GIT quotients of products of projective planes (Q991579)

The group \(G=\text{SL}_3(\mathbb{C})\) acts on the projective complex plane \(\mathbb{P}^2(\mathbb{C})\) in a natural way which can be extended via the diagonal to an action of \(G\) on the variety \(X:=(\mathbb{P}^2(\mathbb{C}))^n\). For an ample \(G\)-linearized line bundle \(L\in\text{Pic}^{G}(X)\) over \(X\) one has the notion of both stable points \(X^{S}(L)\) and semistable points \(X^{SS}(L)\). These allow for the construction of the orbit space \(X^{S}(L) /G\) and the categorical quotient \(X^{SS}(L) //G\) respectively, and it is known that \(X^{S}(L) /G\) is an open subset of \(X^{SS}(L) //G\) under the Zariski topology. This determination depends on the choice of \(L\). In the work under review, the author investigates how varying \(L\) changes both \(X^{SS}(L)\) and \(X^{SS}(L) //G\). Results by Dolgachev-Hu and Thaddeus show that only a finite number of such quotients can be obtained by changing \(L\). Through the use of a numerical criterion for semi-stability the author constructs his own proof of this fact. This numerical criterion, along with a theory of elementary transformations, is then applied to specific cases. In the case \(n=5\) it is shown that there are precisely six (nontrivial) possibilities for the quotient. In short, any \(X^{SS}(L) //G\) is either the product of two projective lines or projective plane with at most four points blown up. In the case \(n=6\) there are at most 38 different quotients. Moreover in the case \(n=6\) the singularities that arise whenever \(X^{S}(G) \neq X^{SS}(G)\) are classified with the help of the Étale Slice Theorem.