betaFeedback

WorksMathematics

Remarks on Fourier–Mukai invariances

Abstract

This note proves that smoothness and Gorensteinness are Fourier–Mukai invariants for proper flat families of Noetherian algebraic spaces. Moreover, a kernel realizing such a partnership is perfect if, and only if, the families are smooth.

Introduction

Fourier–Mukai partnerships provide rich comparisons of geometric objects through their derived categories. It first appeared in the work of [10], and has been extended to relative settings [6, 4, 5]. A fundamental question is which geometric properties are preserved under Fourier–Mukai partnerships.

We first fix notation. Let fi:Yi→Sf_i: Y_i \to S be proper flat morphisms of Noetherian algebraic spaces for i=1,2i=1,2. Denote by pi:Y1×SY2→Yip_i: Y_1 \times_S Y_2 \to Y_i the projections. For E∈Dcohb(Y1×SY2)E \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2), the associated integral transform is ΦE:Dqc(Y1)→Dqc(Y2)\Phi_E: D_{\mathrm{qc}}(Y_1) \to D_{\mathrm{qc}}(Y_2) where A↦R(p2)∗(L(p1)∗A⊗LE)A \mapsto R(p_2)_*(L(p_1)^*A \otimes^L E). We say that Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners if ΦE\Phi_E restricts to an equivalence Dcohb(Y1)→Dcohb(Y2)D^b_{\mathrm{coh}}(Y_1) \to D^b_{\mathrm{coh}}(Y_2) for some such kernel E∈Dcohb(Y1×SY2)E \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2).

Results

For smooth projective schemes over a field, kernels of such transforms are perfect complexes. For nonsmooth families, perfectness need not hold.

Theorem 1.1. Let fi:Yi→Sf_i: Y_i \to S be proper flat morphisms of Noetherian algebraic spaces. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by a kernel E∈Dcohb(Y1×SY2)E \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2). Then the following are equivalent:

(1) f1f_1 is smooth

(2) f2f_2 is smooth

(3) E∈Perf⁡(Y1×SY2)E \in\operatorname{Perf}(Y_1 \times_S Y_2).

Proof. This follows from Propositions 5.7 and 5.10. □

In particular, if either family has a nonsmooth fiber, then kernels giving a Fourier–Mukai partnership cannot be perfect. Consequently, bounded pseudocoherent kernels, rather than perfect ones, are crucial in relative Fourier–Mukai theory.

It is not possible to replace smoothness with regularity in Theorem 1.1. Over a field, regularity can be detected by the equality Perf⁡(X)=Dcohb(X)\operatorname{Perf}(X)=D^b_{\mathrm{coh}}(X). However, over an imperfect field, a regular proper scheme need not be smooth. For such a scheme, no Fourier–Mukai autoequivalence has a perfect kernel. We give a concrete case.

Example 1.2. Consider any regular scheme YY proper over a field kk that is not smooth (i.e. not geometrically regular). For instance, let k:=F3(t)k:=\mathbb{F}_3(t) be the rational function field in one variable, and let YY be the projective plane curve over kk given by the equation

y2z+x3−tz3=0y^2z+x^3-tz^3=0

[7 Remark 16]. By Proposition 5.10, any autoequivalence of the form ΦE\Phi_E on Dcohb(Y)D^b_{\mathrm{coh}}(Y) with E∈Dcohb(Y×kY)E\in D^b_{\mathrm{coh}}(Y\times_kY) satisfies E∉Perf⁡(Y×kY)E\notin\operatorname{Perf}(Y\times_kY). For example, the pushforward of OY\mathcal{O}_Y along the diagonal, i.e. the kernel of the identity functor, is not perfect.

Next, we show that Gorensteinness is preserved under Fourier–Mukai partnerships.

Theorem 1.3. Let fi:Yi→Sf_i:Y_i\to S be proper flat morphisms of Noetherian algebraic spaces. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by a kernel E∈Dcohb(Y1×SY2)E\in D^b_{\mathrm{coh}}(Y_1\times_SY_2). Then f1f_1 is Gorenstein if, and only if, f2f_2 is Gorenstein.

For projective schemes over an algebraically closed field, [6 Theorem 4.4] shows that Gorensteinness passes to Fourier–Mukai partners under Cohen–Macaulay and equidimensionality constraints. Over a field, a proper variety is Gorenstein if, and only if, its category of perfect complexes admits a Serre functor. See [1 Lemma 6.6] for projective varieties and [13 Proposition 2.1] for proper varieties (i.e. integrality included).

Theorem 1.3 strengthens these statements for Gorensteinness to proper flat families of Noetherian algebraic spaces. We do not require the fif_i to be projective, nor do we impose Cohen–Macaulay, equidimensionality, or integrality hypotheses. Moreover, the result applies over any Noetherian base, including in mixed characteristic.

Methodology

For the proof of Theorem 1.1, we base change the partnership along morphisms from fields to SS. This uses [4 Theorem 10.13 and Corollary 6.18]. This reduces the question to regularity, which is characterized by the equality Perf⁡=Dcohb\operatorname{Perf}=D^b_{\mathrm{coh}}. See Lemma 5.6. For the perfectness of kernels, we use that perfectness is detected on fibers. See Lemma 5.9. Moreover, over a regular base, we show that a perfect kernel forces the partners to be regular. See Lemma 5.8.

Our proof of Theorem 1.3 does not pass to fibers, and it is independent of [6] (e.g. l.c.i. cycles are not used). Instead, we work over the base with relative Serre functors. This is motivated by [14 Definition 2.5]. Specifically, we prove that a proper flat morphism is Gorenstein if, and only if, its category of perfect complexes admits a relative Serre functor. See Lemma 4.4. This is a relative version of [1 Lemma 6.6].

The relationship with diagonal morphisms with smoothness is classical: for a finite type morphism of locally Noetherian schemes, smoothness is equivalent to flatness and perfectness of the diagonal [15 Tag 0fDP]. Lunts–Schnürer have given a dg-categorical viewpoint which connects geometric smoothness to homological smoothness of enhancements of Perf⁡\operatorname{Perf} [9]. Calabrese showed a reconstruction result for algebraic spaces with affine diagonals and proper flat morphisms with Gorenstein fibers with suitable ampleness constraints [3]. Our results complement these perspectives by addressing invariance of smoothness and Gorensteinness without projectivity- or ampleness-type constraints.

Acknowledgments. The authors thank Kabeer Manali Rahul and Nebojsa Pavic for discussions. Lank was supported by the ERC Advanced Grant 101095900-TriCatApp. Guisado Villalgordo was supported under the grant PRE2022-104796 by the Spanish Ministry of Science and Innovation.

AI disclosure. The authors used large language models (LLMs) for assistance with proofreading (e.g. grammar, notation, and exposition) and checking arguments (e.g. identifying possible gaps or errors). The main results and methods are due to the authors.

Preliminaries

Generation

We discuss a form of generation for triangulated categories. See [BV03] for details. Let T\mathcal{T} be a triangulated category with shift functor [1]:T→T[1] : \mathcal{T} \to\mathcal{T}. Consider a subcategory S⊆T\mathcal{S} \subseteq\mathcal{T}. A triangulated subcategory of T\mathcal{T} is called thick if it is closed under direct summands. Denote by ⟨S⟩\langle\mathcal{S} \rangle the smallest thick subcategory of T\mathcal{T} containing S\mathcal{S}. If S\mathcal{S} consists of a single object GG, we write ⟨G⟩:=⟨S⟩\langle G \rangle:= \langle\mathcal{S} \rangle. Set add⁡(S)\operatorname{add}(\mathcal{S}) to be the smallest strictly full subcategory of T\mathcal{T} containing S\mathcal{S} that is closed under shifts, finite coproducts, and direct summands. Inductively, let ⟨S⟩0\langle\mathcal{S} \rangle_0 consist of all zero objects in T\mathcal{T}, ⟨S⟩1:=add⁡(S)\langle\mathcal{S} \rangle_1 := \operatorname{add}(\mathcal{S}), and

⟨S⟩n:=add⁡{cone⁡(ϕ):ϕ∈Hom⁡T(⟨S⟩n−1,⟨S⟩1)}.\langle\mathcal{S} \rangle_n := \operatorname{add}\{\operatorname{cone}(\phi) : \phi\in\operatorname{Hom}_{\mathcal{T}}(\langle\mathcal{S} \rangle_{n-1},\langle\mathcal{S} \rangle_1)\}.

It can be checked that ⟨S⟩=⋃n=0∞⟨S⟩n\langle\mathcal{S} \rangle= \bigcup_{n=0}^{\infty} \langle\mathcal{S} \rangle_n. We say E∈TE \in\mathcal{T} is finitely built by S\mathcal{S} if E∈⟨S⟩E \in\langle\mathcal{S} \rangle.

Example 2.1. Let X=Spec⁡(R)X = \operatorname{Spec}(R) where RR is a regular ring. Then ⟨OX⟩=Dcohb(X)\langle\mathcal{O}_X \rangle= D^b_{\mathrm{coh}}(X). See [15 Tag 066Z] and [11 Lemma 1.2].

Algebraic spaces

We follow [15] for conventions. We call a morphism natural if it is functorial, and canonical if it is given by an explicit construction that need not be functorial. Let YY be a quasi-compact algebraic space. An étale presentation of YY is a finitely presented, étale, surjective morphism U→YU \to Y from a scheme.

Notation

We work with the small étale ringed site (Yet,OY)(Y_{\mathrm{et}},\mathcal{O}_Y). Write D(Y)D(Y) for the unbounded derived category of sheaves of OY\mathcal{O}_Y-modules. Define Dqc(Y)D_{\mathrm{qc}}(Y) to be the strictly full subcategory of D(Y)D(Y) consisting of complexes with quasi-coherent cohomology. For ?∈{b,+,−}\mathord{?} \in\{b,+,-\}, let Dqc?(Y)D^?_{\mathrm{qc}}(Y) be the strictly full subcategory of Dqc(Y)D_{\mathrm{qc}}(Y) with prescribed cohomological vanishing. Denote by Perf⁡(Y)\operatorname{Perf}(Y) the strictly full subcategory of Dqc(Y)D_{\mathrm{qc}}(Y) consisting of perfect complexes. If YY is Noetherian, Dcohb(Y)D^b_{\mathrm{coh}}(Y) is the strictly full subcategory of Dqc(Y)D_{\mathrm{qc}}(Y) consisting of bounded pseudocoherent complexes. Recall that the derived internal Hom RHomOY\mathcal{R}\mathcal{H}om_{\mathcal{O}_Y} is formed in the ambient category D(Y)D(Y) [15 Tag 08J7]. In particular, it need not agree with the right adjoint to tensor product in Dqc(Y)D_{\mathrm{qc}}(Y), which is denoted by RHom\mathbf{R}\mathcal{H}om. See [4 Lemma 2.11] and [15 Tag 0A8A].

Duality

We use the duality conventions of [4 §2.4] and [12]. Let f:Y→Xf : Y \to X be a proper morphism of Noetherian algebraic spaces. As ff is proper, f!f^! is naturally isomorphic to the right adjoint f×f^\times of Rf∗\mathbf{R}f_* on quasi-coherent derived categories. See [4 Lemma 2.22]. Recall that f!f^! is the upper shriek functor. It is constructed from Nagata compactifications, and defined on DqcD_{\mathrm{qc}}. See [4 §2.4] for details. In particular, loc. cit. discusses the results of [12 Theorem 1.8] for algebraic spaces, where Neeman’s result constructs the so-called upper shriek functor (e.g. f!f^{!}) on DqcD_{\mathrm{qc}} and not just on Dqc+D_{\mathrm{qc}}^{+}.

Integral transforms

We refer to [4]. Let SS be a Noetherian algebraic space. Suppose Y1Y_{1} and Y2Y_{2} are proper and flat algebraic SS-spaces. Denote by pi:Y1×SY2→Yip_{i}:Y_{1}\times_{S}Y_{2}\to Y_{i} the canonical projections. For E∈Dqc(Y1×SY2)E\in D_{\mathrm{qc}}(Y_{1}\times_{S}Y_{2}), the integral transform with kernel EE is the functor

ΦE:=R(p2)∗(Lp1∗(−)⊗LE):Dqc(Y1)→Dqc(Y2).\Phi_{E}:=\mathop{\mathrm{R}}(p_{2})_{*}\left(\mathop{\mathrm{L}}p_{1}^{*}(-)\otimes^{\mathrm{L}}E\right):D_{\mathrm{qc}}(Y_{1})\to D_{\mathrm{qc}}(Y_{2}).

We say that E∈Dcohb(Y1×SY2)E\in D_{\mathrm{coh}}^{b}(Y_{1}\times_{S}Y_{2}) induces a Fourier–Mukai equivalence (over SS) if ΦE\Phi_{E} restricts to an equivalence Dcohb(Y1)→Dcohb(Y2)D_{\mathrm{coh}}^{b}(Y_{1})\to D_{\mathrm{coh}}^{b}(Y_{2}). In this case, Y1Y_{1} and Y2Y_{2} are called Fourier–Mukai SS-partners, and we say it is given by EE. If clear from context, we omit the ‘SS-’ and say ‘Fourier–Mukai partners’. We say that A∈Dqc(Y1×SY2)A\in D_{\mathrm{qc}}(Y_{1}\times_{S}Y_{2}) is relatively perfect over YiY_{i} if A⊗LLpi∗Dqc(Yi)⊆Dqc(Y1×SY2)A\otimes^{\mathrm{L}}\mathop{\mathrm{L}}p_{i}^{*}D_{\mathrm{qc}}(Y_{i})\subseteq D_{\mathrm{qc}}(Y_{1}\times_{S}Y_{2}). By [4 Corollary 1.4], a pseudocoherent complex EE on Y1×SY2Y_{1}\times_{S}Y_{2} is relatively perfect over Y1Y_{1} (resp. Y2Y_{2}) if, and only if, ΦE(Dcohb(Y1))⊆Dcohb(Y2)\Phi_{E}(D_{\mathrm{coh}}^{b}(Y_{1}))\subseteq D_{\mathrm{coh}}^{b}(Y_{2}) (resp. ΦE(Perf⁡(Y1))⊆Perf⁡(Y2)\Phi_{E}(\operatorname{Perf}(Y_{1}))\subseteq\operatorname{Perf}(Y_{2})). In both cases, any such EE is bounded.

Auxiliary lemmas

Lemma 3.1. Let XX be a quasi-compact quasi-separated algebraic space. Let E∈D(X)E\in D(X). Then EE is invertible if, and only if, there exists an étale presentation s:U→Xs:U\to X such that Ls∗E\mathop{\mathrm{L}}s^{*}E is invertible.

Proof. If EE is invertible, then so is Ls∗E\mathop{\mathrm{L}}s^{*}E for any étale presentation ss. Indeed, the derived pullback is monoidal [15 Tag 07A4]. Conversely, suppose s:U→Xs:U\to X is an étale presentation such that Ls∗E\mathop{\mathrm{L}}s^{*}E is invertible. Fix any étale morphism t:V→Xt:V\to X from a scheme. Consider the fibered square

V×XU→t′Us′↓↓sV→tX\begin{CD} V\times_{X}U @>{t'}>> U \\ @V{s'}VV @VV{s}V \\ V @>{t}>> X \end{CD}

Suppose h:W→V×XUh:W\to V\times_{X}U is an étale surjective morphism from a scheme. Note that s′∘hs'\circ h is an étale covering of VV in XetX_{\mathrm{et}}. As the derived pullback is monoidal, it follows that L(s∘t′∘h)∗E\mathop{\mathrm{L}}(s\circ t'\circ h)^{*}E is invertible. Recall that an algebraic space endowed with the structure on its small étale site is a locally ringed site [15 Tag 04KH]. Furthermore, the derived pullback along an étale morphism from a scheme to XX can be identified with restriction on the small étale site [15 Tag 04M4]. By [15 Tag 0FPY], there exists an étale covering {gi:Wi→W}\{g_{i}:W_{i}\to W\} by schemes such that each E∣WiE|_{W_{i}} can be represented by a shift of an invertible OWi\mathcal{O}_{W_{i}}-module. Consequently, we have a covering {s′∘h∘gi:Wi→V}\{s'\circ h\circ g_{i}:W_{i}\to V\} for which [15 Tag 0FPY] holds. Since VV was arbitrary, this completes the proof. □\square

Lemma 3.2. Let f:Y→Xf:Y\to X be a separated étale morphism of Noetherian algebraic spaces. Then f!f^{!} is naturally isomorphic to Lf∗\mathop{\mathrm{L}}f^{*} on Dqc(X)D_{\mathrm{qc}}(X).

Proof. As ff is flat, [4 Lemma 2.24] gives a natural isomorphism Lf∗E⊗Lf!′OX≅f!′E\mathbf{L}f^{*}E\otimes^{\mathbf{L}}f'_{!}\mathcal{O}_{X}\cong f'_{!}E for all E∈Dqc(X)E\in D_{\mathrm{qc}}(X). Hence, it suffices to prove that f!′OX≅OYf'_{!}\mathcal{O}_{X}\cong\mathcal{O}_{Y}. Consider the fibered square

Y×XY→r2Yr1↓↓fY→fX\begin{CD} Y\times_{X}Y @>{r_{2}}>> Y\\ @V{r_{1}}VV @VV{f}V\\ Y @>{f}>> X \end{CD}

Denote by Δf:Y→Y×XY\Delta_f:Y\to Y\times_XY the diagonal. Note that r1∘Δfr_1\circ\Delta_f and r2∘Δfr_2\circ\Delta_f are the identity of YY. By [15 Tag 05W1], Δf\Delta_f is an open immersion. It also is a closed immersion because ff is separated. The open immersion Δf\Delta_f admits the Nagata compactification given by

Y→ΔfY×XY→1Y×XYY×XY,Y\xrightarrow{\Delta_f}Y\times_XY\xrightarrow{1_{Y\times_XY}}Y\times_XY,

it follows that LΔf∗≅Δf!\mathbf{L}\Delta_f^{*}\cong\Delta_f^{!}. There exists a string of isomorphisms

f!′OX≅LΔf∗Lr2!f!′OX(r2∘Δf=1Y)≅LΔf∗r1!OY(citep‘[GL26, Lemma 2.23‘)≅Δf!r1!OY≅(r1∘Δf)!OY(citep‘[GL26, Lemma 2.21‘)≅OY(r1∘Δf=1Y).\begin{aligned} f'_{!}\mathcal{O}_{X}&\cong\mathbf{L}\Delta_f^{*}\mathbf{L}r_{2!}f'_{!}\mathcal{O}_{X} &&(r_2\circ\Delta_f=1_Y)\\ &\cong\mathbf{L}\Delta_f^{*}r_{1!}\mathcal{O}_{Y} &&(\text{{citep}`[GL26, Lemma 2.23`})\\ &\cong\Delta_f^{!}r_{1!}\mathcal{O}_{Y}\\ &\cong(r_1\circ\Delta_f)^{!}\mathcal{O}_{Y} &&(\text{{citep}`[GL26, Lemma 2.21`})\\ &\cong\mathcal{O}_{Y} &&(r_1\circ\Delta_f=1_Y). \end{aligned}

This completes the proof. □

Lemma 3.3. Let f:Y→Xf:Y\to X be a separated morphism of finite type between Noetherian algebraic spaces. If KK is a dualizing complex on XX, then f!Kf^{!}K is a dualizing complex on YY.

Proof. Choose étale presentations t:V→Yt:V\to Y and s:U→Xs:U\to X from affine schemes. There exists a commutative diagram

V×XU→t′Y×XU→f′Us′′↓s′↓↓sV→tY→fX\begin{CD} V\times_XU @>{t'}>> Y\times_XU @>{f'}>> U\\ @V{s''}VV @V{s'}VV @VV{s}V\\ V @>{t}>> Y @>{f}>> X \end{CD}

obtained by base change. By [4 Lemma 2.19], tt, ss, and f∘tf\circ t are quasi-affine. Base change implies s′s', s′′s'', and t′t' are quasi-affine, finitely presented, and étale. In particular, V×XUV\times_XU is a Noetherian scheme. There is a direct computation,

L(s′′)∗Lt∗f!K≅L(s′′)∗t!′f′!K(Lemma 3.2)≅L(s′′)∗(f∘t)!K(citep‘[GL26, Lemma 2.21]‘)≅(f′∘t′)!Ls∗K(citep‘[GL26, Lemma 2.23]‘).\begin{aligned} \mathbf{L}(s'')^{*}\mathbf{L}t^{*}f^{!}K&\cong\mathbf{L}(s'')^{*}t'_{!}f'^{!}K &&(\text{Lemma 3.2})\\ &\cong\mathbf{L}(s'')^{*}(f\circ t)^{!}K &&(\text{{citep}`[GL26, Lemma 2.21]`})\\ &\cong(f'\circ t')^{!}\mathbf{L}s^{*}K &&(\text{{citep}`[GL26, Lemma 2.23]`}). \end{aligned}

Note that Ls∗K\mathbf{L}s^{*}K is a dualizing complex. As f′∘t′f'\circ t' is a separated finite type morphism of Noetherian schemes, [15 Tag 0AA3] says that (f′∘t′)!Ls∗K(f'\circ t')^{!}\mathbf{L}s^{*}K is a dualizing complex. Since t∘s′′t\circ s'' is an étale presentation of YY, f!Kf^{!}K is a dualizing complex [15 Tag 0E4Y]. □

Gorensteinness

Definition 4.1. Let f:Y→Xf:Y\to X be a proper flat morphism of Noetherian algebraic spaces. An ff-Serre functor is an autoequivalence Sf:Perf⁡(Y)→Perf⁡(Y)\mathcal{S}_f:\operatorname{Perf}(Y)\to\operatorname{Perf}(Y) equipped with a bifunctorial isomorphism

RHom⁡(R⁡f∗RHom⁡(P,Q),OX)≅R⁡f∗RHom⁡(Q,Sf(P))\operatorname{RHom}(\operatorname{R}f_*\operatorname{RHom}(P,Q),\mathcal{O}_X)\cong\operatorname{R}f_*\operatorname{RHom}(Q,\mathcal{S}_f(P))

for all P,Q∈Perf⁡(Y)P,Q\in\operatorname{Perf}(Y).

Remark 4.2. The definition above is an analog of [14 Definition 2.5].

Lemma 4.3. Let f:Y→Xf:Y\to X be a proper flat morphism of Noetherian algebraic spaces. Then ff is Gorenstein if, and only if, f!OXf^!\mathcal{O}_X is invertible.

Proof. Choose an étale presentation s:U→Xs:U\to X from an affine scheme. Note that UU is Noetherian. Consider the fibered square

Y×XU→f′Us′↓↓sY→fX\begin{CD} Y\times_X U @>{f'}>> U\\ @V{s'}VV @VV{s}V\\ Y @>{f}>> X \end{CD}

Choose an étale presentation w:V→Y×XUw:V\to Y\times_X U from an affine scheme. Note that VV is Noetherian, and that ww is separated by [4 Lemma 2.19]. The composition f′∘w:V→Uf'\circ w:V\to U is a flat separated finite type morphism of Noetherian schemes. There is a string of isomorphisms,

(f′∘w)!OU≅w!(f′)!OU(f'\circ w)^!\mathcal{O}_U\cong w^!(f')^!\mathcal{O}_U

([4 Lemma 2.21])

≅w!L(s′)∗f!OX\cong w^!\mathop{\mathrm{L}}(s')^*f^!\mathcal{O}_X

([4 Lemma 2.23])

≅Lw∗L(s′)∗f!OX\cong\mathop{\mathrm{L}}w^*\mathop{\mathrm{L}}(s')^*f^!\mathcal{O}_X

(Lemma 3.2)

≅L(s′∘w)∗f!OX.\cong\mathop{\mathrm{L}}(s'\circ w)^*f^!\mathcal{O}_X.

Since s′∘ws'\circ w is an étale presentation, Lemma 3.1 implies that f!OXf^!\mathcal{O}_X is invertible if, and only if, (f′∘w)!OU(f'\circ w)^!\mathcal{O}_U is invertible. By [15 Tag 0C08], f′∘wf'\circ w is Gorenstein if, and only if, (f′∘w)!OU(f'\circ w)^!\mathcal{O}_U is isomorphic Zariski locally on VV to an invertible object. By [15 Tag 0FPY], the latter holds if, and only if, (f′∘w)!OU(f'\circ w)^!\mathcal{O}_U is invertible. Now, ss and s′∘ws'\circ w are étale, and s′∘ws'\circ w is surjective. Hence, by [15 Tags 0E14 and 04NC], ff is Gorenstein if, and only if, f′∘wf'\circ w is Gorenstein. This completes the proof. □\square

Lemma 4.4. Let f:Y→Xf:Y\to X be a proper flat morphism of Noetherian algebraic spaces. Then Perf⁡(Y)\operatorname{Perf}(Y) admits an ff-Serre functor if, and only if, ff is Gorenstein.

Proof. There is a string of bifunctorial isomorphisms for all P,Q∈Perf⁡(Y)P,Q \in\operatorname{Perf}(Y):

RHom⁡(R⁡f∗RHom⁡(P,Q),OX)≅R⁡f∗RHom⁡(RHom⁡(P,Q),f×OX)(citep‘[GL26,Lemma5.9]‘)≅R⁡f∗RHom⁡(RHom⁡(P,Q),f!OX)(citep‘[GL26,Lemma2.22]‘)≅R⁡f∗RHom⁡(RHom⁡(P,OY)⊗LQ,f!OX)(citep‘[Sta26,Tag08JJand0D0S]‘)≅R⁡f∗RHom⁡(Q,RHom⁡(RHom⁡(P,OY),f!OX))(citep‘[Sta26,Tag08J9and0D0S]‘)≅R⁡f∗RHom⁡(Q,RHom⁡(RHom⁡(P,OY),OY)⊗Lf!OX)(citep‘[Sta26,Tag08JJand0D0S]‘)≅R⁡f∗RHom⁡(Q,P⊗Lf!OX).\begin{aligned} \operatorname{RHom}(\operatorname{R}f_* \operatorname{RHom}(P,Q),\mathcal{O}_X) &\cong\operatorname{R}f_* \operatorname{RHom}(\operatorname{RHom}(P,Q),f^{\times}\mathcal{O}_X) && ({citep}`[GL26, Lemma 5.9]`)\\ &\cong\operatorname{R}f_* \operatorname{RHom}(\operatorname{RHom}(P,Q),f^!\mathcal{O}_X) && ({citep}`[GL26, Lemma 2.22]`)\\ &\cong\operatorname{R}f_* \operatorname{RHom}(\operatorname{RHom}(P,\mathcal{O}_Y)\otimes^{\mathrm{L}} Q,f^!\mathcal{O}_X) && ({citep}`[Sta26, Tag 08JJ and 0D0S]`)\\ &\cong\operatorname{R}f_* \operatorname{RHom}(Q,\operatorname{RHom}(\operatorname{RHom}(P,\mathcal{O}_Y),f^!\mathcal{O}_X)) && ({citep}`[Sta26, Tag 08J9 and 0D0S]`)\\ &\cong\operatorname{R}f_* \operatorname{RHom}(Q,\operatorname{RHom}(\operatorname{RHom}(P,\mathcal{O}_Y),\mathcal{O}_Y)\otimes^{\mathrm{L}}f^!\mathcal{O}_X) && ({citep}`[Sta26, Tag 08JJ and 0D0S]`)\\ &\cong\operatorname{R}f_*\operatorname{RHom}(Q,P\otimes^{\mathrm{L}}f^!\mathcal{O}_X). \end{aligned}

If ff is Gorenstein, then Lemma 4.3 implies f!OXf^!\mathcal{O}_X is invertible. In such a case, (−)⊗Lf!OX:Perf⁡(Y)→Perf⁡(Y)(-)\otimes^{\mathrm{L}}f^!\mathcal{O}_X:\operatorname{Perf}(Y)\to\operatorname{Perf}(Y) is an autoequivalence, and the bifunctorial isomorphisms above show that this defines an ff-Serre functor.

Conversely, if Sf\mathcal{S}_f is an ff-Serre functor on Perf⁡(Y)\operatorname{Perf}(Y), then for all P,Q∈Perf⁡(Y)P,Q\in\operatorname{Perf}(Y) there is a string of natural isomorphisms:

Hom⁡(Q,Sf(P))≅Hom⁡(OY,RHom⁡(Q,Sf(P)))(citep‘[Sta26,Tag08J7and0D0S]‘)≅Hom⁡(Lf∗OX,RHom⁡(Q,Sf(P)))≅Hom⁡(OX,R⁡f∗RHom⁡(Q,Sf(P)))(pull/push)≅Hom⁡(OX,RHom⁡(R⁡f∗RHom⁡(P,Q),OX))(Serre functor)≅Hom⁡(R⁡f∗RHom⁡(P,Q),OX)≅Hom⁡(RHom⁡(P,Q),f×OX)(right adjoint of push)≅Hom⁡(RHom⁡(P,Q),f!OX)(citep‘[GL26,Lemma2.22]‘)≅Hom⁡(Q,P⊗Lf!OX).\begin{aligned} \operatorname{Hom}(Q,\mathcal{S}_f(P)) &\cong\operatorname{Hom}(\mathcal{O}_Y,\operatorname{RHom}(Q,\mathcal{S}_f(P))) && ({citep}`[Sta26, Tag 08J7 and 0D0S]`)\\ &\cong\operatorname{Hom}(\mathrm{L}f^*\mathcal{O}_X,\operatorname{RHom}(Q,\mathcal{S}_f(P)))\\ &\cong\operatorname{Hom}(\mathcal{O}_X,\operatorname{R}f_*\operatorname{RHom}(Q,\mathcal{S}_f(P))) && \text{(pull/push)}\\ &\cong\operatorname{Hom}(\mathcal{O}_X,\operatorname{RHom}(\operatorname{R}f_*\operatorname{RHom}(P,Q),\mathcal{O}_X)) && \text{(Serre functor)}\\ &\cong\operatorname{Hom}(\operatorname{R}f_*\operatorname{RHom}(P,Q),\mathcal{O}_X)\\ &\cong\operatorname{Hom}(\operatorname{RHom}(P,Q),f^{\times}\mathcal{O}_X) && \text{(right adjoint of push)}\\ &\cong\operatorname{Hom}(\operatorname{RHom}(P,Q),f^!\mathcal{O}_X) && ({citep}`[GL26, Lemma 2.22]`)\\ &\cong\operatorname{Hom}(Q,P\otimes^{\mathrm{L}}f^!\mathcal{O}_X). \end{aligned}

Fix P∈Perf⁡(Y)P\in\operatorname{Perf}(Y). As Sf(P)\mathcal{S}_f(P) is perfect, [15 Tag 001P], applied to these isomorphisms as functors of Q∈Perf⁡(Y)Q\in\operatorname{Perf}(Y), yields

ϕP:Sf(P)⟶P⊗Lf!OX\phi_P:\mathcal{S}_f(P)\longrightarrow P\otimes^{\mathrm{L}}f^!\mathcal{O}_X

such that Hom⁡(Q,ϕP)\operatorname{Hom}(Q,\phi_P) is an isomorphism for all Q∈Perf⁡(Y)Q\in\operatorname{Perf}(Y). Pick a compact generator GG for Dqc(Y)D_{\mathrm{qc}}(Y), see [15 Tag 09IY]. Apply Hom⁡(G,−)\operatorname{Hom}(G,-) to the distinguished triangle

Sf(P)→ϕPP⊗Lf!OX⟶cone⁡(ϕP)⟶Sf(P)[1].\mathcal{S}_f(P)\xrightarrow{\phi_P}P\otimes^{\mathrm{L}}f^!\mathcal{O}_X\longrightarrow\operatorname{cone}(\phi_P)\longrightarrow\mathcal{S}_f(P)[1].

There exists a long exact sequence (see [15 Tag 0149]),

⋯⟶Hom⁡(G,Sf(P))→Hom⁡(G,ϕP)Hom⁡(G,P⊗Lf!OX)⟶Hom⁡(G,cone⁡(ϕP))⟶⋯ .\cdots\longrightarrow\operatorname{Hom}(G,\mathcal{S}_f(P))\xrightarrow{\operatorname{Hom}(G,\phi_P)}\operatorname{Hom}(G,P\otimes^{\mathrm{L}}f^!\mathcal{O}_X)\longrightarrow\operatorname{Hom}(G,\operatorname{cone}(\phi_P))\longrightarrow\cdots.

As Hom⁡(G[n],ϕP)\operatorname{Hom}(G[n],\phi_P) is an isomorphism for all n∈Zn\in\mathbb{Z}, it follows that Hom⁡(G,cone⁡(ϕP)[n])≅0\operatorname{Hom}(G,\operatorname{cone}(\phi_P)[n])\cong0 for all n∈Zn\in\mathbb{Z}. Since GG compactly generates Dqc(Y)D_{\mathrm{qc}}(Y), we obtain that cone⁡(ϕP)≅0\operatorname{cone}(\phi_P)\cong0. In other words, ϕP\phi_P is an isomorphism. Taking P:=OYP:=\mathcal{O}_Y implies f!OXf^!\mathcal{O}_X is perfect. Since Sf\mathcal{S}_f is essentially surjective, there exists P0∈Perf⁡(Y)P_0\in\operatorname{Perf}(Y) such that Sf(P0)≅OY\mathcal{S}_f(P_0)\cong\mathcal{O}_Y. Then

OY≅Sf(P0)≅P0⊗Lf!OX.\mathcal{O}_Y\cong\mathcal{S}_f(P_0)\cong P_0\otimes^{\mathrm{L}}f^!\mathcal{O}_X.

Since P0P_0 and f!OXf^!\mathcal{O}_X are perfect, the preceding isomorphism shows that f!OXf^!\mathcal{O}_X is invertible, with inverse P0P_0. Therefore, by Lemma 4.3, ff is Gorenstein.

Lemma 4.5. Let fi:Yi→Sf_i:Y_i \to S be proper flat morphisms of Noetherian algebraic spaces. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by B∈Dcohb(Y1×SY2)B \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2). There are natural isomorphisms for all P,Q∈Perf⁡(Y1)P,Q \in\operatorname{Perf}(Y_1):

R(f1)∗RHom(P,Q)≅R(f2)∗RHom(ΦB(P),ΦB(Q))(1)\mathrm{R}(f_1)_*\mathrm{R}\mathcal{H}om(P,Q) \cong\mathrm{R}(f_2)_*\mathrm{R}\mathcal{H}om(\Phi_B(P),\Phi_B(Q)) \tag*{(1)}
RHom(R(f1)∗RHom(P,Q),OS)≅RHom(R(f2)∗RHom(ΦB(P),ΦB(Q)),OS).(2)\mathrm{R}\mathcal{H}om(\mathrm{R}(f_1)_*\mathrm{R}\mathcal{H}om(P,Q),\mathcal{O}_S) \cong\mathrm{R}\mathcal{H}om(\mathrm{R}(f_2)_*\mathrm{R}\mathcal{H}om(\Phi_B(P),\Phi_B(Q)),\mathcal{O}_S). \tag*{(2)}

Proof. We prove the first claim. By [4 Remark 2.7 and Lemma 10.9], ΦB\Phi_B induces an equivalence on DqcD_{\mathrm{qc}}. Denote by pi:Y1×SY2→Yip_i:Y_1 \times_S Y_2 \to Y_i the projections. For P,Q∈Perf⁡(Y1)P,Q \in\operatorname{Perf}(Y_1) and E∈Dqc(S)E \in D_{\mathrm{qc}}(S), there exist a string of natural isomorphisms:

Hom⁡(E,R(f1)∗RHom(P,Q))≅Hom⁡(Lf1∗E,RHom(P,Q))(pull/push)≅Hom⁡(Lf1∗E⊗LP,Q)([Sta26, Tag 08J7 and 0D0S])≅Hom⁡(ΦB(Lf1∗E⊗LP),ΦB(Q))(ΦB is an equivalence)≅Hom⁡(R(p2)∗(Lp1∗(Lf1∗E⊗LP)⊗LB),ΦB(Q))(def. of ΦB)≅Hom⁡(R(p2)∗(Lp1∗Lf1∗E⊗LLp1∗P⊗LB),ΦB(Q))([Sta26, Tag 07A4])≅Hom⁡(R(p2)∗(Lp2∗Lf2∗E⊗LLp1∗P⊗LB),ΦB(Q))(f1∘p1=f2∘p2)≅Hom⁡(R(p2)∗(Lp1∗P⊗LB)⊗LLf2∗E,ΦB(Q))(projection formula)≅Hom⁡(ΦB(P)⊗LLf2∗E,ΦB(Q))(def. of ΦB)≅Hom⁡(Lf2∗E,RHom(ΦB(P),ΦB(Q)))([Sta26, Tag 08J7 and 0D0S])≅Hom⁡(E,R(f2)∗RHom(ΦB(P),ΦB(Q)))(pull/push).\begin{aligned} \operatorname{Hom}(E,\mathrm{R}(f_1)_*\mathrm{R}\mathcal{H}om(P,Q)) &\cong\operatorname{Hom}(\mathrm{L}f_1^*E,\mathrm{R}\mathcal{H}om(P,Q)) &&\text{(pull/push)}\\ &\cong\operatorname{Hom}(\mathrm{L}f_1^*E\otimes^{\mathrm{L}}P,Q) &&\text{([Sta26, Tag 08J7 and 0D0S])}\\ &\cong\operatorname{Hom}(\Phi_B(\mathrm{L}f_1^*E\otimes^{\mathrm{L}}P),\Phi_B(Q)) &&\text{(}\Phi_B\text{ is an equivalence)}\\ &\cong\operatorname{Hom}(\mathrm{R}(p_2)_*(\mathrm{L}p_1^*(\mathrm{L}f_1^*E\otimes^{\mathrm{L}}P)\otimes^{\mathrm{L}}B),\Phi_B(Q)) &&\text{(def. of }\Phi_B\text{)}\\ &\cong\operatorname{Hom}(\mathrm{R}(p_2)_*(\mathrm{L}p_1^*\mathrm{L}f_1^*E\otimes^{\mathrm{L}}\mathrm{L}p_1^*P\otimes^{\mathrm{L}}B),\Phi_B(Q)) &&\text{([Sta26, Tag 07A4])}\\ &\cong\operatorname{Hom}(\mathrm{R}(p_2)_*(\mathrm{L}p_2^*\mathrm{L}f_2^*E\otimes^{\mathrm{L}}\mathrm{L}p_1^*P\otimes^{\mathrm{L}}B),\Phi_B(Q)) &&\text{(}f_1\circ p_1=f_2\circ p_2\text{)}\\ &\cong\operatorname{Hom}(\mathrm{R}(p_2)_*(\mathrm{L}p_1^*P\otimes^{\mathrm{L}}B)\otimes^{\mathrm{L}}\mathrm{L}f_2^*E,\Phi_B(Q)) &&\text{(projection formula)}\\ &\cong\operatorname{Hom}(\Phi_B(P)\otimes^{\mathrm{L}}\mathrm{L}f_2^*E,\Phi_B(Q)) &&\text{(def. of }\Phi_B\text{)}\\ &\cong\operatorname{Hom}(\mathrm{L}f_2^*E,\mathrm{R}\mathcal{H}om(\Phi_B(P),\Phi_B(Q))) &&\text{([Sta26, Tag 08J7 and 0D0S])}\\ &\cong\operatorname{Hom}(E,\mathrm{R}(f_2)_*\mathrm{R}\mathcal{H}om(\Phi_B(P),\Phi_B(Q))) &&\text{(pull/push).} \end{aligned}

These isomorphisms are natural in E∈Dqc(S)E \in D_{\mathrm{qc}}(S) and in P,Q∈Perf⁡(Y1)P,Q \in\operatorname{Perf}(Y_1). Hence, [15 Tag 001P] gives an isomorphism

R(f1)∗RHom(P,Q)⟶R(f2)∗RHom(ΦB(P),ΦB(Q))\mathrm{R}(f_1)_*\mathrm{R}\mathcal{H}om(P,Q)\longrightarrow\mathrm{R}(f_2)_*\mathrm{R}\mathcal{H}om(\Phi_B(P),\Phi_B(Q))

which is functorial in PP and QQ. This proves the first claim. Applying RHom(−,OS)\mathrm{R}\mathcal{H}om(-,\mathcal{O}_S) gives the second claim, and completes the proof. □

Lemma 4.6. Let fi:Yi→Sf_i:Y_i \to S be proper flat morphisms of Noetherian algebraic spaces. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by B∈Dcohb(Y1×SY2)B \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2). Denote by pi:Y1×SY2→Yip_i:Y_1 \times_S Y_2 \to Y_i the projections. If S1\mathbb{S}_1 is an f1f_1-Serre functor, then there are natural isomorphisms for all P,Q∈Perf⁡(Y1)P,Q \in\operatorname{Perf}(Y_1):

R(f2)∗RHom(ΦB(Q),S2(ΦB(P)))≅R(f1)∗RHom(Q,S1(P))\mathrm{R}(f_2)_*\mathrm{R}\mathcal{H}om(\Phi_B(Q),\mathbb{S}_2(\Phi_B(P)))\cong\mathrm{R}(f_1)_*\mathrm{R}\mathcal{H}om(Q,\mathbb{S}_1(P))

where S2:=ΦB∘S1∘ΦRHom(B,p2∗OY2)\mathbb{S}_2:=\Phi_B\circ\mathbb{S}_1\circ\Phi_{\mathrm{R}\mathcal{H}om(B,p_2^*\mathcal{O}_{Y_2})}. Here, ΦRHom(B,p2∗OY2)\Phi_{\mathrm{R}\mathcal{H}om(B,p_2^*\mathcal{O}_{Y_2})} denotes the integral transform from Dqc(Y2)D_{\mathrm{qc}}(Y_2) to Dqc(Y1)D_{\mathrm{qc}}(Y_1), i.e. the roles of p1p_1 and p2p_2 are interchanged.

Proof. Fix P∈Perf⁡(Y1)P \in\operatorname{Perf}(Y_1). By [4 Remark 2.7 and Lemma 10.9], ΦB\Phi_B is an equivalence on DqcD_{\mathrm{qc}} which restricts to an equivalence on Perf⁡\operatorname{Perf}. By [4 Proposition 9.3], applied with tt the identity, ΦRHom(B,p2∗OY2)\Phi_{\mathrm{R}\mathcal{H}om(B,p_2^*\mathcal{O}_{Y_2})} is right adjoint to ΦB\Phi_B on DqcD_{\mathrm{qc}}. Hence, it is a quasi-inverse of ΦB\Phi_B. We claim that S2(ΦB(P))≅ΦB(S1(P))\mathbb{S}_2(\Phi_B(P))\cong\Phi_B(\mathbb{S}_1(P)). Indeed,

S2(ΦB(P))≅(ΦB∘S1∘ΦRHom(B,p2∗OY2)∘ΦB)(P)(def. of S2)\mathbb{S}_2(\Phi_B(P))\cong(\Phi_B\circ\mathbb{S}_1\circ\Phi_{\mathrm{R}\mathcal{H}om(B,p_2^*\mathcal{O}_{Y_2})}\circ\Phi_B)(P)\quad\text{(def. of }\mathbb{S}_2\text{)}
≅(ΦB∘S1)(P)(quasi-inverse).\cong(\Phi_B\circ\mathbb{S}_1)(P)\quad\text{(quasi-inverse).}

Then Lemma 4.5 applied to QQ and S1(P)\mathcal{S}_1(P) yields

R(f1)∗RHom(Q,S1(P))≅R(f2)∗RHom(ΦB(Q),S2(ΦB(P))).\mathbf{R}(f_1)_*\mathbf{R}\mathcal{H}om(Q,\mathcal{S}_1(P)) \cong\mathbf{R}(f_2)_*\mathbf{R}\mathcal{H}om(\Phi_B(Q),\mathcal{S}_2(\Phi_B(P))).

Thus, the desired claim follows.

Proof of Theorem 1.3. By symmetry, it suffices to show that if f1f_1 is Gorenstein, then so is f2f_2. By [4 Remark 2.7], EE is relatively perfect over each YiY_i. As f1f_1 is Gorenstein, Lemma 4.4 gives an f1f_1-Serre functor S1\mathcal{S}_1. Define S2:=ΦE∘S1∘ΦRHom(E,p2∗OY2)\mathcal{S}_2 := \Phi_E \circ\mathcal{S}_1 \circ\Phi_{\mathbf{R}\mathcal{H}om(E,p_2^*\mathcal{O}_{Y_2})} as in Lemma 4.6. By [4 Lemma 10.9], ΦE\Phi_E restricts to an equivalence Perf⁡(Y1)→Perf⁡(Y2)\operatorname{Perf}(Y_1) \to\operatorname{Perf}(Y_2), and so S2\mathcal{S}_2 is an autoequivalence of Perf⁡(Y2)\operatorname{Perf}(Y_2). For all P,Q∈Perf⁡(Y1)P,Q \in\operatorname{Perf}(Y_1), there is a string of natural isomorphisms

RHom(R(f2)∗RHom(ΦE(P),ΦE(Q)),OS)≅RHom(R(f1)∗RHom(P,Q),OS)(xref‘Lemma 4.5‘)≅R(f1)∗RHom(Q,S1(P))(S1 is an f1-Serre functor)≅R(f2)∗RHom(ΦE(Q),S2(ΦE(P)))(xref‘Lemma 4.6‘).\begin{aligned} \mathbf{R}\mathcal{H}om(\mathbf{R}(f_2)_*\mathbf{R}\mathcal{H}om(\Phi_E(P),\Phi_E(Q)),\mathcal{O}_S) &\cong\mathbf{R}\mathcal{H}om(\mathbf{R}(f_1)_*\mathbf{R}\mathcal{H}om(P,Q),\mathcal{O}_S) &&(\text{{xref}`Lemma 4.5`})\\ &\cong\mathbf{R}(f_1)_*\mathbf{R}\mathcal{H}om(Q,\mathcal{S}_1(P)) &&(\mathcal{S}_1\text{ is an }f_1\text{-Serre functor})\\ &\cong\mathbf{R}(f_2)_*\mathbf{R}\mathcal{H}om(\Phi_E(Q),\mathcal{S}_2(\Phi_E(P))) &&(\text{{xref}`Lemma 4.6`}). \end{aligned}

Note that every object of Perf⁡(Y2)\operatorname{Perf}(Y_2) is isomorphic to one of the form ΦE(P)\Phi_E(P) with P∈Perf⁡(Y1)P \in\operatorname{Perf}(Y_1). Hence, from the equivalence ΦE\Phi_E, it follows that S2\mathcal{S}_2 is an f2f_2-Serre functor. Consequently, Lemma 4.4 implies f2f_2 is Gorenstein.

Smoothness

Lemma 5.1. Let f:Y→Xf : Y \to X be a finitely presented flat morphism of Noetherian algebraic spaces. If the base change of ff along every Spec⁡(k)→X\operatorname{Spec}(k) \to X is smooth where kk is a field, then ff is smooth.

Proof. Choose an étale presentation s:U→Xs : U \to X. Consider the fibered square

Y×XU→f′Us′↓↓sY→fX\begin{CD} Y \times_X U @>{f'}>> U\\ @V{s'}VV @VV{s}V\\ Y @>{f}>> X \end{CD}

Let t:V→Y×XUt : V \to Y \times_X U be an étale presentation. Fix u∈Uu \in U. Denote by iu:Spec⁡(κ(u))→Ui_u : \operatorname{Spec}(\kappa(u)) \to U the canonical morphism. There exists a fibered square

Y×XSpec⁡(κ(u))→fu′Spec⁡(κ(u))iu′↓↓iuY×XU→f′U\begin{CD} Y \times_X \operatorname{Spec}(\kappa(u)) @>{f'_u}>> \operatorname{Spec}(\kappa(u))\\ @V{i'_u}VV @VV{i_u}V\\ Y \times_X U @>{f'}>> U \end{CD}

By the hypothesis, fu′f'_u is smooth. Moreover, we have another fibered square,

V×USpec⁡(κ(u))→tuY×XSpec⁡(κ(u))→fu′Spec⁡(κ(u))iu′′↓↓iu′↓iuV→tY×XU→f′U\begin{CD} V \times_U \operatorname{Spec}(\kappa(u)) @>{t_u}>> Y \times_X \operatorname{Spec}(\kappa(u)) @>{f'_u}>> \operatorname{Spec}(\kappa(u))\\ @V{i''_u}VV @VV{i'_u}V @VV{i_u}V\\ V @>{t}>> Y \times_X U @>{f'}>> U \end{CD}

Base change says tut_u is an étale presentation. Hence, the composition fu′∘tuf'_u \circ t_u is a smooth finitely presented morphism of Noetherian schemes. By [15 Tag 01V8], f′∘tf' \circ t is smooth. Appealing to [15 Tag 03ZF], we have that ff is smooth. □

Lemma 5.2. Let f:Y→Xf : Y \to X be a finitely presented flat morphism of Noetherian algebraic spaces. If for every p∈∣X∣p \in|X| there exists a representative Spec⁡(k)→X\operatorname{Spec}(k) \to X of pp such that the base change of ff along Spec⁡(k)→X\operatorname{Spec}(k) \to X is smooth, then the base change of ff along every Spec⁡(k)→X\operatorname{Spec}(k) \to X is smooth where kk is a field.

Proof. Fix p∈∣X∣p \in|X|. Choose a representative t:Spec⁡(k)→Xt : \operatorname{Spec}(k) \to X of pp such that the base change of ff along tt is smooth. Suppose t′:Spec⁡(k′)→Xt' : \operatorname{Spec}(k') \to X is any representative of pp. By the definition of the points of an algebraic space [15 Tag 03BT], there exist a field ℓ\ell and a commutative diagram

Spec⁡(ℓ)→q2Spec⁡(k)q1↓↓tSpec⁡(k′)→t′X\begin{CD} \operatorname{Spec}(\ell) @>{q_2}>> \operatorname{Spec}(k) \\ @V{q_1}VV @VV{t}V \\ \operatorname{Spec}(k') @>{t'}>> X \end{CD}

Consider the fibered cube

Y×XSpec⁡(ℓ)→q2′Y×XSpec⁡(k)fℓ↓↓fkSpec⁡(ℓ)→q2Spec⁡(k)Y×XSpec⁡(ℓ)→q1′Y×XSpec⁡(k′)→gk′Yfℓ↓↓fk′↓fSpec⁡(ℓ)→q1Spec⁡(k′)→t′XY×XSpec⁡(k)→gkYfk↓↓fSpec⁡(k)→tX\begin{CD} Y \times_X \operatorname{Spec}(\ell) @>{q'_2}>> Y \times_X \operatorname{Spec}(k) \\ @V{f_\ell}VV @VV{f_k}V \\ \operatorname{Spec}(\ell) @>{q_2}>> \operatorname{Spec}(k) \end{CD} \qquad \begin{CD} Y \times_X \operatorname{Spec}(\ell) @>{q'_1}>> Y \times_X \operatorname{Spec}(k') @>{g_{k'}}>> Y \\ @V{f_\ell}VV @VV{f_{k'}}V @VV{f}V \\ \operatorname{Spec}(\ell) @>{q_1}>> \operatorname{Spec}(k') @>{t'}>> X \end{CD} \qquad \begin{CD} Y \times_X \operatorname{Spec}(k) @>{g_k}>> Y \\ @V{f_k}VV @VV{f}V \\ \operatorname{Spec}(k) @>{t}>> X \end{CD}

As fkf_k is smooth, base change implies fℓf_\ell is smooth. Choose an étale presentation s:U→Ys : U \to Y. There exists a further fibered cube

U×XSpec⁡(ℓ)→q2′′U×XSpec⁡(k)sℓ↓↓skY×XSpec⁡(ℓ)→q2′Y×XSpec⁡(k)U×XSpec⁡(ℓ)→q1′′U×XSpec⁡(k′)→sk′Usℓ↓↓sk′↓sY×XSpec⁡(ℓ)→q1′Y×XSpec⁡(k′)→gk′YU×XSpec⁡(k)→s1Usk↓↓sY×XSpec⁡(k)→gkY\begin{CD} U \times_X \operatorname{Spec}(\ell) @>{q''_2}>> U \times_X \operatorname{Spec}(k) \\ @V{s_\ell}VV @VV{s_k}V \\ Y \times_X \operatorname{Spec}(\ell) @>{q'_2}>> Y \times_X \operatorname{Spec}(k) \end{CD} \qquad \begin{CD} U \times_X \operatorname{Spec}(\ell) @>{q''_1}>> U \times_X \operatorname{Spec}(k') @>{s_{k'}}>> U \\ @V{s_\ell}VV @VV{s_{k'}}V @VV{s}V \\ Y \times_X \operatorname{Spec}(\ell) @>{q'_1}>> Y \times_X \operatorname{Spec}(k') @>{g_{k'}}>> Y \end{CD} \qquad \begin{CD} U \times_X \operatorname{Spec}(k) @>{s_1}>> U \\ @V{s_k}VV @VV{s}V \\ Y \times_X \operatorname{Spec}(k) @>{g_k}>> Y \end{CD}

Base change implies sks_k, and hence sℓs_\ell, are étale presentations. Hence, fℓ∘sℓf_\ell\circ s_\ell is smooth and finitely presented. Denote by T#T_\# the set of points of ∣U×XSpec⁡(#)∣|U \times_X \operatorname{Spec}(\#)| where f#∘s#f_\# \circ s_\# is smooth where #∈{k′,ℓ}\# \in\{k',\ell\}. Applying [15 Tag 02V4], (q1′′)−1(Tk′)=Tℓ(q''_1)^{-1}(T_{k'}) = T_\ell. Since q1q_1 is surjective, so is q1′′q''_1. Hence, q1′′((q1′′)−1(Tk′))=Tk′q''_1((q''_1)^{-1}(T_{k'})) = T_{k'}. However, Tℓ=∣U×XSpec⁡(ℓ)∣T_\ell= |U \times_X \operatorname{Spec}(\ell)|, and so fk′∘sk′f_{k'} \circ s_{k'} is smooth. As sk′s_{k'} is an étale presentation from a scheme, [15 Tag 03ZF] says fk′f_{k'} is smooth, which completes the proof. □

Lemma 5.3. Let f:Y→Xf : Y \to X be a finitely presented flat morphism of Noetherian algebraic spaces. Then the following are equivalent:

  1. ff is smooth

  2. the base change of ff along every Spec⁡(k)→X\operatorname{Spec}(k) \to X is smooth where kk is a field

  3. for every p∈∣X∣p \in\lvert X \rvert there exists a representative Spec⁡(k)→X\operatorname{Spec}(k) \to X of pp such that the base change of ff along Spec⁡(k)→X\operatorname{Spec}(k) \to X is smooth.

Proof. (1) ⟹\Longrightarrow (3) follows from [15 Tag 03ZF], (3) ⟹\Longrightarrow (2) is Lemma 5.2, and (2) ⟹\Longrightarrow (1) is Lemma 5.1. □

Lemma 5.4. Let kk be a field. A finitely presented morphism f:X→Spec⁡(k)f : X \to\operatorname{Spec}(k) of algebraic spaces is smooth if, and only if, X×kSpec⁡(k′)X \times_k \operatorname{Spec}(k') is regular for all field extensions k′/kk'/k.

Proof. If ff is smooth, then the base change of ff along all field extensions k′/kk'/k is smooth, and hence, X×kSpec⁡(k′)X \times_k \operatorname{Spec}(k') is regular. Conversely, let s:U→Xs : U \to X be an étale presentation. Fix a field extension k′/kk'/k. If X×kSpec⁡(k′)X \times_k \operatorname{Spec}(k') is regular, then U×kSpec⁡(k′)U \times_k \operatorname{Spec}(k') is regular because the base change of ss along k′/kk'/k yields an étale presentation. By [15 Tag 038V], UU is geometrically regular. Then [15 Tag 038X] implies UU is smooth over kk, i.e. f∘sf \circ s is smooth. Thus, [15 Tag 03ZF] says ff is smooth. □

Remark 5.5. Let XX be a Noetherian algebraic space. Recall that the locus of x∈∣X∣x \in\lvert X \rvert for which the étale local ring OX,x‾\mathcal{O}_{X,\overline{x}} is regular is stable under generalization. Indeed, choose an étale presentation s:U→Xs : U \to X. Let x⇝x′x \rightsquigarrow x' be a specialization in ∣X∣\lvert X \rvert such that OX,x′‾\mathcal{O}_{X,\overline{x'}} is regular. By [15 Tags 03JX, 03JV, and 03K2], there exists a specialization u⇝u′u \rightsquigarrow u' in ∣U∣\lvert U \rvert such that s(u)=xs(u) = x and s(u′)=x′s(u') = x'. By [15 Tag 04KF and 06LN], OU,u′\mathcal{O}_{U,u'} is regular. Hence, OU,u\mathcal{O}_{U,u} is regular because it is a localization of OU,u′\mathcal{O}_{U,u'} [15 Tag 0AFS]. Once more, from [15 Tag 04KF and 06LN], OX,x‾\mathcal{O}_{X,\overline{x}} is regular.

Lemma 5.6. A Noetherian algebraic space XX is regular if, and only if, Perf⁡(X)=Dcohb(X)\operatorname{Perf}(X) = D^b_{\mathrm{coh}}(X).

Proof. Fix an étale presentation s:U→Xs : U \to X with UU affine. If XX is regular, then UU is regular. Hence, for any E∈Dcohb(X)E \in D^b_{\mathrm{coh}}(X), Ls∗E\mathbb{L}s^*E is perfect [15 Tag 0FDC], and so EE is perfect [4 Lemma 3.5]. It follows that Perf⁡(X)=Dcohb(X)\operatorname{Perf}(X) = D^b_{\mathrm{coh}}(X). We prove the converse. Note that every point of ∣X∣\lvert X \rvert specializes to a closed point [15 Tags 03K6 and 005E]. Hence, by Remark 5.5, it suffices to prove that every closed point of ∣X∣\lvert X \rvert belongs to the locus of points whose étale local ring is regular. Let p∈∣X∣p \in\lvert X \rvert be closed. Denote by tp:Spec⁡(κ(p))→Xt_p : \operatorname{Spec}(\kappa(p)) \to X the associated residual space of pp [15 Tag 06R0, 0H1R, and 0318]. By [15 Tag 0H1U], tpt_p is a closed immersion. There exists a fibered square

U×XSpec⁡(κ(p))→spSpec⁡(κ(p))tp′↓↓tpU→sX\begin{CD} U \times_X \operatorname{Spec}(\kappa(p)) @>{s_p}>> \operatorname{Spec}(\kappa(p)) \\ @V{t'_p}VV @VV{t_p}V \\ U @>{s}>> X \end{CD}

By [4 Lemma 2.19], ss is quasi-affine, and hence representable. Base change implies sps_p is an étale presentation. By [15 Tag 02GL], U×XSpec⁡(κ(p))U \times_X \operatorname{Spec}(\kappa(p)) is a disjoint union of spectra of finite separable field extensions of κ(p)\kappa(p). As sps_p is finitely presented, this disjoint union is finite, and hence its source is affine and regular. By Example 2.1, OU×XSpec⁡(κ(p))\mathcal{O}_{U \times_X \operatorname{Spec}(\kappa(p))} finitely builds every object of Dcohb(U×XSpec⁡(κ(p)))D^b_{\mathrm{coh}}(U \times_X \operatorname{Spec}(\kappa(p))). As tpt_p is a closed immersion, R(tp)∗OSpec⁡(κ(p))R(t_p)_*\mathcal{O}_{\operatorname{Spec}(\kappa(p))} belongs to Dcohb(X)D^b_{\mathrm{coh}}(X), which equals Perf⁡(X)\operatorname{Perf}(X) by hypothesis. By flat base change, R(tp′)∗OU×XSpec⁡(κ(p))R(t'_p)_*\mathcal{O}_{U \times_X \operatorname{Spec}(\kappa(p))} is perfect. Hence, for all E∈Dcohb(U×XSpec⁡(κ(p)))E \in D^b_{\mathrm{coh}}(U \times_X \operatorname{Spec}(\kappa(p))), R(tp′)∗ER(t'_p)_*E is perfect. Choose a closed point in ∣U×XSpec⁡(κ(p))∣|U \times_X \operatorname{Spec}(\kappa(p))|. Set u∈∣U∣u \in|U| to be its image under ∣tp′∣|t'_p|. Denote by hu ⁣:Spec⁡(OU,u)→Uh_u \colon\operatorname{Spec}(\mathcal{O}_{U,u}) \to U the canonical morphism. Consider the fibered square

Spec⁡(OU,u)×XSpec⁡(κ(p))→hu′U×XSpec⁡(κ(p))tp′′↓↓tp′Spec⁡(OU,u)→huU\begin{CD} \operatorname{Spec}(\mathcal{O}_{U,u}) \times_X \operatorname{Spec}(\kappa(p)) @>{h'_u}>> U \times_X \operatorname{Spec}(\kappa(p)) \\ @V{t''_p}VV @VV{t'_p}V \\ \operatorname{Spec}(\mathcal{O}_{U,u}) @>{h_u}>> U \end{CD}

Base change implies tp′′t''_p is a closed immersion. By [8 Lemma 4.8], L(hu′)∗L(h'_u)^* is essentially dense on DcohbD^b_{\mathrm{coh}}. Coupled with flat base change, it follows that

R(tp′′)∗Dcohb(Spec⁡(OU,u)×XSpec⁡(κ(p)))⊆⟨R(tp′′)∗L(hu′)∗Dcohb(U×XSpec⁡(κ(p)))⟩R(t''_p)_*D^b_{\mathrm{coh}}(\operatorname{Spec}(\mathcal{O}_{U,u}) \times_X \operatorname{Spec}(\kappa(p))) \subseteq\langle R(t''_p)_*L(h'_u)^*D^b_{\mathrm{coh}}(U \times_X \operatorname{Spec}(\kappa(p))) \rangle
⊆⟨Lhu∗R(tp′)∗Dcohb(U×XSpec⁡(κ(p)))⟩\subseteq\langle Lh_u^*R(t'_p)_*D^b_{\mathrm{coh}}(U \times_X \operatorname{Spec}(\kappa(p))) \rangle
⊆⟨Lhu∗Perf⁡(U)⟩\subseteq\langle Lh_u^*\operatorname{Perf}(U) \rangle
⊆Perf⁡(OU,u).\subseteq\operatorname{Perf}(\mathcal{O}_{U,u}).

Denote by gu ⁣:Spec⁡(κ(u))→Spec⁡(OU,u)g_u \colon\operatorname{Spec}(\kappa(u)) \to\operatorname{Spec}(\mathcal{O}_{U,u}) the canonical morphism. By [15 Tag 0356], gug_u factors through tp′′t''_p. Since gug_u and tp′′t''_p are closed immersions, the induced morphism from Spec⁡(κ(u))\operatorname{Spec}(\kappa(u)) to the source of tp′′t''_p is a closed immersion. Its structure sheaf pushforward is coherent, and pushforwards along closed immersions are exact [15 Tag 04CJ]. Thus, R(gu)∗OSpec⁡(κ(u))R(g_u)_*\mathcal{O}_{\operatorname{Spec}(\kappa(u))} belongs to the left-hand side of the above inclusion, and hence is perfect. This implies OU,u\mathcal{O}_{U,u} is a regular local ring [15 Tag 00OC]. Hence, uu is a regular point of UU. As s(u)=ps(u)=p, it follows that XX is regular at pp [15 Tag 0AH9]. □

Proposition 5.7. Let fi ⁣:Yi→Sf_i \colon Y_i \to S be proper flat morphisms of Noetherian algebraic spaces. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by a kernel E∈Dcohb(Y1×SY2)E \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2). Then f1f_1 is smooth if, and only if, f2f_2 is smooth.

Proof. By [4 Remark 2.7], EE is relatively perfect over each YiY_i. By symmetry, it suffices to prove the case where f1f_1 is smooth. Fix a morphism Spec⁡(k)→S\operatorname{Spec}(k) \to S from a field. Let L/kL/k be a field extension. The composition Spec⁡(L)→S\operatorname{Spec}(L) \to S is affine [15 Tag 09TF]. By [4 Theorem 10.13], Y1×SSpec⁡(L)Y_1 \times_S \operatorname{Spec}(L) and Y2×SSpec⁡(L)Y_2 \times_S \operatorname{Spec}(L) are Fourier–Mukai Spec⁡(L)\operatorname{Spec}(L)-partners. The kernel is the derived pullback of EE to Y1×SY2×SSpec⁡(L)Y_1 \times_S Y_2 \times_S \operatorname{Spec}(L), which is relatively perfect over each Yi×SSpec⁡(L)Y_i \times_S \operatorname{Spec}(L) [4 Corollary 6.18]. Consequently, [4 Lemma 10.9] implies the associated integral transform restricts to equivalences on both DcohbD^b_{\mathrm{coh}} and Perf⁡\operatorname{Perf}. As Y1×SSpec⁡(L)Y_1 \times_S \operatorname{Spec}(L) is smooth over LL, it is regular. By Lemma 5.6, it follows that

Perf⁡(Y1×SSpec⁡(L))=Dcohb(Y1×SSpec⁡(L)).\operatorname{Perf}(Y_1 \times_S \operatorname{Spec}(L)) = D^b_{\mathrm{coh}}(Y_1 \times_S \operatorname{Spec}(L)).

The integral transform takes the left-hand side to Perf⁡(Y2×SSpec⁡(L))\operatorname{Perf}(Y_2 \times_S \operatorname{Spec}(L)) and the right-hand side to Dcohb(Y2×SSpec⁡(L))D^b_{\mathrm{coh}}(Y_2 \times_S \operatorname{Spec}(L)). Hence, every object of Dcohb(Y2×SSpec⁡(L))D^b_{\mathrm{coh}}(Y_2 \times_S \operatorname{Spec}(L)) is perfect, and so Lemma 5.6 implies Y2×SSpec⁡(L)Y_2 \times_S \operatorname{Spec}(L) is regular. Applying Lemma 5.4, Y2×SSpec⁡(k)Y_2 \times_S \operatorname{Spec}(k) is smooth over kk because LL was arbitrary. Therefore, by Lemma 5.3, f2f_2 is smooth. □

Lemma 5.8. Let f1:Y1→Sf_1:Y_1 \to S and f2:Y2→Sf_2:Y_2 \to S be proper flat morphisms to a regular Noetherian algebraic space. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by a kernel B∈Dcohb(Y1×SY2)B \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2). If Y1Y_1 is not regular, then B∉Perf⁡(Y1×SY2)B \notin\operatorname{Perf}(Y_1 \times_S Y_2).

Proof. By [4 Remark 2.7], BB is relatively perfect over each YiY_i. We prove the claim by contradiction. Assume that B∈Perf⁡(Y1×SY2)B \in\operatorname{Perf}(Y_1 \times_S Y_2). By [4 Lemma 10.9], ΦB\Phi_B restricts to triangulated equivalences on DcohbD^b_{\mathrm{coh}} and Perf⁡\operatorname{Perf}. We first show that ΦB(Dcohb(Y1))⊆Perf⁡(Y2)\Phi_B(D^b_{\mathrm{coh}}(Y_1)) \subseteq\operatorname{Perf}(Y_2). Choose a classical generator GiG_i for Perf⁡(Yi)\operatorname{Perf}(Y_i), see [15 Tags 09IY, 09M8, and 09SR]. By [4 Lemma 5.4] and [15 Tag 09SR], Lp1∗G1⊗LLp2∗G2L p_1^*G_1 \otimes^{\mathrm{L}} L p_2^*G_2 is a classical generator for Perf⁡(Y1×SY2)\operatorname{Perf}(Y_1 \times_S Y_2). In particular, BB is finitely built by Lp1∗G1⊗LLp2∗G2L p_1^*G_1 \otimes^{\mathrm{L}} L p_2^*G_2. Fix E∈Dcohb(Y1)E \in D^b_{\mathrm{coh}}(Y_1). As R(p2)∗(Lp1∗E⊗L−)R(p_2)_*(L p_1^*E \otimes^{\mathrm{L}} -) is an exact functor, it follows that

ΦB(E)∈⟨R(p2)∗(Lp1∗E⊗LLp1∗G1⊗LLp2∗G2)⟩.\Phi_B(E) \in\left\langle R(p_2)_*(L p_1^*E \otimes^{\mathrm{L}} L p_1^*G_1 \otimes^{\mathrm{L}} L p_2^*G_2)\right\rangle.

By the projection formula and flat base change, there are isomorphisms

R(p2)∗(Lp1∗E⊗LLp1∗G1⊗LLp2∗G2)≅R(p2)∗Lp1∗(E⊗LG1)⊗LG2R(p_2)_*(L p_1^*E \otimes^{\mathrm{L}} L p_1^*G_1 \otimes^{\mathrm{L}} L p_2^*G_2) \cong R(p_2)_*L p_1^*(E \otimes^{\mathrm{L}} G_1) \otimes^{\mathrm{L}} G_2
≅Lf2∗R(f1)∗(E⊗LG1)⊗LG2.\cong L f_2^*R(f_1)_*(E \otimes^{\mathrm{L}} G_1) \otimes^{\mathrm{L}} G_2.

As f1f_1 is proper, R(f1)∗(E⊗LG1)∈Dcohb(S)R(f_1)_*(E \otimes^{\mathrm{L}} G_1) \in D^b_{\mathrm{coh}}(S). Since SS is regular, Lemma 5.6 implies Dcohb(S)=Perf⁡(S)D^b_{\mathrm{coh}}(S) = \operatorname{Perf}(S). Hence, ΦB(E)\Phi_B(E) is perfect, as desired. Now, ΦB\Phi_B takes Dcohb(Y1)D^b_{\mathrm{coh}}(Y_1) to Dcohb(Y2)D^b_{\mathrm{coh}}(Y_2), so Dcohb(Y2)=Perf⁡(Y2)D^b_{\mathrm{coh}}(Y_2) = \operatorname{Perf}(Y_2). As ΦB\Phi_B restricts to equivalences on DcohbD^b_{\mathrm{coh}} and Perf⁡\operatorname{Perf}, it follows that Dcohb(Y1)=Perf⁡(Y1)D^b_{\mathrm{coh}}(Y_1) = \operatorname{Perf}(Y_1). Thus, Lemma 5.6 implies Y1Y_1 is regular, which is a contradiction. □\square

Lemma 5.9. Let f:Y→Xf:Y \to X be a finitely presented morphism of Noetherian algebraic spaces. For each p∈∣X∣p \in|X|, denote by tp:Spec⁡(κ(p))→Xt_p:\operatorname{Spec}(\kappa(p)) \to X the canonical morphism from the residue field of pp, and by tp′:Y×XSpec⁡(κ(p))→Yt'_p:Y \times_X \operatorname{Spec}(\kappa(p)) \to Y the morphism obtained by base change along tpt_p. If E∈Dcohb(Y)E \in D^b_{\mathrm{coh}}(Y) satisfies L(tp′)∗E∈Perf⁡(Y×XSpec⁡(κ(p)))L(t'_p)^*E \in\operatorname{Perf}(Y \times_X \operatorname{Spec}(\kappa(p))) for all p∈∣X∣p \in|X|, then E∈Perf⁡(Y)E \in\operatorname{Perf}(Y).

Proof. As ff is finitely presented, Y×XSpec⁡(κ(p))Y \times_X \operatorname{Spec}(\kappa(p)) is Noetherian for all p∈∣X∣p \in|X|. Choose q∈∣Y∣q \in|Y|, and set p:=f(q)p := f(q). By [15 Tag 03H4], there exists q′∈∣Y×XSpec⁡(κ(p))∣q' \in|Y \times_X \operatorname{Spec}(\kappa(p))| which maps to qq. Denote by b:Spec⁡(κ(q′))→Y×XSpec⁡(κ(p))b:\operatorname{Spec}(\kappa(q')) \to Y \times_X \operatorname{Spec}(\kappa(p)) the canonical morphism from the residue field of q′q', see [15 Tag 0EMV]. The image of tp′∘bt'_p \circ b is qq, and so tp′∘bt'_p \circ b represents qq [4 Lemma 2.8]. As L(tp′)∗EL(t'_p)^*E is perfect, so is L(tp′∘b)∗E≃Lb∗L(tp′)∗EL(t'_p \circ b)^*E \simeq Lb^*L(t'_p)^*E. In particular, L(tp′∘b)∗EL(t'_p \circ b)^*E has bounded cohomology. Since qq was arbitrary, [4 Proposition 3.12] implies EE is perfect. □\square

Proposition 5.10. Let fi:Yi→Sf_i:Y_i \to S be proper flat morphisms of Noetherian algebraic spaces. Suppose Y1Y_1 and Y2Y_2 are Fourier–Mukai SS-partners given by a kernel E∈Dcohb(Y1×SY2)E \in D^b_{\mathrm{coh}}(Y_1 \times_S Y_2). Then f1f_1 is smooth if, and only if, E∈Perf⁡(Y1×SY2)E \in\operatorname{Perf}(Y_1 \times_S Y_2).

Proof. By [4 Remark 2.7], EE is relatively perfect over each YiY_i. Assume that f1f_1 is smooth. Choose p∈∣S∣p \in|S|. Denote by tp:Spec⁡(κ(p))→St_p:\operatorname{Spec}(\kappa(p)) \to S the canonical morphism associated to the residue field of pp. Denote by tp′:Y1×SY2×SSpec⁡(κ(p))→Y1×SY2t'_p:Y_1 \times_S Y_2 \times_S \operatorname{Spec}(\kappa(p)) \to Y_1 \times_S Y_2 the canonical morphism obtained by base change along tpt_p. As f1f_1 is smooth, Proposition 5.7 implies f2f_2 is smooth. Hence, by base change, Y1×SY2×SSpec⁡(κ(p))Y_1 \times_S Y_2 \times_S \operatorname{Spec}(\kappa(p)) is smooth over κ(p)\kappa(p). Then Lemma 5.6 implies

Perf⁡(Y1×SY2×SSpec⁡(κ(p)))=Dcohb(Y1×SY2×SSpec⁡(κ(p))).(3)\operatorname{Perf}(Y_1 \times_S Y_2 \times_S \operatorname{Spec}(\kappa(p))) = D^b_{\mathrm{coh}}(Y_1 \times_S Y_2 \times_S \operatorname{Spec}(\kappa(p))). \tag*{(3)}

::: By [4 Corollary 6.18], L(tp′∗E)\mathbf{L}(t_p'^*E) is relatively perfect over each Yi×SSpec⁡(κ(p))Y_i \times_S \operatorname{Spec}(\kappa(p)). In particular, L(tp′∗E)\mathbf{L}(t_p'^*E) has bounded cohomology, and so it is perfect. Since this holds for all p∈∣S∣p \in|S|, Lemma 5.9 asserts that EE is perfect.

Assume that EE is perfect. Choose p∈∣S∣p \in|S|. Denote by tp ⁣:Spec⁡(κ(p)‾)→St_p\colon\operatorname{Spec}(\overline{\kappa(p)}) \to S a morphism from an algebraic closure of the residue field of pp. Denote by tp′ ⁣:Y1×SY2×SSpec⁡(κ(p)‾)→Y1×SY2t'_p\colon Y_1 \times_S Y_2 \times_S \operatorname{Spec}(\overline{\kappa(p)}) \to Y_1 \times_S Y_2 the canonical morphism obtained by base change along tpt_p. By [4 Corollary 6.18], L(tp′∗E)\mathbf{L}(t_p'^*E) is relatively perfect over each Yi×SSpec⁡(κ(p)‾)Y_i \times_S \operatorname{Spec}(\overline{\kappa(p)}). Moreover, tpt_p is affine [15 Tag 09TF], so [4 Theorem 10.13] implies L(tp′∗E)\mathbf{L}(t_p'^*E) induces an equivalence on DcohbD_{\mathrm{coh}}^b.

Furthermore, L(tp′∗E)\mathbf{L}(t_p'^*E) is perfect, and so Lemma 5.8 implies Y1×SSpec⁡(κ(p)‾)Y_1 \times_S \operatorname{Spec}(\overline{\kappa(p)}) is regular. As κ(p)‾\overline{\kappa(p)} is perfect, it has no nontrivial finite purely inseparable extensions. Applying [15 Tags 038V and 038X] to an étale presentation of Y1×SSpec⁡(κ(p)‾)Y_1 \times_S \operatorname{Spec}(\overline{\kappa(p)}), it follows from [15 Tag 03ZF] that Y1×SSpec⁡(κ(p)‾)Y_1 \times_S \operatorname{Spec}(\overline{\kappa(p)}) is smooth over κ(p)‾\overline{\kappa(p)}. Since tpt_p represents pp and p∈∣S∣p \in|S| was arbitrary, Lemma 5.3 asserts that f1f_1 is smooth, which completes the proof. □

References

E. Guisado Villalgordo, BCAM – Basque Center for Applied Mathematics, Mazarredo 14, 48009 Bilbao, Basque Country – Spain

Email address: [email protected]

P. Lank, Dipartimento di Matematica “F. Enriques”, Università degli Studi di Milano, Via Cesare Saldini 50, 20133 Milano, Italy

Email address: [email protected]

References

  1. [1][Bal11] Matthew Robert Ballard. Derived categories of sheaves on singular schemes with an application to reconstruction. Adv. Math., 227(2):895–919, 2011.arxiv.org/abs/0801.2599
  2. [3][Cal18] John Calabrese. Relative singularity twisted standard-Orlov. Math. Res. Lett., 25(2):393–414, 2018.
  3. [4][GL26] Elías Guisado Villalgordo and Pat Lank. Relative Fourier–Mukai theory for algebraic spaces. arXiv:2609.01298, 2026.arxiv.org/abs/2609.01298
  4. [5][GLMP25] Elías Guisado Villalgordo, Pat Lank, Kabeer Manali Rahul, and Nebojsa Pavic. Fiberwise criteria for Fourier–Mukai equivalences. arXiv:2512.16503, 2025.arxiv.org/abs/2512.16503
  5. [6][HLS09] Daniel Hernández Ruipérez, Ana Cristina López Martín, and Fernando Sancho de Salas. Relative integral functors for singular fibrations and singular partners. J. Eur. Math. Soc. (JEMS), 11(3):597–625, 2009.arxiv.org/abs/math/0610319
  6. [7][Kol11] János Kollár. Simultaneous normalization and algebra husks. Asian J. Math., 15(3):437–450, 2011.arxiv.org/abs/0910.1076
  7. [8][Lan26] Pat Lank. Germwise singular Fourier–Mukai partnerships. arXiv:2610.29944, 2026.arxiv.org/abs/2610.29944
  8. [9][LS16] Valery A. Lunts and Olaf M. Schnürer. New enhancements of derived categories of coherent sheaves and applications. J. Algebra, 446:203–274, 2016.arxiv.org/abs/1406.7559
  9. [10][Muk81] Shigeru Mukai. Duality between D(X) and D(X̂) with its application to Picard sheaves. Nagoya Math. J., 81:153–175, 1981.
  10. [11][Nee92] Amnon Neeman. The chromatic tower for D(R). With an appendix by Marcel Böckstedt. Topology, 31(3):519–532, 1992.
  11. [12][Nee23] Amnon Neeman. An improvement on the base-change theorem and the functor f!. Bull. Iran. Math. Soc., 49(3):163, 2023. Id/No 25.DOI
  12. [13][Spe23] Dylan Spence. Reconstruction of projective curves from the derived category. Mich. Math. J., 73(1):171–194, 2023.
  13. [14][SS12] Carlos Sancho de Salas and Fernando Sancho de Salas. Reconstructing schemes from the derived category. Proc. Edinb. Math. Soc., II Ser., 55(3):781–796, 2012.DOI
  14. [15][Sta26] The Stacks Project Authors. Stacks Project. https://stacks.math.columbia.edu, 2026.

Paper details

vlnc:W_k3x77bux
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 4

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

large language models (LLMs)

View statement
Looking for related papers…
Contents