Arxiv:1907.03330V3 [Math.AG] 30 Oct 2019 Fpit,Cmatﬁdjacobians

arXiv:1907.03330v3 [math.AG] 30 Oct 2019 fpit,Cmatfidjacobians. Compactified points, of where ue hrceitco hc scmue n[¨9] ec egtthe get we series: generating Hence [Gö90]. the in and computed curves rational is which of characteristic Euler system l ainlfiesaendl n hs r h aoa uvsw w we p curves of raional scheme the Hilbert the are to these birational is and jacobian nodal, compactified character relative are degene Euler maximally fibers the of rational number that all the is equals idea jacobian compactified The relative Jacobian. compactified relative ubro ainlcre na nerllna ytmon system linear integral an in curves rational of number e od n phrases. and words Key 2010 Let ONIGRTOA UVSO 3SRAE WITH SURFACES K3 ON CURVES RATIONAL COUNTING N S X ahmtc ujc Classification. Subject Mathematics n nato yafiiegroup finite a by action an ibr cee fpit nasot rjciesurface projective smooth a on points of schemes Hilbert Abstract. pcs ntecs that case the In spaces. P h eutn eeaigfnto so h form the of is function generating resulting the representation. a as for in surface an dition over K3 curves a cohomo of on the family to system tautological points the of of schemes Jacobian Hilbert compactified of cohomology the relate We |L| ∞ =0 ( n ∞ easot rjcieK ufc over surface K3 projective smooth a be n =0 , N stenme frtoa uvscnandi an in contained curves rational of number the is ) C P ( n n i ∞ stetuooia aiyo uvsover curves of family tautological the is ) =0 t n ( − G ¨ tsh aeafruafrtedmnino h ooooyof cohomology the of dimension the for formula a Göttsche gave = 1) obto uvswt oa iglrte o ocnrbt othe to contribute to not singularities nodal with curves of -orbit X n i ∞ H =0 i uv onig 3srae,Goprpeettos ibr sc Hilbert representations, Group surfaces, K3 counting, Curve ( e IIEGOPACTIONS GROUP FINITE S ( [ J n ] n , S C ( C ) G saK ufc,ec lmn of element each surface, K3 a is q n G n 1. )) rpeettos ial,w ieasffiin con- sufficient a give we Finally, -representations. edsrb h cinof action the describe we , When . t ALNZHAN SAILUN n Introduction = 41,1J8 43,14J50. 14J32, 14J28, 14J15, X n ∞ =0 G 1 csfihul n ypetclyon symplectically and faithfully acts e ( S [ n q/f ] ) t n ( q C = ,where ), n[Z6 n Ba9,the [Bea99], and [YZ96] In . |L| n Y ∞ =1 G (1 ihfiesbigintegral, being fibers with S S f ntecohomology the on When . G − ( scluae sn the using calculated is q ie rc on trace a gives t sacs form. cusp a is ) n n dmninllinear -dimensional ) − erllinear tegral 24 oyo the of logy S = it of oints n.Btthe But ant. admits aefiesif fibers rate ∆( si fthe of istic ubrof number t S t ) , S hemes the , 2 SAILUN ZHAN

[n] n 24 S is the Hilbert scheme of n points of S and ∆(t)= t n≥1(1 − t ) is the unique cusp form of weight 12 for SL2(Z). Q In this paper G will always be a finite group. We will consider a smooth projective K3 surface over C with a G-action, and ask whether we can prove a similar equality for G-representations. Since the main idea is to count the number of rational points in finite fields and then use comparison theorems between singular cohomology and l-adic cohomology, we will also consider the situation in characteristic p. i i H (·, C) denotes the singular cohomology and H (·, Ql) denotes the l-adic cohomology. We will consider power series with coefficients lying in the ring of virtual graded

G-representations Rk(G), of which the elements are the formal diﬀerences of isomorphism classes of ﬁnite dimensional graded k-representations of G. The addition is given by direct sum and the multiplication is given by tensor product.

The main results are as follows.

Theorem 1.1. Let S be a smooth projective surface over Fq with a G-action, where q is a power of p. Suppose p ∤ |G|. Let S[n] be the Hilbert scheme of n points of S, and let S(n) be the n-th symmetric power of S. Then we have the following equality as virtual graded G-representations.

∞ ∞ ∞ ∗ [n] n ∗ (n) mn [H (S , Ql)]t = [H (S , Ql)][−2n(m − 1)]t Fp Fp n=0 m=1 n=0 ! X Y X

b1 b3 ∞ i 1 − mi i 3 − mi i=0[(∧ H ) ⊗ Ql,qi(m 1) ]t i=0[(∧ H ) ⊗ Ql,qi(m 1) ]t = m b2 i i 2 mi m m=1 1 − [QPl,qm−1 ]t (−1) [(∧ H )P⊗ Q i(m−1) ]t 1 − [Ql,qm+1 ]t Y i=0 l,q P ∗ i where the coeﬃcients lie in RQl (G), H := ⊕H and [−2n(m − 1)] indicates shift i j i j n in degrees. ∧ H means ∧ H (SFp , Ql) of degree ij, and Ql,q denotes the one- dimensional trivial G-representation Ql[−2n]. In particular, the above equalities hold in RC(G)[[t]] for the singular cohomology if S is a smooth projective surface over C with a G-action by comparison theorems.

∞ ∞ ∞ [H∗(S[n], C)]tn = [H∗(S(n), C)][−2n(m − 1)]tmn n=0 m=1 n=0 ! X Y X

b1 b3 ∞ i 1 − mi i 3 − mi i=0[(∧ H ) ⊗ Cqi(m 1) ]t i=0[(∧ H ) ⊗ Cqi(m 1) ]t = . m b2 i i 2 mi m m=1 1 − [PCqm−1 ]t (−1) [(∧ H )P⊗ C i(m−1) ]t 1 − [Cqm+1 ]t Y i=0 q P COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 3

Corollary 1.2. Let S be a smooth projective surface over C with a G-action. If we ﬁx i ≥ 0, then Hi(S[n], C) become stable for n ≥ i as G-representations.

Theorem 1.3. Let X and Y be smooth projective algebraic varieties over C with a G-action. If X and Y have trivial canonical bundles and there is a birational map f : X → Y which commutes with the G-action, then

H∗(X, C) =∼ H∗(Y, C) as graded G-representations.

Note that Theorem 1.3 is the ﬁnite group actions version of Batyrev’s result [Ba99]. As explained in [Ba99, Prop 3.1], there exists U ⊂ X (resp. V ⊂ Y ) which is the maximal open subset where f (resp. f −1) is deﬁned, and we have f : U → V is an isomorphism. The requirement of commuting with the G-action in Theorem 1.3 means that U, V are G-stable, and f : U → V commutes with the G-action.

Recall that a linear system |L| is called an integral linear system if every effective divisor in it is integral. |L| is called G-stable if G induces an action on the projective space |L|, which means G maps an effective divisor in |L| again to an effective divisor in |L|.

Corollary 1.4. Let S be a smooth projective K3 surface over C with a G-action, and let Cn be the tautological family of curves over any n-dimensional integral G- stable linear system. Then we have the following equalities as virtual graded G- representations.

∞ ∞ ∞ ∞ n n [n] n (n) mn (∗) [e(J (Cn))]t = [e(S )]t = [e(S )][−2n(m − 1)]t n=0 n=0 m=1 n=0 ! X X Y X

∞ b1 i i 1 mi b3 i i 3 mi i=0(−1) [∧ H ][−2i(m − 1)]t i=0(−1) [∧ H ][−2i(m − 1)]t = m b2 i i 2 mi m m=1 1 − CP[−2(m − 1)]t (−1) [∧ H ][−P2i(m − 1)]t 1 − C[−2(m + 1)]t Y i=0 P i i i j where the coefficients lie in RC(G), e(X) means i(−1) H (X, C), and ∧ H means i j ∧ H (S, C) of degree ij. P Remark 1.5. Note that we are fixing the surface S here, so the equality above should be understood as: if S admits an n-dimensional integral G-stable linear n n system, then [e(J (Cn))] equals the coefficient of t on the right hand side. 4 SAILUN ZHAN

Recall that for a complex K3 surface S with an automorphism g of ﬁnite order 0 n, H (S,KS) = CωS has dimension 1, and we say g acts symplectically on S if it

acts trivially on ωS, and g acts non-symplectically otherwise, namely, g sends ωS to k ζnωS, 0

Theorem 1.6. Let G be a ﬁnite group which acts faithfully and symplectically on a complex K3 surface S. Then

∞ ∞ ∞ k mk n ǫ(ord(g ))t Tr(g)| [n] t = exp [e(S )] k n=0 m=1 ! X X Xk=1

−1 1 for all g ∈ G, where ǫ(n)=24 n p|n 1+ p . In particular, if G is generated by a single element g of order N ≤Q8, then we have

∞ n N n=0 Tr(g)|[e(S[n])]t 24 1 P t/η(t) 2 t/η(t)8η(t2)8 3 t/η(t)6η(t3)6 4 t/η(t)4η(t2)2η(t4)4 5 t/η(t)4η(t5)4 6 t/η(t)2η(t2)2η(t3)2η(t6)2 7 t/η(t)3η(t7)3 8 t/η(t)2η(t2)η(t4)η(t8)2

1/24 ∞ n where η(t)= t n=1(1 − t ). Remark 1.7. QNotice that if g acts symplectically on S, then g has order ≤ 8 by [H16, Corollary 15.1.8]. We know the generating series for topological Euler ∞ [n] n ∞ n characteristic n=0 e(S )t = n=0 Tr(1)|[e(S[n])]t equals t/∆(t), where ∆(t) = 24 η(t) is a levelP 1 cusp form ofP weight 12. But by Theorem 1.6, we deduce that when an element g of order N acts faithfully and symplectically on S, we have ∞ n n=0 Tr(g)|[e(S[n])]t = t/F (t), where F (t) is also a cusp form for Γ0(N) of weight ⌈ 24 ⌉ ([CS17, Proposition 5.9.2]). This coincides with the results by Jim Bryan and PN+1 Ad´am´ Gyenge in [BG19] when G is a cyclic group. See also [BO18, Lemma 3.1]. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 5

Theorem 1.8. Let G = hgi be a ﬁnite group generated by an automorphism g of order p, which acts non-symplectically on a complex K3 surface S. Then we have

∞ ∞ −24 n m Tr(1)|[e(S[n])]t = (1 − t ) n=0 m=1 ! X Y ∞ ∞ dp−24 ∞ −d n m mp Tr(g)|[e(S[n])]t = (1 − t ) (1 − t ) n=0 m=1 ! m=1 ! X Y Y rank T (g) 2 g ⊥ for all g ∈ G, g =16 , where d = p−1 , and T (g) := (H (S, Z) ) is the orthogonal complement of the g-invariant sublattice.

For the left hand side of (∗), recall that in [Bea99], for the curve C in Cn, we have e(J n(C)) = 0 if the normalization C˜ has genus ≥ 1, and e(J n(C)) = 1 if C is n a nodal rational curve. Hence intuitively that is why e(J (Cn)) counts the number of rational curves in Cn if we assume all rational curves in Cn are nodal. But in our situation, e(J n(C)) = 0 does not mean [e(J n(C))] = 0 as G-representations. n Hence non-rational curves may also contribute to [e(J (Cn))], and certain G-orbits of curves contribute certain representations (See Example 1, 2 and 3). Nevertheless, we show that a G-orbit of curves with nodal singularities will contribute nothing if the normalization of the curve quotient by its stablizer is not rational. By this method, we are able to understand certain G-orbits in the linear system. We denote by [e(X)] the alternating sum of the compactly supported l-adic cohomology n n n=0(−1) [Hc (XFp , Ql)] when we are in the situation of characteristic p.

PTheorem 1.9. Let C be an integral curve over Fp with nodal singularities and a G-action. Suppose p ∤ |G|. Denote by C˜ its normalization. If C/˜ hgi is not a rational 1 n curve (i.e. P ) over Fp for every g ∈ G, then [e(J C)]=0 as G-representations.

Corollary 1.10. Let C be an integral curve over Fp with nodal singularities and a G-action. Suppose p ∤ |G|. Denote by C˜ its normalization. If C/G˜ is not a rational 1 n curve (i.e. P ) over Fp, then [e(J C)]=0 as G-representations.

n By the above discussions, we show that the representation [e(J (Cn))] actually

‘counts’ the curves in Cn whose normalization quotient by its stablizer is rational (See Example 1, 2, and 3).

This paper is organized as follows. Section 2 explains how we can compare cohomology groups using the Weil conjectures. In Section 3, we work with Hilbert schemes of points and prove Theorem 1.1 and Corollary 1.2. In Section 4, we review 6 SAILUN ZHAN

the method of p-adic integrals in [Ba99] and prove Theorem 1.3 by applying it in our case. In Section 5, we deal with compactiﬁed jacobians and prove Theorem 1.9 and Corollary 1.10. In Section 6, we prove Corollary 1.4, Theorem 1.6 and Theorem 1.8 by using the results in previous sections. Then we give three explicit examples when

G equals Z/2Z or Z/3Z. Finally, when G = P SL(2, 7), A6, A5 or S5, we determine the smooth projective curve C over C with a faithful G-action if there exists g ∈ G such that C/hgi = P1.

Acknowledgements

I thank my advisor Professor Michael Larsen for his guidance and valuable discussions throughout this work, and in particular, for suggesting the problem. I also thank Professor Jim Bryan for several helpful suggestions and comments.

2. Preliminaries

Let X be a smooth projective variety over C. Then we can choose a ﬁnitely ∼ generated Z-subalgebra R ⊂ C such that X = X ×S SpecC for a regular projective scheme X over S = SpecR, and we can choose a maximal ideal q of R such that X has good reduction modulo q. We denote by X¯ the smooth projective variety over

Fq after reduction, where q is some power of a prime number p. ¯ ¯ Denote X ⊗Fq Fp by XFp . By comparison theorems between étale cohomology and i ¯ ∼ singular cohomology for smooth projective varieties, we have H (XFp , Ql) ⊗ C = i ∼ i H (X, Ql) ⊗ C = H (X, C), and this isomorphism is compatible with the G-action by functoriality if there is a finite group G acting on X. Also by the fact that i ¯ ∼ i ¯ G H (XFp /G, Ql) = H (XFp , Ql) for a projective variety X with a finite group G- action, we have

i ¯ ∼ i ¯ G ∼ i G ∼ i H (XFp /G, Ql) ⊗ C = H (XFp , Ql) ⊗ C = H (X, C) = H (X/G, C) for a smooth projective X.

Now we ﬁrst recall the proof of the following well-known fact, the idea of which will be used throughout the paper.

Proposition 2.1. Let X and Y be two smooth projective varieties over C. By ¯ ¯ considering good reductions of X and Y modulo some q, if |X(Fqn )| = |Y (Fqn )| for every n ≥ 1, then Hn(X, C) ∼= Hn(Y, C) as vector spaces over C. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 7 ¯ Proof. By the Lefschetz ﬁxed point formula for Frobenius, we have |X(Fqn )| = ∞ i i n ¯ n i=0(−1) Tr(Fq ,H (XFp , Ql)), where Fq denotes the geometric Frobenius. Let αPn,i, i =1, 2, ..., an (resp. βn,i, i =1, 2, ..., bn) denote the eigenvalues of Fq acting on n ¯ n ¯ H (XFp , Ql) (resp. H (YFp , Ql)), where an (resp. bn) is the n-th betti number. Then ¯ ¯ ∞ i ai n ∞ i bi n since |X(Fqn )| = |Y (Fqn )|, we have i=0(−1) j=1 αi,j = i=0(−1) j=1 βi,j for + n every n ≥ 1. Then by linear independenceP of theP charactersPχα : Z →PC, n 7→ α i/2 and taking into consideration that αi,j, βi,j, j =1, 2, ... all have absolute value q by

Weil’s conjecture, we deduce that for each i, the list of eigenvalues αi,1, ..., αi,ai must

be the same as the list of eigenvalues βi,1, ..., βi,bi up to reordering, which implies, in

particular, ai = bi for each i. Hence the proposition follows. Since we only know that the multiplicities of the eigenvalues of two sides are the same and we do not know whether the Frobenius action is semisimple, we n ¯ ∼ n ¯ cannot get H (XFp , Ql) = H (YFp , Ql) as Galois representations. Actually what ∗ ¯ ∗ ¯ we proved above is that [H (XFp , Ql)] = [H (YFp , Ql)] in the graded Grothendieck

group G0(Ql[GalFq ]), which is the abelian group generated by the set {[X]|X ∈

graded Ql[GalFq ] − Mod} of isomorphism classes of ﬁnitely generated graded Galois modules modulo the relations [A] − [B] + [C] = 0 if there is a graded short exact sequence 0 → A → B → C → 0. Using the same idea, we can prove Macdonald’s formula [M62] by the Weil conjectures.

Proposition 2.2. Let X be a smooth projective variety of dimension N over C. Then by choosing some good reduction of X over q, we have

(−1)j+1 ∞ 2N bj ∗ ¯ (k) k j+1 [H (X , Ql)]t = (1+(−1) [Ql,αj,i ]t) Fp   j=0 i=1 Xk=0 Y Y  

where the coefficients lie in G0(Ql[GalFq ]), αj,i denote the eigenvalues of Fq acting j ¯ on H (XFp , Ql), and Ql,αj,i denotes the one-dimensional Galois representation of degree j with eigenvalue αj,i by the geometric Frobenius Fq. ¯ ¯ (k) Proof. First we note that for a smooth projective variety X over Fq, X does have the purity property. Namely, the absolute value of the eigenvalues of Frobe- (k) (k) n ¯ n/2 n ¯ ∼ n ¯ k Sk nius Fq on H (X , Ql) is q . This is because H (X , Ql) = H (X , Ql) ⊂ Fp Fp Fp n k k H (X¯ , l), and X¯ is smooth projective, hence pure. Now by the Weil conjectures, Fp Q 8 SAILUN ZHAN we have (−1)j+1 ∞ r ∞ 2N bj ¯ t ¯ (k) k exp( |X(Fqr )| )= |X (Fq)|t = (1 − αj,it) . r   r=1 j=0 i=1 X Xk=0 Y Y   The first equality is a combinatorial fact [Gö94, Remark 1.2.4]. Let {βk,n,i} denote n ¯ (k) the eigenvalues of Fq on H (X , Ql). Then we have Fp

(−1)j+1 ∞ ∞ 2N bj (−1)m β tk = (1 − α t) . k,m,i  j,i  m=0 i j=0 i=1 Xk=0 X X Y Y Replacing q by qn, we get the following equality for every n ≥ 1

(−1)j+1 ∞ ∞ 2N bj (−1)m βn tk = (1 − αn t) . k,m,i  j,i  m=0 i j=0 i=1 Xk=0 X X Y Y   Note that if we deﬁne the weight of α as 2 logq |α|, then αj,i has an odd weight if j is odd and there is a minus sign before it. On the other hand, αj,i has an even weight ∞ m if j is even and there is a plus sign before it since 1/(1 − αt)= m=0(αt) . On the left hand side, the same situation holds, namely if the eigenvaluesP have odd (resp. even) weights, then there are minus (resp. plus) signs before them. Hence by linear independence of the characters, we deduce the desired equality in G0(Ql[Fq]).

∞ n n Recall the deﬁnition of the Poincar´epolynomial P (X, z) := n=0 dimH (X, C)z . Corollary 2.3. Let X be a smooth projective variety ofP dimension N over C. Then the generating series of Poincare´ polynomials of X(n) is given by

∞ 2N j+1 (−1) bj P (X(n), z)tn = 1+(−1)j+1zj t . n=0 j=0 X Y Proof. Take the Poincar´epolynomial of both sides of the equality.

Corollary 2.4. Let X be a smooth projective curve, an abelian variety, or a smooth projective K3 surface over C. Denote the dimension of X by N. Then by choosing some good reduction of X over q, we have

(−1)j+1 ∞ 2N bj ∗ ¯ (k) k (j+1)i i j ¯ i [H (X , Ql)]t = (−1) [∧ H (X , Ql)]t Fp  Fp  j=0 i=0 Xk=0 Y X  i j ¯  where the coeﬃcients lie in RQl (GalFq ), and ∧ H (XFp ) has degree ij. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 9

n ¯ Proof. This follows from the fact that the action of the Frobenius on H (XFp , Ql) is semisimple if X is a smooth projective curve, an abelian variety or a K3 surface n ¯ (k) [De72]. Hence the action of the Frobenius on H (X , Ql) is also semisimple since Fp

k Sk n ¯ (k) ∼ n ¯ k Sk ∼ ni ¯ H (X , Ql) = H (X , Ql) = H (X , Ql) . Fp Fp Fp n +...+n =n i=1 ! 1 Mk O So the equality in Proposition 2.2 is actually an equality as graded Galois modules.

Now if we have a smooth projective variety with ﬁnite group actions, we then need a lemma corresponding to Proposition 2.1. The idea is that if X is a quasi-projective

variety over Fp with an automorphism σ of ﬁnite order, then X and σ can be deﬁned over some Fq. Let Fq be the corresponding geometric Frobenius. Then for n ≥ 1, n the composite Fq ◦ σ is the Frobenius map relative to some new way of lowering the

field of definition of X from Fp to Fqn ([DL76, Prop.3.3] and [Ca85, Appendix(h)]). ∞ k n ∗ k Then the Grothendieck trace formula implies that k=0(−1) Tr((Fq σ) ,Hc (X, Ql)) n is the number of fixed points of Fq σ. P Proposition 2.5. Let X and Y be two smooth projective varieties over C with finite group G-actions. By considering good reductions of X and Y modulo some q

¯ gFqn ¯ gFqn such that G-actions can be deﬁned over Fq, if |X(Fp) | = |Y (Fp) | for every n ≥ 1 and g ∈ G, then Hn(X, C) =∼ Hn(Y, C) as G-representations for every n.

Proof. Fix g ∈ G. Since g commutes with Fq and the action of g on the cohomology group is semisimple, there exists a basis of the cohomology group such

that the actions of g and Fq are in Jordan normal forms simultaneously. Let

αn,i, i = 1, 2, ..., an (resp. βn,i, i = 1, 2, ..., bn) denote the eigenvalues of Fq acting n ¯ n ¯ on H (XFp , Ql) (resp. H (YFp , Ql)) in such a basis, where an (resp. bn) is the n-th betti number. Let cn,i, i = 1, 2, ..., an (resp. dn,i, i = 1, 2, ..., bn) denote the n ¯ n ¯ eigenvalues of g acting on the same basis of H (XFp , Ql) (resp. H (YFp , Ql)). Then since ∞ gF n i ∗ i ¯ q n ¯ |X(Fp) | = (−1) Tr((gFq ) ,H (XFp , Ql)) i=0 X ¯ gFqn ¯ gFqn and |X(Fp) | = |Y (Fp) | for every n ≥ 1, we have

∞ ai ∞ bi i n i n (−1) ci,jαi,j = (−1) di,jβi,j i=0 j=1 i=0 j=1 X X X X 10 SAILUN ZHAN

for every n ≥ 1. By the linear independence and eigenvalue discussions as before, we ai bi have ai = bi and j=1 ci,j = j=1 di,j for each i. But since g is arbitrary, this means that the characters for the G-representations Hi(X¯ , ) and Hi(Y¯ , ) are the P P Fp Ql Fp Ql same. Hence Hn(X, C) ∼= Hn(Y, C) as G-representations for every n by comparison theorems.

By a similar argument as in the proof of Proposition 2.2, we obtain

Corollary 2.6. Let X be a smooth projective curve, an abelian variety, or a smooth projective K3 surface over C. Denote the dimension of X by N. Then by choosing some good reduction of X over q such that the G-action can be deﬁned over

Fq, we have

(−1)j+1 ∞ 2N bj [H∗(X(k), C)]tk = (−1)(j+1)i[∧iHj(X, C)]ti   j=0 i=0 Xk=0 Y X   where the coeﬃcients lie in RC(G). Proof. Similar as before, except that now we have

(−1)j+1 ∞ ∞ 2N bj (−1)m h βn tk = (1 − g αn t) k,m,i k,m,i  j,i j,i  m=0 i j=0 i=1 Xk=0 X X Y Y   m ¯ (k) where hk,m,i are the eigenvalues of g on H (X , Ql), and gj,i are the eigenvalues of Fp j ¯ g on H (XFp , Ql). Hence we deduce that the trace of g on the left hand side equals the trace of g on the right hand side for the equality in Corollary 2.6.

3. Hilbert scheme of points

Let X[n] denote the component of the Hilbert scheme of X parametrizing subschemes of length n of X. For properties of Hilbert scheme of points, see references [I77], [Gö94] and [N99]. The following theorem is proved for smooth projective surfaces over C in [Gö90], and for quasi-projective surfaces over C in [GS93]. Theorem 3.1. The generating function of the Poincarépolynomials of the Hilbert scheme X[n] is given by

∞ ∞ (1 + z2m−1tm)b1(X)(1 + z2m+1tm)b3(X) p(X[n], z)tn = . (1 − z2m−2tm)b0(X)(1 − z2mtm)b2(X)(1 − z2m+2tm)b4(X) n=0 m=1 X Y COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 11

By analyzing the structure of X[n] [G¨o90, Lemma 2.9], we know that

∞ ∞ ∞ ¯ [n] n ¯ (n) kn(m−1) nm |X (Fqk )|t = |X (Fqk )|q t n=0 m=1 n=0 ! X Y X for some q and every k ≥ 1, which implies

∞ ∞ ∞ ∗ ¯ [n] n ∗ ¯ (n) nm [H (X , Ql)]t = [H (X , Ql)](−n(m − 1))[−2n(m − 1)]t . Fp Fp n=0 m=1 n=0 ! X Y X m Now using Proposition 2.2 and replacing t by [Ql,qm−1 ]t , we have

∞ ∞ b1 m b3 m (1 + [ ⊗ m−1 ]t ) (1 + [ ⊗ m−1 ]t ) ∗ ¯ [n] n i=1 Ql,α1,i Ql,q i=1 Ql,α3,i Ql,q [H (X , Ql)]t = . Fp m b2 m m (1 − [ m−1 ]t ) (1 − [ ⊗ m−1 ]t )(1 − [ m+1 ]t ) n=0 m=1 Q Ql,q i=1 Ql,α2Q,i Ql,q Ql,q ! X Y Q Taking the Poincarépolynomials of both sides, we obtain Theorem 3.1 for smooth projective surfaces. We notice that the first term in the product on the right hand side of Theorem 3.1 is the generating series of the Poincarépolynomial for X(n), which seems like a coincidence at first glance. But actually each term is some twisted generating series of the Poincarépolynomial for X(n).

Now suppose we have a smooth projective surface over C with a G-action, and we want a similar equality as above. By considering some good reduction of the surface, we can restrict ourselves to the characteristic p and p ∤ |G|. The following discussion will be used to deduce the key Lemma 3.3.

Let S be a smooth projective surface over Fq with an automorphism g over Fq of gFq ﬁnite order. If x ∈ S(Fq) where Fq is the geometric Frobenius, then x lies over a closed point y ∈ S. Denote the residue degree of y by N. Hence x ∈ S(FqN ) and N−1 there are N geometric points x, Fq(x), ..., Fq (x) lying over y. Let us study the relative Hilbert scheme of n points at a closed point.

n ∼ n Hilb (Spec(OS,y)/SpecFq) = Hilb (Spec(FqN [[s, t]])/SpecFq).

d Since g and Fq ﬁx y, they act on this Hilbert scheme. Over Fq, we have

n ∼ n Hilb (Spec(OS,y)/SpecFq) ⊗Fq Fq = Hilb (Spec(Fq ⊗Fq FqN [[s, t]])/SpecFq)

by the base changed property of the Hilbert scheme. Denote by u a primitive element

of the ﬁeld extension FqN /Fq and denote by f(x) the irreducible polynomial of u 12 SAILUN ZHAN

over Fq. Since we have an Fq-algebra isomorphism ∼ ∼ qN−1 Fq ⊗Fq FqN = Fq ⊗Fq (Fq[x]/(f(x))) = Fq(x)/(x − u) × ... × Fq(x)/(x − u ) by the Chinese Remainder Theorem, we deduce that n ∼ n Hilb (Spec(OS,y)/SpecFq) ⊗Fq Fq = Hilb (Spec((Fq × ... × Fq)[[s, t]])/SpecFq) ∼ n d = Hilb ( SpecFq[[s, t]]/SpecFq). n a Hence the Fq-valued points of Hilb (Spec(OS,y)/SpecFq) correspond to the closed

subschemes of degree n of SpecFq[[s, t]], i.e. the closed subschemes of degree n of d N−1 S whose underlying space` is a subset of the points x, Fq(x), ..., Fq (x). q q q Since Fq acts on FqN [[s, t]] by sending s to s , t to t and c ∈ FqN to c , we deduce q from the above discussion that Fq acts on (Fq × ... × Fq)[[s, t]] by sending s to s , t q to t and (α0,α1, ..., αN−2,αN−1) ∈ Fq × ... × Fq to (α1,α2, ..., αN−1,α0).

Let σ be an element of Gal(FqN /Fq). Recall that for an FqN -vector space V , a σ-linear map f : V → V is an additive map on V such that f(αv) = σ(α)f(v) for all α ∈ FqN and v ∈ V .

Lemma 3.2. Let H = hgi. Suppose p ∤ |H|, then we can choose s and t such that g acts on FqN [[s, t]] σ-linearly, where σ is the inverse of the Frobenius automorphism

of Gal(FqN /Fq).

Proof. g acts as an Fq-automorphism on FqN [[s, t]] sending (s, t) to (s, t) and

FqN to FqN . Since we know Fq sends (α0,α1, ..., αN−2,αN−1) ∈ Fq × ... × Fq to N−1 (α1,α2, ..., αN−1,α0) and gFq ﬁxes the geometric points x, Fq(x), ..., Fq (x), we de-

duce that g sends (α0,α1, ..., αN−2,αN−1) ∈ Fq ×...×Fq to (αN−1,α0, ..., αN−3,αN−2).

Hence g(α)= σ(α) for all α ∈ FqN where σ is the inverse of the Frobenius automorphism.

Now we write g(s)= as+bt+... and g(t)= cs+dt+... where a,b,c,d ∈ Fq since g

commutes with Fq. Deﬁne an automorphism ρ(g) of FqN [[s, t]] by ρ(g)(s)= as + bt,

ρ(g)(t) = cs + dt and the action of ρ(g) on FqN is the same as the action of g. 1 −1 Then we denote the FqN -automorphism |H| h∈H hρ(h) by θ. Notice that θ is an automorphism because the linear term of θ Pis an invertible matrix. We deduce that gθ = θρ(g), which implies θ−1gθ = ρ(g). Hence we are done.

Notice that Lemma 3.2 is the only place we use the assuption p ∤ |G| for Theorem 1.1. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 13

From Lemma 3.2, we now know that the g-action on (Fq ×...×Fq)[[s, t]] is given by sending s to (a, ..., a)s+(b, ..., b)t, t to (c, ..., c)s+(d, ..., d)t and (α0,α1, ..., αN−2,αN−1) ∈

Fq × ... × Fq to (αN−1,α0, ..., αN−3,αN−2).

q Hence the action of gFq on (Fq ×...×Fq)[[s, t]] is given by sending s to (a, ..., a)s + q q q (b, ..., b)t , t to (c, ..., c)s +(d, ..., d)t and (α0,α1, ..., αN−2,αN−1) ∈ Fq × ... × Fq to itself. Hence gFq acts on each complete local ring, which is what we expected since gF ﬁxes each geometric point over y. In particular, it acts on O\ ∼= q SFq ,x ∼ (Fq ×{0} × ... ×{0})[[s, t]] = Fq[[s, t]].

n Recall that Hilb (O\S ,x)( q) parametrizes closed subschemes of degree n of S Fq F Fq supported on x.

Lemma 3.3.

|Hilbn(O\)( )gFq | = |Hilbn( [[s, t]])( )Fq | SFq ,x Fq Fq Fq

Proof. First we deﬁne an Fq-automorphismg ˜ on Fq[[s, t]] by

g˜(s)= as + bt andg ˜(t)= cs + dt

q Recall that the action of Fq on Fq[[s, t]] is an Fq-endomorphism sending s to s and t q to t . By Lemma 3.2 and the above discussion, we observe that the action of gFq on

Fq[[s, t]] on the left is the same as the action ofgF ˜ q on Fq[[s, t]] on the right. Hence we have

n gFq n gF˜ q |Hilb (O\S ,x)( q) | = |Hilb ( q[[s, t]])( q) |. Fq F F F

Now for the right hand side,g ˜ is an automorphism of ﬁnite order and Fq is the geometric Frobenius. Then by the Grothendieck trace formula, we have

∞ n gF˜ q k ∗ k n |Hilb (Fq[[s, t]])(Fq) | = (−1) Tr((˜gFq) ,Hc (Hilb (Fq[[s, t]]), Ql)). Xk=0

But the action ofg ˜ factors through GL2(Fq). Now we use the fact that if G is a connected algebraic group acting on a separated and ﬁnite type scheme X, then the ∗ action of g ∈ G on Hc (X, Ql) is trivial [DL76, Corollary 6.5]. Hence we have ∞ n gF˜ q k ∗ k n |Hilb (Fq[[s, t]])(Fq) | = (−1) Tr((Fq) ,Hc (Hilb (Fq[[s, t]]), Ql)) Xk=0 n Fq = |Hilb (Fq[[s, t]])(Fq) |. 14 SAILUN ZHAN

n gFq From Lemma 3.3 we observe that |Hilb (O\S ,x)( q) | is a number independent Fq F of the choice of the gFq-ﬁxed point x. Lemma 3.4.

∞ ∞ ∞ |Pr(S,gFq)| r r [n] gFq n n g Fq nr |S ( q) |t = |Hilb (O\S ,x)( q) |t F Fq F n=0 r=1 n=0 ! X Y X where Hilbn(O\) is the punctual Hilbert scheme of n points at some grF r-ﬁxed SFq ,x q point x ∈ S(Fq), and Pr(S,gFq) denotes the set of primitive 0-cycles of degree r of r r j j r−1 i i g Fq g Fq gFq on S, whose elements are of the form i=0 g Fq (x) with x ∈ S(Fq) \(∪j

[n] gFq Proof. Let Z ∈ S (Fq) . SupposeP (n1, ..., nr) is a partition of n and Z =

(Z1, ..., Zr) with Zi being the closed subscheme of Z supported at a single point with length ni. Then SuppZ decomposes into gFq orbits. We can choose an ordering ≤ on

S(Fq). In each orbit, we can find the smallest xj ∈ S(Fq). Suppose Zj with length l is supported on xj and xj has order k. Then the component of Z which is supported k−1 i i on the orbit of xj is determined by Zj, namely, it is ∪i=0 g Fq (Zj) with length kl. k k [n] gFq Also notice that Zj is fixed by g Fq . Hence, to give an element of S (Fq) is the same as choosing some gFq orbits and for each orbit choosing some element in k k Hilbn(O\)( )g Fq for some gkF k-fixed point x in this orbit such that the final SFq ,x Fq q length altogether is n. Combining all of these into power series, we get the desired equality.

The idea we used above is explained in detail in [Gö90, lemma 2.7]. We im- [n] plicitly used the fact that π : (S(n))red → S is a locally trivial fiber bundle in n [n] the Zariski topology with fiber Hilb (Fq[[s, t]])red [Gö94, Lemma 2.1.4], where S(n) parametrizes closed subschemes of length n that are supported on a single point. n Denote Hilb (Fq[[s, t]]) by Vn. Now combining Lemma 3.3 and Lemma 3.4, we have

∞ ∞ ∞ |Pr(S,gFq)| r [n] gFq n Fq nr |S (Fq) |t = |Vn(Fq) |t . n=0 r=1 n=0 ! X Y X

Recall the following structure theorem for the punctual Hilbert scheme of points [ES87, Prop 4.2]. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 15

Proposition 3.5. Let k be an algebraically closed ﬁeld. Then Hilbn(k[[s, t]]) over k has a cell decomposition, and the number of d-cells is P (d, n − d), where P (x, y) := #{partition of x into parts ≤ y}.

Denote by p(n, d) the number of partitions of n into d parts. Then p(n, d) = P (n − d,d). Now we can proceed similarly as in the proof of [G¨o90, Lemma 2.9].

Proof of Theorem 1.1 Since we have ∞ 1 ∞ ∞ = p(n, n − i)tnzi, 1 − zi−1ti i=1 n=0 i=0 Y X X by Proposition 3.5 we get ∞ ∞ ∞ 1 #{m-dim cells of Hilbn(k[[s, t]])}tnzm = . 1 − zi−1ti n=0 m=0 i=1 X X Y Let l ∈ N. Then by choosing suﬃciently large q powers Q such that the cell decom-

position of Vn,Fq is deﬁned over FQ, we have ∞ ∞ l nr t 1 |V (FQr )|t ≡ . n,Fq 1 − Qr(i−1)tri n=0 i=1 X Y Hence

∞ ∞ ∞ ∗ [n] n ∗ (n) mn [H (S , Ql)]t = [H (S , Ql)][−2n(m − 1)]t Fp Fp n=0 m=1 n=0 ! X Y X 16 SAILUN ZHAN

m as graded G-representations. Now by Corollary 2.6 and replacing t by Ql,qm−1 t , we have ∞ ∗ [n] n [H (S , Ql)]t Fp n=0 X b1 b3 ∞ i 1 − mi i 3 − mi i=0[(∧ H ) ⊗ Ql,qi(m 1) ]t i=0[(∧ H ) ⊗ Ql,qi(m 1) ]t = . m b2 i i 2 mi m m=1 1 − [QPl,qm−1 ]t (−1) [(∧ H )P⊗ Q i(m−1) ]t 1 − [Ql,qm+1 ]t Y i=0 l,q P

Corollary 3.6. For a smooth projective surface S over C, we have ∞ [e(S[n])]tn n=0 X

∞ b1 i i 1 mi b3 i i 3 mi i=0(−1) [∧ H ][−2i(m − 1)]t i=0(−1) [∧ H ][−2i(m − 1)]t = m b2 i i 2 mi m m=1 1 − CP[−2(m − 1)]t (−1) [∧ H ][−P2i(m − 1)]t 1 − C[−2(m + 1)]t Y i=0 where the coeﬃcients lie in RC(PG). Remark 3.7. Notice that the generating series of Euler numbers of S[n] is ∞ [n] n ∞ m −e(S) n=0 e(S )t = m=1(1 − t ) . But this is not the case if we consider G- ∞ m −[e(S)] ∞ m representationsP andQ regard m=1(1−t ) as exp( m=1[e(S)](− log(1−t ))) = ∞ ∞ mk exp( m=1[e(S)]( k=1 t /kQ))). What we have is actuallyP P ∞ P ∞ b1 m b3 m n ( i=1(1 − g1,it ))( i=1(1 − g3,it )) Tr(g)|[e(S[n])]t = m b2 m m n=0 m=1 (1 − t )( i=1(1 − g2,it ))(1 − t ) ! X Y Q Q Q ∞ ∞ tmk b1 b2 b3 = exp 1 − g k + gk − g k +1 . k 1,i 2,i 3,i m=1 i=1 i=1 i=1 !! X Xk=1 X X X We will use this expression to determine the G-representation [e(S[n])] later when S is a K3 surface.

Proof of Corollary 1.2 We let

∞ G(t) := (1 − t) [H∗(S[n], C)]tn n=0 X

∞ b1 i 1 mi b3 i 3 mi (1 − t) i=0[∧ H ][−2i(m − 1)]t i=0[∧ H ][−2i(m − 1)]t = m b2 i i 2 mi m m=1 1 − C[−2(m −P1)]t (−1) [∧ H ][−2iP(m − 1)]t 1 − C[−2(m + 1)]t Y i=0 P COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 17

j We denote by ai,j the degree i part of the coeﬃcient of t . If i < j, then ai,j(G(t))=0 as G-representation. Now ﬁx i, take n ≥ i, and then we have ∞ i [n] k H (S , C)= ai,n t G(t) k=0 n X

= ai,k(G(t)) k=0 X∞

= ai,k(G(t)) Xk=0 = ai,0(G(1)).

i [n] Notice that ai,0(G(1)) is a representation independent of n. Hence H (S , C) become stable as G-representations for n ≥ i.

4. Birational varieties with trivial canonical bundles

We ﬁrst review the idea in the proof of the following theorem.

Theorem 4.1. [Ba99] Let X and Y be smooth n-dimensional projective varieties n n over C. Assume that the canonical line bundles ωX and ωY are trivial and that X and Y are birational. Then X and Y have the same Betti numbers, that is,

Hi(X, C) =∼ Hi(Y, C) for all i ≥ 0.

Let F be a finite extension of the p-adic field Qp, R ⊂ F be the maximal compact r subring, q ⊂ R the maximal ideal, and Fq = R/q the residue field with |Fq| = q = p . Now let X be a smooth n-dimensional algebraic variety over F . Assume that X admits an extension X to a regular S-scheme, where S = SpecR. Then if the relative n n dualizing sheaf ΩX /S is trivial, there exists a global section ω ∈ Γ(X , ΩX /S) which has no zeros in X . This ω is called a gauge form, and it determines a canonical p- adic measure dµω on the locally compact p-adic topological space X (F ) of F -rational points. This dµω is called the Weil p-adic measure associated with the gauge form ω, and it is defined as follows:

Let x ∈ X (F )bean F -point and let t1, ..., tn be local p-adic analytic parameters at n x. Then t1, ..., tn deﬁne a p-adic homeomorphism θ : U → A (F ) of an open subset n ∗ U ⊂ X (F ) and θ(U) ⊂ A (F ) in the p-adic topology. Let ω = θ (gdt1 ∧ ... ∧ dtn), where g is a p-adic analytic function on θ(U) having no zeros. Then dµω on U is 18 SAILUN ZHAN

deﬁned to be the pullback with respect to θ of the p-adic measure ||g(t)||dt on θ(U),

where || · || can be taken as the absolute value on F extended from || · ||p on Qp, and dt is the standard p-adic Haar measure on An(F ) normalized by the condition dt = 1. An(R) ′ R Notice that if we choose another global section ω , then dµω and dµω′ may not agree on X (F ), but they agree on X (R) since ω′ = ωh for an invertible function h and h has p-adic norm 1 on X (R).

Hence in general even if we do not assume the canonical bundle is trivial, since it is locally trivial, we can still deﬁne a p-adic measure dµ at least on the compact X (R).

Now we list some properties of these measures [Ba99]:

Theorem 4.2. Let X be a regular S-scheme with a gauge form ω, and let dµω be the corresponding Weil p-adic measure on X (F ). Then

|X (Fq)| dµ = ω qn ZX (R) Theorem 4.3. Let X be a regular S-scheme, and let dµ be the p-adic measure on X (R). Then

|X (Fq)| dµ = qn ZX (R) Theorem 4.4. Let X be a regular S-scheme and let Z ⊂ X be a closed reduced subscheme of codimension ≥ 1. Then the subset Z(R) ⊂ X (R) has zero measure with respect to the canonical p-adic measure dµ on X (R).

Combining all of these, let X and Y be smooth projective birational varieties of dimension n over C with trivial canonical bundles. Then there exist Zariski open dense subsets U ⊂ X and V ⊂ Y with codimX (X\U) ≥ 2 and codimY (Y \V ) ≥ 2 and an isomorphism φ : U → V . By a standard argument, we can choose a ﬁnitely generated Z-subalgebra R ⊂ C such that ‘everything’ (e.g. X,Y,U,V ) can be deﬁned over SpecR and have good reductions over some maximal ideal q. Let ωX and ωY be gauge forms on X and Y respectively and ωU and ωV their restriction to U ∗ ∗ and V . Since φ ωV is another gauge form on U, φ ωV = hωU for some nowhere ∗ vanishing regular function h ∈ Γ(U, OX ). The property codimX (X\U) ≥ 2 implies ∗ that h ∈ Γ(X, OX ). Hence ||h(x)|| = 1 for all x ∈ X(F ) (notice that X(F )= X(R)), COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 19

∗ which means the Weil p-adic measures on U(F ) associated with φ ωV and ωU are the same. Hence

dµX = dµY . ZU(F ) ZV (F ) By Theorem 4.4, we have

dµX = dµX = dµX = dµY = dµY = dµY . ZX(R) ZX(F ) ZU(F ) ZV (F ) ZY (F ) ZY (R) Applying Theorem 4.2, we get

|X(Fq)| |Y (Fq)| = qn qn

|X(Fq)| = |Y (Fq)|.

Replacing R by its cyclotomic extension obtained by adjoining all (qr − 1)-th roots

of unity, the residue ﬁeld Fq will become Fqr and we get the same equality regarding

Fqr -rational points for r ≥ 1. Hence we have

Hi(X, C) =∼ Hi(Y, C) for all i ≥ 0.

Now let X and Y be smooth projective algebraic varieties over C with finite group G-actions. If X and Y have trivial canonical bundles and there is a birational map f : X → Y which commutes with G-actions, where the meaning of commuting with G-actions is explained after Theorem 1.3, then as above there exists R, which is the maximal compact subring in a local p-adic field, such that ‘everything’ (e.g. X,Y,U,V ) can be defined over SpecR and have good reductions over some maximal

σFq σFq ideal q. For a fixed σ ∈ G, we have X(Fq) ⊂ X(FqN ) and Y (Fq) ⊂ Y (FqN ) for some positive integer N. Let R′ be the integral closure of R in the unramified extension of F such that the residue field is FqN . Then we have Proposition 4.5.

σFq |X(Fq) | dµω = Nn ˜ ′ q ZX(R ) ˜ ′ ′ σFq where X(R )= {x ∈ X(R )|x¯ ∈ X(Fq) }.

′ Proof. Let φ : X(R ) → X(FqN ), x → x¯ be the reduction map. Ifx ¯ ∈ X(FqN ) is a

closed FqN -point and g1, ..., gn are generators of the maximal ideal of OX,x¯, then they deﬁne a p-adic analytic homeomorphism γ : φ−1(¯x) → An(q′), where φ−1(¯x) is the 20 SAILUN ZHAN

ﬁber of φ overx ¯ and An(q′) is the set of R′-rational points of An whose coordinates lie in q′. Now as in the proof of [Ba99, Theorem 2.5], we know that 1 dµω = dt = Nn . −1 n ′ q Zφ (¯x) ZA (q ) Hence we have σFq |X(Fq) | dµω = Nn . ˜ ′ q ZX(R )

Proof of Theorem 1.3. The Weil p-adic measures on U(F ′) associated with ∗ f ωV and ωU are the same, which implies the following equality

dµX = dµY ˜ ′ ˜ ′ ZU(F ) ZV (F )

˜ ′ ′ σFq ˜ ′ ′ σFq where U(F )= {x ∈ U(F )|x¯ ∈ X(Fq) }, V (F )= {x ∈ V (F )|x¯ ∈ Y (Fq) } and f induces a bijection between U˜(F ′) and V˜ (F ′).

Hence

dµX = dµX = dµY = dµY . ˜ ′ ˜ ′ ˜ ′ ˜ ′ ZX(R ) ZU(F ) ZV (F ) ZY (R ) Then by Proposition 4.5, we have

σFq σFq |X(Fq) | = |Y (Fq) |.

Replacing R by its cyclotomic extension and then by Proposition 2.5, we get

H∗(X, C) =∼ H∗(Y, C)

as graded G-representations.

5. Compactified Jacobians

Recall some facts from [AK76], [A04] and [EGK00]. Let C/S be a ﬂat projective family of integral curves. By a torsion-free rank-1 sheaf I on C/S, we mean an S-ﬂat

coherent OC -module I such that, for each point s of S, the ﬁber Is is a torsion-free

rank-1 sheaf on the ﬁber Cs. We say that I is of degree n if χ(Is) − χ(OCs )= n for each s. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 21

Given n, consider the ´etale sheaf associated to the presheaf that assigns to each locally Noetherian S-scheme T the set of isomorphism classes of torsion-free rank-

1 sheaves of degree n on CT /T . This sheaf is representable by a projective S- ¯n n n scheme, denoted JC/S. It contains J := PicC/S as an open subscheme. For every S-scheme T , we have a natural isomorphism J¯n = J¯n × T . If S = Speck for an CT /T C/S ¯n n algebraically closed field k, we denote JC/S by J C. Recall that at the beginning we are considering C, which is the tautological family of curves over an n-dimensional integral G-stable linear system. Since C has a stratification according to the geometric genus of the fibers and the G-action (see §6), we can temporarily focus our attention on J nC for a single singular curve C with a G-action (note that our G-action on J nC is given by pushing forward the torsion-free rank-1 sheaves). This is reasonable since we have

Lemma 5.1. Let X be an algebraic variety over Fp with a G-action, U an open subvariety of X, Z := X\U the closed subvariety. If both U and Z are G-stable, then

[e(X)] = [e(U)] + [e(Z)]

∞ n n as G-representations, i.e. in RQl (G), and [e(X)] := n=0(−1) [Hc (X, Ql)]. Proof. One way to see this is to consider theP bounded long exact sequence 0 0 0 0 → Hc (U) → Hc (X) → Hc (Z) → ... and check that Tr(g) = 0on −[e(U)] + [e(X)] − [e(Z)] for every g. Another way is to use the description of Tr(g) on e(X) without involving cohomology [Ca85, Appendix(h)].

n Now we have an integral curve C over Fp. Recall that J C parametrizes the isomorphism classes of torsion-free rank-1 sheaves of degree n on C, and we have the following facts [Bea99].

Proposition 5.2. Let C be an integral curve over an algebraically closed ﬁeld k.

n ′ (1) If L ∈ J C is a non-invertible torsion-free rank 1 sheaf, then L = f∗L , where L′ is some invertible sheaf on some partial normalization f : C′ → C.

′ n ′ (2) If f : C → C is a partial normalization of C, then the morphism f∗ : J C → J nC is a closed embedding.

Using these two facts, we obtain the following corollary. 22 SAILUN ZHAN

Corollary 5.3. Let all the singularities of an integral curve C be nodal singularities. Then J nC has the following stratiﬁcation

J nC = J nC′ ′ Ca→C where J nC parametrizes rank-1 torsion-free sheaves of degree n, and C′ goes through all partial normalizations of C (including C itself).

Now let J nC′ be some stratum which is preserved by G. We want to calculate the G-representation [e(J nC′)]. Here we need to make use of the short exact sequence of algebraic groups 0 → L → J nC′ → J nC˜′ → 0

where L is a smooth connected linear algebraic group [BLR90, §9 Corollary 11], and C′ is the normalization of C′. Since L is linear, we have that J nC′ is a principal Zariski ﬁber bundle over J nC˜′ [Se88, Chapter VII, Proposition 6]. Now we need to f prove the following lemma.

Lemma 5.4. Let B and F be separated schemes of ﬁnite type over Fp, and let G be a ﬁnite group acting on B × F and B such that the projection B × F → B is G-equivariant. Suppose B is connected. Then we have

[e(B × F )] = [e(B)][e(F )]

∗ as virtual G-representations, i.e. in RQl (G), where the action of g ∈ G on Hc (F, Ql) ∗ ∗ is given by Hc ({g(b)}×F, Ql) → Hc ({b}×F, Ql) by choosing any closed point b ∈ B. Proof. We begin with a homotopy argument. Fix g ∈ G. By assumption, we have a commutative diagram

g B × F −−−→ B × F

π π  g  B −−−→ B y y Hence we have a map φ =(g, π) from B ×F to the ﬁber product B ×F , which maps ∼ ∼ (b, f) to (b, gb(f)), where gb : F = {b} × F →{g(b)} × F = F , and the diagram

φ B × F −−−→ B × F

π π

 id  B −−−→ B y y COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 23

commutes. On the other hand, we have

B × F −−−→ F

π   B −−−→ Fp y y i i Hence R π!(Z/nZ) is the constant sheaf Hc(F, Z/nZ) on B. The automorphism i φ acts on it and, at b ∈ B it acts the way gb acts on Hc(F, Z/nZ). Since an endomorphism of a constant sheaf over a connected base is constant, the action of

φ is the same everywhere. Passing to limit, we deduce that the actions of gb on ∗ Hc (F, Ql) are the same for every b ∈ B. By the same idea we used before, now it suﬃces to prove the following:

gFQ gFQ gFQ |(B × F )(Fp) | = |B(Fp) ||F (Fp) |

But by what we just proved and the Lefschetz trace formula, we have |({b1} ×

gFQ gFQ F )(Fp) | = |({b2} × F )(Fp) |, where b1, b2 are ﬁxed points of gFQ. Hence the equality follows.

Corollary 5.5. Let B, E and F be separated schemes of finite type over Fp. Suppose E is a Zariski-locally trivial fiber bundle over B with fiber F and let G be a finite group acting on E and B, the action of which is compatible with the projection π : E → B. If B is irreducible, then we have

[e(E)] = [e(B)][e(F )]

as virtual G-representations, i.e. in RQl (G).

∗ Proof. Fixing g ∈ G, it suﬃces to prove that the action of gb1 on Hc (F, Ql) is

the same as the action of gb2 for any b1, b2 ∈ B ﬁxed by g. Take open neighborhoods ∞ n U1, U2 of b1, b2 which trivialize the bundle. Replacing U by ∩n=0g (U), we can

assume U1, U2 are g-stable and connected since B is irreducible. Now let V = U1 ∩U2

and take any closed point b0 ∈ V . By Lemma 5.4, we deduce that the action of gb1

is the same as the action of gb0 , which is the same as the action of gb2 . Hence we have

gFQ gFQ gFQ |E(Fp) | = |B(Fp) ||F (Fp) |

and we are done. 24 SAILUN ZHAN

Corollary 5.6. Let C be an integral projective curve over Fp with ﬁnite group G-actions. Then

[e(J nC)] = [e(L)][e(J nC˜)]

as virtual G-representations, i.e. in RQl (G), where L is a linear algebraic group and C˜ is the normalization of C.

Proof. Let f ∗ : J nC → J nC˜ be the pullback map. Since g is an automorphism ∗ ∗ on C and C˜, we have g∗f = f g∗. Now we use Corollary 5.5. Now to prove Theorem 1.9, we ﬁrst prove the following statement about e(J nC).

Lemma 5.7. Let C be an integral curve over Fp with nodal singularities and a G-action. Suppose p ∤ |G|. If C/˜ hgi is not a rational curve for every g ∈ G, then [e(J nC)]=0 as G-representations.

Proof. By Corollary 5.6, it suﬃces to prove [e(J nC˜)] = 0, which is equivalent n ˜ to Tr(g)|[e(JnC˜)] = 0 for any g ∈ G. But J C is an abelian variety, which means i n ˜ ∼ i 1 n ˜ ˜ ˜ H (J C, Ql) = ∧ H (J C, Ql). Since p ∤ |G|, C/ hgi is smooth. Then since C/ hgi 1 ˜ 1 n ˜ hgi is not rational, we have H (C/ hgi , Ql) =6 0. Hence H (J C, Ql) =6 0, which 1 n ˜ implies Tr(g)|[e(JnC˜)] = 0. This is because H (J C, Ql) = V0 ⊕ V1, where V0 is the non-empty eigenspace of g with eigenvalue 1, and V1 is its complement. We have i i 1 n ˜ i i i i i i (−1) ∧ H (J C, Ql)=( (−1) ∧ V0) ⊗ ( (−1) ∧ V1) and (−1) ∧ V0 has traceP 0. P P P Now with the help of Corollary 5.3, we have

n n ′ Proof of Theorem 1.9. Fix g ∈ G. Recall that we have J C = C′→C J C by Corollary 5.3. Depending on the action of g on the nodes of C, g∗ permutes` or acts n ′ on the strata J C . For any union of two or more strata permuted by g∗ cyclically, i i the trace of g on any Hc equals 0 since g acts by cyclically permuting the Hc of the components. For the stratum which is stable under g, the trace of g is also 0 by Lemma 5.7. Hence [e(J nC)] = 0 by Lemma 5.1.

Proof of Corollary 1.10. If C/G˜ is not a rational curve, then C/˜ hgi is not rational for any g ∈ G.

Remark 5.8. We are dealing with characteristic p in this section because we do not know whether Corollary 5.5 is true for singular cohomology in characteristic 0. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 25

Since B, E, and F are not assumed to be smooth projective, we cannot use the proper smooth base change theorem.

6. Rational curves on surfaces

Let S be a smooth projective K3 surface over C with a G-action, and let C be the tautological family of curves over an n-dimensional integral linear system |L| acted on by G. Then J nC is a smooth projective variety over |L| whose fiber over a point n t ∈ L is the compactified jacobian J Ct. Choose some good reduction over q such n that ‘everything’ (J C, S, G-action etc.) is defined over Fq, and we assume |L| is still

integral. Then |L| has a stratiﬁcation where each stratum B satisﬁes StabG(t)= H

for every t ∈ B and some subgroup H, and the ﬁbers Ct of the stratum have the ′ H ′ H same geometric genus. This is because for any subgroup H in G, |L| \ ∪H %H |L| is a locally closed subspace. The reason for the stratiﬁcation by the geometric genus is that the geometric genus gives a lower semicontinuous function in our case [Sh12].

Now notice that gB = B if g ∈ NG(H) and gB ∩ B = ∅ if g∈ / NG(H). Hence

we have a new stratification of |L| where each stratum ∪g∈GgB is G-invariant. Let π : J nC → |L| be the compactified jacobian of the family of curves. Then by −1 n considering π (∪g∈GgB) for all B, we obtain a stratification of J C, where each

stratum is G-stable and is a Zariski-locally trivial ﬁber bundle over ∪g∈GgB for the corresponding B. Then since

∗ −1 G ∗ −1 Hc (π (∪g∈GgB), Ql) = IndNG(H)Hc (π B, Ql), we know that [e(J nC)] equals

−1 G −1 G n [e(π (∪g∈GgB))] = IndNG(H)[e(π B)] = IndNG(H)[e(J C0)][e(B)]

X X X G by Lemma 5.1 and Corollary 5.5, where C0 is the ﬁber Ct for some t ∈ B, IndNG(H) n is the induced representation, and e(J C0) can be expressed using Corollary 5.3, 5.6 or Theorem 1.9.

Proof of Corollary 1.4. Recall that we have a birational map from J nC to S[n]

[Bea99], which maps a pair (Ct, L) to a unique eﬀective divisor D on Ct which is linearly equivalent to L, and can be viewed as a length d subscheme of S. Note that the G-actions commutes with this map in the sense of Theorem 1.3. Hence by Theorem 1.3 [e(J nC)] = [e(S[n])] as G-representations. Now by using Corollary 3.6 we are done. 26 SAILUN ZHAN

∞ [n] n Now we want to explicitly describe n=0[e(S )]t as a G-representation. For that purpose, we cite the following theoremsP ([Mu88, Proposition 1.2] and [AST11, Lemma 2.3]).

Theorem 6.1. [Mu88] Let g be a symplectic automorphism of a complex K3 surface S of order n < ∞. Then the number of ﬁxed points of g is equal to ǫ(n) = −1 1 24 n p|n 1+ p . Theorem Q 6.2.[AST11] Let g be a non-symplectic automorphism of a complex K3 surface S of prime order p. Then the Euler characteristic of Sg is 24−dp, where g rank T (g) 2 g ⊥ S denotes the ﬁxed locus of g, d = p−1 , and T (g) := (H (S, Z) ) . Proof of Theorem 1.6. By remark 3.7, we deduce that

∞ ∞ ∞ k mk n Tr(g )|[e(S¯)]t Tr(g)| ¯[n] t = exp . [e(S )] k n=0 m=1 ! X X Xk=1

Then using the Lefschetz ﬁxed point formula and Theorem 6.1, we get the equality we want.

When G is a cyclic group of order N, we have N ≤ 8 by [H16, Corollary 15.1.8]. 1/24 ∞ n Recall the deﬁnition of the Dedekind eta function η(t) = t n=1(1 − t ), where 2πiz t = e . Fix a generator g of G. Q If N is a prime number p, we notice that ord(gk) = 1 if p|k, and ord(gk) = p otherwise. Hence

∞ ∞ ∞ mk ∞ ∞ mpk n ǫ(p)t (24 − ǫ(p))t Tr(g)| ¯[n] t = exp exp [e(S )] k pk n=0 m=1 k=1 ! m=1 k=1 ! X X X X Xǫ(p)−24 ∞ −ǫ(p) ∞ p = (1 − tm) (1 − tmp) m=1 ! m=1 ! Y Y 24 ∞ − p+1 = (1 − tm)(1 − tmp) m=1 ! Y 24 p 24 = t/η(t) p+1 η(t ) p+1 . COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 27

If N = 4, we have

4, if k ≡ 1, 3 (mod 4) k Tr(g )| ¯ = 8, if k ≡ 2 (mod 4) [e(S)]   24, if k ≡ 0 (mod 4).  Hence 

∞ ∞ mk ∞ mk ∞ mk n 4t 8t 24t Tr(g)| ¯[n] t = exp + + [e(S )] k k k n=0 m=1 m=1 m=1 ! X X kX≡1,3 X Xk≡2 X Xk≡0 ∞ = (1 − tm)−4(1 − t2m)−2(1 − t4m)−4 m=1 Y = t/η(t)4η(t2)2η(t4)4.

If N = 6, we have

2, if k ≡ 1, 5 (mod 6) 6, if k ≡ 2, 4 (mod 6) k  Tr(g )|[e(S¯)] =  8, if k ≡ 3 (mod 6)  24, if k ≡ 0 (mod 6).    Hence 

∞ ∞ mk ∞ mk ∞ mk ∞ mk n 2t 6t 8t 24t Tr(g)| ¯[n] t = exp + + + [e(S )] k k k k n=0 m=1 m=1 m=1 m=1 ! X X kX≡1,5 X kX≡2,4 X Xk≡3 X Xk≡0 ∞ = (1 − tm)−2(1 − t2m)−2(1 − t3m)−2(1 − t6m)−2 m=1 Y = t/η(t)2η(t2)2η(t3)2η(t6)2.

If N = 8, we have

2, if k ≡ 1, 3, 5, 7 (mod 8) 4, if k ≡ 2, 6 (mod 8) k  Tr(g )|[e(S¯)] =  8, if k ≡ 4 (mod 8)  24, if k ≡ 0 (mod 8).     28 SAILUN ZHAN

Hence

∞ ∞ mk ∞ mk ∞ mk ∞ mk n 2t 4t 8t 24t Tr(g)| ¯[n] t = exp + + + [e(S )] k k k k n=0 m=1 m=1 m=1 m=1 ! X X k≡X1,3,5,7 X kX≡2,6 X Xk≡4 X Xk≡0 ∞ = (1 − tm)−2(1 − t2m)−1(1 − t4m)−1(1 − t8m)−2 m=1 Y = t/η(t)2η(t2)η(t4)η(t8)2.

Proof of Theorem 1.8. By remark 3.7, we have

∞ ∞ ∞ k mk n Tr(g )|[e(S¯)]t Tr(g)| ¯[n] t = exp . [e(S )] k n=0 m=1 ! X X Xk=1 By the topological Lefschetz formula, we have

∞ ∞ ∞ mk ∞ ∞ mpk n (24 − dp)t (dp)t Tr(g)| ¯[n] t = exp exp [e(S )] k pk n=0 m=1 ! m=1 ! X X Xk=1 X Xk=1 ∞ dp−24 ∞ −d = (1 − tm) (1 − tmp) . m=1 ! m=1 ! Y Y

Example 1 (Z/2Z). Here we look at an explicit K3 surface with a symplectic Z/2Z-action. Consider the elliptic K3 surface S deﬁned by the Weierstrass equation

2 3 4 2 12 10 8 6 4 2 y = x +(t + a1t + a2)x +(t + b1t + b2t + b3t + b4t + b5t + b6)

2 6 where (a1, a2) ∈ C , (b1, ..., b6) ∈ C are generic. The ﬁbration has 24 nodal ﬁbers

(Kodaira type I1) over the zeros of its discriminant polynomial and those zeros do not contain 0 and ∞. The automorphsim of order 2

σ(x, y, t)=(x, −y, −t)

acts non-trivially on the basis of the ﬁbration and preserves the smooth elliptic curves over t = 0 and t = ∞. Now denote one of the ﬁbers by L, then |L| is a COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 29

hσi-invariant integral linear system and all of the singular curves in |L| are nodal rational curves. We want to understand the σ-orbits in |L|.

Since we know explicitly the action of σ, by calculation we know that there are 4 σ-fixed points on the fiber over t =0 and 4 σ-fixed points on the fiber over t = ∞. So σ has 8 isolated fixed points, hence it is a symplectic involution.

Now we know from Theorem 1.6 that

∞ ∞ −8 ∞ −8 n k 2k Tr(σ)|[e(S¯[n])]t = (1 − t ) (1 − t ) . n=0 ! ! X Yk=1 kY=1

This implies Tr(σ)|[e(S)] = 8 by looking at the coeﬃcient of t. Hence we have

Tr(σ)|[e(JC)] = 8 by Corollary 1.4. But in this case, there are only two strata contributing to [e(JC)] as hσi-representations (see Remark 6.3). One consists of nodal rational curves which are not σ-stable. The other consists of elliptic curves whose quotient by σ is rational. The ﬁrst stratum contributes n1[Z/2Z] if there are n1 such σ-orbits in |L|, where [Z/2Z] is the regular representation. The second stratum contributes n2(2V1 − 2V−1) if there are n2 such elliptic curves in |L|, where Vs is the 1-dim representation on which σ has eigenvalue s. Hence

[e(JC)] = n1[Z/2Z]+ n2(2V1 − 2V−1)

But since we already know the representation [e(JC)], by calculation we have n1 = 12 and n2 = 2. On the other hand, this coincides with the geometric picture. From the deﬁnition of σ we observe that there are indeed 12 σ-orbits of nodal rational curves. Denote

by C0,C∞ the ﬁbers over t = 0, ∞. Since σ preserve C0 and there are 4 σ-ﬁxed

points, we deduce that the degree 2 morphism C0 → C0/ hσi has 4 ramiﬁcation

points. Hence by Riemann-Hurwitz formula C0/ hσi is smooth rational. By the

same argument, C∞/ hσi is also smooth rational. This is what we have expected since there should be two such curves from the calculation of the representations.

Example 2 (Z/3Z). Here we look at an explicit K3 surface with a non-symplectic Z/3Z-action [AST11, Remark 4.2]. Consider the elliptic K3 surface S deﬁned by the Weierstrass equation

2 3 6 3 12 9 6 3 y = x +(t + a1t + a2)x +(t + b1t + b2t + b3t + b4) 30 SAILUN ZHAN

2 4 where (a1, a2) ∈ C , (b1, ..., b4) ∈ C are generic. The ﬁbration has 24 nodal ﬁbers

(Kodaira type I1) over the zeros of its discriminant polynomial and those zeros do not contain 0 and ∞. The automorphsim of order 3

σ(x, y, t)=(x,y,ζ3t) acts non-trivially on the basis of the ﬁbration and preserves the smooth elliptic curves over t = 0 and t = ∞. Now denote one of the ﬁbers by L, then |L| is a hσi-invariant integral linear system and all of the singular curves in |L| are nodal rational curves. We want to understand the σ-orbits in |L|.

First we observe that σ ﬁxs the ﬁber over t = 0. Hence hσi acts non-symplectically on S. We know from [AST11, Theorem 4.1] that rank T (σ) = 14. Then using Theorem 1.8 we have

∞ ∞ −3 ∞ −7 n k 3k Tr(σ)|[e(S¯[n])]t = (1 − t ) (1 − t ) . n=0 ! ! X Yk=1 kY=1 This implies Tr(σ)|[e(S)] = 3 by looking at the coeﬃcient of t. Hence we have

Tr(σ)|[e(JC)] = 3 by Corollary 1.4. But in this case, there are only two strata contributing to [e(JC)] as hσi-representations (see Remark 6.3). One consists of nodal rational curves which are not σ-stable. The other consists of elliptic curves whose quotient by σ is rational. In particular, the ﬁber over t = 0 will not contribute to [e(JC)]. The ﬁrst stratum contributes n1[Z/3Z] if there are n1 such σ-orbits in |L|, where [Z/3Z] is the regular representation. The second stratum contributes

−1 n2(2V1 − Vζ3 − V ) if there are n2 such elliptic curves in |L| , where Vs is the 1-dim ζ3 representation on which σ has eigenvalue s. Hence

−1 [e(JC)] = n1[ /3 ]+ n2(2V1 − Vζ3 − V ) Z Z ζ3

But since we already know the representation [e(JC)], by calculation we have n1 =8 and n2 = 1. On the other hand, this coincides with the geometric picture. From the deﬁnition of σ we observe that there are indeed 8 σ-orbits of nodal rational curves. Since the action of σ is explicit, by calculation we know that there are 3 σ-ﬁxed points on the

fiber over t = ∞. Denote by C∞ the fiber over t = ∞. Then this implies that the degree 3 morphism C∞ → C∞/ hσi has 3 ramification points. Hence by Riemann-

Hurwitz formula C∞/ hσi is smooth rational, which is what we have expected. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 31

Remark 6.3 (Z/2Z in general). Let us consider a complex K3 surface S with a Z/2Z-action (i.e. an involution σ). Take a σ-invariant integral linear system L of dimension d. Assume all the rational curves in L have nodal singularities.

For the stratum of L which consists of curves that are not σ-stable and of geometric genus > 0, denote by M the corresponding stratum of J dC. Then d [e(M)] = e(M/ hσi)[Z/2Z] = e(J C0)e(B/ hσi)[Z/2Z], where [Z/2Z] is the regu- d ∞ n n d lar representation. But e(J C0) = n=0(−1) dimH (J C0, Ql) = 0 by Theorem 1.9. Hence [e(M)] = 0 as representation.P For the stratum of L which consists of curves that are not σ-stable and are nodal rational curves, we have [e(M)] = n0[Z/2Z] where n0 is the number of σ-orbits d of nodal rational curves, since e(J C0) = 1 by Corollary 5.3, Corollary 5.6 and e(Gm) = 0. For those curves that are σ-stable, we first notice that σ fixes some curve only if σ acts non-symplectically on S, and in that case the fixed curves are always smooth

([AST11, §2]). Let C0 be a smooth curve of genus d ≥ 1 fixed by σ. If d = 1, then the fixed locus of σ consists of two disjoint elliptic curves and they are linearly equivalent. If d> 1, then it is the only curve of genus d fixed by σ ([AST11, Theorem 3.1]). In either case, the stratum consisting of σ-fixed curves contributes 0 to the representation since the Euler characteristic of an abelian variety is 0.

For the stratum of L which consists of curves that are σ-stable, if the normalization of the curves quotient by hσi is not rational, then by Theorem 1.9, we have [e(M)] = 0.

Hence there are only two strata contributing to [e(J dC)]. One consists of σ-orbits of nodal rational curves. The other consists of the curves whose normalization quotient by hσi are smooth rational.

In particular, for the stratum which consists of σ-stable smooth curves whose quotient by hσi are rational, by a similar argument as in the proof of Lemma 5.7, d 2d−1 2d−1 we have [e(J (C0))] = 2 V1 − 2 V−1, where Vs is the 1-dim representation on which σ has eigenvalue s.

Now let us give an example of a 2-dim Z/2Z-invariant linear system. This example is suggested by Jim Bryan. 32 SAILUN ZHAN

Example 3 (2-dim Z/2Z). Let S be a K3 surface given by the double cover of P2 branched over a smooth sextic curve C in P2. Let τ be the involution on P2 sending (x : y : z) to (−x : y : z). Denote the covering involution by i : S → S. Then if we suppose C is τ-invariant, the ‘composition’ of τ and i will give a symplectic involution σ on S ([GS07, Section 3.2]).

The ﬁxed locus of τ on P2 consists of a point x0 = (1 : 0 : 0) and a line l0 = {(x :

y : z)|x = 0}. Denote the six intersection points of l0 and C by x3, x4, ..., x8. Let −1 π : S → P2 be the double cover map. Denote the two points in π (x0) by x1, x2.

Then the ﬁxed locus of σ is the eight points x1, x2, ..., x8. Notice that σ commutes

with i and the induced action of σ on P2 is just τ.

∗ Now let L = π OP2 (1). Then the linear system |L| consists of the curves which are the preimages of the lines in P2 under π. For a generic choice of C, |L| is a σ-invariant integral linear system. A generic line will intersects C in six points, and its preimage is a smooth genus 2 curve. Some lines will intersect C in a tangent point and 4 other distinct points, and their preimages are curves with one node. The other lines are the 324 bitangents of C, which can be seen from the Pl¨ucker 2 ∞ n −24 formula or the coeﬃcient of t in n=1(1 − t ) . Let C → |L| be the tautologicalQ family of curves over |L|. Now we know from Theorem 1.6 that

∞ ∞ −8 ∞ −8 n k 2k Tr(σ)|[e(S[n])]t = (1 − t ) (1 − t ) . n=0 ! ! X Yk=1 kY=1

2 This implies Tr(σ)|[e(S[2])] = 52 by looking at the coeﬃcient of t . Hence we have

Tr(σ)|[e(JC)] = 52 by Corollary 1.4. Since we know Tr(1)|[e(S[2])] = 324, we have

[e(JC)] = 188V1 + 136V−1 where V1 is the 1-dimensional trivial representation and V−1 is the 1-dimensional representation on which σ has eigenvalue −1.

To give a geometric interpretation of this representation, we ﬁrst concentrate on the σ-invariant curves in |L| since non-invariant curves with zero or one node will contribute nothing to the representation, and non-invariant nodal rational curves will contribute some multiple of the regular representation. The σ-invariant curves in |L| are given by the preimages of the τ-invariant lines, which consists of the line COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 33

l0 = {(x : y : z)|x = 0} and all the lines passing through x0 = (1 : 0 : 0), i.e., {(x : y : z)|by + cz =0}, (b : c) ∈ P1.

The preimage of l0 is a smooth genus 2 curve, and it has 6 ramiﬁcation points

x3, x4, ..., x8 under σ. Hence its preimage quotient by σ is a smooth rational curve,

and it will contribute 8V1 − 8V−1 to the representation by a similar argument as in the proof of Lemma 5.7.

The preimages of {(x : y : z)|by + cz = 0} are more complicated. A generic line will intersect C in six points, and its preimage is a smooth curve of genus 2. It has

2 ramiﬁcation points x1, x2 under σ. Hence its preimage quotient by σ is an elliptic curve, and it will contribute nothing to the representation.

Now let us consider those tangent lines. We ﬁrst observe that if a line passes through one of the six points x3, x4, ..., x8, then by looking at the σ-action on the preimage of the line, this point must be a tangent point. So its preimage is a curve 1 2 with one node, and the normalization of it has 4 ramiﬁcation points x1, x2, xi , xi 1 2 under σ, where xi , xi are the two points on the normalization over the point xi if our line passes through xi, i ∈{3, 4, ..., 8}. Hence the normalization of its preimage quotient by σ is a rational curve, and we denote its contribution to the representation by [e(JC1)]. Now if there is a tangent point y of the tangent line which is not one of the six

points x3, x4, ..., x8, then since both of the line and the curve C are τ-invariant, this line must have another tangent point τ(y). Hence this line must be a bitangent. In order to calculate the number of the τ-invariant bitangents, we notice that the degree of the dual curve of C is 30. We already have 6 tangent lines with one tangent point, and all the other τ-invariant tangent lines are bitangents. Hence there should be 12 τ-invariant bitangents. The preimage of the τ-invariant bitangent is a rational curve with two nodes, and τ permutes the nodes. The normalization of the preimage has 2

ramiﬁcation points x1, x2 under σ. Hence the normalization of its preimage quotient by σ is a rational curve, and we denote its contribution to the representation by

[e(JC2)]. Finally, non-invariant curves with two nodes will also contribute to the representation. We know there are 324 curves with two nodes, and 12 of them are σ-invariant by the discussion above. So there are 312 non-invariant nodal rational curves, which will contribute 156[Z/2Z] = 156V1 + 156V−1 to the representation. 34 SAILUN ZHAN

Combining all of the above, we have

188V1 + 136V−1 = [e(JC)]

=8V1 − 8V−1 + 6[e(JC1)] + 12[e(JC2)] + 156V1 + 156V−1. Hence we only need to check

[e(JC1)] + 2[e(JC2)]=4V1 − 2V−1 and this is true by the following two lemmas.

Lemma 6.4. Let C1 be an integral curve of arithmetic genus 2 with one node over Fp. If there is a involution σ acting on it, and the action σ on its normalization

C˜1 has 4 ﬁxed points, two of which are the points over the node, then we have

[e(JC1)]=2V1 − 2V−1 as Z/2Z-representations.

Lemma 6.5. Let C2 be an integral curve of arithmetic genus 2 with two nodes

over Fp. If there is a involution σ acting on it, and σ permutes the nodes, then we have

[e(JC2)] = V1 as Z/2Z-representations.

Proof of Lemma 6.4. Since C1 is an integral curve of arithmetic genus 2 with one node, its normalization π : C˜1 → C1 is an elliptic curve, and we denote it by E. By Corollary 5.3 and Corollary 5.6, we have

[e(JC1)] = [e(JC1)] + [e(JE)] = [e(Gm)][e(JE)] + [e(JE)].

Now we notice that σ has 4 ﬁxed points on E, so E → E/ hσi realizes E as a double cover of P1. Hence

0 1 2 [e(JE)] = [H (E, Ql)] − [H (E, Ql)] + [H (E, Ql)]=2V1 − 2V−1.

On the other hand, since σ ﬁxes the points over the node, we deduce from the short exact sequence ∗ ∗ 0 → OC1 → π∗OE → δ → 0 0 that σ acts trivially on the skyscraper sheaf δ. Hence σ acts trivially on H (C1, δ)=

Gm. So [e(Gm)] = e(Gm)V1 = 0 since the topological Euler characteristic of Gm is 0. COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 35

Combining the above discussion, we have

[e(JC1)]=2V1 − 2V−1.

Proof of Lemma 6.5. Since C2 is an integral curve of arithmetic genus 2 with

two nodes, its normalization π : C˜2 → C2 is a rational curve. It also has two partial ′ ′′ normalizations by resolving one of the nodes π1 : C2 → C2 and π2 : C2 → C2. By Corollary 5.3 and Corollary 5.6, we have

′ ′′ 1 [e(JC2)] = [e(JC2)]+[e(JC2)]+[e(JC2 )]+[e(JP )] = [e(Gm×Gm)]+e(Gm)[Z/2Z]+V1 since σ permutes two nodes.

2 On the other hand, Gm is an aﬃne curve. So dim Hc (Gm, Ql) = 1 and dim 0 Hc (Gm, Ql) = 0. Since the topological Euler characteristic of Gm is 0, we also have 1 dim Hc (Gm, Ql) = 1. Notice that σ permutes two Gm’s, and hence by the K¨unneth formula we have 1 1 1 2 [e(Gm × Gm)] = [Hc (Gm, Ql) ⊗ Hc (Gm, Ql)] − [Hc (Gm, Ql) ⊗ Hc (Gm, Ql)] 2 1 2 2 − [Hc (Gm, Ql) ⊗ Hc (Gm, Ql)] + [Hc (Gm, Ql) ⊗ Hc (Gm, Ql)]

= V−1 − [Z/2Z]+ V1 =0.

Combining the above discussion, we have

[e(JC2)] = V1.

Finally, let us give some discussions when G equals a certain ﬁnite simple group.

Example 4 (PSL(2,7)). Let S be a complex K3 surface acting faithfully by

G = P SL2(F7). Such a K3 surface exists. For example, P SL(2, 7) acts faithfully and symplectically on the surface X3Y + Y 3Z + Z3X + T 4 = 0 in P3 by means of a linear action on P3 [Mu88]. We know from Theorem 1.9 that a G-stable curve C with nodal singularities in an integral linear system does contribute to the representation [e(J dC)] only if there exists some g ∈ G such that C/˜ hgi = P1. It turns out that if this happens, then C˜ must be the Klein quartic, which is the Hurwitz surface of the lowest possible genus. Notice that G acts on C faithfully since any non-trivial element of G acts symplectically on S and cannot ﬁx curves. 36 SAILUN ZHAN

Proposition 6.6. Let C be a smooth projective curve over C with a faithful G = P SL(2, 7)-action. If there exists g ∈ P SL(2, 7) such that C/ hgi = P1, then the genus of C is 3 and g has order 7. In particular, the automorphism group of C reaches its Hurwitz bound, and hence C is the Klein quartic.

Proof. The idea is to use the equivariant Riemann-Hurwitz formula [Se79, Chap- ter VI §4] for π : C → C/G = P1. We have

1 [e(C)] = e(P )Ih1i − (Ih1i − Ihgi) 1 pX∈P as G-representations, where hgi is the stablizer of some point over p, Ihgi denotes

G ½ the induced representation Indhgi ½, and is the 1-dim trivial representation. Notice

that Ihgi is independent of the point we choose over p.

Since G acts trivially on P1, we have

1 H (C, C)= (Ih1i − Ihgi) − 2Ih1i +2½ 1 pX∈P and what we are going to do is to compare the representations on both sides. For this purpose, we need the character table of P SL(2, 7).

1A1 2A21 3A56 4A42 7A24 7B24

χ1 111 1 11 (3) (2) 2 4 3 5 6 χ2 1 1(−1) 1ωω¯ 1i(-i) ζ7ζ7 ζ7 ζ7 ζ7 ζ7 (3) (2) 3 5 6 2 4 χ3 1 1(−1) 1ωω¯ 1i(-i) ζ7 ζ7 ζ7 ζ7ζ7 ζ7 (6) (4) (2) (2) (2) (2) (2) (2) 6 6 χ4 1 1 (−1) 1 ω ω¯ 1 (−1) i(−i) ζ7...ζ7 ζ7...ζ7 (7) (3) (4) (3) (2) (2) (2) (2) (2) 6 6 χ5 1 1 (−1) 1 ω ω¯ 1(−1) i (−i) 1ζ7...ζ7 1ζ7...ζ7 (8) (4) (4) (2) (3) (3) (2) (2) (2) (2) (2) 6 (2) 6 χ6 1 1 (−1) 1 ω ω¯ 1 (−1) i (−i) 1 ζ7...ζ7 1 ζ7...ζ7

This is a reﬁned character table which can be deduced from the usual character table. Each entry denotes the eigenvalues of the element in given conjuacy classes acting on given irreducible representations, nAm denotes the conjugacy class of size m in which each element has order n, a(i)b(j) denotes eigenvalue a with multiplicity 6 2 3 4 5 6 i and eigenvalue b with multiplicity j, and ζ7...ζ7 means ζ7ζ7 ζ7 ζ7 ζ7 ζ7 .

1 −1 For induced representations, we have Ihgi(x)= |hgi| h∈G χ(hxh ), where χ(x)= 1 if x ∈hgi, and χ(x) = 0 otherwise. Hence using ourP character table and calculating COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 37

by Schur orthogonality relations, we have

I2 = χ1 + χ2 + χ3 +4χ4 +3χ5 +4χ6

I3 = χ1 + χ2 + χ3 +2χ4 +3χ5 +2χ6

I4 = χ1 + χ2 + χ3 +2χ4 + χ5 +2χ6

I7 = χ1 + χ5 +2χ6

where In denotes Ihgi for the element g of order n.

Now since there exists g ∈ G such that C/ hgi = P1, we have H1(C, C)g = 1 H (C/ hgi , C) = 0. But for χ1, χ5 and χ6, whatever g is, there are always non- 1 trivial g-fixed vectors. This implies that H (C, C) does not contain χ1, χ5 and χ6 at all. Hence the coefficients of χ1, χ5 and χ6 in p∈P1 (Ih1i − Ihgi) − 2Ih1i +2½ must 1 be 0. This gives us only two possibilities: H (PC, C) = I1 − I2 − I3 − I4 +2½ or 1 H (C, C) = I1 − I2 − I3 − I7 +2½. If we look at the dimensions of the right hand sides, the first one gives dimension -12 and the second gives dimension 6. Hence the 1 only possibility is H (C, C)= I1 − I2 − I3 − I7 +2½ = χ2 + χ3, which shows that the 1 1 genus of C is 2 dimH (C, C) = 3. We also deduce from this argument that g must has order 7 since the element of order not equal to 7 does have fixed vectors in χ2 and χ3. Following this observation, we do the calculations for some other groups in Mukai’s list [Mu88].

Example 5 (A6). G = A6 acts faithfully and symplectically on the K3 surface 6 6 2 6 3 5 5 1 Xi = 1 Xi = 1 Xi = 0 in P via permutation action of coordinates on P . PThen by TheoremP 1.9,P a G-stable curve C with nodal singularities in an integral linear system will not contribute to the representation [e(J dC)].

Proposition 6.7. Let C be a smooth projective curve over C with a faithful 1 G = A6-action. Then for any g ∈ A6, we have C/ hgi= 6 P .

Proof. We have the following character table for A6. 38 SAILUN ZHAN

1A1 2A45 3A40 3B40

χ1 11 1 1 (5) (3) (2) (3) (2) (2) χ2 1 1 (−1) 1 ωω¯ 1ω ω¯ (5) (3) (2) (2) (2) (3) χ3 1 1 (−1) 1ω ω¯ 1 ωω¯ (8) (4) (4) (2) (3) (3) (2) (3) (3) χ4 1 1 (−1) 1 ω ω¯ 1 ω ω¯ (8) (4) (4) (2) (3) (3) (2) (3) (3) χ5 1 1 (−1) 1 ω ω¯ 1 ω ω¯ (9) (5) (4) (3) (3) (3) (3) (3) (3) χ6 1 1 (−1) 1 ω ω¯ 1 ω ω¯ (10) (4) (6) (4) (3) (3) (4) (3) (3) χ7 1 1 (−1) 1 ω ω¯ 1 ω ω¯

4A90 5A72 5B72

χ1 1 1 1 (2) 2 3 4 2 3 4 χ2 1(−1) i(−i) 1ζ5ζ5 ζ5 ζ5 1ζ5ζ5 ζ5 ζ5 (2) 2 3 4 2 3 4 χ3 1(−1) i(−i) 1ζ5ζ5 ζ5 ζ5 1ζ5ζ5 ζ5 ζ5 (2) (2) (2) (2) (2) 2(2) 3(2) 4 (2) (2) 2 3 4(2) χ4 1 (−1) i (−i) 1 ζ5ζ5 ζ5 ζ5 1 ζ5 ζ5 ζ5 ζ5 (2) (2) (2) (2) (2) (2) 2 3 4(2) (2) 2(2) 3(2) 4 χ5 1 (−1) i (−i) 1 ζ5 ζ5 ζ5 ζ5 1 ζ5ζ5 ζ5 ζ5 (3) (2) (2) (2) (2) 2(2) 3(2) 4(2) (2) 2(2) 3(2) 4(2) χ6 1 (−1) i (−i) 1ζ5 ζ5 ζ5 ζ5 1ζ5 ζ5 ζ5 ζ5 (2) (2) (3) (3) (2) (2) 2(2) 3(2) 4(2) (2) (2) 2(2) 3(2) 4(2) χ7 1 (−1) i (−i) 1 ζ5 ζ5 ζ5 ζ5 1 ζ5 ζ5 ζ5 ζ5

For induced representations, we have

I2 = χ1 +3χ2 +3χ3 +4χ4 +4χ5 +5χ6 +4χ7

I3A = χ1 +3χ2 + χ3 +2χ4 +2χ5 +3χ6 +4χ7

I3B = χ1 + χ2 +3χ3 +2χ4 +2χ5 +3χ6 +4χ7

I4 = χ1 + χ2 + χ3 +2χ4 +2χ5 +3χ6 +2χ7

I5 = χ1 + χ2 + χ3 +2χ4 +2χ5 + χ6 +2χ7

Now suppose there exists g ∈ G such that C/ hgi = P1. Then we have H1(C, C)g = H1(C/ hgi , C) = 0. But for all the irreducible representations of G, whatever g is, there are always non-trivial g-ﬁxed vectors. This implies that H1(C, C) = 0. Hence

p∈P1 (Ih1i −Ihgi)−2Ih1i +2½ = 0. But no combination will give this equality. Hence Psuch g does not exist.

Example 6 (A5). G = A5 acts faithfully and symplectically on the K3 surface 5 6 2 5 3 5 1 Xi = 1 Xi = 1 Xi = 0 in P via permutation action of the ﬁrst 5 coordinates 5 Pon P . ThenP by theoremP 1.9, a G-stable curve C with nodal singularities in an COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 39

integral linear system can contribute to the representation [e(J dC)] only if C˜ is rational.

Proposition 6.8. Let C be a smooth projective curve over C with a faithful 1 G = A5-action. If there exists g ∈ A5 such that C/ hgi = P , then C must be a smooth rational curve.

Proof. We have the following character table for A5.

1A1 2B15 3A20 5A12 5B12

χ1 11 1 11 (3) (2) 4 2 3 χ2 1 1(−1) 1ωω¯ 1ζ5ζ5 1ζ5 ζ5 (3) (2) 2 3 4 χ3 1 1(−1) 1ωω¯ 1ζ5 ζ5 1ζ5ζ5 (4) (2) (2) (2) 4 4 χ4 1 1 (−1) 1 ωω¯ ζ5...ζ5 ζ5...ζ5 (5) (3) (2) (2) (2) 4 4 χ5 1 1 (−1) 1ω ω¯ 1ζ5...ζ5 1ζ5...ζ5 For induced representations, we have

I2 = χ1 + χ2 + χ3 +2χ4 +3χ5

I3 = χ1 + χ2 + χ3 +2χ4 + χ5

I5 = χ1 + χ2 + χ3 + χ5

Now suppose there exists g ∈ G such that C/ hgi = P1. Then we have H1(C, C)g = 1 H (C/ hgi , C)=0. But for χ1, χ2, χ3 and χ5, whatever g is, there are always non- 1 trivial g-ﬁxed vectors. This implies that H (C, C) does not contain χ1, χ2, χ3 and

χ5 at all. Hence the coeﬃcients of χ1, χ2, χ3 and χ5 in p∈P1 (Ih1i − Ihgi) − 2Ih1i +2½ 1 must be 0. This gives us only one possibility: H (C, C)=P I1 − I2 − I3 − I5 +2½ = 0, which implies C has genus 0.

Example 7 (S5). G = S5 acts faithfully and symplectically on the K3 surface 5 6 2 5 3 5 1 Xi = 1 Xi = 1 Xi = 0 in P via permutation action of the ﬁrst 5 coordinates 5 Pon P . ThenP by theoremP 1.9, a G-stable curve C with nodal singularities in an integral linear system can contribute to the representation [e(J dC)] only if C˜ has genus 4. In this case C˜ has the largest possible automorphism group for a genus 4 curve and C˜ is Bring’s curve.

Proposition 6.9. Let C be a smooth projective curve over C with a faithful 1 G = S5-action. If there exists g ∈ S5 such that C/ hgi = P , then C has genus 4 and g has order 5. In particular, C is Bring’s curve. 40 SAILUN ZHAN

Proof. We have the following character table for S5.

1A1 2A10 2A15 3A20

χ1 11 1 1

χ2 1 -1 1 1 (4) (3) (2) (2) (2) χ3 1 1 (−1) 1 (−1) 1 ωω¯ (4) (3) (2) (2) (2) χ4 1 1(−1) 1 (−1) 1 ωω¯ (5) (3) (2) (3) (2) (2) (2) χ5 1 1 (−1) 1 (−1) 1ω ω¯ (5) (2) (3) (3) (2) (2) (2) χ6 1 1 (−1) 1 (−1) 1ω ω¯ (6) (3) (3) (2) (4) (2) (2) (2) χ7 1 1 (−1) 1 (−1) 1 ω ω¯

4A30 5A24 6A20

χ1 1 1 1

χ2 -1 1 -1 4 χ3 1(−1)i(−i) ζ5...ζ5 1(−1)ωω¯ 4 χ4 1(−1)i(−i) ζ5...ζ5 1(−1)(−ω)(−ω¯) (2) 4 χ5 1(−1) i(−i) 1ζ5...ζ5 1ωω¯(−ω)(−ω¯) (2) 4 χ6 1 (−1)i(−i) 1ζ5...ζ5 −1ωω¯(−ω)(−ω¯) (2) (2) (2) 4 χ7 1(−1)i (−i) 1 ζ5...ζ5 1(−1)ωω¯(−ω)(−ω¯) For induced representations, we have

I2A = χ1 +3χ3 + χ4 +3χ5 +2χ6 +3χ7

I2B = χ1 + χ2 +2χ3 +2χ4 +3χ5 +3χ6 +2χ7

I3 = χ1 + χ2 +2χ3 +2χ4 + χ5 + χ6 +2χ7

I4 = χ1 + χ3 + χ4 + χ5 +2χ6 + χ7

I5 = χ1 + χ2 + χ5 + χ6 +2χ7

I6 = χ1 + χ3 + χ4 + χ5 + χ7

Now since there exists g ∈ G such that C/ hgi = P1, we have H1(C, C)g = 1 H (C/ hgi , C) = 0. But for χ1, χ5 and χ7, whatever g is, there are always non- 1 trivial g-fixed vectors. This implies that H (C, C) does not contain χ1, χ5 and χ7 at all. Hence the coefficients of χ1, χ5 and χ7 in p∈P1 (Ih1i − Ihgi) − 2Ih1i +2½ must 1 be 0. This gives us only two possibilities: H (C,PC)= I1 − I2 − I4 − I5 +2½ =2χ4 1 or H (C, C) = I1 − I2 − I5 − I6 +2½ = 2χ4 +2χ6. For the second case, we notice that whatever conjugacy class g belongs to, there always exists g-fixed vectors in COUNTING RATIONAL CURVES ON K3 SURFACES WITH FINITE GROUP ACTIONS 41

1 2χ4 +2χ6. Hence H (C, C) = I1 − I2 − I4 − I5 +2½ = 2χ4. It follows that C has genus 4 and g has order 5.

References

[A04] Alexeev, V.: Compactified Jacobians and Torelli map. Publ. RIMS, Kyoto Univ. 40 (2004), 1241–1265. [AK76] Altman, A.; Kleiman, S.: Compactifying the Jacobian, Bull. Amer. Math. Soc. 82(6) (1976), 947–949. [AST11] Artebani, M.; Sarti, A.; Taki, S.: K3 surfaces with non-symplectic automorphisms of prime order. Math. Z. 268 (2011), 507–533. [Ba99] Batyrev, V.V.: Birational Calabi-Yau n-folds have equal Betti numbers. New Trends in Algebraic Geometry, London Mathematical Society Lecture Note Series, Cambridge Uni- versity Press, 1999, 1–12. [Bea99] Beauville, A.: Counting rational curves on K3 surfaces. Duke Math. J. 97 (1999), 99–108. [BG19] Bryan, J.; Gyenge, A.:´ G-fixed Hilbert schemes on K3 surfaces, modular forms, and eta products. arXiv:1907.01535. [BLR90] Bosch, Siegfried; Lütkebohmert, Werner; Raynaud, Michel: Néron models, Ergebnisse der Mathematik und ihrer Grenzgebiete (3)[Results in Mathematics and Related Areas(3)], vol. 21, pp. x+325, Springer-Verlag, Berlin, 1990. [BO18] Bryan, J.; Oberdieck, G.: CHL Calabi-Yau threefolds: Curve counting, Mathieu moon- shine and Siegel modular forms. arXiv:1811.06102. [Ca85] Carter, R.W.: Finite Groups of Lie Type: Conjugacy Classes and Complex Characters, Wiley, 1985. [CS17] Cohen, H.; Strömberg, F: Modular Forms: A Classical Approach, Graduate Studies in Math. 179, Amer. Math. Soc., 2017. [De72] Deligne, P.: La conjecture de Weil pour les surfaces K3. Invent. math. 15 (1972), 206–226. [DL76] Deligne, P.; Lusztig, G.: Representations of reductive groups over finite fields. Ann. of Math. 103 (1976), 103–161. [ES87] Ellingsrud, G.; Strømme, S.A.: On the homology of the Hilbert scheme of points in the plane. Invent. math. 87 (1987), 343–352. [EGK00] Esteves, E.; Gagné, M.; Kleiman, S.: Abel maps and presentation schemes. Comm. in Algebra 28(12) (2000), 5961–5992. [FGvS] Fantechi, B.; Göttsche, L.; van Straten, D.: Euler number of the compactified Jacobian and multiplicity of rational curves. J. Alg. Geom. 8 (1999), 115–133. [Gö90] Göttsche, L.: The Betti numbers of the Hilbert scheme of points on a smooth projective surface. Math. Ann. 286 (1990), 193–207. [Gö94] Göttsche, L.: Hilbert Schemes of Zero-Dimensional Subschemes of Smooth Varieties, LNM 1572, Springer-Verlag Berlin Heidelberg, 1994. 42 SAILUN ZHAN

[GS93] G¨ottsche, L.; Soergel, W.: Perverse sheaves and the cohomology of Hilbert schemes of smooth algebraic surfaces. Math. Ann. 296 (1993), 235–245. [GS07] van Geemen, B.; Sarti, A.: Nikulin involutions on K3 surfaces. Math. Z. 255 (2007), 731–753. [H16] Huybrechts, D.: Lectures on K3 Surfaces, Cambridge studies in advanced mathematics 158, Cambridge University Press, 2016. [I77] Iarrobino, A.A.: Punctual Hilbert Schemes. Mem. Amer. Math. Soc. 10 , no. 188, viii+112 pp, 1977. [M62] Macdonald, I.G.: The Poincar´epolynomial of a symmetric product. Proc. Cambridge Philos. Soc. 58(4) (1962), 563–568. [Mu88] Mukai, S.: Finite groups of automorphisms of K3 surfaces and the Mathieu group. Invent. math. 94 (1988), 183–221. [N99] Nakajima, H.: Lectures on Hilbert Schemes of Points on Surfaces. University Lecture Series, 18. American Mathematical Society, Providence, RI, 1999. xii+132 pp. [Se79] Serre, J.P.: Local Fields, GTM 67, Springer-Verlag New York, 1979. [Se88] Serre, J.P.: Algebraic Groups and Class Fields, GTM 117, Springer-Verlag New York, 1988. [Sh12] Shende, V.: Hilbert schemes of points on a locally planar curve and the Severi strata of its versal deformation. Compos. Math. 148(2) (2012), 531–547. [YZ96] Yau, S.T.; Zaslow, E.: BPS states, String duality, and Nodal curves on K3. Nuclear Physics B 471(3) (1996), 503–512.

Department of Mathematics, Indiana University, Bloomington, IN, 47405, U.S.A.

E-mail address: [email protected]