Complex ADHM equations and sheaves on \(\mathbb P ^3\) (Q926848): Difference between revisions

Let \(V\), \(W\) be complex vector spaces of dimensions \(c\) and \(r\), respectively. An \textit{ADHM datum} is a quadruple \(X=(B_1,B_2,i,j)\) with \(B_1,\;B_2\in \text{End}(V)\), \(i\in \text{Hom}(W,V)\) and \(j\in \text{Hom}(V,W)\). It satisfies the complex ADHM equation if \([B_1,B_2]+ij=0\). These notions, toghether with the real ADHM equation, appear in the famous paper of \textit{M. Atiyah, V. Drinfeld, N. Hitchin} and \textit{Yu. Manin} [Phys. Lett. A 65, 185--187 (1978; Zbl 0424.14004)] where framed \(\text{SU}(r)\)-instantons on \(S^4\) are described in algebro-geometric terms. To an ADHM datum one can associate two morphisms of sheaves on \({\mathbb P}^2={\mathbb C}{\mathbb P}^2\) (with homogeneous coordinates \([x:y:z]\)) \[ V\otimes {\mathcal O}_{{\mathbb P}^2}(-1)\overset {\alpha}\longrightarrow (V\oplus V\oplus W)\otimes {\mathcal O}_{{\mathbb P}^2}\overset {\beta}\longrightarrow V\otimes {\mathcal O}_{{\mathbb P}^2}(1) \] where \(\alpha \) has components \(zB_1+x{\text{id}}_V\), \(zB_2+y{\text{id}}_V\), \(zj\), and \(\beta \) has components \(-zB_2-y{\text{id}}_V\), \(zB_1+x{\text{id}}_V\), \(zi\). The complex ADHM equation is equivalent to \(\beta \alpha =0\). In this case, \(\alpha \) is a monomorphism of sheaves and \(\beta \) is an epimorphism if and only if the ADHM datum is \textit{stable} which means that there is no proper subspace \(S\subset V\) such that \(B_k(S)\subseteq S\), \(k=1,2\), and \(i(W)\subseteq S\). If all of these conditions are satisfied, \({\mathcal F}:=\text{Ker}\beta /\text{Im}\alpha \) is a torsion free sheaf on \({\mathbb P}^2\) such that the restriction of \(\mathcal F\) to the line at infinity \({\ell}_{\infty}:=\{z=0\}\) is canonically isomorphic to \(W\otimes {\mathcal O}_{{\ell}_{\infty}}\). In the paper under review, the authors study pencils of ADHM data : \[ X(z,w)=z(B_{11},B_{12},i_1,j_1)+w(B_{21},B_{22},i_2,j_2),\;[z:w]\in {\mathbb P}^1, \] with all the members satisfying the complex ADHM equation. Such a pencil is called \textit{stable} if all of its members are stable and \textit{semi-stable} if it has at least one stable member. To a pencil of ADHM data one can associate a complex on \({\mathbb P}^3={\mathbb C}{\mathbb P}^3\) (with homogeneous coordinates \([x:y:z:w]\)) : \[ V\otimes {\mathcal O}_{{\mathbb P}^3}(-1)\overset {\tilde \alpha}\longrightarrow (V\oplus V\oplus W)\otimes {\mathcal O}_{{\mathbb P}^3} \overset {\tilde \beta}\longrightarrow V\otimes {\mathcal O}_{{\mathbb P}^3}(1) \] where \(\tilde \alpha \) is a monomorphism and \(\tilde \beta \) is an epimorphism if and only if the pencil is stable. The authors show that putting \({\mathcal F}:=\text{Ker}{\tilde \beta}/\text{Im}{\tilde \alpha}\) one gets a 1-1 correspondence between pencils of ADHM data and torsion free sheaves \(\mathcal F\) on \({\mathbb P}^3\) with \(c_1({\mathcal F})=0\), \(c_2({\mathcal F})=c\), \(\text{rk}{\mathcal F}=r\) and \({\text{H}}^0({\mathcal F}(-1))= {\text{H}}^1({\mathcal F}(-2))={\text{H}}^2({\mathcal F}(-2))= {\text{H}}^3({\mathcal F}(-3))=0\) plus a fixed isomorphism \({\mathcal F}| {\ell}_{\infty}\simeq W\otimes {\mathcal O}_{{\ell}_{\infty}}\), where \({\ell}_{\infty}=\{z=w=0\}\). They also prove that if the pencil is only semi-stable then \(Q:=\text{Coker}{\tilde \beta}\) is supported at finitely many points and \({\mathcal F}:=\text{Ker}{\tilde \beta}/\text{Im}{\tilde \alpha}\) is a torsion free sheaf with \({\text{H}}^0({\mathcal F}(-1))={\text{H}}^1({\mathcal F}(-2))= {\text{H}}^3({\mathcal F}(-3))=0\). When \(r=1\), such an \(\mathcal F\) must be the ideal sheaf of a curve in \({\mathbb P}^3\). The authors conjecture that any torsion free sheaf \(\mathcal F\) satisfying these vanishing conditions and which is trivial on the line at infinity \({\ell}_{\infty}\) arises from a semi-stable pencil of ADHM data, but there are rank 3 locally free sheaves on \({\mathbb P}^3\) contradicting this conjecture.