1: \documentclass[a4paper]{article}

2: \usepackage{amsmath,amsfonts,amssymb,amscd,graphicx,bez123}

3: 

4: %---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

5: %

6: %   The Dynkin diagrams Rational Double Points

7: %

8: %   Benjamin Friedrich

9: %   15. 05. 2002

10: %

11: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

12: 

13: \title{The Dynkin diagrams of Rational Double Points}

14: \author{Benjamin Friedrich\thanks{Supervisor: Prof. P.M.H. Wilson}}

15: 

16: \date{15. May 2002}

17: 

18: \begin{document}

19: 

20: % A2 - big

21: \newcommand{\Azweibig}

22: {

23: \setlength{\unitlength}{1mm}

24: \begin{picture}(15,10)(0,-5)

25: 

26: \put (0,0) {\circle{3}}

27: 

28: \put (2,0) {\line(1,0){6}}

29: 

30: \put (10,0) {\circle{3}}

31: 

32: \end{picture}

33: }

34: 

35: % E6

36: \newcommand{\Esechs}

37: {

38: \setlength{\unitlength}{0.6mm}

39: \begin{picture}(50,30)(-5,-20)

40: 

41: \put (0,0) {\circle{3}}

42: 

43: \put (-3,4) {\makebox{-2}}

44: 

45: \put (2,0) {\line(1,0){6}}

46: 

47: \put (10,0) {\circle{3}}

48: 

49: \put (7,4) {\makebox{-2}}

50: 

51: \put (12,0) {\line(1,0){6}}

52: 

53: \put (20,0) {\circle{3}}

54: 

55: \put (17,4) {\makebox{-2}}

56: 

57: \put (22,0) {\line(1,0){6}}

58: 

59: \put (30,0) {\circle{3}}

60: 

61: \put (27,4) {\makebox{-2}}

62: 

63: \put (32,0) {\line(1,0){6}}

64: 

65: \put (40,0) {\circle{3}}

66: 

67: \put (37,4) {\makebox{-2}}

68: 

69: \put (20,-8) {\line(0,1){6}}

70: 

71: \put (20,-10) {\circle{3}}

72: 

73: \put (17,-17) {\makebox{-2}}

74: 

75: \end{picture}

76: }

77: 

78: % E6 - Znum

79: \newcommand{\Esechsznum}

80: {

81: \setlength{\unitlength}{0.6mm}

82: \begin{picture}(50,30)(-5,-20)

83: 

84: \put (0,0) {\circle{3}}

85: 

86: \put (-2,4) {\makebox{1}}

87: 

88: \put (2,0) {\line(1,0){6}}

89: 

90: \put (10,0) {\circle{3}}

91: 

92: \put (8,4) {\makebox{2}}

93: 

94: \put (12,0) {\line(1,0){6}}

95: 

96: \put (20,0) {\circle{3}}

97: 

98: \put (18,4) {\makebox{3}}

99: 

100: \put (22,0) {\line(1,0){6}}

101: 

102: \put (30,0) {\circle{3}}

103: 

104: \put (28,4) {\makebox{2}}

105: 

106: \put (32,0) {\line(1,0){6}}

107: 

108: \put (40,0) {\circle{3}}

109: 

110: \put (38,4) {\makebox{1}}

111: 

112: \put (20,-8) {\line(0,1){6}}

113: 

114: \put (20,-10) {\circle{3}}

115: 

116: \put (18,-17) {\makebox{2}}

117: 

118: \end{picture}

119: }

120: 

121: % E7

122: \newcommand{\Esieben}

123: {

124: \setlength{\unitlength}{0.6mm}

125: \begin{picture}(60,30)(-5,-20)

126: 

127: \put (0,0) {\circle{3}}

128: 

129: \put (-3,4) {\makebox{-2}}

130: 

131: \put (2,0) {\line(1,0){6}}

132: 

133: \put (10,0) {\circle{3}}

134: 

135: \put (7,4) {\makebox{-2}}

136: 

137: \put (12,0) {\line(1,0){6}}

138: 

139: \put (20,0) {\circle{3}}

140: 

141: \put (17,4) {\makebox{-2}}

142: 

143: \put (22,0) {\line(1,0){6}}

144: 

145: \put (30,0) {\circle{3}}

146: 

147: \put (27,4) {\makebox{-2}}

148: 

149: \put (32,0) {\line(1,0){6}}

150: 

151: \put (40,0) {\circle{3}}

152: 

153: \put (37,4) {\makebox{-2}}

154: 

155: \put (42,0) {\line(1,0){6}}

156: 

157: \put (50,0) {\circle{3}}

158: 

159: \put (47,4) {\makebox{-2}}

160: 

161: \put (20,-8) {\line(0,1){6}}

162: 

163: \put (20,-10) {\circle{3}}

164: 

165: \put (17,-17) {\makebox{-2}}

166: 

167: \end{picture}

168: }

169: 

170: % E7 - Znum

171: \newcommand{\Esiebenznum}

172: {

173: \setlength{\unitlength}{0.6mm}

174: \begin{picture}(60,30)(-5,-20)

175: 

176: \put (0,0) {\circle{3}}

177: 

178: \put (-2,4) {\makebox{2}}

179: 

180: \put (2,0) {\line(1,0){6}}

181: 

182: \put (10,0) {\circle{3}}

183: 

184: \put (8,4) {\makebox{3}}

185: 

186: \put (12,0) {\line(1,0){6}}

187: 

188: \put (20,0) {\circle{3}}

189: 

190: \put (18,4) {\makebox{4}}

191: 

192: \put (22,0) {\line(1,0){6}}

193: 

194: \put (30,0) {\circle{3}}

195: 

196: \put (28,4) {\makebox{3}}

197: 

198: \put (32,0) {\line(1,0){6}}

199: 

200: \put (40,0) {\circle{3}}

201: 

202: \put (38,4) {\makebox{2}}

203: 

204: \put (42,0) {\line(1,0){6}}

205: 

206: \put (50,0) {\circle{3}}

207: 

208: \put (48,4) {\makebox{1}}

209: 

210: \put (20,-8) {\line(0,1){6}}

211: 

212: \put (20,-10) {\circle{3}}

213: 

214: \put (18,-17) {\makebox{2}}

215: 

216: \end{picture}

217: }

218: 

219: % E8

220: \newcommand{\Eacht}

221: {

222: \setlength{\unitlength}{0.6mm}

223: \begin{picture}(70,30)(-5,-20)

224: 

225: \put (0,0) {\circle{3}}

226: 

227: \put (-3,4) {\makebox{-2}}

228: 

229: \put (2,0) {\line(1,0){6}}

230: 

231: \put (10,0) {\circle{3}}

232: 

233: \put (7,4) {\makebox{-2}}

234: 

235: \put (12,0) {\line(1,0){6}}

236: 

237: \put (20,0) {\circle{3}}

238: 

239: \put (17,4) {\makebox{-2}}

240: 

241: \put (22,0) {\line(1,0){6}}

242: 

243: \put (30,0) {\circle{3}}

244: 

245: \put (27,4) {\makebox{-2}}

246: 

247: \put (32,0) {\line(1,0){6}}

248: 

249: \put (40,0) {\circle{3}}

250: 

251: \put (37,4) {\makebox{-2}}

252: 

253: \put (42,0) {\line(1,0){6}}

254: 

255: \put (50,0) {\circle{3}}

256: 

257: \put (47,4) {\makebox{-2}}

258: 

259: \put (52,0) {\line(1,0){6}}

260: 

261: \put (60,0) {\circle{3}}

262: 

263: \put (57,4) {\makebox{-2}}

264: 

265: \put (20,-8) {\line(0,1){6}}

266: 

267: \put (20,-10) {\circle{3}}

268: 

269: \put (17,-17) {\makebox{-2}}

270: 

271: \end{picture}

272: }

273: 

274: % E8 - Znum

275: \newcommand{\Eachtznum}

276: {

277: \setlength{\unitlength}{0.6mm}

278: \begin{picture}(70,30)(-5,-20)

279: 

280: \put (0,0) {\circle{3}}

281: 

282: \put (-2,4) {\makebox{2}}

283: 

284: \put (2,0) {\line(1,0){6}}

285: 

286: \put (10,0) {\circle{3}}

287: 

288: \put (8,4) {\makebox{4}}

289: 

290: \put (12,0) {\line(1,0){6}}

291: 

292: \put (20,0) {\circle{3}}

293: 

294: \put (18,4) {\makebox{6}}

295: 

296: \put (22,0) {\line(1,0){6}}

297: 

298: \put (30,0) {\circle{3}}

299: 

300: \put (28,4) {\makebox{5}}

301: 

302: \put (32,0) {\line(1,0){6}}

303: 

304: \put (40,0) {\circle{3}}

305: 

306: \put (38,4) {\makebox{4}}

307: 

308: \put (42,0) {\line(1,0){6}}

309: 

310: \put (50,0) {\circle{3}}

311: 

312: \put (48,4) {\makebox{3}}

313: 

314: \put (52,0) {\line(1,0){6}}

315: 

316: \put (60,0) {\circle{3}}

317: 

318: \put (58,4) {\makebox{2}}

319: 

320: \put (20,-8) {\line(0,1){6}}

321: 

322: \put (20,-10) {\circle{3}}

323: 

324: \put (18,-17) {\makebox{3}}

325: 

326: \end{picture}

327: }

328: 

329: % Dn

330: \newcommand{\Dn}

331: {

332: \setlength{\unitlength}{0.6mm}

333: \begin{picture}(70,30)(-5,-20)

334: 

335: \put (0,0) {\circle{3}}

336: 

337: \put (-3,4) {\makebox{-2}}

338: 

339: \put (2,0) {\line(1,0){6}}

340: 

341: \put (10,0) {\circle{3}}

342: 

343: \put (7,4) {\makebox{-2}}

344: 

345: \put (12,0) {\line(1,0){6}}

346: 

347: \put (20,0) {\circle{3}}

348: 

349: \put (17,4) {\makebox{-2}}

350: 

351: \put (22,0) {\line(1,0){6}}

352: 

353: \put (32,0) {\circle*{2}}

354: \put (35,0) {\circle*{2}}

355: \put (38,0) {\circle*{2}}

356: 

357: \put (42,0) {\line(1,0){6}}

358: 

359: \put (50,0) {\circle{3}}

360: 

361: \put (47,4) {\makebox{-2}}

362: 

363: \put (52,0) {\line(1,0){6}}

364: 

365: \put (60,0) {\circle{3}}

366: 

367: \put (57,4) {\makebox{-2}}

368: 

369: \put (10,-8) {\line(0,1){6}}

370: 

371: \put (10,-10) {\circle{3}}

372: 

373: \put (8,-17) {\makebox{-2}}

374: 

375: \end{picture}

376: }

377: 

378: % Dn - Znum

379: \newcommand{\Dnznum}

380: {

381: \setlength{\unitlength}{0.6mm}

382: \begin{picture}(70,30)(-5,-20)

383: 

384: \put (0,0) {\circle{3}}

385: 

386: \put (-2,4) {\makebox{1}}

387: 

388: \put (2,0) {\line(1,0){6}}

389: 

390: \put (10,0) {\circle{3}}

391: 

392: \put (8,4) {\makebox{2}}

393: 

394: \put (12,0) {\line(1,0){6}}

395: 

396: \put (20,0) {\circle{3}}

397: 

398: \put (18,4) {\makebox{2}}

399: 

400: \put (22,0) {\line(1,0){6}}

401: 

402: \put (32,0) {\circle*{2}}

403: \put (35,0) {\circle*{2}}

404: \put (38,0) {\circle*{2}}

405: 

406: \put (42,0) {\line(1,0){6}}

407: 

408: \put (50,0) {\circle{3}}

409: 

410: \put (48,4) {\makebox{2}}

411: 

412: \put (52,0) {\line(1,0){6}}

413: 

414: \put (60,0) {\circle{3}}

415: 

416: \put (58,4) {\makebox{1}}

417: 

418: \put (10,-8) {\line(0,1){6}}

419: 

420: \put (10,-10) {\circle{3}}

421: 

422: \put (8,-17) {\makebox{1}}

423: 

424: \end{picture}

425: }

426: 

427: % An

428: \newcommand{\An}

429: {

430: \setlength{\unitlength}{0.6mm}

431: \begin{picture}(70,15)(-5,-5)

432: 

433: \put (0,0) {\circle{3}}

434: 

435: \put (-3,4) {\makebox{-2}}

436: 

437: \put (2,0) {\line(1,0){6}}

438: 

439: \put (10,0) {\circle{3}}

440: 

441: \put (7,4) {\makebox{-2}}

442: 

443: \put (12,0) {\line(1,0){6}}

444: 

445: \put (20,0) {\circle{3}}

446: 

447: \put (17,4) {\makebox{-2}}

448: 

449: \put (22,0) {\line(1,0){6}}

450: 

451: \put (32,0) {\circle*{2}}

452: \put (35,0) {\circle*{2}}

453: \put (38,0) {\circle*{2}}

454: 

455: \put (42,0) {\line(1,0){6}}

456: 

457: \put (50,0) {\circle{3}}

458: 

459: \put (47,4) {\makebox{-2}}

460: 

461: \put (52,0) {\line(1,0){6}}

462: 

463: \put (60,0) {\circle{3}}

464: 

465: \put (57,4) {\makebox{-2}}

466: 

467: \end{picture}

468: }

469: 

470: % An - Znum

471: \newcommand{\Anznum}

472: {

473: \setlength{\unitlength}{0.6mm}

474: \begin{picture}(70,15)(-5,-5)

475: 

476: \put (0,0) {\circle{3}}

477: 

478: \put (-2,4) {\makebox{1}}

479: 

480: \put (2,0) {\line(1,0){6}}

481: 

482: \put (10,0) {\circle{3}}

483: 

484: \put (8,4) {\makebox{1}}

485: 

486: \put (12,0) {\line(1,0){6}}

487: 

488: \put (20,0) {\circle{3}}

489: 

490: \put (18,4) {\makebox{1}}

491: 

492: \put (22,0) {\line(1,0){6}}

493: 

494: \put (32,0) {\circle*{2}}

495: \put (35,0) {\circle*{2}}

496: \put (38,0) {\circle*{2}}

497: 

498: \put (42,0) {\line(1,0){6}}

499: 

500: \put (50,0) {\circle{3}}

501: 

502: \put (48,4) {\makebox{1}}

503: 

504: \put (52,0) {\line(1,0){6}}

505: 

506: \put (60,0) {\circle{3}}

507: 

508: \put (58,4) {\makebox{1}}

509: 

510: \end{picture}

511: }

512: 

513: % cross

514: \newcommand{\cross}

515: {

516: \setlength{\unitlength}{1mm}

517: \begin{picture}(80,30)(-15,-15)

518: 

519: \put (0,0) {\circle{3}}

520: 

521: \put (-3,4) {\makebox{$e_1$}}

522: 

523: \put (2,0) {\line(1,0){6}}

524: 

525: \put (10,0) {\circle{3}}

526: 

527: \put (7,4) {\makebox{$e_2$}}

528: 

529: \put (12,0) {\line(1,0){6}}

530: 

531: \put (22,0) {\circle*{1}}

532: \put (25,0) {\circle*{1}}

533: \put (28,0) {\circle*{1}}

534: 

535: \put (32,0) {\line(1,0){6}}

536: 

537: \put (40,0) {\circle{3}}

538: 

539: \put (37,4) {\makebox{$e_{n-1}$}}

540: 

541: \put (42,0) {\line(1,0){6}}

542: 

543: \put (50,0) {\circle{3}}

544: 

545: \put (47,4) {\makebox{$e_n$}}

546: 

547: \put (52,2) {\line(1,1){4}}

548: 

549: \put (58,8) {\circle{3}}

550: 

551: \put (60,7) {\makebox{$f_3$}}

552: 

553: \put (52,-2) {\line(1,-1){4}}

554: 

555: \put (58,-8) {\circle{3}}

556: 

557: \put (60,-9) {\makebox{$f_4$}}

558: 

559: \put (-6,6) {\line(1,-1){4}}

560: 

561: \put (-8,8) {\circle{3}}

562: 

563: \put (-14,7) {\makebox{$f_1$}}

564: 

565: \put (-6,-6) {\line(1,1){4}}

566: 

567: \put (-8,-8) {\circle{3}}

568: 

569: \put (-14,-9) {\makebox{$f_2$}}

570: 

571: \end{picture}

572: }

573: 

574: % T-tree

575: \newcommand{\Ttree}

576: {

577: \setlength{\unitlength}{0.7mm}

578: \begin{picture}(120,60)(-5,-50)

579: 

580: \put (0,0) {\circle{3}}

581: 

582: \put (2,0) {\line(1,0){6}}

583: 

584: \put (10,0) {\circle{3}}

585: 

586: \put (12,0) {\line(1,0){6}}

587: 

588: \put (22,0) {\circle*{1}}

589: \put (25,0) {\circle*{1}}

590: \put (28,0) {\circle*{1}}

591: 

592: \put (32,0) {\line(1,0){6}}

593: 

594: \put (40,0) {\circle{3}}

595: 

596: \put (42,0) {\line(1,0){6}}

597: 

598: \put (50,0) {\circle{3}}

599: 

600: \put (52,0) {\line(1,0){6}}

601: 

602: \put (60,0) {\circle{3}}

603: 

604: \put (62,0) {\line(1,0){6}}

605: 

606: \put (72,0) {\circle*{1}}

607: \put (75,0) {\circle*{1}}

608: \put (78,0) {\circle*{1}}

609: 

610: \put (82,0) {\line(1,0){6}}

611: 

612: \put (90,0) {\circle{3}}

613: 

614: \put (92,0) {\line(1,0){6}}

615: 

616: \put (100,0) {\circle{3}}

617: 

618: \put (50,-8) {\line(0,1){6}}

619: 

620: \put (50,-10) {\circle{3}}

621: 

622: \put (50,-18) {\line(0,1){6}}

623: 

624: \put (50,-22) {\circle*{1}}

625: \put (50,-25) {\circle*{1}}

626: \put (50,-28) {\circle*{1}}

627: 

628: \put (50,-38) {\line(0,1){6}}

629: 

630: \put (50,-40) {\circle{3}}

631: 

632: \end{picture}

633: }

634: 

635: % cyclus

636: \newcommand{\cyclus}

637: {

638: \setlength{\unitlength}{1mm}

639: \begin{picture}(30,25)(-17,-5)

640: 

641: \put (0,0) {\circle{3}}

642: 

643: \put (-1,-5) {\makebox{$e_1$}}

644: 

645: \put (2,2) {\line(1,1){4}}

646: 

647: \put (8,8) {\circle{3}}

648: 

649: \put (10,7) {\makebox{$e_2$}}

650: 

651: \put (8,10) {\line(0,1){6}}

652: 

653: \put (8,18) {\circle{3}}

654: 

655: \put (10,17) {\makebox{$e_3$}}

656: 

657: \put (-3,18) {\circle*{1}}

658: \put (0,18) {\circle*{1}}

659: \put (3,18) {\circle*{1}}

660: 

661: \put (-6,6) {\line(1,-1){4}}

662: 

663: \put (-8,8) {\circle{3}}

664: 

665: \put (-14,7) {\makebox{$e_n$}}

666: 

667: \put (-8,10) {\line(0,1){6}}

668: 

669: \put (-8,18) {\circle{3}}

670: 

671: \put (-17,17) {\makebox{$e_{n-1}$}}

672: 

673: \end{picture}

674: }

675: 

676: % Hirzebruch - Jung

677: \newcommand{\HiJu}

678: {

679: \setlength{\unitlength}{1mm}

680: \begin{picture}(60,20)(-5,-5)

681: 

682: \put (0,0) {\circle{3}}

683: 

684: \put (-4,4) {\makebox{$-b_1$}}

685: 

686: \put (2,0) {\line(1,0){6}}

687: 

688: \put (10,0) {\circle{3}}

689: 

690: \put (6,4) {\makebox{$-b_2$}}

691: 

692: \put (12,0) {\line(1,0){6}}

693: 

694: \put (20,0) {\circle{3}}

695: 

696: \put (16,4) {\makebox{$-b_3$}}

697: 

698: \put (22,0) {\line(1,0){6}}

699: 

700: \put (32,0) {\circle*{1}}

701: \put (35,0) {\circle*{1}}

702: \put (38,0) {\circle*{1}}

703: 

704: \put (42,0) {\line(1,0){6}}

705: 

706: \put (50,0) {\circle{3}}

707: 

708: \put (46,4) {\makebox{$-b_k$}}

709: 

710: \end{picture}

711: }

712: 

713: % E8 labelled

714: \newcommand{\Eachtlab}

715: {

716: \setlength{\unitlength}{1mm}

717: \begin{picture}(70,30)(-5,-20)

718: 

719: \put (0,0) {\circle{5}}

720: 

721: \put (-2,4) {\makebox{-2}}

722: 

723: \put (-2,-1) {\makebox{$e_1$}}

724: 

725: \put (3,0) {\line(1,0){4}}

726: 

727: \put (10,0) {\circle{5}}

728: 

729: \put (8,4) {\makebox{-2}}

730: 

731: \put (8,-1) {\makebox{$e_2$}}

732: 

733: \put (13,0) {\line(1,0){4}}

734: 

735: \put (20,0) {\circle{5}}

736: 

737: \put (18,4) {\makebox{-2}}

738: 

739: \put (18,-1) {\makebox{$e_3$}}

740: 

741: \put (23,0) {\line(1,0){4}}

742: 

743: \put (30,0) {\circle{5}}

744: 

745: \put (28,4) {\makebox{-2}}

746: 

747: \put (28,-1) {\makebox{$e_5$}}

748: 

749: \put (33,0) {\line(1,0){4}}

750: 

751: \put (40,0) {\circle{5}}

752: 

753: \put (38,4) {\makebox{-2}}

754: 

755: \put (38,-1) {\makebox{$e_6$}}

756: 

757: \put (43,0) {\line(1,0){4}}

758: 

759: \put (50,0) {\circle{5}}

760: 

761: \put (48,4) {\makebox{-2}}

762: 

763: \put (48,-1) {\makebox{$e_7$}}

764: 

765: \put (53,0) {\line(1,0){4}}

766: 

767: \put (60,0) {\circle{5}}

768: 

769: \put (58,4) {\makebox{-2}}

770: 

771: \put (58,-1) {\makebox{$e_8$}}

772: 

773: \put (20,-7) {\line(0,1){4}}

774: 

775: \put (20,-10) {\circle{5}}

776: 

777: \put (18,-17) {\makebox{-2}}

778: 

779: \put (18,-11) {\makebox{$e_4$}}

780: 

781: \end{picture}

782: }

783: 

784: % E6 - plain

785: \newcommand{\Esechsplain}

786: {

787: \setlength{\unitlength}{0.6mm}

788: \begin{picture}(50,20)(-5,-15)

789: 

790: \put (0,0) {\circle{3}}

791: 

792: \put (2,0) {\line(1,0){6}}

793: 

794: \put (10,0) {\circle{3}}

795: 

796: \put (12,0) {\line(1,0){6}}

797: 

798: \put (20,0) {\circle{3}}

799: 

800: \put (22,0) {\line(1,0){6}}

801: 

802: \put (30,0) {\circle{3}}

803: 

804: \put (32,0) {\line(1,0){6}}

805: 

806: \put (40,0) {\circle{3}}

807: 

808: \put (20,-8) {\line(0,1){6}}

809: 

810: \put (20,-10) {\circle{3}}

811: 

812: \end{picture}

813: }

814: 

815: % E7 - plain

816: \newcommand{\Esiebenplain}

817: {

818: \setlength{\unitlength}{0.6mm}

819: \begin{picture}(60,20)(-5,-15)

820: 

821: \put (0,0) {\circle{3}}

822: 

823: \put (2,0) {\line(1,0){6}}

824: 

825: \put (10,0) {\circle{3}}

826: 

827: \put (12,0) {\line(1,0){6}}

828: 

829: \put (20,0) {\circle{3}}

830: 

831: \put (22,0) {\line(1,0){6}}

832: 

833: \put (30,0) {\circle{3}}

834: 

835: \put (32,0) {\line(1,0){6}}

836: 

837: \put (40,0) {\circle{3}}

838: 

839: \put (42,0) {\line(1,0){6}}

840: 

841: \put (50,0) {\circle{3}}

842: 

843: \put (20,-8) {\line(0,1){6}}

844: 

845: \put (20,-10) {\circle{3}}

846: 

847: \end{picture}

848: }

849: 

850: % E8 - plain

851: \newcommand{\Eachtplain}

852: {

853: \setlength{\unitlength}{0.6mm}

854: \begin{picture}(70,20)(-5,-15)

855: 

856: \put (0,0) {\circle{3}}

857: 

858: \put (2,0) {\line(1,0){6}}

859: 

860: \put (10,0) {\circle{3}}

861: 

862: \put (12,0) {\line(1,0){6}}

863: 

864: \put (20,0) {\circle{3}}

865: 

866: \put (22,0) {\line(1,0){6}}

867: 

868: \put (30,0) {\circle{3}}

869: 

870: \put (32,0) {\line(1,0){6}}

871: 

872: \put (40,0) {\circle{3}}

873: 

874: \put (42,0) {\line(1,0){6}}

875: 

876: \put (50,0) {\circle{3}}

877: 

878: \put (52,0) {\line(1,0){6}}

879: 

880: \put (60,0) {\circle{3}}

881: 

882: \put (20,-8) {\line(0,1){6}}

883: 

884: \put (20,-10) {\circle{3}}

885: 

886: \end{picture}

887: }

888: 

889: % Dn - plain

890: \newcommand{\Dnplain}

891: {

892: \setlength{\unitlength}{0.6mm}

893: \begin{picture}(70,20)(-5,-15)

894: 

895: \put (0,0) {\circle{3}}

896: 

897: \put (2,0) {\line(1,0){6}}

898: 

899: \put (10,0) {\circle{3}}

900: 

901: \put (12,0) {\line(1,0){6}}

902: 

903: \put (20,0) {\circle{3}}

904: 

905: \put (22,0) {\line(1,0){6}}

906: 

907: \put (32,0) {\circle*{1}}

908: \put (35,0) {\circle*{1}}

909: \put (38,0) {\circle*{1}}

910: 

911: \put (42,0) {\line(1,0){6}}

912: 

913: \put (50,0) {\circle{3}}

914: 

915: \put (52,0) {\line(1,0){6}}

916: 

917: \put (60,0) {\circle{3}}

918: 

919: \put (10,-8) {\line(0,1){6}}

920: 

921: \put (10,-10) {\circle{3}}

922: 

923: \end{picture}

924: }

925: 

926: % An - plain

927: \newcommand{\Anplain}

928: {

929: \setlength{\unitlength}{0.6mm}

930: \begin{picture}(70,10)(-5,-5)

931: 

932: \put (0,0) {\circle{3}}

933: 

934: \put (2,0) {\line(1,0){6}}

935: 

936: \put (10,0) {\circle{3}}

937: 

938: \put (12,0) {\line(1,0){6}}

939: 

940: \put (20,0) {\circle{3}}

941: 

942: \put (22,0) {\line(1,0){6}}

943: 

944: \put (32,0) {\circle*{1}}

945: \put (35,0) {\circle*{1}}

946: \put (38,0) {\circle*{1}}

947: 

948: \put (42,0) {\line(1,0){6}}

949: 

950: \put (50,0) {\circle{3}}

951: 

952: \put (52,0) {\line(1,0){6}}

953: 

954: \put (60,0) {\circle{3}}

955: 

956: \end{picture}

957: }

958: 

959: % roots

960: \newcommand{\roots}

961: {

962: \setlength{\unitlength}{1.5mm}

963: \begin{picture}(30,27)(-15,0)

964: 

965: \put (0,4) {\circle{3}}

966: 

967: \put (-1,0) {\makebox{$\alpha_{3,2}$}}

968: 

969: \put (2,6) {\line(2,1){4}}

970: 

971: \put (8,9) {\circle{3}}

972: 

973: \put (10,8) {\makebox{$\alpha_{1,2}$}}

974: 

975: \put (8,11) {\line(0,1){4}}

976: 

977: \put (8,17) {\circle{3}}

978: 

979: \put (10,16) {\makebox{$\alpha_{1,3}$}}

980: 

981: \put (-6,7) {\line(2,-1){4}}

982: 

983: \put (-8,9) {\circle{3}}

984: 

985: \put (-14,8) {\makebox{$\alpha_{3,1}$}}

986: 

987: \put (-8,11) {\line(0,1){4}}

988: 

989: \put (-8,17) {\circle{3}}

990: 

991: \put (-14,16) {\makebox{$\alpha_{2,1}$}}

992: 

993: \put (3,12) {\line(2,-1){3}}

994: 

995: \put (-6,16) {\line(2,-1){3}}

996: 

997: \put (2,21) {\line(2,-1){4}}

998: 

999: \put (-6,19) {\line(2,1){4}}

1000: 

1001: \put (3,14) {\line(2,1){3}}

1002: 

1003: \put (-6,10) {\line(2,1){3}}

1004: 

1005: \put (0,15) {\line(0,1){5}}

1006: 

1007: \put (0,6) {\line(0,1){5}}

1008: 

1009: \put (0,22) {\circle{3}}

1010: 

1011: \put (-1,25) {\makebox{$\alpha_{2,3}$}}

1012: 

1013: \put (0,13) {\circle{3}}

1014: 

1015: \end{picture}

1016: }

1017: 

1018: 

1019: % toric description of blow-up

1020: \newcommand{\toric}{

1021: \setlength{\unitlength}{.3mm}

1022: \begin{picture}(110,110)(-10,-10)

1023: 

1024: \put (0,0) {\vector(1,0){90}}

1025: 

1026: \put (0,0) {\vector(0,1){90}}

1027: 

1028: \put (0,0) {\line(1,2){40}}

1029: 

1030: \multiput (0,0) (15,15) {4} {\line(1,1){12}}

1031: 

1032: \put (30,0) {\circle*{2}}

1033: 

1034: \put (30,30) {\circle*{2}}

1035: 

1036: \put (30,60) {\circle*{2}}

1037: 

1038: \put (33,-10) {\makebox{$x \equiv (1,0)$}}

1039: 

1040: \put (33,21) {\makebox{$y \equiv (1,1)$}}

1041: 

1042: \put (33,55) {\makebox{$z \equiv (1,2)$}}

1043: 

1044: \end{picture}

1045: }

1046: 

1047: 

1048: % cone

1049: \newcommand{\cone}

1050: {

1051: \setlength{\unitlength}{0.5mm}

1052: \begin{picture}(75,80)(-35,-65)

1053: 

1054: \cbezier(-20,0)(-20,5.522847498)(-11.045694996,10)(0,10)

1055: \cbezier(0,10)(11.045694996,10)(20,5.522847498)(20,0)

1056: \cbezier(20,0)(20,-5.522847498)(11.045694996,-10)(0,-10)

1057: \cbezier(0,-10)(-11.045694996,-10)(-20,-5.522847498)(-20,0)

1058: 

1059: \put (-30,-53) {\vector(1,1){60}}

1060: \put (32,5) {\makebox{$x$}}

1061: 

1062: \put (30,-53) {\vector(-1,1){60}}

1063: \put (-35,5) {\makebox{$z$}}

1064: 

1065: \put (0,-23) {\vector(1,0){35}}

1066: \put (30,-30) {\makebox{$y$}}

1067: 

1068: \cbezier[20](-20,-46)(-20,-40.477152502)(-11.045694996,-36)(0,-36)

1069: \cbezier[20](0,-36)(11.045694996,-36)(20,-40.477152502)(20,-46)

1070: \cbezier(20,-46)(20,-51.522847498)(11.045694996,-56)(0,-56)

1071: \cbezier(0,-56)(-11.045694996,-56)(-20,-51.522847498)(-20,-46)

1072: 

1073: \end{picture}

1074: }

1075: 

1076: 

1077: % resolution - real image

1078: \newcommand{\resreal}{

1079: \setlength{\unitlength}{0.5mm}

1080: \begin{picture}(170,105)(-150,-85)

1081: 

1082: % cone

1083: 

1084: \cbezier(-20,0)(-20,5.522847498)(-11.045694996,10)(0,10)

1085: \cbezier(0,10)(11.045694996,10)(20,5.522847498)(20,0)

1086: \cbezier(20,0)(20,-5.522847498)(11.045694996,-10)(0,-10)

1087: \cbezier(0,-10)(-11.045694996,-10)(-20,-5.522847498)(-20,0)

1088: 

1089: \put (-30,-53) {\line(1,1){60}}

1090: 

1091: \put (30,-53) {\line(-1,1){60}}

1092: 

1093: \put (0,-23) {\circle{3}}

1094: 

1095: \cbezier[20](-20,-46)(-20,-40.477152502)(-11.045694996,-36)(0,-36)

1096: \cbezier[20](0,-36)(11.045694996,-36)(20,-40.477152502)(20,-46)

1097: \cbezier(20,-46)(20,-51.522847498)(11.045694996,-56)(0,-56)

1098: \cbezier(0,-56)(-11.045694996,-56)(-20,-51.522847498)(-20,-46)

1099: 

1100: % cylinder

1101: 

1102: \cbezier[20](-100,-23)(-100,-17.477152502)(-91.045694996,-13)(-80,-13)

1103: \cbezier[20](-80,-13)(-68.95430504,-13)(-60,-17.477152502)(-60,-23)

1104: \cbezier(-60,-23)(-60,-28.522847498)(-68.95430504,-33)(-80,-33)

1105: \cbezier(-80,-33)(-91.045694996,-33)(-100,-28.522847498)(-100,-23)

1106: 

1107: \put (-100,-53) {\line(0,1){60}}

1108: \put (-60,-53) {\line(0,1){60}}

1109: 

1110: % morphism

1111: 

1112: \put (-51,-23) {\vector(1,0){34}}

1113: 

1114: % text

1115: 

1116: \put (-119,-65) {\makebox{

1117: $\widetilde{X}_{\rm real} = T^{\ast} {\mathbb R}{\mathbb P}^1 $}}

1118: \put (-100,-75) {\makebox{

1119: $ = T^{\ast} S^1 = S^1 \times {\mathbb R}$}}

1120: 

1121: \put (-10,-65) {\makebox{$X_{\rm real}$}}

1122: 

1123: \put (-143,-23) {\makebox{$ E_{\rm real} = {\mathbb R}{\mathbb P}^1 $}}

1124: \put (-127,-33) {\makebox{$=S^1$}}

1125: 

1126: \end{picture}

1127: }

1128: 

1129: 

1130: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1131: %

1132: % settings and self-defined commands

1133: %

1134: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1135: 

1136: \newcommand{\GG}{\ensuremath{\mathfrak g}}

1137: \newcommand{\OO}{\mathcal O}

1138: \newcommand{\Ox}{\ensuremath{\OO_X}}

1139: \newcommand{\Oxx}{\ensuremath{ \OO_{\widetilde{X}}}}

1140: \newcommand{\res}{\ensuremath{ \pi : \widetilde{X} \rightarrow X}}

1141: \newcommand{\Znum}{\ensuremath{ Z_{\rm num}}}

1142: \newcommand{\PP}{\ensuremath{{\mathbb P}^1}}

1143: \newcommand{\CC}{\ensuremath{\mathbb C}}

1144: \newcommand{\ZZ}{\ensuremath{\mathbb Z}}

1145: \newcommand{\degz}{\text{deg}_{\Znum}}

1146: \newcommand{\II}{\ensuremath{\mathcal J}}

1147: \newcommand{\mm}{\mathfrak m}

1148: \newcommand{\XX}{\ensuremath{\widetilde{X}}}

1149: \newcommand{\XXX}{\ensuremath{\XX ^{\prime}}}

1150: 

1151: \pagestyle{plain}

1152: \setlength{\parindent}{0mm}

1153: 

1154: \newtheorem{theorem}{Theorem}[section]

1155: \newtheorem{proposition}[theorem]{Proposition}

1156: \newtheorem{lemma}[theorem]{Lemma}

1157: \newtheorem{corollary}[theorem]{Corollary}

1158: \newtheorem{defprop}[theorem]{Definition and Proposition}

1159: \newtheorem{remark}[theorem]{Remark}

1160: 

1161: %\include{pictures}

1162: 

1163: \begin{titlepage}

1164: 

1165: \begin{flushright}

1166:   Part III Essay \\ Easter 2002

1167: \end{flushright}

1168: 

1169: \bigskip

1170: 

1171: \begin{center}

1172:   \LARGE The Dynkin Diagrams of Rational Double Points

1173: \end{center}

1174: 

1175: \bigskip

1176: 

1177: \begin{center}

1178:   \normalsize Benjamin Friedrich \\

1179:   \smallskip

1180:   \sl Trinity College \\

1181:   \bigskip

1182:   {\it Home address:} \\

1183:   Kleiner Warnowdamm 10 \\

1184:   18109 Rostock \\

1185:   Germany \\

1186:   {\tt bf216@hermes.cam.ac.uk} \\

1187:   \bigskip

1188:   {\rm 15 May 2002}

1189: \end{center}

1190: 

1191: \bigskip

1192: 

1193: \begin{center}

1194:   I declare that this essay is work done as part of the Part III Examination. It is the

1195:   result of my own work, and except where stated otherwise, includes nothing which was

1196:   performed in collaboration. No part of this essay has been submitted for a degree or

1197:   any such qualification. \\[2cm]

1198:   Benjamin Friedrich

1199: \end{center}

1200: 

1201: \end{titlepage}

1202: 

1203: \begin{titlepage}

1204: 

1205: \pagebreak

1206: \thispagestyle{empty}

1207: 

1208: \begin{flushright}

1209:   Part III Essay \\ Easter 2002

1210: \end{flushright}

1211: 

1212: \bigskip

1213: 

1214: \begin{center}

1215:   \LARGE The Dynkin Diagrams of Rational Double Points

1216: \end{center}

1217: 

1218: \bigskip

1219: 

1220: \setlength{\parindent}{0mm}

1221: \begin{abstract}

1222: 

1223: \begin{center}

1224: Rational double points are the simplest surface singularities. In

1225: this essay we will be mainly concerned with the geometry of the

1226: exceptional set corresponding to the resolution of a rational

1227: double point. We will derive the classification of rational

1228: double points in terms of Dynkin diagrams.

1229: \end{center}

1230: 

1231: \end{abstract}

1232: 

1233: \end{titlepage}

1234: 

1235: \goodbreak

1236: 

1237: \tableofcontents

1238: 

1239: \goodbreak

1240: 

1241: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1242: \section{Introduction}

1243: 

1244: Rational double points\footnote{In the literature, they are also

1245: called Du Val singularities, Kleinian singularities or simple

1246: critical points. } are the simplest surface singularities and were

1247: first studied by Du Val \cite{DV}. One may think of them as

1248: neglible singularities. They play an important r\^{o}le in the

1249: classification of surfaces and occur in the theory of

1250: simultaneous resolutions of singularities.

1251: 

1252: In this essay we will be mainly concerned with the geometry of the

1253: exceptional set corresponding to a resolution of a rational

1254: double point. We will derive the classification of rational

1255: double points in terms of Dynkin diagrams. It should be noted,

1256: that the proof of this classification is rather lengthy. However,

1257: the author was unable to find the complete proof in a single

1258: source and therefore decided to present it in full detail. Most

1259: ideas are taken from two papers of Artin \cite{A2}, \cite{A1},

1260: balanced with a slightly different approach in Reid's draft

1261: \cite{Reid}. Further parts of the argument are taken from Durfee

1262: \cite{Df}, Mumford \cite{Mf} and Brieskorn \cite{B3}. The second

1263: article by Pinkham in \cite{DPT} treats the topic very nicely,

1264: although some difficult steps are omitted.

1265: 

1266: Finally, we will find a connection between the most simple objects

1267: in different fields of mathematics: Rational double points are

1268: linked with Platonic solids and simple Lie groups.

1269: 

1270: 

1271: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1272: \section{Basic facts on surface singularities}

1273: 

1274: 

1275: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1276: \subsection{Definitions}

1277: 

1278: We want to study {\em surface singularities} $(X,x)$; here $X$ is a nor\-mal, two-dimen\-sional, projective variety over $\mathbb C$ which is non-singular,

1279: except maybe at $x \in X$.

1280: 

1281: Two singularities are isomorphic, if there exist open neighbourhoods of the singular points which are isomorphic.

1282: 

1283: A {\em resolution} of $(X,x)$ is a birational, proper and

1284: surjective morphism

1285: 

1286: $$ \res $$

1287: 

1288: where $\widetilde{X}$ is a non-singular projective variety over \CC.

1289: 

1290: See section \ref{blowup} for an example.

1291: 

1292: It is an important and difficult theorem, that resolutions always exist; for a general discussion we refer to \cite{Lp}, \cite{Lf}.

1293: 

1294: \paragraph{Immediate properties of the exceptional set}

1295: The {\em exceptional set} $E:=\pi^{-1}(x)$ is compact (since $X$

1296: is proper) and one-dimensional (since $\pi$ is birational).

1297: Moreover it is connected by Zariski's connectedness theorem

1298: \ref{app5}. Therefore $E$ is a bunch of irreducible projective

1299: curves

1300: 

1301: $$E=\bigcup\limits_{i=1}^{n}E_i .$$

1302: 

1303: We say that a surface singularity is

1304: \begin{description}

1305: \item{\em rational}, if for a resolution

1306: $$ \res $$

1307: the first higher direct image sheaf of $\XX\text{'s}$ structure sheaf vanishes

1308: \begin{equation}

1309: \label{ratcond}

1310: R^1 \pi_{\ast} \Oxx = 0,

1311: \end{equation}

1312: and

1313: \item{\em a double point}, if the local ring $\OO _{X,x}$ has multiplicity two, i.e. the

1314: leading coefficient of its Hilbert-Samuel polynomial is two

1315: (\cite{Ng} III \S 23, \cite{ZS} vol. 2, VIII \S 10).

1316: \end{description}

1317: 

1318: \paragraph{Remarks:}

1319: \begin{enumerate}

1320: \item

1321: \begin{enumerate}

1322: \item The definition of a rational singularity is independent of the chosen resolution:

1323: Since $ R^1 \pi_{\ast} \Ox~ $ is a coherent sheaf (\cite{Ha} III.8.8.(b))

1324: concentrated on $x$,

1325: all we are interested in is $h^0(X,R^1 \pi_{\ast} \Oxx)$.

1326: However, we will see soon in section \ref{leray} that

1327: $$ p_a(X) - p_a(\widetilde{X}) = h^0(X,R^1 \pi_{\ast} \Oxx) $$

1328: and the arithmetic genus of a {\bf non-singular} projective variety is a birational invariant (\cite{Ha} V.5.6).

1329: 

1330: \item The condition (\ref{ratcond}) may appear opaque at a first glance, but will hopefully become more transparent in the sequel.

1331: For example, it implies that the $E_i$ are rational curves.

1332: 

1333: \end{enumerate}

1334: 

1335: \item Since we are in the normal case, the condition for a double point means, that two general curves on $X$ through $x$ have local intersection number

1336: two at $x$ (\cite{Reid} 4.6). If $X$ is a hypersurface $f^{-1}(0)$, yet another way to state this condition is

1337: 

1338: $$ f \in m_x^2 \text{ and } f \notin m_x^3 ,$$

1339: 

1340: where $m_x$ is the ideal of functions vanishing at $x$ (\cite{Dm} (7.48)).

1341: \end{enumerate}

1342: 

1343: 

1344: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1345: \subsection{A first consequence of the rationality condition (\ref{ratcond})}

1346: Let us mention a simple consequence of the rationality condition (\ref{ratcond}).

1347: 

1348: \begin{proposition}

1349: Let $\res$ be a resolution as above. If

1350: $$ R^1 \pi_{\ast} \Oxx = 0 $$

1351: then

1352: $$ p_a(X) = p_a(\widetilde{X}) . $$

1353: \end{proposition}

1354: 

1355: We need a lemma.

1356: 

1357: \begin{lemma}

1358: For any resolution $\res$, we have

1359: $$ \pi_{\ast} \Oxx = \Ox . $$

1360: \end{lemma}

1361: 

1362: 

1363: {\bf Proof} (\cite{Ha} p. 280):

1364: Since the question is local on $X$, we can assume $X$ is affine, say $X=\text{Spec } A$.

1365: By \cite{Ha} II.5.8.(b), $\pi_{\ast} \Oxx$ is a coherent sheaf of $\Ox$-algebras, hence $B:=H^0(X,\pi_{\ast} \Oxx)$ is a finitely generated $A$-module.

1366: But $A$ and $B$ are integral domains with the same quotient field (since $\pi$ is birational)

1367: and $A$ is algebraically closed (since $X$ is normal), thus $B=A$, and $ \pi_{\ast} \Oxx = \Ox$. \hfill $\Box$

1368: 

1369: \vspace{3mm} {\bf Proof of the proposition} (\cite{Ha} Ex. III.8.1):

1370: Let

1371: \begin{equation}

1372: \label{injres}

1373: 0 \longrightarrow I^0 \stackrel{d^0}{\longrightarrow} I^1 \stackrel{d^1}{\longrightarrow} I^2 \longrightarrow \dots

1374: \end{equation}

1375: be an injective resolution for $\Oxx$. We have not only $R^1

1376: \pi_{\ast} \Oxx = 0$, but $R^i \pi_{\ast} \Oxx = 0$ for $i \ge 1$,

1377: because the fibers of $\pi$ have dimension $\le 1$  (\ref{app7}).

1378: Therefore, by applying $\pi_{\ast}$ to (\ref{injres}), we obtain

1379: again an exact sequence

1380: \begin{equation}

1381: \label{acres}

1382: 0 \longrightarrow \pi_{\ast}\Oxx \longrightarrow \pi_{\ast}I^0 \stackrel{\pi_{\ast}d^0}{\longrightarrow} \pi_{\ast}I^1 \stackrel{\pi_{\ast}d^1}

1383: {\longrightarrow} \pi_{\ast}I^2 \longrightarrow \dots .

1384: \end{equation}

1385: Since injectives are flasque (\cite{Ha} III.2.4), direct images of flasque sheaves are flasque, and flasque sheaves are acyclic for the global section

1386: functor (\cite{Ha} III.2.5), we see that (\ref{acres}) is an acyclic resolution for $\Ox = \pi_{\ast} \Oxx$.

1387: Thus

1388: \begin{eqnarray*}

1389: H^i(X,\Ox) & = & \frac{\text{ker }H^0(X,\pi_{\ast}d^i)}{\text{im }H^0(X,\pi_{\ast}d^{i-1})} \\

1390: & = & \frac{\text{ker }H^0(\widetilde{X},d^i)}{\text{im }H^0(\widetilde{X},d^{i-1})} \\

1391: & = & H^i(\widetilde{X},\Oxx)

1392: \end{eqnarray*}

1393: and our claim follows. \hfill $\Box$

1394: 

1395: \paragraph{Remark:}

1396: \label{leray}

1397: This proof is just a degenerated case of the {\em Leray spectral sequence}

1398: \begin{eqnarray*}

1399: && E_2^{p,q} = H^p(X,R^q\pi_{\ast}\Oxx) \\

1400: && \Rightarrow \\

1401: && E_{\infty}^{p,q}=H^{p+q}(\widetilde{X},\Oxx)

1402: \end{eqnarray*}

1403: (cf. \cite{Gd} II.4.17.1, \cite{Wb} V; for general information on spectral sequences cf. \cite{McC}, \cite{Sh} II \S  4),

1404: which takes in our setting the simple form (theorem \ref{app6} and \ref{app7})

1405: 

1406: \goodbreak

1407: 

1408: $$ E_2^{p,q}:

1409: \begin{array}{cccc}

1410: 0 & 0 & 0 & 0 \\

1411: H^0(X,R^1\pi_{\ast}\Oxx) & ? & ? & 0 \\

1412: H^0(X,\Ox) & H^1(X,\Ox)  & H^2(X,\Ox) & 0

1413: \end{array}

1414: $$

1415: \nopagebreak[4] \hrulefill \nopagebreak[4]

1416: \begin{eqnarray*}

1417: & H^i(\widetilde{X},\Oxx)=E_{\infty}^{i,0}=E_2^{i,0}=H^i(X,\Ox) \text{ for } i=0,1, \\

1418: & H^2(\widetilde{X},\Oxx)=E_{\infty}^{2,0}=E_2^{2,0}/E_2^{0,1}=H^2(X,\Ox)/H^0(X,R^1\pi_{\ast}\Oxx).

1419: \end{eqnarray*}

1420: From this we get

1421: $$ p_a(X) - p_a(\widetilde{X}) = h^0(X,R^1 \pi_{\ast} \Oxx) $$

1422: and our proposition (and its converse!) follow at once.

1423: 

1424: 

1425: % -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1426: \subsection{Further properties of the exceptional set $E$}

1427: 

1428: It will be a great technical convenience to work with {\em good resolutions}; for them we require that

1429: 

1430: \begin{enumerate}

1431: 

1432: \item all $E_i$ are non-singular,

1433: 

1434: \item \label{goodcond2}

1435: $E_i \cap E_j \cap E_k = \varnothing $ for mutually distinct $i, j, k$,

1436: 

1437: \item the intersection of $E_i$ and $E_j$ is transverse for $i \ne j$.

1438: 

1439: \end{enumerate}

1440: 

1441: Any resolution $\res$ of a {\bf surface} $X$ can be brought in

1442: such a nice form by successively blowing up points of

1443: $\widetilde{X}$ (cf. again \cite{Lp}, \cite{Lf}, and also

1444: \cite{Ha} V.3.8, V.3.9).

1445: 

1446: In the following we do always assume that

1447: $ \res $

1448: is good.

1449: 

1450: A fundamental fact about good resolutions is the following

1451: 

1452: \begin{proposition} {\rm (\cite{Mf} p. 6)}

1453: \label{negdef}

1454: The {\em intersection matrix of the resolution} $\left( E_i \cdot E_j \right)_{i,j=1 \dots n}$ is negative definite.

1455: \end{proposition}

1456: 

1457: {\bf Proof}:

1458: We take a meromorphic function $f \in k(X)$ with $f(x)=0$ and define two effective divisors

1459: 

1460: $$H_0:=f^{-1}(0) \text{ and } H_{\infty}:=f^{-1}(\infty) . $$

1461: 

1462: Denote the proper transform of $H_i$ with $\widetilde{H}_i$ for $i=0, \infty$ respectively.

1463: Then we have a linear equivalence of divisors

1464: 

1465: $$ \widetilde{H}_{\infty} \sim \widetilde{H}_0 + \sum\limits_{i=1}^n m_i E_i$$

1466: 

1467: where $m_i=\text{ord} _{E_i} f \circ \pi > 0$.

1468: 

1469: It suffices to prove that the matrix

1470: 

1471: $$M:=\left(m_i E_i \cdot m_j E_j\right)_{i,j=1, \dots ,n}$$

1472: 

1473: is negative definite. Now we have

1474: $M_{i,j} \ge 0$ if $i \ne j$ (since the $E_i$ are irreducible)

1475: and

1476: \begin{eqnarray*}

1477: \sum\limits_{i=1}^n M_{i,j} & = & \sum\limits_{i=1}^n (m_i E_i \cdot m_j E_j) \\

1478: & = & (\widetilde{H}_{\infty} - \widetilde{H}_0) \cdot m_j E_j \\

1479: & = & 0 - \widetilde{H}_0 \cdot m_j E_j \le 0 .

1480: \end{eqnarray*}

1481: 

1482: This implies that $M$ is negative semi-definite:

1483: 

1484: \begin{eqnarray}

1485: \sum\limits_{i,j=1}^n a_i a_j M_{i,j} & = & \sum\limits_{i=1}^n a_i^2 M_{i,i} + 2\sum\limits_{i,j=1 \atop i<j}^n a_i a_j M_{i,j} \nonumber\\

1486: & = & \sum\limits_{j=1}^n

1487: \underbrace{

1488: \left(

1489: \sum\limits_{i=1}^n M_{i,j}

1490: \right)

1491: }_{\le 0}

1492: a_j^2

1493: \label{mf1}

1494: - \sum\limits_{i,j=1 \atop i<j}^n \underbrace{M_{i,j} (a_i-a_j)^2}_{\ge 0} \le 0 .

1495: \end{eqnarray}

1496: 

1497: To show definiteness, we note that $\widetilde{H}_0$ must pass through some $E_i$, hence

1498: $$ \sum \limits_{i=1}^n M_{i,j_0} < 0 \text{ for some } j_0 . $$

1499: Suppose we have equality in (\ref{mf1}). Then $a_{j_0}=0$.

1500: Furthermore, we get $a_i=a_j$ if $M_{i,j} > 0$, or inductively $a_i=a_j$ if $E_i$ and $E_j$

1501: are connected in $E$. But $E$ is connected, hence $a_i=0$ for $i=1, \dots , n$ in this case. \hfill $\Box$

1502: 

1503: \vspace{3mm}

1504: In the proof of proposition \ref{negdef}, we encountered an effective {\em exceptional divisor} (i.e. an divisor supported on $E$)

1505: $$ Z:=\widetilde{H}_{\infty}-\widetilde{H}_0=\sum \limits_{i=1}^n m_i E_i > 0 $$

1506: which had the note-worthy property

1507: \begin{equation}

1508: \label{numcond}

1509: Z \cdot E_i \le 0 \text{ for } i=1, \dots , n.

1510: \end{equation}

1511: 

1512: Since $E$ is connected, any exceptional divisor $Z$ with this property (\ref{numcond}) must satisfy $Z \ge E$, by the arguments used in that proof. If two

1513: exceptional divisors

1514: $Z^1=\sum \limits_{i=1}^n r_i^1 E_i > 0$ and $Z^2=\sum \limits_{i=1}^n r_i^2 E_i > 0$ both satisfy (\ref{numcond}) then obviously so does

1515: $Z:=\min(Z^1,Z^2)=\sum \limits_{i=1}^n r_i E_i > 0$ where $r_i:=\min(r_i^1,r_i^2)$:

1516: $

1517: Z \cdot E_i \le Z^j \cdot E_i \le 0$ whenever $r_i=r_i^j

1518: $.

1519: Hence there exists a minimal positive exceptional divisor, called

1520: the {\em numerical divisor} $\Znum$ (\cite{Reid} 4.5) (also

1521: called {\em fundamental divisor} \cite{A1}), for which

1522: (\ref{numcond}) holds.

1523: 

1524: This divisor $\Znum$ will provide a useful tool to describe the exceptional set of a rational singularity.

1525: 

1526: 

1527: % ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1528: \section{The geometry of the exceptional set $E$ of a resolution of a rational singularity}

1529: 

1530: Throughout this section, we assume that \res\ is good resolution

1531: of a rational singularity $(X,x)$ and $E$ its exceptional set.

1532: 

1533: We will prove that the $E_i$ are rational curves $E_i \cong {\mathbb P}^1$.

1534: Moreover, we will be able to read off the multiplicity of $(X,x)$ as the self-intersection-number $-(\Znum)^2$.

1535: 

1536: The idea is to study fatter and fatter infinitesimal neighbourhoods of $E$ in order to examine the embedding of $E$ in $\widetilde{X}$.

1537: 

1538: We will identify an exceptional divisor $Z=\sum\limits_{i=1}^n r_i E_i$ with its associated {\em positive cycle}: this is the, generally {\bf non-reduced},

1539: scheme $(\text{Supp } Z, \OO _Z)$.

1540: Recall that $\OO_Z = \text{coker} (\Oxx (-Z) \rightarrow \Oxx )$, i.e. $(\text{Supp } Z, \OO _Z)$ is the subscheme of $\widetilde{X}$ defined by the coherent

1541: sheaf of ideals on $\widetilde{X}$ whose sections on an open $U \subset \widetilde{X}$ are the

1542:  rational functions $f \in \Gamma(U,\Oxx)$ which have zeros of order at

1543: least $r_i$ along $E_i$ for all $i$ with $E_i \cap U \neq \varnothing $ \cite{A2}.

1544: Note $\text{Supp } Z = \bigcup\limits_{r_i > 0} E_i$.

1545: 

1546: 

1547: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1548: \subsection{The exceptional curves $E_i$ are rational}

1549: 

1550: \begin{theorem}

1551: \label{theo1}

1552: {\rm (\cite{A1} prop. 1, \cite{B3} Lemma 1.3)}

1553: The exceptional set of a good resolution of a rational singularity

1554: consists of rational projective curves $E_i \cong {\mathbb P}^1$.

1555: \end{theorem}

1556: 

1557: {\bf Proof} (\cite{A1}, \cite{A2}):

1558: 

1559: The proof relies on Grothendieck's theorem on formal functions \ref{app3}, which takes in our case the form

1560: 

1561: $$ 0 = \left( \left( R^1 \pi_{\ast} \Oxx \right)_x \right)\widehat{}\ \cong\

1562:   \mathop{\varprojlim}_{k=1}^{\infty} H^1\left( E, \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^k} \right) $$

1563: 

1564: where $m_x \subset \OO_{X,x}$ is the maximal ideal corresponding to $x$ and completion is taken with respect to the $m_x$-adic topology.

1565: 

1566: (We will see later in lemma \ref{mznum}, that $\Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^k } = \OO_{k \Znum}$.)

1567: 

1568: Since $E$ is one-dimensional, the natural map

1569: $$ \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^{k+1} } \twoheadrightarrow \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^k }$$

1570: induces a surjection on cohomology

1571: $$ H^1\left( E, \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^{k+1} } \right) \twoheadrightarrow

1572:   H^1\left( E, \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^k } \right)$$

1573: by a vanishing theorem of Grothendieck \ref{app6}.

1574: Thus we see

1575: $$ H^1 \left( E, \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^{k+1} } \right) = 0 \text{ for all } k \in {\mathbb N}. $$

1576: 

1577: We denote the sheaf of ideals of functions vanishing at $x$ by $\mm _x$.

1578: Clearly every function in $\mm _x \cdot \Oxx$ vanishes on $E$; hence for every positive cycle $Z$ we can find an integer $k$ such that every

1579: function in ${\mm _x}^k \cdot \Oxx$ vanishes on $Z$. We now have

1580: $$ \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^k } \twoheadrightarrow \OO _Z $$

1581: and (\ref{app6})

1582: $$ 0= H^1 \left( E, \Oxx \otimes _{\Ox} {\OO_{X,x} / {m_x}^{k+1} } \right) \twoheadrightarrow H^1(E, \OO _Z). $$

1583: 

1584: In particular $H^1(E, \OO _{E_i}) =0$ for $i=1,\dots,n$, from which we conclude

1585: $p_a(E_i) \nolinebreak = \nolinebreak 0$, i.e. $E_i \cong {\mathbb P}^1$. \hfill $\Box$

1586: 

1587: \begin{corollary}

1588: \label{H10}

1589: In the proof of theorem \ref{theo1} we have just seen $H^1(E, \OO _Z)=0$ for every positive cycle $Z$.

1590: \end{corollary}

1591: 

1592: We can make a more precise statement for the numerical divisor \Znum.

1593: 

1594: \begin{corollary}

1595: \label{znum0}

1596: {\rm (\cite{A1} thm. 3)} With the assumptions of the

1597: theorem we have $$ p_a(\Znum) =0 . $$

1598: \end{corollary}

1599: 

1600: {\bf Proof} (\cite{Reid} 3.11):

1601: The statement follows immediately from corollary \ref{znum0}

1602: and the general fact

1603: \fbox{$h^0(E,\OO_{\Znum}) = 1$}.

1604: This can be proved easily by induction:

1605: We know $h^0(E)=1$. Assume $h^0(Y)=1$ for a positive cycle

1606: $E \le Y \lneq \Znum$.

1607: We have

1608: $$ Y \cdot E_i > 0 \text{ for some } i, $$

1609: or equivalently $\text{deg}_{E_i} \OO_E(-Y) \le -1$,

1610: by the very definition of \Znum. Certainly $Y+E_i \le \Znum$ in this situation.

1611: From

1612: $$

1613: 0=H^0(E_i,\OO_{E_i}(-Y)) \rightarrow H^0(E,\OO_{Y+E_i}) \rightarrow H^0(E,\OO_Y)

1614: $$

1615: we conclude

1616: $h^0(E,\OO_{Y+E_i})=1$. \hfill $\Box$

1617: 

1618: 

1619: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1620: \subsection{A criterion for rationality}

1621: 

1622: \begin{theorem}

1623: \label{theo2}

1624: {\rm (\cite{A1}, thm. 3)}

1625: 

1626: Conversely, if we have

1627: 

1628: $$p_a(\Znum) = 0$$

1629: 

1630: for the numerical cycle of a good resolution of a singularity $(X,x)$, then $(X,x)$ is rational.

1631: \end{theorem}

1632: 

1633: We need the following lemma.

1634: 

1635: \begin{lemma}

1636: Let $Z=\sum\limits_{i=1}^n r_i E_i$ be a positive cycle with the property that

1637: $p_a(Y) \le 0$ for all positive cycles $Y \le Z$. Then $H^1(E, \OO _Z)=0$.

1638: \end{lemma}

1639: 

1640: {\bf Proof}:

1641: In particular $p_a(E_i)=0$ for all $i$ with $r_i \ge 1$, i.e. $E_i \cong {\mathbb P}^1$.

1642: We use induction on $\sum\limits_{i=1}^n r_i$. Assume $H^1(E,\OO_Z) \neq 0$.

1643: Let $Z_i:=Z-E_i$ for $r_i \ge 1$. By induction hypothesis

1644: $$H^1(E, \OO _{Z_i})=0 , $$

1645: hence for the kernel $M$

1646: $$ 0 \rightarrow M \rightarrow \OO _Z \rightarrow \OO _{Z_i} \rightarrow 0 $$

1647: we get (\ref{app6})

1648: $$ H^1(E,M) \rightarrow H^1(E, \OO _Z) \rightarrow 0 , $$

1649: i.e. $H^1(E,M) \neq 0$.

1650: By the snake-lemma

1651: $$

1652: \begin{CD}

1653: {}   @.   {}   @.   0   @>>>   M   @.   {} \\

1654: @.        @.         @VVV        @VVV    @. \\

1655: 0 @>>> \Oxx(-Z) @>>> \Oxx @>>> \OO _Z @>>> 0 \\

1656: @.        @VVV    @VVV        @VVV    @. \\

1657: 0 @>>> \Oxx(-Z_i) @>>> \Oxx @>>> \OO _{Z_i} @>>> 0 \\

1658: @.        @VVV    @VVV       @.          @. \\

1659: {} @.  \Oxx(-Z_i) \otimes \OO _{E_i} @>>> 0 @. {} @. {}

1660: \end{CD}

1661: $$

1662: 

1663: we deduce $M  \cong \Oxx(-Z_i) \otimes \OO _{E_i}$. Hence we can write

1664: $$ 0 \neq H^1(E, \Oxx (-Z_i) \otimes \OO _{E_i}) = H^1(E_i,\Oxx(-Z_i) \otimes \OO _{E_i}) . $$

1665: But $E_i$ is just ${\mathbb P}^1$, thus (\cite{Ha} III.5.1)

1666: $$ \text{deg } \Oxx(-Z_i) \otimes \OO _{E_i} \le -2 . $$

1667: On the other hand

1668: $$ \text{deg } \Oxx(-Z_i) \otimes \OO _{E_i} = - Z_i \cdot E_i$$

1669: and by the adjunction formula \ref{app2} we get

1670: $$ Z \cdot E_i = (Z_i + E_i) \cdot E_i \ge 2 + {E_i}^2 = -K \cdot E_i. $$

1671: Summing up $(Z+K) \cdot E_i \ge 0$ yields with the adjunction formula \ref{app2}

1672: $$ 2 p_a(Z) - 2 = ( Z+K) \cdot Z \ge 0, $$

1673: i.e. $p_a(Z) \ge 1$, a contradiction. \hfill $\Box$

1674: 

1675: \vspace{3mm}

1676: 

1677: {\bf Proof of the theorem} (\cite{A1} prop. 1, thm. 3):

1678: We have seen in the proof of theorem \ref{theo1}, that by Grothendieck's theorem on formal functions (\ref{app3})

1679: $$ 0 = \left( \left( R^1 \pi_{\ast} \Oxx \right)_x \right)\widehat{}\ \cong\

1680: \mathop{\varprojlim}_{k=1}^{\infty} H^1( E, \OO _{k\Znum}), $$

1681: hence it is sufficient to prove

1682: $$ H^1(E, \OO _{k \Znum}) = 0 \text{ for all } k. $$

1683: We already know $H^1(E, \OO _{\Znum})=0$ (since $p_a(\Znum )=0$ and $h^0(E,\OO_{\Znum})=1$).

1684: From the surjection (\ref{app6})

1685: $$

1686: H^1(E, \OO _{\Znum}) \twoheadrightarrow H^1(E, \OO _{E_i})

1687: $$

1688: we find that $p_a(E_i)=0$, i.e. $E_i \cong {\mathbb P}^1$.

1689: 

1690: By the lemma, it is enough to show $p_a(Y) \le 0$ for all positive cycles $Y$. Let $Y_1:=Y$ and define $Y_{n+1}$ inductively as follows

1691: 

1692: \begin{enumerate}

1693: \item if $Y_n \ge \Znum$, then $Y_{n+1}:=Y_n-\Znum \ge 0$.

1694: \item if $Y_n \not\ge \Znum$, then $Y_n \cdot E_i > 0$ for some $i$ by the definition of \Znum. Choose such an $i$ with smallest possible multiplicity in

1695: $Y_n$ and set $Y_{n+1}:=Y_n + E_i$.

1696: \end{enumerate}

1697: Stop when $Y_n=0$. We use the equation (\cite{Ha} Ex. V.1.3)

1698: \begin{equation}

1699: \label{lichtenbaum}

1700: p_a(Z_1 + Z_2) = p_a(Z_1) + p_a(Z_2) + Z_1 \cdot Z_1 -1

1701: \end{equation}

1702: to calculate the arithmetic genus:

1703: 

1704: In case 1:

1705: \begin{eqnarray*}

1706: p_a(Y_n) & = & p_a(Y_{n+1}+\Znum) \\

1707: & = & p_a(Y_{n+1}) + p_a(\Znum) + Y_{n+1} \cdot \Znum -1 \\

1708: & \le & p_a(Y_{n+1})-1

1709: \end{eqnarray*}

1710: 

1711: In case 2:

1712: \begin{eqnarray*}

1713: p_a(Y_{n+1}) & = & p_a(Y_n) + p_a(E_i) + Y_n \cdot E_i -1 \\

1714: & \ge & p_a(Y_n).

1715: \end{eqnarray*}

1716: 

1717: Steps of type $(2)$ cannot be repeated infinitely often without reaching a stage where $Y_n \ge \Znum$. Using equation (\ref{lichtenbaum}) once again, we see

1718: that $p_a(Y)$ is a quadratic form in the coefficients $s_i$ of $Y=\sum\limits_{i=1}^n s_i E_i$ whose quadratic term is

1719: $\frac{1}{2} \sum\limits_{i,j=1}^n s_i s_j E_i \cdot E_j$. But the matrix $(E_i \cdot E_j)_{i,j=1,\dots,n}$

1720: is negative definite, hence $p_a(Y)$ is bounded above.

1721: Consequently, there

1722: can be only a finite number of steps and the algorithm must terminate with $Y_n=0$. Then $Y_{n-1}=\Znum$ and we have

1723: $$ p_a(Y_1) \le \dots \le p_a(Y_{n-1})=0 . $$

1724: \hfill $ \Box $

1725: 

1726: 

1727: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1728: \subsection{Invertible sheaves on a positive cycle $Z$}

1729: 

1730: It is a natural question to ask what are the invertible sheaves on

1731: a positive cycle $Z$, i.e. what is $\text{Pic } Z$?

1732: 

1733: In our particular case the answer is quite simple. It will provide an important tool in exploring the geometry of $E$ further.

1734: 

1735: For any invertible sheaf $\mathcal F$ on the positive cycle $Z \ge E$, we can define its {\em multidegree}

1736: 

1737: $$ \text{deg}_Z: \text{Pic }Z \rightarrow {\mathbb Z}^n $$

1738: 

1739: via the composite maps

1740: $$

1741: \text{Pic } Z \rightarrow \text{Pic } E_i \stackrel{\text{deg}}{\longrightarrow} {\mathbb Z}

1742: \text{ for } i=1, \dots , n .

1743: $$

1744: 

1745: Using {\em local transverse cuts} it is easily seen that this map is surjective

1746: $ \text{deg}_Z \nolinebreak \text{Pic } \nolinebreak Z \nolinebreak \twoheadrightarrow \nolinebreak {\mathbb Z}^n $:

1747: Choose a general point $p$ on any $E_i$ and construct a Cartier divisor

1748: $\{ (U_j, f_j ) \}$ with support

1749: $p$

1750: and degree $1$ on $E_i$ whose local equation $s \in \OO _{Z,p}$ restricts to a local equation of $p$ in $\OO _{E_i, p}$. This gives

1751: $$ \text{deg} _Z \{ (U_j, f_j ) \} = (0, \dots ,0 , \mathop{1}_i, 0, \dots ,0 ) . $$

1752: 

1753: In fact, we will prove that $ \text{deg}_Z \text{Pic } Z \twoheadrightarrow {\mathbb Z}^n $ is an isomorphism.

1754: 

1755: It is a well-known fact that (\cite{Ha} Ex. III.4.5) $$ \text{Pic

1756: } Z = H^1(E, \OO _Z ^{\ast} ). $$ (One may think of a

1757: \v{C}hech-$1$-cocycle $\{ ( U_i \cap U_j , g_{i,j} ) \} \in H^1(E,

1758: \OO_Z^{\ast})$ as a set of transition functions $g_{i,j} :

1759: \OO_{U_i \cap U_j} \stackrel{\sim}{\longrightarrow} \OO_{U_i \cap

1760: U_j} $ which define a line bundle on $Z$.)

1761: 

1762: In the reduced case $Z=E$, it is easy to see what $H^1(E, \OO_E^{\ast})$ is, if

1763: we allow transcendental methods. Let $\circ_h$ denote the functor

1764: form the category of schemes of finite type over \CC\ to the category of complex analytic spaces.

1765: (cf. \cite{Ha} B and section \ref{complexana}).

1766: 

1767: Since $E$ is projective over \CC, a theorem by Serre (\cite{Ha}

1768: B.2.1) tells us that

1769: $$

1770: H^i(E,{\mathcal F}) \cong H^i(E_h,{\mathcal F}_h)

1771: $$

1772: for every coherent sheaf $\mathcal F$ on $E$.

1773: The exponential sequence (\cite{Ha} V.5)

1774: $$

1775: 0 \rightarrow \ZZ \rightarrow \OO_{E_h} \stackrel{\text{exp} 2 \pi i }{\longrightarrow} \OO_{E_h}^{\ast} \rightarrow 0

1776: $$

1777: yields (corollary \ref{H10} and theorem \ref{app6})

1778: 

1779: $$ 0 \rightarrow H^1(E_h, \OO_{E_h}^{\ast}) \rightarrow H^2(E_h, \ZZ) \rightarrow 0, $$

1780: 

1781: i.e.

1782: 

1783: $$ \text{Pic } E \cong H^2(E_h, \ZZ). $$

1784: 

1785: As we have already seen, E is built up out of $n$ spheres $S^2

1786: \cong \PP$, which intersect each other transversely. Hence by the

1787: Mayer-Vietoris sequence for, say singular cohomology

1788: 

1789: $$ H^2(E_h, \ZZ) \cong \ZZ^n . $$

1790: 

1791: (We will see later, that $E$ has the homotopy type of a bouquet of $n$ spheres $E \simeq (S^2)^{\vee n}$.)

1792: Therefore, we obtain

1793: \begin{equation*}

1794: \text{Pic } E \cong \ZZ^n .

1795: \end{equation*}

1796: 

1797: Unfortunately, there is no analogue of the exponential sequence in

1798: the non-reduced case. Instead, we need the following proposition

1799: by Artin, whose proof uses a "first order exponential".

1800: 

1801: \begin{proposition}

1802: \label{picz}

1803: {\rm (\cite{A2} lemma 1.4)}

1804: We have

1805: $$

1806: H^1(E,\OO_Z) \cong H^1(E,\OO_E) \cong {\mathbb Z}^n

1807: $$

1808: for every positive cycle $Z \ge E$.

1809: \end{proposition}

1810: 

1811: {\bf Proof}: We will proceed by induction: The case $Z=E$ is

1812: trivial, so assume the proposition holds for

1813: $Z^{\prime} =  Z - E_i \ge E$. We fix our notation for the

1814: following kernels

1815: 

1816: \newcommand{\Nc}{\mathcal N}

1817: \newcommand{\Mc}{\mathcal M}

1818: \newcommand{\Ic}{\mathcal J}

1819: \newcommand{\Kc}{\mathcal K}

1820: 

1821: \begin{eqnarray*}

1822: & 0 \rightarrow \Nc \rightarrow \OO_Z \rightarrow

1823: \OO_E \rightarrow 0 , \\

1824: & 0 \rightarrow \Mc \rightarrow \OO_Z^{\ast} \rightarrow

1825: \OO_E^{\ast} \rightarrow 0 , \\

1826: & 0 \rightarrow \Nc ^{\prime} \rightarrow \OO_{Z^{\prime}} \rightarrow

1827: \OO_E \rightarrow 0 , \\

1828: & 0 \rightarrow \Mc ^{\prime} \rightarrow \OO_{Z^{\prime}}^{\ast} \rightarrow

1829: \OO_E^{\ast} \rightarrow 0 , \\

1830: & 0 \rightarrow \Ic \rightarrow \OO_Z \rightarrow

1831: \OO_{Z^{\prime}} \rightarrow 0 & \text{ and } \\

1832: & 0 \rightarrow \Kc \rightarrow \OO_Z^{\ast} \rightarrow

1833: \OO_{Z^{\prime}}^{\ast} \rightarrow 0 .

1834: \end{eqnarray*}

1835: By \ref{app6}, it suffices to prove $H^1(E,\Mc)=0$.

1836: 

1837: Note that $H^0(E,\OO_E) = \CC$ (and also

1838: $H^0(E,\OO_E^{\ast})=\CC^{\ast}$), since $E$ is connected. In

1839: particular, we get a surjection

1840: $$

1841: H^0(E,\OO_Z) \twoheadrightarrow H^0(E,\OO_E) ,

1842: $$

1843: which implies (corollary \ref{H10})

1844: $$

1845: H^1(E,\Nc)=0 .

1846: $$

1847: Similarly, $H^1(E,\Mc ^{\prime})=0$ (using the induction

1848: hypothesis). Now, these kernels are linked by the short exact

1849: sequences

1850: \begin{eqnarray*}

1851: & 0 \rightarrow \Ic \rightarrow \Nc \rightarrow \Nc ^{\prime}

1852: \rightarrow 0 \\

1853: & 0 \rightarrow \Kc \rightarrow \Mc \rightarrow \Mc ^{\prime}

1854: \rightarrow 0

1855: \end{eqnarray*}

1856: and we obtain

1857: \begin{eqnarray*}

1858: & H^0(E,\Nc ^{\prime}) \stackrel{\delta}{\longrightarrow}

1859: H^1(E,\Ic) \rightarrow 0 \\

1860: & H^0(E,\Mc ^{\prime})

1861: \stackrel{\delta ^{\prime}}{\longrightarrow}

1862: H^1(E,\Kc) \rightarrow H^1(E, \Mc) \rightarrow 0 .

1863: \end{eqnarray*}

1864: 

1865: Because of $\Ic \cdot \Nc = 0$ (thus $\Ic ^2 = 0$), we have an

1866: isomorphism

1867: $$

1868: \epsilon : \Ic \stackrel{\sim}{\longrightarrow} \Kc

1869: $$

1870: via

1871: $$

1872: s \in \Gamma(U,\Ic) \mapsto 1+s \in \Gamma(U,\Kc),

1873: $$

1874: hence $H^1(E,\Ic) \cong H^1(E, \Kc)$.

1875: Analogously, we have a bijection (not a morphism, in general!)

1876: \begin{eqnarray*}

1877: \epsilon ^{\prime} : H^0(E, \Nc ^{\prime})

1878: & \rightarrow &

1879: H^0(E, \Mc ^{\prime}) \\

1880: s^{\prime} & \mapsto & 1+s^{\prime} .

1881: \end{eqnarray*}

1882: Therefore, it suffices to show that the following diagram commutes

1883: $$

1884: \begin{CD}

1885: H^0(E,\Nc ^{\prime}) @>{\delta}>> H^1(E, \Ic) \\

1886: @V{\epsilon^{\prime}}VV @VV{\epsilon}V \\

1887: H^0(E,\Mc ^{\prime}) @>{\delta^{\prime}}>> H^1(E, \Kc) .

1888: \end{CD}

1889: $$

1890: 

1891: Pick an element $s^{\prime} \in H^0(E,\Nc ^{\prime})$ and choose

1892: an open covering $\{U_i\}$ of $E$ such that $s^{\prime}$ can be

1893: lifted to $s_i \in \Gamma(U_i,\Nc)$. Now we can write

1894: $\delta(s^{\prime})$ as the \v{C}ech-$1$-cocycle

1895: $$

1896: \{(U_i \cap U_j, s_i - s_j )\} \in H^1(E, \Ic)

1897: $$

1898: and get

1899: $$

1900: \epsilon(\delta(s^{\prime})) =

1901: \{(U_i \cap U_j, 1+ s_i - s_j )\} \in H^1(E, \Kc) .

1902: $$

1903: In the same way, we can lift

1904: $\epsilon^{\prime}(s^{\prime})=1+s^{\prime}$

1905: to $1+s_i \in \Gamma(U_i, \Mc)$ and obtain

1906: $$

1907: \epsilon^{\prime}(\delta^{\prime}(s^{\prime})) =

1908: \{(U_i \cap U_j, \frac{1 + s_i}{1 + s_j} )\} \in H^1(E, \Kc) .

1909: $$

1910: We use $\Ic \cdot \Nc$ in order to show

1911: $$

1912: \epsilon(\delta(s^{\prime})) =

1913: \epsilon^{\prime}(\delta^{\prime}(s^{\prime})) .

1914: $$

1915: Since

1916: $$

1917: s_i - s_j \in \Gamma(U_i \cap U_j, \Ic)

1918: $$

1919: we get

1920: $$

1921: s_j(s_i - s_j) = 0 ,

1922: $$

1923: hence

1924: $$

1925: (1+s_j)(1+s_i-s_j)=1+s_i,

1926: $$

1927: i.e.

1928: $$

1929: 1+s_i-s_j = \frac{1+s_i}{1+s_j}.

1930: $$

1931: This finishes the proof. \hfill $\Box$

1932: 

1933: 

1934: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1935: \subsection{The multiplicity of a rational singularity}

1936: 

1937: The following theorem is the main result of this section.

1938: 

1939: \begin{theorem}

1940: \label{theo3} {\rm (\cite{A1} cor. 6)} The multiplicity of the

1941: rational singularity $(X,x)$ is equal to the negative of the

1942: self-intersection-number of the numerical cycle $-(\Znum)^2$.

1943: \end{theorem}

1944: 

1945: For the proof, we need two lemmas, which are interesting in their own rights.

1946: 

1947: \begin{lemma}

1948: \label{mznum} {\rm (\cite{A1} thm. 4)} We have

1949: 

1950: $$ \mm _x \cdot \Oxx = \Oxx(-\Znum) . $$

1951: \end{lemma}

1952: 

1953: {\bf Proof} (\cite{A1}, \cite{Reid} 4.17):

1954: 

1955: The inclusion $ \mm _x \cdot \Oxx \subseteq \Oxx(-\Znum)$

1956: is easy \cite{A1}: For $f \in \Gamma(U,\mm _x \cdot \Oxx)$ we can

1957: split the principal divisor $(f)$ in a part $Z$ supported on $E$

1958: and a part $D$, which does not involve any of the $E_i$ at all:

1959: $(f) = Z + D$. Obviously $Z > 0$. We have

1960: $$

1961: (f) \cdot E_i = 0 \text{ and } D \cdot E_i \ge 0 \text{ for } i=1, \dots , n,

1962: $$

1963: since $f$ is regular in a neighbourhood of $E$. Thus $Z \cdot E_i \le 0$ for $i=1, \dots, n$, that is $Z \ge \Znum$. Hence

1964: $f \in \Gamma(U,\Oxx(-\Znum))$.

1965: 

1966: For the other inclusion we have to show that for each point $p \in

1967: E$ there exists a local section $f$ of $\mm _x \cdot \Oxx$ such that $(f)_{|U} =

1968: {\Znum}_{|U} $ for a neighbourhood $U$ of $p$ (\cite{Reid} 4.17).

1969: 

1970: Let $X^{\prime}$ be an affine neighbourhood of $x \in X$ and set

1971: $\widetilde{X}^{\prime} := \widetilde{X} \times _X X^{\prime} = \pi^{-1}(X^{\prime})$.

1972: 

1973: We will write for short ${\II} := \OO_{\widetilde{X}^{\prime}}(-\Znum)$.

1974: 

1975: We can construct a divisor $A$ on \Znum\ as a sum of local transverse cuts such that $p \not\in \text{Supp} A$ and $\degz A = \degz \II_{|\Znum} $.

1976: The crucial point is, that proposition \ref{picz} implies now $\OO_{\Znum}(A) \cong \II_{|\Znum} $.

1977: Hence there exists a section $s \in H^0(E, \II_{|\Znum})$ which does not vanish at $p$.

1978: 

1979: To finish the proof, all we have to do is to lift $s$ to a section on $\widetilde{X}^{\prime}$.

1980: 

1981: From the short exact sequence

1982: 

1983: $$ 0 \rightarrow \II^{\otimes 2} \rightarrow \II \rightarrow \II_{|\Znum} \rightarrow 0 $$

1984: 

1985: (obtained from $ 0 \rightarrow \II \rightarrow \OO_{\widetilde{X}^{\prime}} \rightarrow \OO_{\Znum} \rightarrow 0 $ by tensoring with \II)

1986: we get

1987: 

1988: $$ H^0(\widetilde{X}^{\prime},\II) \rightarrow H^0(E,\II_{|\Znum}) \rightarrow H^1(\widetilde{X}^{\prime},\II^{\otimes 2}). $$

1989: 

1990: So it suffices to proof $H^1(\widetilde{X}^{\prime},\II^{\otimes 2}) =0$. We will prove more generally:

1991: 

1992: \subparagraph{Useful fact:} $H^1(\widetilde{X}^{\prime},\II^{\otimes k}) = 0 \text{ for all } k \in {\mathbb N}$.

1993: 

1994: {\bf Proof of the useful fact}: Since $X^{\prime}$ is affine

1995: 

1996: $$ H^1(\widetilde{X}^{\prime},\II^{\otimes k}) = H^0(X^{\prime},R^1 \pi_{\ast} \II^{\otimes k}) $$

1997: 

1998: by \cite{Ha} III.8.5.

1999: 

2000: The sheaf $R^1 \pi_{\ast} \II^{\otimes k}$ is concentrated in $x$; hence it is enough to prove

2001: $\left( R^1 \pi_{\ast} \II^{\otimes k} \right)_x = 0 $.

2002: By Grothendieck's theorem on formal functions (\ref{app3})

2003: 

2004: $$ \left( \left( R^1 \pi_{\ast} \II^{\otimes k} \right)_x \right) \widehat{}\, =

2005: \mathop{\varprojlim}_Z H^1(E, {\II^{\otimes k}}_{|Z} )$$

2006: 

2007: (cf. the proof of theorem \ref{theo1}), so we are left to show $H^1(E,{\II^{\otimes k}}_{|Z})=0$ for all positive cycles $Z$. Again, we can construct a divisor $A$ on $Z$

2008: as a sum of local transverse cuts such that $\text{deg}_Z A = \text{deg}_Z \II^{\otimes k}$.

2009: 

2010: From the short exact sequence

2011: 

2012: $$ 0 \rightarrow \OO_Z \rightarrow \OO_Z (A) \rightarrow \OO_A(A) \rightarrow 0 $$

2013: 

2014: we get

2015: 

2016: $$ H^1(E, \OO_Z) \rightarrow H^1(E, \OO_Z(A)) \rightarrow H^1(A,\OO_A(A)) . $$

2017: 

2018: We have $H^1(E, \OO_Z) = 0$ by corollary \ref{H10} and $H^1(A, \OO_A(A) ) = 0$ for dimension reasons \ref{app6},

2019: hence also $ H^1(E, \OO_Z(A)) = 0 $.

2020: \hfill $\blacksquare$

2021: 

2022: This shows

2023: 

2024: $$ H^0(\widetilde{X}^{\prime},\OO_{\widetilde{X}^{\prime}}(-\Znum)) \twoheadrightarrow H^0(E,\II_{|\Znum})$$

2025: 

2026: and we can fetch a preimage

2027: $ s^{\prime} \in H^0(\widetilde{X}^{\prime},\OO_{\widetilde{X}^{\prime}}(-\Znum))$ of $s$. By construction, ${s^{\prime}}_p \neq 0$, thus

2028: ${s^{\prime}}_q \neq 0$ for all $q \in U$ for some neighbourhood $U$ of $p$. Or put differently,

2029: 

2030: $$ (s^{\prime})_{|U} = {\Znum}_{|U} . $$

2031: 

2032: We have already seen $\pi_{\ast} \Oxx = \Ox$; therefore $s^{\prime}$ gives rise to a section in $\Gamma(\pi(U), \Ox)$,

2033: and thus in

2034: $\Gamma(\pi(U), \mm _x)$, since $\Znum \ge E$. \hfill $\Box$

2035: 

2036: 

2037: \begin{lemma}

2038: \label{gen} {\rm (\cite{Reid} 4.18)} The ring $\bigoplus_{k \ge 0}

2039: H^0(E,\II^{\otimes k})$ is generated in degree $1$, where

2040: $\II:=\OO_{\Znum}(-\Znum)$.

2041: \end{lemma}

2042: 

2043: {\bf Proof}:

2044: We can use local transverse cuts to construct a divisor $A$ on \Znum\ with $\OO_{\Znum}(A) \cong \II$. Therefore there exists a global section

2045: $s_0 \in H^0(E, \II)$ whose divisor of zeros is precisely $A$. Since we had a lot of freedom in choosing $A$, we see that the linear system

2046: $|A| = |\II |$ is basepoint-free.

2047: Thus we can choose a $s \in H^0(E,\II)$ such that $s$ provides a local base at every point $q \in A$.

2048: We can use $s^{k-1}$ to identify

2049: $\II^{\otimes k-1} \otimes \OO_A \cong \OO _A$.

2050: The short exact sequence

2051: 

2052: $$ 0 \rightarrow \OO_{\Znum} \rightarrow \II \rightarrow \OO _A \rightarrow 0 $$

2053: 

2054: yields (corollary \ref{H10})

2055: 

2056: \begin{equation}

2057: \label{ses1}

2058: 0 \rightarrow H^0(E,\OO _{\Znum}) \rightarrow H^0(E,\II) \rightarrow H^0(A,\OO _A) \rightarrow 0

2059: \end{equation}

2060: 

2061: and (\ref{app6})

2062: 

2063: $$ H^1(E, \II)=0 . $$

2064: 

2065: Let $s_1, \dots, s_d \in H^0(E, \II)$ map to a basis of $H^0(A,\OO _A)$, $d:=h^0(A,\OO _A)$.

2066: Our lemma will follow from the following claim:

2067: 

2068: \begin{eqnarray*}

2069: & H^1(E,\II ^{\otimes k})=0, \\

2070: & H^0(E,\II ^{\otimes k})=\text{span}_{\CC}\{{s_0}^k, {s_0}^{k-l}s^{l-1}s_i : 1 \le l \le k, 1 \le i \le d \}

2071: \end{eqnarray*}

2072: 

2073: Note that the sections

2074: $ {s_0}^{k-l}s^{l-1}s_i ,  l = 1, \dots,  k, i = 1, \dots, d$

2075: and ${s_0}^k$

2076: are linearly independent over \CC.

2077: 

2078: For $k=1$ our claim follows from (\ref{ses1}) and $h^0(E,\OO_{\Znum})=1$:

2079: $$

2080: h^0(E,\II)=h^0(E,\OO _{\Znum}) + h^0(A,\OO _A) = 1 + d.

2081: $$

2082: For the induction step we argue similarly using

2083: $$

2084: \begin{array}{ccccccc}

2085: 0 & \rightarrow & H^0(E,\II ^{\otimes k-1}) & \rightarrow & H^0(E,\II ^{\otimes k}) & \rightarrow & H^0(A,\OO_A) \\

2086: & \rightarrow & 0 & \rightarrow & H^1(E, \II ^{\otimes k}) & \rightarrow & 0

2087: \end{array}

2088: $$

2089: to get

2090: $$

2091: h^0(E,\II ^{\otimes k}) = h^0(E,\II ^{\otimes k-1}) + h^0(A,\OO _A) = (k-1)d+1+d

2092: $$

2093: and

2094: $$

2095: H^1(E,\II ^{\otimes k})=0 .

2096: $$

2097: \hfill $\Box$

2098: 

2099: {\bf Proof of the theorem} (\cite{Reid} 4.18): Let $X^{\prime}=\text{Spec }R$ be

2100: an affine neighbourhood of $x \in X$ and set

2101: $\widetilde{X}^{\prime}:=\widetilde{X} \times _X X^{\prime}$. By

2102: lemma \ref{mznum}, we know

2103: \begin{equation}

2104: \label{k1}

2105: H^0(\XXX, \mm _x \cdot \Oxx) = H^0(\XXX, \Oxx(-\Znum)) .

2106: \end{equation}

2107: 

2108: We want to generalize (\ref{k1}) to

2109: 

2110: \begin{equation}

2111: \label{kk}

2112: H^0(\XXX, {\mm _x }^k \cdot \Oxx) = H^0(\XXX, \Oxx(-k\Znum)) .

2113: \end{equation}

2114: 

2115: With (\ref{kk}) the proof of our assertion is straightforward:

2116: Since

2117: $$

2118: H^1(\XXX, \Oxx(-(k+1)\Znum))=0

2119: $$

2120: by corollary \ref{H10}, we have

2121: $$

2122: \frac{H^0(\XXX,{\mm _x} ^k \cdot \Oxx)}{H^0(\XXX,{\mm _x} ^{k+1} \cdot \Oxx)} =

2123: \frac{H^0(\XXX,\Oxx(-k\Znum))}{H^0(\XXX,\Oxx(-(k+1)\Znum))} =

2124: H^0(E, \OO _{\Znum} (-k\Znum)) .

2125: $$

2126: The Riemann-Roch theorem for curves tells us

2127: \begin{eqnarray*}

2128: h^0(E,\OO _{\Znum}(-k\Znum)) & = & 1 - p_a(\Znum) + \text{deg } \OO _{\Znum} (-k\Znum) \\

2129: & = & 1 - k (\Znum)^2,

2130: \end{eqnarray*}

2131: that is the leading coefficient of the Hilbert-Samuel polynomial of $(\OO _{X^{\prime} , x}, m_x)$ is $-(\Znum)^2$.

2132: 

2133: We will prove (\ref{kk}) by induction, so assume (\ref{kk}) holds for $k < l$.

2134: Clearly (\ref{k1}) implies the inclusion

2135: 

2136: $$ H^0(\XXX, {\mm _x} ^l \cdot \Oxx) \subseteq H^0(\XXX, \Oxx(-l\Znum)). $$

2137: 

2138: We want to show surjectivity. We take a $g \in H^0(\XXX, \Oxx(-l\Znum))$ and restrict it to

2139: $ \bar{g} \in H^0(E, \OO _{\Znum} (-l\Znum)). $ By lemma \ref{gen} we have a surjection

2140: 

2141: $$ H^0(E,\OO _{\Znum} (-\Znum)) \otimes H^0(E,\OO _{\Znum} (-(l-1)\Znum)) \twoheadrightarrow H^0(E,\OO _{\Znum} (-l\Znum)), $$

2142: 

2143: i.e. we can write $\bar{g}$ in the form $\bar{g} = \sum\limits_{j=1}^m \bar{x}_j \bar{y}_j $ with

2144: \begin{eqnarray*}

2145: && \bar{x}_j \in H^0(E,\OO _{\Znum} (-\Znum))  \text{ and }\\

2146: && \bar{y}_j \in H^0(E,\OO _{\Znum} (-(l-1)\Znum)) \text{ for } j = 1,\dots, m .

2147: \end{eqnarray*}

2148: Lifting $\bar{x}_j$ and $\bar{y}_j$ to sections on \XXX

2149: \begin{eqnarray*}

2150: && x_j \in H^0(\XXX,\Oxx (-\Znum)) \text{ and } \\

2151: && y_j \in H^0(\XXX,\Oxx (-(l-1)\Znum)) \text{ for } j = 1,\dots, m

2152: \end{eqnarray*}

2153: gives for

2154: $f_2:=\sum\limits_{j=1}^m x_j y_j$ by induction hypothesis

2155: \begin{eqnarray*}

2156: & f_2 \in H^0(\XXX,{\mm _x} ^l \cdot \Oxx) \text{ and }\\

2157: & g-f_2 \in H^0(\XXX, \Oxx (-(l+1)\Znum)).

2158: \end{eqnarray*}

2159: Continuing in this fashion gives

2160: \begin{eqnarray*}

2161: & f_2, \dots , f_p \in H^0(\XXX,{\mm _x} ^l \cdot \Oxx) \text{ and }\\

2162: & g-f_2- \dots -f_p \in H^0(\XXX, \Oxx (-(l+p-1)\Znum))  .

2163: \end{eqnarray*}

2164: 

2165: Hence it suffices to prove

2166: $$

2167: H^0(\XXX, \Oxx (-p\Znum)) \subseteq H^0(\XXX,{\mm _x} ^l \cdot \Oxx) \text{ for } p \gg 0 .

2168: $$

2169: The point is that

2170: $$

2171: \bigoplus_{k \ge 0} H^0(\XXX, \Oxx(-k\Znum))

2172: $$

2173: is a finitely generated $R$-algebra. Assuming this, let $M$ be the maximal degree in a fixed set of generators. For $p > l M$, each element of

2174: $H^0(\XXX, \Oxx(-p\Znum))$ is a sum of products of at least $l$ generators, thus

2175: $$

2176: H^0(\XXX, \Oxx (-p\Znum)) \subseteq H^0(\XXX,{\mm _x} ^l \cdot \Oxx) \text{ for } p > l M  .

2177: $$

2178: For the proof of the assumption, note that the complete linear system $| \OO _{\XX ^{\prime}}(-\Znum) |$ is free:

2179: By the useful fact, $H^1(\XX ^{\prime} , \OO _{\XX ^{\prime}}(-2\Znum))=0$, i.e. we have a surjection

2180: $$

2181: H^0(\XX ^{\prime} , \OO _{\XX ^{\prime}}(-\Znum)) \twoheadrightarrow H^0(E,\OO _{\Znum}(-\Znum)) .

2182: $$

2183: We have seen in the proof of lemma \ref{mznum} that  $| \OO _{\Znum}(-\Znum) |$ is free and hence so is

2184: $| \OO _{\XX ^{\prime}}(-\Znum) |$.

2185: Thus we have a well-defined morphism

2186: $$

2187: \phi_{|- \Znum |} : \XX ^{\prime} \rightarrow X^{\prime} \times {\mathbb P}^N = {\mathbb P}_{X^{\prime}}^N .

2188: $$

2189: We denote its image by $Y:=\text{im }  \phi_{|- \Znum |} $. We want to show that $Y$ is closed. Since $\XX \rightarrow X$ is proper,

2190: so is $\XX^{\prime} \rightarrow X^{\prime}$ (\cite{Ha} II.4.8.(c)). Because of the separatedness of

2191: ${\mathbb P}_{X^{\prime}}^N \rightarrow X^{\prime}$ (\cite{Ha} II.4.9), we see that

2192: $\phi_{|- \Znum |} : \XX ^{\prime} \rightarrow  {\mathbb P}_{X^{\prime}}^N $ is proper (\cite{Ha} II.4.8.(e)), in particular

2193: $\phi_{|- \Znum |}$ is closed. Hence $Y$ is closed.

2194: 

2195: The pullback under $ \phi_{|- \Znum |}  $

2196: of the relatively ample line bundle $\OO(1) := \OO _{X^{\prime}} \otimes _{\CC} \OO_{{\mathbb P}^N}(1)$ is by

2197: definition of $\phi_{|- \Znum |}$ simply $\OO_{\XX^{\prime}} (-\Znum)$ (\cite{Reid} 4.18).

2198: 

2199: Thus it suffices to show that

2200: $$

2201: S^{\prime}(Y) := \bigoplus_{k \ge 0} H^0(Y,\OO _Y  (k))

2202: $$

2203: is finitely generated as a $R$-algebra. The homogenous coordinate ring

2204: $S(Y)=A[x_0, \dots , x_N]/I(Y)$

2205: of $Y$ is certainly a finitely generated $R$-algebra.

2206: By \cite{Ha} Ex. II.5.9, there exists a natural graded morphism

2207: $$

2208: S(Y) \rightarrow S^{\prime}(Y) ,

2209: $$

2210: which is an isomorphism in high degrees, i.e.

2211: $$

2212: S(Y)^d \stackrel{\sim}{\longrightarrow}  S^{\prime}(Y)^d \text{ for } d \gg 0,

2213: $$

2214: and we are done. \hfill $\Box$

2215: 

2216: \begin{corollary}

2217: \label{znum2}

2218: For a rational double point $(X,x)$, we have

2219: $(\Znum)^2 = -2$.

2220: \end{corollary}

2221: 

2222: 

2223: % -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2224: \section{The geometry of the exceptional set $E$ of a re\-solution of a rational double point}

2225: 

2226: Once the hard work has been done in proving theorems \ref{theo1} and \ref{theo3}, it is now easy to say explicitly

2227: what configurations can arise for $E$, if $(X,x)$ is a rational double point.

2228: 

2229: From now on, we will assume that \res\ is a good resolution of a rational double point $(X,x)$ and $E$ its exceptional set.

2230: 

2231: By proposition \ref{negdef}, we have $E_i^2 \le -1$ for $i=1, \dots ,

2232: n$. If $E_{i_0}^2 = -1 $ for some $i_0$, then $E_{i_0} \cong \PP$

2233: can be contracted by Castelnuovo's criterion \ref{app4} to give a

2234: resolution $\pi^{\prime}: \XXX \rightarrow X$ with fewer $E_i$.

2235: (In general, the resolution $\pi^{\prime} : \XXX \rightarrow X$

2236: needs not to be good anymore, since the condition \ref{goodcond2} in the definition of a good resolution

2237: might be violated. However, it is a simple consequence of the

2238: following theorem \ref{theo4} and, again, lemma \ref{negdef}, that

2239: this cannot happen in our case. Note that we do not use this condition

2240: \ref{goodcond2} in the proof of \ref{theo4}.)

2241: 

2242: Therefore, we can assume $E_i^2 \le -2$ for $i=1, \dots, n$ without loss of generality.

2243: \begin{theorem}

2244: \label{theo4} {\rm (\cite{Df})} The $E_i$ have

2245: self-intersection-number $-2$.

2246: \end{theorem}

2247: 

2248: {\bf Proof}: Let $K$ be a canonical divisor on $\widetilde{X}$

2249: (\cite{Ha} V.1.4.4). The adjunction formula \ref{app2} tells us

2250: 

2251: \begin{equation}

2252: \label{adjdf}

2253: -E_i \cdot K = E_i^2 + 2

2254: \end{equation}

2255: 

2256: and thus

2257: 

2258: $$ E_i \cdot K \ge 0 . $$

2259: 

2260: We apply the adjunction formula \ref{app2} for \Znum

2261: 

2262: $$ 2 p_a(\Znum)-2 = (\Znum)^2 + \Znum \cdot K, $$

2263: 

2264: and get by the corollaries \ref{znum0} and \ref{znum2}

2265: 

2266: $$ 0 = \Znum \cdot K = \sum\limits_{i=1}^n r_i(E_i \cdot K) \ge 0,

2267: \text{ i.e. } E_i \cdot K =0. $$

2268: 

2269: Using (\ref{adjdf}) again, we see $E_i^2=-2$. \hfill $\Box$

2270: 

2271: \vspace{3mm} We define the {\em Dynkin diagram} of the resolution

2272: \res\ to be the weighted dual graph $\Gamma$ associated to $E$:

2273: The vertices $e_i$ of $\Gamma$ correspond to the $E_i$. Whenever $E_i$

2274: and $E_j$ intersect for $i \neq j$, the corresponding vertices are

2275: joined by an edge. Finally, we associate to every vertex $e_i$ of

2276: $\Gamma$ the self-intersection-number $E_i^2$.

2277: 

2278: Every weighted graph $\Gamma$ defines a bilinear form $\langle

2279: \cdot,\cdot \rangle$ on the free module with the vertices $e_i,

2280: i=1, \dots, n$ of $\Gamma$ as basis in the following way: We take

2281: \begin{eqnarray*}

2282: & \langle e_i,e_i \rangle := \text{the weight of }e_i \text{ and }

2283: \\ & \langle e_i,e_j \rangle := \text{number of edges joining }e_i

2284: \text{ and }e_j.

2285: \end{eqnarray*}

2286: 

2287: The bilinear form of the Dynkin diagram of a resolution is obviously given by the matrix

2288: $(E_i \cdot E_j)_{i,j=1,\dots,n}$ and hence negative definite by proposition

2289: \ref{negdef}. This puts very strong restrictions on the possible Dynkin diagrams $\Gamma$.

2290: 

2291: \begin{proposition} {\rm (\cite{Df})}

2292: Let $\Gamma$ be a connected graph weighted by $-2$ whose

2293: associated bilinear form is negative definite. Then $\Gamma$ is a

2294: $T$-tree $T_{p,q,r}$

2295: 

2296: \begin{center}

2297: \Ttree

2298: \end{center}

2299: 

2300: with ${1 \over p} + {1 \over q} + {1 \over r}

2301: > 1$.

2302: \end{proposition}

2303: 

2304: {\bf Proof}: Every connected subgraph $\Gamma^{\prime}$ of

2305: $\Gamma$ satisfies the hypothesis as well, hence can be

2306: \begin{itemize}

2307: \item neither a loop

2308: 

2309: \begin{center}

2310: \cyclus,

2311: \end{center}

2312: 

2313: since then $(e_1 + \dots + e_n)^2=0$ contradicting the negative

2314: definiteness condition

2315: \item nor of the form

2316: 

2317: \begin{center}

2318: \cross,

2319: \end{center}

2320: 

2321: since then $(2e_1 + \dots + 2e_n + f_1 + \dots + f_4)^2=0$.

2322: \end{itemize}

2323: 

2324: Thus $\Gamma$ must be of the form $T_{p,q,r}$.

2325: 

2326: The condition ${1 \over p} + {1 \over q} + {1 \over r} > 1$ follows by an elementary argument: With respect to the standard basis given by the vertices

2327: of $\Gamma$, the associated bilinear form is expressed by the matrix

2328: 

2329: $$

2330: \left(

2331: \begin{array}{ccccccccccc}

2332: -2 & 1 &&&&&&&&& \\

2333: 1 & -2 & 1 &&&&&&&& \\

2334: & 1 & -2 &&&& \mathop{\fbox{1}}\limits_{p,p+q} &&&& \\

2335: &&& \ddots &&&&&&& \\

2336: &&&& -2 & 1 &&&&& \\

2337: &&&& 1 & -2 & \mathop{\fbox{0}}\limits_{p+q-1,p+q} &&&& \\

2338: &&& \mathop{\fbox{1}}\limits_{p+q,p} && \mathop{\fbox{0}}\limits_{p+q,p+q-1} & -2 & 1 && \\

2339: &&&&&& 1 & -2 & 1 && \\

2340: &&&&&&& 1 & -2 && \\

2341: &&&&&&&&& \ddots & 1 \\

2342: &&&&&&&&& 1 & -2

2343: \end{array}

2344: \right).

2345: $$

2346: 

2347: But, up to congruence, this is equal to

2348: 

2349: $$

2350: \left(

2351: \begin{array}{cccccccccc}

2352: -\frac{2}{1} &&&&&&&&& \\

2353: & -\frac{3}{2} &&&&&&&& \\

2354: && \ddots &&&&&&& \\

2355: &&& -\frac{p+1}{p} & \mathop{\fbox{1}}\limits_{p,p+1} &&& \mathop{\fbox{1}}\limits_{p,p+q} && \\

2356: &&& \mathop{\fbox{1}}\limits_{p+1,p} & -\frac{q}{q-1} &&&&& \\

2357: &&&&& \ddots &&&& \\

2358: &&&&&& -\frac{2}{1} &&& \\

2359: &&& \mathop{\fbox{1}}\limits_{p+q,p} &&&& -\frac{r}{r-1} && \\

2360: &&&&&&&& \ddots & \\

2361: &&&&&&&&& -\frac{2}{1}

2362: \end{array}

2363: \right).

2364: $$

2365: 

2366: Now, this matrix is congruent  to a diagonal matrix with negative

2367: main diagonal entries, except maybe a single one

2368: $1-p^{-1}-q^{-1}-r^{-1}$. \hfill $\Box$

2369: 

2370: \begin{corollary}

2371: The Dynkin diagram associated to a rational double point $(X,x)$

2372: must be one of the following diagrams

2373: {\rm

2374: \begin{center}

2375: $A_n$ \raisebox{-2mm}{\An} ($n$ vertices), \\

2376: $D_n$ \raisebox{-11mm}{\Dn}($n$ vertices), \\

2377: $E_6$ \raisebox{-11mm}{\Esechs},\hspace{5mm} $E_7$ \raisebox{-11mm}{\Esieben} and \\

2378: $E_8$ \raisebox{-11mm}{\Eacht}.

2379: \end{center}

2380: }

2381: \end{corollary}

2382: 

2383: We will see in section \ref{C2G} that all Dynkin diagrams actually occur. We say that a rational double point is of type $A_n$, $D_n$ or $E_n$

2384: according to its Dynkin diagram.

2385: 

2386: % -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2387: \section{Example: The singularities $\CC/G$ for finite $G \subset \text{SL}(2,\CC)$}

2388: \label{example1}

2389: After a bit of the theory of rational double points has been presented, we want to study an example.

2390: 

2391: 

2392: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2393: \subsection{Conjugacy classes of finite subgroups of $\text{SL}(2,\CC)$}

2394: 

2395: As a preliminary, we recall briefly the classification of

2396: conjugacy classes of finite subgroups of $\text{SL}(2,\CC)$. We

2397: consider first $\text{SO}(3,{\mathbb R})$. Up to conjugacy, the

2398: finite subgroups of $\text{SO}(3,{\mathbb R})$ are the rotational

2399: symmetry groups of

2400: \begin{itemize}

2401: \item a pyramid (giving the {\em cyclic subgroups} $C_n$)

2402: 

2403: \begin{center}

2404: \includegraphics[scale=0.1]{pyramid},

2405: \end{center}

2406: 

2407: \item an orange (corresponding to the {\em dihedral subgroups} $D_n$)

2408: 

2409: \begin{center}

2410: \includegraphics[scale=0.3]{orange},

2411: \end{center}

2412: 

2413: \item and the Platonic solids, which give

2414: \begin{itemize}

2415: \item the {\em tetrahedral subgroup} $T=A_4$

2416: 

2417: \begin{center}

2418: \includegraphics[scale=0.35]{tetrahedron},

2419: \end{center}

2420: 

2421: \item the {\em octahedral subgroup} $O=S_4$

2422: 

2423: \begin{center}

2424: \includegraphics[scale=0.35]{octahedron.eps},

2425: \end{center}

2426: 

2427: \item and the {\em icosahedral subgroup} $I=A_5$

2428: 

2429: \begin{center}

2430: \includegraphics[scale=0.35]{icosahedron.eps},

2431: \end{center}

2432: 

2433: \end{itemize}

2434: \end{itemize}

2435: respectively \cite{Sl}.

2436: 

2437: If we identify $S^2 \cong \PP$, we get an inclusion of the group of isometries of $\PP$ (with respect to the usual metric) into the group of

2438: conformal transformations

2439: 

2440: $$ \text{SO}(3,{\mathbb R}) \subset {\mathbb P}\text{GL}(2,\CC) . $$

2441: 

2442: Under the double cover

2443: 

2444: $$ \rho: \text{SL}(2,\CC) \twoheadrightarrow {\mathbb P}\text{GL}(2,\CC) = \text{SL}(2,\CC) / \{\pm 1\}  $$

2445: 

2446: this inclusion corresponds to

2447: 

2448: $$ \rho^{-1}(\text{SO}(3,{\mathbb R})) = \text{SU}(2,\CC). $$

2449: 

2450: Since for any finite subgroup $G$ of \text{SL}(2,\CC) we can find

2451: a $G$-invariant Hermitian metric by averaging an arbitrary one,

2452: every finite subgroup $G$ of $\text{SL}(2,\CC)$ is conjugated to a

2453: subgroup of $\text{SU}(2,\CC)$. Hence it corresponds to a finite

2454: subgroup of $\text{SO}(3,{\mathbb R})$, unless it is a cyclic

2455: group of odd order. Thus we have derived the following

2456: classification of the conjugacy classes of finite subgroups of

2457: $\text{SL}(2,\CC)$ (\cite{Lm} II \S 1):

2458: \begin{itemize}

2459: \item the cyclic subgroup of order $n$ $C_n$,

2460: \item the binary dihedral subgroups $\widetilde{D}_n = \rho^{-1}(D_n)$,

2461: \item the binary tetrahedral, octahedral and icosahedral subgroup $\widetilde{T}=\rho^{-1}(T)$, $\widetilde{O}=\rho^{-1}(O)$

2462: and $\widetilde{I}=\rho^{-1}(I)$ respectively.

2463: \end{itemize}

2464: 

2465: 

2466: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2467: \subsection{The singularities $\CC^2 / G$}

2468: \label{C2G}

2469: 

2470: Now let $G$ be any of these subgroups; the affine orbit variety

2471: 

2472: $$ \CC^2 / G = \text{Spec }\CC[ x_1, x_2]^G $$

2473: 

2474: has an isolated singularity at the origin. The singularities

2475: obtained in this fashion are all rational double points \cite{Df}.

2476: It is a result from classical invariant theory that these

2477: singularities embed in codimension one

2478: 

2479: $$ \CC^2 / G = \text{Spec }\CC[x,y,z] / (f) \text{ where } f \in \CC[x,y,z] . $$

2480: 

2481: See for example Klein's influential book \cite{Klein}, or also \cite{DuVal3} 5.39. For a modern treatment, we refer to \cite{DPT}, p. 5.

2482: The following table \ref{rdptable} contains the basic information about these singularities.

2483: 

2484: \begin{table}[h]

2485: \caption{The singularities $\CC^2 / G = \text{Spec }\CC[x,y,z] /

2486: (f) $ for finite $G \subset \text{SL}(2,\CC)$} \label{rdptable}

2487: \begin{minipage}{\linewidth}

2488: \begin{tabular}{|c|c|c|c|c|}

2489: \hline

2490: $G$ & $f$ & Name & Dynkin diagram &

2491: \Znum\footnote{The number above a vertex denotes the multiplicity of the corresponding

2492: projective line $E_i \cong \PP$ in the numerical cycle \Znum.} \\

2493: \hline \hline

2494: &&& \An & \Anznum \\

2495: $C_n$  & $x^n+y^2+z^2$ & $A_{n-1}$ &

2496: ($n-1$ vertices) & \\

2497: \hline

2498: &&& \Dn & \Dnznum \\

2499: $\widetilde{D}_n$& $x^{n+1}+xy^2+z^2$ & $D_{n+2}$ &

2500: ($n+2$ vertices) & \\

2501: \hline

2502: $\widetilde{T}$ & $x^2+y^3+z^4$ & $E_6$ &

2503: \Esechs & \Esechsznum \\

2504: \hline

2505: $\widetilde{O}$ & $x^2+y^3+yz^3$ & $E_7$ &

2506: \Esieben & \Esiebenznum \\

2507: \hline

2508: $\widetilde{I}$ & $x^2+y^3+z^5$ & $E_8$ &

2509: \Eacht & \Eachtznum \\

2510: \hline

2511: \end{tabular}

2512: \end{minipage}

2513: \end{table}

2514: 

2515: 

2516: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2517: \subsection{The icosahedral case $G=\widetilde{I}$}

2518: \label{icosahedral}

2519: 

2520: We will sketch the proof of the assertions made so far in the special case $G=\widetilde{I}$. We see that

2521: 

2522: $$ X:= \CC^2 / \widetilde{I} = \text{Spec }\CC [ x, y, z ] / (x^2 + y^3 + z^5) $$

2523: 

2524: has a singularity at $x_0:=(0,0,0)$, which must be a double point, since

2525: $$\text{rank}_{\CC}\frac{(x,y,z)^k+(x^2+y^3+z^5)}{(x,y,z)^{k+1}+(x^2+y^3+z^5)} \le 2k+1 .$$

2526: 

2527: We want to show that $(X,x_0)$ is a {\bf rational} double point. Indeed, we will show how to resolve the singularity $(X,x_0)$ (\cite{Lm} IV \S 9,

2528: \cite{DPT} p. 15). Blowing up $\CC^2$ instead of $\CC^2 / \widetilde{I}$ will reveal the relevant information more clearly. The blow-up of $\CC^2$

2529: at the origin is known to be ${\mathbf v}(\OO _{\PP} (-1))$ (\cite{Ha} V.3.1);

2530: here ${\mathbf v}({\mathcal F})$ denotes the vector bundle determined by the locally trivial

2531: sheaf $\mathcal F$. The quotient of ${\mathbf v}(\OO _{\PP} (-1))$ by $\{\pm 1\}$ is ${\mathbf v}(\OO _{\PP} (-2))$:

2532: 

2533: \begin{eqnarray*}

2534: & {\mathbf v}(\OO _{\PP} (-1)) / \{\pm 1\} = \{ (w_1:w_2;z_1,z_2)

2535: \in \PP \times \CC ^2 : w_1 z_2 = w_2 z_1 \} / \{\pm 1\} \cong \\

2536: & \{ (w_1:w_2;z_1,z_2) \in \PP \times \CC ^2 : w_1^2 z_2 = w_2^2

2537: z_1 \} =  {\mathbf v}(\OO _{\PP} (-2)) \\ & (w_1:w_2;z_1,z_2)

2538: \mapsto (w_1:w_2;z_1^2,z_2^2) .

2539: \end{eqnarray*}

2540: Note that the $\widetilde{I}$-action on $\CC^2$ lifts to an action on ${\mathbf v}(\OO _{\PP} (-1))$.

2541: Thus we have the following commutative diagram

2542: 

2543: $$

2544: \begin{CD}

2545: {\mathbf v}(\OO _{\PP}(-2)) \cong {\mathbf v}(\OO _{\PP}(-1)) / \{ \pm 1 \}  @<<< {\mathbf v}(\OO _{\PP}(-1)) @>\sigma>> \CC^2 \\

2546: @VVV @VVV @VVV \\

2547: {\mathbf v}(\OO _{\PP}(-2)) / I @<\sim<< {\mathbf v}(\OO _{\PP}(-1)) / \widetilde{I} @>\bar{\sigma}>> \CC^2 / \widetilde{I}

2548: \end{CD}.

2549: $$

2550: 

2551: In particular, $\bar{\sigma}^{-1}(x_0)$ is a copy of \PP\ with self-intersection-number $-2$. A precise analysis of the $I$-action on

2552: ${\mathbf v}(\OO _{\PP} (-2))$ shows, that $I$ acts on the zero-section $S^2 \cong \PP \subseteq {\mathbf v}(\OO _{\PP} (-2))$ in the usual way

2553: as rotations which leave  an inscribed icosahedron invariant. Furthermore the action of $I$ on $T^{\ast}\PP \cong {\mathbf v}(\OO _{\PP} (-2))$

2554: is simply the cotangent action induced by the action of $I$ on $\PP$ (\cite{Lm} IV \S 7). Now, the group $I$  is acting free on $S^2$, except on three

2555: exceptional orbits, which consist of the vertices, the mid-edge-points and the mid-face-points of the inscribed icosahedron, respectively.

2556: Moreover, we see that these three orbits are also the only exceptional orbits for the action of $I$ on ${\mathbf v}(\OO _{\PP} (-2))$.

2557: Therefore, the quotient variety

2558: ${\mathbf v}(\OO _{\PP} (-2)) / I $

2559: is smooth except at the three points corresponding to these orbits.

2560: An explicit calculation using local coordinates shows that these three singular points are cyclic quotient singularities of type $(5,4)$, $(3,2)$ and

2561: $(2,1)$, respectively (\cite{DPT} p. 17), this notion being defined as follows: A {\em cyclic quotient singularity of type} $(n,q)$ is a singularity, which

2562: is isomorphic to $\CC^2 / {{\mathbb \mu}_{n,q}}$ where ${{\mathbb \mu}_{n,q}}$ is the cyclic group generated by

2563: $\left(

2564: \begin{array}{cc}

2565: \xi & 0 \\

2566: 0 & \xi^q

2567: \end{array}

2568: \right)$

2569: for a $n$-th root of unity $\xi$.

2570: Note that the numbers $5$, $3$, $2$ are the ramification indices at the ramification points of the map

2571: 

2572: $$ S^2 \twoheadrightarrow S^2 / I $$

2573: 

2574: corresponding to the exceptional orbits.

2575: 

2576: A cyclic quotient singularity of type $(n,q)$ can be resolved by the Hirzebruch-Jung algorithm using successive blow-ups of points (\cite{Fulton} 2.6).

2577: The exceptional set of a resolution of a cyclic quotient singularity obtained in this way is a bunch of rational curves; the associated Dynkin

2578: diagram is of the form

2579: 

2580: \begin{center}

2581: \HiJu ,

2582: \end{center}

2583: 

2584: where the $b_i$'s are calculated by a modified Euclidean algorithm

2585: 

2586: $$ {n \over q} = b_1 - {1 \over {b_2 - {1 \over \ddots - {1 \over b_k }}}} .$$

2587: 

2588: Applying the Hirzebruch-Jung algorithm three times for the three singular points we got, we obtain a resolution

2589: \res\ with associated Dynkin diagram

2590: 

2591: \begin{center}

2592: \Eachtlab

2593: \end{center}

2594: 

2595: The numerical divisor \Znum\ is easily verified to be

2596: 

2597: $$ \Znum = 2E_1 + 4E_2 + 6E_3 + 3E_4 + 5E_5 + 4E_6 + + 3E_7 + 2E_8 . $$

2598: 

2599: A straightforward computation using equation (\ref{lichtenbaum}) shows

2600: 

2601: $$ p_a(\Znum) = 0, $$

2602: 

2603: i.e. the singularity $(X,x_0)$ is rational by the criterion of theorem \ref{theo2}.

2604: 

2605: 

2606: % --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2607: \section{Changing to the complex analytic category}

2608: \label{complexana}

2609: 

2610: The examples introduced in the preceding section already exhaust

2611: all possibilities of rational double points up to isomorphism in

2612: the complex analytic category. Of course, such a statement cannot

2613: be true in the category of algebraic varieties, since rational

2614: double points can live on the various kinds of surfaces and the

2615: birationality class of the surface is encoded locally due to the

2616: coarseness of the Zariski topology.

2617: 

2618: We give the following (simplified) definition of the {\em complex analytic category}:

2619: \begin{itemize}

2620: \item Its objects are called {\em complex analytic spaces} and can be constructed as follows. Let $U$ be a simply connected open subset of $\CC^n$,

2621: $\OO _U$ the sheaf of complex analytic functions on $U$, and $\II _X$ a sheaf of ideals on $(U,\OO _U)$. Denote by $X \subseteq U$ the zeroset

2622: of $\II _X$ equipped with the {\bf standard topology} and set $\OO _X := \OO _U / \II _X$. The pair $(X,\OO _X)$ is then a complex analytic space.

2623: (For the general definition one allows such simple building blocks to be glued together as in the

2624: definition of schemes.)

2625: \item a {\em morphism of complex analytic spaces} $f: (X,\OO _X) \rightarrow (X^{\prime},\OO _{X^{\prime}})$ is a continuous map

2626: $f: X \rightarrow X^{\prime}$ such that $f^{\ast}: \OO_{X^{\prime}} \rightarrow \OO _X$ is well-defined.

2627: \end{itemize}

2628: To a great extent, the complex analytic category is similar to the category of algebraic varieties: For example, stalks $\OO _{X,x}$ are

2629: noetherian local rings and for reduced complex analytic spaces $(X,\OO _X)$ R\"{u}ckert's Nullstellensatz holds (\cite{Lm} III \S 8).

2630: 

2631: 

2632: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2633: \section{Tautness of rational double points}

2634: 

2635: 

2636: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2637: \subsection{Definition and theorem}

2638: 

2639: To pick up the question of classifying rational double points in the complex analytic category, we introduce the notion of tautness.

2640: 

2641: Let $(X,x)$ be a two-dimensional normal singularity with a good resolution,

2642: whose exceptional set is a bunch of rational curves \PP\, and let

2643: $\Gamma$ be its Dynkin diagram.

2644: 

2645: We say $(X,x)$ is {\em taut}, if up to analytic isomorphism,

2646: $(X,x)$ is the unique such singularity, that has a good resolution with a bunch of rational curves \PP\ as

2647: exceptional set and $\Gamma$ as its Dynkin diagram \cite{B3}.

2648: 

2649: We have the following theorem.

2650: 

2651: \begin{theorem}

2652: \label{rdptaut} The singularities listed in table \ref{rdptable}

2653: are taut.

2654: \end{theorem}

2655: 

2656: This gives us a complete classification of rational double points up to analytic isomorphism. There are several proofs available for theorem \ref{rdptaut}.

2657: 

2658: 

2659: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2660: \subsection{Tjurina's proof}

2661: 

2662: Maybe the most natural one is the proof by Tjurina \cite{Tj}:

2663: Suppose there were another singularity $(X^{\prime},x^{\prime})$ with an exceptional set consisting only of rational curves and

2664: the same Dynkin diagram, i.e. with an isomorphic exceptional variety $(E^{\prime},\OO _{E^{\prime}}) \cong (E,\OO _E)$. Then a sufficient condition for

2665: the existence of an isomorphism of neighbourhoods of $E$ and $E^{\prime}$ is by \cite{Gr} Thm. 3, that $(E,\OO _{n E})$ and

2666: $(E^{\prime},\OO _{n E^{\prime}})$ are isomorphic for $n$ large enough. The proof given in \cite{Tj} proceeds by induction. Assuming

2667: we are given some isomorphism of $(E,\OO _Z)$ and $(E^{\prime},\OO _{Z^{\prime}})$ (here $Z$ and $Z^{\prime}$ are exceptional divisors supported on

2668: $E=\bigcup_{i=1}^n E_i$, $E^{\prime}=\bigcup_{i=1}^n E_i^{\prime}$, respectively), then this can be extended to an isomorphism of $(E,\OO_{Z+E_i})$ and

2669: $(E^{\prime},\OO _{Z^{\prime} + E_i^{\prime}})$, unless some

2670: obstruction occurs, which lies in some cohomology group \cite{Gr}.

2671: Grauert argues, that if all these cohomology groups vanish, then the singularity in question is taut. In general, these cohomology groups

2672: do not vanish and Tjurina's proof is more subtle. Essentially, he shows that the cohomology groups are too small to put obstructions on the lifting

2673: of every possible isomorphism of $(E,\OO _Z)$ and $(E^{\prime},\OO _{Z^{\prime}})$.

2674: 

2675: 

2676: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2677: \subsection{Brieskorn's first proof}

2678: 

2679: For the sake of historical correctness, we mention that the first proof of theorem \ref{rdptaut}

2680: was given by Brieskorn (\cite{B4} Satz 1). He showed that

2681: rational double points can be resolved by blowing up points alone, that is,

2682: it is not necessary to normalize or blow up curves. Such singularities

2683: are called {\em absolutely isolated} and were studied by Kirby (\cite{Kirby} 2.6, 2.7),

2684: who gave a classification of absolutely isolated double points:

2685: they are precisely those listed in table \ref{rdptable}.

2686: 

2687: 

2688: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2689: \subsection{Brieskorn's second proof}

2690: \label{bries}

2691: 

2692: However, we want to sketch another proof of theorem \ref{rdptaut}, which, also due to Brieskorn \cite{B3}, is of compelling beauty and combines

2693: ideas from different fields of mathematics:

2694: 

2695: The {\em local fundamental group} of $(X,x)$ is defined as

2696: 

2697: $$ \pi _{X,x} := \mathop{\varprojlim}_U \pi_1(U \setminus \{x\} )$$

2698: 

2699: where $U$ runs over all neighbourhoods of $x \in X$ (\cite{B3} \S 2). Equivalently, we can calculate $\pi _{X,x}$ as

2700: 

2701: $$ \pi _{X,x} \cong \mathop{\varprojlim}_{\widetilde{U}} \pi_1(\widetilde{U} \setminus \{x\} )$$

2702: 

2703: where the limit is now taken over all neighbourhoods $\widetilde{U}$ of $E \subset \widetilde{X}$. To actually compute $\pi_{X,x}$ it is sufficient

2704: to work out $\pi_1$ for a {\em good} neighbourhood $U$. According to \cite{Prill}, a neighbourhood $U$ of $x \in X$ is called {\em good}, if there exists

2705: a neighbourhood basis $\{ U_i \}$ for $x$ such that $U_i \setminus \{ x \}$ is a deformation retract of $U \setminus \{ x \}$ for all $i$.

2706: Such a good neighbourhood has the homotopy type of a tubular neighbourhood $M$ of $E$ in the sense of Mumford \cite{Mf}.

2707: Intuitively spoken, a tubular neighbourhood is a levelset of the potential distribution due to a uniform charge on $E$. Mumford studied these tubular

2708: neighbourhoods $M$ and showed that they are built out of standard pieces $S^1 \times S^1 \times [ 0,1]$ "plumbed" together in a certain fashion

2709: determined by $(E_i \cdot E_j)_{i,j=1,\dots,n}$.

2710: This description and the Seifert-van-Kampen theorem enables him to give a presentation for $\pi_1 (M)$ in terms of generators and relations. (The ideas

2711: of his proof can also be found in \cite{Lm} IV \S\S 10 - 14 on plumbed surfaces.)

2712: It turns out that for the intersection matrices $(E_i \cdot E_j)_{i,j=1,\dots,n}$

2713: of the resolutions of rational double points this group $\pi _{X,x} = \pi_1 (M)$ is finite.

2714: 

2715: \begin{center}

2716: A rational double point has finite local fundamental group.

2717: \end{center}

2718: 

2719: For the following see \cite{B3} Satz 2.8, \cite{Prill} Thm. 3.

2720: From a merely topological point of view, $x \in X$ possesses a

2721: neighbourhood $U$ with $U^{\prime} := U \setminus \{ x \}$ having

2722: a finite universal cover $V^{\prime} \twoheadrightarrow

2723: U^{\prime}$. This can be uniquely extended to a ramificated cover

2724: $V \twoheadrightarrow U$ by adding a point $y$ to $V^{\prime}$.

2725: Moreover, we can equip $V$ with a normal analytic structure such

2726: that $ V \twoheadrightarrow U$ becomes an analytically ramificated

2727: cover. Since $V^{\prime}$ is simply

2728: connected, we see $\pi _{V,y}=1$. By another fundamental theorem

2729: in Mumford's paper (\cite{Mf} p. 18), this shows the non-singularity

2730: of $V$ at $y$. Now $\pi _{X,x}$ is operating via cover

2731: transformations on $V^{\prime}$, hence also on $V$ with fixed

2732: point $y$. We need another definition to state our results so far.

2733: 

2734: \begin{defprop}{\rm  (Two-dimensional quotient singularities)}

2735: For a neighbourhood $V$ of the origin $O$ in $\CC^2$ and a finite group $G$ of analytic automorphisms of $V$ fixing $O$, the quotient space

2736: $V / G$ has the structure of a normal complex analytic surface and the projection $V \twoheadrightarrow V / G$ is analytic \cite{B3}.

2737: We say that $U$ is a {\em two-dimensional quotient singularity},  if $U$ is isomorphic to a singularity of the form $V/G$.

2738: \end{defprop}

2739: 

2740: We have just seen:

2741: 

2742: \begin{center}

2743: A rational double point is a quotient singularity.

2744: \end{center}

2745: 

2746: By a simple linearization argument (\cite{B3} Lemma 2.2), we can restrict ourselves to the study of quotient singularities of the form $\CC^2 / G$ where

2747: $G$ is a finite subgroup of $\text{GL}(2,\CC)$. Obviously, conjugated subgroups yield isomorphic quotient spaces.

2748: The conjugacy classes of finite subgroups of $\text{GL}(2,\CC)$ have been listed by Du Val (\cite{DuVal3} \S 21).

2749: Later, Prill showed that only a particular class of finite subgroups has to be studied: the so-called {\em small} subgroups \cite{Prill}.

2750: He also classified them (\cite{Prill} Satz 2.3). Using Prill's results, Brieskorn gave a complete classification of the quotient spaces that can

2751: arise in terms of the Dynkin diagram of their resolution (\cite{B3} p. 348). This shows that two-dimensional quotient singularities are taut and finishes

2752: Brieskorn's proof.

2753: 

2754: 

2755: % ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2756: \section{Seven characterizations of rational double points}

2757: 

2758: The following remark allows us to use the intermediate results in

2759: the above discussion \ref{bries} to give alternative

2760: characterizations of rational double points in the analytic

2761: category.

2762: 

2763: \begin{remark}

2764: \label{lastbit}

2765: It can be shown that $\CC^2 / G$, for $G \subset \text{\rm GL}(2,\CC)$ finite,

2766: embeds in codimension one if and only if $G$ is a subgroup

2767: of $\text{\rm SL}(2,\CC)$ {\rm (\cite{Df} cor. 5.3)}.

2768: \end{remark}

2769: 

2770: \begin{theorem}

2771: \label{characterizations}

2772: {\rm (\cite{Df})}

2773: 

2774: Let $(X,x)$ be a normal surface singularity that {\bf embeds in codimension one}. Then the following conditions are equivalent in the

2775: complex analytic category.

2776: \begin{enumerate}

2777: \item $(X,x)$ is a rational double point.

2778: \item $(X,x)$ has a good resolution with an exceptional set consisting of rational curves with self-intersection-number $-2$.

2779: \item $(X,x)$ has a good resolution with an exceptional set consisting of rational curves and a Dynkin diagram listed in table \ref{rdptable}.

2780: \item The local fundamental group of $(X,x)$ is finite.

2781: \item $(X,x)$ is a two-dimensional quotient singularity.

2782: \item $(X,x)$ is isomorphic to $\CC^2 / G$ for finite $G \subset \text{\rm GL}(2,\CC)$.

2783: \item $(X,x)$ is isomorphic to one of the affine varieties studied in section \ref{example1}.

2784: \end{enumerate}

2785: 

2786: \end{theorem}

2787: 

2788: {\bf Remark on the proof:} We have already seen a proof of the

2789: implications $(1) \Rightarrow (2) \Rightarrow (3) $ and presented

2790: some ideas for $ (3) \Rightarrow (4) \Rightarrow (5) \Rightarrow

2791: (6)$. In the example \ref{icosahedral}, we studied a special case

2792: of $(7) \Rightarrow (1)$. The last implication $(6) \Rightarrow

2793: (7)$ is precisely remark \ref{lastbit}.

2794: 

2795: \vspace{3mm} In his survey article \cite{Df}, Durfee presents

2796: further ten characterizations of rational double points. The

2797: characterizations he gives provide a connection of rational double

2798: points for example with weighted homogeneous

2799: polynomials, vanishing cycles, a certain limit

2800: involving volumes, monodromy groups and Morse functions.

2801: A more number-theoretical characterization in terms

2802: of almost factorial rings (fast-faktoriell) rings is due to

2803: Brieskorn (\cite{B3} Satz 1.5).

2804: Finally, a link with elementary catastrophes is discussed in a survey article by Slodowy (\cite{Sl} 9).

2805: 

2806: 

2807: % -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2808: \section{Example: The conical double point}

2809: 

2810: Although we did not give a proof of theorem

2811: \ref{characterizations}, we shall at least study an example to

2812: illustrate the phenomena encountered there. We will consider the

2813: {\em conical double point} $X:=V(f) \subset \CC^3$ where $f=xz -

2814: y^2 \in \CC [x,y,z]$. Note that after a change of variables $xz -

2815: y^2$ becomes $x_1^2 + x_2^2 +x_3^2$; so $X$ is just the surface

2816: singularity labeled $A_1$ from table \ref{rdptable}. $X$ is a

2817: double cone with vertex a rational double point:

2818: 

2819: \begin{center}

2820: \cone.

2821: \end{center}

2822: 

2823: Obviously, $X$ has a normal singularity at $x=(0,0,0)$ and $X$ is embedded in codimension one.

2824: $(X,x)$ is a double point, since

2825: 

2826: $$\text{rank}_{\CC}\frac{(x,y,z)^k + (f)}{(x,y,z)^{k+1} +(f)} = 2k+1,  $$

2827: 

2828: i.e. the leading coefficient of the Hilbert-Samuel polynomial of the local ring $\OO _{X,x}$ at  $x$ it two.

2829: 

2830: Furthermore, $(X,x)$ is absolutely isolated, because the singularity can be resolved by a single blow-up at $x$, as can be easily seen by the

2831: toric description of $X$ \cite{Fulton}:

2832: 

2833: \begin{center}

2834: \toric.

2835: \end{center}

2836: 

2837: We can work out the blow-up explicitly and obtain

2838: \label{blowup}

2839: 

2840: $$ \widetilde{X}^{\prime} \subset \mathop{{\mathbb A}^3 \times {\mathbb P}^2}_{(x,y,z;p:q:r)} $$

2841: 

2842: given by equations

2843: 

2844: $$ xz - y^2, pr - q^2, py=qx, pz=rx, qz=ry .$$

2845: 

2846: $\widetilde{X}^{\prime}$ is isomorphic to

2847: 

2848: $$ \widetilde{X} \subset \mathop{{\mathbb A}^2 \times {\mathbb P}^1}_{(x,z;u:v)} $$

2849: 

2850: cut out by

2851: 

2852: $$ xv^2 = zu^2  $$

2853: 

2854: via

2855: 

2856: \begin{eqnarray*}

2857: \widetilde{X} & \stackrel{\sim}{\longrightarrow} & \widetilde{X}^{\prime} \\

2858: (x,z;u:v) & \mapsto & (x,\frac{v}{u}x\text{ or }\frac{u}{v}z,z;u^2:uv:v^2)

2859: \end{eqnarray*}

2860: 

2861: Now $\widetilde{X}$ is a line bundle on \PP\, whose zero-section has self-intersection-number $-2$. Thus $\widetilde{X}$  is just the line bundle

2862: on the projective line associated to the sheaf $\OO _{\PP}(-2)$, i.e. the cotangent bundle $T^{\ast} \PP$.

2863: See also \cite{Lm} IV \S7.

2864: 

2865: Since $\widetilde{X}$ is a smooth variety, we have a resolution of $(X,x)$

2866: 

2867: $$ \pi : \widetilde{X} \cong \widetilde{X}^{\prime} \twoheadrightarrow X . $$

2868: 

2869: The exceptional set $E$ of  $\pi$ is precisely the zero-section of  $\widetilde{X}$, hence isomorphic to \PP. Moreover $E^2 = -2 $.

2870: The real picture reflects the situation very nicely

2871: 

2872: \begin{center}

2873: \resreal.

2874: \end{center}

2875: 

2876: For the numerical cycle we get $\Znum = E$, i.e. $(X,x)$ is

2877: rational by theorem \ref{theo2}, and characterization $(1)$ is

2878: verified.

2879: 

2880: As already mentioned in section \ref{C2G}, $X$ is isomorphic to the affine orbit variety $\CC^2 / \{ \pm 1 \}$, where we write $-1$ for the reflection in the

2881: origin of the complex plane $\CC^2$. This corresponds to characterizations $(5)$, $(6)$ and $(7)$.

2882: 

2883: Finally, let us calculate the local fundamental group $\pi_{X,x}$. We have a covering map

2884: 

2885: $$ \CC^2 \setminus \{O\} \twoheadrightarrow X \setminus \{ x\} $$

2886: 

2887: with covering transformation group $\{ \pm 1 \}$.

2888: For every $\{ \pm 1 \}$-invariant simply connected neighbourhood

2889: $U$ of $O \in \CC^2$, we observe that $U \setminus \{ O \}$ is

2890: also simply connected, hence

2891: $$

2892: \pi_1 ((U \setminus \{ O \}) / \{ \pm 1 \}) = \{ \pm 1 \} .

2893: $$

2894: But such an $U$ can be chosen arbitrarily small, thus

2895: $$

2896: \pi_{X,x} = \{ \pm 1\},

2897: $$

2898: which is finite and shows characterization $(4)$.

2899: 

2900: 

2901: % -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2902: \section{Lie groups and rational double points}

2903: 

2904: The "$A_n$-$D_n$-$E_n$" - labeling of the various types of

2905: rational double points was actually borrowed from the

2906: classification theory of Lie groups. In this last section we will

2907: sketch some of the deep connections between Lie groups and

2908: rational double points. Essentially, we shall

2909: give a summary of \cite{Sl} 10.

2910: 

2911: 

2912: % -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

2913: \subsection{Dynkin diagrams of simple Lie groups}

2914: 

2915: A connected complex Lie group is called {\it (almost) simple}, if it

2916: contains no normal subgroup of positive dimension.

2917: In their classification theory, the simply connected simple Lie

2918: groups play a special r\^{o}le as their universal coverings

2919: (which are finite (\cite{Sl} 10)).

2920: These groups are classified by their corresponding

2921: Dynkin diagrams \cite{FH}. Surprisingly, the Dynkin diagrams of

2922: table \ref{rdptable} occur again.

2923: 

2924: We recall briefly the relevant part of this classification; note that most of the

2925: following facts hold in a more general context (\cite{FH},

2926: \cite{Sl2} 3.1).

2927: 

2928: Let $G$ be a simply connected simple Lie group of rank $r$ and

2929: \GG\ its Lie algebra. We fix a maximal torus $T\cong(\CC^{\ast})^r$

2930: of $G$ with {\it character group}

2931: 

2932: $$ X^{\ast} ( T ) = \text{Hom} ( T, \CC^{\ast}) = \ZZ^r . $$

2933: 

2934: We denote the normalizer of $T$ in $G$ by $N_G(T)$. The group

2935: $W:=N_G(T) / T$ is finite and is called the {\it Weyl group} of $G$

2936: with respect to $T$.

2937: 

2938: The restriction of the adjoint representation of $G$ on \GG\ to $T$

2939: has eigenspaces $\GG_{\alpha}$ on which $T$ acts by the character

2940: $\alpha \in X^{\ast}(T)$ and we obtain the {\it Cartan

2941: decomposition} of \GG\

2942: 

2943: $$ \GG = \bigoplus_{\alpha \in X^{\ast}(T)} \GG_{\alpha} . $$

2944: 

2945: The finite set $\Sigma := \{ \alpha \in X^{\ast}(T) : \alpha \neq

2946: 0 , \GG_{\alpha} \neq \{0\}\} $ is called the {\it root space}.

2947: Clearly, $\Sigma$ is invariant under the action of $W$.

2948: 

2949: We can define a $W$-invariant scalar product $\langle\cdot,\cdot\rangle$ on $X^{\ast}(T)$

2950: (called the {\it Killing form}) such that the elements of $W$

2951: become reflections in the hyperplane perpendicular to a root

2952: $\alpha$

2953: 

2954: $$ \beta \mapsto \beta - \frac{2 \langle \alpha, \beta \rangle}{\langle

2955: \alpha , \alpha \rangle} \alpha . $$

2956: 

2957: Here $\frac{2 \langle \alpha, \beta \rangle}{\langle

2958: \alpha , \alpha \rangle}$ must be integer.

2959: (The main tool in proving this  and similar facts is identifying

2960: the subalgebra $\GG_{\alpha} \oplus \GG_{- \alpha} \oplus [ \GG_{\alpha},\GG_{-\alpha} ]$

2961: with $\mathfrak{sl}(2,\CC)$ and applying the representation

2962: theory of $\mathfrak{sl}(2,\CC)$.)

2963: 

2964: The geometry of how $\Sigma$ sits in the Euclidean lattice $(X^{\ast}(T),\langle \cdot,

2965: \cdot \rangle)$ is very rigid. For example

2966: 

2967: $$ 4 \cos^2 \angle (\alpha, \beta) =\frac{ 4 \langle \alpha, \beta

2968: \rangle^2 }{\langle \alpha, \alpha \rangle \langle \beta, \beta

2969: \rangle}$$

2970: 

2971: must be an integer between zero and four, i.e. there are just a

2972: few possibilities for the angle between two roots $\alpha$ and

2973: $\beta$.

2974: 

2975: By choosing a direction $l \in T$ (in general position with

2976: respect to $\Sigma$) we can specify the {\it positive roots}

2977: $\alpha \in \Sigma$ to be those with $\alpha(l) > 0$.

2978: In particular, we can focus on {\it simple roots}: these are

2979: positive roots that are not the sum of two other positive roots.

2980: The system of simple roots gives rise to a Dynkin diagram, where

2981: we take a vertex for each simple root and join two vertices by

2982: exactly

2983: $ 4 \cos^2 \angle (\alpha, \beta) $

2984: lines. If we insist on all roots $\alpha$ having the same

2985: length $\langle \alpha, \alpha \rangle$, the only possibilities for

2986: the Dynkin diagram are

2987: 

2988: \begin{center}

2989: $A_n$ \raisebox{-2mm}{\Anplain} (n vertices), \\

2990: $D_n$ \raisebox{-7mm}{\Dnplain} (n vertices), \\

2991: $E_6$ \raisebox{-7mm}{\Esechsplain}, \\

2992: $E_7$ \raisebox{-7mm}{\Esiebenplain} and\\

2993: $E_8$ \raisebox{-7mm}{\Eachtplain}.

2994: \end{center}

2995: 

2996: 

2997: These diagrams $A_n$, $D_n$, $E_6$, $E_7$ and $E_8$ actually occur

2998: for the simply connected simple Lie groups corresponding to the

2999: classical Lie algebras $\mathfrak{sl}(n+1,\CC)$,

3000: $\mathfrak{so}(2n,\CC)$ and the exceptional Lie algebras

3001:  $\mathfrak{e}_6$, $\mathfrak{e}_7$ and $\mathfrak{e}_8$, respectively.

3002: 

3003: It can be shown that a simply connected simple Lie group can be recovered from

3004: its Dynkin diagram.

3005: 

3006: 

3007: % -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

3008: \subsection{A theorem of Brieskorn}

3009: 

3010: Let $G$ be a simply connected simple Lie group.

3011: We consider the quotient $H$ of $G$ by its adjoint action in the

3012: category of algebraic varieties $p: G \twoheadrightarrow H$. There

3013: is an explicit way to describe $p$. Let $r$ be the rank of $G$ and $\rho_i: G \rightarrow

3014: \text{GL}(V_i), i=1, \dots, r$ be the $r$ fundamental irreducible

3015: representations of $G$ on finite-dimensional vector spaces. Then the character map

3016: 

3017: \begin{eqnarray*}

3018: \chi: G & \rightarrow & \CC^r \\

3019: g & \mapsto & (\dots,\text{trace}_{V_i} \rho_i(g),\dots)

3020: \end{eqnarray*}

3021: 

3022: coincides with $p$.

3023: 

3024: \paragraph{Example:} For $G=\text{SL}(n,\CC)$ we have $\text{rank }G=n-1$ and the $n-1$

3025: fundamental irreducible representations are given by the exterior powers

3026: 

3027: $$V_i = {\bigwedge}^i\, \CC^n . $$

3028: 

3029: The corresponding characters are, up to sign, just the

3030: non-trivial coefficients of the characteristic polynomial

3031: 

3032: $$\text{char}(g)=\text{det}(\lambda - g) = \lambda^n -

3033: \text{trace}(g) \lambda ^{n-1}

3034: + \text{trace}({\wedge}^2 g)\lambda^{n-2} -+ \dots .$$

3035: 

3036: Thus we can regard $\chi$ as associating to $g \in

3037: \text{SL}(2,\CC)$ its characteristic polynomial.

3038: 

3039: \vspace{3mm}

3040: 

3041: The point is now to study the {\it unipotent variety}

3042: 

3043: $$\text{Uni}(G) := p^{-1}(p(e)). $$

3044: 

3045: \paragraph{Example:} For $G=\text{SL}(n,\CC)$ the unipotent

3046: variety consists precisely of the unipotent matrices.

3047: 

3048: \vspace{3mm}

3049: 

3050: The variety $\text{Uni}(G)$ is a finite union of conjugacy classes

3051: and contains a unique conjugacy class of dimension $d:=\text{dim }G

3052: -r$ (since $p$ is flat) --- the {\it regular class}.

3053: The complement of the regular class in $\text{Uni}(G)$ is the

3054: closure of a unique conjugacy class of dimension $d-2$ --- the

3055: {\it subregular class} $\text{Sub}(G)$.

3056: 

3057: We choose a $S \subseteq G$ such that

3058: 

3059: \begin{eqnarray*}

3060: & S \text{ is smooth of dimension }\text{dim }G-d+2, \\

3061: & S \cap \text{Sub}(G)=\{ x \} \text{ and } \\

3062: & T_x S + T_x \text{Sub}(G) = T_x G,

3063: \end{eqnarray*}

3064: 

3065: i.e. we require that $S$ is a {\it slice} of codimension $d-2$ transversal to

3066: $\text{Sub}(G)$ at an element $x \in \text{Sub}(G)$.

3067: 

3068: Let $X:=S \cap \text{Uni}(G)$.

3069: 

3070: The following theorem was conjectured by Grothendieck and proved

3071: by Brieskorn.

3072: 

3073: \begin{theorem}

3074: \label{lie}

3075: {\rm (\cite{B5}, \cite{Sl} 10)}

3076: If $G$ is a simply connected simple Lie group of type $A_n$, $D_n$

3077: or $E_n$, then $(X,x)$ is a rational double point.

3078: \end{theorem}

3079: 

3080: \paragraph{Example:} We want to illustrate this theorem in the

3081: simplest possible case $G=\text{SL}(2,\CC)$. All regular unipotent

3082: elements are conjugate to the matrix

3083: $$

3084: \left(

3085: \begin{array}{cc}

3086: 1 & 1 \\

3087: 0 & 1

3088: \end{array}

3089: \right)

3090: $$

3091: and there exists just a single subregular unipotent element

3092: $$

3093: x:=

3094: \left(

3095: \begin{array}{cc}

3096: 1 & 0 \\

3097: 0 & 1

3098: \end{array}

3099: \right) .

3100: $$

3101: As a transversal slice we can simply take $S:=\text{SL}(2,\CC)$.

3102: Now $p=\chi$ is given by the trace

3103: \begin{eqnarray*}

3104: \chi : \text{SL}(2,\CC) & \rightarrow & \CC \\

3105: \left(

3106: \begin{array}{cc}

3107: a & b \\

3108: c & d

3109: \end{array}

3110: \right)

3111: & \mapsto & a+d

3112: \end{eqnarray*}

3113: and we get

3114: \begin{eqnarray*}

3115: X & = & S \cap \text{Uni}(G) \\

3116: & =& \text{Uni}(G)=\chi^{-1}

3117: \left(

3118: \chi

3119: \left(

3120: \begin{array}{cc}

3121: 1 & 0 \\

3122: 0 & 1

3123: \end{array}

3124: \right)

3125: \right) \\

3126: & = & \left\{

3127: \left(

3128: \begin{array}{cc}

3129: 1+x & y \\

3130: z & 1+u

3131: \end{array}

3132: \right) : x+u=0 \text{ and } xu-yz=0

3133: \right\} \\

3134: & = & \{ (x,y,z) : x^2 + yz = 0\},

3135: \end{eqnarray*}

3136: 

3137: i.e. $X$ has a conical double point at

3138: $x=(0,0,0)$.

3139: 

3140: 

3141: % -------------------------------------------------------------------------------------------------------------------

3142: \subsection{Resolutions of rational double points in the Lie group context}

3143: 

3144: A closed subgroup $P\subseteq G$ is called {\it parabolic}, if the

3145: quotient space $G / P$ is a projective variety. The minimal parabolic

3146: subgroups are the {\it Borel subgroups}. All Borel subgroups are

3147: conjugate to each other in $G$ and the normalizer $N_G(P)$ of a

3148: parabolic subgroup coincides with P. Thus the set of all Borel

3149: subgroups $\mathcal B$ becomes a projective variety

3150: 

3151: $$ {\mathcal B} = G / B $$

3152: 

3153: where $B$ is any Borel subgroup of $G$. More generally, ${\mathcal

3154: P}:= G/P$ may be identified with the set of subgroups conjugate to

3155: the parabolic subgroup $P$.

3156: 

3157: \paragraph{Example:} The parabolic subgroups of $\text{SL}(n,\CC)$

3158: are exactly the stabilizer of the flags

3159: 

3160: $$ 0 \subset V_{i_1} \subset \dots \subset V_{i_k} \subset \CC^n

3161: \text{ with } \text{rank}_{\CC}V_{i_j}=i_j, j=1,\dots,k .$$

3162: 

3163: Hence the Borel subgroups correspond to maximal flags

3164: 

3165: $$ 0 \subset V_1 \subset \dots \subset V_{n-1} \subset \CC^n , $$

3166: 

3167: i.e. are conjugate to the subgroup of upper triangular matrices.

3168: 

3169: \vspace{3mm}

3170: 

3171: It was Springer who showed that the natural projection from the

3172: incidence variety

3173: 

3174: $$ I := \{ (x,B) \in \text{Uni}(G) \times {\mathcal B} : x \in B

3175: \} $$

3176: 

3177: to $\text{Uni}(G)$ is a $G$-equivariant resolution of the

3178: singularities of $\text{Uni}(G)$

3179: 

3180: $$ \pi : I \twoheadrightarrow \text{Uni}(G) . $$

3181: 

3182: Let $G,x,S,X$ be as in theorem \ref{lie}. It is a consequence of

3183: the $G$-equivariance of $\pi$ that the restriction

3184: 

3185: $$ \pi : \widetilde{X} := \pi^{-1}(X) \twoheadrightarrow X $$

3186: 

3187: is again a resolution. In fact, it is a {\it minimal} one, that

3188: is, we cannot apply theorem \ref{app4} to obtain a resolution with smaller

3189: exceptional set.

3190: 

3191: We can interpret the exceptional set

3192: 

3193: $$ E:= \pi^{-1} (x) $$

3194: 

3195: in two different ways.

3196: 

3197: On one hand, we know from theorem \ref{lie}, that $(X,x)$ is a

3198: rational double point. Hence $E$ must be a bunch of projective

3199: lines \PP\ intersecting each other as prescribed by the Dynkin

3200: diagram $\Gamma$ of $(X,x)$.

3201: 

3202: On the other hand, we can write

3203: 

3204: $$ E = \{ (x,B) \in \{x\} \times {\mathcal B} : x \in B\} . $$

3205: 

3206: The vertices of the Dynkin diagram $\Gamma_G$ of $G$ correspond to

3207: the simple roots of $G$ (after a maximal torus $T_0$ and a

3208: direction $l \in T_0$, or equivalently, a Borel subgroup $B_0

3209: \supset T_0$ have been specified). Let $P_{\alpha}$ be the minimal

3210: proper (i.e. non-Borel) parabolic subgroup generated by $B_0$ and

3211: the root subgroup $U_{-\alpha}$, where $\alpha$ is a simple root.

3212: Because of $N_G(P_{\alpha})=P_{\alpha}$, we can identify the set

3213: of subgroups conjugate to $P_{\alpha}$ with the projective

3214: variety

3215: 

3216: $$ {\mathcal P}_{\alpha}:= G / P_{\alpha} .$$

3217: 

3218: The natural map

3219: 

3220: $$ f_{\alpha} : {\mathcal B} \cong G/B_0 \rightarrow {\mathcal

3221: P}_{\alpha} \cong G/P_{\alpha} $$

3222: 

3223: maps each Borel subgroup $B$ to the unique parabolic subgroup $P\in

3224: {\mathcal P}_{\alpha}$ containing $B$. Since each $P \in {\mathcal

3225: P}_{\alpha}$ has semisimple rank 1, this map has projective lines

3226: as fibres.

3227: 

3228: Steinberg and Tits showed that $E$ is a bunch of projective lines

3229: --- one line of the form $f_{\alpha}^{-1}(P), P\in {\mathcal

3230: P}_{\alpha}$ for every simple root $\alpha$ --- which intersect as

3231: prescribed by the edges of $\Gamma_G$.

3232: 

3233: \paragraph{Example:} We verify these statements by explicit

3234: calculation in the simplest non-trivial case

3235: $G=\text{SL}(3,\CC)$.

3236: 

3237: As maximal torus $T_0$ we may take the diagonal matrices and as

3238: Borel subgroup the upper triangular matrices.

3239: 

3240: The root space of $\text{SL}(3,\CC)$ is

3241: \begin{center}

3242: \roots

3243: \end{center}

3244: where the character $\alpha_{i,j} \in X^{\ast}(T)$ is defined by

3245: 

3246: $$ \alpha_{i,j}

3247: \left(

3248: \begin{array}{ccc}

3249: a_1 && \\

3250: & a_2 & \\

3251: && a_3

3252: \end{array}

3253: \right)

3254: =

3255: a_i - a_j . $$

3256: 

3257: The eigenspace $\mathfrak{sl}(3,\CC)_{\alpha_{i,j}}$, for

3258: example, consists of those matrices

3259: $(a_{i,j})_{i,j=1,\dots,3}\in\mathfrak{sl}(3,\CC)$

3260: whose single non-zero entry is $a_{2,1}$. Using the

3261: exponential map

3262: $$

3263: \text{exp} : \mathfrak{sl}(3,\CC) \rightarrow \text{SL}(3,\CC)

3264: $$

3265: we see that the root subgroup $U_{\alpha_{21}}$ is

3266: $$

3267: U_{\alpha_{21}} =

3268: \left\{

3269: \left(

3270: \begin{array}{ccc}

3271: 1 & 0 & 0 \\

3272: \lambda & 1 & 0 \\

3273: 0 & 0 & 1

3274: \end{array}

3275: \right) : \lambda \in \CC

3276: \right\} .

3277: $$

3278: 

3279: The simple roots are $\alpha_{1,2}$ and $\alpha_{2,3}$ and we

3280: obtain

3281: \begin{center}

3282: \Azweibig\

3283: \end{center}

3284: as Dynkin diagram of $\text{SL}(3,\CC)$ (\cite{FH} 12).

3285: The unipotent variety $\text{Uni}(\text{SL}(3,\CC))$ is given by

3286: the unipotent matrices, all regular unipotent elements are

3287: conjugate to

3288: $$

3289: \left(

3290: \begin{array}{ccc}

3291: 1 & 1 & 0 \\

3292: 0 & 1 & 1 \\

3293: 0 & 0 & 1

3294: \end{array}

3295: \right)

3296: $$

3297: and all subregular unipotent elements are conjugate to

3298: $$

3299: x:=\left(

3300: \begin{array}{ccc}

3301: 1 & 0 & 0 \\

3302: 0 & 1 & 1 \\

3303: 0 & 0 & 1

3304: \end{array}

3305: \right) .

3306: $$

3307: We have the minimal proper parabolic subgroups

3308: $$

3309: P_{\alpha_{1,2}} = \langle B_0 , U_{\alpha_{2,1}} \rangle

3310: $$

3311: and

3312: $$

3313: P_{\alpha_{2,3}} = \langle B_0 , U_{\alpha_{3,2}} \rangle

3314: $$

3315: which are the stabilizer of the flags

3316: $$

3317: 0 \subset \text{span}_{\CC}\{ v_1 , v_2 \} \subset \text{span}_{\CC}\{ v_1 , v_2 ,v_3

3318: \}

3319: $$

3320: and

3321: $$

3322: 0 \subset \text{span}_{\CC}\{ v_1 \} \subset \text{span}_{\CC}\{ v_1 , v_2 ,v_3

3323: \},

3324: $$

3325: respectively. Clearly, $P_{\alpha_{2,3}}$ is conjugate to the

3326: parabolic subgroup $P_{\alpha_{2,3}}^{\prime}$ stabilizing the flag

3327: $$

3328: 0 \subset \text{span}_{\CC}\{ v_2 \} \subset \text{span}_{\CC}\{ v_1 , v_2 ,v_3

3329: \}.

3330: $$

3331: The set of Borel subgroups containing $x$, which we had identified

3332: with $E$, is given by

3333: $$

3334: f_{\alpha_{1,2}}^{-1}(P_{\alpha_{1,2}}) \cup

3335: f_{\alpha_{2,3}}^{-1}(P_{\alpha_{2,3}}^{\prime}).

3336: $$

3337: (All the other fibres of the maps $f_{\alpha_{1,2}}$ and

3338: $f_{\alpha_{2,3}}$ contain only a finite number of Borel subgroups

3339: which contain $x$.)

3340: 

3341: The two fibres intersect in a single Borel

3342: subgroup: the subgroup stabilizing the maximal flag

3343: $$ 0 \subset \text{span}_{\CC}\{ v_2  \} \subset \text{span}_{\CC}\{ v_1 , v_2 \} \subset \text{span}_{\CC}\{ v_1 , v_2 , v_3

3344: \}  . $$

3345: Hence the Dynkin diagram of $E$ looks like

3346: \begin{center}

3347: \Azweibig .

3348: \end{center}

3349: 

3350: 

3351: % -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

3352: \appendix

3353: \section{Appendix - results from Algebraic Geometry}

3354: 

3355: \begin{theorem}

3356: \label{app1}

3357: {\rm (Riemann-Roch on a surface)(\cite{Ha} V.1.6)}

3358: If $D$ is any divisor on the non-singular surface $X$, then

3359: 

3360: $$ \chi(\Ox(D)) = \frac{1}{2} D \cdot (D-K) + 1 + p_a(X) $$

3361: 

3362: where $K$ is a {\em canonical divisor} on $X$ (\cite{Ha} V.1.4.4).

3363: \end{theorem}

3364: 

3365: A simple corollary is the following.

3366: 

3367: \begin{theorem}

3368: \label{app2}

3369: {\rm (General adjunction formula)(\cite{Ha} Ex. V.1.3.(a))}

3370: If $D$ is an ef\-fec\-tive divisor on the non-singular surface $X$, then

3371: 

3372: $$ 2 p_a(D) -2 = D \cdot (D+K) . $$

3373: \end{theorem}

3374: 

3375: \begin{theorem}

3376: \label{app3}

3377: {\rm (Grothendieck's theorem on formal functions)(\cite{EGA3} 4.2.1 or for projective morphisms \cite{Ha} III.11.1)}

3378: Let $f: X \rightarrow Y$ be a proper morphism of noetherian schemes and $\mathcal F$ a coherent sheaf on $X$. For $y \in Y$ denote by

3379: $m_y \subset \OO _{Y,y}$ the maximal ideal of the stalk at $y$. Then we have a natural isomorphism

3380: 

3381: $$ 0 = \left( \left( R^i f_{\ast} {\mathcal F} \right)_y \right)\widehat{}\ \cong\

3382:   \mathop{\varprojlim}_{k=1}^{\infty} H^i\left( f^{-1}(y), {\mathcal F} \otimes _{\OO _Y} {\OO_{Y,y} \over {m_y}^k} \right)

3383: \text{ for all } i \in {\mathbb N}$$

3384: 

3385: where the completion is taken with respect to the $m_y$-adic topology.

3386: \end{theorem}

3387: 

3388: \begin{theorem}

3389: \label{app4}

3390: {\rm (Castelnuovo's criterion for contracting a curve)(\cite{Lm} IV \S 15, \cite{Ha} V.5.7)}

3391: If $C$ is a curve on a non-singular surface $X$ with $C \cong \PP$ and $C^2=-1$, then

3392: there exists a morphism $f: X \rightarrow X^{\prime}$ to a non-singular surface $X^{\prime}$ which contracts $C$ to a point $p$,

3393: such that $X$ is isomorphic via $f$ to the

3394: blow-up of $X^{\prime}$ with center $p$, and $C$ is the exceptional curve.

3395: \end{theorem}

3396: 

3397: \begin{theorem}

3398: \label{app5} {\rm (Zariski's connectedness theorem)(\cite{EGA3}

3399: 4.3.1, \cite{Ha} III.11.4)} Let $f: X \rightarrow Y$ be a

3400: birational morphism between projective varieties and assume that

3401: $Y$ is normal. Then $f$ has connected fibers.

3402: \end{theorem}

3403: 

3404: \begin{theorem}

3405: \label{app6}

3406: {\rm (A vanishing theorem of Grothendieck)(\cite{Ha} III.2.7)}

3407: For any sheaf of abelian groups ${\mathcal F}$ on a noetherian scheme $X$ of dimension $n$, we have

3408: 

3409: $$ H^i(X,{\mathcal F}) =0 \text{ for } i>n . $$

3410: \end{theorem}

3411: 

3412: \begin{theorem}

3413: \label{app7} {\rm (A vanishing theorem for higher direct image

3414: sheaves)(\cite{Ha} III.11.2)} Let $f: X \rightarrow Y$ be a

3415: projective morphism of noetherian schemes and denote by $r$ the

3416: maximal dimension of its fibers. Then for all coherent sheaves

3417: ${\mathcal F}$ on $X$, we have

3418: 

3419: $$ R^i f_{\ast}\mathcal F = 0 \text{ for } i>r .$$

3420: \end{theorem}

3421: 

3422: \newpage

3423: 

3424: % -----------------------------------------------------------------------------------------------------------------------

3425: \begin{thebibliography}{99}

3426: \bibitem{A2}

3427:   Artin, M.: {\it Some numerical criteria for the contractability of curves on algebraic surfaces.}

3428:   Amer. Journ. of Math. {\bf 84} (1962), pp. 485 - 496.

3429: \bibitem{A1}

3430:   Artin, M.: {\it On isolated rational singularities of surfaces.}

3431:   Amer. Journ. of Math. {\bf 88} (1966), pp. 129 - 136.

3432: \bibitem{B4}

3433:   Brieskorn, E.: {\it \"{U}ber die Aufl\"{o}sung gewisser Singularit\"{a}ten von holomorphen

3434:   Abbildungen.} Math. Ann. {\bf 166} (1966), pp. 76 - 102.

3435: \bibitem{B3}

3436:   Brieskorn, E.: {\it Rationale Singularit\"{a}ten komplexer F\"{a}chen.}

3437:   Inventiones math. {\bf 4} (1968), pp. 336 - 358.

3438: \bibitem{B2}

3439:   Brieskorn, E.: {\it Die Aufl\"{o}sung der rationalen Singularit\"{a}ten holomorpher Abbildungen.}

3440:   Math. Ann. {\bf 178} (1968), pp. 255 - 270.

3441: \bibitem{B5}

3442:   Brieskorn, E.: {\it Singular elements of semisimple algebraic

3443:   groups.} Actes, Congr\`{e}s intern. Math. 1970, vol. 2, pp. 279

3444:   - 284.

3445: \bibitem{Dm}

3446:   Dimca, A.: {\it Topics on real and complex singularities.}

3447:   Vieweg, Braunschweig, 1987.

3448: \bibitem{DPT}

3449:   Demazure, M., H. Pinkham and B. Teissier: {\it S\'{e}minaire sur les singularit\'{e}s des surfaces.}

3450:   Springer, Berlin, 1980.

3451: \bibitem{Df}

3452:   Durfee, A.: {\it Fifeteen characterizations of rational double points and simple critical points.}

3453:   L'Enseignement Mathematique {\bf 25} (1979), pp. 131 - 163.

3454: \bibitem{DV}

3455:   Du Val, P.: {\it On isolated singularities of surfaces that do not affect the condition of adjunction I, II, III.}

3456:   Proc. Cambridge Phil. Soc. {\bf 30} (1934), pp. 483 - 491.

3457: \bibitem{DuVal3}

3458:   Du Val, P.: {\it Homographies, quaternions and rotations.}

3459:   Clarendon Press, Oxford, 1964.

3460: \bibitem{Fulton}

3461:   Fulton, W.: {\it Introduction to toric varieties.}

3462:   Princeton Univ. Press, Princeton, 1993.

3463: \bibitem{FH}

3464:   Fulton, W. and J. Harris: {\it Representation theory: A first

3465:   course.} Springer, New York, 1991.

3466: \bibitem{Gd}

3467:   Godement, R.: {\it Topologie alg\'{e}brique et th\'{e}orie des faisceaux.}

3468:   Herman, Paris, 1958.

3469: \bibitem{Gr}

3470:   Grauert, H.: {\it \"{U}ber Modifikationen und exzeptionelle Mengen.}

3471:   Math. Ann. {\bf 146} (1962), pp. 331 - 368.

3472: \bibitem{EGA3}

3473:   Grothendieck, A. and J. Dieudonn\'{e}: {\it \'{E}l\'{e}ments de

3474:   g\'{e}om\'{e}trie alg\'{e}brique: vol. 3: \'{E}tude cohomologique des

3475:   faisceaux coh\'{e}rents.}

3476:   Institut des Hautes \'{E}tudes, Paris, 1961.

3477: \bibitem{Ha}

3478:   Hartshorne, R.: {\it Algebraic Geometry.}

3479:   Springer, New York, 1977.

3480: \bibitem{Kirby}

3481:   Kirby, D.: {\it The structure of an isolated multiple point of a surface I, II, III.}

3482:   I, II: Proc. London Math. Soc. (3) {\bf 6} (1956), pp. 597 - 609. III: London Math. Soc. (3) {\bf 7}

3483:   (1957), pp. 1 - 28.

3484: \bibitem{Klein}

3485:   Klein, F.: {\it Vorlesung \"{u}ber das Ikosaeder und die Aufl\"{o}sung der Gleichung vom f\"{u}nften Grade.}

3486:   Teubner, Leipzig, 1884.

3487: \bibitem{Lm}

3488:   Lamotke, K.: {\it Regular solids and isolated singularities.}

3489:   Vieweg, Braunschweig, 1986.

3490: \bibitem{Lf}

3491:   Laufer, H.: {\it Normal, two-dimensional singularities.}

3492:   Princeton Univ. Press, Princeton, 1971.

3493: \bibitem{Lp}

3494:   Lipman, J.: {\it Introduction to resolution of singularities.}

3495:   Published in: {\it Algebraic Geometry.} (Proc. Symp. Pure Math. 29, Arcata 1974).

3496:   AMS, Providence, 1975.

3497: \bibitem{McC}

3498:   McCleary, J.: {\it User's guide to spectral sequences.}

3499:   Publish or Perish, Wilmington, 1985.

3500: \bibitem{Mf}

3501:   Mumford, D.: {\it The topology of normal singularities of an algebraic surface and a criterion for simplicity.}

3502:   Publ. Math. Inst. des Hautes \'{E}tudes Sci. {\bf 9} (1961), pp. 229 - 246.

3503: \bibitem{Ng}

3504:   Nagata, M.: {\it Local Rings.}

3505:   Wiley, New York, 1962.

3506: \bibitem{Prill}

3507:   Prill, D.: {\it Local classification of quotients of complex manifolds by discontinous groups.}

3508:   Duke Math. Journ. {\bf 34} (1967), p. 375 - 386.

3509: \bibitem{Reid}

3510:   Reid, M.: {\it Chapters on algebraic surfaces.}

3511:   Published in: {\it Complex algebraic varieties,} Koll\'{a}r, J. (Ed.). AMS, Providence, 1997.

3512: \bibitem{Sh}

3513:   Shatz, S.: {\it Profinite groups, arithmetic and geometry.}

3514:   Princeton Univ. Press, Princeton, 1972.

3515: \bibitem{Sl2}

3516:   Slodowy, P.: {\it Simple singularities and simple algebraic

3517:   groups.} Springer, Berlin, 1980.

3518: \bibitem{Sl}

3519:   Slodowy, P.: {\it Platonic solids, Kleinian singularities and Lie groups.}

3520:   Published in: {\it Algebraic Geometry: Proceedings, Ann Arbour 1981.}

3521:   Springer, Berlin, 1983.

3522: \bibitem{Tj}

3523:   Tjurina, G.: {\it On the tautness of rationally contractible curves on a surface.}

3524:   Izv. Akad. Nauk. USSR Ser. Mat. {\bf 32} (1968), pp. 943 - 970; English transl.: Math. USSR Izv. {\bf vol. 2, no. 4}

3525:   (1968), pp. 907 - 954.

3526: \bibitem{Wb}

3527:   Weibel, C.: {\it An introduction to homological algebra.}

3528:   Cambridge Univ. Press, Cambridge, 1994.

3529: \bibitem{ZS}

3530:   Zariski, O. and P. Samuel: {\it Commutative Algebra, vol. 2.}

3531:   Van Nostrand, Princeton, 1960.

3532: \end{thebibliography}

3533: 

3534: \end{document}

3535: