Arxiv:2104.12618V1 [Math.AG] 26 Apr 2021 Ul Akt H Orsodn Tilting Corresponding the to Back Pulls Lsdﬁl Fcaatrsi .Let 0

Home , Abuse of notation

arXiv:2104.12618v1 [math.AG] 26 Apr 2021 ul akt h orsodn tilting corresponding the to back pulls lsdfil fcaatrsi .Let 0. characteristic of field closed ilfcso w ainso utetsevs tbepisadfra and pairs stable sheaves: quotient of strang variants s towards two to attempt on recent [PT09] focus more Thomas will a and t For Pandharipande [Tha94] 3-folds. by Thaddeus Calabi–Yau by used on used also was was It often It pairs. Quite numbers. stable suit of more space are scheme. moduli variants the certain Quot problems, Grothendieck’s some Towards is oversized. sheaves decorated of aro Coh on pair iigaseilzto a ewe ouispaces. moduli between map properness and specialization separatedness a of giving criteria valuative the satisfies orsod oafml fqoin uk,udrsial condition suitable under husks, quotient of family a to corresponds a tblt oyoili ag,apair a large, is polynomial stability eaiestig Suppose setting. relative a standard the of tilts certain pair of the hearts call we case, this In bs ftriooy ecl ha qipdwt morphism a with equipped sheaf a call we terminology, of abuse pair Let most The structure. extra with sheaves are sheaves Decorated hr sada notion: dual a is There e od n phrases. and words Key 2020 ii tbepisaeas nw sqoin uk Kl8.Tee Koll´ There, [Kol08]. husks quotient as known also are pairs stable Limit EOAE HAE N OPIM NTLE HEARTS TILTED IN MORPHISMS AND SHEAVES DECORATED hr safml fsaiiycniin Dfiiin41 endo pairs on defined 4.1) (Definition conditions stability of family a is There . X ahmtc ujc Classification. Subject Mathematics rjciiyo h utsae nteecss eas rv for a prove tilt. a also spa We under Quot spaces cases. Quot corresponding these the in the relating spaces with unde algebra Quot spaces heart, the moduli of standard projectivity the the identify of then tilts We in respectively, monomorphisms, Abstract. eanniglrpoetv ait ihafie oaiainoe na an over polarization fixed a with variety projective nonsingular a be ( X ) Lma31.Tetilting The 3.1). (Lemma eietf ii tbepisadsal rmdsevsa epimorphism as sheaves framed stable and pairs stable limit identify We IBN I,S-HN AG N IGUXIA BINGYU AND WANG, SZ-SHENG LIN, YINBANG eoae sheaf, decorated rmdsheaves framed X ii stable limit → S ( E sasot rjciemrhs.W en torsion a define We morphism. projective smooth a is 1. ,α E, 0 ( ( t t rmr 42,1F8 eodr 42,14D23. 14D22, secondary 14F08; 14D20, primary ,α E, srcue ouisae utsae tilt. space, Quot space, moduli -structure, Introduction ,α E, srcue(rpsto 4.3). (Proposition -structure eafie oeetsefon sheaf coherent fixed a be ewl dniysal ar ihqoinsi the in quotients with pairs stable identify will We . t srcueo ahfie.Aqoin nCoh in quotient A fiber. each on -structure ) H9] rmdsefi fteform the of is sheaf framed A [HL95]. t ssal fadol if only and if stable is ∶ srcue hs er sdntda Coh as denoted is heart whose -structure, ∶ E E 1 0 → → E E 0 ) ) . hscnb nesodas understood be can This . uai oii Hall motivic a in mula α be mn hm n is one them, Among able. utbeconditions. suitable r α ult,se[L9.We [GL19]. see duality, e ( s e sheaves. med sgnrclysurjective. generically is aclt h Verlinde the calculate o e,otiigthe obtaining ces, X § h utshm is scheme Quot the , aosmdl space moduli famous ) h ouispace moduli The 3). ya unfortunate an By . uycrecounting curve tudy rsuidin studied ar and s hnthe When . lgebraically # # ( ( X X ) ) , 2 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA

There is also a family of stability conditions (Definition 4.5) on framed sheaves. We will identify stable framed sheaves with monomorphisms in certain tilted hearts (Proposition 4.7). Over an algebraically closed field of characteristic 0, these moduli spaces coincide with the corresponding Quot spaces [BLM+19], which are generalizations of Quot schemes. As a consequence, these Quot spaces are projective. We also point out that torsion pairs where we tilt the standard heart are defined using the corresponding stability polynomials. Over a Calabu–Yau 3-fold, when E0 ≅ OX and E has dimension 1, the equivalence between stable pairs and quotients in a tilted heart has been obtained by Bridgeland [Bri11] and used to derive a wall-crossing formula between the Donaldson–Thomas (DT) invariants and Pandharipande–Thomas (PT) invariants. As an immediate application of our results, we also derive a formula relating the moduli space of limit stable pairs and the Quot scheme under certain assumptions (Theorem 5.6). Over a nonsingular curve, stable pairs of the form (E,α∶ OX → E) have been identified as epimorphisms in the corresponding tilted heart in [Rot20]. A variant of the result on framed sheaves has also been obtained there. We organize the paper as follows. In §2, we review some basic notions on t-structures. In §3, we compare quotient husks and quotients in a tilted heart. In §4, we identify stable pairs and framed sheaves as epimorphisms and monomorphisms in different tilted hearts. In §5, we review the formalism of motivic Hall algebra and derive the formula relating the Quot spaces under a tilt.

2. t-structures and torsion pairs In this section, we review the basic notions of t-structures, hearts, torsion pairs, and tilts, and prove an observation important to us. A t-structure τ [BBD82] on a triangulated category D is a pair of full subcategories (D⩽0, D⩾0) satisfying the following conditions: (i) Hom(F,G) = 0 for every F ∈ D⩽0 and G ∈ D⩾1; (ii) every object E ∈ D fits into an exact triangle ⩽0 ⩾1 ⩽0 τ E → E → τ E → (τ E)[1] with τ ⩽0E ∈ D⩽0 and τ ⩾1E ∈ D⩾1. ⩽n ⩽0 ⩾n ⩾0 Here, we use the notation D ∶= D [−n] and D ∶= D [−n], for n ∈ Z. The truncation functors τ ⩽n and τ ⩾n are similarly defined. The full subcategory A ∶= D⩽0 ∩ D⩾0 is called the heart of the t-structure, which is an abelian category. The t-structure is bounded if ⩽n ⩾m D = ∪n,m∈ZD ∩D . The cohomology objects of an object E ∈ D with respect to the heart n = ⩽n ⩾n A are defined by HA(E) ∶ (τ τ E)[n]. Assume that A is an abelian category. A pair of full subcategories (T , F) of A is called a torsion pair if Hom(T , F) = 0 and every object E ∈ A fits into an exact sequence (2.1) 0 → T → E → F → 0 with T ∈ T and F ∈ F. It follows that T and F are respectively closed under taking ⊥ quotients and sub-objects. Note that if (T , F) is a torsion pair, then F = T is the right orthogonal to T in A. DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 3

Example 2.1. Let X be a noetherian scheme of dimension n. For any integer 0 ⩽ d < n, we deﬁne = ∈ ⩽ Coh⩽d(X) ∶ {E Coh(X)S dim E d}. Here and henceforward, we use dim E to denote the dimension of the support of the sheaf = ⊥ E. When d 0, we simply write Coh0(X). Then (T , T ) is a torsion pair in the abelian category Coh(X) (cf. [HL10, Deﬁnition 1.1.4]). Let A ⊂ D be the heart of a t-structure and (T , F) a torsion pair in A. We can tilt the # t-structure on D to obtain a new one. The resulting t-structure has heart A = ⟨F, T [−1]⟩, # the extension-closure. Also, the shift A [1] can be described as # = ∈ 0 ∈ −1 ∈ i = ≠ A [1] E D S HA(E) T , HA (E) F, HA(E) 0 for i 0, −1 . The following observation is important for us. Lemma 2.2. Let A be the heart of a bounded t−structure on a triangulated category D and # (T , F) a torsion pair of A. Let A be the tilt in A at this pair. # (i) If E0 ∈ F, then a morphism α∶ E0 → E in A is an epimorphism if and only if α is a morphism in A with E ∈ F and the cokernel coker (α) taken in A lies in T . ∈ # (ii) If E0 T , then a morphism β∶ E → E0 in A [1] is a monomorphism if and only if β is a morphism in A with E ∈ T and the kernel ker(β) taken in A lies in F. Proof. We only show (ii), as the proof of (i) is similar (cf. [Bri11, Lemma 2.3]). Given ∈ # E0 T and a short exact sequence in A [1] β 0 → E Ð→ E0 → Q → 0, we have

−1 −1 0 β 0 (2.2) 0 → HA (E) → 0 → HA (Q) → HA(E) Ð→ E0 → HA(Q) → 0 by taking cohomology with respect to the original t-structure. Note that an object F ∈ D lies # ⊂ −1 ∈ 0 ∈ i = ≠ in A [1] D precisely when HA (F ) F, HA(F ) T and HA(F ) 0 for i 0, −1. It follows i = ≠ ∈ # = = −1 ∈ that HA(E) 0 for all i 0, and thus E A ∩ A [1] T . Moreover, ker(β) HA (Q) F. ∈ ∈ For the converse, take a morphism β∶ E → E0 in A with E T , ker(β) F and embed it in a distinguished triangle β (2.3) E Ð→ E0 → Q → E[1].

Since T is closed under taking quotients and E0 ∈ T , the same long exact sequence (2.2) 0 ∈ −1 = ∈ i = ≠ then shows that HA(Q) T , HA (Q) ker(β) F and HA(Q) 0 for i 0, −1. Thus # # Q ∈ A [1]. It follows that (2.3) deﬁnes a short exact sequence in A [1], and hence β is a # monomorphism in A [1].

3. Quot spaces and moduli of quotient husks We construct a family of t-structures for a smooth projective morphism, and compare the Quot space in the sense of [BLM+19, §11] with the moduli space of quotient husks constructed by Koll´ar [Kol08]. 4 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA

Over a field k, let S be a regular noetherian scheme of finite type and f∶ X → S be a smooth projective morphism, with a fixed relatively ample line bundle. On each fiber Xs for s ∈ S, we can define the following tilting t-structure for some fixed non-negative integer m: = ∈ ⩽ Ts {E Coh(Xs)S dim E m}, = ∈ = ∈ Fs {F Coh(Xs)S Hom(E, F ) 0, ∀E Ts}. Then (Ts, Fs) defines a torsion pair of Coh(Xs). If we denote the standard t-structure on b ⩽0 ⩾0 D (Xs) as (Ds ,Ds ), the tilting t-structure is defined by #,⩽0 = ∈ ⩽1 1 ∈ #,⩾0 = ∈ ⩾0 0 ∈ (3.1) Ds {E Ds S H (E) Ts} and Ds {E Ds S H (E) Fs}. The cohomologies are taken with respect to the standard t-structure. The heart of the # ⩽0 ⩾0 t-structure is As = Ds ∩Ds . It is an abelian category, so we can talk about epimorphisms in it. We will show that the family of t-structures defined above integrates over S, namely, they are pullbacks of a t-structure on the total space X. b We have two full subcategories of D (X): = ∈ ∈ ⩽ T {E Coh(X)S ∀s S, dim Es m}, F = {F ∈ Coh(X)S Hom(E, F ) = 0, ∀E ∈ T }. For brevity, we say objects in T have relative dimension ⩽ m over S.

Lemma 3.1. (T , F) is a torsion pair of Coh(X). Proof. This is similar to the absolute case. For any coherent sheaf E, We can take its maximal subsheaf E′ whose support has a relative dimension ⩽ m over S. Then the quotient ′ E~E belongs to F. b The torsion pair (T , F) induces the following t-structure on D (X): #,⩽0 ⩽1 1 #,⩾0 ⩾0 0 (3.2) D = {E ∈ D S H (E) ∈ T } and D = {E ∈ D S H (E) ∈ F} ⩽0 ⩾0 b + where (D ,D ) is the standard t-structure on D (X). By [BLM 19, Theorem 5.7], there b is a canonical way to pullback this t-structure to D (XT ) for an arbitrary morphism T → S, which is faithful [Kuz11] with respect to f since f is ﬂat. Here,

XT = T ×S X. Suppose s is a point of S (possibly nonclosed) and we denote # ⩽0 # ⩾0 ((D )s , (D )s ) to be the pullback t-structure on the fiber Xs. We have the following: Lemma 3.2. Pullbacks of the t-structure agree with the fiberwise defined ones, namely, # ⩽0 = #,⩽0 # ⩾0 = #,⩾0 (D )s Ds and (D )s Ds . DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 5

+ Proof. If s is a closed point, we can use [BLM 19, Theorem 5.7 (3)]. Let T = {s} and φ = = (−∞,0] = # ⩽0 be the closed embedding of {s} into S. Take a −∞, b 0, then Ds (D )s and (−∞,0] #,⩽0 D = D . Let is∶ Xs → X be the inclusion of the fiber Xs in X, we have # ⩽0 = ∈ b ∈ #,⩽0 (D )s {F D (Xs)S is∗F D } = ∈ b ∈ ⩽1 1 ∈ = #,⩽0 {F D (Xs)S F D , H (F ) T } Ds . The other equality can be shown similarly. For a non-closed point s, we can still apply [BLM+19, Theorem 5.3 & 5.7] and obtain the result. Remark 3.3. Suppose X = S ×Y → S is a trivial family with Y over an algebraic closed field b k, the t-structure (3.2) may not be the pullback of the one on D (Y ), which is similar to (3.1). Namely, it is not constant in the sense of [AP06,Pol07], as illustrated in the following example. Example 3.4. Let Y = P1 and S = P1. Let ∆ ≅ P1 ⊂ X = S × Y be the diagonal. Then #,⩽0 #,⩾0 O∆[−1] lies in the the heart D ∩ D . However, it does not lie in the heart of the pullback of the corresponding t-structure, according to [Pol07, Lemma 3.3.2]. ∈ ⊆ 1 Let F0 F Coh(X) be pure and flat over S. Let T be an S-scheme and F0T denote the pullback of F0 to XT . A family of epimorphisms in the tilted hearts of the family of b t-structures (3.1), parametrized by T , is a morphism F0T → F in D (XT ) satisfying the following conditions.

(a) F is ﬂat over T , i.e. the derived pullback Ft of F to a ﬁber Xt lies in the tilted # b heart Coh (Xt) of the induced t-structure on D (Xt). # ∈ (b) The morphism F0t → Ft in Coh (Xt) is an epimorphism for all t T . Here, = Xt Spec κ(t) ×S X.

Remark 3.5. In this setting, F is automatically a sheaf because its derived pullback Ft to # each fiber is a quotient of F0t in Coh (Xt) and by Lemma 2.2, it must be a sheaf in Ft . Let # op Quotf (F0,P )∶ (S-Schemes) → Sets be the functor that sends an S-scheme T to the set of families of epimorphisms in the tilted hearts where the quotient has Hilbert polynomial P . If the family of t-structures universally satisfies openness of flatness, then the functor is represented by an algebraic space locally of finite presentation over S [BLM+19, Proposition 11.6]. We next recall the notion of a husk. A husk of a coherent sheaf F on a normal scheme Y is a homomorphism q∶ F → E such that (a) q is an isomorphism on all n-dimensional points, where n = dim F ; (b) E is pure of dimension n.

1The Quot functor can be deﬁned in a much broader context, but let us restrict to this situation, which is enough for our purpose. 6 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA

A quotient husk of a fixed coherent sheaf E0 is a homomorphism q∶ E0 → E such that it factors as E0 → F → E where the first arrow is an epimorphism and the second arrow is a husk. For a morphism f∶ X → S, one can similarly define the notion of a family of quotient husks to be a morphism q∶ E0 → E such that (a) E is pure and flat over S; (b) On each fiber Xs, the homomorphism qs∶ E0s → Es is a quotient husk.

We say a sheaf E on X is pure over S if for every s ∈ S, the restriction Es is pure of the same dimension. It is immediate to check that this definition of a family of quotient husks has base-change property, therefore we can define a moduli functor of families of quotient husks op QHuskf (E0,P )∶ (S-Schemes) → Sets ∈ where P is a fixed polynomial with rational coefficients and E0 Coh(X). It sends an ∗ S-scheme T → S to the set of families of quotient husks fT E0 → ET on XT such that when restricted to each fiber Xt for t ∈ T , the Hilbert polynomial of Et is P . Then we have the following existence theorem on the moduli space of quotient husks by Kollár:

Theorem 3.6. QHuskf (E0,P ) is represented by a proper algebraic space. The following proposition shows that the notion of a family of quotient husks is equivalent to a family of epimorphisms with respect to the family of t-structures (3.1).

Proposition 3.7. Assume E0, E are two sheaves ﬂat over S and the restriction Es has (m + 1)-dimensional support, then a homomorphism of sheaves q∶ E0 → E is a family of quotient husks if and only if it is a family of quotients with respect to the family of t- structures (3.1).

Proof. Suppose q∶ E0 → E is a family of quotient husk, then for every point s ∈ S, the restriction qs∶ E0s → Es is a quotient husk on Xs. In particular, Es is a pure sheaf with (m + 1)-dimensional support. This implies Es is contained in Fs. The fact that qs is surjective at all (m + 1)-dimensional points implies its cokernel is supported in a locus of dimension m or less. Together with Lemma 2.2, this shows that qs is an epimorphism in # Coh (Xs). Conversely, if q is a family of quotients, then on each fiber, the homomorphism qs factors as E0s → Im(qs) → Es where the first arrow is an epimorphism. By Lemma 2.2, the cokernel of qs is in Ts so it is supported in a locus of dimension ⩽ m. Therefore the second arrow above has to be an isomorphism at all (m + 1)-dimensional points. Also by Lemma 2.2, Es is contained in Fs and therefore does not contain any subsheaf supported on a locus of dimension m or less. This implies Es is pure and E is pure over S. This finishes the proof that qs is a quotient husk. = # Let E0 F0 and P be a polynomial of degree m + 1, then the functor Quotf (E0,P ) is the same as the functor QHuskf (E0,P ), that is, we have the following proposition. # = Proposition 3.8. Given an S-scheme T , Quotf (E0,P )(T ) QHuskf (E0,P )(T ). # ∈ Proof. Suppose we are given an element αT ∶ E0T → E in Quotf (E0,P )(T ). For t T , the # restriction αt∶ E0t → Et lies in the heart Coh (Xt) of the pullback t-structure. By Lemma DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 7 ∈ 2.2, Et Coh(Xt). Thus, E is a coherent sheaf flat over T . On the other hand, αt has cokernel (taken in Coh(Xt)) in Tt. Therefore, αT is a family of quotient husks. Given a family of quotient husks αT ∶ E0T → E, the restriction αt is a quotient husk. By # Proposition 3.7 and Lemma 2.2, E is flat over T and αt is a quotient in Coh (Xt). By identifying the moduli functors, we can easily obtain the projectivity of the Quot space in some cases. Theorem 3.9. Let k be an algebraic closed field of characteristic 0 and f∶ X → S = Spec k be a nonsingular projective scheme. Let P be a polynomial of degree m + 1. Then the Quot # space Quotf (E0,P ) is projective.

Proof. This follows from the projectivity of QHuskf (E0,P ) in this set-up, which is obtained via a geometric invariant theoretic construction [Lin18]. Remark 3.10. Quotient husks are also known as limit stable pairs [Lin18]. Assuming the universal openness of ﬂatness for the family of t-structures, one would be able to obtain the projectivity of the Quot space over a general base S in characteristic 0, by carrying out a GIT construction [Lin18, Remark 4.6]. For stable pairs with respect to a smaller stability condition, we can also identify them as quotients in the heart of a tilting t-structure. We will study this next.

4. Stable pairs and framed sheaves Over an algebraically closed field of characteristic 0, let X be a nonsingular projective ∈ ∈ variety with a fixed polarization OX (1). Let E0 Coh(X) be fixed. Let P Q[m] be a fixed polynomial of degree d, which will be used as a Hilbert polynomial. Let δ, τ ∈ Q[m] be polynomials with positive leading coefficients. Given a coherent sheaf E on (X, OX (1)), we denote its Hilbert polynomial by PE, its = multiplicity by rE and its reduced Hilbert polynomial by pE PE~rE.

4.1. Stable pairs. We consider homomorphisms of the form E0 → E. Definition 4.1. A pair (E,α∶ E0 → E) with α ≠ 0 is δ-stable if E is pure and for every subsheaf F ⊂ E, < ⊂ (i) pF + δ~rF pE + δ~rE if im α F , < (ii) pF pE + δ~rE otherwise. We can replace the strong inequalities by weak inequalities to define δ-semistability. Stability can be equivalently defined in terms of quotients. When deg δ ⩾ dim E, a δ-stable pair is called a limit stable pair, which is the same as a quotient husk.

Theorem 4.2 (Y. Lin). There is a projective coarse moduli space SE0 (P, δ) of S-equivalence classes of δ-semistable pairs with Hilbert polynomial P . It contains an open subscheme Ss P, δ δ E0 ( ) as the ﬁne moduli space of -stable pairs. 8 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA = We consider δ-stable pairs (E,α) with PE P for a small δ: deg δ < deg P = d.

Let r = rE and P + δ λ = . r λ λ We deﬁne a torsion pair (T , F ) on Coh(X): λ = ∈ ⩽ < > T {E Coh(X)S dim E d and ∀ quotient sheaf E ↠ G, dim G d or pG λ} and λ = ∈ ⊂ ⩽ ⩽ F {E Coh(X)S ∀ subsheaf F E with dim F d, F is pure of dim. d and pF λ}. λ,# Then, we denote the heart of the tilting t-structure as Coh (X). λ Proposition 4.3. Suppose E has Hilbert polynomial P and E0 ∈ F . We also assume that δ is not a critical value, namely, there are no strictly semistable pairs with respect to λ,# δ. Then, α∶ E0 → E is an epimorphism in Coh (X) if and only if (E,α∶ E0 → E) is a δ-stable pair. λ,# ∈ λ Proof. Suppose that α is an epimorphism in Coh (X). By the assumption E0 F and λ λ Lemma 2.2, we know that α is a morphism in Coh(X) with E ∈ F and coker (α) ∈ T . Given a quotient q∶ E ↠ G in Coh(X), if q ○ α = 0, then G is a quotient of coker (α). Thus, pG > λ. If q ○ α ≠ 0, let F = ker q. Then pF ⩽ λ. Since δ is not critical, this is a strict < inequality. Therefore, λ (PG + δ)~rG. We have shown (E,α) is a δ-stable pair. ⊂ Conversely, suppose (E,α∶ E0 → E) is a δ-stable pair. Then for every subsheaf F E, < ∈ λ ⩽ pF λ. Thus, E F . On the other hand, coker (α) has dimension d. Given a quotient G of coker (α), it is also a quotient of E and the composition E0 → E → G is zero. > ∈ λ Thus, pG λ. Hence, coker (α) T . Again by Lemma 2.2, α is an epimorphism in λ,# Coh (X). Similar to the quotient husks case, by identifying two moduli functors, we can identify the moduli space with the corresponding Quot space. Theorem 4.4. With assumptions and notation as in the previous proposition, the moduli Ss P, δ = S P, δ δ λ,# E ,P space E0 ( ) E0 ( ) of -stable pairs is isomorphic to the Quot space Quot ( 0 ) λ,# parametrizing quotients of E0 with Hilbert polynomial P in the heart Coh (X). In particular, the Quot space is projective. 4.2. Framed sheaves. We have a notion dual to pairs: framed sheaves. They are homomorphisms of the form E → E0.

Deﬁnition 4.5. A framed sheaf is a coherent sheaf E with Hilbert polynomial PE = P , together with a nonzero framing α∶ E → E0. It is τ-stable if ker α is zero or pure of dimension d, and for every nonzero subsheaf F ⊂ E of dimension d, < ⊂ (i) pF pE − τ~rE if F ker α, < (ii) pF − τ~rF pE − τ~rE otherwise.

Theorem 4.6 (Huybrechts-Lehn). There is a projective coarse moduli space FE0 (P, τ) of S-equivalence classes of τ-semistable framed sheaves with Hilbert polynomial P . It contains F s P, τ τ an open subscheme E0 ( ) as the ﬁne moduli space of -stable framed sheaves. DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 9

When deg τ ⩾ d, the moduli space is isomorphic to a Quot scheme. Therefore, we again consider a small stability parameter τ: deg τ < deg P = d. Now, we let P − τ λ = . r In some cases, we can also identify stable framed sheaves as monomorphisms in a tilted heart.

λ Proposition 4.7. Suppose E has Hilbert polynomial P and E0 ∈ T . We also assume that λ,# τ is not a critical value. Then, α∶ E → E0 is a monomorphism in Coh (X)[1] if and only if (E,α∶ E → E0) is a τ-stable framed sheaf. The proof is similar to that of Proposition 4.3. For completeness, we include it here. λ,# Proof. Suppose that α∶ E → E0 is a monomorphism in Coh (X)[1]. By the assumption ∈ λ ∈ λ E0 T and Lemma 2.2, we know that α is a morphism in Coh(X) with E T and ∈ λ ⊂ > ∈ λ < ker(α) F . Given a subsheaf F E, pE~F λ, because E T . Then, pF − τ~rF λ. If F ⊂ ker α, then pF ⩽ λ, which is actually a strict inequality, since we assume τ is not critical. Therefore, (E,α∶ E → E0) is τ-stable. Conversely, we assume that (E,α∶ E → E0) is τ-stable. First, τ-stability implies that for > > ∈ λ any dimension d quotient sheaf Q of E, pQ − τ~rQ λ or pQ λ. Therefore, E T . On the other hand, the τ-stability also implies that if nonzero, ker(α) is pure of dimension d λ and has reduced Hilbert polynomial ⩽ λ. Furthermore, ker(α) ∈ F . Again by Lemma 2.2, λ,# α is a monomorphism in Coh (X)[1]. Remark 4.8. This is a variant of [Rot20, Lemma 5.5]. Thus, by identifying two moduli functors, we can identify the moduli space of τ-stable framed sheaves with the corresponding Quot space. Theorem 4.9. With assumptions and notation as in the previous proposition, the moduli F s P, τ = F P, τ τ space E0 ( ) E0 ( ) of -stable framed sheaves is isomorphic to the Quot space λ,#[1] Quot (E0,PE0 − P ) parametrizing quotients of E0 with Hilbert polynomial PE0 − P in λ,# the heart Coh (X)[1]. In particular, the Quot space is projective. 5. Change of Quot space under tilting In this section, we will prove a formula relating the moduli space of quotient husks and the Grothendieck’s Quot scheme, which parameterize quotient sheaves supported in dimension no more than one. We follow Bridgeland’s treatment of Hall algebra identities in [Bri11, §6]. 5.1. The stack of pairs. We first introduce a modification of stack of sheaves with sections, which were constructed in [Bri11, §2.3]. Let X be a nonsingular projective variety ∈ over C and E0 Coh(X) fixed. We denote by M the stack of coherent sheaves on X. It is an Artin stack, locally of finite type over C. There is another stack M(E0) with a morphism q∶M(E0) →M parameterizing pairs (E,α∶ E0 → E). Indeed, the objects of M(E0) lying 10 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA over a scheme S are pairs (E,α) consisting of an S-flat coherent sheaf E on S ×X together with α∶ E0S → E where E0S denotes the pullback of E0 under the projection S × X → X. = Let f∶ T → S be a morphism of schemes and fX f ×idX . Given an object (F, β) lying over ∗ T , a morphism θ∶ (F, β) → (E,α) lying over f is an isomorphism θ∶ fX E → F on T ×X with ∗ = ∗ θ ○ fX α β ○ κ, where the morphism κ∶ fX E0S → E0T denotes the canonical isomorphism of pullbacks. The morphism q of stacks is defined by forgetting the data of the morphism α in the obvious way. By an easy modification of the argument of [Bri11, Lemma 2.4], we have the following lemma.

Lemma 5.1. The stack M(E0) is an Artin stack and the morphism q is representable and of finite type. The following lemma is a result of the fibers of the morphism q. ∈ Lemma 5.2. Let E0 Coh(X) be fixed. There is a stratification of M by locally closed ⊂ ∈ = substacks Mr M such that the objects of M(C) are E Coh(X) with hom(E0, E) r. The pullback of q∶M(E0) →M to Mr is a locally trivial fibration in the Zariski topology, with fiber Cr.

Proof. Let S be a scheme. Given an S-flat coherent sheaf E on S×X, we write hom(E0S , E) op for the set-valued covariant funtor on (S-Schemes) , which associates to any S-scheme ∗ ∗ f∶ T → S the set Hom(fX E0S, fX E) of OT ×X -linear morphism . By a standard limit argument (cf. [Gro66, (8.5.2), (8.8.2), (8.9.1), (11.2.6)]), we may assume that S is noetherian. According to the results of Grothendieck (see [Nit05, Theorem 5.8] and references therein), there is a coherent sheaf G(E0S , E) on S such that the funtor hom(E0S , E) is represented by the linear scheme Spec (SymOS G(E0S , E)). Then the remaining proof is essentially the same as in [Bri11, Lemma 2.5]. 5.2. Motivic Hall algebra. We are going to recall the notion of motivic Hall algebras. For a more detailed discussion, we refer to [Bri11, Bri12]. We denote C by the subcategory Coh⩽1(X) of Coh(X), and this corresponds to an open and closed substack C ⊂ M by the usual abuse of notation. There exists a stack C(2) of short exact sequences in the category C. It comes with three distinguished morphisms (2) a1, a2 and b∶C → C. These morphisms correspond to sending a short exact sequence 0 → A1 → B → A2 → 0 to the sheaves A1, A2 and B respectively. We remark that (a1, a2) is of finite type [Bri12, Lemma 4.2]. The motivic Hall algebra, denoted by H(C), is the relative Grothendieck group K(St~C) over the stack C. By definition, it is defined to be the complex vector space spanned by isomorphism classes of symbols [X → C] where X is an Artin stack of finite type over C with affine geometric stabilizers , modulo three relations: the scissor relations for finite disjoint satcks, geometric bijection relations and Zariski fibration relations (see [Bri12, Definition 3.10]). It is equipped with a noncommutative product ∗ given explicitly by the rule f1 f2 = b○h [X1 Ð→ C] ∗ [X2 Ð→ C] [Z Ð→ C], DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 11 where h is defined by the Cartesian diagram

h (2) b Z C C. (a1,a2)

f1×f2 X1 × X2 C × C The unit is given by 1 = [Spec C → C], which corresponds to the zero object in C and the product ∗ is associative [Bri12, Theorem4.3]. On the other hand, there is a natural grading on H(C) by the monoid ∆ consisting of ⩽ classes of sheaves supported in dimension 1. More precisely, let N1(X) denote the abelian group of cycles of dimension one modulo numerical equivalence. We define the monoid by = ∈ > = ⩾ ∆ {(β, n) N1(X) ⊕ Z S β 0 or β 0 and n 0} ⊂ (cf. [Bri11, §2.1]). There are open and closed substacks Cγ C, the stacks of objects of class γ ∈ ∆. Thus elements of H(C) are naturally graded by the monoid ∆. An element ∈ [f∶ X → C] is homogeneous of degree γ ∆ if f factors through the substack Cγ. 5.3. Laurent subsets. Let us summarize sections 5.2 and 6.1 of [Bri11]. A subset S ⊂ ∆ ∈ ∈ ∈ is Laurent if for all β N1(X), the collection {n Z S (β, n) S} is bounded below. Let Φ denote the set of all Laurent subsets. For the ∆-graded Hall algebra H(C), we can use Φ to define a new algebra, denoted = ∈ by H(C)Φ. Elements of such new algebra are of the form a ∑γ∈S aγ where S Φ and ∈ ⊂ aγ H(C)γ H(C). There is a natural topology and product ∗ on H(C)Φ induced by projection operators (see [Bri12, §5.2]). To define a stability condition, we fix an ample divisor H on X. Given a class γ = (β, n) ∈ ⋅ −1 −∞ ∞ ∞ ∆ define the slope µ(γ) = n(β H) ∈ ( , ]. If β = 0, we consider γ to have slope , and otherwise µ(γ) ∈ Q. −∞ ∞ Given an interval I ⊂ ( , ], define SS(I) ⊂ C to be the full subcategory consisting of zero objects together with those one-dimensional sheaves whose Harder-Narasimhan factors ∞ all have slope in I (see [Bri11, §6.1]). We write SS(I) = SS(⩾ µ) if I = [µ, ]. Then the following lemma follows from Lemmas 5.3, 6.2 and (31) of [Bri11]. ∞ Lemma 5.3. The subcategory SS([µ, )) defines an invertible element 1SS([µ,∞)) in H(C)Φ. = = ∞ 5.4. Identities in the Laurent Hall algebra. Let T Coh0(X) SS( ). Consider the ∩ ⊥ # ∩ − torsion pair (T , F C) of C, where F = T . Then C = ⟨F C, T [ 1]⟩ is the tilt of C. For the Grothendieck’s Quot scheme Quot E0 and the moduli space of quotient husks ( ) 2 (or limit stable pairs, Proposition 3.8) QHusk(E0), we introduce ∶= ∩ Quot(E0)⩽1 Quot(E0) C, ∶= ∩ ∈ QHusk(E0)⩽1 QHusk(E0) C if E0 F which parameterize quotients of E0 supported in dimension ⩽ 1. By the same argument of [Bri11, Lemma 2.6], we can view Quot(E0)⩽1 and QHusk(E0)⩽1 as open substacks of

2 # The condition E0 ∈F implies that if α∶ E0 → E is an epimorphism in Coh(X) , then E ∈ F ⊂ Coh(X) by Lemma 2.2 12 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA the moduli stack C(E0). Namely, these C-valued points are morphisms E0 → E that are # epimorphisms in the categories C and C respectively. The morphisms Quot(E0)⩽1 → C and QHusk(E0)⩽1 → C, ∶ which are the restrictions to Quot(E0)⩽1 and QHusk(E0)⩽1 of q C(E0) → C, deﬁne elements # Q⩽1 and Q⩽1 of H(C)Φ by the similar argument of [Bri11, Lemma 5.5]. ∶ ∶= Given a substack i N → C, we write 1N [N → C] in H(C). Pulling back the morphism ∶ ⊂ q C(E0) → C to N C given a stack N (E0) with a morphism N (E0) → N and hence an E0 ∶= element 1N [N (E0) → C] in H(C). By abuse of notation, we use the same symbol for an open substack of C and the corresponding full subcategory of C deﬁned by its C-valued points. Following [Bri11] we establish the torsion pair and Quot space identities in the next two lemmas.

Lemma 5.4. The following identities hold in the Laurent Hall algebra H(C)Φ. (a) 1SS(⩾µ) = 1T ∗ 1SS([µ,∞)). E0 → ∗ − = (b) limµ −∞ Q⩽1 1SS(⩾µ) 1SS(⩾µ) 0. # E0 → ∗ − = ∈ (c) limµ −∞ Q⩽1 1SS([µ,∞)) 1SS([µ,∞)) 0 if E0 F. The proof of Lemma 5.4 is essentially the same as in [Bri11, Proposition 6.5], noticing the boundedness of the Quot scheme and the moduli space of stable pairs. Remark that the ∈ geometric bijection relations plays an essential role, and we need the assumption E0 F of (c) to use Lemma 2.2 instead of [Bri11, Lemma 2.3].

Lemma 5.5. Assume that E0 is locally free. There is an identity E0 = E0 ∗ E0 1SS(⩾µ) 1T 1SS([µ,∞)) in H(C)Φ. Proof. Form Cartesian squares

p j (2) b Y XC C (a1,a2) × ∞ (q,id) × ∞ (i,q) × T (E0) SS([µ, ))(E0) T SS([µ, ))(E0) C C.

E0 ∗ E0 ○ ○ ∶ Then 1T 1SS([µ,∞)) is represented by the composite morphism b j p Y → C. Note that, ∶ by Lemma 5.2, the morphism of stacks q T (E0) → T is a Zariski fibration, with fiber over a sheaf T being the vector space Hom(E0, T ). By pullback the same is true for the map p. Since the morphism (a1, a2) satisfies the iso-fibration property of [Bri12, Lemma A.1], the groupoid of S-valued points of X is as follows. The objects are a short exact sequences of S-flat sheaves on S × X γ (5.1) 0 → T → E Ð→ F → 0 ∈ ∈ ∞ ∶ together with T T , F SS([µ, )) and a map α E0S → F . The morphisms are isomor- phisms of short exact sequences commuting with the map α. DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 13 ○ ∶ Recall that by Lemma 5.4 (a), b s Z → SS(⩾ µ) induces an equivalence on C-valued points. On the other hand, consider a Cartesian diagram h ⩾ ⊂ W SS( µ)(E0) C(E0) q

b○s Z SS(⩾ µ) ⊂ C. ○ E0 Since b s is a geometric bijection, so too is h. Thus the element 1SS(⩾µ) can be represented by the morphism q ○ h. The groupoid of S-valued points of W can be represented by the short exact sequences ∶ = ○ (5.1) with a map δ E0S → E. Setting α γ δ defines a morphism of stacks W → X . It is easy to see that this is a Zariski fibration, with fiber over a C-valued point of X represented by a sequences (5.1) with a map α being a vector space for Hom(E0, T ). Indeed, we have a long exact sequence 1 0 → Hom(E0, T ) → Hom(E0, E) → Hom(E0, F ) → Ext (E0, T ) ∨ ⊗ on X. Since the dimension of support of T is zero, so too is that of E0 T . For a locally free sheaf E0, we get 1 ≅ 1 ∨ ⊗ = Ext (E0, T ) H (X, E0 T ) 0 ∶ by the dimensional reason. Since W → X is the same fiber Hom(E0, T ) as the map p Y → X , the result follows from the Zariski relation [W → X → C] = [Y → X → C]. We are now in a position to give the formula relating Quot(E0)⩽1 and QHusk(E0)⩽1. ∈ Theorem 5.6. Assume that E0 F and it is locally free. There is an identity ∗ = E0 ∗ # Q⩽1 1T 1T Q⩽1 in H(C)Φ. Proof. By Lemma 5.4 (a) and Lemma 5.5, the expression (b) of Lemma 5.4 can be rewritten ∗ ∗ − E0 ∗ E0 −∞ Q⩽1 1T 1SS([µ,∞)) 1T 1SS([µ,∞)) → 0 as µ → . E0 Multiplying (c) of Lemma 5.4 on the left by 1T gives E0 ∗ # ∗ − E0 ∗ E0 −∞ 1T Q⩽1 1SS([µ,∞)) 1T 1SS([µ,∞)) → 0 as µ → . Thus E0 ∗ # ∗ − ∗ ∗ −∞ 1T Q⩽1 1SS([µ,∞)) Q⩽1 1T 1SS([µ,∞)) → 0 as µ → . By Lemma 5.3, we can multiply the inverse of 1SS([µ,∞)) and deduce the result. Remark 5.7. In [Tod20], Toda studied the higher rank DT/PT correspondence, via stable objects in the derived category of coherent sheaves. He applied the integration map to the moduli stacks. By Behrend’s result [Beh09], the integrations are related to higher rank DT and PT invariants. The invariants are defined using the virtual fundamental classes, whose existence are guaranteed by the symmetric obstruction theories. The moduli space of quotient husks/limit stable pairs can also be viewed as a version of higher rank PT moduli space. Over a Calabi–Yau 3-fold, we can also apply the integration map. However, the question whether the result is a deformation invariant remains, due to the absence of a result on a virtual fundamental class at the moment. 14 YINBANGLIN,SZ-SHENGWANG,ANDBINGYUXIA

Acknowledgements. We would like to thank Dingxin Zhang for helpful discussions. YL was supported by Grant 2017M620726 from China Postdoctoral Science Foundation. SSW thanks Shing-Tung Yau Center of Southeast University for providing a stimulating environ- ment, and was supported by the Fundamental Research Funds for the Central Universities 2242020R10048.

References [AP06] Dan Abramovich and Alexander Polishchuk. Sheaves of t-structures and valuative criteria for stable complexes. J. Reine Angew. Math., 590:89–130, 2006. [BBD82] A. A. Be˘ılinson, J. Bernstein, and P. Deligne. Faisceaux pervers. In Analysis and topology on singular spaces, I (Luminy, 1981), volume 100 of Astérisque, pages 5–171. Soc. Math. France, Paris, 1982. [Beh09] Kai Behrend. Donaldson-Thomas type invariants via microlocal geometry. Ann. of Math. (2), 170(3):1307–1338, 2009. [BLM+19] Arend Bayer, Mart´ıLahoz, Emanuele Macr`ı,Howard Nuer, Alexander Perry, and Paolo Stellari. Stability conditions in families. arXiv:1902.08184, 2019. [Bri11] Tom Bridgeland. Hall algebras and curve-counting invariants. J. Amer. Math. Soc., 24(4):969– 998, 2011. [Bri12] Tom Bridgeland. An introduction to motivic Hall algebras. Adv. Math., 229(1):102–138, 2012. [GL19] Thomas Goller and Yinbang Lin. Rank-one sheaves and stable pairs on surfaces. arXiv:1907.05180v2, 2019. [Gro66] Alexander Grothendieck. Eléments´ de géométrie algébrique. IV. Etude´ locale des schémas et des morphismes de schémas. III. Inst. Hautes Etudes´ Sci. Publ. Math., (28):255, 1966. [HL95] Daniel Huybrechts and Manfred Lehn. Framed modules and their moduli. Internat. J. Math., 6(2):297–324, 1995. [HL10] Daniel Huybrechts and Manfred Lehn. The geometry of moduli spaces of sheaves. Cambridge Mathematical Library. Cambridge University Press, Cambridge, second edition, 2010. [Kol08] János Kollár. Hulls and husks. arXiv:0805.0576, 2008. [Kuz11] Alexander Kuznetsov. Base change for semiorthogonal decompositions. Compos. Math., 147(3):852–876, 2011. [Lin18] Yinbang Lin. Moduli spaces of stable pairs. Pacific J. Math., 294(1):123–158, 2018. [Nit05] Nitin Nitsure. Construction of Hilbert and Quot schemes. In Fundamental algebraic geometry, volume 123 of Math. Surveys Monogr., pages 105–137. Amer. Math. Soc., Providence, RI, 2005. [Pol07] Alexander Polishchuk. Constant families of t-structures on derived categories of coherent sheaves. Mosc. Math. J., 7(1):109–134, 167, 2007. [PT09] R. Pandharipande and R. P. Thomas. Curve counting via stable pairs in the derived category. Invent. Math., 178(2):407–447, 2009. [Rot20] Franco Rota. Some quot schemes in tilted hearts and moduli spaces of stable pairs. arXiv:2005.02202, 2020. [Tha94] Michael Thaddeus. Stable pairs, linear systems and the Verlinde formula. Invent. Math., 117(2):317–353, 1994. [Tod20] Yukinobu Toda. Hall algebras in the derived category and higher-rank DT invariants. Algebr. Geom., 7(3):240–262, 2020. DECORATED SHEAVES AND MORPHISMS IN TILTED HEARTS 15

School of Mathematical Sciences, Tongji University, Shanghai, China Email address: yinbang [email protected] Shing-Tung Yau Center of Southeast University, Southeast University, Nanjing, China Email address: [email protected] Morningside Center of Mathematics, Chinese Academy of Science, Beijing, China Email address: [email protected]