1: \documentclass[11pt]{amsart}

2: 

3: \usepackage{amssymb, amsmath, epsfig, graphics, amscd, ifthen}

4: 

5: \newcommand{\pics}{true}

6: 

7: \setlength{\oddsidemargin}{0cm} \setlength{\evensidemargin}{0cm}\setlength{\textwidth}{16cm} \setlength{\topmargin}{1cm}\setlength{\textheight}{22cm} \setlength{\headheight}{.1in}\setlength{\headsep}{.3in}\setlength{\parskip}{.5mm}

8: 

9: \renewcommand{\theequation}{\thesection.\arabic{equation}}

10: \renewcommand{\thefigure}{\thesection.\arabic{figure}}

11: 

12: \newcommand{\susy}{{$N=1$}}

13: \newtheorem{theorem}{Theorem}[section]

14: \newtheorem{nothing}[theorem]{$\!\!$}

15: \newtheorem{proposition}[theorem]{Proposition}

16: \newtheorem{lemma}[theorem]{Lemma}

17: \newenvironment{pf}{\paragraph{\sc Proof}}{\hspace*{\fill}$\Box$\par\medskip}

18: \newtheorem{construction}[theorem]{Construction}

19: \newtheorem{example}[theorem]{Example}

20: \newtheorem{conjecture}[theorem]{Conjecture}

21: \newtheorem{remark}[theorem]{Remark}

22: \newtheorem{definition}[theorem]{Definition}

23: 

24: \newcommand{\sortout}[1]{{\bf [#1]}}

25: \newcommand{\tofollow}{[Details to follow.]}

26: 

27: \newcommand{\spec}{\mathop{\rm Spec\,}\nolimits}

28: \newcommand{\coker}{\mathop{\rm coker}\nolimits}

29: \newcommand{\coh}{\mathop{\rm Coh\,}\nolimits}

30: \renewcommand{\mod}{\mathop{\rm Mod\,}\nolimits}

31: \newcommand{\proj}{\mathop{\rm Proj\,}\nolimits}

32: \newcommand{\pic}{\mathop{\rm Pic}\nolimits}

33: \newcommand{\Coh}{\mathop{\rm Coh}\nolimits}

34: \newcommand{\supp}{\mathop{\rm Supp}\nolimits}

35: \renewcommand{\phi}{\varphi}

36: \newcommand{\df}{\mathop{\rm Def\,}\nolimits}

37: \newcommand{\ext}{\mathop{\rm ext}\nolimits}

38: \newcommand{\Hom}{\mathop{\rm Hom}\nolimits}

39: \newcommand{\sHom}{\mathop{{\mathcal{H}om}}\nolimits}

40: \newcommand{\Ext}{\mathop{\rm Ext}\nolimits}

41: \newcommand{\shExt}{\mathop{{\mathcal E}\hspace{-1pt}xt}\nolimits}

42: \newcommand{\shHom}{\mathop{Hom}\nolimits}

43: \newcommand{\im}{\mathop{\rm Im}\nolimits}

44: \newcommand{\rk}{\mathop{\rm rk}\nolimits}

45: \newcommand{\dR}[1]{{\bf R}#1_*}

46: \newcommand{\shRhom}{\mathop{{\bf R}Hom}\nolimits}

47: \newcommand{\Db}[1]{D^b(#1)}

48: \newcommand{\ch}{{\rm ch}}

49: \newcommand{\td}{{\rm td}}

50: \newcommand{\SL}{{\rm SL}}

51: \newcommand{\GL}{{\rm GL}}

52: \newcommand{\SU}{{\rm SU}}

53: \newcommand{\PGL}{{\rm PGL}}

54: \newcommand{\Gm}{{\rm G_m}}

55: 

56: \renewcommand{\Box}{\square}

57: \renewcommand{\tilde}{\widetilde}

58: 

59: \newcommand{\al}{\alpha}

60: \newcommand{\be}{\beta}

61: \newcommand{\la}{\lambda}

62: \newcommand{\del}{\delta}

63: \newcommand{\ep}{\epsilon}

64: \newcommand{\F}{\mathcal F}

65: \renewcommand{\S}{\Sigma}

66: \newcommand{\M}{\mathcal M}

67: \newcommand{\B}{\mathcal B}

68: \newcommand{\D}{\mathcal D}

69: \newcommand{\N}{\mathcal N}

70: \newcommand{\I}{\mathcal I}

71: \newcommand{\J}{\mathcal J}

72: \newcommand{\A}{{\bf A}}

73: \renewcommand{\AA}{\mathbb A}

74: \newcommand{\sA}{\mathcal A}

75: \newcommand{\sP}{\mathcal P}

76: \newcommand{\X}{\mathcal X}

77: \newcommand{\T}{\mathcal T}

78: \newcommand{\Y}{\mathcal Y}

79: \newcommand{\E}{\mathcal E}

80: \newcommand{\R}{\mathbb R}

81: \newcommand{\U}{\mathcal U}

82: \newcommand{\V}{\mathcal V}

83: \newcommand{\W}{\mathcal W}

84: \newcommand{\C}{\mathbb C}

85: \newcommand{\sC}{\mathcal C}

86: %\newcommand{\C}{k}

87: \newcommand{\Z}{\mathbb Z}

88: \renewcommand{\O}{\mathcal O}

89: \renewcommand{\L}{\mathcal L}

90: \newcommand{\PP}{\mathbb P}

91: \renewcommand{\P}{{\bf P}}

92: \newcommand{\Q}{\mathbb Q}

93: \newcommand{\sQ}{\mathcal Q}

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

95: \renewcommand{\sl}{{\mathfrak sl}}

96: 

97: \newcommand{\sH}{{\mathcal H}} % \lH-bundle

98: \newcommand{\sh}{{\mathcal H}_0} % \lh-bundle

99: \newcommand{\lH}{{\mathfrak h}} % affine Cartan

100: \newcommand{\lh}{{\mathfrak h}_0} % finite Cartan

101: \newcommand{\G}{{\mathfrak g}} % affine Lie

102: \newcommand{\g}{{\mathfrak g}_0} %finite Lie

103: \newcommand{\De}{\Delta} % affine simple roots

104: \newcommand{\de}{\Delta_0} % finite simple root

105: \newcommand{\ro}{R} % set of finite roots

106: \newcommand{\roo}{\eta} % a root

107: \newcommand{\posro}{R_+} % set of finite positive roots

108: \newcommand{\posroa}{R_+^a} % finite positive roots containing a given simple root

109: \newcommand{\spvec}{{\bf v_0}} % special vector in a rep

110: 

111: \newcommand{\defpar}{s}

112: \newcommand{\dpar}{s}

113: \newcommand{\defpoly}{\Theta}

114: \newcommand{\defspace}{{\mathcal S}}

115: 

116: %\newcommand{\z}{{\mathfrak z}}

117: \newcommand{\z}{{Z}}

118: 

119: \newcommand{\e}{_{\text{\'et}}}

120: 

121: \newcommand{\color}[6]{}

122: 

123: \begin{document}

124: \thispagestyle{empty}

125: \begin{center}

126: {\LARGE Sheaves on fibered threefolds and quiver sheaves} 

127: 

128: \vspace{0.2in}

129: 

130: {\large Bal\'azs Szendr\H oi} 

131: 

132: \vspace{0.15in}

133: 

134: {\large August 2007}

135: 

136: \end{center}

137: 

138: \vspace{0.1in}

139: 

140: {\small

141: \begin{center} {\sc Abstract} \end{center}

142: {\leftskip=50pt \rightskip=50pt

143: \noindent This paper classifies a class of holomorphic $D$-branes, closely related 

144: to framed torsion-free sheaves, on threefolds fibered in resolved $ADE$ surfaces 

145: over a general curve~$C$, in terms of representations with relations of a twisted 

146: Kronheimer--Nakajima-type quiver in the category $\Coh(C)$ of coherent sheaves 

147: on~$C$. For the local Calabi--Yau case $C\cong\A^1$ and special choice of 

148: framing, one recovers the \susy\ $ADE$ quiver studied by Cachazo--Katz--Vafa. 

149: \par}}

150: 

151: \section*{Introduction}

152: 

153: The purpose of this paper is to study, via dimensional reduction, certain 

154: holomorphic $D$-branes, closely related to torsion-free sheaves, on 

155: threefolds $X\to C$ fibered in resolved $ADE$ surfaces over a curve. 

156: Fibered local Calabi--Yau threefolds $X\rightarrow \AA^1$ of this type, 

157: as well as their deformations $X_\defpar\rightarrow \AA^1$ and extremal transitions,

158: were thoroughly analized in~\cite{ckv, cifkv} from the point of view of 

159: supersymmetric gauge theory. The paper~\cite{ckv} contains an assertion, made 

160: explicit in~\cite{katz} and studied in~\cite{zhu}, that exceptional components

161: of a natural threefold contraction $X_\defpar\to \bar X_\defpar$ are classified 

162: by irreducible representations of a certain quiver with loop edges, 

163: the \susy\ $ADE$ quiver (see Figure~\ref{fig!quiv2} for an example), 

164: satisfying a specific set of relations. This statement is in the spirit of 

165: Gabriel's theorem classifying exceptional (not necessarily irreducible) 

166: rational curves in resolved $ADE$ surfaces in terms of irreducible

167: representations of the corresponding Dynkin quiver.

168: 

169: In this paper we generalize the work of~\cite{ckv, katz, zhu} 

170: in two directions: we consider holomorphic $D$-branes, 

171: objects in the derived category of coherent sheaves,

172: instead of exceptional components, and we study the semi-local case: 

173: the neighbourhood of a deformed $ADE$ fibration $X_\defpar\rightarrow C$ over 

174: a general curve~$C$. The main result is Theorem~\ref{thm!class}, which shows that 

175: certain holomorphic $D$-branes on the fibered threefold $X_\defpar$ are classified 

176: by representations with relations of a Kronheimer--Nakajima-type quiver 

177: in the category $\Coh(C)$ of coherent sheaves on the curve~$C$. In particular, 

178: moduli spaces of such holomorphic $D$-branes are quiver bundle varieties over $C$.

179: If $C\cong\AA^1$, a further dimensional reduction leads to Theorem~\ref{thm!A1}, 

180: relating sheaves on the threefold to the zero-dimensional problem of ordinary 

181: matrix representations of the \susy\ $ADE$ quiver of~\cite{ckv, katz, zhu}.

182: The loops in the \susy\ $ADE$ quiver arise as the action by multiplication 

183: of a parameter $t\in H^0(\O_{\AA^1})$ on spaces of sections of sheaves on 

184: the base~$\AA^1$.

185: 

186: The geometry considered in this paper is non-monodromic, meaning that there

187: is no global~\cite{sz} nor local~\cite{ckv} monodromy in the fibration of

188: $ADE$ surfaces over the curve~$C$. It appears to be an interesting question

189: to extend the results proved here to these more general cases involving 

190: monodromy.

191: 

192: In recent work~\cite{penn}, the moduli space of certain very special 

193: holomorphic $D$-branes on resolved $A_1$-fibered geometries $X\rightarrow C$ 

194: has been connected, via imposing a superpotential and going through a large~$N$ 

195: transition, to the Hitchin system on~$C$. The branes studied in~\cite{penn} are 

196: not of the type classified by our results; they should rather correspond to a 

197: complex of quiver representations. Understanding the precise connection 

198: between~\cite{penn} and the present paper is left for future work.

199: 

200: After introducing basic notation in Section~\ref{sec!surf}, 

201: Section~\ref{sec!defs} describes the threefolds we study, and defines

202: some auxiliary sheaves of non-commutative algebras over the curve~$C$. 

203: Section~\ref{sec!res} contains our results, in particular the

204: general statement Theorem~\ref{thm!class} connecting quiver bundles 

205: to holomorphic $D$-branes on ADE fibrations, as well as the 

206: statement for the affine case. Proofs are discussed in Section~\ref{sec!pfs}. 

207: 

208: \section{Finite groups of type $ADE$ and surfaces}

209: \label{sec!surf}

210: 

211: Let $\Gamma<\SL(2,\C)$ be a finite subgroup of type~$A, D$ or~$E$. 

212: Let~$\lh$ be the Cartan subalgebra of the finite dimensional 

213: Lie algebra of the same type. Fix a set of simple roots $\{\roo_a: a\in\de\}$

214: indexed by nodes of the Dynkin diagram~$\de$, and let~$\posro$ be 

215: the set of positive roots. Let $\lH$ be the corresponding affine 

216: Cartan with simple roots indexed by nodes of the Dynkin diagram $\De\supset \de$. 

217: 

218: The group ring $\C\Gamma$ has center $Z(\C\Gamma)\cong\C^\De$; 

219: explicitly, for $\lambda \in Z(\C\Gamma)$, the isomorphism is obtained by 

220: taking the trace of $\lambda$ on a set of 

221: irreps, indexed by the nodes of $\De$ according to the McKay correspondence.   

222: There is also a natural identification

223: \[\lh=\{\lambda\in\C^\De\, |\, \lambda\cdot\delta=0\}\subset \lH\cong\C^\De,\]

224: where $\delta=(\delta_a)$ 

225: are the dimensions of the irreps of $\Gamma$.

226: 

227: \begin{lemma} The centralizer $C_{\GL(2,\C)}(\Gamma)$ of $\Gamma$ in 

228: ${\GL(2,\C)}$ is 

229: \begin{enumerate}

230: \item the full group $\GL(2,\C)$ for type $A_1$;

231: \item a torus $(\C^\star)^2$ in~$\GL(2,\C)$ for type $A_n$ with $n>1$;

232: \item the center $\C^\star$ of~$\GL(2,\C)$ for types $D$ and $E$.

233: \end{enumerate}

234: \label{lemma!cent}

235: \end{lemma}

236: 

237: Let $\bar Y=\AA^2/\Gamma$ be the singular affine quotient,  

238: $Y\rightarrow \bar Y$ its minimal resolution. Exceptional curves in the 

239: resolution are in one-to-one correspondence with the nodes of $\de$,

240: and thus with a set of simple roots of $\lh$; 

241: the positive roots $\roo\in\posro$ correspond to connected, 

242: possibly reducible exceptional rational curves. The 

243: universal deformations $\Y\rightarrow\lh$ and $\bar\Y\rightarrow\lh/W$ of 

244: $Y$ and $\bar Y$, where $W$ denotes the Weyl group, 

245: are connected by the well known commutative diagram

246: \[\begin{array}{ccccc}

247: \Y & \longrightarrow & p^*\bar\Y & \longrightarrow &\bar\Y \\

248: &\searrow & \downarrow&&\downarrow\\

249: && \lh & \stackrel{p}\longrightarrow & \lh/W.

250: \end{array}

251: \]

252: 

253: \section{Threefolds: definitions} \label{sec!defs} 

254: \subsection{The geometry}

255: Let $C$ be a curve, and let $\sQ$ be a rank-two

256: vector bundle on $C$ whose structure group reduces from $\GL(2,\C)$ to the

257: centralizer $C_{\GL(2,\C)}(\Gamma)$. Thus, by Lemma~\ref{lemma!cent}, 

258: \begin{itemize}

259: \item for type $A_1$, $\sQ$ is an arbitrary rank-two vector bundle; 

260: \item for type $A_n$ with $n>1$, $\sQ\cong\sQ_1\oplus\sQ_2$ 

261: is the direct sum of two line bundles;

262: \item for types $D,E$, $\sQ\cong\sQ_0^{\oplus 2}$ for some line bundle $\sQ_0$.

263: \end{itemize}

264: There is a fiberwise $\Gamma$-action on the total space of the 

265: vector bundle $\sQ$, and the quotient $\bar{X}=\sQ/\Gamma$ is a threefold 

266: with a curve of compound Du Val singularities

267: along the image of the zero section. 

268: Let $f\colon X\rightarrow \bar X$ be the crepant resolution,

269: with a map $\pi\colon X\rightarrow C$ whose fibres are minimal resolutions 

270: of the corresponding surface singularity, with trivial monodromy in the fibres. 

271: The canonical bundle of $X$ is

272: \[\omega_X\cong \pi^*(\omega_C\otimes \det\sQ^\vee).\]

273: In particular,~$X$ is Calabi--Yau if and only if~$\sQ$ has canonical 

274: determinant on~$C$. 

275: 

276: Part of the deformation theory of the threefold~$X$ was described 

277: in~\cite{sz}. Let $\sh=\det\sQ\otimes\lh$, a vector bundle over $C$, and let

278: $\defspace=H^0(C, \sh)$ be its space of sections. Then there is a smooth family

279: of threefolds $\X\rightarrow \defspace$, with injective Kodaira--Spencer 

280: map and central fibre $X_0\cong X$, together with a fibration 

281: $\X\to C\times S$ and a contraction $\X\to\bar\X$ over $S$. 

282: Thus, for every $\defpar\in\defspace$, the threefold fibre $X_\defpar$ possesses 

283: a fibration $\pi_\dpar\colon X_\defpar\rightarrow C$ in surfaces and a contraction 

284: $f_\defpar\colon X_\defpar\rightarrow \bar X_\defpar$ to a singular threefold 

285: with compound Du Val singularities. 

286: More precisely, for every positive root $\roo\in\posro$ of $\lh$, 

287: there is a map $p_\roo\colon \sh\rightarrow \det\sQ$, whose vanishing locus 

288: is a family of root hyperplanes in the $\lh$ fibers, and we have 

289: 

290: \begin{lemma} Let $\defpar\in \defspace=H^0(C,\sh)$ be a section of $\sh$, and 

291: let $\roo\in\posro$ be a positive root of $\lh$. 

292: The contraction $f_\defpar\colon X_\defpar\rightarrow \bar X_\defpar$ 

293: contracts a (possibly reducible) rational curve corresponding to the 

294: root~$\roo$ over a point~$P\in C$, if and only if the projected section

295: $p_\roo(\defpar)\in H^0(C, \det\sQ)$ vanishes at~$P\in C$. 

296: \end{lemma} 

297: \noindent

298: Thus if the projected section $p_\roo(\defpar)$ is not identically zero

299: for any root~$\roo$, 

300: then $f_\defpar$ is a small contraction, contracting rational curves to 

301: isolated singularities in certain configurations. If for different 

302: roots~$\roo$, the sections~$p_\roo(\defpar)$ have different simple zeros, 

303: then $f_\defpar$ contracts a set of isolated $(-1, -1)$-curves

304: to simple nodes. If the linear system~$\det\sQ$ has no base points on $C$,

305: then this holds for generic~$\defpar\in\defspace$.

306: 

307: In the special case~$C\cong\AA^1$, the central fiber~$X_0=\AA^1\times Y$ is 

308: Calabi--Yau, and its deformations are parameterized by an $\lh$-valued 

309: polinomial $\defpar\in\lh[t]$. Under the isomorphism

310: $\lh\cong\{\la\, | \, \defpar\cdot \delta=0\}\subset\C^\De$, we can also

311: parameterize deformations by a set of ordinary polinomials 

312: $\defpoly_a\in \C[t]$ 

313: indexed by nodes of the affine Dynkin diagram $\De$, satisfying 

314: $\sum_a \delta_a \defpoly_a=0$. The exceptional fibres of 

315: $f_\defpar\colon X_\defpar\rightarrow \bar X_\defpar$

316: lie over roots of the various polynomials $\defpoly_{\roo_a}=\defpoly_a$, 

317: corresponding to simple roots $\roo_a$, as well as over roots of their linear 

318: combinations $\defpoly_\roo=\sum_a \mu_a \defpoly_a$, corresponding to other 

319: positive roots $\roo=\sum_a\mu_a\roo_a\in\posro$. 

320: For generic choice of parameter $\defpar\in\defspace$, equivalently for 

321: generic choice of $\{\defpoly_a\}$, 

322: the polynomials $\{\defpoly_\roo: \roo\in\posro\}$ have 

323: distinct simple roots, and the exceptional set of 

324: $f_\defpar\colon X_\defpar\rightarrow \bar X_\defpar$ 

325: consists of isolated $(-1, -1)$-curves.

326: 

327: \subsection{Sheaves of non-commutative algebras and their sheaves of modules} 

328: 

329: Given $(C, \sQ)$, let $\sH=\det\sQ\otimes \lH$, a vector bundle on the 

330: curve~$C$ containing~$\sh$ as a subbundle. 

331: Given a section $\defpar\in H^0(C, \sH)$, consider the natural composition

332: \[ \sigma_\defpar\colon\ \  \sQ^\vee\otimes\sQ^\vee\stackrel{\wedge^2}\longrightarrow \det\sQ^\vee \stackrel{\cdot \defpar}\longrightarrow \lH\otimes \O_C\stackrel{\sim}\longrightarrow Z(\C\Gamma)\otimes\O_C,\]

333: a family of $Z(\C\Gamma)$-valued symplectic forms in the fibres of the vector 

334: bundle $\sQ^\vee$. Also fix, once and for all, a trivializing section 

335: $z\in H^0(\O_C)$. 

336: 

337: \begin{definition}\rm Let $\sA_\defpar$ be the sheaf of non-commutative algebras 

338: on $C$ whose sections on an open set $U\subset C$ are 

339: \[\sA_\defpar(U)= T\sQ^\vee(U)*\C\Gamma \big/\big\langle\!\big\langle[x_1,x_2]+\sigma_\defpar(x_1,x_2)\big\rangle\!\big\rangle,\]

340: where $T\sQ^\vee(U)$ is the full tensor algebra of $\sQ^\vee(U)$, 

341: $x_i\in\sQ^\vee(U)$ are local sections, and 

342: $\big\langle\!\big\langle\ldots\big\rangle\!\big\rangle$ denotes the two-sided

343: ideal generated by all given expressions. Define also

344: \[\sP_\defpar(U)=T(\sQ^\vee\oplus\O_C)(U)*\C\Gamma \big/\big\langle\!\big\langle[x_1,x_2]+\sigma_\defpar(x_1,x_2)z^2, [x_i, z]\big\rangle\!\big\rangle,\]

345: where the fixed section $z\in H^0(\O_C)$ commutes with elements 

346: of~$\C\Gamma$. The sheaf~$\sP_\defpar$ becomes a sheaf of graded algebras  

347: by assigning degree $1$ to local sections~$x_i\in\sQ^\vee(U)$ 

348: as well as to~$z\in H^0(\O_C)$; thus its degree-zero piece is 

349: \[\sP_{\defpar,0}\cong \O_C\otimes\C\Gamma.

350: \]

351: \end{definition}

352: 

353: \begin{remark}\rm The sheaf of algebras~$\sA_\defpar$ is a relavitive version 

354: of the following non-commutative deformation of the skew group algebra, 

355: introduced by Crawley--Boevey and Holland in~\cite{cb-h}, depending on a 

356: deformation parameter $\lambda\in\lH\cong Z(\C\Gamma)$:

357: \[ A_\lambda= \C\langle x_1, x_2\rangle * \Gamma \big/\big\langle

358: \!\big\langle[x_1,x_2]+\lambda\big\rangle \!\big\rangle.

359: \]

360: The graded version is

361: \[ P_\lambda = \C\langle y_0, y_1, y_2\rangle * \Gamma \big/\big\langle \!\big\langle[y_0, y_i], [y_1, y_2]+\lambda y_0^2\big\rangle \!\big\rangle.\]

362: 

363: For~$\Gamma=\{1\}$, $\lambda$ is just a complex number; if $\lambda\neq 0$, 

364: $A_\lambda$ is isomorphic to the first Weyl algebra, whereas~$P_\lambda$ is 

365: a degenerate Sklyanin algebra deforming the algebra of functions on 

366: the commutative projective plane~$\PP^2$. As proved in~\cite{cb-h}, 

367: for general $\Gamma$ and $\lambda\in\lh\subset Z(\C\Gamma)$ the algebra

368: $A_\lambda$ is finite over its center

369: \[ Z A_\lambda \cong \C[\bar Y_\lambda]. 

370: \]

371: The latter is the coordinate ring of the affine variety $\bar Y_\lambda$ 

372: corresponding to the deformation parameter $\lambda\in\lh$, a deformation 

373: of the invariant ring $\C[x_1,x_2]^\Gamma\cong\C[\bar Y]$. For 

374: $\lambda\in \lH\setminus\lh$, $A_\lambda$ is ``genuinely'' 

375: non-commutative.

376: \end{remark}

377: 

378: By abuse of notation, we will refer to $\P_\defpar=\proj_C\sP_\defpar$ as the

379: non-commutative projective bundle corresponding to $\defpar\in \defspace$, with 

380: fibration $\pi_\dpar\colon\P_\defpar\rightarrow C$. 

381: Setting $z=0$, we have its divisor at infinity

382: \[i_\dpar\colon D_\defpar\hookrightarrow\P_\defpar.\] 

383: The divisor $D_\defpar$ has the structure of an ordinary (commutative) 

384: $\PP^1$-bundle

385: \[\pi_\dpar|_{D_\defpar}=\tau_\defpar\colon D_\defpar\rightarrow C\] 

386: equipped with a $\Gamma$-action on the fibres. Its complement 

387: $\A_\defpar=\P_\defpar\setminus D_\defpar = \spec_C \sA_\defpar$ is

388: a non-commutative affine bundle.

389: 

390: The sheaf $\sP_\defpar$ is a sheaf of regular graded algebras in the 

391: sense of~\cite{az}; sheaf theory on~$\P_\defpar$ works in 

392: complete analogy with the absolute case discussed in~\cite{bgk}. 

393: The category of coherent sheaves $\Coh(\P_\defpar)$ is by definition

394: the quotient of the category of sheaves of finitely generated 

395: graded right $\sP_\defpar$-modules by the subcategory of sheaves of 

396: torsion $\sP_\defpar$-modules; we will sometimes refer to objects in this

397: category as $\sP_\defpar$-modules. The trivial module, graded in

398: degree~$n$, defines the object $\O_{\P_\defpar}(n)\in\Coh(\P_\defpar)$; 

399: given a sheaf $\E$, its twists $\E(n)$ are obtained by shifting the grading. 

400: We have $\Ext$ groups as the derived functors of $\Hom$, and also 

401: functors $\shExt^i(-,\O_{\P_\defpar})$; the latter take values in the

402: category of left $\sP_\defpar$-modules (compare~\cite{bgk}). 

403: 

404: Pushforward 

405: \[\pi_{\defpar*}\colon \Coh(\P_\defpar)\rightarrow \Coh^\Gamma(C)\]

406: along the morphism $\pi_\dpar\colon\P_\defpar\rightarrow C$ is defined in 

407: the usual way, as the coherent $\Gamma$-sheaf on $C$ defined by sections over 

408: preimages of open sets of $C$, the section spaces being (right) $\C\Gamma$-modules; 

409: the action of $\Gamma$ on $C$ is taken to be trivial.

410: The higher pushforwards ${\rm R}^p\pi_{\defpar*}(-)$ are the derived functors 

411: of $\pi_{\defpar*}$. Given a $\P_\defpar$-module $\E$, we will 

412: also use the relative $\Hom$-functor

413: \[\Hom_C(\E, -)\colon \Coh(\P_\defpar)\rightarrow \Coh^\Gamma(C)\]

414: defined by homomorphisms on preimages of open sets in $C$, as well as its

415: derived functors 

416: \[\Ext^i_C(\E, -)\colon \Coh(\P_\defpar)\rightarrow \Coh^\Gamma(C).\]

417: We also have a pullback functor 

418: \[\pi_\dpar^*\colon \Coh^\Gamma(C)\rightarrow \Coh(\P_\defpar)\]

419: taking a sheaf of (right) $\C\Gamma$-modules $\F$ to the sheaf 

420: $\F\otimes_{\C\Gamma}\sP_\defpar$ of (right) $\sP_\defpar$-modules.

421: The pair $(\pi_\dpar^*,\pi_{\defpar*})$ forms an adjoint pair as 

422: in the commutative case. Similarly, for the inclusion 

423: $i_\dpar\colon D_\defpar\to\P_\defpar$, we have a pullback

424: (restriction) functor

425: \[i_\dpar^*\colon\coh(\P_\defpar)\rightarrow\coh^\Gamma(D_\defpar),\] 

426: defined by factoring modules of 

427: local sections by the ideal $\langle z\rangle$ (recall that $z$ is central),

428: as well as a pushforward 

429: \[i_{\defpar*}\colon\coh^\Gamma(D_\defpar)\rightarrow\coh(\P_\defpar),\] 

430: with $z$ acting on local sections by zero. 

431: There is also a restriction functor to the finite part $\A_\defpar$, defined

432: by factoring the ideal $\langle z-1\rangle$.

433: 

434: \begin{definition}\rm A $\pi_\dpar$-free sheaf on $\P_\defpar$ is an

435: object~$\E\in\Coh(\P_\defpar)$, which admits an embedding

436: \[  \E\hookrightarrow \pi_\dpar^*(\U)(n)

437: \]

438: for some $\U\in\Coh^\Gamma(C)$ and $n\in\Z$.

439: A framed $\pi_\dpar$-free sheaf $(\E, \phi)$ 

440: on~$(\P_\defpar, D_\defpar)$ is a $\pi_\dpar$-free sheaf~$\E$ on~$\P_\defpar$, 

441: together with a fixed isomorphism

442: \[ \phi\colon i_{\defpar}^*\E \stackrel{\sim}\longrightarrow \tau_\defpar^* \W,\]

443: on the divisor $D_\defpar$ at infinity, for some $\W\in\Coh^\Gamma(C)$.

444: \end{definition}

445: 

446: \begin{remark}\rm If $\pi\colon\P\rightarrow \{*\}$ is a (non-commutative) 

447: projective space over a point, the $\pi$-free sheaves are exactly the torsion free

448: ones (compare~\cite[Section 2]{bgk}). 

449: To see this, note that a $\pi$-free sheaf is certainly torsion free, 

450: since it embeds into a locally free sheaf. Conversely, a torsion free sheaf

451: embeds into some locally free sheaf, which in turn embeds into some~$\O^m_\P(n)$.

452: \end{remark}

453: 

454: \begin{lemma} If $\E$ is $\pi_\dpar$-free, then $L^ji_\dpar^*\E=0$ for $j>0$.

455: \end{lemma}

456: \begin{pf} As in the commutative case, the structure 

457: sheaf $i_{\defpar*}\O_{D_\defpar}$ has a resolution 

458: \[ 0 \to \O_{\P_\defpar}(-1) \stackrel{z}\to \O_{\P_\defpar}\to i_{\defpar*}\O_{D_\defpar}\to 0,\]

459: which implies that $L^ji_\dpar^*\E=0$ for $j>1$ for any $\E\in\Coh(\P_\defpar)$, 

460: and also that $L^1i_\dpar^*$ is left exact. If $\E$ is $\pi_\dpar$-free, 

461: applying the latter to an embedding $\E\hookrightarrow \pi_\dpar^*(\U)(n)$

462: gives the vanishing of $L^1$ also. 

463: \end{pf}

464: 

465: %If $\defpar=0$, then $\sP_\defpar$ is a sheaf of commutative algebras twisted 

466: %by a $\Gamma$-action, and $\P_\defpar\rightarrow C$ is an 

467: %ordinary $\PP^2$-bundle with a $\Gamma$-action on the fibres. 

468: %All sheaf-theoretic notions agree with the usual $\Gamma$-equivariant ones.

469: 

470: \section{Threefolds: the results}\label{sec!res} 

471: 

472: \subsection{Twisted quiver representations and quiver sheaves}

473: \label{sec!quiv}

474: 

475: Recall that, given a quiver with arrows $a\to b$ marked by 

476: objects $O_{ab}\in\sC$ of an abelian tensor category $\sC$, a representation 

477: of the marked quiver in $\sC$ consists of a set of objects $O_a$ of 

478: $\sC$ associated to 

479: nodes, and a set of morphisms $\phi_{ab}\in\Hom_\sC(O_a\otimes O_{ab}, O_b)$ 

480: associated to the arrows $a\to b$. Representations of a marked quiver in 

481: the category~$\Coh(X)$ of an algebraic variety~$X$ are also called

482: {\em quiver sheaves}~\cite{gk} on~$X$. 

483: 

484: In the specific context of classifying holomorphic $D$-branes 

485: on the threefold~$X$ and its 

486: deformations, the following quiver marked in~$\Coh(C)$ will arise naturally. 

487: The quiver is the standard extended McKay quiver of~\cite{nak}, 

488: obtained from the original 

489: one by adding an extra leaf at each node with arrows in both directions.

490: Using the data of the vector bundle $\sQ$ on $C$, we mark this quiver 

491: in $\Coh(C)$ as follows.

492: \begin{itemize}

493: \item The marked $A_n$ quiver for $n>1$ is illustrated on 

494: Figure~\ref{fig!quiv1}; recall that in this case, there is a decomposition 

495: $\sQ=\sQ_1\oplus \sQ_2$ into a sum of line bundles, 

496: since the structure group of $\sQ$ reduces to the diagonal torus. 

497: \item The marked $A_1$ quiver consists of only two nodes $0$ and $1$ 

498: and two arrows $0\to 1, 1\to 0$ marked by the rank-two bundle $\sQ^\vee$, as well 

499: as leaves marked as in the higher $A_n$ case.  

500: \item For types $D$ and $E$, arrows between nodes are all marked by the line 

501: bundle $\sQ_0^\vee$, where $\sQ=\sQ_0^{\oplus 2}$; leaves are marked as before. 

502: \end{itemize}

503: 

504: \begin{figure}[ht]

505: \centering

506: \ifthenelse{\equal{\pics}{true}}

507: {\input{quiver1.pstex_t}}{}

508: \caption{The marked extended McKay quiver for $A_2$} 

509: \label{fig!quiv1}

510: \end{figure}

511: 

512: \subsection{The main classification result}

513: 

514: \begin{theorem} Given $\defpar\in H^0(C, \sH)$, there is a 

515: 1-to-1 correspondence between the following sets of data.

516: \label{thm!class}

517: \begin{enumerate} \renewcommand{\labelenumi}{\rm(\arabic{enumi})}

518: \item Isomorphism classes of framed $\pi_\dpar$-free sheaves $(\E, \phi)$

519: on $(\P_\defpar, D_\defpar)$.

520: \item Quintuples $(\V, \W, \B, \I, \J)$, where $\W, \V$ are coherent 

521: $\Gamma$-sheaves on $C$, and 

522: \[\begin{array}{rcl}

523: \B& \in&\Hom^\Gamma_C(\V\otimes \sQ^\vee, \V), \\ \\

524: \I &\in &\Hom^\Gamma_C(\W, \V),\\ \\ 

525: \J &\in&\Hom^\Gamma_C(\V\otimes \det\sQ^\vee, \W),\end{array}\] 

526: satisfying the following two conditions: 

527: \begin{enumerate}

528: \item the ADHM relation 

529: \[  \B\wedge \B + \I\circ\J + \defpar =0 \in \Hom^\Gamma_C(\V\otimes \det\sQ^\vee, \V),

530: \] 

531: where

532: \[H^0(C, Z(\C\Gamma)\otimes\det\sQ)\hookrightarrow \Hom^\Gamma_C(\V\otimes \det\sQ^\vee, \V)\]

533: is the natural embedding as the central subspace;

534: \item non-degeneracy: if $\V'\subset\V$ is a $\Gamma$-subsheaf such that 

535: $\B(\V'\otimes\sQ^\vee)\subset \V'$ and $\I\W\subset\V'$, then~$\V'=\V$. 

536: \end{enumerate} 

537: Sets of quintuples 

538: are identified under the action of invertible elements of $\Hom^\Gamma_C(\V,\V)$. 

539: \item Representations 

540: $(\{\V_a\}, \{\W_a\}, \{\B_{ab}\}, \{\I_a\}, \{\J_a\})$ 

541: in $\Coh(C)$ of the marked McKay-type qui\-ver introduced in~\ref{sec!quiv}, 

542: satisfying 

543: \begin{enumerate}

544: \item the ADHM relations

545: \[ \sum_b \epsilon_{ab}\B_{ba}\circ \B_{ab} + \I_a\circ \J_a + \defpar_a =0 \in \Hom_C(\V_a\otimes \det\sQ^\vee, \V_a)

546: \]

547: at each node~$a$, where $\epsilon_{ab}\in\{\pm 1\}$ is a standard assingment of 

548: signs to arrows with $\epsilon_{ab}=-\epsilon_{ba}$, and 

549: $\defpar_a=P_{\roo_a}(\defpar)$ is the projected section corresponding to 

550: the simple root $\roo_a$, and 

551: \item non-degeneracy: if $\{\V_a'\}$ is a $\B$-invariant set of subsheaves 

552: containing the images of $\I_a$'s, then $\V_a'=\V_a$ at all nodes. 

553: \end{enumerate}

554: Two representations

555: are identified under invertible elements of $\prod_a\Hom_C(\V_a,\V_a)$.

556: \end{enumerate}

557: 

558: \noindent If $\defpar\in\defspace=H^0(C, \sh)$ is a deformation parameter of

559: the threefold $X=X_0$, then the same data also parametrizes

560: \begin{enumerate} \renewcommand{\labelenumi}{\rm(\arabic{enumi})}

561: \addtocounter{enumi}{3}

562: \item certain objects in $\D(\Coh X_\defpar)$, the derived category 

563: of coherent sheaves on $X_\defpar$. 

564: \end{enumerate}

565: \end{theorem} 

566: \begin{pf}

567: The equivalence ${\rm (1)}\iff {\rm (2)}$ follows from a version of 

568: the relative Beilinson resolution for the non-commutative 

569: projective bundle $\P_\defpar\rightarrow C$; details are given in 

570: Section~\ref{sec!koszul}. 

571: McKay's definition of the quiver describing the representation theory 

572: of $\Gamma$ implies ${\rm (2)}\iff {\rm (3)}$ in the standard way.

573: Finally the mapping ${\rm (1)}\implies {\rm (4)}$ in the geometric 

574: case~$\defpar\in\defspace=H^0(C, \sh)$ is given by a derived 

575: equivalence to be discussed in Section~\ref{sec!derived}.

576: \end{pf}

577: 

578: \begin{remark}\rm As $X=X_0$ and its deformations $X_\defpar$ for 

579: $\defpar\in \defspace$ are not projective, one needs to rigidify 

580: before holomorphic $D$-branes, in other words objects in $\D^b(X_\defpar)$, 

581: have a sensible moduli space. 

582: For the central fibre $X=X_0$, a crepant resolution of the singular threefold

583: $\sQ/\Gamma$, one has a derived equivalence~\cite{bkr}

584: \[ \D(X_0)\cong \D^\Gamma(\sQ)

585: \]

586: between the derived categories of coherent sheaves on~$X_0$ and that of

587: $\Gamma$-equivariant sheaves on the total space of the bundle~$\sQ\to C$. 

588: One can easily rigidify

589: on the latter by considering~$\Gamma$-sheaves on the 

590: projective bundle $\P_0=\PP(\sQ\oplus\O_C)\rightarrow C$, framed on the divisor 

591: at infinity $D_0 =\PP(\sQ)\hookrightarrow\P_0$.

592: Theorem~\ref{thm!class} is the appropriate generalization of this approach 

593: which also works for deformations: for the analogous problem on $X_\defpar$, we 

594: consider framed sheaves on the non-commutative projective 

595: bundle~$\P_\defpar\rightarrow C$.

596: 

597: In the surface case, this approach was used earlier in~\cite{bgk}. 

598: To quote the result, let $\lambda\in Z(\C\Gamma)$. 

599: Then for $\Gamma$-modules $V,W$, Nakajima's non-singular quiver 

600: variety $\M_{V, W, \lambda}$ parametrizes torsion free

601: sheaves on the non-commutative space $\PP^2_\lambda=\proj P_\lambda$, 

602: framed on the commutative $\Gamma$-line at $\infty$. This statement 

603: generalizes earlier work of \cite{don, kn, nak1, n-s, kko}

604: and others. The origin of all such results is of course the ADHM 

605: classification~\cite{adhm} of finite-action $\SU(\dim(W))$-instantons 

606: on~$\R^4$ of charge $\dim(V)$.

607: \end{remark}

608: 

609: \subsection{Some holomorphic $D$-branes on $ADE$ fibrations over $\AA^1$}

610: 

611: If $C\cong\AA^1$, Theorem~\ref{thm!class} can in some cases be re-written in 

612: terms of classical quiver representations: representations of a quiver in 

613: vector spaces. This will give an interpretation of an assertion

614: of~\cite{ckv, katz, zhu}.

615: 

616: Recall that for $C\cong\AA^1$, a deformation parameter $\defpar\in\defspace$ 

617: of the central fibre $X_0=\AA^1\times Y$ can be specified by a 

618: set of polynomials $\{\defpoly_a\in \C[t] : a\in\De\}$ indexed by the vertices of 

619: the affine quiver, subject to $\sum_a \delta_a\cdot \defpoly_a = 0$. The 

620: following definition is due to Cachazo--Katz--Vafa~\cite{ckv, katz}. 

621: 

622: \begin{definition}\rm The {affine \susy\ $ADE$ quiver} is the McKay quiver 

623: extended by a loop $a\to a$ at each vertex. For a (finite-dimensional)

624: representation $(\{V_a\}, \{B_{ab}\}, \{\Psi_a\})$ of this quiver, 

625: the ADHM-type relations are 

626: \begin{equation}

627: \label{rel1}

628:  \sum_{b} \epsilon_{ab} B_{ba} B_{ab} + \defpoly_a(\Psi_a)=0\in \Hom(V_a, V_a)

629: \end{equation}

630: at each vertex $a\in\De$ of the quiver, where $\defpoly_a(\Psi_a)$ is to be 

631: interpreted as the evaluation of a polynomial on an endomorphism of $V_a$, 

632: as well as

633: \begin{equation}

634: \label{rel2}

635: \Psi_a B_{ba} = B_{ba}\Psi_b\in\Hom(V_a, V_b)

636: \end{equation}

637: along each arrow $a\to b$ of the quiver~$\De$.

638: \end{definition}

639: 

640: \begin{figure}[ht]

641: \centering

642: \ifthenelse{\equal{\pics}{true}}{\input{quiver2.pstex_t}}{}

643: \caption{A representation of the affine \susy\ $A_2$ quiver} 

644: \label{fig!quiv2}

645: \end{figure}

646: 

647: Consider quadruples $(\{V_a\}, \{B_{ab}\}, \{\Psi_a\}, \spvec)$, 

648: where $(\{V_a\}, \{B_{ab}\}, \{\Psi_a\})$ is a representation of the 

649: affine \susy\ $ADE$ quiver satisfying the ADHM-type relations, 

650: and $\spvec\in V_0$ is a fixed vector in the vector space attached to the 

651: affine node. Call a quadruple non-degenerate if

652: there is no $(B, \Psi)$-invariant collection of subspaces $\{V_a'\subset V_a\}$

653: with~$\spvec\in V_0'$. 

654: 

655: \begin{theorem} Equivalence classes of non-degenerate quadruples 

656: $(\{V_a\}, \{B_{ab}\}, \{\Psi_a\}, \spvec)$ satisfying the ADHM relations, 

657: identified under the action of $\prod_a \GL(V_a)$, parametrize certain objects 

658: in $\D(\coh X_\defpar)$, holomorphic $D$-branes on the threefold~$X_\defpar$. 

659: \label{thm!A1}

660: \end{theorem}

661: 

662: \begin{pf} Quiver sheaf data on $C$ parametrize certain branes on~$X_\defpar$ 

663: by Theorem~\ref{thm!class}. The correspondence between representations 

664: of the \susy\ $ADE$ quiver and a special class of quiver sheaf data will 

665: be discussed in Section~\ref{sec!affpf}. 

666: \end{pf}

667: 

668: \begin{remark}\rm As explained in~\cite{ckv}, the quiver 

669: relations~(\ref{rel1})-(\ref{rel2}) come from the natural superpotential 

670: of the quiver gauge theory on $\De$, involving adjoint 

671: fields $\Psi_a$ as well as bifundamental fields $B_{ab}$. 

672: \end{remark}

673: 

674: \begin{remark}\rm Let the finite \susy\ $ADE$ quiver be obtained from the affine 

675: one by deleting the affine node. Representations of the finite \susy\ $ADE$ quiver, 

676: satisfying the ADHM-type relations~(\ref{rel1})-(\ref{rel2}), 

677: parametrize holomorphic $D$-branes supported on exceptional fibres of

678: $f_\dpar\colon X_\defpar\rightarrow \bar X_\defpar$. This follows from the

679: statement that the vanishing of the affine component of $\V$ forces 

680: all other $\V_a$ be supported on points $P\in C$ at which some projected section

681: $p_\roo(\defpar)$ vanishes for some positive root $\roo\in\posro$, 

682: in other words on points of the base curve over which 

683: the surface fiber $\pi_\dpar^{-1}(P)$ contains exceptional curves.

684: Observing that the section $\defpar\in H^0(C, \Z(\C\Gamma)\otimes\det\sQ)$ 

685: is central in $\Hom^\Gamma_C(\V\otimes \det\sQ^\vee, \V)$, 

686: so commutes with all components of $\B$, the latter statement is essentially 

687: proved in~\cite[4.1--4.2]{ckv}.

688: This establishes a direct link to~\cite{katz, zhu}, according to 

689: which (in the generic case) irreducible representations of the finite \susy\ quiver

690: with the given relations parametrize exceptional components of 

691: the contraction $f_\dpar\colon X_\defpar\rightarrow \bar X_\defpar$.

692: \end{remark}

693: 

694: \section{Proofs}\label{sec!pfs} 

695: \subsection{The Beilinson argument} 

696: \label{sec!koszul}

697: 

698: The aim of this section is to prove of the equivalence 

699: ${\rm (1)}\iff {\rm (2)}$ of the classification result Theorem~\ref{thm!class} 

700: via an analysis of framed $\pi_\dpar$-free sheaves on $\P_\defpar$. 

701: 

702: Given $\defpar\in H^0(C, \sH)$, recall the sheaf of algebras~$\sP_\defpar$ 

703: on the curve~$C$, and the associated non-commutative bundle

704: $\pi_\dpar\colon\P_\defpar\rightarrow C$. 

705: Define $\sP_\defpar$-modules~$\T_i$ by 

706: %middle one is a generalized Euler sequence

707: %Here $\T_1$ is the analogue of the relative tangent bundle, 

708: %Here $\T_1$ is the analogue of the relative tangent bundle, 

709: %so it's $\sQ$ rather than its dual; $\T_2$ would be its determinant 

710: %$\wedge^2 \T_1$ if it had one.

711: \begin{equation}

712: \begin{array}{c} 

713: \T_0 = \O_{\P_\defpar},\\ \\

714: 0 \longrightarrow \O_{\P_{\defpar}} \longrightarrow \pi_\dpar^*(\sQ\oplus\O_C)(1) \longrightarrow \T_1 \longrightarrow 0, \\ \\

715: \T_2 =  \pi_\dpar^* (\det\sQ)(3).

716: \label{seq!defN1}

717: \end{array}

718: \end{equation}

719: 

720: \begin{proposition} 

721: A $\pi_\dpar$-free sheaf $\E$ on $\P_\defpar$, framed on the divisor $D_\defpar$, 

722: is the cohomology of a monad

723: \[ \pi_\dpar^*\Ext^1_C\left(\T_2(-1), \E\right)(-1)\rightarrow \pi_\dpar^*\Ext^1_C\left(\T_1, \E\right)\rightarrow\pi_\dpar^*\Ext^1_C\left(\T_0(1), \E\right)(1)

724: \]

725:  of $\sP_\defpar$-modules.

726: \label{prop!Emon}

727: \end{proposition}

728: \begin{pf} 

729: %The individual terms in the monad form sheaves of (right) 

730: %$\P_\defpar$-modules, as by definition~(\ref{seq!defN1})

731: %there is a ${\sP_{\defpar,0}}$-bimodule

732: %structure on $\T_i$ and hence their twists. 

733: %I don't think we need this.

734: Given a $\sP_\defpar$-module $\F$, a Koszul duality argument, in an analogous way 

735: to the absolute case in~\cite[Section 7]{bgk} following~\cite[Thm 2.6.1]{bgs}, 

736: leads to a Beilinson-type spectral sequence with $E_1$ term

737: \[ E_1^{p,q} = \pi_\dpar^*\Ext^q_C\left(\T_{-p}(p), \F\right)(p),

738: \] 

739: nonzero only for $-2\leq p\leq 0, \ 0\leq q \leq 2$, converging to~$\F$ 

740: in the limit. The vanishing results 

741: \[ \Ext^q_C\left(\T_{-p}(p), \E(-1)\right) =0 \mbox{ for } q=0,2,\ p=-1, -2

742: \]

743: which follow from the existence of the framing of $\E$ on the divisor $D_\defpar$

744: (compare~\cite[Lemma 6.2]{kko},~\cite[Lemma 4.2.12]{bgk}),

745: reduce the spectral sequence for $\F=\E(-1)$ to the monad given in the statement. 

746: Details are left to the reader.

747: \end{pf}

748: 

749: We also record an auxiliary result. 

750: 

751: \begin{lemma}

752: \label{lem!natiso}

753: There are natural isomorphisms

754: \[\Hom_{\P_\defpar}\!\!\left(\pi_\dpar^*\det\sQ^\vee, \pi_\dpar^*\sQ^\vee(1)\right) \cong \Hom_{\P_\defpar}\!\!\left(\pi_\dpar^*\sQ^\vee, \O_{\P_\defpar}(1)\right) \cong \Hom^\Gamma_C\!\left(\sQ^\vee,\sQ^\vee\oplus \O_C\right).

755: \]

756: \end{lemma}

757: 

758: \begin{pf} The first isomorphism follows from Lemma~\ref{lem!rk2} below. The 

759: second one follows from adjunction for the pair~$(\pi_\dpar^*,\pi_{\defpar*})$, 

760: together with 

761: \[\pi_{\defpar*}\O_{\P_\defpar}(1)\cong \sP_{\defpar,1}\cong (\sQ^\vee\oplus \O_C)*\C\Gamma\in\Coh^\Gamma(C),\]

762: an identity well known from the commutative context. 

763: \end{pf}

764: 

765: \begin{lemma} Let $\sQ$ be a rank-two bundle on a (commutative) space. 

766: Then there is a natural isomorphism

767: \[ \sQ\otimes\det\sQ^\vee \cong \sQ^\vee.

768: \] 

769: \label{lem!rk2}

770: \end{lemma}

771: \begin{pf} The embedding 

772: $\iota\colon\det\sQ^\vee\rightarrow \sQ^\vee\otimes\sQ^\vee$ 

773: induces a natural map

774: \[\Hom\!\left(\det\sQ^\vee, \det\sQ^\vee\right)\rightarrow\Hom\!\left(\det\sQ^\vee, (\sQ^\vee)^{\otimes 2}\right)\cong \Hom\!\left(\sQ\otimes\det\sQ^\vee, \sQ^\vee\right). 

775: \]

776: The image of the identity of the first $\Hom$-group gives a natural morphism 

777: as in the statement, which can be checked on a local basis to 

778: be an isomorphism.

779: \end{pf}

780: 

781: Now return to the context of the classification result Theorem~\ref{thm!class}, 

782: and consider a quintuple $(\V, \W, \B, \I, \J)$ as in Theorem~\ref{thm!class}(2); 

783: recall that \[\W, \V\in\Coh^\Gamma(C),\] and 

784: \[\begin{array}{rcl}

785: \B& \in&\Hom^\Gamma_C(\V\otimes \sQ^\vee, \V), \\ \\

786: \I &\in &\Hom^\Gamma_C(\W, \V),\\ \\ 

787: \J &\in&\Hom^\Gamma_C(\V\otimes \det\sQ^\vee, \W).\end{array}\] 

788: Let

789: \[ c\in\Hom_{\P_\defpar}\!\!\left(\pi_\dpar^*\det\sQ^\vee, \pi_\dpar^*\sQ^\vee(1)\right), \ \ \ d\in \Hom_{\P_\defpar}\!\!\left(\pi_\dpar^*\sQ^\vee, \O_{\P_\defpar}(1)\right)

790: \]

791: denote the images, under the isomorphisms of~Lemma~\ref{lem!natiso}, of the 

792: canonical element 

793: \[{\rm Id}\in\Hom^\Gamma_C( \sQ^\vee,\sQ^\vee)\subset\Hom^\Gamma_C\!\left(\sQ^\vee,\sQ^\vee\oplus \O_C\right)\]

794: Note also that we have a fixed section 

795: \[z\in\Hom_{\P_\defpar}(\O_{\P_\defpar}, \O_{\P_\defpar}(1)).\] 

796: Define 

797: \[

798: a = \left(\begin{array}{c} \pi_\dpar^*\left(\B\circ({\rm Id}_\V\otimes\iota)\right)\otimes z - \pi_\dpar^*({\rm Id_\V})\otimes c\,(-1) \\ \pi_\dpar^*(\J)\otimes z \end{array} \right) \colon \pi_\dpar^*(\V\otimes\det\sQ^\vee)(-1) \longrightarrow \pi_\dpar^*(\V\otimes\sQ^\vee\oplus \W)

799: \]

800: where $\iota\colon \det\sQ^\vee\to(\sQ^\vee)^{\otimes 2}$ is the natural map.

801: Define similarly

802: \[

803: b= \left(\begin{array}{cc}\pi_\dpar^*(\B)\otimes z + \pi_\dpar^*({\rm Id_\V})\otimes d & \pi_\dpar^*(\I)\otimes z \end{array}\right) :  \pi_\dpar^*(\V\otimes\sQ^\vee\oplus \W)\to \pi_\dpar^*(\V)(1),

804: \]

805: to obtain the chain of morphisms

806: \begin{equation}

807: \pi_\dpar^*(\V\otimes\det\sQ^\vee)(-1)\stackrel{a}\longrightarrow \pi_\dpar^*(\V\otimes\sQ^\vee\oplus \W)\stackrel{b}\longrightarrow \pi_\dpar^*(\V)(1).

808: \label{seq!monad}

809: \end{equation}

810: 

811: The following result completes the proof of the equivalence 

812: ${\rm (1)}\iff {\rm (2)}$ of the classification result Theorem~\ref{thm!class}.

813: 

814: \begin{proposition} If the quintuple satisfies the ADHM relation, then

815: {\rm (\ref{seq!monad})}~is a complex of $\sP_\defpar$-modules.

816: Furthermore, it is a monad defining a framed $\pi_\dpar$-free sheaf $\E$ if

817: and only if the quintuple  $(\V, \W, \B, \I, \J)$ is non-degenerate.

818: Conversely, every $\pi_\dpar$-free $\sP_\defpar$-module~$\E$, framed 

819: on~$D_\defpar$, arises from this construction.

820: \end{proposition}

821: \begin{pf} The standard direct computation shows that $b\circ a=0$ is equivalent

822: to the ADHM relation. The proof of the equivalence of the monad property and 

823: non-degeneracy is analogous to the absolute case~\cite[Section 4.1]{bgk}. 

824: For the converse, given a framed sheaf $(\E, \phi)$, 

825: let $\V=\Ext^1_C(\O_{\P_\defpar}(1), \E)$. Then by Proposition~\ref{prop!Emon}, 

826: $\E$ is the middle cohomology of the monad

827: \[ \pi_\dpar^*(\V\otimes\det\sQ^\vee)(-1)\rightarrow \pi_\dpar^*\Ext^1_C\left(\T_1, \E\right)\rightarrow\pi_\dpar^*\V(1).

828: \]

829: The usual arguments~\cite[Theorem 6.7]{kko} show that, since $\E$ is framed

830: on~$D_\defpar$, this monad is isomorphic to a monad of the form~(\ref{seq!monad}) 

831: for some quintuple $(\V, \W, \B, \I, \J)$.

832: \end{pf}

833: 

834: \subsection{A derived equivalence}\label{sec!derived}

835: 

836: In this section we complete the proof of Theorem~\ref{thm!class} by establishing

837: the missing link ${\rm (1)}\implies {\rm (4)}$. 

838: 

839: \begin{proposition} Let $\defpar\in \defspace$ be a deformation 

840: parameter of the central fibre $X=X_0$.

841: There is a distinguished equivalence of triangulated categories 

842: \[ \D(\coh X_\defpar) \cong \D(\mod\sA_\defpar ),

843: \] 

844: where $\mod\sA_\defpar$ is the category of sheaves of 

845: finitely generated right $\sA_\defpar$-modules, and $\D(-)$ denotes the bounded 

846: derived category on both sides.

847: \end{proposition}

848: \begin{pf} This assertion is a fibered version of the analogous two-dimensional 

849: equivalence proved in~\cite{gs}, and the proof carries over verbatim.

850: A deformation argument starting from 

851: the central fibre $X=X_0$ shows that a certain specific component $M_\defpar$ 

852: of a fine moduli space of torsion sheaves on $\A_\defpar$ maps by a 

853: semi-small birational map to the singular variety $\bar X_\defpar$. 

854: By~\cite{vdb}, generalizing an argument of~\cite{bkr},

855: this implies that $M_\defpar$ is a crepant resolution of $\bar X_\defpar$, 

856: and one has a derived equivalence

857: \[ \D(\coh M_\defpar) \cong \D(\mod\sA_\defpar)

858: \]

859: defined by the universal sheaf.  

860: But since $X_\defpar$ is the unique crepant resolution of $\bar X_\defpar$, 

861: necessarily $M_\defpar\cong X_\defpar$ and the proposition follows. 

862: Details are left to the reader.

863: \end{pf}

864: 

865: This equivalence gives the mapping ${\rm (1)}\implies {\rm (4)}$ 

866: of Theorem~\ref{thm!class} from framed $\pi_\dpar$-free sheaves on $\P_\defpar$

867: to objects in $\D(\coh X_\defpar)$. Indeed, a right $\sP_\defpar$-module 

868: can be restricted to the affine part $\A_\defpar$ to give a right

869: $\sA_\defpar$-module, and then mapped using the derived equivalence to an object

870: in $\D(\coh X_\defpar)$, in other words a holomorphic $D$-brane on $X_s$.

871: 

872: \subsection{Fibrations over the affine line}

873: \label{sec!affpf}

874: 

875: In this section, we take a fibration $X_\defpar\rightarrow C\cong\AA^1$ and 

876: discuss the proof of Theorem~\ref{thm!A1}. From Theorem~\ref{thm!class}, 

877: we know that certain holomorphic $D$-branes on $X_\defpar$ are classified by 

878: non-degenerate quintuples $(\V,\W, \B, i, j)$ satisfying the ADHM equation.

879: Consider the subclass of representations in $\Coh(\AA^1)$ with the simplest 

880: possible framing

881: $\W\cong\O_{\AA^1}$ and $\V$ a torsion $\Gamma$-sheaf on $\AA^1$. It follows

882: that $\J=0$ and $\I\in H^0(\AA^1, \V^\Gamma)$.

883: Decompose $\V$ and the map $B$ into $\Gamma$-components to obtain

884: torsion sheaves $\V_a$ and sheaf homomorphisms 

885: $\B_{ab}\colon \V_a\rightarrow \V_b$

886: indexed by nodes and edges of the McKay quiver.

887: 

888: Set $V_a=H^0(\AA^1, \V_a)$, and let $B_{ab}=H^0(\B_{ab})\colon V_a\rightarrow V_b$ 

889: to be the map on global sections induced by $\B_{ab}$. Let 

890: $\spvec\in V_0$ be the section corresponding to $\I$.

891: Let also $\Psi_a\colon V_a\rightarrow V_a$ be the map induced by 

892: multiplication by the section $t\in H^0(\AA^1, \O_{\A_1}) \cong \C[t]$. 

893: Theorem~\ref{thm!A1} follows from Theorem~\ref{thm!class}, together with

894: 

895: \begin{proposition} The map

896: \[ (\V, \O_C, \B, 0, 0) \mapsto (\{V_a\}, \{B_{ab}\}, \{\Psi_a\}, \spvec\in V_0)

897: \] 

898: sets up a one-to-one correspondence from this restricted set

899: of quiver ADHM data to representations of the affine \susy\ ADE quiver 

900: satisfying the relations~(\ref{rel1})-(\ref{rel2}).

901: \end{proposition} 

902: \begin{pf} Given $(\V, \B)$, the edge relations~(\ref{rel2}) 

903: $\Psi_a B_{ba} = B_{ba}\Psi_b$ 

904: for the data $(\{V_a\}, \{B_{ab}\}, \{\Psi_a\})$ hold by definition. 

905: Further, the ADHM equation for $(\V, \B)$ is

906: \[ \B\wedge \B + \defpar = 0 \in \Hom(\V, \V\otimes\det\sQ),\]

907: which in $\Gamma$-components says that 

908: \[ \sum_{b}\epsilon_{ab}\B_{ba}\circ\B_{ab}+\defpar_a=0\in \Hom(\V_a, \V_a).

909: \]

910: Replacing $\defpar_a$ by the polynomial $\defpoly_a$, and remembering

911: that the effect of $t\in H^0(\O_{\AA^1})$ on $H^0(\V)$ is exactly $\Psi_a$, 

912: for global sections we obtain

913: \[\sum_{b} \epsilon_{ab} B_{ba}\circ B_{ab}+\defpoly_a(\Psi_a)=0\in \Hom(V_a, V_a)

914: \]

915: which is exactly relation~(\ref{rel1}) for the node~$a$. 

916: 

917: Conversely, given a representation $(\{V_a\}, \{B_{ab}\}, \{\Psi_a\},\spvec\in V_0)$

918: of the \susy\ ADE quiver, define torsion sheaves attached to the nodes by

919: \[\begin{CD}\V_a = \coker\Big(V_a\otimes\O_{\AA^1}@>{1\otimes t - \Psi_a\otimes 1}>> V_a\otimes \O_{\AA^1}\Big).\end{CD}\]

920: Using Lemma~\ref{linalg} below, for adjacent nodes $a,b$ we have a diagram

921: \[\begin{CD} 0 & @>>> & {\ \ V_a\otimes\O_{\AA^1}} & @>{1\otimes t - \Psi_a\otimes 1}>> & {\ \ V_a\otimes\O_{\AA^1}} & @>>> \V_a @>>> 0 \\

922: & & & & @V{B_{ab}\otimes 1}VV && @V{B_{ab}\otimes 1}VV \\

923: 0 & @>>> & {\ \ V_b\otimes\O_{\AA^1}} & @>{1\otimes t - \Psi_b\otimes 1}>> & {\ \ V_b\otimes\O_{\AA^1}} & @>>> \V_b @>>> 0

924: \end{CD}

925: \]

926: which, by commutativity $\Psi_a B_{ba} = B_{ba}\Psi_b$, induces a map

927: $\B_{ab}\colon \V_a\rightarrow \V_b$. The converse of the above argument 

928: shows that the ADHM relation follows from the relations~(\ref{rel1}). 

929: By Lemma~\ref{linalg}, the two constructions are inverses to each other. 

930: \end{pf}

931: 

932: The proof used the elementary

933: 

934: \begin{lemma} Given a torsion sheaf $\V$ on $\AA^1=\spec \C[t]$, 

935: let $V=H^0(\AA^1, \V)$ and let $\Psi\colon V\rightarrow V$ be the map 

936: given by multiplication by $t\in H^0(\O_{\AA^1})$. Then the sequence of sheaves 

937: \[\begin{CD} 0  @>>> V\otimes\O_{\AA^1} @>{1\otimes t - \Psi\otimes 1}>>  V\otimes\O_{\AA^1}  @>c>> \V @>>> 0\end{CD}\]

938: is exact on~$\AA^1$, where $c\colon H^0(\V)\otimes \O_{\AA^1} \rightarrow \V$ is 

939: the canonical map. Conversely, given a vector space with an endomorphism

940: $(V, \Psi)$, the exact sequence defines a torsion sheaf $\V$ on $\AA^1$, 

941: and the two constructions are mutual inverses. 

942: \label{linalg}\end{lemma} 

943: %\begin{pf} \end{pf}

944: 

945: \begin{remark}\rm In this Lemma, $\V\cong\O_Z$ is a structure sheaf of 

946: a $0$-dimensional subscheme $Z\subset\AA^1$ if and only if $\Psi$ is a

947: regular endomorphism. Their moduli space is  

948: \[{\rm Mat}(n,\C)/\!/\GL(n,\C)\cong\{\mbox{regular endomorphisms}\}/\GL(n,\C)\cong \AA^n\cong(\AA^1)^{[n]},\] 

949: where the map is given by taking the coefficients of the characteristic polynomial 

950: of $\Psi$, which is also the equation of the corresponding subscheme.

951: \end{remark}

952: \subsection*{Acknowledgements} Thanks to Sheldon Katz, Eduard Looijenga, 

953: Tom Nevins and Tony Pantev for helpful remarks and correspondence. 

954: Special thanks to Ian Grojnowski for many conversations on subjects 

955: related to this paper. Support by a European Union Marie Curie 

956: Individual Fellowship and by OTKA grant $\#046878$ is also gratefully

957: acknowledged.

958: 

959: \begin{thebibliography}{99} 

960: \bibitem{az} M. Artin and J. Zhang, {\it Non-commutative projective schemes}, Adv. Math. 109 (1994), 228--287.

961: \bibitem{adhm} M. F. Atiyah, V. Drinfeld, N. Hitchin and Yu. Manin, {\it Geometric construction of instantons}, Phys. Lett. A 65 (1978), 185--187.

962: \bibitem{bgk} V. Baranovski, V. Ginzburg and A. Kuznetsov, {\it Quiver varieties and a non-commu\-ta\-tive~$\PP^2$}, Compositio Math. 134 (2002), 283--318. 

963: \bibitem{bgs} A.~Beilinson, V. Ginzburg and C. Soergel, {\em Koszul duality patterns in representation theory}, J. Am. Math. Soc. 9 (1996), 473--527.

964: \bibitem{bkr} T. Bridgeland, A. King and M. Reid, {\it The McKay correspondence as an equivalence of derived categories}, J. Am. Math. Soc. 14 (2001), 535--554. 

965: \bibitem{ckv} F. Cachazo, S. Katz and C. Vafa, {\it Geometric transitions and $N=1$ quiver theories}, hep-th/0108120.

966: \bibitem{cifkv} F. Cachazo, B. Fiol, K. Intriligator, S. Katz and C. Vafa, {\it A geometric unification of dualities}, Nucl. Phys. B628 (2002), 3--78. 

967: \bibitem{cb-h} W. Crawley-Boevey and Holland, {\it Non-commutative deformations of Kleinian surface singularities}, Duke Math. J. 92 (1998), 605--635.

968: \bibitem{penn} D.-E. Diaconescu, R. Dijkgraaf, R. Donagi, C. Hofman and T. Pantev, {\it Geometric transitions and integrable systems}, Nucl. Phys. B752 (2006) 329--390.

969: %\bibitem{dv} R. Dijkgraaf and C. Vafa, {\it Matrix models, topological strings and supersymmetric gauge theories}, Nucl. Phys. B644 (2002), 3--20.  

970: \bibitem{don} S. Donaldson, {\it Instantons and geometric invariant theory}, Comm. Math. Phys, Comm. Math. Phys. 93 (1984), 453--460.

971: \bibitem{gs} I. Gordon and S.P. Smith, {\it Representations of symplectic reflection algebras and resolutions of deformations of symplectic quotient singularities}, Math. Ann. 330 (2004), 185--200.

972: \bibitem{gk} P. Gothen and A. King, {\it Homological algebra of twisted quiver bundles}, J. London Math. Soc. 71 (2005), 85--99.

973: \bibitem{kko} A. Kapustin, A. Kuznetsov and D. Orlov, {\it Non-commutative instantons and twistor transform}, Comm. Math. Phys.  221  (2001), 385--432.

974: \bibitem{katz} S. Katz, {\it $ADE$ geometry and dualities}, minicourse, Workshop on Algebraic Geometry and Physics, Lisbon, September 2004.

975: %\bibitem{king} A. King, {\it Moduli of representations of finite-dimensional algebras}, Quart. J. Math. Oxford 45 (1994), 515--530.

976: %\bibitem{kronheimer}  {P. Kronheimer}, {\it The construction of ALE spaces as hyper-K\"ahler quotients}, J. Differential Geom. 29 (1989) 665--683.

977: \bibitem{kn} P. Kronheimer, H. Nakajima, {\it Yang--Mills instantons on ALE gravitational instantons}, Math. Ann. 288 (1990), 263--307.

978: \bibitem{nak} H. Nakajima, {\it Instantons on ALE spaces, quiver varieties and Kac--Moody algebras}, Duke M. J. 76 (1994), 365--416.

979: \bibitem{nak1} H. Nakajima, {\it Heisenberg algebra and Hilbert scheme of surfaces}, Ann. of Math. 145 (1997), 379--388. 

980: %\bibitem{nak2} H. Nakajima, {\it Quiver varieties and Kac--Moody algebras}, Duke 91

981: %\bibitem{nak3} H. Nakajima, {\it Lectures on Hilbert schemes of points on surfaces}, AMS, Providence, Rhode Island, 1999.

982: \bibitem{n-s} N. Nekrasov and J. Schwarz, {\it Instantons on noncommutative ${\mathbb R}^4$, and $(2,0)$ superconformal six dimensional theory}, Comm. Math. Phys. 198 (1998) 689--703.

983: %\bibitem{ns} T. Nevins and T. Stafford, {\it Sklyanin algebras and Hilbert schemes of points}, math.AG/0310045.

984: \bibitem{sz} B. Szendr\H oi, {\it Artin group actions on derived categories of coherent sheaves}, J. Reine Angew. Math. 572 (2004), 139--166.

985: \bibitem{vdb} M. Van den Bergh, {\it Non-commutative crepant resolutions}, The legacy of Niels Henrik Abel,  749--770, Springer, Berlin, 2004.

986: \bibitem{zhu} X. Zhu, {\it  Representations of $N=1$ $ADE$ quivers via reflection functors}, Mich. Math. J. 54 (2006) 671--686.

987: \end{thebibliography}

988: 

989: \noindent {\small {\sc Department of Mathematics, Utrecht University}

990: 

991: \noindent Current address:

992: 

993: \noindent {\sc Mathematical Institute, University of Oxford}

994: 

995: \noindent E-mail address: \tt szendroi@maths.ox.ac.uk}

996: 

997: \end{document}

998: 

999: