Gushel-Mukai fourfolds are smooth four-dimensional algebraic varieties defined as intersections of quadrics and Grassmannians. They form a rich family with a moduli space that has been studied intensively, but a basic question has resisted answer: are they rational? Can they be birationally mapped to projective space?
Benedetti, Manivel, and Perrin (arXiv:2603.17487) prove that very general Gushel-Mukai fourfolds are irrational, using quantum cohomology as the obstruction.
The small quantum cohomology ring of an algebraic variety encodes counts of rational curves — the enumerative geometry of lines, conics, and higher-degree curves lying on the variety. Computing this ring requires understanding how curves of each degree contribute to the multiplication table. For Gushel-Mukai fourfolds, this computation reveals that the quantum cohomology has a structure incompatible with rationality.
The proof also characterizes the exceptional rational cases: those Gushel-Mukai fourfolds that are rational turn out to have quantum cohomology identical to that of K3 surfaces — two-dimensional objects with rich geometry and trivial canonical bundle. The rational fourfolds are, in a precise cohomological sense, behaving like surfaces. The generic fourfold doesn't, and that failure of surface-like behavior is what prevents rationality.
This is enumerative geometry solving a birational problem. The curves on the variety — how many there are, how they intersect — determine whether the variety can be straightened out into projective space. The answer is usually no, and the quantum cohomology ring provides a computable certificate of the obstruction. Count the curves, and the irrationality follows.