Derived equivalence for stratified Mukai flop on $G\left(2,4\right)$

November 27, 2006

1 Introduction

We consider a conjecture on the derived equivalence for $K$  -equivalent varieties in the case of a certain flop between $9$  -dimensional varieties. The following is a very general version of this conjecture on the fully faithful derived embedding for log $K$  -related varieties ([11 Conjecture 2.2):
Conjecture 1.1. Let $\left(X,B\right)$  and $\left(Y,C\right)$  be pairs of quasi-projective varieties with $\mathbb{Q}$  -divisors such that there exist quasi-finite and surjective morphisms $\pi :U\to X$  and $\sigma :V\to Y$  from smooth varieties, which may be reducible, such that ${\pi }^{*}\left({K}_{X}+B\right)={K}_{U}$  and ${\sigma }^{*}\left({K}_{Y}+C\right)={K}_{V}$  . Let $\mathcal{X}\to X$  and $\mathcal{Y}\to Y$  be natural morphisms from the associated Deligne-Mumford stacks. Assume that there are proper birational morphisms $\mu :W\to X$  and $\nu :W\to Y$  from a third variety $W$  such that ${\mu }^{*}\left({K}_{X}+B\right)\le {\nu }^{*}\left({K}_{Y}+C\right)$  . Then there exists a fully faithful exact functor ${D}^{b}\left(\text{Coh}\left(\mathcal{X}\right)\right)\to {D}^{b}\left(\text{Coh}\left(\mathcal{Y}\right)\right)$  .
The conjecture is proved to be true in some cases in [2, [5, [4, [9, [6, [13, [10, [11and [1.
On the other hand, Namikawa [14proved that a naturally defined functor between the derived categories for the stratified Mukai flop on $G\left(2,4\right)$  is not an equivalence. This is a flop between $9$  -dimensional varieties which is the total space of a $3$  -parameter degeneration of standard $3$  -dimensional flops of $\left(-1,-1\right)$  -curves and the fiber over the most degenerate point is isomorphic to $G\left(2,4\right)$  .
So it is worthwhile to check the conjecture in this special case. We shall prove that there is nevertheless another functor between the same categories which is an equivalence.

2 Stratified Mukai flop

We recall the construction of stratified Mukai flops due to Markman [12(see also [14).
Let $G=G\left(r,n\right)$  be the Grassmann variety of $r$  -dimensional subspaces in an $n$  -dimensional vector space $V$  . We assume that $2r\le n$  . Let $S$  (resp. $Q$  ) be the tautological subbundle (resp. quotient) bundle on $G$  : $0\to S\to V\otimes {\mathcal{O}}_{G}\to Q\to 0.$  The polarization of $G$  is given by $\text{det}Q={\mathcal{O}}_{G}\left(1\right).$  We have ${\Omega }_{G}^{1}\sim =\mathcal{ℋ}om\left(Q,S\right)$  . Since ${\text{Ext}}^{1}\left({\mathcal{O}}_{G},{\Omega }_{G}^{1}\right)\sim ={\text{Ext}}^{1}\left(Q,S\right)\sim =\mathbb{C}$  there exists a non-trivial extension of vector bundles on $G$  induced by the natural homomorphism ${\mathcal{O}}_{G}\to \mathcal{ℋ}om\left(S,S\right)$  : $0---\to {\Omega }_{G}^{1}---\to {\stackrel{~}{\Omega }}_{G}^{1}---\to {\mathcal{O}}_{G}---\to 0\parallel ↓↓0---\to \mathcal{ℋ}om\left(Q,S\right)---\to \mathcal{ℋ}om\left(V\otimes {\mathcal{O}}_{G},S\right)---\to \mathcal{ℋ}om\left(S,S\right)---\to 0.$  Let ${X}_{0}={T}^{*}G$  and $X$  be the total spaces of ${\Omega }_{G}^{1}$  and ${\stackrel{~}{\Omega }}_{G}^{1}$  , respectively, and $\pi :X\to G$  the projection. We write ${\pi }^{*}S={S}_{X}$  and ${\pi }^{*}{\mathcal{O}}_{G}\left(1\right)={\mathcal{O}}_{X}\left(1\right)$  . A point of $X$  is given by a pair consisting of a point $p\in G$  and a homomorphism $A:V\to {S}_{p}\subset V$  which induces a homothety $t:{S}_{p}\to {S}_{p}$  ( $t\in \mathbb{C}$  ). We have $\left(p,A\right)\in {X}_{0}$  if and only if $t=0$  . Let $Z=\left\{A\in \text{End}\left(V\right)|\exists t\in \mathbb{C}\text{s.t.}{A}^{2}=tA,\text{rank}A\le r\right\}$  with a morphism $t:Z\to \mathbb{C}$  , and set ${Z}_{0}={t}^{-1}\left(0\right)$  . Then there is a natural morphism $\phi :X\to Z$  given by $\phi \left(p,A\right)=A$  , which induces a morphism ${\phi }_{0}:{X}_{0}\to {Z}_{0}$  . We know that ${\phi }_{0}$  and $\phi$  are projective birational morphisms.
We write ${W}^{\left(k\right)}=\left\{\left(p,A\right)\in X|\text{rank}A\le k\right\}.$  Then the exceptional loci of both ${\phi }_{0}$  and $\phi$  are equal to $W={W}^{\left(r-1\right)}$  . We have $dimG=r\left(n-r\right)$  , $dim{X}_{0}=2r\left(n-r\right)$  , $dimX=2r\left(n-r\right)+1$  , and $dimW=2\left(r-1\right)\left(n-r+1\right)+n-2r+1$  . We have a dual construction starting from the dual vector space ${V}^{*}$  . Let ${G}^{+}$  be the Grassmann variety of $r$  -dimensional subspaces in ${V}^{*}$  , ${S}^{+}$  (resp.
${Q}^{+}$  ) the tautological subbundle (resp. quotient bundle) on ${G}^{+}$  , $\text{det}{Q}^{+}={\mathcal{O}}_{{G}^{+}}\left(1\right)$  , ${X}_{0}^{+}$  (resp. ${X}^{+}$  ) the total space of ${\Omega }_{{G}^{+}}^{1}$  (resp. ${\stackrel{~}{\Omega }}_{{G}^{+}}^{1}$  ), ${\pi }^{+}:{X}^{+}\to {G}^{+}$  the projection, ${\pi }^{+*}{S}^{+}={S}_{{X}^{+}}^{+}$  , and ${\pi }^{+*}{\mathcal{O}}_{{G}^{+}}\left(1\right)={\mathcal{O}}_{{X}^{+}}\left(1\right)$  . There is a natural projective birational morphism ${\phi }^{+}:{X}^{+}\to Z$  given by ${\phi }^{+}\left(q,B\right){=}^{t}B$  which induces a projective birational morphism ${\phi }_{0}^{+}:{X}_{0}^{+}\to {Z}_{0}$  . We write ${W}^{+\left(k\right)}=\left\{\left(q,B\right)\in {X}^{+}|\text{rank}B\le k\right\}.$  The exceptional locus of $\phi$  is ${W}^{+}={W}^{+\left(r-1\right)}$  . The diagram $X---\to \phi Z←---{\phi }^{+}{X}^{+}$  thus obtained is called a stratified Mukai flop. By restricting to the subspaces defined by $t=0$  , we have a smaller diagram ${X}_{0}---\to \phi {Z}_{0}←---{\phi }^{+}{X}_{0}^{+}$  which is also called a stratified Mukai flop.
If $r=1$  , then the above diagrams are reduced to a standard flop and a Mukai flop.
The birational map $\left({\phi }^{+}{\right)}^{-1}\circ \phi :X-\to {X}^{+}$  is decomposed into blow-ups and downs ([12). We consider only the case $r=2$  in the following. Let ${f}_{1}:{X}_{1}\to X$  be the blowing up along the center $G={W}^{\left(0\right)}$  , the $0$  -section of the projection $\pi$  . Then the strict transform ${W}^{\prime }$  of $W={W}^{\left(1\right)}$  is smooth.
Indeed, the projection ${W}^{\prime }\to G$  is smooth and its fibers are isomorphic to the cone over ${\mathbb{P}}^{1}×{\mathbb{P}}^{n-3}\subset {\mathbb{P}}^{2n-5}$  . Let ${f}_{2}:{X}_{2}\to {X}_{1}$  be the blowing up along the center ${W}^{\prime }$  . Let ${E}_{i}$  be the exceptional divisor of ${f}_{i}$  for $i=1,2$  , and ${E}_{1}^{\prime }={f}_{2*}^{-1}{E}_{1}$  , the strict transform of ${E}_{1}$  . We set $f={f}_{2}\circ {f}_{1}$  . If $r\ge 3$  , then we have similar construction with more blow-ups ([12).
On the dual side, let ${f}_{1}^{+}:{X}_{1}^{+}\to {X}^{+}$  be the blowing up along the center ${G}^{+}={W}^{+\left(0\right)}$  , the $0$  -section of the projection ${\pi }^{+}$  . Then the strict transform $\left({W}^{+}{\right)}^{\prime }$  of ${W}^{+}={W}^{+\left(1\right)}$  is smooth. Let ${f}_{2}^{+}:{X}_{2}^{+}\to {X}_{1}^{+}$  be the blowing up along the center $\left({W}^{+}{\right)}^{\prime }$  . Let ${E}_{i}^{+}$  be the exceptional divisor of ${f}_{i}^{+}$  for $i=1,2$  , and $\left({E}_{1}^{+}{\right)}^{\prime }=\left({f}_{2*}^{+}{\right)}^{-1}{E}_{1}^{+}$  . We set ${f}^{+}={f}_{2}^{+}\circ {f}_{1}^{+}$  .
The birational map $\left({\phi }^{+}{\right)}^{-1}\circ \phi :X-\to {X}^{+}$  induces an isomorphism ${X}_{2}\to {X}_{2}^{+}$  ([12). We write $Y={X}_{2}={X}_{2}^{+}$  under this identification. Then we have ${E}_{1}^{\prime }=\left({E}_{1}^{+}{\right)}^{\prime }$  , ${E}_{2}={E}_{2}^{+}$  , and a commutative diagram $A\in Z←---\phi X←---{f}_{1}{X}_{1}←---{f}_{2}{X}_{2}\parallel \parallel {A}^{+}\in {Z}^{+}←---{\phi }^{+}{X}^{+}←---{f}_{1}^{+}{X}_{1}^{+}←---{f}_{2}^{+}{X}_{2}^{+}$  Since $dimX=4n-7$  , $dimG=2n-4$  and $dimW=3n-5$  , we have ${K}_{Y}={f}^{*}{K}_{X}+\left(2n-4\right){E}_{1}^{\prime }+\left(n-3\right){E}_{2}$  .
The isomorphism ${X}_{2}\to {X}_{2}^{+}$  can be described set theoretically in the following way.
(1) For a point $\left(p,A\right)\in X\W$  , we have $\text{rank}A=2$  and ${S}_{p}=\text{Im}A$  . The corresponding point $\left(q{,}^{t}A\right)\in {X}^{+}\{W}^{+}$  is given by ${S}_{q}^{+}={\text{Im}}^{t}A=\left(\text{Ker}A{\right)}^{\perp }$  .
(2) For a point ${lim}_{\epsilon \to 0}\left(p,\epsilon {A}_{1}\right)\in {E}_{1}\{W}^{\prime }$  , with $\text{rank}{A}_{1}=2$  and ${S}_{p}=\text{Im}{A}_{1}$  , the corresponding point is ${lim}_{\epsilon \to 0}\left(q,{\epsilon }^{t}{A}_{1}\right)\in {E}_{1}^{+}\\left({W}^{+}{\right)}^{\prime }$  , where ${S}_{q}^{+}={\text{Im}}^{t}{A}_{1}$  .
(3) For a point $\left(p,A\right)\in W\G$  , we have $\text{rank}A=1$  and ${S}_{p}\supset \text{Im}A$  . We take ${A}_{2}$  such that ${S}_{p}=\text{Im}\left(A+{A}_{2}\right)$  . Then a point ${lim}_{\epsilon \to 0}\left(p,A+\epsilon {A}_{2}\right)\in {E}_{2}\{E}_{1}^{\prime }$  corresponds to a point ${lim}_{\epsilon \to 0}\left({q}_{\epsilon }{,}^{t}\left(A+\epsilon {A}_{2}\right)\right)\in {E}_{2}^{+}\\left({E}_{1}^{+}{\right)}^{\prime }$  which is over a point $q\in {G}^{+}$  given by ${S}_{q}^{+}={lim}_{\epsilon \to 0}{\text{Im}}^{t}\left(A+\epsilon {A}_{2}\right)$  .
(4) For a point ${lim}_{\epsilon \to 0}\left(p,\epsilon {A}_{1}\right)\in {E}_{1}\cap {W}^{\prime }$  with $\text{rank}{A}_{1}=1$  and ${S}_{p}\supset \text{Im}{A}_{1}$  , we take ${A}_{2}$  such that ${S}_{p}=\text{Im}\left({A}_{1}+{A}_{2}\right)$  . Then a point ${lim}_{\epsilon \to 0}\left(p,\epsilon {A}_{1}+{\epsilon }^{2}{A}_{2}\right)\in {E}_{1}^{\prime }\cap {E}_{2}$  corresponds to a point ${lim}_{\epsilon \to 0}\left({q}_{\epsilon }{,}^{t}\left(\epsilon {A}_{1}+{\epsilon }^{2}{A}_{2}\right)\right)\in \left({E}_{1}^{+}{\right)}^{\prime }\cap {E}_{2}^{+}$  which is over a point $q\in {G}^{+}$  given by ${S}_{q}^{+}={lim}_{\epsilon \to 0}\text{Im}{\left(}^{t}\left({A}_{1}+\epsilon {A}_{2}\right)\right)$  .
The above description shows that the birational map ${X}_{2}-\to {X}_{2}^{+}$  induced by $\left({\phi }^{+}{\right)}^{-1}\circ \phi$  is an isomorphism by the Zariski main theorem. As a consequence, we have the following lemma:
Lemma 2.1. ${S}_{X}^{*}{|}_{X\W}={S}_{{X}^{+}}^{+}{|}_{{X}^{+}\{W}^{+}}.$  Moreover, ${f}^{*}{S}_{X}^{*}$  is a locally free subsheaf of ${f}^{+*}{S}_{{X}^{+}}^{+}$  such that a local section of the latter belongs to the former if and only if its value in ${S}_{q}^{+}$  at any point in $\left({f}^{+}{\right)}^{-1}\left({x}^{+}\right)$  for ${x}^{+}=\left(q,B\right)$  is contained in $\text{Im}B$  .
• Proof. For $\left(p,A\right)\in X\W$  , we have ${\text{Ker}}^{t}A=\left(\text{Im}A{\right)}^{\perp }$  from (1) above.
Therefore, ${}^{t}A:{V}^{*}\to {V}^{*}$  induces an isomorphism ${S}_{p}^{*}={V}^{*}/{\text{Ker}}^{t}A\to {S}_{q}^{+}$  , hence the first assertion.
Any element $v\in {V}^{*}$  determines a global section of ${f}^{*}{S}_{X}^{*}$  , and the sheaf ${f}^{*}{S}_{X}^{*}$  is generated by such sections. The value at the point $\left(q{,}^{t}A\right)\in {X}^{+}\{W}^{+}$  of the corresponding section of ${S}_{{X}^{+}}^{+}$  is given by ${}^{t}Av\subset {S}_{q}^{+}$  , hence the second assertion.
By taking the determinants, we obtain:
Corollary 2.2. ${f}^{*}{\mathcal{O}}_{X}\left(1\right)={f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-1\right)\otimes {\mathcal{O}}_{Y}\left(-2{E}_{1}^{\prime }-{E}_{2}\right).$  $\square$
Let $P$  be a fiber of ${f}_{1}$  above a point in $G$  , ${P}_{0}=P\cap {f}_{1*}^{-1}{X}_{0}$  and ${P}_{1}=P\cap {W}^{\prime }$  .
Then the sequence ${P}_{1}\subset {P}_{0}\subset P$  is isomorphic to ${\mathbb{P}}^{1}×{\mathbb{P}}^{n-3}\subset {\mathbb{P}}^{2n-5}\subset {\mathbb{P}}^{2n-4}$  , where the first inclusion is the Segre embedding and the second is linear. Let ${P}^{\prime }={f}_{2*}^{-1}P$  , ${P}_{0}^{\prime }={f}_{2*}^{-1}{P}_{0}$  , and ${P}_{1}^{\prime }={f}_{2}^{-1}\left({P}_{1}\right)$  . Let ${l}_{1}$  be a line on ${P}_{0}$  which meets ${W}^{\prime }$  at $2$  points, ${l}_{1}^{\prime }={f}_{2*}^{-1}{l}_{1}$  , and ${l}_{2}$  a fiber of ${f}_{2}$  over a point in ${W}^{\prime }$  .
Lemma 2.3. The cone of curves $\overline{NE}\left(Y/X\right)$  is generated by the classes of ${l}_{1}^{\prime }$  and ${l}_{2}$  . The intersection numbers are given by the following table: $\left({E}_{1}^{\prime }\cdot {l}_{1}^{\prime }\right)=-1,\left({E}_{1}^{\prime }\cdot {l}_{2}\right)=0,\left({E}_{2}\cdot {l}_{1}^{\prime }\right)=2,\left({E}_{2}\cdot {l}_{2}\right)=-1.$
• Proof. ${P}_{1}$  is a determinantal variety defined by quadratic equations. Therefore we obtain the formula for $\overline{NE}\left(Y/X\right)$  . The intersection numbers are obvious.

3 Derived equivalence

Let $X\to Z←{X}^{+}$  and ${X}_{0}\to {Z}_{0}←{X}_{0}^{+}$  be stratified flops defined in the previous section. We shall compare the bounded derived categories of coherent sheaves ${D}^{b}\left(\text{Coh}\left(X\right)\right)$  with ${D}^{b}\left(\text{Coh}\left({X}^{+}\right)\right)$  and ${D}^{b}\left(\text{Coh}\left({X}_{0}\right)\right)$  with ${D}^{b}\left(\text{Coh}\left({X}_{0}^{+}\right)\right)$  . By [10 Lemma 5.6 and Corollary 5.7, the derived equivalence for the former pair implies the latter.
We begin with an easy case $r=1$  , i.e., the standard flop and the usual Mukai flop.
Proposition 3.1. Let $l$  be an arbitrary integer. Then a functor ${\Phi }_{l}:{D}^{b}\left(\text{Coh}\left(X\right)\right)\to {D}^{b}\left(\text{Coh}\left({X}^{+}\right)\right)$  defined by ${\Phi }_{l}\left(a\right)={f}_{*}^{+}\left({f}^{*}a\otimes {\mathcal{O}}_{Y}\left(lE\right)\right)$  is an equivalence.
• Proof. The adjoint functor ${\Psi }_{l}:{D}^{b}\left(\text{Coh}\left({X}^{+}\right)\right)\to {D}^{b}\left(\text{Coh}\left(X\right)\right)$  is given by ${\Psi }_{l}\left(b\right)={f}_{*}\left({f}^{+*}a\otimes {\mathcal{O}}_{Y}\left(\left(n-1-l\right)E\right)\right).$  The category ${D}^{b}\left(\text{Coh}\left(X\right)\right)$  is spanned by the set of sheaves ${\mathcal{O}}_{X}\left(k\right)$  for $l-n+1\le k\le l$  . By the Kodaira vanishing theorem, we have ${\Phi }_{l}:{\mathcal{O}}_{X}\left(k\right)↦{\mathcal{O}}_{Y}\left(k,0\right)\left(lE\right)={\mathcal{O}}_{Y}\left(0,-k\right)\left(\left(l-k\right)E\right)↦{\mathcal{O}}_{{X}^{+}}\left(-k\right)$  and  $\begin{array}{cccc}{\Psi }_{l}:& {\mathcal{O}}_{{X}^{+}}\left(-k\right)↦{\mathcal{O}}_{Y}\left(0,-k\right)\left(\left(n-1-l\right)E\right)={\mathcal{O}}_{Y}\left(k,0\right)\left(\left(n-1-l+k\right)E\right)& & ↦{\mathcal{O}}_{X}\left(k\right).\end{array}$
There is an adjunction morphism of functors $F:{\text{Id}}_{{D}^{b}\left(\text{Coh}\left(X\right)\right)}\to {\Psi }_{l}{\Phi }_{l}$  , which is reduced to the identity when restricted to the open subset $X\W$  . By the above argument, we have isomorphisms $\omega \sim =\Psi \Phi \left(\omega \right)$  for a spanning class $\Omega =\left\{\omega \right\}$  . Since the $\omega$  are invertible sheaves, it follows that the morphisms $F\left(\omega \right)$  are isomorphisms. Therefore, the natural homomorphisms $\Phi :{\text{Hom}}^{p}\left({\omega }_{1},{\omega }^{2}\right)\to {\text{Hom}}^{p}\left({\omega }_{1},\Psi \Phi \left({\omega }^{2}\right)\right)\to {\text{Hom}}^{p}\left(\Phi \left({\omega }_{1}\right),\Phi \left({\omega }^{2}\right)\right)$  for any ${\omega }_{1},{\omega }_{2}\in \Omega$  and $p\in \mathbb{Z}$  are isomorphisms. Then $\Phi$  is an equivalence by [2and [3.
We assume that $r=2$  in the rest of the paper. We consider exact functors between bounded derived categories
 $\begin{array}{c}\begin{array}{cccc}& \Phi :{D}^{b}\left(\text{Coh}\left(X\right)\right)\to {D}^{b}\left(\text{Coh}\left(Y\right)\right)& & \Psi :{D}^{b}\left(\text{Coh}\left(Y\right)\right)\to {D}^{b}\left(\text{Coh}\left(X\right)\right)\end{array}\end{array}$ (3.1)
defined by
 $\begin{array}{cccc}& \Phi \left(a\right)=R{f}_{*}^{+}\left(L{f}^{*}\left(a\right)\otimes {\mathcal{O}}_{Y}\left(\left(2n-5\right){E}_{1}^{\prime }+\left(n-3\right){E}_{2}\right)\right)& & \Psi \left(b\right)=R{f}_{*}\left(L{f}^{+*}\left(b\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)\right).\end{array}$
They are adjoints each other because ${K}_{Y}={f}^{*}{K}_{X}+\left(2n-4\right){E}_{1}^{\prime }+\left(n-3\right){E}_{2}$  .
Lemma 3.2. ${D}^{b}\left(\text{Coh}\left(X\right)\right)$  is spanned by the set of the following locally free sheaves: ${\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right)$  for $0\le i$  , $0\le j$  and $i+j\le n-2$  .
• Proof. By [8, any point sheaf ${\mathcal{O}}_{p}$  for $p\in G$  has a finite locally free resolution whose terms are direct sums of the sheaves ${\text{Sym}}^{i}{S}^{*}\otimes {\mathcal{O}}_{G}\left(j\right)$  for $0\le i$  , $0\le j$  and $i+j\le n-2$  . Hence ${\mathcal{O}}_{{\pi }^{-1}\left(p\right)}$  is resolved by our set. Then so is any point sheaf ${\mathcal{O}}_{x}$  for $x\in X$  because ${\pi }^{-1}\left(p\right)$  is an affine space.
Let ${\mathcal{ℰ}}^{+}$  be a subsheaf of ${S}_{{X}^{+}}^{+}$  such that a local section of the latter belongs to the former if and only if its value in ${S}_{q}^{+}$  at a point $\left(q,B\right)$  is contained in $\text{Im}B$  . We denote by ${\mathcal{ℰ}}_{i}^{+}$  for $i>0$  the image of the natural homomorphism ${\text{Sym}}^{i-1}{S}_{{X}^{+}}^{+}\otimes {\mathcal{ℰ}}^{+}\to {\text{Sym}}^{i}{S}_{{X}^{+}}^{+}$  .
Lemma 3.3. (1) For $0\le j\le n-3$  ,
 $\begin{array}{cccc}& \Phi \left({\mathcal{O}}_{X}\left(j\right)\right)={\mathcal{O}}_{{X}^{+}}\left(-j\right)& & \Psi \left({\mathcal{O}}_{{X}^{+}}\left(-j\right)\right)={\mathcal{O}}_{X}\left(j\right).\end{array}$
(2) Let ${I}_{{W}^{+}}$  be the ideal sheaf of ${W}^{+}\subset {X}^{+}$  . Then $\Phi \left({\mathcal{O}}_{X}\left(n-2\right)\right)={I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-n+2\right).$  (3) For $0  , $0\le j$  and $i+j\le n-3$  ,
 $\begin{array}{cccc}& \Phi \left({\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right)\right)={\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-j\right)& & \Psi \left({\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-j\right)\right)={\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right).\end{array}$
(4) For $0  , $\Phi \left({\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(n-2-i\right)\right)={\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-n+2+i\right).$
• Proof. (1) By Corollary  2.2   $\begin{array}{cccc}& {f}^{*}\left({\mathcal{O}}_{X}\left(j\right)\right)\otimes {\mathcal{O}}_{Y}\left(\left(2n-5\right){E}_{1}^{\prime }+\left(n-3\right){E}_{2}\right)& & ={f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-j\right)\otimes {\mathcal{O}}_{Y}\left(\left(2n-5-2j\right){E}_{1}^{\prime }+\left(n-3-j\right){E}_{2}\right).\end{array}$
Since ${K}_{Y}={f}^{*}{K}_{X}+\left(2n-4\right){E}_{1}^{\prime }+\left(n-3\right){E}_{2}$  , the higher direct images for ${f}^{+}$  vanish because $-\left(2j+1\right){E}_{1}^{\prime }-j{E}_{2}$  is nef. Since $\left(2n-5-2j\right){E}_{1}^{\prime }+\left(n-3-j\right){E}_{2}$  is effective, we have the first formula.
Similarly, we have ${f}^{+*}\left({\mathcal{O}}_{{X}^{+}}\left(-j\right)\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)={f}^{*}{\mathcal{O}}_{X}\left(j\right)\otimes {\mathcal{O}}_{Y}\left(\left(2j+1\right){E}_{1}^{\prime }+j{E}_{2}\right).$  The higher direct images for $f$  vanish because $\left(2j-2n+5\right){E}_{1}^{\prime }+\left(j-n+3\right){E}_{2}$  is nef. $\left(2j+1\right){E}_{1}^{\prime }+j{E}_{2}$  is effective, hence the second formula.
(2) We already proved in (1) that the higher direct images vanish. We obtain our formula from ${f}_{*}^{+}{\mathcal{O}}_{Y}\left(-{E}_{1}^{\prime }-{E}_{2}\right)={I}_{{W}^{+}}.$  (3) We have an exact sequence $0\to {\mathcal{O}}_{{E}_{2}}\left(-{E}_{1}^{\prime }\right)\to {f}^{+*}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{E}_{2}}\left(-{E}_{1}^{\prime }\right)\to {\mathcal{O}}_{{E}_{2}}\left(-{E}_{1}^{\prime }\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-1\right)\to 0.$  where the first term is the image of ${f}^{*}{S}_{X}^{*}$  . We define a decreasing filtration of the sheaf ${f}^{*}{\text{Sym}}^{i}{S}_{X}^{*}$  by locally free subsheaves ${\mathcal{ℱ}}^{k,l}$  for $0\le l\le k\le i$  by  $\begin{array}{cccc}{\mathcal{ℱ}}^{k,l}=& {f}^{*}{\text{Sym}}^{i}{S}_{X}^{*}\cap {f}^{+*}{\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\left(-i{E}_{1}^{\prime }-k{E}_{2}\right)& & \cap \left({f}^{*}{\text{Sym}}^{i}{S}_{X}^{*}\left(-l{E}_{2}\right)+{f}^{+*}{\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\left(-i{E}_{1}^{\prime }-\left(k+1\right){E}_{2}\right)\right).\end{array}$
We have  $\begin{array}{cccc}& {f}^{*}{\text{Sym}}^{i}{S}_{X}^{*}={F}^{0,0}\supset {F}^{1,0}\supset {F}^{1,1}\supset {F}^{2,0}\supset & & \cdot \cdot \cdot \supset {F}^{i-1,i-1}\supset {F}^{i,0}={f}^{+*}{\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\left(-i{E}_{1}^{\prime }-i{E}_{2}\right).\end{array}$
and ${\mathcal{ℱ}}^{k,l}/{\mathcal{ℱ}}^{k,l+1}\sim ={\mathcal{O}}_{{E}_{2}}\left(-i{E}_{1}^{\prime }-k{E}_{2}\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-k+l\right)$  for $0\le l\le k  , where we put ${\mathcal{ℱ}}^{k,k+1}={\mathcal{ℱ}}^{k+1,0}$  . Since $R\left({f}_{2}^{+}{\right)}_{*}{\mathcal{O}}_{{E}_{2}}\left(t{E}_{2}\right)=0$  for $0  , we have $R\left({f}_{2}^{+}{\right)}_{*}\left({\mathcal{ℱ}}^{k,l}/{\mathcal{ℱ}}^{k,l+1}\otimes {\mathcal{O}}_{Y}\left(\left(2n-5-2j\right){E}_{1}^{\prime }+\left(n-3-j\right){E}_{2}\right)\right)=0$  because $i+j\le n-3$  . Therefore,  $\begin{array}{cccccccccc}\Phi \left(& {\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right)\right)& =& {\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-j\right)& & \otimes R{f}_{*}^{+}{\mathcal{O}}_{Y}\left(\left(2n-5-i-2j\right){E}_{1}^{\prime }+\left(n-3-i-j\right){E}_{2}\right)& =& {\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-j\right)\otimes R{f}_{1*}^{+}{\mathcal{O}}_{{X}_{1}^{+}}\left(\left(2n-5-i-2j\right){E}_{1}^{+}\right)& =& {\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-j\right).\end{array}$
For the inverse direction, we climb back along the filtration ${\mathcal{ℱ}}^{k,l}$  . Since $R{f}_{2*}\left({\mathcal{ℱ}}^{k,l}/{\mathcal{ℱ}}^{k,l+1}\otimes {\mathcal{O}}_{Y}\left(\left(i+2j+1\right){E}_{1}^{\prime }+\left(i+j\right){E}_{2}\right)\right)=0$  for $0\le l\le k  , we have  $\begin{array}{cccccccc}& \Psi \left({\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-j\right)\right)& & ={\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right)\otimes R{f}_{*}{\mathcal{O}}_{Y}\left(\left(i+2j+1\right){E}_{1}^{\prime }+\left(i+j\right){E}_{2}\right)& & ={\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right)\otimes R{f}_{1*}{\mathcal{O}}_{{X}_{1}}\left(\left(i+2j+1\right){E}_{1}\right)& & ={\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(j\right).\end{array}$
(4) We already proved in (3) that  $\begin{array}{cccc}& R{f}_{*}^{+}\left({f}^{*}\left({\text{Sym}}^{i}{S}_{X}^{*}\otimes {\mathcal{O}}_{X}\left(n-2-i\right)\right)\otimes {\mathcal{O}}_{Y}\left(\left(2n-5\right){E}_{1}^{\prime }+\left(n-3\right){E}_{2}\right)\right)& & \sim =R{f}_{*}^{+}\left({F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{X}\left(-n+2+i\right)\otimes {\mathcal{O}}_{Y}\left(\left(2i-1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\right).\end{array}$
Moreover  $\begin{array}{cccc}& {R}^{p}{f}_{*}^{+}\left({f}^{+*}\left({\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-n+2+i\right)\right)\otimes {\mathcal{O}}_{Y}\left(\left(i-1\right){E}_{1}^{\prime }-{E}_{2}\right)\right)& & \to {R}^{p}{f}_{*}^{+}\left({F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{X}\left(-n+2+i\right)\otimes {\mathcal{O}}_{Y}\left(\left(2i-1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\right)\end{array}$
is surjective for $p>0$  . Therefore, there are no higher direct images for ${f}^{+}$  .
Since  $\begin{array}{cccc}& 0\to {F}^{i-1,0}\to {f}^{+*}{\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{Y}\left(-i{E}_{1}^{\prime }-\left(i-1\right){E}_{2}\right)& & \to {\mathcal{O}}_{{E}_{2}}\left(-i{E}_{1}^{\prime }-\left(i-1\right){E}_{2}\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-i\right)\to 0\end{array}$
we have  $\begin{array}{cccc}& 0\to {f}_{*}^{+}\left({F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{X}\left(-n+2+i\right)\otimes {\mathcal{O}}_{Y}\left(\left(2i-1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\right)& & \to {\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{X}\left(-n+2+i\right)\to {\mathcal{O}}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-n+2\right)\to 0.\end{array}$
On the other hand, we have  $\begin{array}{cccc}& {f}_{*}^{+}\left({F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{X}\left(-n+2+i\right)\otimes {\mathcal{O}}_{Y}\left(\left(2i-1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\right){|}_{{X}^{+}\{G}^{+}}& & \sim ={\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-n+2+i\right){|}_{{X}^{+}\{G}^{+}}.\end{array}$
By taking the direct image sheaves of both sides under the inclusion morphism ${X}^{+}\{G}^{+}\to {X}^{+}$  , we obtain  $\begin{array}{cccc}& {f}_{*}^{+}\left({F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{X}\left(-n+2+i\right)\otimes {\mathcal{O}}_{Y}\left(\left(2i-1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\right)& & \sim ={\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(-n+2+i\right).\end{array}$
We assume that $n=4$  besides $r=2$  in the following. Then $dimG=4$  , $dim{X}_{0}=8$  , $dimX=9$  , and $dimW=7$  . In particular, $W$  is locally complete intersection. We have ${K}_{Y}={f}^{*}{K}_{X}+4{E}_{1}^{\prime }+{E}_{2}$  .
In this case, Namikawa [14proved that the functors ${\Phi }^{\prime }:{D}^{b}\left(\text{Coh}\left(X\right)\right)\to {D}^{b}\left(\text{Coh}\left(Y\right)\right),{\Psi }^{\prime }:{D}^{b}\left(\text{Coh}\left(Y\right)\right)\to {D}^{b}\left(\text{Coh}\left(X\right)\right)$  defined by ${\Phi }^{\prime }\left(a\right)=R{f}_{*}^{+}\left(L{f}^{*}\left(a\right)\otimes {\mathcal{O}}_{Y}\left(4{E}_{1}^{\prime }+{E}_{2}\right)\right),{\Psi }^{\prime }\left(b\right)=R{f}_{*}\left(L{f}^{+*}\left(b\right)\otimes {\mathcal{O}}_{Y}\right).$  are not equivalences.
Lemma 3.4. Assume that $n=4$  . Then (1) $\Psi \left({I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\right)={\mathcal{O}}_{X}\left(2\right)$  .
(2) $\Psi \left({\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)\right)={\text{Sym}}^{i}{S}^{*}\otimes {\mathcal{O}}_{X}\left(2-i\right)$  for $0  .
• Proof. (1) ${W}^{+}$  is a divisor on ${X}_{0}^{+}$  and ${\phi }^{+}$  is the contraction of $\left(-2\right)$  -curves along the generic points of $\phi \left({W}^{+}\right)$  . Hence ${\mathcal{O}}_{{X}_{0}^{+}}\left(-{W}^{+}\right)={\mathcal{O}}_{{X}_{0}^{+}}\left(2\right)$  . We have an exact sequence $0\to {\mathcal{O}}_{{X}^{+}}\left(-2-{X}_{0}^{+}\right)\to {I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\to {\mathcal{O}}_{{X}_{0}^{+}}\to 0.$  Since ${X}_{0}^{+}$  is a Cartier divisor, we have ${L}_{k}{f}^{+*}{\mathcal{O}}_{{X}_{0}^{+}}=0$  for $k>0$  . Hence $0\to {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2-{X}_{0}^{+}\right)\to {f}^{+*}\left({I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\right)\to {f}^{+*}{\mathcal{O}}_{{X}_{0}^{+}}\to 0.$  On the other hand, $0\to {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2-{X}_{0}^{+}\right)\to {I}_{{W}^{+}}{\mathcal{O}}_{Y}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\to {\mathcal{O}}_{\left({X}_{0}^{+}{\right)}^{\prime \prime }}\to 0$  where $\left({X}_{0}^{+}{\right)}^{\prime \prime }=\left({f}_{*}^{+}{\right)}^{-1}{X}_{0}^{+}$  . Thus  $\begin{array}{cccc}& 0\to {\mathcal{O}}_{\left({E}_{1}^{+}{\right)}^{\prime }+{E}_{2}^{+}}\left(-\left({X}_{0}^{+}{\right)}^{\prime \prime }\right)\to {f}^{+*}\left({I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\right)& & \to {I}_{{W}^{+}}{\mathcal{O}}_{Y}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\to 0.\end{array}$
We have ${\mathcal{O}}_{\left({E}_{1}^{+}{\right)}^{\prime }+{E}_{2}^{+}}\left(-\left({X}_{0}^{+}{\right)}^{\prime \prime }\right)\sim ={\mathcal{O}}_{{E}_{1}^{\prime }+{E}_{2}}\left({E}_{1}^{\prime }+{E}_{2}\right).$  Since $R{f}_{2*}{\mathcal{O}}_{{E}_{2}}\left({E}_{2}\right)=0$  and $R{f}_{1*}{\mathcal{O}}_{{E}_{1}}\left(2{E}_{1}\right)=0$  , we have $R{f}_{*}{\mathcal{O}}_{{E}_{1}^{\prime }+{E}_{2}}\left(2{E}_{1}^{\prime }+{E}_{2}\right)\sim =R{f}_{1*}{\mathcal{O}}_{{E}_{1}}\left(2{E}_{1}\right)\sim =0.$  Therefore, we have  $\begin{array}{cccc}& R{f}_{*}\left({f}^{+*}\left({I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)\right)& & \sim =R{f}_{*}\left({I}_{{W}^{+}}{\mathcal{O}}_{Y}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)\right).\end{array}$
Let ${F}^{+}=\left({X}_{0}^{+}{\right)}^{\prime \prime }\cap \left({E}_{1}^{+}{\right)}^{\prime }\subset Y$  . The projection ${f}^{+}:{F}^{+}\to {G}^{+}$  is a ${\mathbb{P}}^{3}$  -bundle whose fiber $\left({P}_{0}^{+}{\right)}^{\prime }$  is contracted by $f$  . We have ${I}_{{W}^{+}}{\mathcal{O}}_{Y}={I}_{{F}^{+}}\left(-{E}_{1}^{\prime }-{E}_{2}\right)$  Thus ${I}_{{W}^{+}}{\mathcal{O}}_{Y}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)={I}_{{F}^{+}}{f}^{*}{\mathcal{O}}_{X}\left(2\right)\otimes {\mathcal{O}}_{Y}\left(4{E}_{1}^{\prime }+{E}_{2}\right).$  On the other hand, there is an exact sequence we have ${\mathcal{O}}_{\left({P}_{0}^{+}{\right)}^{\prime }}\left({E}_{1}^{\prime }\right)\sim ={\mathcal{O}}_{{\mathbb{P}}^{3}}\left(-1\right)$  and ${\mathcal{O}}_{\left({P}_{0}^{+}{\right)}^{\prime }}\left({E}_{2}\right)\sim ={\mathcal{O}}_{{\mathbb{P}}^{3}}\left(2\right)$  . Hence ${\mathcal{O}}_{\left({P}_{0}^{+}{\right)}^{\prime }}\left(4{E}_{1}^{\prime }+{E}_{2}\right)\sim ={\mathcal{O}}_{{\mathbb{P}}^{3}}\left(-2\right)$  . It follows that $R{f}_{*}{\mathcal{O}}_{{F}^{+}}\left(4{E}_{1}^{\prime }+{E}_{2}\right)=0$  . Therefore, $R{f}_{*}\left({I}_{{F}^{+}}{\mathcal{O}}_{Y}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)\right)\sim ={\mathcal{O}}_{X}\left(2\right).$  (2) We have an exact sequence $0\to {\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)\to {\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)\to {\mathcal{O}}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\to 0.$  Thus the torsion part of ${f}^{+*}{\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)$  is isomorphic to that of ${f}^{+*}{I}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)$  . Therefore, by the argument in (1), we have $\Psi \left({\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)\right)\sim =R{f}_{*}\left({f}^{+*}{\mathcal{ℰ}}_{i}^{+}/\text{torsion}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(i-2\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)\right).$  We have an exact sequence  $\begin{array}{cccccc}& 0\to {f}^{+*}{\mathcal{ℰ}}_{i}^{+}/\text{torsion}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(i-2\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)& & \to {f}^{+*}\left({\text{Sym}}^{i}{S}_{{X}^{+}}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)& & \to {f}^{+*}\left({\mathcal{O}}_{{W}^{+}}\otimes {\mathcal{O}}_{{X}^{+}}\left(-2\right)\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)\to 0.\end{array}$
Therefore, we have an exact sequence  $\begin{array}{cccc}& 0\to {f}^{+*}{\mathcal{ℰ}}_{i}^{+}/\text{torsion}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(i-2\right)\otimes {\mathcal{O}}_{Y}\left({E}_{1}^{\prime }\right)& & \to {F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(i-2\right)\otimes {\mathcal{O}}_{Y}\left(\left(i+1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\to Q\to 0\end{array}$
where the cokernel $Q$  has a decomposition as follows $0\to {\mathcal{O}}_{{F}^{+}}\left(-{E}_{2}\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\to Q\to {\mathcal{O}}_{{E}_{1}^{\prime }}\left({E}_{1}^{\prime }-{E}_{2}\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\to 0.$  As in (1), ${F}^{+}=F={X}_{0}^{\prime \prime }\cap {E}_{1}^{\prime }$  is a ${\mathbb{P}}^{3}$  -bundle over $G$  with fibers ${P}_{0}^{\prime }$  such that ${\mathcal{O}}_{{P}_{0}^{\prime }}\left({E}_{1}^{\prime }\right)\sim ={\mathcal{O}}_{{\mathbb{P}}^{3}}\left(-1\right)$  and ${\mathcal{O}}_{{P}_{0}^{\prime }}\left({E}_{2}\right)={\mathcal{O}}_{{\mathbb{P}}^{3}}\left(2\right)$  . Thus ${\mathcal{O}}_{{F}^{+}}\left(-{E}_{2}\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)\otimes {\mathcal{O}}_{{P}_{0}^{\prime }}\sim ={\mathcal{O}}_{{P}_{0}^{\prime }}\left(4{E}_{1}^{\prime }+{E}_{2}\right)\sim ={\mathcal{O}}_{{P}_{0}^{\prime }}\left(-2\right)$  and we have $R{f}_{*}{\mathcal{O}}_{{F}^{+}}\left(-{E}_{2}\right)=0$  . We have also  $\begin{array}{cccccc}& R{f}_{*}{\mathcal{O}}_{{E}_{1}^{\prime }}\left({E}_{1}^{\prime }-{E}_{2}\right)\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(-2\right)& & \sim =R{f}_{*}{\mathcal{O}}_{{E}_{1}^{\prime }}\left(3{E}_{1}^{\prime }+{E}_{2}\right)\otimes {\mathcal{O}}_{X}\left(2\right)& & \sim =R{f}_{1*}{\mathcal{O}}_{{E}_{1}}\left(3{E}_{1}\right)\otimes {\mathcal{O}}_{X}\left(2\right)\sim =0.\end{array}$
Therefore  $\begin{array}{cccccc}& \Psi \left({\mathcal{ℰ}}_{i}^{+}\otimes {\mathcal{O}}_{{X}^{+}}\left(i-2\right)\right)& & \sim =R{f}_{*}\left({F}^{i-1,0}\otimes {f}^{+*}{\mathcal{O}}_{{X}^{+}}\left(i-2\right)\otimes {\mathcal{O}}_{Y}\left(\left(i+1\right){E}_{1}^{\prime }+\left(i-1\right){E}_{2}\right)\right)& & \sim =R{f}_{*}\left({F}^{i-1,0}\otimes {\mathcal{O}}_{Y}\left(\left(5-i\right){E}_{1}^{\prime }+{E}_{2}\right)\right)\otimes {\mathcal{O}}_{X}\left(2-i\right).\end{array}$
If $i=1$  , then ${F}^{0,0}={f}^{*}{\text{Sym}}^{i}{S}^{*}$  and $R{f}_{*}{\mathcal{O}}_{Y}\left(4{E}_{1}^{\prime }+{E}_{2}\right)={\mathcal{O}}_{X}$  , hence the result.
If $i=2$  , then ${\mathcal{ℱ}}^{0,0}/{\mathcal{ℱ}}^{1,0}\sim ={\mathcal{O}}_{{E}_{2}}\left(-2{E}_{1}^{\prime }\right)$  , hence $R{f}_{*}\left({\mathcal{ℱ}}^{0,0}/{\mathcal{ℱ}}^{1,0}\otimes {\mathcal{O}}_{Y}\left(3{E}_{1}^{\prime }+{E}_{2}\right)\right)=0$  and we complete the proof.
Theorem 3.5. If $n=4$  , then $\Phi$  and $\Psi$  in ( 3.1 ) are equivalences.
• Proof. There is an adjunction morphism of functors $F:{\text{Id}}_{{D}^{b}\left(\text{Coh}\left(X\right)\right)}\to \Psi \Phi$  , which is reduced to the identity when restricted to the open subset $X\W$  .
By Lemmas  3.2 ,  3.3 and  3.4 , we have isomorphisms $\omega \sim =\Psi \Phi \left(\omega \right)$  for the spanning class $\Omega =\left\{\omega \right\}$  given by Lemma  3.2 . Since the $\omega$  are locally free sheaves, it follows that the morphisms $F\left(\omega \right)$  are isomorphisms. Therefore, the natural homomorphisms $\Phi :{\text{Hom}}^{p}\left({\omega }_{1},{\omega }^{2}\right)\to {\text{Hom}}^{p}\left({\omega }_{1},\Psi \Phi \left({\omega }^{2}\right)\right)\to {\text{Hom}}^{p}\left(\Phi \left({\omega }_{1}\right),\Phi \left({\omega }^{2}\right)\right)$  for any ${\omega }_{1},{\omega }_{2}\in \Omega$  and $p\in \mathbb{Z}$  are isomorphisms. Then $\Phi$  is an equivalence by [2and [3.
Remark 3.6. (1) Let $F=\oplus \omega$  be the sum of the spanning sheaves given in Lemma  3.2 . Then $F$  is not an almost exceptional object in the sense of [1in our case $n=4$  and $r=2$  , because ${R}^{1}{\phi }_{*}{\mathcal{O}}_{X}\left(-2\right)\ne 0$  implies ${\text{Hom}}^{1}\left(F,F\right)\ne 0$  .
(2) In order to extend our argument to the case $n\ge 5$  , the Eagon-Northcott resolution on ${X}_{0}^{+}$  would be useful ([7):
 $\begin{array}{cccccc}& 0\to {\pi }_{0}^{+*}\left({\text{Sym}}^{n-2}{S}^{+2}{\right)}^{*}{\otimes }^{n}\bigwedge {V}^{*n}\stackrel{{B}_{*}}{\to }\cdot \cdot \cdot \to {\pi }_{0}^{+*}\left({\text{Sym}}^{k}{S}^{+2}{\right)}^{*}{\otimes }^{k+2}\bigwedge {V}^{*n}& & \stackrel{{B}_{*}}{\to }{\pi }_{0}^{+*}\left({\text{Sym}}^{k-1}{S}^{+2}{\right)}^{*}{\otimes }^{k+1}\bigwedge {V}^{*n}\stackrel{{B}_{*}}{\to }...\stackrel{{B}_{*}}{\to }{\pi }_{0}^{+*}\left({S}^{+2}{\right)}^{*}{\otimes }^{3}\bigwedge {V}^{*n}& & \stackrel{{B}_{*}}{\to }{\mathcal{O}}_{{X}_{0}^{+}}{\otimes }^{2}\bigwedge {V}^{*n}\stackrel{{}^{2}\bigwedge B}{\to }{{\pi }_{0}^{+*}}^{2}\bigwedge {S}^{+2}\to {\mathcal{O}}_{{W}^{+}}\left(-1\right)\to 0.\end{array}$
References

1. R. Bezrukavnikov and D. Kaledin. McKay equivalence for symplectic resolutions of singularities. math.AG/0401002.
2. A. Bondal, D. Orlov. Semiorthogonal decomposition for algebraic varieties. alg-geom/9506012.
3. Tom Bridgeland. Equivalences of triangulated categories and Fourier-Mukai transforms. math.AG/9809114. Bull. London Math. Soc. 31 (1999), no. 1, 25–34.
4. Tom Bridgeland. Flops and derived categories. math.AG/0009053. Invent. Math. 147 (2002), no. 3, 613–632.
5. Tom Bridgeland, Alastair King and Miles Reid. Mukai implies McKay: the McKay correspondence as an equivalence of derived categories. math.AG/9908027. J. Amer. Math. Soc. 14 (2001), no. 3, 535–554.
6. J. -C. Chen. Flops and equivalences of derived categories for threefolds with only Gorenstein singularities. math.AG/0202005.
7. David Eisenbud. Commutative Algebra with a View Toward Algebraic Geometry. Springer-Verlag, 1995.
8. M. M. Kapranov. Derived category of coherent sheaves on Grassmann manifolds. (Russian) Izv. Akad. Nauk SSSR Ser. Mat. 48 (1984), no. 1, 192–202.
9. Yujiro Kawamata. Francia's flip and derived categories. math.AG/0111041. Algebraic geometry, 197–215, de Gruyter, Berlin, 2002.
10. Yujiro Kawamata. D-equivalence and K-equivalence. math.AG/0205287. J. Differential Geom. 61 (2002), no. 1, 147–171.
11. Yujiro Kawamata. Log Crepant Birational Maps and Derived Categories. math.AG/0311139.
12. Eyal Markman. Brill-Noether duality for moduli spaces of sheaves on K3 surfaces. math.AG/9901072. Journal of Algebraic Geometry 10 (2001), no. 4, 623-694.
13. Yoshinori Namikawa. Mukai flops and derived categories. math.AG/0203287. J. Reine Angew. Math. 560 (2003), 65–76.
14. Yoshinori Namikawa. Mukai flops and derived categories II. math.AG/0305086.

Department of Mathematical Sciences, University of Tokyo, Komaba, Meguro, Tokyo, 153-8914, Japan kawamata@ms.u-tokyo.ac.jp