Algebraic geometers study spaces called varieties, and one powerful approach involves encoding the geometry of a space using combinatorial data like polytopes and lattices (think of a grid of points with extra structure). Over the past few decades, researchers have developed several generalizations of a classical theory called toric geometry, including frameworks called Newton-Okounkov bodies, cluster varieties, and toric degenerations. A key feature of these generalizations is that the relevant combinatorial objects are connected by piecewise-linear transformations called mutations, which can change the underlying grid structure in a controlled way.
Polyptych lattices were introduced to provide a single unified combinatorial framework that captures all these different mutation-related structures at once. Rather than working with just one lattice, a polyptych lattice is a collection of lattices glued together by these piecewise-linear mutation maps. An important property some polyptych lattices enjoy is called detropicalizability, which means they can be translated back into honest geometric spaces, allowing the combinatorial machinery to produce real algebraic varieties.
The paper presents a concrete recipe for building new pairs of polyptych lattices from simpler ingredients. Starting with a pair of well-behaved geometric objects called Gorenstein-Fano polytopes, the authors construct a matching pair of polyptych lattices that are dual to each other in a precise sense. Importantly, if the starting ingredients already have the detropicalizability property, the newly constructed lattices inherit it automatically via an explicit formula. This provides a large and concrete library of examples that researchers can use to study and test the broader theory, and the construction also works with data coming from cluster algebras, another active area of modern mathematics.