sign in sign up
<<
Home , Cohomology ring, Complex manifold, Complex projective space

arXiv:1510.02242v1 [math.DG] 8 Oct 2015 hrceiaino yeqarc ntrso h aoinde Fano complex the compact of terms a in hyperquadrics of characterization 1727 n h udmna eerhFnsfrteCentra the for Funds Research Fundamental the and 11471247) inequality. number Chern hcto phic e ] oaah n ciigv n[ in gave Ochiai and Kobayashi 6], rem [ result, 1.1 Theorem c oio’ eutta h ainlPnrai lse r i are classes relaxed Pontrjagin be rational ([ the can that “diffeomorphic” result hypothesis Novikov’s the and inequality to l.Hrerc n oar hwdi [ in showed Kodaira and topologi Hirzebruch and/or geometrical ble. via complex compact C 1 No65 P ( h uhrwsprilyspotdb h ainlNtrlS Natural National the by supported partially was author The phrases. and words Key 2010 nodrt euetermi euti [ in result main their deduce to order In ti nipratpolmt hrceietesadr comp standard the characterize to problem important an is It h atta h oa otjgncasof class Pontrjagin total the that fact The C M n (1) (2) P ) ly e oei hi ro.LtrYuntcdta ([ that noticed Yau Later proof. their in role key a plays n ) nsmay ehave we summary, In ]). OERMRSO H NQEESO H COMPLEX THE OF UNIQUENESS THE ON REMARKS SOME ahmtc ujc Classification. Subject 6= then , nerlchmlg igadPnrai lse stecomp the as classes Pontrjagin and integral Abstract. ti iooopi to biholomorphic is it ¨ he aiodhvn h aeitga ooooyrn a ring cohomology th integral showing same by classes the Pontrjagin having K¨ahler on manifold requirement the drop can oar n a.Ti bevto,tgte iharsl of result a all with characterize together observation, to This Yau. and Kodaira suethat assume hc mrvso eut fFjt n Libgober-Wood. and Fujita of results on improves which M M − C HK57 P ( − n sbhlmrhcto biholomorphic is n sbhlmrhcto biholomorphic is ( 1) + tms ebhlmrhcto biholomorphic be must it , n .20 hoe ] nprdb h dao h ruet f[ of arguments the of idea the by Inspired 6]. Theorem 210, p. , 1) + HzbuhKdia[ (Hizebruch-Kodaira g where , efis oiei hsatceta facmatK¨ahler manif compact a if that article this in notice first We M g ssml-once.Ti ehial ensacasclr classical a refines technically This simply-connected. is ntecs of case the in C nqees ope rjciesae otjgncas P class, Pontrjagin space, projective complex uniqueness, P g n C stepstv eeao of generator positive the is ntrso oooytp ne idsmer.When symmetry. mild under type homotopy of terms in P M n C RJCIESPACES PROJECTIVE provided C P scalled is P n n HK57 rvddthat provided n 1. 21,5C5 32Q55. 53C55, 32Q15, rvddthat provided en vncnb eoe yYusCennumber Chern Yau’s by removed be can even being KO73 Introduction n HK57 ,Yu[ Yau ], C Fano sod h aehlsfreven for holds same The odd. is IGLI PING HK57 P M n nte hrceiainof characterization another ] 1 . hti K¨aher manifold a if that ] fisfis hr class Chern first its if a h tnadfr 1+ (1 form standard the has ,Hrerc n oar hwdarelated a showed Kodaira and Hirzebruch ], Yau77 n sodd; is n Universities. l H ]) see n h rtclass first the and even is . 2 ineFudto fCia(rn No. (Grant China of Foundation cience ( faKalrmnfl shomeomor- is K¨ahler manifold a If s de oemrhs invariants homeomorphism ndeed e rjciespace projective lex M tasml-once compact simply-connected a at esiadWlig nbe us enables Wilking, and Dessai faFn aiod ealthat Recall manifold. Fano a of x C a nomto sltl spossi- as little as information cal P ; Z 4 Yau77 ). sbhlmrhcto biholomorphic is e rjciespaces projective lex o“oemrhc u to due “homeomorphic” to sl fHirzebruch- of esult )teetaassumption extra the ]) l a h same the has old c 1 n ti’ ojcue Yau’s conjecture, etrie’s ( M M fw further we if C g spstv.By positive. is ) C sdiffeomorphic is n P 2 P ) ,we 4, = n n n HK57 +1 then , swl sa as well as C P sta of that as c 4 1 , Theo- , C ( M P n ) as 6= 2 PING LI Kodaira’s embedding theorem a Fano manifold is projective and thus automatically K¨ahler. The Fano index of a Fano manifold M is defined to be the largest positive integer I such that c (M)/I H2(M; Z). We denote by I(M) the Fano index of M. Kobayashi and Ochiai 1 ∈ showed in [KO73] that if the Fano index of an n-dimensional Fano manifold is no less than n + 1, it must be biholomorphic to CP n. A later result of Michelsohn (cf. [LM89, p. 366] or [Mi80, p. 1143]) indeed showed that the Fano index of a Fano manifold can not be larger than n + 1. We summarize them into the following Theorem 1.2 (Kobayashi-Ochiai [KO73], Michelsohn [Mi80]). Suppose M is a Fano mani- fold. Then I(M) n + 1, with equality if and only if M is biholomorphic to CP n. ≤ Remark 1.3. A recent exposition paper [To15] by Tosatti presents a detailed proof of Theo- rem 1.1 and some results in [KO73] as well as some necessary background knowledge. More- over, he gave a detailed proof of the fact that the nonexistence of exotic complex structures on CP 3 implies the nonexistence of complex structures on S6 ([To15, Prop. 3.1]), which was originally observed by Hirzebruch ([Hi54, p. 223]).

The next natural question is whether we are able to relax the hypotheses “K¨ahlerness” and “homeomorphism” in Theorem 1.1 to guarantee that its conclusion remains true. For general n, we have no essentially stronger results up to now, at least to the author’s best knowledge. But when n are small enough, we indeed have some stronger results. For n = 2, still applying his Chern number inequality, together with some well-known facts on compact complex surfaces, Yau showed that ([Yau77]) a compact complex homotopy equivalent to CP 2 is biholomorphic to CP 2, which also solved an old conjecture in algebraic posed by Severi. For n = 3, Lanteri and Struppa showed that ([LS86]) a compact K¨ahler threefold having the same integral as CP 3 is biholomorphic to CP 3, in whose proof Yau’s Chern number inequality is still a major ingredient. For n = 4 or 5, by applying Theorem 1.2, Fujita showed that a Fano manifold having the same integral cohmology ring as CP 4 or CP 5 is biholomorphic to CP 4 or CP 5. By applying Theorem 1.2 and a formula relating the Chern number c1cn−1 to Hodge numbers discovered by themselves in [LW90], Libgober and Wood showed that if a compact K¨ahler manifold is homotopically equivalent to CP n for n = 4, 5 or 6, then it is biholomorphic to CP n. We collect the above-mentioned results into the following Theorem 1.4. We denote by (S) the following statement: (S) := A compact M is biholomorphic to CP n. Then (1) (Yau,[Yau77]) When n = 2, (S) holds if we assume that M is homotopically equivalent to CP n; (2) (Lanteri-Struppa,[LS86]) When n = 3, (S) holds if we assume that M is K¨ahler and has the same integral cohomology ring as CP n; (3) (Fujita,[Fu80]) When n = 4 or 5, (S) holds if we assume that M is Fano and has the same integral cohomology ring as CP n; (4) (Libgober-Wood,[LW90]) When n = 4, 5 or 6, (S) holds if we assume that M is K¨ahler and homotopically equivalent to CP n. Remark 1.5. Note that, when n = 4 or 5, the assumptions in (3) and (4) of Theorem 1.4 can not imply each other and thus their results are independent. Also note that the proof in [Fu80] is sketchy and many details were omitted. UNIQUENESSOFTHECOMPLEXPROJECTIVESPACES 3 2. Main observations

We shall present in this section our main observations of this article, Theorems 2.1, 2.3 and 2.4 and postpone their proofs to the next section. Our first observation is that, if we combine some of Hirzebruch and Kodaira’s original arguments in [HK57] and Kobayashi-Ochiai’s criterion in Theorem 1.2, the original hypothesis “homeomorphism” in Theorem 1.1 can be relaxed to yield the following

Theorem 2.1. Suppose M is a compact K¨ahler manifold having the same integral cohomology ring and Pontrjagin classes as CP n. Then

(1) M is biholomorphic to CP n provided that n is odd; (2) M is biholomorphic to CP n provided that n is even and M is simply-connected. Remark 2.2.

(1) Since H∗(CP n; Z) has no torsion, rational Pontrjagin classes coincide with integral Pontrjagin classes and thus our hypotheses in Theorem 2.1 is strictly weaker than those in Theorem 1.1. (2) It must be known to some experts that the original hypothesis “homeomorphism” in Theorem 1.1 can be relaxed to some extent. However, to the author’s best knowledge, there is no literature where this assumption was explicitly refined in the form as in our Theorem 2.1.

In view of Theorem 1.2, in order to complete the proof of Theorem 2.1, it suffices to show 2 that c1(M) = (n + 1)g with g being a positive generator of H (M; Z). We shall see in the next section in this process the invariance of Pontrjagin classes play a key role. However, as we have mentioned in Theorem 1.4, when n 6, only assuming homotopy equivalence and ≤ without requirement on Pontrjagin classes, Libgober and Wood can still be able to show that c1(M) = (n + 1)g by applying some subtle invariants of homotopy equivalence in algebraic . But their methods are ad hoc and fail to treat the general n. Our second observation is that, if we allow the manifold M to have mild symmetry, the same result still holds for general n. A smooth closed 2n-dimensional manifold is called an n-dimensional homotopy complex if it is homotopically equivalent to CP n. Recall that a classical conjecture in transformation group theory, which was posed by Petrie in [Pe72] and is still open in its full generality, asserts that if an n-dimensional homotopy M admits an (effective and smooth) action, then its total Pontrjagin class agrees with that of CP n, i.e., p(M)=(1+ g2)n+1 for a generator g of H2(M; Z). Petrie himself verified this conjecture ([Pe73]) under the stronger hypothesis that if an n-dimensional torus acts (effectively and smoothly) on M. Dessai and Wilking improved on Petrie’s result by showing n+1 that the conjecture holds if a torus whose is larger than 4 acts on M ([DW04, p. 506]). Now combining Theorem 2.1 with Dessai-Wilking’s this result, our second observations reads

Theorem 2.3. If a compact K¨ahler manifold is homotopically equivalent to CP n and acted n+1 on effectively and smoothly by a torus whose dimension is larger than 4 , then it must be biholomorphic to CP n. When n 6, the latter hypothesis can be dropped by various results ≤ in Theorem 1.4. 4 PING LI We now turn to our third observation in this article. We have mentioned in Remark 1.5 that, when n = 4, the hypotheses of Fujita and Libgober-Wood can not imply each other and thus are independent. Our third observation is to present a weaker hypothesis than both of them. As is now well-known that a Fano manifold is simply-connected, which is a corollary of the celebrated Calabi-Yau theorem (cf. [Zh00, p. 225]), the conditions of simply-connectedness and having the same integral cohomology ring are strictly weaker than the assumptions in (3) and (4) of Theorem 1.4. Therefore our third observation, which improves on the results of Fujita and Libgober-Wood when n = 4, asserts that the assumption on the invariance of Pontrjagin classes in Theorem 2.1 can be dropped if n = 4: Theorem 2.4. A simply-connected compact K¨ahler manifold having the same integral coho- mology ring as CP 4 is biholomorphic to CP 4.

3. Proofs of Theorems 2.1 and 2.4

3.1. Proof of Theorem 2.1. We first show the following key lemma under the hypotheses in Theorem 2.1 , which is [HK57, p. 208, Lemma2]. Here the idea of our proof was still adopted from [HK57] but is more direct and compact. We shall also see from this process that the technical assumption we need is only the invariance of the integral cohomology ring and Pontrjagin classes. Lemma 3.1. Suppose M is a compact K¨ahler manifold and its integral cohomology ring and n Pontrjagin classes are the same as those of CP . Then c1(M) = (n + 1)g (resp. c1(M) = (n + 1)g) provided n is odd (resp. even). Here g is the positive generator of H2(M; Z). ± Proof. By assumptions we have

∗ H (M; Z)= Z[g]/(gn+1), gn = 1, ZM and the total Pontrjagin class of M is given by p(M)=(1+ g2)n+1. We first note that the Hodge numbers of M are the same as those of CP n. Indeed, the famous relations of Hodge numbers for compact K¨ahler manifolds tell us that hp,p(M) 1, hp,q(M) 0 for 0 p,q n, ≥ ≥ ≤ ≤ and 1 + ( 1)i hp,q(M) = the i-th of M = the i-th Betti number of CP n = − , 2 p+Xq=i which imply that hp,p(M) = 1 (0 p n), and hp,q(M)=0(p = q). ≤ ≤ 6 These lead to the value of the Todd genus of M, td(M), via n td(M)= ( 1)qh0,q(M) = 1. − Xq=0 The Todd genus is a complex genus in the sense of Hirzebruch ([Hi66, Ch.1,3] or [HBJ92, 1.8]) whose associated power series is § x x x 2 −x = e x − x . 1 e · e 2 e 2 − − UNIQUENESSOFTHECOMPLEXPROJECTIVESPACES 5 x Note that x − x is nothing but the even power series whose associated genus is the Aˆ-genus e 2 −e 2 and can be defined for oriented closed smooth manifolds in terms of Pontrjagin classes ([Hi66, Ch.1],[HBJ92, 1.6]). We now suppose c (M) = kg for k Z. In view of the fact that § 1 ∈ p(M)=(1+ g2)n+1, we have

kg g 2 n+1 1 = td(M)= e ( g − g ) ZM · e 2 e 2 − (k+n+1)g g 2 n+1 = e ( g ) ZM · e 1 − n+1 n (k+n+1)g g = the coefficient of g in e 2 · (eg 1)n+1 − (k+n+1)g 1 = the residue of e 2 at g = 0 · (eg 1)n+1 − 1 (k+n+1)g 1 = e 2 dg 2π√ 1 I · (eg 1)n+1 − k+n−1 − 1 (y + 1) 2 = dy (eg 1 =: y) 2π√ 1 I yn+1 − − − n k+n 1 = the coefficient of y in (y + 1) 2 j(j 1) (j n + 1) k + n 1 = − · · · − . ( − =: j) n! 2 This yields n!= j(j 1) (j n + 1) − · · · − and thus j = n (resp. j = n or 1) provided n is odd (resp. even). This implies k = n + 1 − (resp. k = (n + 1)) provided n is odd (resp. even) and thus completes the proof.  ±

In view of Theorem 1.2, in order to complete the proof of Theorem 2.1, it suffices to show that, when n is even, the additional hypothesis of simply-connectedness on M can rule out the possibility of c (M) = (n + 1)g. This can follows from equality case of the following 1 − inequality due to Yau ([Yau77]):

Theorem 3.2 (Yau’s Chern number inequality, negative case). Suppose M is an n-dimensional compact K¨ahler manifold with c1(M) < 0. Then we have the following Chern number inequal- ity

2(n + 1) − (3.1) ( c )n 2c ( c )n, n − 1 2 ≥ − 1 where the equality holds if and only if M has constant holomorphic , i.e., M is holomorphically covered by the unit ball in Cn.

Completion of proof of Theorem 2.1.

Now suppose n is even and c1(M)= (n + 1)g. Thus means c1(M) < 0 and the inequality − 2 (3.1) holds for M. The first Pontrjagin class of M, p1(M), is equal to (n + 1)g as the total Pontrjagin class p(M)=(1+g2)n+1. Recall that for a complex manifold M we have the basic fact p (M)= c2(M) 2c (M). Then c (M)= n(n+1) g2 and so M satisfies the equality case 1 1 − 2 2 2 of (3.1), contradicting to the simply-connectedness of M. 6 PING LI 3.2. Proof of Theorem 2.4. We shall adopt the same strategy as in [LW90, p. 147] to prove Theorem 2.4. 4 If M has the same integral chomology ring as that of CP , then the Chern number c1c3 of M equals to 50 by [LW90, Corollary 2.5]. This, together with the fact that c (M) 5g from 1 ≤ Theorem 1.2, tells us that the possible values of c1(M) are (3.2) g, 2g, 5g, 10g, 25g, 50g. ± ± ± − − − We also know from last subsection that the Todd genus td(M) = 1. By the formula of Todd genus in terms of Chern numbers for n = 4 ([Hi66, p.14]) we have 1 1 = td(M)= ( c + c c + 3c2 + 4c2c c4). 720 − 4 1 3 2 1 2 − 1 Note that the top Chern number c4 = 5 as it equals to the Euler number of M. Combining this with c1c3 = 50 yields the following relation (3.3) 3c2 + 4c2c + ( c4 675) = 0. 2 1 2 − 1 − By the abuse of notation we may view c1 and c2 as integers in (3.3) via the identifications 2 4 2 H (M; Z)= Zg and H (M; Z)= Zg . Thus, if we view (3.3) as a equation of c2, we have 4c2 7c4 + 2025 (3.4) c = − 1 ± 1 . 2 p3

It can be checked directly that the only values among (3.2) which make c2 in (3.4) integral are c = 5g. In this case c = 10g2. The possibility (c , c ) = ( 5g, 10g2) still satisfies the 1 ± 2 1 2 − equality case in (3.1) and thus can be ruled out by the simply-connected hypothesis as before. This means the only possibility is c1(M) = 5g, which completes the proof. Remark 3.3. The only difference between this proof and [LW90] is that, under the assump- tion of homotopy equivalence made in [LW90], the parity of c1 must be the same as that of n + 1 (in the case of n = 4 c1 must be odd) as its modulo two reduction is exactly the second Stiefel-Whitney class, which is an invariant under homotopy equivalence due to the classical Wu formula. So the possible values considered in [LW90] are smaller than ours in 4 (3.2). Fortunately, all the other values in (3.2) make the discriminant 7c1 + 2025 in (3.3) square-free and thus can still be ruled out. So the strategy in [LW90] can be carried over to deal with Theorem 2.4.

Acknowledgements

Part of the work was done when I visited Taida Institute for Mathematical Sciences at National Taiwan University in July and August 2015. I would like to thank the institute and Professor Chang-Shou Lin for their hospitality. I also would like to thank Professor Valentino Tosatti for his interest and useful comments on the content of this note, and Dr. Zhaohu Nie for some useful discussions during our stay in Taiwan.

References

[DW04] A. Dessai, B. Wilking: Torus actions on homotopy complex projective spaces, Math. Z. 247 (2004), 505-511. [Fu80] T. Fujita: On topological characterizations of complex projective spaces and affine linear spaces, Pro. Japan Acad. Sci. 56 (1980), 231-234. UNIQUENESSOFTHECOMPLEXPROJECTIVESPACES 7

[Hi54] F. Hirzebruch: Some problems on differentiable and complex manifolds, Ann. of Math. 60 (1954), 213- 236. [Hi66] F. Hirzebruch: Topological methods in , 3rd Edition, Springer, Berlin (1966). [HBJ92] F. Hirzebruch, T. Berger, R. Jung: Manifolds and modular forms, Aspects of Mathematics, E20, Friedr. Vieweg and Sohn, Braunschweig, 1992. [HK57] F. Hirzebruch, K. Kodaira: On the complex projective spaces, J. Math. Pures Appl. 36 (1957), 201-216. [KO73] S. Kobayashi, T. Ochiai: Characterizations of complex projective spaces and hyperquadrics, Ann. Math. 13 (1973), 31-47. [LS86] A. Lanteri, D. Stuppa: Projective manifolds with the same Pk, Monatsh. Math. 101 (1986), 53-58. [LM89] H.B. Lawson, M.L. Michelson: Spin Geometry, Princeton University Press, Princeton, New Jersey, 1989. [LW90] A. Libgober, J. Wood: Uniqueness of the complex structure on K¨ahler manifolds of certain homology types, J. Differential Geom. 30 (1990), 139-154. [Mi80] M.L. Michelson: Clifford and spinor cohomology of K¨ahler manifolds, Amer. J. Math. 102 (1980), 1083-1146. [No65] S. Novikov: Rational Pontrjagin classes. Homeomorphism and homotopy type of closed manifolds. I, Izv. Akad. Nauk SSSR Ser. Mat. 29 (1965), 1373-1388. [Pe72] T. Petrie: Smooth S1 actions on homotopy complex projective spaces and related topics, Bull. Amer. Math. Soc. 78 (1972), 105-153. [Pe73] T. Petrie: Torus actions on homotopy complex projective spaces, Invent. Math. 20 (1973), 139-146. [To15] V. Tosatti: Uniqueness of CP n, arXiv:1508.05641. [Yau77] S.-T. Yau: Calabi’s conjecture and some new results in algebraic geometry, Proc. Natl. Acad. Sci. USA, 74 (1977), 1798-1799. [Zh00] F. Zheng: Complex differential geometry, AMS/IP Studies in Advanced Mathematics 18, American Mathematical Society, Providence, RI 2000.

Department of Mathematics, Tongji University, Shanghai 200092, China E-mail address: [email protected], [email protected]