The length classification of threefold flops via noncommutative algebras (Q1711917)

Starting in the late 1980's, techniques of (commutative) algebraic geometry have been used to study noncommutative algebra. See for example [\textit{M. Artin}, Lond. Math. Soc. Lect. Note Ser. 238, 1--19 (1997; Zbl 0888.16025)] for an introduction to these ideas. More recently, noncommutative algebra has been used to study algebraic geometry. For instance, the paper under review uses quivers to refine a classification of threefold flops. \par A \textit{flop} is a type of birational transformation. In the seminal paper [\textit{J. Kollár}, Nagoya Math. J. 113, 15--36 (1989; Zbl 0645.14004)], it was shown that if $X$ and $Y$ are threefolds, with minor singularities and nef canonical bundles, then any birational transformation between $X$ and $Y$ can be factored into flops. Thus flops are an important part of the minimal model program for threefolds. In [\textit{S. Katz} and \textit{D. R. Morrison}, J. Algebraic Geom. 1, No. 3, 449--530 (1992; Zbl 0788.14036)], flops were classified by a length invariant. The length $\ell$ can take on the integer values 1--6, and for each $\ell$ there exists a \textit{universal flop} of length $\ell$. Any flop of length $\ell$ can then be constructed from this universal flop, at least in theory. \par From a practical standpoint, these constructions are difficult. The paper under review simplifies these calculations using noncommutative algebra. A \textit{universal flopping algebra} of length $\ell$ is introduced, realized as the path algebra of certain quivers. The main theorem (Theorem 1.3) gives explicit, compact descriptions of the quivers involved. \par The construction works for all possible lengths 1--6. Previously, [\textit{C. Curto} and \textit{D. R. Morrison}, J. Algebr. Geom. 22, No. 4, 599--627 (2013; Zbl 1360.14053)] used matrix factorization to study flops, with theorems for lengths 1 and 2, and conjectures for higher lengths. The current paper recovers those theorems and calculates factorizations that had been conjectured for lengths 3--6. Contraction algebras, introduced by \textit{W. Donovan} and \textit{M. Wemyss} [Duke Math. J. 165, No. 8, 1397--1474 (2016; Zbl 1346.14031)], are also calculated. \par Several examples of universal flops and universal flopping algebras are also given, as well as noncommutative crepant resolutions [\textit{M. van den Bergh}, in: The legacy of Niels Henrik Abel. Papers from the Abel bicentennial conference, University of Oslo, Oslo, Norway, June 3--8, 2002. Berlin: Springer. 749--770 (2004; Zbl 1082.14005)]. An appendix gives Magma code for calculating examples that would be too unwieldy to include in the paper. \par This paper is a nice example of noncommutative algebra giving back to algebraic geometry. Such cross-pollination will undoubtedly continue.