\bt$ we find $\dim(S[V]^G/\cF S[V]^G)=0$,

762: where $|\cF|\leq n-1$. Hence by the Graded Nakayama Lemma again

763: we conclude that $\cF\sseq S[V]^G$ is set of primary invariants.

764: 

765: \abs

766: Finally, we note the following. As $\hat{f}\in S[U]^G$ is not a 

767: zero-divisor, and hence regular, we conclude that $S[U]^G$ is 

768: Cohen-Macaulay, if and only if $S[V]^G$ is. Moreover,

769: if $S[V]^G$ is Cohen-Macaulay, then there is a maximal regular

770: homogeneous sequence in $S[U]^G$ beginning with $\hat{f}$, which

771: thus is a set of primary invariants of $S[U]^G$, and the above 

772: construction indeed yields a a set of primary invariants of $S[V]^G$,

773: which is optimal in the sense of \cite[Ch.3.3.2]{derksen} if the used

774: set of primary invariants of $S[U]^G$ was.

775: 

776: \Rem\label{secrem}

777: To compute secondary invariants in our particular situation 

778: in Section \ref{ex}, we use a special adaptation of the method 

779: typically used in the case where $\chr(F)$ does not divide $|G|$, 

780: see \cite[Ch.3.5]{derksen}.

781: 

782: \abs

783: Let us assume that $R:=S[V]^G$ is known to be Cohen-Macaulay,

784: and that the Hilbert series $H_R\in\Q(t)$ and a set $\cF\sseq R_+$

785: of primary invariants are known.

786: Hence by Remark \ref{hilserrem} we have 

787: $f:=\prod_{i=1}^n(1-t^{d_i})\cdot H_R\in\Z^{\geq 0}[t]$,

788: and hence the cardinality $s$ of any minimal homogeneous $F[\cF]$-module 

789: generating set $\cG$ of $R$ is given as $s=f(1)$, while the degrees $e_j$ 

790: can be determined from the monomials occurring in $f$.

791: 

792: \abs

793: Let $\cG\sseq R$ be a set having the appropriate cardinality, 

794: whose elements are homogeneous of the appropriate degrees.

795: By the Graded Nakayama Lemma, see \cite[La.3.5.1]{derksen},

796: the set $\cG$ generates the $F[\cF]$-module $R$, if and only if 

797: $\cG$ generates the $F$-vector space $R/F[\cF]_+ R$.

798: By the assumptions made on $\cG$ we conclude that 

799: $\cG$ is a generating set of the $F[\cF]$-module $R$,

800: if and only if $\cG\sseq R/F[\cF]_+ R$ is $F$-linearly independent.

801: 

802: \abs

803: As we are developing a method to find secondary invariants, the ring $R$ and

804: hence $R/F[\cF]_+ R$ are not yet known. Thus  we proceed as follows.

805: Let $H\leq G$ be a subgroup such that $\chr(F)$ does not divide 

806: $[G\cn H]$, and let $S:=S[V]^H$. As we have $F[\cF]\sseq R\sseq S$,

807: we may consider the natural map $\pi\cn R\ra S\ra S/(\sum_{i=1}^n f_iS)$

808: of $F$-algebras. Hence we have $F[\cF]_+ R\sseq\ker(\pi)$. 

809: Conversely, let $h\in\ker(\pi)\unlhd R$, hence we have

810: $h=\sum_{i=1}^n f_i h_i$, where $h_i\in S$. Thus we have

811: $h=\cR_H^G(h)=\sum_{i=1}^n f_i\cdot\cR_H^G(h_i)\in F[\cF]_+ R$, and

812: hence $\ker(\pi)=F[\cF]_+ R$. Thus we have an embedding

813: $\pi\cn R/F[\cF]_+ R\ra S/(\sum_{i=1}^n f_iS)$. Hence

814: $\cG\sseq R/F[\cF]_+ R$ is $F$-linearly independent,

815: if and only if $\pi(\cG)\sseq S/(\sum_{i=1}^n f_iS)$ is.

816: 

817: \abs

818: Let us finally assume that $S=S[V]^H$ is Cohen-Macaulay, and that

819: $\cG'\sseq S$ is a minimal set of secondary invariants; note

820: that $\cF\sseq S$ is a set of primary invariants.

821: As $S$ is the free $F[\cF]$-module generated by $\cG'$, a description

822: of $S$ as a finitely presented commutative $F$-algebra can be derived

823: using linear algebra techniques, see \cite[Ch.3.6]{derksen}.

824: From that, a description of $S/(\sum_{i=1}^n f_iS)$ as a finitely

825: presented commutative $F$-algebra is immediately derived. Hence 

826: the $F$-linear independence of $\pi(\cG)\sseq S/(\sum_{i=1}^n f_iS)$ 

827: is easily verified or falsified using Gröbner basis techniques,

828: see \cite[Ch.3.5]{derksen}.

829: \QED

830: 

831: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

832: \section{The invariant ring $S[W'^\ast]^G$}\label{ex}

833: 

834: \abs

835: Let again $G:=GL_3(\F_2)$ and let $S[W'^\ast]^G$ be

836: the invariant ring introduced in Section \ref{ext}. 

837: We are prepared to analyze its structure, and begin with a

838: module theoretic property of the $G$-module $W$.

839: 

840: \Prop\label{tsmodprop}

841: The $G$-module $W$ is a transitive permutation module.

842: 

843: \Pf

844: If $W$ were a transitive permutation module, 

845: the corresponding point stabilizer would be a subgroup of order $24$, 

846: leading to the following sensible guess.

847: As in Section \ref{ext}, let $\cS_4\cong H\leq G$ be the subgroup 

848: permuting the set $\{x^\ast,y^\ast,z^\ast,(x+y+z)^\ast\}\sseq M^\ast$.

849: Hence $H$ fixes $w_0=[0,0,0,0,0,0,1]\in W$, and we are led to conjecture 

850: that $H=\Stab_G(w_0)$ and that $w_0\cdot G\sseq W$ is an $\F_2$-basis 

851: of $W$ being permuted by $G$.

852: To check this, let $C\in G$ be as in Section \ref{gdef}.

853: %the following element of order $7$

854: %$$ C:=\left[\begin{array}{ccc}

855: %. & 1 & . \\

856: %. & . & 1 \\

857: %1 & 1 & . \\ \end{array}\right].$$

858: Its action on $W$ is given as

859: $$D_W\cn\quad C\mt\left[\begin{array}{ccc|ccc|c}

860: . & . & 1 & . & . & . & . \\

861: 1 & . & 1 & . & . & . & . \\

862: . & 1 & . & . & . & . & . \\

863: \hline

864: . & . & 1 & . & . & 1 & . \\

865: . & . & . & 1 & 1 & . & . \\

866: . & . & . & . & 1 & . & . \\

867: \hline

868: . & . & . & . & 1 & . & 1 \\

869: \end{array}\right] .$$

870: 

871: \abs

872: Hence $\{C^i\in G;i=0,\ldots,6\}\sseq G$ is a set of representatives 

873: of the right cosets $H|G$, and 

874: $\Om:=\{w_0\cdot C^i\in W;i=0,6,1,2,3,4,5\}\sseq W$ 

875: is given as follows, where the rows indicate the elements of $\Om$ 

876: in terms of the standard basis of $W$,

877: $$\left[\begin{array}{ccccccc}

878:  . & . & . & . & . & . & 1 \\

879:  . & . & . & . & . & 1 & 1 \\ 

880:  . & . & . & . & 1 & . & 1 \\

881:  . & . & . & 1 & . & . & 1 \\

882:  . & . & 1 & . & 1 & 1 & 1 \\

883:  . & 1 & . & 1 & 1 & . & 1 \\

884:  1 & . & . & 1 & . & 1 & 1 \\

885: \end{array}\right] .$$

886: 

887: \abs

888: Hence $\Om$ is an $\F_2$-basis of $W$, and it is easily checked 

889: that it is permuted by $G$, where in particular

890: $A\mt (1,4)(2,7)\in\cS_7$ and $B\mt (2,4,3)(5,7,6)\in\cS_7$.

891: \QED

892: 

893: \abs\abs

894: By Proposition \ref{tsmodprop} the $G$-module $W'^\ast$ is a 

895: direct summand of the permutation module $W^\ast$.

896: Note that direct summands of permutation modules are also called 

897: {\bf trivial source modules}, see \cite[Ch.3.10.4]{derksen}.

898: 

899: \Prop\label{hilprop}

900: The Hilbert series $H_{S[W'^\ast]^G}\in\Q(t)$ of $S[W'^\ast]^G$ is given as

901: $$ H_{S[W'^\ast]^G}(t)=

902: \frac{1\!+\!t^4\!+\!2 t^5\!+\!t^6\!+\!t^7\!+\!t^8\!+\!2 t^9

903:       \!+\!2 t^{10}\!+\!t^{11}\!+\!t^{12}\!+\!t^{13}\!+\!

904:       2 t^{14}\!+\!t^{15}\!+\!t^{19}}

905:      {(1-t^2)\cdot (1-t^3)^2\cdot (1-t^4)\cdot (1-t^6)\cdot (1-t^7)}.$$

906: 

907: \Pf

908: As the $G$-module $W'^\ast$ is a trivial source module,

909: by \cite[Cor.3.11.4]{benson} the $G$-module $W'^\ast$ has a 

910: unique lift to a trivial source $\Z_2 G$-module $\wh{W'^\ast}$, 

911: where $\Z_2\sseq \Q_2$ is the integral closure of $\Z$ in the

912: $2$-adic completion $\Q_2$ of $\Q$. 

913: Let $\wti{W'^\ast}:=\wh{W'^\ast}\otimes_{\Z_2}\Q_2$.

914: By \cite[Prop.3.10.15]{derksen} we have

915: $H_{S[W'^\ast]^G}(t)=H_{S[\wti{W'^\ast}]^G}(t)\in\Q(t)$,

916: where by Molien's Theorem, see \cite[Thm.2.5.2]{benson2},

917: the latter is given as

918: $$ H_{S[\wti{W'^\ast}]^G}(t)=

919:   \frac{1}{|G|}\cdot\sum_{\sg\in G} 

920:   \frac{1}{\det_{\wti{W'^\ast}}(1-t\sg)} \in\Q_2(t) .$$

921: Moreover, we have 

922: $\det_{\wti{W'^\ast}}(1-t\sg)=\prod_{i=1}^7 (1-\lb_i(\sg)\cdot t)$,

923: where %$n=\dim_{\Q_2}(\wti{W'^\ast})$ and 

924: $\{\lb_1(\sg),\ldots,\lb_7(\sg)\}$

925: are the eigenvalues of the action of $\sg\in G$ in a suitable extension

926: field of $\Q_2$. Hence $\det_{\wti{W'^\ast}}(1-t\sg)$ can be evaluated

927: from the ordinary character table of $G$, see \cite[p.3]{Atlas}, 

928: and the character $\chi_{\wti{W'^\ast}}$ of $\wti{W'^\ast}$, 

929: see \cite[Ch.2.5]{benson2}. This method is implemented in \GAP,

930: where also the ordinary character table of $G$ is available.

931: 

932: \abs

933: Hence it remains to find the character $\chi_{\wti{W'^\ast}}$.

934: Thus we have to determine the trivial source lift $\wh{W'^\ast}$. 

935: By the proof of Proposition \ref{tsmodprop} we

936: have $(\F_2)_H^G\cong W\cong W'\oplus\F_2$, where $\F_2$ denotes the 

937: trivial $G$-module. Hence we have the trivial source lifts

938: $(\Z_2)_H^G\cong\wh{W}\cong\wh{W'}\oplus\Z_2$, where 

939: again $\Z_2$ denotes the trivial $G$-module. 

940: Since we have $(\wh{W})^\ast\cong\wh{W^\ast}$ as $G$-modules,

941: we obtain $((\Z_2)_H^G)^\ast\cong\wh{W^\ast}\cong\wh{W'^\ast}\oplus\Z_2$,

942: Moreover, we conclude

943: $((\Q_2)_H^G)^\ast\cong\wh{W^\ast}\otimes_{\Z_2}\Q_2

944:  \cong\wti{W'^\ast}\oplus\Q_2$.

945: By \cite[p.3]{Atlas} the character $1_H^G$ of the permutation 

946: $G$-module $(\Q_2)_H^G$ is given as $1_G+\chi_6$, where $\chi_6$ is

947: the unique irreducible character of degree $6$ and $1_G$ is

948: the trivial character. As $\chi_6$ is real-valued, we hence have 

949: $\chi_{\wti{W'^\ast}}=1_H^G-1_G=\chi_6$.

950: \QED

951: 

952: \Prop\label{cmprop}

953: The invariant ring $S[W'^\ast]^G$ is Cohen-Macaulay.

954: 

955: \Pf

956: Let $D\leq G$ be a $2$-Sylow subgroup of $G$, hence we have $|D|=8$.

957: Using the standard methods to compute primary invariants, 

958: see \cite[Ch.3.3]{derksen}, and secondary invariants in the modular case,

959: see \cite[Ch.3.5]{derksen}, implemented in \MAGMA, 

960: we find primary invariants $\{f'_1\ld f'_6\}\sseq S[W'^\ast]^D$ 

961: having degrees $\{1,1,2,2,2,4\}$, and a minimal set of secondary invariants

962: $\cG':=\{g'_0\ld g'_3\}\sseq S[W'^\ast]^D$ having degrees 

963: $\{0,3,3,6\}$, where of course $g'_0=1$. 

964: As we moreover have $|\cG'|\cdot |D|=\prod_{i=1}^6 \deg(f'_i)$, 

965: by \cite[Thm.3.7.1]{derksen} we conclude that $S[W'^\ast]^D$ 

966: is Cohen-Macaulay, and thus by Proposition \ref{cmsubgprop} 

967: the ring $S[W'^\ast]^G$ also is.

968: \QED

969: 

970: \Abs

971: We are prepared to compute primary invariants of $S[W'^\ast]^G$.

972: To do this, we first consider 

973: the permutation module $W^\ast=W'^\ast\oplus\F_2$, 

974: and compute the homogeneous components $S[W^\ast]^G_d$, for $d\leq 7$, 

975: as follows. Let $\Om^\ast=\{\om_1^\ast\ld \om_7^\ast\}\sseq W^\ast$ 

976: be the $\F_2$-basis of $W^\ast$ dual to the $\F_2$-basis 

977: $\Om\sseq W$ given in the proof of Proposition \ref{tsmodprop}.

978: As $S[W^\ast]_d$ also is a permutation module, whose $\F_2$-basis 

979: $(\Om^\ast)^d$, consisting of the monomials of degree $d$

980: in the indeterminates $\Om^\ast$, is permuted by $G$.

981: Hence $(\Om^\ast)^d$ is partitioned into $G$-orbits

982: $(\Om^\ast)^d=\coprod_{i=1}^{n_d}\cO_i$, 

983: where $n_d=\dim_{\F_2}(S[W^\ast]^G_d)$. Letting 

984: $\cO_i^+:=\sum_{f\in\cO_i}f\in S[W^\ast]^G_d$

985: denote the corresponding orbit sum, the set

986: $\{\cO_i^+;i\in\{1\ld n_d\}\}$ forms an 

987: $\F_2$-basis of $S[W^\ast]^G_d$, see also \cite[Ch.3.10]{derksen}.

988: 

989: \abs

990: As by Remark \ref{primrem} we have 

991: $H_{S[W^\ast]^G}=\frac{1}{1-t}\cdot H_{S[W'^\ast]^G}\in\Q(t)$,

992: we look for primary invariants having degrees $\{1,2,3,3,4,6,7\}$.

993: By \cite[Prop.3.3.1]{derksen}, a set 

994: $\hcF=\{\hat{f}_0\ld\hat{f}_6\}\sseq S[W^\ast]^G$ 

995: of homogeneous elements is a set of primary invariants, if and only if

996: $\dim(S[W^\ast]/\hcF S[W^\ast])=0$.

997: Krull dimensions can be computed using Gröbner basis techniques,

998: which are implemented in \MAGMA{}, and we indeed find the 

999: following set $\hcF$ of primary invariants of $S[W^\ast]^G$, 

1000: consisting of certain of the orbit sums computed above,

1001: 

1002: $$ \begin{array}{rcl}

1003: \hat{f}_0 & := & 

1004: \om^\ast_1+\om^\ast_2+\om^\ast_3+\om^\ast_4

1005: +\om^\ast_5+\om^\ast_6+\om^\ast_7, \\ 

1006: \hat{f}_1 & := & 

1007: \om^\ast_1\om^\ast_2+\om^\ast_1\om^\ast_3+\om^\ast_1\om^\ast_4

1008: +\om^\ast_1\om^\ast_5+\om^\ast_1\om^\ast_6+\om^\ast_1\om^\ast_7

1009: +\om^\ast_2\om^\ast_3+ \\

1010: & &

1011: \om^\ast_2\om^\ast_4+\om^\ast_2\om^\ast_5+\om^\ast_2\om^\ast_6

1012: +\om^\ast_2\om^\ast_7+\om^\ast_3\om^\ast_4+\om^\ast_3\om^\ast_5

1013: +\om^\ast_3\om^\ast_6+ \\

1014: & &

1015: \om^\ast_3\om^\ast_7+\om^\ast_4\om^\ast_5+\om^\ast_4\om^\ast_6

1016: +\om^\ast_4\om^\ast_7+\om^\ast_5\om^\ast_6+\om^\ast_5\om^\ast_7

1017: +\om^\ast_6\om^\ast_7, \\ 

1018: \hat{f}_2 & := & 

1019: \om^\ast_1\om^\ast_2\om^\ast_3+\om^\ast_1\om^\ast_2\om^\ast_4

1020: +\om^\ast_1\om^\ast_2\om^\ast_5+\om^\ast_1\om^\ast_2\om^\ast_7

1021: +\om^\ast_1\om^\ast_3\om^\ast_4+\om^\ast_1\om^\ast_3\om^\ast_5+\\

1022: & &

1023: \om^\ast_1\om^\ast_3\om^\ast_6+\om^\ast_1\om^\ast_4\om^\ast_6

1024: +\om^\ast_1\om^\ast_4\om^\ast_7+\om^\ast_1\om^\ast_5\om^\ast_6

1025: +\om^\ast_1\om^\ast_5\om^\ast_7+\om^\ast_1\om^\ast_6\om^\ast_7+\\

1026: & &

1027: \om^\ast_2\om^\ast_3\om^\ast_5+\om^\ast_2\om^\ast_3\om^\ast_6

1028: +\om^\ast_2\om^\ast_3\om^\ast_7+\om^\ast_2\om^\ast_4\om^\ast_5

1029: +\om^\ast_2\om^\ast_4\om^\ast_6+\om^\ast_2\om^\ast_4\om^\ast_7+\\

1030: & &

1031: \om^\ast_2\om^\ast_5\om^\ast_6+\om^\ast_2\om^\ast_6\om^\ast_7

1032: +\om^\ast_3\om^\ast_4\om^\ast_5+\om^\ast_3\om^\ast_4\om^\ast_6

1033: +\om^\ast_3\om^\ast_4\om^\ast_7+\om^\ast_3\om^\ast_5\om^\ast_7+\\

1034: & &

1035: \om^\ast_3\om^\ast_6\om^\ast_7+\om^\ast_4\om^\ast_5\om^\ast_6

1036: +\om^\ast_4\om^\ast_5\om^\ast_7+\om^\ast_5\om^\ast_6\om^\ast_7, \\

1037: %

1038: \hat{f}_3 & := &

1039: \om^\ast_1\om^\ast_2\om^\ast_6+\om^\ast_1\om^\ast_3\om^\ast_7

1040: +\om^\ast_1\om^\ast_4\om^\ast_5+\om^\ast_2\om^\ast_3\om^\ast_4+\\

1041: & &

1042: \om^\ast_2\om^\ast_5\om^\ast_7+\om^\ast_3\om^\ast_5\om^\ast_6

1043: +\om^\ast_4\om^\ast_6\om^\ast_7, \\

1044: \hat{f}_4 & := &

1045: \om^\ast_1\om^\ast_2\om^\ast_3\om^\ast_5

1046: +\om^\ast_1\om^\ast_2\om^\ast_4\om^\ast_7

1047: +\om^\ast_1\om^\ast_3\om^\ast_4\om^\ast_6

1048: +\om^\ast_1\om^\ast_5\om^\ast_6\om^\ast_7+\\

1049: & &

1050: \om^\ast_2\om^\ast_3\om^\ast_6\om^\ast_7

1051: +\om^\ast_2\om^\ast_4\om^\ast_5\om^\ast_6

1052: +\om^\ast_3\om^\ast_4\om^\ast_5\om^\ast_7, \\

1053: \end{array} $$

1054: 

1055: $$ \begin{array}{rcl}

1056: \hat{f}_5 & := &

1057: \om^\ast_1\om^\ast_2\om^\ast_3\om^\ast_4\om^\ast_5\om^\ast_6

1058: +\om^\ast_1\om^\ast_2\om^\ast_3\om^\ast_4\om^\ast_5\om^\ast_7

1059: +\om^\ast_1\om^\ast_2\om^\ast_3\om^\ast_4\om^\ast_6\om^\ast_7+ \\

1060: & &

1061: \om^\ast_1\om^\ast_2\om^\ast_3\om^\ast_5\om^\ast_6\om^\ast_7

1062: +\om^\ast_1\om^\ast_2\om^\ast_4\om^\ast_5\om^\ast_6\om^\ast_7

1063: +\om^\ast_1\om^\ast_3\om^\ast_4\om^\ast_5\om^\ast_6\om^\ast_7+ \\

1064: & &

1065: \om^\ast_2\om^\ast_3\om^\ast_4\om^\ast_5\om^\ast_6\om^\ast_7, \\

1066: \hat{f}_6 & := &

1067: \om^\ast_1\om^\ast_2\om^\ast_3\om^\ast_4\om^\ast_5\om^\ast_6\om^\ast_7 .\\

1068: \end{array} $$

1069: 

1070: \abs

1071: Actually, it turns out that there is no set of primary invariants 

1072: of $S[W^\ast]^G$ having a strictly smaller degree product,

1073: hence $\hcF$ is optimal in the sense of \cite[Ch.3.3.2]{derksen}.

1074: As $\hat{f}_0\in S[W^\ast]^G_1$, 

1075: using the technique described in Remark \ref{primrem}, we find

1076: an optimal set $\cF=\{f_1,\ld f_6\}\sseq S[W'^\ast]^G$ of 

1077: primary invariants, having degrees $\{2,3,3,4,6,7\}$.

1078: 

1079: \Abs

1080: Next we compute secondary invariants of $S[W'^\ast]^G$.

1081: Using the Hilbert series given in Proposition \ref{hilprop},

1082: we by Proposition \ref{cmprop} and Remark \ref{hilserrem}

1083: conclude that there is a minimal 

1084: set of $18$ secondary invariants, having degrees

1085: $\{0, 4, 5, 5, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13, 14, 14, 15, 19 \}.$

1086: To find such a set of secondary invariants, we first compute the homogeneous 

1087: components $S[W'^\ast]^G_d$, for $d\leq 7$, using linear algebra 

1088: techniques, see \cite[Ch.3.1]{derksen}, implemented in \MAGMA, 

1089: and then consider products of the homogeneous invariants thus found

1090: having appropriate degrees. Thus we successively generate homogeneous 

1091: elements $\cG:=\{g_1,g_2\ld g_{18}\}\in S[W'^\ast]^G$,

1092: repeatedly using the method described in Remark \ref{secrem} 

1093: to ensure that we have 

1094: $\dim_{\F_2}(\lr{\pi(g_j);j\in\{1\ld k\}}_{\F_2})=k$, 

1095: for $k\in\{1\ld 18\}$.

1096: 

1097: \abs

1098: To apply the method described in Remark \ref{secrem}, we again consider

1099: the invariant ring $S[W'^\ast]^D$, where $D\leq G$ be a $2$-Sylow 

1100: subgroup of $G$, see the proof of Proposition \ref{cmprop}.

1101: As $S[W'^\ast]^D$ is Cohen-Macaulay, using linear algebra

1102: techniques, implemented in \MAGMA, we obtain the finite presentation

1103: $S[W'^\ast]^D\cong\lr{F_1\ld F_6,G_1\ld G_3|R_1\ld R_3}$ as commutative 

1104: $\F_2$-algebras, where the relations are given as

1105: $$

1106: \begin{array}{rcl}

1107:  R_1 & := &

1108:  (F_1+F_2)^2(F_1F_2F_4+F_3F_5+F_4^2+F_4F_5) + (F_3+F_4)F_5^2 + \\

1109: && (F_1^3+F_1F_2^2+ F_1F_5+ F_2F_5)\cdot G_1

1110:   +(F_1^2F_2+ F_2^3)\cdot G_2 

1111:   +G_1^2, \\

1112:  R_2 & := &

1113:    (F_1F_2+F_2^2+F_3)F_3F_4 +

1114:    (F_1F_2^2+F_1F_3+F_2^3+F_2F_5)\cdot G_1 + \\

1115: && (F_1F_2^2+F_1F_5+F_2^3+F_2F_5)\cdot G_2 

1116:   +G_1G_2 + G_3, \\

1117:  R_3 & := & (F_2^2F_6 + F_3^2F_5) + F_2F_3\cdot G_2 + G_2^2, \\

1118: \end{array}$$

1119: where the isomorphism from the finitely presented algebra to 

1120: $S[W'^\ast]^D$ is given by $F_i\mt f'_i$ and $G_j\mt g'_j$, where

1121: $\{f'_1\ld f'_6\}\sseq S[W'^\ast]^D$ and 

1122: $\{g'_1\ld g'_3\}\sseq S[W'^\ast]^D$ 

1123: are as in the proof of Proposition \ref{cmprop}.

1124: 

1125: \abs

1126: Decomposing the set of primary invariants 

1127: $\cF\sseq S[W'^\ast]^G\sseq S[W'^\ast]^D$ into the $\F_2$-algebra 

1128: generators $\{f'_1\ld f'_6\}\cup\{g'_1\ld g'_3\}$ 

1129: of $S[W'^\ast]^D$, again using linear algebra techniques, 

1130: implemented in \MAGMA,

1131: finally yields the finite presentation

1132: $$ S[W'^\ast]^D/(\sum_{i=1}^6 f_iS[W'^\ast]^D)\cong 

1133:  \lr{F_1\ld F_6,G_1\ld G_3|R_1,\ld R_3,R'_1\ld R'_6} $$

1134: as commutative $\F_2$-algebras, where the additional relations are given as

1135: $$

1136: \begin{array}{rcl}

1137:  R'_1 & := & F_2^2 + F_4 + F_5, \\

1138:  R'_2 & := & F_1F_4 + F_2F_5 + G_1 + G_2, \\

1139:  R'_3 & := & F_2F_4 + G_1, \\

1140:  R'_4 & := & F_1^2(F_1+F_2)^2 \!+\! F_1(F_1+F_2)(F_4+F_5) \!+\! F_3(F_3+F_4) 

1141:              \!+\! F_6 \!+\! F_1\cdot G_1, \\

1142:  R'_5 & := &   F_1^2(F_3^2+F_3F_4+F_6)

1143:              + F_1F_2(F_2^2F_4 F_3F_4+F_4F_5+F_6) + \\

1144:       &    &   F_2^2(F_3F_4+F_3F_5+F_4^2) 

1145:              + (F_3+F_4)(F_5^2+F_6) + F_4^2F_5 + \\

1146:       &    &   (F_1F_3+F_1F_5+F_2^3+F_2F_4)\cdot G_1 + \\

1147:       &    &   (F_1F_2^2+F_1F_3+F_1F_4+F_1F_5)\cdot G_2 

1148:              + G_3, \\

1149:  R'_6 & := & F_1F_6(F_3+F_4) .\\

1150: \end{array}$$

1151: 

1152: \abs

1153: Finally, we end up with secondary invariants 

1154: $\cG:=\{g_1\ld g_6\}\cup\{g_7\ld g_{18}\}$,

1155: where $g_1=1$ and $\{g_1\ld g_6\}$ have degrees $\{0,4,5,5,6,7\}$, while

1156: $$ \begin{array}{rcrr|}

1157: g_7    & := & g_2^2     & (8),\\

1158: g_8    & := & g_2g_3    & (9),\\

1159: g_9    & := & g_2g_4    & (9),\\

1160: g_{10} & := & g_2g_5    &(10),\\

1161: \end{array}\quad

1162: \begin{array}{rcrr|}

1163: g_{11} & := & g_3g_4    &(10),\\

1164: g_{12} & := & g_2g_6    &(11),\\

1165: g_{13} & := & g_2^3     &(12),\\

1166: g_{14} & := & g_2^2g_3  &(13),\\

1167: \end{array}\quad

1168: \begin{array}{rcrr}

1169: g_{15} & := & g_2^2g_5  &(14),\\

1170: g_{16} & := & g_2g_3g_4 &(14),\\

1171: g_{17} & := & g_2^2g_6  &(15),\\

1172: g_{18} & := & g_2^3g_6  &(19),\\

1173: \end{array}$$

1174: where the bracketed numbers indicate the degrees. Hence in particular 

1175: we conclude that $\{f_1\ld f_6\}\cup\{g_1\ld g_6\}\sseq S[W'^\ast]^G$ 

1176: is a minimal $\F_2$-algebra generating set of $S[W'^\ast]^G$, and 

1177: in particular $S[W'^\ast]^G$ is as an $\F_2$-algebra 

1178: generated by invariants of degree at most $7$.

1179: 

1180: \Rem

1181: In conclusion we note the following observations.

1182: 

1183: \bfa

1184: As $W$ is a transitive permutation module of odd degree in 

1185: characteristic $2$, there is a non-zero $G$-invariant quadratic 

1186: form on $W$. Indeed, this form coincides with the primary invariant 

1187: $\hat{f}_1\in S[W^\ast]^G$ of degree $2$. Moreover, we have

1188: $$ f_1=a^\ast e^\ast + b^\ast f^\ast + c^\ast d^\ast 

1189:      + d^\ast e^\ast + d^\ast f^\ast + e^\ast f^\ast 

1190:      + d^{\ast 2} + e^{\ast 2} + f^{\ast 2}

1191:    \in S[W'^\ast]^G .$$

1192: Hence this shows that $D_{W'}$ is an embedding 

1193: $D_{W'}\cn G\ra SO_6^+(\F_2)$, where $SO_6^+(\F_2)$ 

1194: denotes the special orthogonal group of degree $6$ over $\F_2$

1195: of maximal Witt index type, see \cite[p.22]{Atlas}.

1196: 

1197: \abs\bfb

1198: We have $\{0\}\ra (W'/W'')^\ast\ra W'^\ast\ra W''^\ast\ra\{0\}$,

1199: %where $W''^\ast$ is obtained from the tautological representation

1200: %$W''$ of $G$ by applying the graph automorphism, 

1201: see Section \ref{gdef}.

1202: Note that this extension does not split: Assume to the 

1203: contrary it splits. Then we also have

1204: $W'\cong W''\oplus W'/W''$, and hence 

1205: $(\F_2)_H^G\cong W\cong W''\oplus W'/W''\oplus\F_2$

1206: is semisimple, and thus $\dim_{\F_2}(\End_G((\F_2)_H^G))=3$. 

1207: By the proof of Proposition \ref{hilprop} we have

1208: $\dim_{\Q_2}(\End_G((\Q_2)_H^G))=2$, a contradiction to 

1209: \cite[Thm.3.11.3]{benson}.

1210: 

1211: \abs

1212: Still we are tempted to apply the technique described at the 

1213: beginning of Remark \ref{primrem}, yielding an embedding

1214: $S[W'^\ast]^G/((W'/W'')^\ast S[W'^\ast])^G\ra S[W''^\ast]^G$.

1215: The question arises whether this map is still surjective.

1216: While the authors do not see a structural reason why this should 

1217: be the case, from the primary invariants 

1218: $\{f_4,f_5,f_6\}\sseq S[W'^\ast]^G$ 

1219: of degrees $\{4,6,7\}$ we obtain $\{c_0,c_1,c_2\}\sseq S[W''^\ast]^G$,

1220: given as

1221: $$ \begin{array}{rcl}

1222: c_2 & := & 

1223: a^{\ast 4} + b^{\ast 4} + c^{\ast 4}

1224: + a^{\ast 2}b^{\ast 2} 

1225: + a^{\ast 2}c^{\ast 2} 

1226: + b^{\ast 2} c^{\ast 2} + \\

1227: & &

1228:   a^{\ast 2}b^\ast c^\ast 

1229: + a^\ast b^{\ast 2} c^\ast 

1230: + a^\ast b^\ast c^{\ast 2}, \\

1231: c_1 & := & 

1232:   a^{\ast 4} b^{\ast 2}

1233: + a^{\ast 2} b^{\ast 4} 

1234: + a^{\ast 4} c^{\ast 2} 

1235: + a^{\ast 2} c^{\ast 4} 

1236: + b^{\ast 4} c^{\ast 2} 

1237: + b^{\ast 2} c^{\ast 4} + \\

1238: & & 

1239:   a^{\ast 4} b^\ast c^\ast 

1240: + a^\ast b^{\ast 4} c^\ast 

1241: + a^\ast b^\ast c^{\ast 4} 

1242: + a^{\ast 2} b^{\ast 2} c^{\ast 2}, \\

1243: c_0 & := & 

1244:   a^{\ast 4} b^{\ast 2} c^\ast 

1245: + a^{\ast 4} b^\ast c^{\ast 2}

1246: + a^{\ast 2} b^{\ast 4} c^\ast 

1247: + a^\ast b^{\ast 4} c^{\ast 2} 

1248: + a^{\ast 2} b^\ast c^{\ast 4} 

1249: + a^\ast b^{\ast 2} c^{\ast 4}. \\

1250: \end{array} $$

1251: It turns out that these are the {\bf Dickson invariants} of 

1252: $S[W''^\ast]^G$, see \cite[Ch.8.1]{benson2}. 

1253: By Dickson's Theorem, see \cite[Thm.8.1.1]{benson2},

1254: the set $\{c_0,c_1,c_2\}\sseq S[W''^\ast]^G$ is algebraically independent, 

1255: and we have $S[W''^\ast]^G=\F_2[c_0,c_1,c_2]$. 

1256: Hence the above map

1257: $S[W'^\ast]^G/((W'/W'')^\ast S[W'^\ast])^G\ra S[W''^\ast]^G$

1258: indeed is surjective.

1259: 

1260: \abs\bfc

1261: Using the technique described in Remark \ref{primrem}, we may also 

1262: find an optimal set of primary invariants of the invariant ring 

1263: $S[\wti{W'^\ast}]^G$, which by the Hochster-Eagon Theorem, 

1264: see \cite[Thm.4.3.6]{benson2} or Proposition \ref{cmsubgprop}, 

1265: is Cohen-Macaulay. These optimal primary invariants turn out to have 

1266: degrees $\{2,3,3,4,4,7\}$. Indeed, the Hilbert series 

1267: $H_{S[\wti{W'^\ast}]^G}(t)=H_{S[W'^\ast]^G}(t)\in\Q(t)$,

1268: see the proof of Proposition \ref{hilprop}, can be rewritten as

1269: $$ H_{S[\wti{W'^\ast}]^G}(t)=

1270:    \frac{1+2t^5+2t^6+t^7+t^{10}+2t^{11}+2t^{12}+t^{17}}

1271:         {(1-t^2)\cdot (1-t^3)^2\cdot (1-t^4)^2\cdot (1-t^7)}.$$

1272: 

1273: \abs

1274: Since the optimal set $\cF$ of primary invariants 

1275: has degrees $\{2,3,3,4,6,7\}$, 

1276: there is no set of primary invariants of $S[W'^\ast]^G$ 

1277: having degrees $\{2,3,3,4,4,7\}$. This shows, although $S[\wti{W'^\ast}]^G$ 

1278: and $S[W'^\ast]^G$ do have the same Hilbert series, that their ring 

1279: structures are different. 

1280: Still, as the underlying modules $W'^\ast$ and $\wti{W'^\ast}$ are 

1281: closely related, the corresponding invariant rings should be closely 

1282: related as well. But how this relationship might look like, for the time

1283: being remains mysterious to the authors.

1284: 

1285: 

1286: 

1287: 

1288: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1289: \begin{thebibliography}{References}

1290: \setlength{\parskip}{0em}

1291: 

1292: \bibitem{benson} {\sc D. Benson}: 

1293:    Representations and cohomology I, 

1294:    Cambridge Studies in Advanced Mathematics 30,

1295:    Cambridge Univ. Press, 1991. 

1296: 

1297: \bibitem{benson2} {\sc D. Benson}:

1298:    Polynomial invariants of finite groups,

1299:    London Mathematical Society Lecture Note Series 90,

1300:    Cambridge Univ. Press, 1993.

1301: 

1302: \bibitem{Atlas} {\sc J. Conway, R. Curtis, S. Norton, R. Parker, R. Wilson}:

1303:    Atlas of finite groups, maximal subgroups and ordinary characters for

1304:    simple groups, Oxford Univ. Press, 1985.

1305: 

1306: \bibitem{MAGMA} {\sc The Computational Algebra Group}:

1307:    {\sf MAGMA-V2.10} --- The Magma Computational Algebra System,

1308:    School of Mathematics and Statistics, University of Sydney, 2003,

1309:    \url{http://magma.maths.usyd.edu.au/magma/}.

1310: 

1311: \bibitem{derksen} {\sc H. Derksen, G. Kemper}:

1312:    Computational invariant theory,

1313:    Encyclopedia of Mathematical Sciences 130, Springer, 2000.

1314: 

1315: \bibitem{dixmier} {\sc J. Dixmier}:

1316:    On the projective invariants of quartic plane curves,

1317:    Adv. in Math. 64, 1987, 279--304.

1318: 

1319: \bibitem{elkies} {\sc N. Elkies}:

1320:    The Klein quartic in number theory, The eightfold way, 51--101, 

1321:    Math. Sci. Res. Inst. Publ. 35, Cambridge Univ. Press, 1999. 

1322: 

1323: \bibitem{GAP} {\sc The GAP Group}:

1324:    {\sf GAP-4.3} --- Groups, Algorithms and Programming,

1325:    Aachen, St. Andrews, 2003, \url{http://www-gap.dcs.st-and.ac.uk/gap/}.

1326: 

1327: \bibitem{ModAtlas} {\sc C. Jansen, K. Lux, R. Parker, R. Wilson}:

1328:    An atlas of Brauer characters,

1329:    London Mathematical Society Monographs, New Series 11,

1330:    Oxford Univ. Press, 1995.

1331: 

1332: \bibitem{nartritz} {\sc E. Nart, C. Ritzenthaler}: 

1333:    Non-hyperelliptic curves of genus three over finite fields 

1334:    of characteristic two, 2003, 

1335:    \url{http://front.math.ucdavis.edu/math.NT/0312366}.

1336: 

1337: \bibitem{smith} {\sc L. Smith}: 

1338:    Polynomial invariants of finite groups,

1339:    Research Notes in Mathematics 6, Peters, 1995.

1340: 

1341: \end{thebibliography}

1342: 

1343: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1344: \abs

1345: 

1346: \abs

1347: {\sc J. M.: \\

1348: Lehrstuhl D f\"ur Mathematik, RWTH Aachen, \\

1349: Templergraben 64, D-52062 Aachen, Germany}; \\

1350: \url{Juergen.Mueller@math.rwth-aachen.de}.

1351: 

1352: \abs

1353: {\sc

1354: Institut f\"ur Experimentelle Mathematik (IEM), \\

1355: Universit\"at Duisburg-Essen, \\

1356: Ellernstra{\ss}e 29, D-45326 Essen, Germany}.

1357: 

1358: \abs

1359: {\sc C. R.: \\

1360: Institut f\"ur Experimentelle Mathematik (IEM), \\

1361: Universit\"at Duisburg-Essen, \\

1362: Ellernstra{\ss}e 29, D-45326 Essen, Germany}.\\

1363: \url{ritzenth@math.jussieu.fr}.

1364: 

1365: \end{document}

1366:  

1367: